Molecular Aspects of Medicine 30 (2009) 191–296
Contents lists available at ScienceDirect
Molecular Aspects of Medicine journal homepage: www.elsevier.com/locate/mam
Review
The proteasomal system Tobias Jung a, Betül Catalgol a,b, Tilman Grune a,* a b
Institute of Biological Chemistry and Nutrition, University Hohenheim, Garbenstrasse 28, 70593 Stuttgart, Germany Department of Pharmaceutical Toxicology, Faculty of Pharmacy, Istanbul University, Istanbul, Turkey
a r t i c l e
i n f o
Article history: Received 3 March 2009 Received in revised form 7 April 2009 Accepted 7 April 2009
Keywords: Proteasome UPS Ubiquitin Aging Cancer Protein oxidation
a b s t r a c t Rising interest in the mechanism and function of the proteasomes and the ubiquitin system revealed that it is hard to find any aspect of the cellular metabolic network that is not directly or indirectly affected by the degradation system. This includes the cell cycle, the ‘‘quality control’’ of newly synthesized proteins (ERAD), transcription factor regulation, gene expression, cell differentiation, immune response or pathologic processes like cancer, neurodegenerative diseases, lipofuscin formation, diabetes, atherosclerosis, inflammatory processes or cataract formation and in addition to that the aging process itself and the degradation of oxidized proteins, in order to maintain cell homeostasis. But also this seems to be only a small aspect of the general view. The various regulator proteins that are able to change the rate or specificity of proteolysis, fitting it out for highly specialized tasks, or the precise regulation of the half-life of cellular proteins by ubiquitin-mediated degradation shape the proteasome and the ubiquitin–proteasome system into a fascinating and essential part of cellular function in the three kingdoms of bacteria, plants and animals. This review tries to summarize the current knowledge on the proteasome and the ubiquitin–proteasomal system, including the cellular functions of this system. Ó 2009 Elsevier Ltd. All rights reserved.
Contents 1. 2.
3.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The 20S proteasome . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. The proteasomal alpha-subunits (a-subunits). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. The proteasomal beta-subunits (b-subunits) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. Intracellular assembly of the 20S proteasome. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4. Modeling of the 20S proteasomal proteolysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Regulation of the 20S proteasome . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. The 19S regulator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. The ubiquitin–proteasome system (UPS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. The immunoproteasome . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. The thymus specific proteasome (thymoproteasome) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . NO as a regulator of proteasomal activity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5. 3.6. The 11S regulator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.7. The hybrid proteasome (PA28-20S-PA700) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.8. The PA200 regulator protein. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.9. UPS inhibitors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.9.1. Inhibitors of the proteasome. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.9.2. Inhibitors of deubiquitinating enzymes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
* Corresponding author. Tel.: +49 711 459 24060; fax: +49 711 459 23386. E-mail address:
[email protected] (T. Grune). 0098-2997/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.mam.2009.04.001
. . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . .
192 192 196 197 200 201 202 202 203 205 206 207 207 210 211 211 212 216
192
4.
5.
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
The proteasome in cell physiology, pathology and aging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Proteasome and transcription factor degradation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. The ‘‘quality control’’ in protein folding and endoplasmic-reticulum-associated degradation (ERAD) . . . . . . . . . . . 4.3. The role of the proteasome in antigen processing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4. The degradation of damaged proteins – a function of the 20S proteasome . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4.1. Oxidative stress and protein modification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4.2. Different stages of protein oxidation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4.3. Proteasomal degradation of oxidized proteins. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5. The proteasomal system and aging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6. The proteasome in neurodegeneration and stroke . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6.1. The role of the proteasome in Alzheimer’s disease (AD). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6.2. Huntington’s disease (HD). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6.3. Amyotrophic lateral sclerosis (ALS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6.4. Friedreich ataxia (FA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6.5. Parkinson’s disease (PD) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6.6. Stroke . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.7. Cystic fibrosis (CF). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.8. Atherosclerosis (AS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.9. Rheumatism and autoimmune disease . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.10. The proteasome and viral infections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.11. The proteasome and age related macular degeneration (AMD) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.12. The proteasome in the eye lens – involvement in cataract formation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.13. The proteasome and ethanol-induced liver injury . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.14. The proteasome and diabetes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.15. Proteasome and skin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.15.1. Skin carcinogenesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.15.2. Scleroderma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.15.3. Psoriasis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.15.4. Photoaging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.15.5. Intrinsic skin aging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.16. Role of the proteasomal system in cancer and therapy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.16.1. General role of the UPS in cancer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.16.2. Von Hippel–Lindau disease . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.16.3. Proteasome inhibitors in cancer therapy (multiple myeloma) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.17. Cachexia (muscle wasting) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
217 217 219 223 225 226 227 231 233 234 236 239 241 242 243 244 245 246 247 248 248 249 250 250 251 252 254 254 255 255 257 257 258 260 262 263 264 264
1. Introduction To maintain the cellular functionality and viability, it is important to subject proteins to a defined turnover. In order to realize this, damaged, modified, misfolded proteins or proteins that have become‘ unnecessary must be recognized and degraded. During evolution several systems have developed for this purpose, including the proteasomal system, which is the most important of them, at least in the cytosol. The proteasomal system consists of the 20S ‘‘core’’ proteasome and a set of regulator proteins that can change its activities and specificities. In this review we will describe the current knowledge on structure and assembly of the proteasome its intracellular functions and its role in pathologies. For simplicity in the following manuscript, the 20S ‘‘core’’ proteasome will be termed as ‘‘proteasome’’. All other forms will be termed according to the attached regulators. 2. The 20S proteasome The 20S ‘‘core’’ proteasome is the main particle of the proteasomal system, a very complex cellular structure involved in the proteolytic degradation of oxidized proteins, life-span regulation of proteins (Dick et al., 1991; Grune et al., 1996; Takeuchi and Toh-e, 1997; Sitte et al., 1998; Friguet et al., 2000; Davies, 2001), protein ‘‘quality control’’ (Schmitz and Herzog, 2004; Lord et al., 2000; Plemper and Wolf, 1999; Hampton, 2002; Brodsky and McCracken, 1999), cell cycle regulation (Takeuchi and Toh-e, 1997), gene expression (Zimmermann et al., 2001; Wu et al., 2000; Blagosklonny et al., 1996; Dembla-Rajpal et al., 2004; Stitzel et al., 2001), stress (Mathew et al., 1998; Sulahian et al., 2006; Kahn et al., 2008; Hahn et al., 2006; Bregegere et al., 2006) and immune responses (Preckel et al., 1999; Kloetzel et al., 1999; Schwarz et al., 2000; Khan et al., 2001; Osterloh et al., 2006; Borissenko and Groll, 2007), carcinogenesis (Clawson, 1996; Scheffner and
193
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
Model
5 nm
X-ray structure
X-ray structure (open)
α
Archaea proteasome T. acidophilum
β
S. cerevisia α6
α7 α1
Eukaryotic 20S proteasome
β1 β7 α7
α5
β6
α6
α3
α2
β2
α4
β4
β3 β6
β5 α5
α4
B. bovis
Fig. 1. The archaea, yeast and mammalian proteasome. The left column shows a simplified model of the ancestral archaea proteasome (upper image) and the ‘‘evolutionary developed’’ yeast/mammalian proteasome (lower image). The single subunits are color-coded and represented by spheres. The archaeic proteasome contains only one kind of and a- (blue) and b-subunits (green), arranged in two homodimeric heptameric stacked rings (a7b7b7a7). The 20S proteasome from yeast and mammals shows the same arrangement of subunits, but contains each seven different a- and b-subunits, grouped in four heteromeric rings. The middle column shows a three dimensional reconstruction of the holoprotein from X-ray crystallographic data: the upper image (archaea) has a resolution of 3.4 Å (Lowe et al., 1995), the middle (yeast) 2.40 Å (Groll et al., 1997), and the bottom image (bovine) 2.75 Å (Unno et al., 2002). The three dimensional structure was calculated from X-ray crystallographic data (available at RCSB Protein Data Bank, www.rcsb.org) using the electron density to limit the effective surface of the molecule respective its single subunits. The different subunits show the same color-code like the simplified models on the left. The most right column displays the same as the middle one, except that three subunits from the front of every ring are removed in order to depict the inside of the holoprotein. (For interpretation of the references in colour in this figure legend, the reader is referred to the web version of this article.)
Whitaker, 2003; Moriishi et al., 2007), DNA repair (Walters et al., 2003; Bergink et al., 2006; Krogan et al., 2004) and probably many more cellular functions and regulatory pathways that might be partially not even discovered yet. The term ‘‘20S’’ points towards the sedimentation constant of the isolated ‘‘core’’ proteasome (Hough et al., 1987). The 20S proteasome is the central part of the proteasomal system which in turn is an essential part of every eukaryotic (both plants and animals) and of some prokaryotic cells. The mammalian 20S ‘‘core’’ proteasome is a hollow cylindrical structure (160 100 Å) build of four homologous rings (two a- and two b-rings), arranged in the sequence abba. Each a- and b-ring contains seven subunits. The tree dimensional structure of the proteasome has been investigated via X-ray crystallography in various organisms (Lowe et al., 1995; Groll et al., 1997; Tomisugi et al., 2000; Unno et al., 2002,) (Fig. 1). Two basic kinds of the 20S proteasomes are known: the ancestral form found in archaea (like Thermoplasma acidophilum), showing the same arrangement of subunits as in the mammalian proteasome, but contain only a single form of a- and b-subunits, building heptameric rings (7a and 7b). The single a- and b-subunit of the archaeic proteasome is depicted in Fig. 2. The 20S proteasome as found in animals, yeast and plants contains seven different a- and b-subunits, resulting in the complete arrangement a17b17b17a17 (see also Fig. 1). The mammalian and yeast 20S proteasome are sufficiently homologous to compare and extrapolate experimental results. Each ring of both the yeast and mammalian proteasome contains seven different subunits: a1–a7, respectively b1–b7 (Fig. 3), showing individual molecular masses between 20 and 30 kDa (Table 1). The whole proteasome summarizes to an overall mass of about 700 kDa. The outer a-rings regulate the substrate access to the inner proteolytic chamber of the proteasome that is formed by the b-rings. The inner chamber is subdivided into two fore chambers found between an a- and a b-ring, and the main chamber that is located between the b-rings (Fig. 4). One function of the outer a-rings is the binding of different regulator proteins (see Chapter 3), that are able to change the activity and/or the specificity of the ‘‘core’’ particle. Until now, several different
194
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
proteasomal axis
proteasomal axis H3
H0
H4
H4
Thr1
H8 H5 C
H5
H1
N
N
H3
H2 H2
α-subunit
H1
β-subunit
Fig. 2. The a- and b-subunits of the proteasome of Thermoplasma acidophilum. The left image displays the a-subunits of the archaeic proteasome and the proteasomal axis as axis of symmetry. The subunit contains nine major helical structures (H0–H8), while ‘‘N’’ indicates the N-terminal end of the protein. The right image shows the same for the proteolytic active b-subunit of the archaeic proteasome, containing five helices (H1–H5), while ‘‘N’’ and ‘‘C’’ indicate the corresponding ends of the amino acids chain. ‘‘Thr1’’ indicates the proteolytic active centre of the b-subunit. The high amount of homology of these both subunits is evident. Both images are recalculated from X-ray crystallographic data (available at RCSB Protein Data Bank, www.rcsb.org), original data from Lowe et al. (1995).
proteasome regulators are known: the 11S regulator particle, that is termed in most organisms ‘‘PA28’’ (build of three different subunits, PA28a, PA28b and PA28c in diverse combinations) or ‘‘REG’’ and its analogue in Trypanosoma brucei called ‘‘PA26’’ that are ATP-independent. Furthermore, the ATP-dependent 19S regulator (also called ‘‘PA700’’) and an analogous particle found in archaea called ‘‘PAN’’. A further regulator in the nucleus is called PA200. Until now, three forms (PA200i, PA200ii and PA200iii) are known, that are involved in spermatogenesis (Khor et al., 2006) and DNA-repair (Ustrell et al., 2002). Only PA200i seems to bind to the proteasome, the others are found in the nuclear foci (Blickwedehl et al., 2007) not associated with the 20S proteasomal particle. Furthermore, PR39 and PI31 are known, two natural inhibitors of the 20S proteasome. PR39, a short peptide containing 39 amino acids, is a non-competitive mammalian and yeast proteasome inhibitor, first extracted from porcine bone marrow. It shows an unique mechanism of inhibition: while binding to the 20S ‘‘core’’ particle via interaction with the a7 subunit, it induces an allosteric change of the whole proteasomal structure, resulting in reduced activity and affecting the binding of the 19S regulator protein (Gaczynska et al., 2003). The PI31 was discovered by the lab of DeMartino (Ma et al., 1992). It is a mammalian protein, competing with PA28 (a and b) to bind the ‘‘core’’ particle (Zaiss et al., 1999). Another cellular negative regulator of the 20S proteasome is the heat shock protein 90 (Hsp90) (Wagner and Margolis, 1995; Conconi and Friguet, 1997; Imai et al., 2003). The proteolytic activity of the proteasome is located in the inside of the proteasome in the main chamber. Three of the seven b-subunits, b1, b2 and b5 are showing proteolytic activity, the others do not (Fig. 3). These three b-subunits can be replaced in some mammalian cells (in de novo synthesized proteasomes) under certain conditions by their c-interferon inducible forms b1i, b2i and b5i resulting in a proteasomal form producing products with a changed fragment length. These proteasomal forms are playing an important role in the immune response (see Chapter 4.3). A fourth form of a subunit exchange is the thymus specific replacement of b5 by b5t (Hirano et al., 2008). Also in eubacteria proteasome related proteins have been found (Chuang et al., 1993): eubacteria contain no 20S proteasomes, despite of the actinobacteria, that provide the only group of eubacteria known to contain proteasomal genes and expressed proteasomes (De Mot, 2007). Normally in eubacteria HslVU (also termed as ClpQY), an ATP-dependent protease (Rohrwild et al., 1997; Song et al., 2003), is expressed, a hexameric structure showing structural similarities to the b-subunits of the 20S proteasome. HslVU is a dimer consisting of the proteasome-related peptidase HsllV and HslU, an ATPase (Li et al., 2008). Proteasomal b-subunits providing an own class of proteases, not showing any relations to other kinds of proteases and a high degree of evolutionary relationship, suggesting a common ancestor (Lowe et al., 1995). Resulting from the shape of the holoproteasome and subunit arrangement of the proteasome found in Thermoplasma acidophilum, the subunits are divided in two different groups, the a- and the b-subunits (Zwickl et al., 1992). The proteasome is not the only cellular proteolytic
195
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
5 nm α1 α6
α5
α7
β1
β2
α4
α1
α2
β7
α3 α7
α6
α7
α1
α3
α5
α2 β2
β4
β3
β6 α6
β α
α4
β5 β4
β7
β6 β3
α3
α7
α6
β2 α2
β1 α1
back
α5 α4
β2
α5
α5
β
β4 β5
α6
α4
front
α1 α2
β3 β6
top
α
α3
α2
β7
β5 α4
α3 β3
β6 α7
perspective
α4 β4
β5 α6
α5 β5
β4 α5
α6 β6
β3 α4
α7 β7
β2 α3
β1 β1
α2
α1
unrolled
Fig. 3. Structure of the mammalian 20S ‘‘core’’ proteasome. The top row of images shows different perspectives of a simplified model structure of the subunit arrangement of the 20S ‘‘core’’ proteasome. The a-rings are dark, the b-rings light grey. The whole proteasome has a diameter of about 100 Å and is 160 Å in height (Borissenko and Groll, 2007). The bottom row shows the same model in a perspective view and the image termed ‘‘unrolled’’ shows a planar presentation of all subunits (while the outside of the proteasome points to the viewer). The three b-subunits of each half-proteasome (b1, b2 and b5) containing proteolytic activity are represented by blue circles. (For interpretation of the references in colour in this figure legend, the reader is referred to the web version of this article.)
Table 1 Names and molecular masses of the ‘‘core’’ proteasomal subunits in yeast and human. The systematic names of the subunits are defined according the position of the subunit in the mature proteasome (a- or b-ring). The molecular masses represent the masses after post-transcriptional subunit processing, i.e. the mass of the subunit present in the final proteasome. The systematic names with an ‘‘i’’ refers to the optional expressed subunits induced by c-interferon. Molecular masses are taken from Coux et al. (1996). Systematic
S. cerevisiae
Homo sapiens
Mass (kDa)
Literature
a1 a2 a3 a4 a5 a6 a7
C7/Prs2 Y7 Y13 Pre6 Pup2 Pre5 C1/ Prs1 Pre3 – Pup1 – Pup3 C11/Pre1 Pre2 – C5/Prs3 Pre4
HsPROS27/HsIota HsC3 HsC9 HsC6/XAPC7 HsZeta HsC2/HsPROS30 HsC8 HsDelta/Y Lmp2 Z Mecl1 HsC10-II HsC7-I X/MB1 Lmp7 HsC5 HsN3/HsBPROS26
27.4 25.9 29.5 27.9 26.4 30.2 28.4 25.3 (21.9) 23.2 (20.9) 30.0 (24.5) 28.9 (23.8) 22.9 22.8 nd (22.4) 30.4 (21.2) 26.5 (23.3) 29.2 (24.4)
Petit et al. (1997) Tokunaga et al. (1990) and Du et al. (2000) Feist et al. (1996) and Castano et al. (1996) Dong et al. (2004) and Mukherjee et al. (2005) Mayau et al. (1998), Van et al. (1999) and Jorgensen and Hendil (1999) Kania et al. (1996, 2005) Bose et al. (2004), Gerards et al. (1997), Gerards et al. (1998) and Shu et al. (2003) Chung et al. (2000), Lequeu et al. (2005), Madding et al. (2007) Dissemond et al. (2003), Wang et al. (2006), Mishto et al. (2006) and Krause et al. (2006) Schweisguth (1999), Ramos et al. (2004) and Baldisserotto et al. (2007) Scheffler et al. (2008) and Eleuteri et al. (1997) Nishimura et al. (1993) and Nothwang et al. (1994) Nothwang et al. (1994) and Reidlinger et al. (1997) Reidlinger et al. (1997) Scheffler et al. (2008), Egerer et al. (2006), Caudill et al. (2006) and Heink et al., 2006 Trachtulec et al. (1997) and Rodriguez-Vilarino et al. (2000) Nothwang et al. (1994), Kopp et al. (1995) and Thomson and Rivett (1996)
b1 b1i b2 b2i b3 b4 b5 b5i b6 b7
machinery, including also the lysosomal system (de, 2005, 1983; Kurz et al., 2008) with its variety of cathepsins (Turk and Guncar, 2003; Turk et al., 2001; Roberts, 2005) and the cytoskeleton localized calpains, but the most important concerning
196
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
α forechamber
β main chamber
β forechamber
α
Fig. 4. Inner structure of the mammalian 20S proteasome. The left image shows the structure of a mammalian 20S proteasome (bovine), the right image the inner structure after removing of several a- and b-subunits. The inner subdivision in two forechambers and one main chamber resulting from the interfaces of a- and b-rings is clearly visible. Both images are recalculated from X-ray crystallographic data (available at RCSB Protein Data Bank, www.rcsb.org), original data at a resolution of 2.75 Å from Unno et al. (2002).
the recognition and degradation of proteins. It was stated that between 70% (Rock et al., 1994) and 90% (Lee and Goldberg, 1998) of misfolded/damaged/not longer needed proteins are degraded via the proteasomal system. Regulatory degradation of proteins is mostly ATP-dependent. In this ATP-requiring proteolytic process ‘‘doomed’’ proteins have to be labeled with a chain of ubiquitin molecules (polyubiquitinylation), whereas in the ATP-independent degradation pathway of damaged and thus unfolded proteins no ubiquitination is required. In mammalian cells the proteasomal protein can be about 1% of the whole cellular protein pool (in liver and kidney cells) (Peters, 1994) and it can be found in the cytosolic and nuclear compartment, bound to the endoplasmic reticulum or associated to the cytoskeleton (Scherrer and Bey, 1994). The mammalian 20S ‘‘core’’ proteasome was first discovered and isolated from human erythrocytes by Harris in 1968 and then called ‘‘cylindrin’’ (Harris, 1969, 1968) pointing to the structure of the protein complex. It has been also termed ‘‘macroxyproteinase’’ (Pacifici et al., 1989; Pacifici and Davies, 1990; Salo et al., 1990), ‘‘multicatalytic proteinase complex’’ (Rivett, 1989; Orlowski, 1990; Djaballah and Rivett, 1991) and ‘‘prosome’’ (Scherrer and Bey, 1994; Nothwang et al., 1992), but the name ‘‘proteasome’’ is the most widespread used today. 2.1. The proteasomal alpha-subunits (a-subunits) In eukaryotic cells the recognition and substrate access to the proteolytic inner chamber is only possible after contact with the a-rings followed by conformational change that ‘‘unlocks’’ the gate formed by the N-terminal ends of the a2, a3 and a4 subunits (Groll et al., 2000). The N-terminal ends of these subunits are turned to the axis of the proteasome, blocking the channel into the proteolytic chamber build by the b-rings (Fig. 5). Experiments incubating isolated proteasomes with low doses of sodium dodecylsulfate (SDS), inducing a slight unfolding of native proteasome structure, resulted in a significant increase of proteolytic activity of the treated proteasomes due an SDS-induced opening of the gate to the proteolytic chamber (Coux et al., 1996). Analogous effects where found using repeated freeze–thaw cycles or conditions of low ionic strength. These results were pointing to an involvement of structural changes of the gating proteasome subunits. Another possibility is the attachment of a regulatory subunits (like 11S or 19S) to the proteasomal a-rings, resulting in an increased activity (up to 10-fold) (Adams et al., 1998; Glickman et al., 1998) and a changed substrate specificity (see Chapter 3), caused by a maximal opening of the gated substrate channel by conformational changes of the blocking N-terminal ends. Furthermore, the exposure of hydrophobic structures of misfolded or by (oxidative) damage unfolded proteins (usually these amino acids are covered in the inside of the folded globular protein) is assumed to ‘‘activate’’ the proteasome by inducing of conformational changes in the gating structures of the a-rings (Liu et al., 2003). The completely opened gate shows a diameter of about 13 Å (determined in an archae proteasome), enough to allow the access of an unfolded substrate protein, i.e. a single chain of amino acids. Each of the N-terminal ends of the a-subunits seems to have an unique three dimensional conformation, necessary for the gating process. Some of these structures are highly conserved in eukaryotic cells. One motif found in every single a-subunit (both in archaea and eukaryotes) is the so-called YDR-motif (Tyr8-Asp9-Arg10), maybe working as a joint, bending away the blocking structures from the gated channel and is found in archaea, too (Groll et al., 2000). A major role seems to be played by the a3 subunit (Fig. 5), since a3DN mutants of yeast, missing the last nine amino-residues (GSRRYDSRT) found in the wild type, show a constant high proteolytic activity and are not longer susceptible to SDS induced activation (Groll et al., 2000), while characteristics of degradation or binding to regulatory subunits are not or very less affected. The a7DN mutant showed almost no increase in proteasomal activity, while the a3a7DN mutant was significantly more efficient in casein degradation than either of the single deletions (Bajorek et al., 2003). The YDR motif and especially in a3 seem to play an important role in the stabilization of the gate, showing allosteric effects of a2, a3 and a4. Cryo-electron microscopic investigations of the archae proteasome (Rabl et al., 2008) (from Thermoplasma acidophilum), containing a-rings
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
197
Fig. 5. Structure of the a-rings of the mammalian 20S ‘‘core’’ proteasome. The left image shows a ‘‘tube view’’ of the single amino acid chains of the seven proteasomal alpha-subunits (a1–a7) as seen along the axis of the particle. The right image shows a three dimensional reconstruction of the alpha-ring from the electron density (from X-ray crystallography. Both images are recalculated from X-ray crystallographic data (available at RCSB Protein Data Bank, www.rcsb.org)), original data from bovine liver proteasome at 2.75 Å resolution (Unno et al., 2002). The different subunits are color-coded. The a3-subunit, essential for the gating-mechanism of the proteasome, is yellow, the N-terminal end of it in the centre of the a-ring is encircled. The arrow on the top of the stylized proteasome is indicating the direction of the view. (For interpretation of the references in colour in this figure legend, the reader is referred to the web version of this article.)
build of seven identical subunits, showed the ability of some oligopeptides to induce an opening of the gate by conformational changes of the N-terminal ends, involving the last 13 amino acids, and a slight turn of each a-subunit (about 4°, as found by the attachment of PAN in it’s ATP-bound conformation). The used oligopeptides (seven or more residues) were the different C-termini of PAN’s subunits (Smith et al., 2007) binding to gaps between the single a-subunits. Such sequences are known as the HbXY motif. The diameter of the closed gate was about 9 Å, significantly smaller than the channel to the inner proteolytic chamber of the proteasome (23 Å). The activated open gate showed a diameter of 13–20 Å. Studies using the PA26 binding revealed a similar mechanism of opening the gate-structures by its C-termini and a so-called ‘‘activation loop’’ that induces both – a movement of the concerning N-terminal structures and a slight turning of each a-subunit (Rabl et al., 2008). PAN in contrast needed only its C-terminal structures, resulting a similar ‘‘open’’-state of the core particle induced by to different kinds of regulator binding. The ‘‘closed’’- and the ‘‘opened gate’’-structures of the archae proteasome after binding of an oligopeptide are depicted in Fig. 6. In conclusion it should be mentioned that the 20S ‘‘core’’ proteasome is usually found in its ‘‘inactive’’ form but can be activated by regulator proteins, unfolded proteins or proteasomal substrates. It has also to be remarked that one has to discriminate between the proteasomal peptidase (degradation of small amino acid fragments) and protease activities (degradation of whole unfolded proteins): whereas the protease activity depends on the opening status of the gate, the peptidase activity shows almost no effects of the degradation rate of oligopeptides in dependence of the gate status, suggesting that the gating alpha rings have only little interactions with small peptides and play only a key-role in the degradation of larger peptides or proteins. 2.2. The proteasomal beta-subunits (b-subunits) The main task of the b-rings in the assembled proteasome is the proteolytic degradation of polypeptides/proteins. The archaea proteasome is composed of seven identical b-subunits, resulting in seven active centers. Unno et al. (2002) proposed a novel N-terminal nucleophile hydrolase activity of the b7-subunit located at Thr8 of the protein after determining the X-ray crystal structure of the bovine proteasome. In the yeast/mammalian proteasome there are only three b-subunits that show proteolytic activity: b1, b2 and b5. The proteolytic centers are localized in the inside of the proteasome in the main chamber (Fig. 7). Each of the three subunits shows a different proteolytic activity: b1 reflects a peptidyl-glutamyl-peptide-hydrolysing-activity (caspase-like, cleavage after acidic amino acids, also termed as ‘‘post-glutamyl-peptide hydrolytic’’ activity) (Loidl et al., 1999), a trypsin-like (after basic amino acids) activity at b2 and a chymotrypsin-like activity (after neutral amino acids) at the b5-subunit (Groll and Huber, 2004). In each case the catalytic centre is build by the N-terminal threonine residue (Thr1) (Figs. 8 and 9). Proteases that show similar specificity are serine-proteases, like chymotrypsin (Klegeris and McGeer, 2005) and trypsin (Ofosu et al., 1998). The oligopeptidic products of proteasomal degradation showing length
198
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
Fig. 6. The a-rings of the Thermoplasma acidophilum 20S proteasome. This figure shows the proteasomal gate in the closed (upper left image) and opened state (upper right image) induced by short oligopeptides. The image beyond shows both structures, revealing a slight turn of the subunits (about 4°) and a movement of the N-terminal structure of each subunit, turning away from the central annulus, leading to a widening of the opening. The dashed line shows the effective radius of the opening in ‘closed’, the continuous line in ‘opened’ state. The protein backbone is recalculated from X-ray crystallographic data (available at RCSB Protein Data Bank, www.rcsb.org); original data from Lowe et al. (1995).
Fig. 7. Arrangement of the proteolytic b-subunits and active centers in the yeast 20S proteasome. The left image shows the assembled yeast 20S proteasome, with color-coded b-subunits, while the a-rings are dark grey. The right image shows the assembled proteasome after removing of the proteolytic inactive bsubunits. b1 is white, b2 purple and b5 light blue. The position of the N-terminal Thr1 residue is indicated by a yellow-glowing red dot. b20 is the active subunit b2 of the other (‘lower’) b-ring. The molecular structure is recalculated from X-ray crystallographic data (available at RCSB Protein Data Bank, www.rcsb.org); original data from Groll et al. (1997). (For interpretation of the references in colour in this figure legend, the reader is referred to the web version of this article.)
between 2 and 35 amino acids (Luciani et al., 2005), with three maxima at 2–3, 8–10 and 20–30 amino acids (Kohler et al., 2001), the average length of the products is 8–12 amino acids.
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
199
Fig. 8. Proteolytic subunits of the mammalian 20S proteasome. This figure shows the arrangement of the b-subunits found in the 20S ‘‘core’’ proteasome by ‘‘tube view’’ and the molecular surface represented by electron density. The proteolytic active b-subunits (b1, b2 and b5) are light blue, the inactive light grey. The proteolytic centers are indicated by a white or red ring, respectively. In every case it is build from the N-terminal threonin (Thr1), pointing to the inside of the main chamber. The circular images show a magnification of the area around the Thr1. The protein backbone is recalculated from X-ray crystallographic data (available at RCSB Protein Data Bank, www.rcsb.org); original data from Lowe et al. (1995). The stylized proteasome top right indicates the view in the arrow direction on the second b-ring (black rectangle) is presented, both the top a- and b-rings and the bottom a-ring are removed in this figure. (For interpretation of the references in colour in this figure legend, the reader is referred to the web version of this article.)
Fig. 9. The active center at N-terminal Thr1 of the yeast b1-subunit. The left image displays the whole b1-subunit, representing the amino acid backbone as a tube scheme. The high conserved amino acids that are found in the yeast and mammalian 20S proteasome are color-coded: Thr1 is red, Asp17, Lys33, Ser129, Ser166 and Ser169 are blue. The right image shows the Thr1 in its close environment in detail. The molecular structures in this image show the positions of the highly conserved amino acids surrounding the proteolytic active N-terminus; nitrogen is blue, oxygen red, hydrogen is not displayed, while the carbon-structure of the amino acids are dyed green. The side chain of the residue asparagine 17 (Asp17) is not displayed due to steric causes. The protein backbone is recalculated from X-ray crystallographic data (available at RCSB Protein Data Bank, www.rcsb.org); original data from Lowe et al. (1995). (For interpretation of the references in colour in this figure legend, the reader is referred to the web version of this article.)
The residues Thr1, Asp17 and Lys33 have turned out to be important for the proteolytic mechanism (Fig. 9). The amino acids Ser129/166/169, localized around Thr1 seem to be important for the stabilization of the three dimensional structure of the whole proteolytic centre. Allosteric effects between the single catalytic centers are still under discussion. Some kinetic
200
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
findings point into this direction (Orlowski and Wilk, 2000), but since most of the known natural proteasome inhibitors preferentially bind to the catalytic centre of b5 subunit, resulting in the reduction of the chymotrypsin-like activity only, without affecting the other two activities, suggested allosteric effects can perhaps be neglected. At least partially in contradiction to these results is Kisselevs ‘‘bite and chew model’’ of proteasomal degradation (Kisselev et al., 1999): this model proposes allosteric relations of the active subunits. Experiments with substrates for the chymotrypsin-like activity (b5-subunit), in this case the Suc-LLVY-MNA or Suc-FLF-MNA, strongly increased the caspase-like cleaving activity (b1-subunit). Inhibiting substrates for one activity also affected other activities in an indirect way. Kisselev concluded a cyclic mechanism of peptide-chain degradation; where the chymotrypsin-like activity initially ‘‘cleaves’’ the substrate protein and triggers further cleavage (‘‘chewing’’) by b1 (peptidyl-glutamyl-peptide-hydrolysing-activity) while b5 is temporarily inhibited. If no further b1-mediated substrate degradation is possible, then b5 is ‘‘reactivated’’ and the cycle starts off again. According to this model the allosteric effects of the single active subunits are essential for substrate degradation (Kisselev et al., 1999). Interestingly, experiments with knock out mutants of the catalytic centers in yeast showed that the existing active proteasomal subunits are of various importances for cellular survival (cell division). The order of importance (respective proteolytic capacity) is b5 b2 > b1, since b5/b1 and b2/b1 mutants are viable, b5/b2 mutants however are not (Heinemeyer et al., 1997; Jager et al., 1999).
2.3. Intracellular assembly of the 20S proteasome The intracellular assembly of the proteasomal subunits to the functional 20S proteasomal holoprotein is a very complex and not yet fully understood process. The present model involves the interaction of the proteasomal a- and b-subunits and different proteasome-specific chaperones, called ‘‘proteasome assembling chaperones’’ (PAC). Until now four different forms of the mammalian PACs are known (PAC1–PAC4) (Yashiroda et al., 2008; Kusmierczyk et al., 2008; Le et al., 2007); the mammalian PAC1 is sometimes termed in yeast as Pba1 (Hirano et al., 2005; Li et al., 2007) or POC1 (Le et al., 2007), PAC2 as Pba2 or POC2 (yeast), PAC3 as Pba3, Dmp2 (Yashiroda et al., 2008) or POC3 (yeast), and PAC4 as Pba4, Dmp1, respectively POC4 (yeast). To be functional, the mammalian PAC1 and PAC2 form heterodimers (PAC1–PAC2) that are corresponding to the yeast Pba1–Pba2, respectively POC1–POC2 heterodimers and show the same function in proteasome assembly. PAC3 and PAC4 build a heterodimer, too, and the corresponding yeast forms of Dmp1–Dmp2/Pba3–Pba4 (Kusmierczyk et al., 2008) heterodimers have been identified as well. The PAC1–PAC2 dimer and its yeast equivalents first assemble one alpha-ring. After this the PAC3–PAC4 heterodimer is bound, starting the stepwise incorporation of the b-subunits, ending up in the formation of a single half proteasome. Two of these half proteasomes have to be associated in order to form the holoproteasome, and thus the functional 20S ‘‘core’’ proteasome particle. Furthermore, the proteasome maturation factor Ump1 (Li et al., 2007; McIntyre et al., 2007; Chen et al., 2006) (also termed as ‘‘POMP’’ (Fricke et al., 2007; Chondrogianni and Gonos, 2007; Hoefer et al., 2006)) for proteasome maturation protein or ‘‘proteassemblin’’ (Jayarapu and Griffin, 2004; Griffin et al., 2000), that was found in yeast (the human form is called ‘‘hUmp1’’), takes part in the assembly of two half proteasomes, build of one a- and one b-ring, to the holoproteasome. After this step, the proteolytic b-subunits are activated by autocatalytic degradation of their N-terminal ends and the degradation of both Ump1 molecules in the proteolytic chamber. Additionally, the two attached PAC1–PAC2 dimers are degraded by the now functional 20S proteasome. In more detail (see also Fig. 10): after the dimerization of PAC1 and PAC2, the resulting heterodimer binds to the subunits a5 and a7, after this the remaining five a-subunits associate stepwise driven by their intermolecular mutual interactions (Jayarapu and Griffin, 2004) and supplemented furthermore by the orchestrating PAC3–PAC4. According to their interactions (Jayarapu and Griffin, 2004) the a-subunits should bind to the PAC1–PAC2-a5–a7-cluster in the following order: a6 and a1 (both bind on the a7 side), followed by a2 (binds to a1), a3 (binds a2) and a4 (binds both to a3 and a5). The formation of the a-ring is assisted by the binding of a PAC3–PAC4 dimer, that attaches to the a2-subunit, incorporating the first two b-subunits, b2 and b3. In this stage the unprocessed forms of some b-subunits are assembled, that still contain a later to be removed prosequence. As usual b-subunits containing a prosequence (as b1, b2, b5, b6 and b7) (Fricke et al., 2007) are called prob1, prob2 and so on. The binding of b3 is accompanied by the detachment of PAC3–PAC4 from the cluster and the binding of Ump1. The PAC3–PAC4 dimer can be recycled for the next proteasome to be assembled. Ump1 integrates the b-subunits in the following order b4, b5, b6, b7 and b1 (in fact, b1 can also be incorporated already after b4), as clarified by experiments using siRNA, investigating the formed intermediates of proteasome assembly, dependent from the suppressed subunits (Hirano et al., 2008). One intermediate in this step is the 13S intermediate, consisting of a whole aring, b2, b3, b4, b6 and Ump1 (Schmidt et al., 1997). The b-subunits attach to Ump1 guided by their prosequences. After this step, the intermediate contains one complete a- and one complete b-ring as well as an Ump1 and PAC1–PAC2. This intermediate is called the 16S intermediate. Two of these 16S intermediates dimerize while binding to a Hsc73 heat shock protein each, resulting in a large particle containing every proteasomal subunit twice. In this dimerization the C-terminal end of b7 seems to play a major role (Marques et al., 2007): the last integrated b-subunits seems to trigger the dimerization of 16S, thus being possibly a rate limiting step in proteasome formation, since overexpression of prob7 resulted in a massive increase of 16S dimerization. Hsc73 detaches and by rearrangement of the molecular structures of Ump1 and the propeptides of the proteolytic active subunits, inducing the autocatalytic processing of that propeptides, that sets the N-terminal threonine residues Thr1 on b1, b2 and b5 free. The proteolytic centres are activated and degrade their first
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
201
Fig. 10. Assembly of the mammalian 20S proteasome. This image depicts the currently discussed model of mammalian 20S proteasome assembly according to Hirano et al. (2008). The assembly starts with the dimerization of PAC1 and PAC2, binding the subunits a7 and a5 (the resulting structure as Step A), followed by the incorporation of all remaining a-subunits into a ring structure and the binding of the PAC3–PAC4 dimer (Step B). PAC3–PAC4 binds the b2subunit and Ump1 (Step C). The next incorporated subunit is b3, accompanied by the detachment of PAC3–PAC4 (Step D). The remaining b-subunits are assembled in the sequence b4, b5, b6, b7 and b1. The result is the so-called 16S intermediate, build of a complete a- and b-ring as well as Ump1 (Step E). Two of this 16S intermediated dimerize while binding the heat shock protein chaperone Hsc73 (Step F). Hsc73 detaches and the proteolytic active b-subunits b1/ 2/5 are activated by an intramolecular structural shift in the arrangement of Ump1 and the b-subunit propeptides (Step G), resulting in an autocatalytic processing of the propeptides. This activates the N-terminal Thr1 residues of the active centers. In the following the two molecules of Ump1 in the main chamber of the proteasome are degraded (Step H), followed by the degradation of the two still attached PAC1–PAC2 dimers (Step I). After the release of all degradation products the newly synthesized 20S proteasome is fully functional (Step J).
substrate–Ump1. After this, the still attached two PAC1–PAC2 dimers are degraded, too. The single steps of this process are visualized in Fig. 10. In recent investigations using intracellular immunolocalization it turned out, that during the mammalian proteasome maturation Ump1 is attached very stable to the outer membrane of the endoplasmic reticulum (ER) (Fricke et al., 2007), suggesting that the complete assembled a-ring docks to Ump1 and is not released before the attachment of all b-subunits. Ump1 was almost completely colocalized with the ER membrane and the intermediates of proteasome maturation up to an amount of 75–88%. Dimerization of the 16S intermediate to the whole proteasome seems to occur unbound in the cytosol (Schmidtke et al., 1997).
2.4. Modeling of the 20S proteasomal proteolysis Several attempts have been made to produce mathematical models that are able to describe the cleaving properties of the 20S proteasome, with and without regulatory proteins (see Chapter 3). Prediction of fragment length or cleaving sites in a given oligopeptide may be among other things interesting for the prediction of antigens presented by the immune system. There are two main strategies in the prediction of oligopeptide fragments of proteasomal degradation. One strategy concentrates on the prediction of the resulting fragments (Holzhutter et al., 1999); this attempt becomes imprecise, if the resulting fragments overlap. Another one concentrates on the potential cleaving sites, using the probability of a certain degradation product to occur (Ginodi et al., 2008). Usually cleaving sites are determined by the locus between two amino acids that is to be cleft with respect to one or two neighboring amino acids on each side. Other models try to imitate the mechanism of degradation itself, using kinetic rates in order to determine the velocity of degradation of a certain substrate protein. Though in every case the results have to be experimentally validated, and the model itself is based on so-called learning sets, resulting from analysis of the degradation products of a certain peptide.
202
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
3. Regulation of the 20S proteasome The highly effective protein degradation machinery of the proteasome must be strictly regulated in order to prevent the uncontrolled degradation of cellular proteins. Therefore, several regulators evolved in order to control the proteasomal recognition and degradation of substrate proteins. The gating of the 20S proteasome requires a regulated entrance of substrates into the proteasome and, there, a controlled opening of the a-rings. Hydrophobic proteins are perhaps able to open the asubunit gate due to their hydrophobic stretches (see degradation of oxidized proteins), whereas all undamaged, normally folded proteins require some kind of targeting. The attachment of several regulatory complexes specializes the proteasome for such tasks. One of the most important of these regulators is the already mentioned 19S regulator complex acing in context with another cellular machinery, the ubiquitination system. 3.1. The 19S regulator The 19S regulator (also termed ‘‘PA 700’’ or ‘‘proteasome activator 700 kDa’’) (DeMartino et al., 1994; Lam et al., 1997; Strickland et al., 2000; Liu et al., 2002) is build of a ring shaped base and a lid-structure, that regulates the entrance of substrates to the attached 20S proteasome. The base structure contains at least 10 different subunits (Rpt1–Rpt6, Rpn1, Rpn2, Rpn10 and Rpn13). The ‘‘lid’’ structure is formed by nine subunits, too: Rpn3, Rpn5–Rpn9, Rpn11, Rpn12 and Rpn15 (also termed as DSS1 or SHFM1 in humans, and SEM1 in yeast). The Rpt-subunits show an ATPase activity, the Rpn-subunits not. The Rpn11-particle in the lid contains a Zn2+-dependent proteolytic centre that catalyzes the degradation of poly-ubiquitin chains, releasing single ubiquitin molecules for re-use. Rpt2 (in humans termed also S4 or p56; in yeast as YTA5 or mts2), Rpt3 (human form termed as S6, Tbp7 or P48; the yeast one as YTA2) and Rpt5 (S60 or Tbp1 in humans, YTA1 in yeast) are involved in gate opening of the 20S ‘‘core’’ proteasome (Tanaka, 2009). Rpn10 (S5a or Mbp1 in humans, SUN1, MCB1 or pus1 in yeast) and Rpn13 (ADRM1 in humans, DAQ1 in yeast) are ubiquitin-receptors (Tanaka, 2009). The central role of ubiquitin, a small protein that enables the specific labeling of intracellular proteins in order to regulate their amount and live-span will be described below. One 19S regulator attaches to each of the ends of the 20S proteasome forming a large particle, the so-called 26S proteasome with an overall mass of over 2 MDa (Glickman and Ciechanover, 2002). In a similar mechanism as shown by other proteasome regulator particles (see below) the 19S regulator enables enhanced substrate access to the inner proteolytic chamber by changing the structure of the gating a-rings. In yeast the base-ring Rpt2 ATPase has been shown to be involved (Kohler et al., 2001) in this process. Fig. 11 displays a hypothetical structure of the 19S regulator protein, since only a few interactions of the single subunits are known (Chen et al., 2008) and no X-ray crystallography data are available yet (Tanaka, 2009).
Fig. 11. The 19S regulator particle. These two images show a simplified model structure of the 19S regulator particle. The subunit arrangement is shown from two different perspectives. The a-rings of the 20S proteasome are binding from below (not shown).
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
203
3.2. The ubiquitin–proteasome system (UPS) The UPS is a machinery, virtually involved in any cellular process that depends on the degradation, regulation or life-span determination of intracellular proteins. In contrast to the 20S proteasome, the substrates of the 26S proteasome have to be polyubiquitinated to be recognized. Only a few proteins have been found that can be degraded by the 26S proteasome without polyubiquitination. One of these exceptions that can be degraded by the 26S proteasome without ubiquitination is the murine ornithine decarboxylase (mODC) (Murakami et al., 1992), showing a sequence of 37 C-terminal amino acids, that seem to be a needed and sufficient structure for substrate recognition by 26S proteasome, binding to the same sites of the 19S regulator that are responsible for ubiquitin recognition (Hoyt et al., 2005; Zhang et al., 2003). mODC together with its inhibitor antizyme (AZ) is often used to determine in vitro 26S proteasomal activity, due to the difficulty in preparation of large amount of polyubiquitinated substrate proteins (Porat et al., 2008; Coffino, 2001; Hayashi, 1995; Tokunaga et al., 1994; Murakami et al., 1993). The 26S proteasome is able to degrade natively folded intact proteins in an ATP-dependent fashion. As mentioned, most of the 26S substrates have to be polyubiquitinated. In this process a very complex system, containing four different kinds of enzymes (E1–E4) is involved. The first step in substrate poly-ubiquitination is the ATP-dependent activation of ubiquitin (Ub) by the E1 enzyme. Ub is a small protein containing 76 amino acids (Fig. 12), highly conserved in higher eukaryotic cells (Wilkinson et al., 1980). E1 transfers Ub to a lysine-residue of E2, and both E2 and the doomed substrate are bound by an E3 enzyme, inducing the transfer of Ub from E2 to a lysine amino groups of the substrate. After this, E2 and E3 are released. The cyclic transfer of more Ub-molecules to the first Ub attached to the substrate is performed by these enzymes and by another enzyme – the E4. The substrate specificity is provided by the E3 class of enzymes, containing thousands of Ub-ligases, each one only specific for a limited number of substrate proteins. In the following we will describe the different steps of ubiquitination in more detail (see also Fig. 13): (i) Ubiquitin activation by E1: As mentioned the initial step is an ATP-dependent (ATP ? AMP + PPi) activation of Ub by one of the very few E1 ‘‘ubiquitin-activating enzymes’’. Until now, only eight (Groettrup et al., 2008) of these E1enzymes are known: UBE1, UBE1L (Dou et al., 2005; Krug et al., 2005; Kim et al., 2006; Pelzer et al., 2007; Zheng et al., 2008; Lai et al., 2009), AOS1 (Johnson et al., 1997) (also termed as SAE1), UBA2 (SAE2) (Tatham et al., 2001), APP-BP1 (Park et al., 2008), UBA3 (Huang and Schulman, 2005), UBA4 (Schmitz et al., 2008) (MOCS3), UBA5 (Liu et al., 2009) (UBE1DC1), UBA6 (Groettrup et al., 2008) (also termed as UBE1L2 or E1–L2) and ATG7 (Mizushima et al., 1998). In this reaction the C-terminal end of Ub (glycin76) is attached through a thioester bond to a cysteine residue in the active centre of E1 (Sun, 2003). Different forms of the E1 enzyme are present in the cytosol and the nucleus (Grenfell et al., 1994).
Fig. 12. The ubiquitin molecule. Here the human ubiquitin (hUb) is shown reconstructed from X-ray crystallographic data at a resolution of 1.8 Å (Schlesinger and Goldstein, 1975). The three dimensional structure of that molecule was calculated from X-ray crystallographic data (available at RCSB Protein Data Bank, www.rcsb.org) using the electron density to limit the effective surface. In the background the surface according to the detected electron density is displayed (grey), in the front the secondary structure of the molecule (a-helices in blue, b-sheets in red). The lysine residues enabling the attachment to other Ub-molecules are marked in yellow (Lys6/27/29/48/63). While ubiquitination the C-terminal Gly76 is bound to an internal lysine residue of the target protein destined for 26S proteasomal degradation. Polyubiquitination by the UPS is usually realized by linking the C-terminal Gly76 and the Lys48 of the previous Ub until a chain of at least four Ub-molecules is attached to the target protein, representing the minimum target signal for 26S proteasomal degradation (Thrower et al., 2000). (For interpretation of the references in colour in this figure legend, the reader is referred to the web version of this article.)
204
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
Ub
Cysteine-residue
E1 ATP AMP + PPi
Ub
E1 E2 E1 Ub
E2 RING E3
HECT E3
Ub
Ub Ub
E2
E2 Ub
Ub
E3 Substrate
Lysine-residue
E3 Substrate
Lysine-residue
Ub
Substrate
Fig. 13. Substrate-ubiquitination in mammalian cells. This figure depicts a simplified scheme of mammalian substrate ubiquitination according to Hochstrasser (2006). After an ATP-dependent activation of an ubiquitin-molecule (Ub, yellow) by E1 (blue hexagon) and binding of Ub to a cysteine residue (purple dot) of E1, Ub is transferred to E2 (green hexagon), binding to a cysteine residue again (purple dot on E2), while E1 is released. Substrate specificity is delivered from a huge amount of E3 (yellow) enzymes, specific only for a very limited number of substrate (blue) proteins. After binding of E2-Ub to the E3-substrate complex, the Ub is transferred either directly from E2 to a lysine-residue on the substrate protein (RING-bearing E3 enzyme) or from E2 to a cysteine-residue on E3 (purple dot on E3) and then to a lysine-residue in the substrate protein (HECT-domain E3 enzyme). After this, both E2 and E3, are released. (For interpretation of the references in colour in this figure legend, the reader is referred to the web version of this article.)
(ii) Ubiquitin-conjugation and E2 enzymes: In a kind of intermediate step the E1 enzyme transfers the activated Ub to an ‘ubiquitin-conjugating enzyme’ (=E2). Again the C-terminal glycine76 of Ub is attached to a cysteine residue of E2 via a thioester bond. E2 enzymes represent a group of enzymes. It is estimated that several dozens of E2 enzymes exist, each responsible for the interaction with a special set of E3 enzymes or substrates. (iii) Ubiquitin transfer to the substrate by E3 enzymes: Ubiquitination of substrate proteins is realized by a complex containing an Ub-loaded E2, a substrate specific E3 and the substrate protein itself bound to E3. After this substrate–ubiquitination can take place in two ways: Ub is transferred either from E2 directly to the substrate protein (RING E3) or from E2 to E3 and from E3 to the substrate protein (HECT E3) (Hochstrasser, 2006). Most E3s are bearing the RING motif (Hochstrasser, 2006) (Fig. 13). A complex of E1–E2–E3 can be excluded, since the binding sites of E1 and E3 are overlapping at E2 (Eletr et al., 2005). A common characteristic of all E2 enzymes is a highly conserved asparagine residue close to the cysteine in the active-site (Wu et al., 2003). This asparagine residue might be required for the catalyzation of the E2/E3 isopeptide formation by stabilization of the oxyanion intermediate resulting from the nucleophilic lysine attack from ubiquitin to the substrate protein that is to be ubiquitinated (Wu et al., 2003). On the other hand E2 enzymes might form dimers, especially if one of the E2 is Ub-loaded (Varelas et al., 2003). This might be necessary for their functionality. Each E2 can activate a special set of E3 enzymes. E3 represents the largest group of enzymes involved in the UPS, the socalled ‘‘ubiquitin ligases’’. The exact number of E3s in a mammalian cell remains unknown until now, but there are at least several hundred assumed to exist (Wong et al., 2003). According to the N-end rule, the intracellular half-life of a protein depends from its N-terminal residue. This ‘‘rule’’ is conserved from bacteria (E. coli) (Tobias et al., 1991; Shrader et al., 1993) over yeast (S. cerevisiae) (Bachmair and Varshavsky, 1989; Bachmair et al., 1986) and plants (Arabidopsis thaliana) (Zwirn et al., 1997; Potuschak et al., 1998) to mammals (Gonda et al., 1989; Varshavsky, 1996; Mogk et al., 2007) that may show different proteolytic systems, but share a common manner
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
205
of substrate recognition. Eukaryotic cells realize that recognition by so called N-recognins, a special sub-class of E3-ubiquitin-ligases, that recognize the N-terminal amino acids and attach ubiquitin-residues, in order to label it as a substrate for 26S-mediated proteolysis. The bacterial recognizing protein is ClpAP that shows homologies in substrate recognition and binding to the N-recognins. The N-terminal degradation signal is termed as ‘‘N-degron’’ (Tasaki et al., 2005; Rajagopalan et al., 2004; Huang et al., 1998). As already stated the E3 enzymes can be assigned by their structure in two groups: the HECT (homology to E6AP C-terminus) E3s, and the RING (really interesting new gene) finger E3s. The RING finger E3s consist of two subfamilies: the U-Box (Xu et al., 2006; Loscher et al., 2005; Hatakeyama et al., 2004; Hatakeyama and Nakayama, 2003; Ohi et al., 2003) and the PHD (plant homeo-domain). The RING finger E3s bind both the substrate protein and the Ub-loaded E2 and catalyze the transfer of Ub to the substrate, whereas the HECT E3s bind Ub–E2 and the substrate at the same time as well, but transfer the Ub first to E3 and then to the substrate protein; in this case, the Ub is bound again to an E3 highly conserved cysteine residue. Although E3 has usually an orientating function arranging the substrate protein and E2 in a way enabling an Ubtransfer when the reactive Ub–E2–thioester bond comes close to the lysine target on the substrate (Hochstrasser, 2006). Most of these ‘‘orientating’’ E3s subtypes show a RING finger structure. How exactly further Ub-molecules are attached to the first one in order to achieve poly-ubiquitination is not yet clear. Several models explain this elongation by a cyclic repetition of the attaching step, other models involve an ‘‘ubiquitin-chain elongation factor’’ the so-called ‘‘E4 enzymes’’ that attaches in an unspecific way (Hoppe, 2005). These E4 enzymes have been found in yeast, catalyzing the assembly of polyubiquitin-chains to oligoubiquitylated substrates by interplay with E1, E2 and E3. The first discovered E4 enzyme in yeast is Ufd2p, but its mechanism of action is unknown yet. Ufd2p shows the hallmarks of a RING-finger containing Ub-ligase (Tu et al., 2007). In mammalian cells polyubiquitination is supposed to be catalyzed by an E2/E3/UEV-complex, whereas UEV is an ‘‘ubiquitin E2 variant’’ lacking the cysteine residue in their active site (Hoege et al., 2002). The time limiting step in polyubiquitination is the transfer of the first Ub to the substrate protein. Experiments revealed that after the first Ub is attached to a protein, polyubiquitination takes place very quick and much faster than in the nonubiquitylated form of the same protein (Petroski and Deshaies, 2005). Polyubiquitinated substrates are the targets of 26S proteasomal degradation. To be recognized for degradation at least four ubiquitin molecules have to be attached to the substrate protein (Thrower et al., 2000). The recognition of the substrate is mediated by subunits of the 19S regulator, the Rpn10 and the Rpn13 subunits. Furthermore, Rpn1 (Hrd2, or Nas1 in yeast, S2 or Trap2 in human), the Rpt5 (Yta in yeast, S60 or Tbp1 in human) and the Rpn10 (Mcb1 or Sun1 in yeast, S5a in human) subunits are involved, while Rpt5 shows ATPase-activity and Rpn10 does not. Furthermore, the Rpn1 (Hrd2 or Nas1 in yeast, S2 or Trap2 in human) subunit of the base binds to the deubiquitinating enzyme Upb6. Ubp6 is suggested to function as a timer: if a bound substrate protein is deubiquitinated until the attached chain is shorter than four Ub-molecules the affinity of the substrate may decrease until it is released (Kraut et al., 2007); this could be a strategy preventing 26S proteasomal inhibition in futile attempts of degrading an indegradable substrate. Furthermore, a decreasing pool of free cytosolic available ubiquitin leads to an increased expression of Upd6 (Hanna et al., 2007). Another protein bound to the 19S is Hul5 (Kohlmann et al., 2008), an enzyme that elongates unspecifically existing polyubiquitin-chains, fulfilling an E4-like function. Hul5 seems to be counteracting Upb6 (Kraut et al., 2007), but its precise function attached to the 19S regulator is still unclear, since 26S-proteasomal substrates have to be already polyubiquitinated. The Rpt2 (Yta5 in yeast, Rpn1 in human) subunit of the 19S-base-ring regulates both substrate entry and release of degradation products from the 20S ‘‘core’’ particle, while showing ATPase-activity. The Rpn11 (Mpr1 in yeast, Poh1 in human) subunit, a Zn2+ dependent metalloprotease, located in the 19S-lid-structure, provides deubiquitination of the substrate protein in a fully ATP-dependent manner (Amerik and Hochstrasser, 2004). The ubiquitin molecules are released into the cytosol during substrate degradation (it is still unclear if release precedes the substrate degradation or occurs later) in the form of polyubiquitin-chains. These chains are decomposed by deubiquinating enzymes (DUBs) to single ubiquitin molecules (Amerik and Hochstrasser, 2004), in order to prevent competition of these free Ub-chains with polyubiquitinated substrates. As described above the re-use of ubiquitin is both ATP- and E1-dependent. 3.3. The immunoproteasome The 20S immunoproteasome (i20S) is a special inducible form of the 20S proteasome. It is characterized by the replacement of the three constitutive proteolytic active subunits (b1, b2 and b5) by their inducible equivalents (b1i, b2i and b5i). Moreover a special proteasome regulator, the so-called 11S regulator (also termed as PA28, PA26 or REG) is induced by the same cellular signals. One of the most important functions of the i20S proteasome is its role in immune response. It is suggested that the pattern of oligopeptides, produced by proteolytic attack of the immunoproteasome shows a different length distribution than the peptides produced by constitutive 20S proteasome (c20S). The products of the i20S proteasome, displaying an average length of 8–10 amino acids are ‘‘optimized’’ for presentation by the major histocompatibility complex I (MHC-I) on the cell’s surface. In contrast to the typical proteasomal degradation, the immunoproteasome i20S shows a higher yield in producing antigens proper for MHC-I presentation. It is suggested, considering the fact that i20S induction is mostly dependent on the amount of cytokines released in the tissue, that immunoproteasomes produce new self-determinants in surrounding uninfected cells to prevent autoimmune response (Yewdell, 2005). However, it was proposed by Yewdell (2005), that the main
206
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
functions of the immunoproteasome are other than the generations of MHC-I presented antigens and further investigations are needed. As already mentioned, the three proteolytic active b-subunits of the ‘‘constitutive’’ 20S proteasome (c20S) can be replaced by interferon c (IFN-c)-inducible forms (Ortiz-Navarrete et al., 1991; Aki et al., 1994; Groettrup et al., 2001; Piccinini et al., 2003; Stratford et al., 2006; Stohwasser et al., 2000; Rivett et al., 2001). Also tumor necrosis factor alpha (TNF-a) (Lin et al., 2005; Llovera et al., 1997), or by lipopolysaccharides (LPS) (Nelson et al., 2000) are able to induce these subunits. In general, these subunits and the formed proteasome are indicated by an additional ‘‘i’’ (i20S). The inducible proteasomal subunits are homologs of the catalytic active ones b1, b2 and b5 that are replaced by b1i (also termed as LMP2 for low-molecular-weight protein 2), b2i (also MELC1 for multicatalytic endopeptidase complex-like-1) and b5i (or LMP7) (Driscoll et al., 1993; Glynne et al., 1991; Cerundolo et al., 1995). It has been shown that these inducible subunits are only incorporated in de novo synthesized proteasomes (Kloetzel, 2001; Kruger et al., 2003). Therefore, INF-c induces the expression of Ump1 and the inducible forms of b-subunits, b1i, b2i and b5i. Paradoxically even though the mRNA of Ump1 is increased, the amount of the free protein decreases as found in HeLa cells, and the half-life of Ump1 is lowered from 82 to 21 min (Heink et al., 2005). This is found to be due to an increasing proteasome formation that directly correlates with the degradation of Ump1, the first proteasomal substrate (see above), thus Ump1 turnover can be used as a direct indicator for proteasomal formation rate. Two main factors turn the formation of the immunoproteasome i20S into a highly efficient process. First that b5i shows a higher affinity to Ump1 than b5 does, resulting in a higher formation of i20S compared to c20S. It is interesting that Ump1 is able to bind both the propeptide prob5i and to the processed form b5i, suggesting two different binding sites (Heink et al., 2005). Second: the i20S proteasome has only a short half life of 27 h compared to that of the c20S of about 8–12 days (Tanaka and Ichihara, 1989, 1995). Thus both a quick increase and a fast removal of i20S are assured. In continuous stimulation after a period of about 7 days the c20S can be completely replaced (Khan et al., 2001) by i20S, whereas in short time stimuli, the ratio of i20S to c20S increases. Surely not only i20S and c20S proteasomes are assembling, but forms that contain only between one and five inducible b-subunits. Whether these intermediate-type proteasomes have a special function is still unclear. In every case both Ump1 and b5i seem to be crucial for the formation of i20S. In cells not expressing b5i, Ump1 is increased in IFN-c treatment, but i20S it not formed. In Ump1 knockdown cells, proteasome mediated proteolysis decreases very fast to about 60% of the normal amount after 24 h and 40% after 48 h (Heink et al., 2005) and the amount of fully assembled proteasomes is lowered significantly (Witt et al., 2000). On the other hand, an overexpression of Ump1 has been shown to increase the cellular proteolysis mediated by the proteasome (Chondrogianni and Gonos, 2007). The inducible b-subunits show a constant increased intracellular amount as found in the muscle tissue of aged rats compared to young ones (3–6-fold increased amounts). The same was found in neurons, astrocytes and endothelial cells in the hippocampus of aged humans (about 70 years) compared to a younger control (about 42 years) (Mishto et al., 2006). Thus the inducible form seems to accumulate over age in cells that contain usually only the ‘‘housekeeping’’ c20S proteasome, especially in the main mammalian postmitotic aging tissue like nerve and muscle cells. 3.4. The thymus specific proteasome (thymoproteasome) Another proteasomal subunit is the so-called beta5t (b5t) that has been found in mice, exclusively in cortical thymic epithelial cells (cTECs) (Murata et al., 2007), playing a role in the positive selection of thymocytes (Murata et al., 2008). The term ‘‘thymoproteasome’’ has been suggested for the ‘‘configuration’’ b1i, b2i and b5t. It is hypothesized that b5t is responsible for the presentation of antigens on the cell surface that are not found in other cells, resulting in positive selection of CD8+ T-cells (Murata et al., 2008). The gene coding of b5t is an extra gene adjacent to the one coding b5 (human and mouse genome), the expressed protein shows a close relation to both b5 and b5i. About 20% of the thymic proteasomes were found to contain b5t instead of the constitutive b5. Both b1i and b2i were preferred incorporated compared to the constitutive subunits in proteasomes containing b5t. The antigens presented by the major histocompatibility complex class I (MHC-I) (Fruh and Yang, 1999; York and Rock, 1996; Pamer and Cresswell, 1998) complexes show hydrophobic C-termini functioning as an anchor in MHC-I binding (Young et al., 1995), resulting from the proteolytic characteristics of b5 cleavage The proteolytic pocket of b5t mainly contains hydrophilic amino acid residues in contrast to b5 and b5i and thus incorporation of b5t (via de novo synthesis) reduces the chymotrypsin-like proteasomal activity by 60–70%, without influencing to the other two activities (Murata et al., 2007). Both maximal velocity of protein degradation and Michaelis constant are lower in b5t compared to b5 and b5i. This decreases the amount of produced oligopeptides showing a hydrophobic C-terminus. These hydrophobic C-termini are preferred for incorporation into the MHC-I binding grove, suggesting b5t reduces the amount of MHC-I presentable antigens. This might result in a diminished production and thus presentation of MHC-I high affinity binding oligopeptides, and as mentioned resulting in a lowered interaction of cTECs with the ab T-cell antigen receptor (TCR) causing a higher chance of positive selection of these cells (Murata et al., 2007, 2008). In a mouse model b5t-deficient animals showed imperfect development of CD8+ T-cells, manifested by a decrease of about 80% (Bevan, 2007), suggesting that b5t enhances the selection of CD8+. Despite of that the amount of antigen loaded MHC-I presented on b5t deleted cells does not decrease. In most cells lysosomal cathepsin S is a necessary factor in antigen presentation, whereas this task is performed by cathepsin L in thymus cortical epithelial cells; deletion of cathepsin L in these cells reduced CD4+ selection without any influence of the MHC-II amount. In these two ways of presentation for
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
207
MHC-I (by proteasomes for CD8+ cells) and MHC-II (by lysosomes for CD4+ cells) the different presented antigenic peptides decide about positive or negative selection of mature T-cells. Negative selection sorts out T-cells with high affinity to selfantigens that would otherwise result in autoimmune reactions, while a weak interaction with MHC results in T-cell maturation. A mediocre affinity to self-antigens usually results in a positive selection. 3.5. NO as a regulator of proteasomal activity A recent study by Kotamraju et al. (2006) showed a strong increase of proteasomal activity mediated by the NO-radical in bovine aortic endothelial cells (BAECs) via the NO/cGMP/cAMP pathway. In this aortic cell line the important vasodilator NO triggered increased expression of two inducible proteasomal subunits: b1i (Lmp2) and b5i (Lmp7). The effect was also obtained using NO-donors, among them DETA/NO (Jung et al., 2007a), causing an about 2-fold increase in the expression of caspase- and trypsin-like proteasomal activity, respectively, the according subunits b1i and b5i. The same effect was induced by cell-treatment with cell permeable analogons of cAMP or cGMP. On the other hand, b1i- and b5i-upregulation was inhibited by co-treatment of the cells using guanylyl (Lucas et al., 2000) or adenylyl cyclase (Dobson et al., 2003) inhibitors. Further experiments revealed that DETA/NO or cyclic nucleotides activated downstream both protein kinase A (PKA) (Ryu et al., 2005) and protein kinase G (PKG) (Cudmore et al., 2006). Since pretreatment of the cells with inhibitors for PKA/G significantly reduced b1i- and b5i-upregulation, too. Both in iNOS/ mice and mice treated with N-[3-(aminomethyl)benzyl]acetamidine (1400 W) (Alderton et al., 2001), an inhibitor of iNOS, the amounts of b1i and b5i were significantly decreased in aortic endothelial cells compared to wild type animals (Kotamraju et al., 2006). It is noteworthy to mention, that the proteasomal activity in tissues apart from the aorta (liver, kidney and lung) of iNOS/ mice was not affected, though the exact mechanism is not yet known. In the mouse genome a cAMP regulated region was found that is responsible for the expression of the b5i proteasomal subunit (Zanelli et al., 1993), that may explain the upregulation of that subunit without the presence of IFN-c or TNF-a. The mechanism presented by Kotamraju was a direct upregulation of cGMP by NO, that triggered an increase of the intracellular cAMP mount. Both increased cGMP and cAMP activated the protein kinases G (by cGMP) and A (by cAMP) that activated the ‘‘cAMP-response element-binding protein’’ (CREB) (Seo and Chung, 2008) by phosphorylation. CREB (inducible also by IFN-c) in its phosphorylated form was now able to induce the expression of both b1i and b5i, while the effect of b1i was the strongest. Higher expression of the inducible subunits increased in summary the overall proteasomal activity. 3.6. The 11S regulator The mammalian PA28 (also termed as ‘‘11S’’, ‘‘REG’’ or ‘‘PA26’’ (in Trypanosoma brucei)) regulator particle is found as a heterohexameric, a heteroheptameric or a homoheptameric complex. The complex is able to bind to an a-ring of the proteasome and to increase the rate of peptide degradation. Protein degradation after the binding of the 11S to the 20S is performed in an ATP-independent way, and only the degradation of already unfolded substrate proteins is increased, but not the proteolysis of natively folded forms (Rechsteiner et al., 2000). Studies investigating the inducible form of the proteasome with two PA28 regulator proteins attached, showed in human red blood cells a 10-fold increase of the activity of b2 (trypsin-like) and a 50-fold increase in both b1-(peptidyl-glutamyl-like) and b5-(chymotrypsin-like) activities (Di, 1992). Kuehn and Dahlman even detected a 60-fold increase in the activity of b1 (peptidyl-glutamyl-like) in human red blood cells (Kuehn and Dahlmann, 1996). Attachment of the PA28 activator increases Vmax and reduces Km for proteasomal degradation (Dubiel et al., 1992). The PA28 activator protein is build of three different subunits, PA28a (28589 Da), PA28b (27230 Da) and PA28c (29365 Da) able to associate in different forms to built up the PA28 regulator particle (Fig. 14). The single subunits show about 35% similarity in their amino acid sequences (Noda et al., 2000). Each of the subunits contains four long helices: helix 1 (residues 7–46), helix 2 (107–139), helix 3 (147–191) and helix 4 (195–239) (Fig. 14). Another loop, that is critical for proteasome-activation, connects helix 2 and helix 3 on the base of the subunits where they attach to the a-ring of the proteasome. The amino acids 141–149 (binding of the regulator particle to the 20S proteasome) are highly conserved as well as the C-termini 240–249 (also binding to the 20S proteasome). For example the PA28a N146Y mutant is able to bind, but not to activate the 20S proteasome (Zhang et al., 1998). PA28a and PA28b show some highly conserved structures: the deletion of some C-terminal amino acids from PA28a or -b results in the inability of the regulator protein to bind the proteasome (Ma et al., 1993; Zhang et al., 1998). Mutants lacking 1–9 of the C-terminal amino acids were unable to bind 20S proteasome. The Pro240-residue has been revealed in PA28a as essential for proteasome activation (Zhang et al., 1998). Activation of the core proteasome is presumable induced by a conformational change in the gating N-termini of its a-rings. PA28a and -b proteins are inducible by IFN-c, while PA28c (a major auto-antigen in lupus erythematosus (Nikaido et al., 1990), formerly termed as Ki, sharing about 25% of sequence identity with PA28a and -b) is not. In HeLa cells treated with IFN-c showed a 3–5-fold increase of the PA28a and b-subunits, as well as IFN-a was able to induce PA28a and -b expression in HeLa, Raji and Jurkat cells (Ahn et al., 1996), while PA28c amounts remained unchanged. The distributions of the three PA28 isoforms in tissues are still under discussion and in part inconsistent; though it seems to be clear, that PA28a and b are mainly found in immuno-competent tissues like thymus, spleen and lung, while almost undetectable in brain and nervous tissue (Rechsteiner et al., 2000). Investigations of the intracellular distributions showed homogenous distribution of a
208
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
Fig. 14. The PA28a subunit and the PA28a7 complex. This image shows the PA28a7 regulator cap and a model of a single PA28a-subunit. Top left the whole regulator cap is depicted in front view. The single subunits are color coded to enable differentiation. The right the top view of the regulator particle is shown, displaying the inner channel. Bottom left the inner pore is visible after removing of three PA28a-subunits. The channel is defined by helix 3 (H3) is the seven subunits. The image bottom right shows a tube view representation of a single PA28a-subunit, while the four main helices (H1–H4) are indicated. The link between H2 and H3 is the binding site to the a-rings of the 20S core proteasome (20S binding site). PA28a7 can be synthesized in vitro but plays no relevant role in vivo due to its instability and low probability of formation, but is until now the only 11S regulator particle that has been analyzed by X-ray crystallographic analysis (Knowlton et al., 1997). The molecular structure is computed from the correspondent data set available on ‘‘RCSB Protein Data Bank’’ using electron density to define the surface of both the particle and its single subunits at a resolution of 2.80 Å (Knowlton et al., 1997).
and b in both cytosol and nucleus, while PA28c was almost exclusively found in the nucleus (Soza et al., 1997). PA28a and -b activate all three proteolytic activities of the 20S proteasome, whereas PA28c only activates the trypsin-like one on b2 (Realini et al., 1997). The stoichiometry of the PA28a/b complex was under heavy discussion: chemical cross-linked activator complexes led to the assumption of an a3b3 complex first, with an alternating arrangement of the single subunits (Song et al., 1996; Ahn et al., 1995). Finally Zhang found the a3b4 structure containing a b b dimer, but no a a, what was confirmed by mass spectrometry (Zhang et al., 1999). Until now, in biological systems the complexes PAa3b3 (in each case with alternating arrangement of the PA28a- and PA28b-subunits), PA28a4b3, PA28a3b4 and the homoheptamer PA28c7 are known (Rechsteiner et al., 2000). It is suggested that the most stable form is the a3b4-arrangement of the subunits, when PA28a and -b subunits are mixed in vitro using an a/b-ratio of 1.2, an a3b4 and a4b3 are found with a predominant formation of the first structure (Zhang et al., 1999). Yet pure solutions of PA28a resulted in the formation of a heptamer, even though PA28a7 is very instable and thus not found in a cell in the presence of PA28b. PA28c only binds to itself, while PA28a and -b bind strongly between each other and only weakly to themselves (Realini et al., 1997). Interestingly, in pure solutions PA28b is only found as monomer, in contrast to PA28a that is self associating (Fig. 14). The a3b4 structure has a molecular mass of 196 kDa. The affinity of the different PA28 subunits to the 20S core proteasome is in the order PA28ab (as a dimer) > PA28c > PA28a > PA28b (Li and Rechsteiner, 2001). The complete PA28a3b4 complex is barrel shaped and has dimensions of about 60 Å in height and a diameter of about 90 Å at the base binding an a-ring of the 20 S proteasome. It has a central opening with a distal diameter of about 20 Å and a base diameter of about 30 Å. The channel formed by the subunits is mainly shaped by hydrophilic and charged amino acid residues from helix 3. Binding of PA28a3b4 induces a conformational change in the 20S core particle, that enables the already mentioned ‘‘dual cleavage’’. In the process of dual cleavage a conformational change induced by the PA28 proteasome activator, enables a coordinated cleavage of a protein in the inner chamber at the same time by two b-subunits in different b-rings, that occurs in low and medium (10–180 lM) substrate concentrations. At very high concentrations (300 lM) the model suggests com-
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
209
petition of the substrate proteins for active sites of the proteasome, resulting in single cleavage again, but at significantly higher rates compared to the 20S alone (Dick et al., 1996). Only dual cleavage enables the proteasome to generate quickly large amounts of MHC-I presentable oligopeptides in a rapid one-step reaction. The probability of a second cleavage close enough to the first cleavage resulting in useful peptides would be much to low under in vivo conditions without the induced changes by PA28. The inducible proteasomal subunits, as PA28a and -b, are involved in the immune response and are inducible by IFN-c, too. Knockout mice of PA28b showed significant decreases in immune response, in contrast to PA28c knockout animals, suggesting that PA28c may not be involved (Murata et al., 1999). PA28a and -b are mainly found in the cytosol and only in low amounts in the nucleus of a cell, while PA28c shows an inverted distribution (Wojcik et al., 1998). PA28a and -b seem to be involved exclusively in antigen presentation, since no other biological function was revealed until now. As mentioned in contrast to the PA28a/b, the PA28c7 complex does only enhance the trypsin-like proteolytic activity on b2 about 10–12-fold. The other two subunits are inhibited (0.4–0.6-fold of normal activity) (Li and Rechsteiner, 2001). This may be due to an improper conformational changing of the b1- and b5-subunit, since a decreased substrate entrance, selective filtering or reduced substrate binding were excluded. PA28c-knock-out animals showed defects in their regulation of mitosis and apoptosis, though the biological mechanism is unclear (Murata et al., 1999; Barton et al., 2004). The PA28c7 complex was shown to be involved in cell cycle regulation: some regulating proteins seem to be exclusively degraded in a PA28cdependent manner. One of them is the cell cycle inhibitor p21Cip1. This protein has been shown to be degraded independent of polyubiquitination, even if it is found intracellularly mostly in a native and folded state. So PA28c7 seems to be a new and ubiquitin-independent way of native protein degradation. A recent study showed not only the regulation of p21Cip1 but also p16Ink4a and p19Arf. Knock-down of PA28c stabilized the intracellular amount of p21Cip1, particularly in the brain of mice, where PA28c is expressed in large amounts in contrast to PA28a and -b. In p21Cip1 six C-terminal amino acids residues (amino acid 156-161) have been identified, that are essential for the binding to PA28c7; these six amino acids are proven to bind the a7 subunit of the 20S core proteasome (Touitou et al., 2001). In addition to that PA28c turned out to be important for protein degradation of lysine-free or -low proteins (as mentioned, lysine is needed for poly-ubiquitination). Since viral proteins only show very low amounts of lysine, PA28c7 might be a pathway of degradation for viral pathogenic proteins. This seems plausible, considering the fact that PA28c is highly conserved in the nuclei from worms to humans (Chen et al., 2007). Another protein, that is specifically regulated by PA28c7 is the steroid receptor coactivator-3 (SRC-3/AIB1), an oncogene overexpressed in breast cancers (Li et al., 2006).
Fig. 15. The relative cytosolic amount of hybrid proteasomes as found in HeLa cells. This image shows the relative distribution of 20S proteasomes (first from left) and the different combinations of regulator particles attached to the core proteasome as found in the cytosol of HeLa cells according to Tanahashi et al. (2000). Complexes of proteasomes with only one attached regulator protein where not detectable by immunochemical methods, suggesting they only represent an extremely short-lived intermediate. ‘‘Immuno’’ indicates the immunoproteasome containing two 11S regulator caps, ‘‘hybrid’’ the hybrid proteasome with both 11S and 19S caps and ‘‘26S’’ the ATP-dependent form of the proteasome fit out with two 19S regulators.
210
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
Fig. 16. The molecular-coupling hypothesis. This image depicts the highly hypothetic coupling of the immunoproteasome, mediated by the PA28 regulator cap, to the TAP1–TAP2-complex and thus to the whole MHC-I loading machinery, containing each four molecules of ERp57, MHC-I, tapasin and calreticulin at the inside of the endoplasmic reticulum. The advantage of this model is the direct channeling of antigenic oligopeptides to the MHC-I target without exposing them to cytosolic proteases. Image according to Rechsteiner et al. (2000).
3.7. The hybrid proteasome (PA28-20S-PA700) Hybrid proteasomes, as termed by Tanahashi et al. (2000), are proteasomes that contain not two regulator proteins of the same kind but are built of one 20S ‘‘core’’ proteasome (or i20S) with both a PA700 and a PA28 regulator particle attached. The relative amounts of the different possible hybrid proteasomes have been determined by Tanahashi et al. as displayed in Fig. 15. The hexameric PA28 (ab)3 and the two heptameric forms (both (ab)3a and (ab)3b) are all involved in the formation of hybrid proteasomes, while PA28c (c7) is not. Under the influence of IFN-c both 26S proteasome and hybrid proteasomes can be isolated by immunoprecipitation (Hendil et al., 1998). Tanahashi et al. (2000) showed that PA28 is actually involved in the ATP-dependent proteolysis by forming PA28-20SPA700 complexes, and that ATP is needed to attach the PA28 regulator particle to the 20S proteasome, comparable to the formation of the 26S proteasome. The exact function of this proteasome is unknown, but it might be possible that the substrate protein is recognized and bound by the 19S regulator and degraded by the core particle, that changes its proteolytic specificities upon PA28-binding (Hendil et al., 1998), though the proteolytic activity of the 26S proteasomes was higher than the activity shown by the hybrid forms. Nevertheless cooperation in the cellular antigen processing of the immunoproteasome and the hybrid form is possible, since both complexes are induced by IFN-c. As mentioned, most of the proteolysis for MHC-I antigen presentation is performed by both the 26S proteasome and the so-called hybrid-proteasome (PA28-20S–19S). Considering the fact, that the PA28 activator protein is not able to perform degradation of folded proteins, it seems possible that the substrate has to be unfolded first by the ATP-dependent 19S-regulator cap. On the other hand whether the short oligopeptides might be released after 19S–20S-mediated degradation and further processed by another proteasome that is fit out with a PA28-regulator, is not clear. In the same way it is unknown, whether the 26S-proteasomal substrates that are degraded by a c20S core particle have to be further processed by an immunoproteasome. The ‘‘molecular coupling hypothesis’’ suggests the attachment of a hybrid proteasome to a TAP protein channel in the ERmembrane via the attached PA28 regulator protein (Fig. 16). In this case the polyubiquitinated, but natively folded antigenic protein is recognized and unfolded by the 19S regulator cap, while guided into the core proteasome (in this case an immunoproteasome), where it is degraded. Both fragment length and characteristics are influenced by the PA28 regulator that transfers the antigenic oligopeptides directly into the TAP transporter protein, in order to protect the fragments from degradation by cytosolic proteases. Further antigen processing (N-terminal trimming) is performed by ER-resident proteases, like ERAAP and ERAP1/2. This hypothesis is based upon highly conserved structures at the distal end of the PA28 regulator that are neither involved in binding or activation of the core particle. These sequences are termed as ‘‘KEKE’’-motifs (Realini et al., 1994). KEKE-motifs have been hypothesized to mediate protein–protein interactions and have been found in four pro-
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
211
teasomal subunits and in five different subunits of the 19S regulator cap. Moreover they were revealed in Hsp90 and calnexin, two proteins involved in the process of epitope loading to MHC-I. This directly led to another hypothesis concerning the binding of heat shock proteins, delivering protein substrates to the hybrid-proteasome. This hypothesis was based on findings, that heat shock proteins are involved in immune response (Wells et al., 1998; Binder et al., 2001). But this idea was already disproven in 2006 using the SIINFEKL epitope of ovalbumin: experiments showed no promotion of its MHC-I presentation (Gonciarz-Swiatek and Rechsteiner, 2006). 3.8. The PA200 regulator protein The PA200 proteasome activator is a protein, exclusively located in the nucleus of mammalian cells. The yeast homolog (S. cerevisiae) Blm10 shows about 20% of sequence homology (Iwanczyk et al., 2006) to the mammalian one. First investigations of the PA200 activator structure and the binding of PA200 to the 20S proteasome were done by electron microscopy. Three dimensional reconstructions from this data showed an asymmetric dome-shaped structure (100 Å in diameter, like the 20S proteasomal a-ring, and about 60 Å high) with an inner cavity, sitting on one or both a-rings like a cap (Ortega et al., 2005). In contrast to other proteasomal regulator proteins, PA200 is a monomeric structure of about 200 kDa. The binding of PA200 to the a-ring seems to be due to a structure containing multiple HEAT-repeats while contacting almost every single a-subunit, except a7, while the yeast Blm10 really binds every single one (Iwanczyk et al., 2006). Since the N-terminal end of the a-subunits normally enclose the gate formed by the a-rings and thus regulating substrate access to the inner chamber, the PA200 regulator must change their structure in a way resulting in enhanced substrate access. This was shown to be the primary mechanism of activation by reconstructions from electron microscopic data, too (Ortega et al., 2005). Like the 11S regulator, the PA200 increases the degradation of small oligopeptides, releasing fluorescent products after degradation, but does not enable the processing of natively folded proteins. Electron microscopic studies of isolated proteasome and PA200 proteasome complexes from bovine testis by Ortega et al. (2005) showed a ratio of 50:40:10 of 20S:PA200-20S:PA200-20SPA200. Only little is known about the function of PA200. Experiments of gene deletion or overexpression did not result in a significant phenotype (Iwanczyk et al., 2006). On the other hand the role of PA200 in the repair of DNA after exposure to ionizing radiation or oxidizing agents seems to be ensured (Ustrell et al., 2002). PA200 is expressed in response to ionizing radiation and accumulates in its hybrid form on chromatin (Blickwedehl et al., 2007). Knockdown cells of PA200 showed both instability of the genome and a reduced survival rate after been exposed to ionizing radiation. The genome-stabilizing functions of PA200 seems to be due to its ability to enhance the peptidyl-glutamyl-like (b1) cleavage of the proteasome (Blickwedehl et al., 2008). On the other hand deletion of Blm10 in the yeast A364a-strain showed neither a significant effect on growth nor viability after cell treatment with the DNA-damaging agents bleomycin or phleomycin (Iwanczyk et al., 2006). Most interestingly no increased sensitivity of A364a yeast was shown to UV- or gamma-irradiation, methyl methane sulfonate, camptothecin or hydroxyurea. Blm10 overexpression showed a reduced growth, but this might be an effect of increased binding to the 20S proteasomes, thus detracting it from other important functions. Interestingly, due to the presence of the PA200 protein in the nucleus another form of the proteasome can be formed. In yeast the Blm10 homolog is known to form the PA200-20S–19S proteasome complex (Ustrell et al., 2002; Schmidt et al., 2005). After treatment of HeLa cells with ionizing radiation an immunoprecipitation showed a coprecipitation of PA200 in complex with 20S–19S, though the total levels of both 20S and 19S were not increased. Thus irradiation seems to induce an increased formation of the PA700-20S–19S hybrid proteasome, mediated by DNA damage (Blickwedehl et al., 2008). Twenty four hours after irradiation the PA200-20S–19S complex accumulated on chromatin. Looking at the proteolytic activity associated with chromatin, the trypsine-like one (b2 subunit) is increased 6-fold, the peptidyl-glutamyl-like one (at b1) up to 19-fold, accompanied by a 5- to 8-fold accumulation of 20S at the chromatin after irradiation (Rapic-Otrin et al., 2002). Further investigations showed the accumulation of PA200 that is not dependent from ATM, a PI3-kinase-like kinase (Vantler et al., 2006; Zaiss et al., 2002), starting the signaling-cascade after cell stress by irradiation and the triggers the tumor suppressor p53 (Das et al., 2001; Jiang et al., 2004;Essmann et al., 2005). The functions of PA200 might be to increase b1-proteasomal subunit activity that is essential for the survival of cells after exposure to ionizing radiation (Blickwedehl et al., 2008). 3.9. UPS inhibitors Investigations using numerous proteasome inhibitors in the past 15 years served as an excellent tool in the discovery of many new substrates of the ubiquitin–proteasome pathway and its role in different cellular and physiological processes. Since the proteasome is included in many cellular functions, for example, control of cell cycle progression by degradation of cyclins and inhibitors of cyclin-dependent kinases (Koepp et al., 1999), regulation of cell growth and gene expression by the degradation of transcriptional regulators, such as c-Jun, E2F-1 and L-catenin (Hershko and Ciechanover, 1998), termination of certain signal transduction cascades by degradation of activated protein kinases, e.g. src and protein kinase C (Harris et al., 1999; Lu et al., 1998), proteasome inhibitors have multiple and complex effects on cells (Kisselev and Goldberg, 2001). The primary consequence of proteasome inhibition is a decrease of overall rates of protein breakdown in cells leading to a rapid accumulation of short-lived proteins conjugated to ubiquitin. Proteasome inhibition also causes accumulation of misfolded and damaged proteins and these accumulated unfolded polypeptides trigger expression of heat shock proteins and
212
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
Table 2 Different classes of proteasome inhibitors. Synthetic inhibitors
Natural inhibitors
Peptide aldehydes
Epoxyketones
– – – – – – – –
– – – – – – –
ALLN MG132 MG115 PSI Leupeptin Peptide glyoxal CEP1612 Fellutamide B
Epoxomicin Eponemycin Dihydroeponemycin TMC-86A and B TMC-89A TMC-96 YU101
Peptide boronates
Lactacystin
– PS-341 – DFLB – PS-273
– Clasto-b-lactone – PS-519
Peptide vinyl sulfones – NLVS – [125I]YL3VS – NIP-Leu-Leu-Asn-VS – Ada-[125I]Tyr-Ahx3-Leu3-VS – Ada-Lys(Bio)-Ahx3-Leu3-VS a-Ketoaldehydes and a-ketoamides
DCI TMC-95A Gliotoxin Syringolin A Glidobactin A
endoplasmic reticulum stress proteins expressed as protection mechanisms against various toxic conditions (Kisselev and Goldberg, 2001). This large number of changes with proteasome inhibition raised the interest on synthetic and naturally occurring inhibitors (Table 2) of proteolytic activities of the proteasome and still several studies are executed to find out new compounds. 3.9.1. Inhibitors of the proteasome To understand the mechanisms of proteasome inhibitors, it is important to understand the proteasomal structure. As mentioned before, proteasomes form a class of proteolytic enzymes called threonine proteases. Unlike any other protease, all the proteolytic sites in proteasomes utilize N-terminal threonines of the b-subunits as the active site nucleophiles. Although the 26S proteasome exhibits many different enzymatic activities, small molecule inhibitors are available for the proteolytic sites of the 20S proteasomes (Kisselev and Goldberg, 2001). Enzyme-specific inhibitors of proteases are usually short peptides linked to a pharmacophore, generally located at its C-terminus. The pharmacophore interacts with a catalytic residue with the formation of reversible or irreversible covalent adducts, while the peptide portion specifically associates with the enzymes substrate binding unit in the active site. Since the proteasome has multiple active sites, the inhibitors are classified according to the sites of proteasome. For a reduction in the protein breakdown the inhibitor is not required to inhibit all active sites. In fact, inhibition of the chymotrypsin-like site or its inactivation by mutation alone causes a large reduction in the rates of protein breakdown (Rock et al., 1994; Heinemeyer et al., 1997; Chen and Hochstrasser, 1996). In contrast, inactivation of trypsin-like or caspase-like sites had little effect on overall proteolysis (Heinemeyer et al., 1997; Arendt and Hochstrasser, 1997; Kisselev et al., 1999). In addition, most inhibitors of chymotrypsin-like sites are highly hydrophobic and consequently much more cell-permeable than inhibitors of the trypsin or caspase-like sites, which contain charged residues. The chymotrypsin-like active site of proteasomes cleaves primarily after large hydrophobic residues, similar to the preference of intracellular cysteine proteases such as cytosolic calpains and many lysosomal cathepsins (Chapman et al., 1997). Therefore, high selectivity of proteasome inhibition by peptide based compounds would be hard to achieve just by simply manipulating the peptide portion of the inhibitor. Instead, the use of a pharmacophore with preference for the proteasome’s N-terminal threonine is required. So proteasome inhibitors are broadly categorized as synthetic analogs and natural products due to their chemical structures (Table 2). Synthetic inhibitors are peptide based compounds formed by combining a peptide moiety with a reactive pharmacophore group, an aldehyde, a boronate, a vinylsulfone, a benzamide, an a-ketoamide or an a-ketoaldehyde (Bochtler et al., 1999). Peptide aldehyde inhibitors were the first proteasome inhibitors identified (Rock et al., 1994; Vinitsky et al., 1992) and researchers have developed a myriad of aldehyde inhibitors because of the convenience in synthesis and optimization. They are the most widely used inhibitors with higher potency and increased selectivity toward the chymotrypsin-like activity of 20S proteasome. Aldehyde inhibitors of the chymotrypsin-like site are slow-binding (Vinitsky et al., 1992), but they are cell permeable and reversible. These inhibitors have fast dissociation rates, are rapidly oxidized into inactive acids by cells and are transported out of cell by the multi-drug resistance (MDR) system carrier (Kisselev and Goldberg, 2001). Consequently, in experiments involving cultured mammalian cells and yeast, effects of these inhibitors can be rapidly reversed by removal of the inhibitor under physiological conditions (Lee and Goldberg, 1996). In the inhibition process the
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
213
functional aldehydic group is readily subject to a nucleophilic attack by hydroxyl and thiol groups. The N-terminal threonine hydroxyl group as a nucleophile on 20S proteasome b-subunits interacting with the aldehyde by forming a covalent hemiacetal (Vinitsky et al., 1992). Peptide aldehydes have been used for long time as inhibitors of cysteine and serine proteases, and thus can inhibit such proteases in vivo. Leupeptin, a standard serine and cysteine protease inhibitor, was shown to be trypsin-like activity blocking agent of 20S proteasome (Wilk and Orlowski, 1983). ALLN (Ac-Leu-Leu-nLeu-al), which was used in earlier studies, was first described as a calpain inhibitor I (Wang and Yuen, 1994), and is 25-fold more potent against cathepsin B and calpain than against the proteasome (Rock et al., 1994). MG132 (Z-Leu-Leu-Leu-al, also termed Cbz-LLL or zLLL) is not only significantly more potent than ALLN against the proteasome (Palombella et al., 1994), but is much more selective, as shown by the fact that inhibition of calpains and cathepsins requires at least 10-fold higher concentrations (Tsubuki et al., 1996). Another peptide aldehyde, PSI (Z-Ile Glu(Ot-Bu)-Ala-Leu-al), inhibits the proteasome 10-fold better than calpain but is less potent than MG132 (Figueiredo-Pereira et al., 1994). Also a peptide glyoxal adduct is produced by additional ketone moiety at the a-position of the peptide aldehyde to improve its efficacy (Lynas et al., 1998). Finally, the dipeptide aldehyde CEP1612 appears at least as good as MG132 in potency and selectivity (Harding et al., 1995), but is not available commercially. Fellutamide B is a very potent inhibitor of the chymotrypsin-like sites and it also inhibits the trypsin-like and caspase-like sites, albeit at higher concentrations (Hines et al., 2008). It binds to all three catalytic sites of the yeast 20S proteasome with the formation of a hemiacetal bond, and the N-terminal aliphatic tail, which distinguished this compound from other peptide aldehydes reported to date. Therefore, it adopts different conformations at different active sites. Since MG132, PSI, MG115 (Z-Leu-Leu-nVal-al) and ALLN can all inhibit calpains and various lysosomal cathepsins in addition to the proteasome, when using these inhibitors in cell culture it is important to perform control experiments to confirm that the observed effects are due to the inhibition of the proteasome. For this purpose, agents can be used which block intracellular cysteine proteases, but do not inhibit proteasomes (Rock et al., 1994). As the most potent and selective of commercially available aldehydes, MG132 is preferable to ALLN, MG115 (Z-Leu-Leu-nVal-al), or even PSI. On the other hand, the least selective inhibitor, ALLN, because of its ability to inhibit most major proteases in mammalian cells, is probably the best tool for prevention of unwanted proteolysis, for example during isolation of proteins from mammalian cells. Although peptide aldehydes are potent inhibitors of the 20S proteasome and widely used in biological studies their highly reactive aldehydic groups are a major limitation for their use as therapeutic agents. Peptide boronates are much more potent and selective inhibitors of the proteasome than the aldehydes. The boronate analogue of MG132, MG262 (Z-Leu-Leu-Leu-boronate), is 100-fold more potent than the aldehyde (Adams et al., 1998). The mechanism of inhibition by these slow-binding compounds is yet to be confirmed by X-ray analysis, but it is presumed that boronates, like peptide aldehydes, form a tetrahedral adduct with the active site threonine. An empty p-orbital on a boron atom is positioned to accept the oxygen electron pair of the amino terminal threonine residue of the 20S proteasome to form a stable tetrahedral intermediate (Myung et al., 2001) (Fig. 17). The stable tetrahedral borane complex shortens the peptide chain length of boronic acid based inhibitor to a dipeptide and this increases solubility and membrane permeability and provides, therefore, advantages by using it as a therapeutic agent (Lynas et al., 1998; Bogyo et al., 1998). The boronateproteasome adducts have much slower dissociation rates than aldehyde-proteasome adducts, and although boronates are considered reversible inhibitors, the inhibition is practically irreversible over hours. Boronates are also more selective inhibitors than aldehydes and are very poor inhibitors of thiol proteases, due to the weak interactions between sulfur and boron. Inhibition of serine proteases by many peptide boronates, such as PS-341 (pyrazylcarbonyl-Phe-Leu-boronate), is also 1000fold weaker than that of the proteasome (Adams et al., 1998). Boronates, unlike aldehydes, are not inactivated by oxidation and are not rapidly secreted from cells by MDR (Kisselev and Goldberg, 2001). This combination of potency, selectivity and metabolic stability makes the peptide boronates, better drug candidates than other classes of proteasome inhibitors, and one of the dipeptide boronates, PS-341, or with other widely used name bortezomib (Adams et al., 1998) is currently in phase III clinical trials in cancer patients. In contrast to other proteasome inhibitors, bortezomib exhibits enzyme specificity as well as a metabolic stability. It binds to the proteasomal chymotrypsin-like active site with high affinity (Ki = 0.6 nM) (Adams et al., 1998). Bortezomib is a dipeptide boronic acid that has been shown by Groll et al. (2006) in complex with the active centres of the 20S proteasome via X-ray crystallography. Bortezomib shows different affinities to the active proteasomal subunits
Proteasome H N
HN O
O H N
N O
N H
B
H2N OH
O
OH
PS-341 Fig. 17. Pseudo-covalent adduct formation in PS-341 and 20S proteasome interaction. The p-orbital on boron and the electron pair on oxygen form a pseudocovalent adduct between PS-341 and 20S proteasome.
214
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
(b5 > b1 b2), leaving the trypsin-like activity of b2 almost unaffected. In contrast to other known inhibitors, boronic acid shows a high specificity to oxygen nucleophiles, but not to cysteine nucleophiles, thus cysteine proteases are affected to a much lesser extent. The boron atom directly interacts with the Oc on the active Thr1 residue. Furthermore one of the –OHgroups attached to the boron atom forms a hydrogen bridge to the -NH2-group of the active Thr1, stabilizing the binding of bortezomib to the active centre. Interest in dipeptidyl boronates (and bortezomib specifically) was piqued when these agents were shown to inhibit growth and cell proliferation potently in a standard National Cancer Institute (NCI) screen of 60 cell lines derived from multiple human tumors (Adams et al., 1999). Two other dipeptide boronates, dansyl-Phe-Leu-boronate (DFLB) and PS-273 (morpholino-naphthylalanine-Leu-boronate, also termed MNLB or MG273), are useful fluorescent probes of the active sites, because binding of these inhibitors to them enhances the fluorescence of the environment-sensitive dansyl and naphthyl moieties (Mc et al., 1997). Peptide vinyl sulfones are one of the synthetic irreversible inhibitors of proteasome, covalently modifying its catalytic bsubunits. The hydroxyl group of the proteasomes catalytic threonine reacts with the double bond of the vinyl sulfone moiety in a Michael addition (Bogyo et al., 1997). The selectivity of inhibition by vinyl sulfones depends on the peptide part of the inhibitor. Vinyl sulfones are easier to synthesize compared to other irreversible proteasome inhibitors. Two compounds most widely used are NLVS (4-hydroxy-5-iodo-3-nitrophenylacetyl-Leu-Leu-leucinal-vinyl-sulfone), which modifies chymotrypsin-like subunits and [125I]YL3VS (Tyr-Leu-Leu-Leu-VS) which reacts with all b-subunits (Bogyo et al., 1998, 2000). Later, the compounds NIP-Leu-Leu-Asn-VS (Nazif and Bogyo, 2001) and Ada-[125I]Tyr-Ahx3-Leu3-VS, which react with all three active sites at comparable rates, and a biotinylated analogue of one of them, Ada-Lys(Bio)-Ahx3-Leu3-VS, have been synthesized (Kessler et al., 2001). There are reversible peptide a-ketoaldehydes (Lynas et al., 1998) and a-ketoamides (Chatterjee et al., 1999) which are similar to aldehydes in potency and selectivity and indanone peptides (Lum et al., 1998) which are even less potent. Thus these compounds do not appear to offer any advantage over other classes of inhibitors. In addition to the synthetic inhibitors, there are several natural inhibitors of the proteasome showing differences according to the origin and chemical structure. They display a variety of core structures and pharmacophores (Myung et al., 2001). Epoxyketones such as epoxomicin isolated from an unidentified actinomycete strain and eponemycin that was isolated from Streptomyces hygroscopicus are the most selective proteasome inhibitors because of their unique mechanism. These compounds react with both the hydroxyl and amino groups of the catalytic N-terminal threonine of the proteasome. The epoxomicin–proteasome complex reveals a six-membered morpholine ring, formed by the N-terminal threonine and epoxyketone moiety of the inhibitor (Groll et al., 2000) (Fig. 18). This structure suggests that the catalytic hydroxyl attacks
O N
N H
O
H N O
O
H
O
N H O OH
H O
N-terminal Thr on Proteasome
HN O H2N
O
H
Epoxomicin
O N O
N H
H N O
O N H
OH O
OH
N H OH
H N O
Proteasome
Morpholino adduct Fig. 18. Morpholino adduct formation between proteasome and epoxomicin. Following the reaction between epoxomicin and both the hydroxyl and amino groups of the catalytic N-terminal threonine of the proteasome a morpholine ring is formed.
215
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
O
O NH OH O
H O S
(in H2O)
N
:B HO
O
Lactacystin
NH
- NAcCys
N-terminal Thr on Proteasome
OH O
O
H
HN O
O
O
H2N
clasto-Lactacystin β-lactone
O NH
Proteasome
OH
HO O
O H2N
HN O
Ester adduct Fig. 19. 20S proteasome–lactacystin interaction. Lactacystin decomposed in water to clasto-lactacystin-b-lactone in H2O. b-Lactone covalently binds to the Nterminal threonine on the proteasome, resulting in an ester adduct formation and causes irreversible inhibition.
first the carbonyl group of the pharmacophore. Then the free amino group of the threonine opens up the epoxy ring and completes the formation of the morpholino adduct. Proteasomal subunits are the only target proteins modified by epoxyketones. The reason for this high specificity is the 2(R) configuration of epoxyketones to take specific advantage of the proteasomes unique catalytic mechanism by forming a cyclical morpholino ring. Inverting the configuration from R to S dramatically decreased the potency of proteasome inhibition (Meng et al., 1999). In addition epoxomicin reacts primarily with the chymotrypsin-like active site, and eponemycin and its synthetic analogue dihydroeponemycin react with the caspase-like and chymotrypsin-like sites at similar rates. More recently other linear peptide-epoxyketone natural products have been isolated from microbial metabolites. These include TMC-86A and B (Koguchi et al., 2000), TMC-89A and B (Koguchi et al., 2000), TMC96 (Koguchi et al., 2000). YU101 was developed by the optimization of P2–P4 amino acids to maximize the potency of inhibition in chymotrypsin-like activity (Elofsson et al., 1999). Among the natural inhibitors lactacystin is a well known and cell permeable inhibitor of proteasome. Fenteany et al. (1995) found that lactacystin, a Streptomyces lactacystinaeus metabolite, selectively modifies the b5 subunit of mammalian proteasome and irreversibly blocks its activity. Subsequent studies showed that lactacystin itself is not active and its decomposition product in aqueous solution and in neutral pH the clasto-lactacystin-b-lactone is suggested to have the activity. This b-lactone covalently binds to N-terminal threonine on proteasome results in ester adduct formation and causes irreversible inhibition (Dick et al., 1996) (Fig. 19). Also it was seen that the plasma membrane of mammalian cells is not permeable to lactacystin, the b-lactone which is formed in cell culture media spontaneously enters cells (Dick et al., 1997). b-Lactone is a significantly more selective proteasome inhibitor than peptide aldehydes. The potent analogue of lactacystin-b-lactone, PS519 (Phillips et al., 2000; Shah et al., 2002; Vanderlugt et al., 2000) has anti-inflammatory effects in animal models of asthma (Elliott et al., 1999) and is currently in phase I clinical trials for the treatment of stroke. 3,4-Dichloroisocoumarin (DCI) is a natural compound which is not often employed for in vivo studies due to its toxicity and lack of sensitivity. It inhibits proteasome by covalent modification of its N-terminal threonine (Akopian et al., 1997; Orlowski et al., 1997) and this inhibition occurs with the formation of non-hydrolysable acyl enzyme by cyclical ester in the structure of DCI. Recently new natural compounds are of interest for their proteasome inhibition properties. TMC-95A, a cyclic peptide metabolite of Apiospora montagnei, is a potent inhibitor of the chymotrypsin-like activity of pure 20S proteasomes (Kohno et al., 2000). It does not modify catalytic threonine, but binds to the active site by an array of hydrogen bonds between inhibitor and enzyme active sites (Groll et al., 2001). Gliotoxin is a fungal epipolythiodioxopiperazine toxin, inhibits proteasome by an unusual mechanism (Kroll et al., 1999). Unlike the other proteasome inhibitors, it binds to an unknown non-catalytic site and inhibits chymotrypsin-like activity of 20S proteasome. The disulfide bridge that gliotoxin has in its heterobicyclic
216
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
core forms a mixed disulfide bond with a proteasomal subunit. Syringolin A (SylA) is a virulence factor of the plant pathogen Pseudomonas syringae (Kisselev, 2008). Groll et al. (2008) found that it irreversibly inhibits all three types of proteasomal proteolytic sites. The structure of SylA-20S proteasome complex revealed a novel mode of inhibition whereby the hydroxylgroup of proteasomes catalytic threonine performs a Michael type 1,4-addition to the vinyl ketone moiety in the 14-membered ring of the inhibitor. This mechanism resembles mechanisms of inhibition by peptidyl vinyl sulfones (Groll and Huber, 2004). They also found that another microbial metabolite, Glidobactin A (GlbA), inhibited the chymotrypsin- and the trypsinlike activities of the proteasome and reacted with active site threonines in a similar fashion. As conclusion, relying on the truth that the ubiquitin–proteasome pathway is involved in many cellular processes, proteasome inhibitors are of potential interest in several research areas and especially in therapeutics and this leads to the importance of highlighting the chemical structures of different proteasome inhibitors. On the other hand taken into account, that the inhibition of the proteasome itself always blocks a multitude of pathways, for some approaches it might be more promising to target the selective ubiquitination machinery. 3.9.2. Inhibitors of deubiquitinating enzymes Proteins, which are to be degraded by the 26S proteasome, undergo covalent attachment of ubiquitin. This ubiquitin conjugation to proteins is a reversible process. The catalysis of this process is achieved by a large family of cysteine proteases called deubiquitinating enzymes (DUBs) consisting of ubiquitin COOH-terminal hydrolases (UCHs) and ubiquitin-specific proteases (USPs or UBPs) that remove ubiquitin units attached to proteins. UCH isozymes are involved in the processing of ubiquitin-fusion proteins, extended by small peptides or larger substrates with a flexible peptide linking the C-terminal domain. USPs are responsible for removing ubiquitin from larger proteins and disassembling the polyubiquitin chains from substrates before degradation by the proteasome (Wilkinson, 2000). This reaction allows ubiquitin to be used in subsequent conjugations. Deubiquitinating enzymes (DUBs) can also direct the turnover of specific proteins by removing ubiquitin or polyubiquitin chains from the substrate and avoid their degradation by the proteasome. In the processes such as transcriptional silencing, growth regulation and myogenesis, specific control of protein turnover is crucial (Debigare and Price, 2003). Besides ubiquitin, also ubiquitin like modifiers have more specialized roles. For example SUMO functions as a nuclear targeting signal (Hochstrasser, 2001), Nedd8 regulates the activity of E3 ligases and thereby influence ubiquitinylation (Ohh et al., 1998) and autophagy related modifiers including ATG12 control autophagy (Mizushima et al., 2003). Ub like (Ubl) ligases are counteracted by Ubl specific proteases (ULPs). UCHs, contain a 230 amino acid core catalytic domain, are papain like thiol proteases. BAP1, a related UCH family member, differs from the other family members in having a 500 amino acid catalytic domain interacting with the ring-finger domain of breast cancer tumor suppressor BRCA1 (Wilkinson, 2000). USPs as thiol proteases have an about 350 amino acid long core catalytic domain. Today are known sixteen different USPs in yeast. They vary from 50 to 250 kDa with a variety of Nterminal extensions and occasional C-terminal extensions. These extensions contribute to substrate specificity and/or localization (Wilkinson, 2000). A newly identified DUB (Ubp-M) is phosphorylated at the onset of mitosis and dephosphorylated during the metaphase/anaphase transition. Ubp-M may deubiquitinate one or more critical proteins involved in the condensation of mitotic chromosomes (Cai et al., 1999). Mammalian deubiquitinating enzymes were described to involve in the regulation of growth following cytokine-induced signaling events. These DUBs were shown to be immediate early gene products produced upon cytokine stimulation (D’Andrea and Pellman, 1998). Other processes regulated by DUBs include; receptor internalization by UBP4, inhibition of transcriptional silencing by UBP3 in yeast. UBPs also bind to a component called Sir4, a telomeric silencing complex. These regulatory processes suggest that DUBs are targeted to specific sites in the cell and point to several roles for deubiquitinating enzymes in cellular compartments (Wilkinson, 2000). DUBs have also been shown to have profound effects on development. The FAF gene in Drosophila codes for a DUB that is involved in development of the photoreceptors in the eye (Wu et al., 1999), modulating the activity of the Ras/receptor tyrosine kinase pathway. A similar enzyme exists in mouse where it is found to bind at the cell membrane at cell–cell contacts, and to bind to AF6, a downstream target of Ras that is degraded by the ubiquitin-dependent pathway and thought to be involved in adhesiondependent regulation processes (Taya et al., 1998). In normal cells, the ubiquitination and subsequent degradation of tumor suppressor protein p53 is mediated by Mdm2 as an ubiquitin ligase (E3) (Haupt et al., 1997). Stabilization of p53 is essential for its tumor suppressor function. Li et al. (2002) have identified a Herpes virus-associated ubiquitin-specific protease (HAUSP) also known as human USP7 as a novel p53-interacting protein. HAUSP belongs to the UBP family of DUBs and contains the characteristic Cys and His motifs at the core enzymatic domain. The amino- and carboxy-terminal extensions of HAUSP have no significant homology to other members of the UBP family, and make it critical for the substrate specificity for p53. HAUSP strongly stabilizes p53 even in the presence of excess Mdm2, and also induces p53-dependent cell growth repression and apoptosis. These findings imply that HAUSP might function as a tumor suppressor in vivo through the stabilization of p53. The BRCA2 protein known as tumor suppressor protein has been proposed to function in the repair of DNA double-strand breaks by homologous recombination. Mammalian cells lacking functional BRCA2 are sensitive to DNA damaging agents, show genomic instability (Patel et al., 1998), and are deficient in homology-directed DNA repair (Moynahan et al., 2001). Schoenfeld et al. (2004) showed specific interaction of USP11 with BRCA2 using an immunopurification-mass spectrometry approach. USP11 is able to deubiquitinate BRCA2 and provide the stabilization as tumor suppressor. Several proteins implicated in growth and development, including the mammalian proteins Tre-2 and Unp and the Drosophila fat facets protein, were either shown to be deubiquitinating enzymes or to have sequence similarity to such enzymes (Papa and Hochstrasser, 1993). There is increas-
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
217
ing evidence that links ubiquitination to certain pathologies. Changes in the DUB activities can contribute to diseases. Some known examples of DUBs implicated in diseases are CYLD, a tumor suppressor that prevents cylindromatosis (Bignell et al., 2000), UCH-L1 linked to Parkinson’s disease (Meray and Lansbury, 2007) and USP14 and ataxin-3 (Wilson et al., 2002) linked to ataxias. The first generation of USP and Ubl-specific protease inhibitors based on the entire Ub/Ubl protein itself modified at the C terminus with electrophilic entities capable of reacting with the active site cysteine thiol present in most Ub and Ubl specific proteases. These electrophilic traps include aldehydes (Pickart and Rose, 1986; Hershko and Rose, 1987), nitrile derivatives, Michael acceptors (including vinyl sulfone, vinyl methyl ester) and alkyl halides (Borodovsky et al., 2002). Borodovsky et al. (2005) reported the synthesis of a panel of peptide vinyl sulfones harboring various portions of the C terminus of ubiquitin and showed that depending on their length such compounds can efficiently target USPs and Ubl-specific proteases. To date only a few examples of small molecules with inhibitory potential toward DUBs have been reported (Liu et al., 2003, 2004; Verbitski et al., 2004). However these inhibitors show moderate activity and selectivity. A class of O-acyloxime derivatives of isatins was discovered as UCH-L1 inhibitors (Liu et al., 2003). Treatment of lung tumor cell lines with these inhibitors indicated that UCH-L1 activity is antiproliferative. LDN-91946 is a heteroaryl carboxylic acid for UCH-L1 inhibition. It is also a selective inhibitor for other enzymes with cysteine residues (Love et al., 2007). Identification of these new compounds that inhibit DUBs, may bring a new therapeutic approach in cancer.
4. The proteasome in cell physiology, pathology and aging 4.1. Proteasome and transcription factor degradation Posttranslational modification of proteins plays an important role in the transcriptional regulation in cells and causes, therefore, changes in gene expressions. Such modifications include phosphorylation or dephosphorylation as reversible processes (Hunter and Karin, 1992) or proteolysis as an irreversible process. Proteasomal degradation regulates numerous transcriptional factors including NFjB, p53, c-jun, b-catenin, E2F-1 and consequently activates or inactivates related gene expression. The proteasome might eliminate the nuclear transcription factors or attenuate the response gene expression or might trigger the transcriptional activation (NFjB inhibitor IjB degradation). p53: One of the targets of ubiquitin proteasome-mediated degradation is the tumor suppressor p53, which acts as a negative regulator of cell growth and is implicated in many cancers (Eliyahu et al., 1989; Levine et al., 1991). In many human tumor cells, mutant p53 accumulates preventing the cell cycle arrest and causes uncontrolled cell proliferation. The role of proteasome in p53 degradation is confirmed by studies in vitro and in vivo (Maki et al., 1996; Ciechanover et al., 1991). Degradation is shown to be accelerated by the human papilloma virus oncoprotein E6 (Scheffner et al., 1993). The E6 dependent degradation is mediated by the E3 enzyme E6-AP (E6 associated protein). In cells not transformed by HPV, ubiquitination is mediated by an unidentified species of E3 (Hershko and Ciechanover, 1998). UBC4 and E2–F1 are also suggested as specific enzymes in the ubiquitination of p53 (Rolfe et al., 1995; Ciechanover et al., 1994). It has also been reported that Mdm2 oncoprotein promotes rapid ubiquitin-targeted degradation of p53 (Haupt et al., 1997). A transient stabilization of p53 occurs following UV irradiation or DNA damage induced by other means, and leads to an increase in its level and enables the protein to cope with the damage. Concomitantly increased p53 dependent transcription of Mdm2 targets p53 for degradation and ensures it to be removed after its repair function has been fulfilled (Hartwell, 1992). p53 induces the expression of many genes, some of which, such as p21/Waf-1/Cip-1, cause cell cycle arrest; others, including Gadd-45, engage in DNA repair (Raycroft et al., 1990). The proapoptotic genes upregulated by p53 are plentiful and include APAF-1, SIAH-1, PTEN, Fas, DR5, Pmp22, Bax, Noxa, p53AIP1, and PUMA (Schuler and Green, 2001, p. 53) can also inhibit the expression of Bcl-2 (Miyashita et al., 1994) and b-catenin (Sadot et al., 2001), genes responsible for survival, proliferation, and overall disease progression in various forms of cancer. AP-1, c-jun and c-fos: Transcription factor AP-1 binds to the promotor regions in several important genes such as IL-2, IFNc (McMurray et al., 2001), IL-5 (Mori et al., 2000), CD95L (Melino et al., 2000), MMP-1 (Behren et al., 2005), G1-cyclin-dependent kinases (Hennigan and Stambrook, 2001), TGF-b (Kuo et al., 2007). AP-1 comprises a dimer of the subunits jun and fos which are degraded by UPS (Wu, 2002). Proteasomal degradation of these both proteins was confirmed by several in vitro and in vivo studies (Stancovski et al., 1995; Jariel-Encontre et al., 1997). The UPS-mediated degradation of proto-oncoprotein c-jun is controlled by the d domain, a 27 amino acid segment in its structure. The d domain is a cis-acting signal required for ubiquitination and subsequent degradation for c-jun. Since v-jun, which is the transforming retroviral counterpart of c-jun, lacks the d domain it can not be ubiquitinated. Phosphorylation by mitogen-activated protein kinases (MAPKs), such a Jun kinase 1 (JNK1), reduces ubiquitination and increases jun stability. In addition it was also shown that ubiquitination may not be necessary for the degradation of c-jun (Jariel-Encontre et al., 1995). Degradation of c-fos is shown to be stimulated by cjun. Similar to v-jun, v-fos is also not degraded by UPS (Stancovski et al., 1995). NFjB: The transcription factor NFjB regulates a number of genes encoding cytokines, chemokines, growth factors, cell adhesion molecules, surface receptors and acute phase proteins (Baeuerle and Henkel, 1994). It is identified as nuclear transcription factor binding to the B site of the intronic promoter of the Ig j chain (Sen and Baltimore, 2006). In unstimulated cells, NFjB is sequestered in the cytoplasm as a complex consisting of DNA-binding subunits p50 and p65 and bound to its inhibitor, IjB. It can be p50 or p65 homodimers or p50/p65 heterodimers. When cells are stimulated (by cytokines, stress, or
218
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
Fig. 20. NFjB-degradation by proteasome and gene expression. Following the IjB phosphorylation by IjB kinase (IKK) and degradation by proteasome, p50 and p65 dimers are released to the nucleus and activate gene expression (IL, interleukin; TNF, tumor necrosis factor; VCAM, vascular cell adhesion molecule; ICAM, intercellular adhesion molecule).
chemotherapy), signaling cascades are triggered that lead to activation of IjB kinase, a heterodimeric protein kinase that catalyzes IjB phosphorylation. IjB kinase phosphorylates two serine residues Ser32 and Ser36 in the amino-terminal regulatory domain of IjB (Wu, 2002). Mutations in these serine residues results in the prevention of phosphorylation. The phosphorylated sites on IjB are then recognized by E3RS (IjB/b-TrCP), an SCF-type E3 ubiquitin ligase, leading to ubiquitination. IjB is then degraded by the proteasomal pathway, releasing free active p50–p65 dimer, which translocates to the nucleus and binds to promoter regions of several target genes, thereby triggering their transcription (Fig. 20) (Brown et al., 1995). The p50 activation process is also controlled by the UPS. p50 is the product of a large cytoplasmic precursor called p105. Whereas the amino terminal of p105 constitutes inactivated p50, the carboxy-terminal contains a number of closely adjacent sequence repeats called ankyrin repeats required for the inhibition of p50 DNA binding. It is known that p50 requires a proteolytic event to be released from the carboxy-terminus. Studies revealed that processing of the p105 requires ubiquitin, E2– F1 and a novel species of E3. p105 is the first known molecule to be processed by the ubiquitin system rather than being completely destroyed. Besides the degradation of p105 by UPS, the p50 subunit is known to be generated by novel cotranslational biogenesis requiring the 26S proteasome (Fan and Maniatis, 1991). b-Catenin: b-Catenin is a subunit of cadherin protein complex. It is found in the plasma membrane, cytoplasm and nucleus, known to function as an oncogene and be produced in higher amounts in patients with basal cell carcinoma leading to the increase in proliferation of related tumors. When it is not bound to cadherin, it can interact with several different proteins inside the cell such as axin. As a complex with axin, b-catenin is phosphorylated by glycogen synthase kinase-3 (GSK-3) that creates a signal for the rapid degradation of b-catenin by proteasomes (Miller and Moon, 1996). GSK-3 phosphorylates b-catenin in three conserved Ser residues and one Thr residue in the N-terminal domain. Mutations in these residues resulted in increased stability and increased activity of the protein (Morin et al., 1997). Also various signals can inhibit the phosphorylation, allowing b-catenin to the nucleus, to interact with transcription factors and regulate gene expression. E2F-1: The E2F family plays important role in regulating G1-S transition in the cell-cycle progression. E2F-1 is one of the best studied proteins of the E2F group. E2F-1 is known as an oncogene and inducer of apoptosis in different conditions. E2F-1 is degraded by UPS (Hofmann et al., 1996). Retinoblastoma tumor suppressor (Rb) protein binds to E2F-1 and prevents ubiquitinylation and results in a marked stabilization. Others: There are so many other transcription factors which are eliminated by proteasome such as Smad1 and 2 that activates TGF-b responsive genes (Zhang et al., 2001; Lo and Massague, 1999), c-Myc modulates the gene expression of FasL and CXCR4 (Kasibhatla et al., 2000; Hasegawa et al., 2001), OBF-1 plays major role in the formation of the germinal center of sec-
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
219
Table 3 Transcription factors regulated by the proteasome. Proteins
Functions and the related genes
b-Catenin c-jun, c-fos c-Myb c-Myc E2F-1 HIF-1a IRF-1 NFjB OBF-1 p53 Smad 1 Smad 2 STAT 3 STAT 4 STAT 5b
Oncogene Protooncogene CD4 gene expression FasL and CXCR4 gene expression Cell cycle progression Gene expression during hypoxia IFN and IFN-inducible genes Cytokines, chemokines, growth factors, cell adhesion Formation of germinal center of secondary lymphoid organs Cell growth TGF-b responses Cytokine stimuli
CD4, is a co-receptor that assists the T-cell receptor to activate its T-cell following an interaction with an antigen presenting cell; FasL (Fas Ligand), is a type II transmembrane protein that belongs to the tumor necrosis factor family. The binding of FasL with its receptor induces apoptosis; CXCR4, a CXC chemokine receptor specific for stromal derived factor-1 which is a molecule endowed with chemotoxic activity for lymphocytes; IFN, interferon; TGF-b, acts as an antiproliferative factor in normal epithelial cells and at early stages of oncogenesis.
ondary lymphoid organs (Tiedt et al., 2001), STAT 3, 4 and 5b respond to many cytokine stimuli and activate cytokine responsive genes (Daino et al., 2000; Wang et al., 2001; Gebert et al., 1999), c-Myb enhances of the pre-T-cell receptor a chain gene and plays positive and negative roles in CD4 gene expression (Feikova et al., 2000), HIF-1a induces target genes during hypoxia (Kallio et al., 1999), and IRF-1 activates IFN genes and IFN-inducible genes (Nakagawa and Yokosawa, 2000). Some of those factors are listed in Table 3. 4.2. The ‘‘quality control’’ in protein folding and endoplasmic-reticulum-associated degradation (ERAD) One of the major tasks of the proteasome is the so-called ‘‘quality control’’ of newly synthesized proteins. This is best investigated in the ER. Proteins are synthesized and enter the endoplasmic reticulum (ER) for further processing in an unfolded state, mediated by the Sec61p complex (Sanders et al., 1992; Wirth et al., 2003). After entering the ER, proper folding is mediated by a complex system of chaperones, in order to reach the functional three dimensional structure of the protein. About 7500 different proteins (Chen et al., 2005) are folded in the ER and about 30–80% of the newly synthesized proteins are misfolded (Rivett and Hearn, 2004) and have to be selectively recognized and degraded in order to prevent the accumulation of non-functionally proteins and to recycle their amino acids. This is realized by a complex system of chaperones and lectines able to recognize improper folded intermediates (Hinault et al., 2006; Leidhold and Voos, 2007). These targets can be labeled for ubiquitin-mediated proteasomal degradation. Since the UPS is not located in the ER-lumen, ER-resident proteins targeted for degradation have to be retrotranslocated into the cytosol. The access of the UPS to transmembrane proteins (‘‘ERAD-M’’ class) is still unclear, but some models have been developed from a small number of studied proteins (Sayeed and Ng, 2005; Carvalho et al., 2006). This whole process is termed as ER quality control (Ellgaard and Helenius, 2003). How important ERAD for the functional integrity of the cell is, becomes clear by the fact that more and more ERAD substrates are known to be linked to human diseases (Aridor, 2007) and an accumulation of non-degraded substrates triggers apoptosis, known as the ‘‘unfolded protein response’’ (UPR) (Ron and Walter, 2007). In the following the mechanisms and models of ERAD will be discussed resulting in the final step of substrate-polyubiquitination (Eisele et al., 2006; McCracken and Brodsky, 2005). Soluble proteins that are recognized as misfolded by chaperones are kept bound to their chaperones thus remaining soluble. Misfolded/unfolded proteins tend to form insoluble aggregates driven by their exposed hydrophobic structures, but retranslocation can only be ensured with soluble substrates. Chaperones like the heat shock proteins BiP (also termed Grp78, an ER homologue of Hsp70) (Oida et al., 2008; Awad et al., 2008; Kimura et al., 2008) and its co-chaperone Hsp40 turned out to prevent aggregate-formations of attached proteins (Nishikawa et al., 2001). In the case of glycoproteins the ‘‘quality control’’ machinery uses glucosidases (Gls) I and II, UDP-glucose:glycoprotein glucosyltransferase (UGT) and ER-mannosidase I (ER-Man I) (Mancini et al., 2003) (Fig. 21). During ER protein synthesis or shortly after translocation of the nascent protein into the ER lumen those proteins are fit out with an N-linked oligosaccharide (NAc-Gln2-Man9-Glc3). Translocation starts with binding of the protein to a specific channel, while all cytosolic chaperones detach during this process. The unfolded protein enters the channel driven by stochastic Brownian motion, while binding to BiP, an essential chaperone molecule for protein ER luminal folding, in an ATP-dependent way on the ER luminal sites of the translocated peptide. This reduces the degrees of freedom resulting in a net forward movement (Rapoport, 2007). BiP recognizes and binds the exposed hydrophobic sequences of unfolded protein structures. Two of the three terminal glucose residues will be cut off by Gls I and II, resulting in a substrate for calnexin and calreticulin binding of these monoglucosylated oligosaccharide–proteins. Calnexin is a 90 kDa membrane bound protein of the ER, calreticulin a soluble 60 kDa
220
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
Fig. 21. The endoplasmic reticulum (ER) quality control. Nascent proteins are translocated into the endoplasmic reticulum (ER) via the so called ‘‘translocon’’ either co- or post-translational. In the co-translational way, the nascent protein binds to the protein conducing channel, a heterotrimer containing three to four copies of Sec61 (a heterotrimer of an a-, b- and c-subunit). In mammalians almost all proteins are transported into the ER via this pathway (Meyer et al., 2000). ER-transport of the nascent protein in the post-translational pathway is mediated by the Sec61–Sec62–Sec63 complex, while binding to the channel is realized by a singnal sequence of the nascent protein (Meyer et al., 2000). During transport into the ER, the nascent protein is manno- and glycolsylated. Two of the glucose-residues are quickly removed by the glucosidases I and II (GI and GII) and the complex binds to ERp57. This complex binds to the cytosolic calreticulin (Crt) and subsequently to the ER-membrane-bound calnexin (Cnx), for subsequent steps of folding, assembly and posttranslational modification. After GII-mediated removing of the last remaining glucose-residue, proper folded proteins are recognized by ER a-mannosidase II, de-mannosylated and excreted to the Golgi apparatus via a transport-process involving the proteins ERGIC53, ERGL, and VIP36. If the protein is still misfolded, it is re-glucosylated by UDP-glucose:glycoprotein glucosyltransferase (UGT) and re-enters the calnexin–calreticulin-cycle until it is proper folded and recognized by ER a-mannosidase II. If the protein does not reach a proper state of folging, it is de-mannosylated by ER a-mannosidase I, and recognized by the so-called EDEMS, that initiate a BiP-binding-mediated transport of the terminally misfolded proteins into the cytosol, where they are degraded by the UPS (for details of the ERAD-mechanism, see Fig. 22).
protein found in the ER lumen; furthermore the associated oxidoreductase ERp57 (also termed as ‘‘GRp58’’ that has shortly been proven to be identical to the 1,25D3-MARRS receptor (Khanal and Nemere, 2007)) is involved, binding the substrate protein together with calnexin (Ellgaard et al., 1999; Trombetta and Parodi, 2003; Ellgaard and Frickel, 2003). Binding to the so-called calnexin/calreticulin cycle provides time for the protein to reach its correct folding state. Only after removing of the last glucose residue by Gls II the protein is released from the calnexin/calreticulin cycle (Roth et al., 2008). Normally, the now correctly folded native protein is exported to the secretory pathway by a mechanism containing the proteins ERGIC53, ERGL and VIP36 (Fig. 21). If the protein is not correctly folded, it enters the calnexin/calreticulin cycle by de-(Gls II) and re-glycosylation (UGT) and is kept there until is reaches native folding in a process called ‘‘retention’’ (Olivari et al., 2006). If native folding is reached, it returns to the normal pathway by Gls II deglycosylation and export to the secretory pathway. This is termed as the ‘‘first phase of ER retention’’. Several studies have been performed using knockout mice or animals overexpressing key-proteins of ER luminal protein folding machinery and the process of ERAD. GRp58/ERp57 knockout mice showed embryonic lethality (Garbi et al., 2006) and calreticulin deficiency was manifested in imperfect cardiac development, being lethal, too (Guo et al., 2001; Michalak et al., 2002). Unexpectedly calnexin deficient mice are viable, but about 50% of the calnexin/ animals die within the first two days after birth, the survivors are smaller and show motoric disorders (Denzel et al., 2002). An overexpression of BiP, calnexin and Grp170 (also termed ‘‘ORP150’’) (Gao et al., 2008) resulted in a reduction of b-amyloid, a protein playing an important role in the pathology of Alzheimer’s disease. Further studies suggested an inhibition of APP maturation by BiP caused by retention of the protein in the ER (Hoshino et al., 2007). BiP has been shown to protect tumor cells by calciumbinding under glucose restriction or severe hyperoxia and thus preventing the cell from apoptotic cascades initiated by Ca2+-efflux from the ER. Moreover BiP inhibits pro-apoptotic factors like BIK (Zhu et al., 2005; Li et al., 2008) and BAX (Hanke, 2000; Su et al., 1997) and suppresses the cleavage of procaspase-7 and -12 by forming complexes with these proteins (Fu et al., 2007; Rao et al., 2002; Reddy et al., 2003). Moreover the involvement of different ER chaperones has been shown in
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
221
Fig. 22. The retrotranlocation of misfolded proteins and proteasomal degradation. After de-mannosylation misfolded proteins are recognized by a protein complex and transported into the cytosol (ERAD-L). Misfolded membrane proteins and proteins with cytosolic domains are also using some of these proteins to be transported into the cytosol (ERAD-M and ERAD-L, respectively). Within this complex several proteins of the ubiqutination cascade are present. For detailed function of individual proteins see the text. Ubiquitination is taking place on the cytosolic side, where also the proteasomal degradation occurs.
several pathologic processes like neurodegenerative diseases, cancer progression, atherosclerosis, diabetes and viral/bacterial infections (Ni and Lee, 2007). The attempts of proper protein folding in the mentioned calnexin/calreticulin cycle can be terminated by ER-Man I by substrate de-mannosylation and proteins of the EDEM family, recognizing these de-mannosylated products, resulting in the ‘‘second phase of ER retention’’ and retrotranslocation into the cytosol, where the misfolded proteins are recognized and degraded by the UPS (Fig. 22). It has been shown that specific inhibition of ER-Man I by the alkaloid kifunensine (Liu et al., 1999; Vallee et al., 2000), a specific inhibitor of a1,2-mannosidases, prevents the hydrolysis of the Man9-structure, slowing down both release and degradation of terminally misfolded glycoproteins from the calnexin/calreticulin cycle (Molinari et al., 2002). ER-Man I de-mannosylates the substrate proteins independent of their folding state. But the fact that the enzyme has a slow kinetic, suggests that it may serve as a ‘‘timer’’ terminating futile attempts of proper protein folding. The EDEM-family (for ‘‘ER degradation enhancing a-mannosidase-like protein’’ in yeast) contains until now three known members, EDEM1 (Olivari et al., 2006; Zuber et al., 2007; Cali et al., 2008), EDEM2 (Mast et al., 2005) and EDEM3 (Hirao et al., 2006). These proteins show a high homology (about 35% of the sequence, conserving all catalytically active amino acids without differences in their location (Hosokawa et al., 2001; Karaveg and Moremen, 2005; Karaveg et al., 2005)) to ERMan I. An overexpression of EDEM1 and EDEM3 enhanced the degradation of misfolded proteins from the calnexin/calreticulin cycle by accelerating de-mannosylation, while RNA interference decreases the disposal of misfolded glycoproteins (Yoshida et al., 2003) by preventing correct folding and reducing the amount of proteins exported by the secretory pathway (Eriksson et al., 2004). EDEM1 is found as soluble and membrane bound ER protein (Olivari et al., 2005), able to interact with calnexin (Oda et al., 2003) in a process called ‘‘dislocation’’. The human forms of EDEM1 are named Htm1p/Mn11p. EDEM2, a soluble member of the family, has been shown to be transcriptional up-regulated by the stress-activated transcription factor Xbp1 (IRE1a-activated (Plongthongkum et al., 2007; Schroder and Kaufman, 2006)). Like all EDEMs (Olivari et al., 2005), EDEM2 is enhancing ERAD of misfolded proteins by extracting them from the calnexin/calreticulin cycle. The degradation rate of non-glycosylated misfolded proteins is not affected by EDEM regulation (Olivari et al., 2005). How exactly the mechanism works, enabling EDEM to induce degradation of defective proteins is still unclear. The EDEMs can be induced by UPR to reduce the accumulation of potentially harmful misfolded proteins.
222
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
Other proteins involved in the ERAD, despite of N-linked-glycan-trimming, are BiP, recognizing hydrophobic sequences, or oxidoreductases binding free thiol-residues, protein disulfide isomerases (PDIs) (Appenzeller-Herzog and Ellgaard, 2008; Nuss et al., 2008; Otsu et al., 2006), redox driven chaperones, and Eps1p (Kimura et al., 2005; Xiao et al., 2004), targeting misfolded membrane proteins for proteasomal degradation. Since the exact process of retrotranslocation of misfolded proteins into the cytosol is still under investigation, in the following some of the suggested mechanisms will be described. The first one involves Sec61, supposing a bidirectional transport of proteins both ER-importing and -exporting them. Though Sec61-mediated transport requires unfolding of the respective protein, an idea supported by results showing the reduction and unfolding of proteins by PDI while being prepared for dislocation (Tsai et al., 2001). In contrast some ERAD substrates have been found that are transferred to the cytosol in a folded state. Another mechanism suggests an export channel formed by the mammalian proteins Derlin-1 (the yeast homologue is termed Der1p) and the AAA ATPase p97 (also termed as Cdc48p or VCP, for valosin-containing protein) (Woodman, 2003; Halawani and Latterich, 2006) in the ER (Sun et al., 2006). Both proteins are found in complex with each other (Wang et al., 2006), ER-membrane associated (Latterich et al., 1995) and in the cytosolic volume (Ghislain et al., 1996; Hitchcock et al., 2001), involved in the degradation of some (but not all) ERAD substrates. The driving foce of the translocation process is considered to be provided by ATP-hydrolysis of p97 (Raasi and Wolf, 2007), while substrate binding is mainly due to recognition of ubiquitin-binding domains of p97. For example, Derlin-1 has been found interacting with US11 peptide-labeled MHC I molecules (Ye et al., 2004). Other hypotheses include the proteins Hrd1p (Bazirgan et al., 2006; Gauss et al., 2006) and Doa10p, two ER located transmembrane proteins that could be able to form pores, enabling substrate dislocation into the cytosol and thus to the UPS. One third mechanism suggests an involvement of the AAA-ATPases (Zwickl and Baumeister, 1999) of the proteasomal 19S regulator particle in the dislocation of misfolded proteins from the ER. Meanwhile another cytosolic protein complex have been found, containing an AAA-ATPase: the mammalian Cdc48p (also termed as p97 or VCP) (Goder et al., 2008; Shcherbik and Haines, 2007; Wilson et al., 2006) may be a key protein in the dislocation process. In complex with two other proteins, Npl4 and Ufd1 (Shcherbik and Haines, 2007; Lass et al., 2008; Cao et al., 2007) that bind ubiquitinated proteins, this complex (Cdc48p-Npl4Ufd1) (Heubes and Stemmann, 2007; Nowis et al., 2006; Alzayady et al., 2005; Cao and Zheng, 2004) may be responsible for the breakdown of large protein complexes before proteolytic degradation of its elements. Experiments revealed that the process of translocation from ER by Cdc48p-Npl4-Ufd1 is ATP dependent; a comparable function has been attributed to chaperones of the Hsp70 family in importing newly synthesized proteins into the ER and mitochondria (Hood et al., 2003). Large amounts of Cdc48p-Npl4-Ufd1 have been found interacting with Derlin-1 and the ER-protein E3gp78, furthermore it interacts with the yeast Ub-chain elongation factor Ufd2p (Tu et al., 2007; Mahoney et al., 2002), until now the first identified E4 protein, showing a structure similar to the ‘‘really interesting new gene’’ (RING) (Itahana et al., 2007; Kedar et al., 2004) domain that is found in several ubiquitin ligases. Other interactions of Cdc48p-Npl4-Ufd1 include Rad23p and Dsk2p (Medicherla et al., 2004; Ohno et al., 2005), while Dsk2p contains a ubiquitin-associated (UBA) domain, able to recognized ubiquitin tags. Both proteins have been found in certain pathways involved in the transfer of ubiquitinated substrate proteins to the UPS. Most recently the so-called ‘‘escort pathway’’ is intensively discussed: a model substrate protein (Spt23p-p90) (Bhattacharya et al., 2008; Shcherbik et al., 2004) with a short chain of ubiquitin molecules (1–3) attached, binds to the Cdc48p-Npl4Ufd1 complex. The short Ub-chain is prolongated by Ufd2p and Spt23p-p90 is transferred to Rad23p, respectively Dsk2p that ‘‘prefer’’ the binding of longer Ub-chains, while Rad23p shows interactions with the glycan-removing protein Png1p (Chantret et al., 2003), before the substrate is delivered to the UPS for terminal degradation. In this model the substrates follow the gradient of ubiquitin-affinity until degradation. Therefore, ERAD contains a first ubiquitin independent and a second ubiquitin dependent step. In the first step substrate proteins are dislocated from the ER to the cytosol. Here lysine residues must be exposed as targets for ubiquitin attachment and the proteins are polyubiquitinated. Transmembrane proteins may undergo another mechanism of dislocation, while the first step is Ub-attachment. The type of protein and the location of the misfolded sequence determine the pathway of degradation. Until now in yeast the pathways ERAD-L (proteins showing defects in their luminal domain) (Willer et al., 2008), ERAD-M (for membrane bound proteins, showing defects in their transmembrane domain) (Bazirgan et al., 2006; Bordallo et al., 1998; Boisrame et al., 2006; Gardner et al., 2000; Mueller et al., 2006) and ERAD-C (for nuclear and cytosolic proteins) (Neuber et al., 2005; Ravid et al., 2006) are identified. Currently/temporarely exist several models of ERAD substrate recognition, ubiquitination and degradation: different complexes bind, ubiquitinate and retrotranslocate the substrates of the ERAD-C, -M or -L pathways. In yeast, the ERAD-C substrates are recognized by cytosolic Hsp70 (Ssa1 (Needham and Masison, 2008; Zhang et al., 2006; Li et al., 2006; Hainzl et al., 2004)) and Hsp40 (Ydj1 (Mandal et al., 2008; Tutar and Tutar, 2008; Wright et al., 2007)) and Hlj1 (Tsai et al., 2006; ncevska-Taneva et al., 2006) chaperones. This is exclusively specific for ERAD-C, and the membrane bound Doa10 (TEB4 in mammals) E3-ligase (Ravid et al., 2006; Kreft et al., 2006), ubiquitinating the substrate protein. The co-factors of Doa10 are the E2-enzymes Ubc6 (Oh et al., 2006; Botero et al., 2002; Lenk et al., 2002) (Ubc6e in mammals), and the E2-complex Ubc7-Cue1, delivering the ubiquitin, and the protein Ubx2 (Wilson et al., 2006; Neuber et al., 2005; Schuberth and Buchberger, 2005). Ubx2 and Ubc7 (Bazirgan and Hampton, 2008; Arai et al., 2006) (same name for mammals) are both membrane bound and associate Cdc48 to the ER-membrane (Nakatsukasa and Brodsky, 2008). After this, the substrate is transported to the cytosol for 26S proteasome-mediated degradation. This transport is performed by a complex containing Cdc48, Npl4 and Udf1 (in mammals the according complex is p97-UFD1-NPL4) that binds the ubiquitinated substrate; pro-
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
223
longation of the attached ubiquitin is performed by Ufd2p, an E4-enzyme, responsible for polyubiquitination of the substrate protein, followed by degradation by the UPS. Cdc48 (p97 in mammals) is an AAA-ATPase, that is involved in the ERAD-process after ubiquitination, but before degradation of the substrate, and that extracts ubiquitinated substrates into the cytosol (Nakatsukasa and Brodsky, 2008). Both the ERAD-M and ERAD-L pathway depend on the same mediators: the substrate is retrotranslocated by a complex formed of Cdc48, Ufd1, Npl4, and the membrane bound Ubx2. That whole complex is associated to (among other proteins) the E2-complex Ubc7-Cue1, the E3-ubiquitin ligases Hrd1 (Omura et al., 2008, 2006; Yang et al., 2007; Kikkert et al., 2004) (HRD1 in mammals) and Hrd3 (Gardner et al., 2000; Mueller et al., 2006; Plemper et al., 1999) (SEL1L in mammals) and again the ‘‘transport complex’’ Cdc48-Npl4-Ufd1. After ubiquitination of the substrate it is released into the cytosol while bound to the ‘‘transport complex’’, polyubiquitinated and degraded by the 26S proteasome. Membrane bound proteins seem to be solubilised in an ATP-dependent and Cdc48-dependent manner (Nakatsukasa et al., 2008). This model of the ERAD-C, -M and -L pathways were described according to Nakatsukasa and Brodsky (2008). 4.3. The role of the proteasome in antigen processing Immunoproteasomes are involved in the generations of small antigenic peptides having a length between 8 and 10 amino acids that are presented on the cell surface by MHC-I molecules. Both in vivo and in vitro it has been shown that the generation of those small oligopeptides is largely increased in the presence of i20S (Deol et al., 2007; Chapiro et al., 2006; Van den Eynde and Morel, 2001). If this is to be caused by an enhanced substrate degradation or the changed specificity in substrate cleavage is still under investigation. In contrast to the c20S the i20S favors cleavage after hydrophobic residues that are fulfilling later the function of an anchor while being presented as antigen on MHC-I. The C-termini of the presented antigens are thus clearly formed by proteasomal degradation, while the N-termini are usually trimmed by two or three amino acid residues by cytosolic peptidases (Kloetzel, 2004; Kloetzel and Ossendorp, 2004; Strehl et al., 2005; Tenzer et al., 2005). The rate of product release by i20S have been proven in several studies to correlate significantly with the rate of antigenic peptide formation (Deol et al., 2007; Sijts et al., 2000), and substrate degradation by i20S has been shown in several experiments to be faster than by the c20S (Strehl et al., 2008), pointing both in the direction of changed substrate specificities and an increased cleavage rate. This might be due to slight structural changes in the i20S particle induced by binding of the PA28 regulator particle, however without affecting the active sites (Textoris-Taube et al., 2007), an effect termed as ‘‘dual cleavage’’, enabling the quick generation of large amount of MHC-I presentable oligopeptides. The processing of an intracellular antigen from the native protein to the trimmed short antigenic oligopeptides is a very complex mechanism. The presentation of antigens on the cellular surface is an essential part of the friend and foe recognition of the mammalian immune system. The proteasomal degradation of cellular proteins is the first step: the nascent protein pool is one of the main sources of MHC I-presented antigens, while about 1% of proteasomal released oligopeptides are further processed for this purpose (Lehner, 2003; Reits et al., 2003). Due to the fact that about 30–80% of all nascent proteins turn out to be defective ribosomal products (DRiPs), this is a considerable pool, quickly providing antigenic oligopeptides for cellular proteins (Rivett and Hearn, 2004). These defective proteins enable the presentation of protein parts as antigens even for proteins that show stability for many days or weeks in their native form within 30 minutes. This is important since after viral infections the de novo synthesized viruses can be released after a few hours, so the immune response must follow quickly (Yewdell et al., 2003). Though, it was shown that only a single oligopeptide is bound to a MHC-I molecule for every 10,000 misfolded, defective, damaged or unwanted proteins degraded (Yewdell, 2001). Moreover it is still not clear how exactly these presentable antigens are not completely degraded before reaching the endoplasmic reticulum, since an oligopeptide of 9 amino acids has a half-life of about 7 s, while it needs about 6 s to diffuse across the cellular diameter (Reits et al., 2003) and a chaperone stabilization of these fragments could not be demonstrated yet. Other studies demonstrated that peptides, that are stabilized by the binding of heat shock proteins become at a far higher rate presented by MHC-I molecules and that inhibition of the heat shock proteins Hsp70 and Hsp90 by deoxyspergualin almost completely inhibit MHC-I antigen presentation (Binder et al., 2001), showing an important function of these chaperones in the presentation of antigens. The amount of antigenic fragments is significantly increased in cells exposed to INF-c, changing the proteasomal system in a way that results in an enlarged formation of ‘‘presentable’’ antigens. IFN-c, a cytokine released by T-helper cells, CD4+, CD8+ lymphocytes and natural killer cells, after forming a homodimer induces the cellular antiviral response by binding to the interferon gamma receptor 1 (IFN-cR1), that activates IFN-cR2, and the Janus kinases Jak1 and Jak2, initiating the dimerization of the phosphor-signal transducer and transcription-activator-1 (STAT-1), that are translocated into the nucleus, where the genes, that are thus induced by IFN-c are upregulated (Rosenzweig and Holland, 2005). This enhances the formation of both immunoproteasomes and the PA28 proteasome regulator particle, accumulating near the endoplasmic reticulum (Brooks et al., 2000). While the levels of immunoproteasomes are strongly increased, their distribution instead seems not to change. Possibly the close range to the TAP (TAP = transporter associated with antigen processing) transporter enhanced the transport of antigenic oligopeptides into the endoplasmic reticulum (Rivett and Hearn, 2004) (Fig. 23). Since native and correctly folded viral proteins can not be degraded by c20S or i20S, even if bound to a PA28 regulator cap, the proteins have to be polyubiquitinated first (Fig. 23). Only polyubiquitinated (viral) proteins or protein-pathogens can be recognized, unfolded, and degraded in an ATP-dependent manner by both 26S and hybrid proteasomes (PA28-20S–19S) (see Chapter 3.7). 26S and hybrid proteasomes have revealed different sets of products but approximately the same rates of
224
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
Fig. 23. Proteasomal contribution to antigen processing. The current model of the proteasomal contribution to antigen presentation starts with a viral infection and infiltration of the cell by viral DNA (or by invading protein–pathogens). The native and correctly folded viral proteins have to be polyubiquitinated to be recognized and degraded by the proteasome in an ATP-dependent manner, mediated by both 26S and the so-called ‘‘hybrid proteasomes’’. The N-termini of the released antigenic oligopeptides are further trimmed by cytosolic proteases like tripeptidyl peptidase II (TPP II) or the puromycin-sensitive aminopeptidase (PSA). The antigens are now transported into the endoplasmic reticulum (ER) by the ‘‘transporter associated with antigen processing’’ (TAP). At the same time in the ER MHC-I proteins are post processed by calnexin (Cnx) mediated attachment of b2-microglobulines (b2m). The mature but unloaded MHC-I molecule binds ERp57 and calreticulin (Crn). The present antigenic oligopeptides are coupled onto MHC-I in a complex interaction of the completely assembled MHC-I, tapasin, calreticulin, ERp57 and the TAP transporter itself. The loaded MHC-I is first transferred into the Golgi-apparatus and from there via the secretory pathway to the surface of the cell, where they can be presented to cytotoxic T-cells (CD8+) of the immune systems. These cells recognize presented antigens and are stimulated to proliferate and initiate destruction of the infected cell if necessary (according to Groothuis et al. (2005)).
protein degradation (Cascio et al., 2002); though 26S seems to be downregulated by IFN-c, possibly by an induced decrease of a7-phosphorylation of the core proteasome – a process not essential for 26S formation, but with a significant influence on it (Bose et al., 2004). Such regulator-capped proteasomes may contain different amounts and compositions of inducible i20Ssubunits, that may enhance antigen-presentation, but have turned out to be not essential for (Rivett and Hearn, 2004). The released fragments may be further degraded by ATP-independent immunoproteasomes, or even from the PA28-capped side after processing by the 19S-capped one of the same proteasome in an ATP-dependent way. In a last step the short oligopeptides are N-terminal trimmed by cytosolic peptidases like tripeptidyl peptidase II (TPP II) (Kohan et al., 2005) or the puromycin-sensitive aminopeptidase (PSA) (Cascio et al., 2001; Levy et al., 2002). Other cytosolic peptidases involved might be the IFN-c inducible leucine aminopeptidase (Beninga et al., 1998), the thimet oligopeptidase (TOP), that is found in many animals and plants, acting on oligopeptides but not proteins (York et al., 2003; Sigman et al., 2003), neurolysin (Oliveira et al., 2003) or the bleomycin hydrolase (Stoltze et al., 2000). The post-processed antigenic peptides are then incorporated into the endoplasmic reticulum by the ‘‘transporter associated with antigen processing’’ (TAP) (Tenzer et al., 2005) in an ATPhydrolysing manner (Fig. 23). The transmembrane complex TAP is composed of two different subunits, TAP1 (748 amino acids in human) and TAP2 (686 amino acids in human), key proteins in antigen presentation. The TAP1–TAP2 dimer is sta-
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
225
bilized by the membrane bound glycoprotein tapasin, bridging the MHC I complex and TAP (Tan et al., 2002; Lehner et al., 1998). The genes coding the human forms of TAP1 and TAP2 are found on chromosome 6p21, closely to the MHC-II locus and the two inducible proteasomal subunits b1i and b5i. The expression of all four genes is inducible by IFN-c, suggesting a connected function both in processing and presentation of antigens (Herget and Tampe, 2007). TAP significantly prefers oligopeptides with C-terminal hydrophobic or basic residues over acidic ones (Rivett and Hearn, 2004) and the preferred C-termini are both the preferred anchor structures for antigen binding to MHC-I and the mainly generated C-termini by immunoproteasomal degradation of proteins. The main TAP-substrate are peptides of 12 amino acids, peptides with 8–16 amino acids are transported, too, but some with significantly lower affinities (Reits et al., 2003). Other ER resident aminopeptidases have been identified, that play a similar role in antigen processing like the cytosolic ones: the ‘‘ER aminopeptidase associated with antigen processing’’ (ERAAP) (Serwold et al., 2002) and the ‘‘ER aminopeptidase 1’’ and ‘‘2’’ (ERAP1 and ERAP2) (Evnouchidou et al., 2008; Fruci et al., 2008; York et al., 2006). ERAP1 processes also the N-terminal ends and is upregulated by IFN-c; it shows large substrate specificity but prefers peptides containing P10 amino acids, while showing a decreased affinity to chains of 8 amino acids, the ideal length for MHC-I presentation. Attachment of the prepared antigens to MHC-I in the endoplasmic reticulum is performed by a large complex of TAP, ERp57 (four polypeptides), unloaded MHC-I (four polypeptides), tapasin (four polypeptides), calreticulin (four polypeptides) and the antigen itself as a critical part of this complex (Ortmann et al., 1994; Sadasivan et al., 1996; Herget and Tampe, 2007): this is named the MHC class-I loading complex. Both tapasin and calreticulin play an important role in stabilizing this whole complex, while calreticulin is involved in proper antigen loading of MHC-I. The loaded MHC-I protein is exported into the Golgi-apparatus and transported via the common secretory pathway to the cells surface, where the presented antigens are inspected by the immune system (Fig. 23). Antigens, that are not recognized to be bodies own by thymus born cytotoxic T-cells, stimulate such a T-cell clone to proliferate and to destroy the infected cell. Activated CD8+ cells release the antiviral cytokines IFN-c and TNF-a the two main peptides inducing the expression of inducible proteasomal subunits in order to stimulate the antigen presentation of the surrounding tissue. Other cytokines released are the macrophage inflammatory protein-1a and -1b (MIP-1a, respectively -1b) suppressing viral production (Lieberman et al., 2001), followed by lysis of the infected target cell induced by secretion of perforin and granzymes. Considering the link between viral infection of a cell and antigen presentation, an effective strategy of viruses has become the targeting of the proteasomal systems or proteins, involved in the presentation of antigens. For example, the human papillomavirus virus (HPV) is able to downregulate the transcription of b1i, TAP1 and MHC (Georgopoulos et al., 2000). The herpes simplex viruses types 1 and 2 are able to interfere with TAP function by expressing a protein (ICP47, expressed in the immediate early phase (Stone et al., 2006) that blocks TAP by binding at the cytosolic side of the transporter (Aisenbrey et al., 2006; Furukawa et al., 2000), while the human cytomegalovirus expresses the protein US6 (Kim et al., 2008; Pande et al., 2005), blocking TAP from the ER-lumeninal side. A recently discovered strategy is used by the bovine and equine herpes virus 1 (BHV-1, respectively EHV-1): these viruses express a protein (UL49.5, also termed as ‘‘glycoprotein gN’’) that arrests TAP in a dysfunctional conformation and induces degradation of TAP by the proteasome (Koppers-Lalic et al., 2005; Koppers-Lalic et al., 2008) (see Chapter 3.7). 4.4. The degradation of damaged proteins – a function of the 20S proteasome Protein synthesis, environmental influences on the protein pool or the cellular metabolism itself are leading to a constant formation of damaged or unfolded proteins. The malfunction of the protein synthesis and folding machineries, which is leading to simply non-native folded proteins with normal amino acids, the damage of proteins with the formation of posttranslational modified amino acids and misfolded protein as a result of that, are the major sources of unfolded proteins. Protein Table 4 Reactive oxygen and nitrogen species important in biological systems. Radicals Reactive oxygen species (ROS) Alkoxyl Peroxyl Hydroxyl Superoxide Reactive nitrogen species (RNS) Nitrogen dioxide Nitrogen oxide
Non-radicals RO RO 2 OH O2
Hydrochlorous acid Hypobromous acid Hydrogen peroxide Singulet oxygen Hydroperoxides
HOCl HOBr H2O2 O2 ROOH
NO2 NO
Alkyl peroxynitrites Dinitrogen tetroxide Dinitrogen trioxide Nitronium cation Nitrosyl anion Nitrosyl cation Nitrous acid Peroxynitrite Peroxynitrous acid
ROONO N2O4 N2O3 NOþ 2 NO NO+ HNO2 ONOO ONOOH
226
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
damage with the formation of modified amino acids might be induced by numerous agents, including oxidants, sugars and xenobiotics. The formation of oxidatively modified proteins is one of the major consequences of aerobic life and the cell has to deal with the removal of such proteins. 4.4.1. Oxidative stress and protein modification The continuous formation of free radicals and other highly reactive oxidants is an inevitable byproduct of aerobic cellular metabolism. The main sources of intracellular radical formation are the mitochondria, releasing large amounts of the superoxide anion ðO 2 Þ, resulting from electron leakage from the respiratory chain. This primary radical can generate further radicals/highly reactive oxidants in numerous reactions. Generally, these oxidants are divided into reactive oxygen (ROS) and nitrogen species (RNS), sometimes summarized as ‘‘RONS’’. A few biological relevant RONS are listed in Table 4. Also several other cellular systems releasing reactive species are described, among them: the peroxisomes, generating H2O2 as a byproduct, as well as the xanthine oxidase releasing superoxide anions, the oxidative burst of macrophages while responding to infections, the monoamine oxidase (MAO), tyrosine hydrolase or L-amino acid oxidase, all of them releasing H2O2 during their normal activity. Another radical generating system is the family of the nitric oxide synthetases (NOS) with four known members: the inducible (iNOS), the endothelial (eNOS), the neuronal (nNOS) and a mitochondrial one (mtNOS). They produce the NO-radical, that originally shows protecting and antioxidative effects in tissues, but that forms the highly reactive peroxynitrite (ONOO) once it reacts with superoxide. Peroxynitrite forms peroxynitrous acid at physiological pH and becomes very instable in its protonated form (ONOOH), and decays to the two most aggressive radicals found in cellular systems: the hydroxyl radical ( OH) with an intracellular half-life of only 109 s and the nitrifying NO2 radical. Since these reactive species are able to (oxidatively) damage virtually every cellular structure, like proteins, lipids and DNA/RNA, systems are required that are able to reduce this continuous damage. The cellular antioxidative machinery can be divided into three ‘‘lines of defense’’, depicted in Fig. 24. The ‘‘first line’’ is able to scavenge free radicals/reactive species by ‘‘outcompeting’’ those mentioned cellular structures. This ‘‘line’’ contains low molecular antioxidants that react with free radicals and thus forming compounds that are much less reactive. Increasing the amount of radical scavengers’ reduces, but naturally does not completely prevent, cellular damage. An important intracellular antioxidant is glutathione GSH, a virtually universal antioxidant. This tripeptide (c-L-Glu-L-Cys-Gly) is abundant in mammalian cells (Sies, 1999). It is mainly found in its reduced state (GSH) while only a low amount is oxidized to the glutathione disulfide (GSSG). The ratio of the redox pair GSH/GSSG can be used to quantify the ‘‘redox-state’’ of a cell (Jones, 2002): an unstressed cell shows a physiological GSH/GSSG ratio of about 100:1 (Borras et al., 2004). GSH is used by several reducing enzymes and may also play a role in signal transduction (Paolicchi et al., 2002; Adler et al., 1999), cell proliferation
Oxidants
Low molecular antioxidants
Enzymatic antioxidants
- Protein repair systems for oxidized cysteine and methionine residues - Protein degradation
Vitamin C Vitamin E (mostly α-tocopherol) Carotenoides Glutathione (GSH) Lipoic acid Ubiquinol Uric acid Flavanoids Cu,ZnSOD MnSOD Catalase Glutathione peroxidase Peroxiredoxins Ferritin, transferrin GSSG-reductase Lipases
(O2•- → H2O2) (O2•- → H2O2) (H2O2 → O2 + H2O) (H2O2 → O2 + H2O) (H2O2 → O2 + H2O) (binding of Fe2+)
Thioredoxin/Thioredoxin reductase Glutaredoxin-/Glutathione-/Glutathione reductase-system Methionine sulfoxide reductase Proteasome Proteases DNAses/RNAses Lipases
I.
II.
III.
Fig. 24. The tree cellular antioxidative defense lines. This scheme displays the different antioxidative defense lines found in cellular systems. The first one (I) contains low molecular antioxidants. The second one (II) is build by enzymatic antioxidants, proteins that convert highly reactive particles into less reactive. The third line (III) contains enzymes that are able to repair cellular damage. The left part of the scheme depicts the protection from protein oxidation: the amino acids cysteine and methionine when oxidized might be repaired. If a protein is oxidatively modified in an irreversible way, it is recognized by the proteasome and degraded.
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
227
(Kwon et al., 2003), gene regulation (Arrigo, 1999; Kretz-Remy and Arrigo, 2002) and apoptosis (Marchetti et al., 1997; Hampton and Orrenius, 1998). Moreover it is considered to be a storage and transfer molecule for the NO-radical (S-nitrosoglutathione, GSNO) (Ji et al., 1999). The second ‘‘line of defense’’ is formed by enzymatic antioxidants, capable to transform reactive particles in less reactive ones; the common first step in the transformation of superoxide to H2O2, the second one the disproportionation of H2O2 to molecular oxygen and water. Considering oxidative stress, iron and iron-binding proteins play a special role. The redox-active form of iron (Fe2+) is able to catalyze the Fenton-Reaction, transferring an electron to H2O2 and thus inducing a decay to OH and OH, causing large amounts of oxidative damage by the highly reactive hydroxyl radical. Since the cytosol is a highly reducing environment, Fe3+ is reduced to Fe2+ very quickly. The so called Fenton-Reaction can be driven by all transition metals, but only copper (Cu+/Cu2+) and iron (Fe2+/Fe3+) are playing a significant role in mammalian cells. Therefore, the concentration of these ions must be regulated tightly either by transport or storage proteins, like ferritin/transferin and ferroportin for iron and ceruloplasmin for copper. Disruptions of iron homeostasis are known to be involved in several ocular diseases like glaucoma, cataract, age related macular degeneration (AMD, see Chapter 4.11) and intraocular hemorrhage, as well as Parkinson’s disease (PD, see Chapter 4.6.5) and mood disorders (Schmechel, 2007), whereas Menkes’s or Wilson’s disease are associated with changes in the copper metabolism. The ‘‘third line’’ of defense contains enzymes, able to repair oxidative damage inflicted to cellular structures. This includes a wide array of enzymes involved in DNA repair, in membrane repair and degradation of lipid peroxidation products. In the case of oxidatively damaged amino acids, it is known today that the two amino acids most susceptible to oxidative modification (cysteine and methionine) (Mary et al., 2004), are the only two amino acids that can be enzymatically reduced after oxidation. In the case of cysteine oxidation the result is the generation of disulfides, or sulfenic, sulfonic or sulfinic residues (Poole et al., 2004). A slight oxidation of a protein only affecting these amino acids can still be repaired: methionine by the methionine sulfoxide-reductases (Cabreiro et al., 2006; Moskovitz et al., 1999, 2001; Koc and Gladyshev, 2007) or the thioredoxin/thioredoxin-reductase (Petropoulos and Friguet, 2005; Brigelius-Flohe et al., 2003) and cysteine oxidation by the glutaredoxin-/glutathione-/glutathione-reductase-system (Bjornstedt et al., 1994; Trotter and Grant, 2003; Fernandes and Holmgren, 2004), using GSH to reduce substrate proteins. Another important cellular factor for proper protein folding is the family of so-called ‘‘protein disulfide isomerases’’ (PDI) (Puig and Gilbert, 1994; Martens et al., 2000; Leichert and Jakob, 2004). PDI provides a subfamily of the thioredoxins. Thioredoxin is mainly found in the endoplasmic reticulum (ER), where it reaches millimolar concentrations (about 10 mg per ml) (Ferrari and Soling, 1999; Freedman et al., 1994), playing an important role in the refolding of non-native or misfolded proteins. The formation of proper disulfide bridges and their influence on the three dimensional form of a protein, at the same time being essential for its functionality, is an important factor in protein maturation (for further information about the subsequential steps in protein expression, maturation and the cellular quality control, see Chapter 4.2). Since virtually every thiol of a protein can form a disulfide bridge with another one, that formation represents important steps in proper protein maturation and post-translational modification. The chaperone activity of PDI has already been shown in the ability of PDI to avoid the formation of protein aggregates (Song and Wang, 1995; Winter et al., 2002). The affinity of PDI to unfolded or non-native, proteins is both high and unspecific. This seems to be due to three substrate binding-sites with low affinities each (Hatahet and Ruddock, 2007), recognizing exposed hydrophobic structures of non-native proteins, a kind of substrate recognition that reminds to that of the 20S proteasome (see Chapter 2.2). Due to the multiplicative affinities, a substrate that is able to bind all three recognition-sites will be recognized with higher affinity than other ones (Hatahet and Ruddock, 2007). PDI can be found in two different forms, an oxidized and a reduced one. The oxidized form of PDI is reduced by forming a disulfide bridge in a substrate protein and needs to be reoxidized before further action (Thorpe and Coppock, 2007). The reduced form of PDI can perform isomerization and reduction of substrate disulfides and thus contribute to the reduction of oxidatively damaged proteins (Hatahet and Ruddock, 2007), while the reduction of PDI itself depends on GSH. PDI has been proven to be both degraded and newly synthesized after H2O2-mediated cellular stress (in clone 9 rat liver cells). The turnover of proteins in oxidatively stressed cells is naturally increased, but PDI has turned out to be degraded in a much larger amount (up to 90% in 24 h) than other proteins. At the same time the degradation can be blocked by inhibition of the proteasome (Grune et al., 2002). If in proteins amino acids other than the sulfur containing ones are oxidized and thus repairing of the protein is not possible, the damaged protein has to be degraded. A function largely performed by the proteasome, therefore, also being a part of the third line of defense. The recognition of the fact that oxidants and radicals are formed inevitable lead to the formulation of a condition defined as oxidative stress: the amount of free radicals/reactive species in a particular system overwhelms the antioxidative repair capacities (Sies, 1991, 1997). During oxidative stress the amount of irreducible oxidized proteins significantly increases. Thus a system, able to recognize and to degrade (oxidatively) modified proteins is needed.
4.4.2. Different stages of protein oxidation As every biological molecule, proteins can be oxidatively modified by a variety of free radicals and oxidants. Since proteins provide the largest group of cellular molecules, the probability of protein oxidation is increased in cells undergoing oxidative stress and thus the amount of dysfunctional proteins in the cell is increased under such conditions. If proteins are oxidatively damaged by reactive species, many different amino acid modifications can occur (Stadtman, 1993; Dean et al.,
228
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
Fig. 25. The three stages of protein oxidation. The left image shows the native form of a fully functional globular protein. The second image (from left) the slightly oxidized form, showing a protein having a (slightly) reduced activity, depending of the location of the oxidative modification; the oxidative modification is indicated by colored blisters, representing different modifications resulting from different oxidants (oxidation Stage I). The next image shows the protein in a more oxidized and completely unfolded state (Stage II). In this stage of oxidation no protein activity remains and the protein is the ideal substrate for proteasomal degradation. Probably the degradation signal is consisting of hydrophobic amino residues, in the native protein enclosed in its inside structure, but now exposed to the surface and able to bind to the a-rings of the proteasome. If the protein is not degraded at this stage of oxidation, it can be further oxidized and covalently cross-linked to other proteins in similar stages of modification (Stage III). The formation of these aggregates can be driven thermodynamically by its exposed hydrophobic residues and covalent cross linking can be caused by products of lipid peroxidation like MDA or HNE (Jung et al., 2007b,c, 2008).
1997; Naskalski and Bartosz, 2000). Furthermore, the oxidant can cleave the backbone of the protein, releasing smaller fragments. If proteins are oxidatively modified, perhaps three different stages can be differentiated (Fig. 25), dependent on the amount of oxidation, the protein was exposed to (Jung et al., 2007b). The third stage is the result of long-term accumulation of oxidatively modified proteins and further (chronic) oxidation: a highly oxidized protein aggregate containing covalently cross-linked proteins (30–70%), lipids (20–50%) (Double et al., 2008) and low amounts of sugars termed ‘‘lipofuscin’’ (Jung et al., 2007c; Terman et al., 2007; Kurz, 2008). In the first stage of oxidation (see Fig. 25) the protein is only slightly modified, but the main structure is still intact, resulting perhaps in a moderate reduction of activity. In this stage it is possible to reduce inflicted damage by the above mentioned repair-systems, though only for oxidatively modified methionine and cysteine. Due to introduced negative charges by amino acid oxidation, increasing its polarity, the protein might show a higher solubility (Chevallet et al., 2003; Rabilloud et al., 2002). In the next stage the amount of inflicted damage is sufficient to cause a partial unfolding of the protein, while the hydrophobic sequences that are usually covered inside globular soluble proteins are exposed at the surface. There is evidence, that these exposed hydrophobic structures are a degradation signal for the 20S proteasome and recognized by its a-rings. In most cells during and after oxidative stress an increased amount of proteolysis can be detected. According to Pacifici et al. (1993) there are three survival benefits of ATP-independent and specific degradation of oxidatively damaged proteins: (i) the quick removing of oxidized proteins, (ii) the prevention of an intracellular accumulation of these dysfunctional proteins, (iii) and the ‘‘recycling’’ of undamaged amino acids for de novo synthesis of proteins. The prevention of the accumulation of the most damaged proteins is the main function of the proteolytic degradation systems of damaged proteins. If the damaged protein is not proteasomal recognized and degraded, further oxidation can occur as well as a covalent crosslinking to other proteins by products of lipid peroxidation like 4-hydroxy-2-trans-nonenal (HNE) (Friguet and Szweda, 1997; Cohn et al., 1996; Xu et al., 2000) and malondialdehyde (MDA) (Voitkun and Zhitkovich, 1999; Montine et al., 1996; Burcham and Kuhan, 1997), two abundant bifunctional aldehydic oxidation products. If the protein is not sufficiently rapid degraded and/or the cell is exposed to an extreme amount of oxidative stress, the probability increases for proteins in the second stage to reach the third one. In this stage the proteins are not longer degradable by the proteasome. The result is a hydrophobic and insoluble protein aggregate, also termed ‘‘aggresomes’’ (Amidi et al., 2007; Grune et al., 2004; Johnston et al., 1998; Wong
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
229
et al., 2008), preliminary stages of the so-called ‘‘lipofuscin’’ (Harman, 1989; Terman and Brunk, 2004; Brunk and Terman, 2002), also termed as ‘‘age pigment’’ (Klimenko et al., 1987; Brody, 1960) or ‘‘ceroid’’ (Hartroft and Porta, 1965; Levine et al., 1968; Lee and Nicholson, 1976). Lipofuscin is not the direct result of extreme oxidative stress and heavily protein oxidation, but rather long-term effect of chronic low dosed (non-lethal) stress. The exact mechanism of lipofuscin-formation is still unknown, provided that there is only one mechanism. Lipofuscin is a highly oxidized mixture of different cellular components, differing from cell to cell and, therefore, also the lipofuscin is no unique compound. Lipofuscin can be detected using lipophilic staining methods of histology (Sudan Black, Berlin Blue, Nile Blue, Fontana-Masson, osmic acid, hematoxilin, ferric ferrycianide, Ziehl-Neelson or Eosin) (Porta, 2002) or, most often by its autofluorescence. Common transmission light microscopy displays lipofuscin as a transparent or yellow-brownish opaque material, forming globular structures in the perinuclear volume of the cell, showing diameters from 0.1 to 5.0 lm. The absorption peak of lipofuscin is found in the blue range of visible light at about 366 nm, the emission from 570 to 605 nm. The exact structure causing absorption and emission characteristics is still unclear, since native lipofuscin as occurring in animals is a mixture containing several fluorescent molecules. Until now A2E (Iriyama et al., 2008; Sparrow et al., 2008, 2006; Lakkaraju et al., 2007; Broniec et al., 2005), a pyridinium bis-retinoid found in retinal pigment cells is one of a few identified fluorophores. A2E plays a role in the pathology of age related macular degeneration (AMD) (Nowak, 2006; Ethen et al., 2007; Algvere and Seregard, 2002), mainly by its blue-light phototoxicity (Algvere et al., 2006). Retinal lipofuscin is able to generate reactive oxygen intermediates by photochemistry: one of the experimental described examples is the photochemical oxidation of retinal palmitate, resulting in the release of anhydroretinol, a molecule involved in the intracellular signal transduction cascade and able to induce apoptosis by the production of reactive oxygen species (Lamb et al., 2001).
Fig. 26. The intracellular network of lipofuscin formation. This image shows a simplified scheme of the very complex currently discussed network, involved in the intracellular lipofuscin formation. During cellular aging the amount of non-functional enlarged mitochondria increases, producing larger amounts of free radicals compared to normal mitochondria, resulting in an increased amount of protein oxidation. Normally oxidized proteins are degraded by the proteasome and only a remaining part is taken up by lysosomes. Damaged mitochondria are also taken up by these organelles and could be degraded via the lysosomal pathway. Though, by the intralysosomal accumulation of lipofuscin, the capacity of this system is reduced. Additionally both lipofuscin and covalently cross-linked proteins are able to inhibit the proteasome, thus driving the vicious cycle of lipofuscin formation.
230
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
Today, the preferred hypothesis of lipofuscin formation is the ‘‘mitochondrial–lysosomal axis theory’’ (Brunk and Terman, 2002), a very complex network involving mitochondria, lysosomes, the proteasome, a decrease of proteolysis and autophagy, lipid peroxidation and the cytotoxic effects of lipofuscin itself, catalyzing its own generation. A simplified scheme of formation of lipofuscin is depicted in Fig. 26. To explain this whole network would go beyond the scope of a proteasome-review. For extensive reviews of the intracellular mechanisms of lipofuscin formation, see Jung et al. (2007c), Terman et al. (2007), Terman and Brunk (2004, 2006), Brunk and Terman (2002), Kurz et al. (2008), Terman et al. (2006), and Sitte et al. (2001), but there are two main characteristics of lipofuscin, contributing to protein oxidation and decreased proteolysis that have to be mentioned: (i) Lipofuscin has the ability to bind transition metals like iron and copper, resulting from the absorption of the according proteins. In fact in some cells about 50% of lipofuscin are the remnants of the subunits c of mitochondrial ATP-synthase (Elleder et al., 1997). In the highly reducing environment of the cytosol these metals create a surface, able to catalyze Fenton-reaction and to generate quickly large amounts of highly reactive OH radicals. Data show that only a very small amount of lipofuscin is free in the cytosol, while most of that material accumulates in the lysosomal system (Kurz et al., 2008; Kurz et al., 2008; Jung et al., 2008). Normally only very few reducing compounds can be found in a lysosome, able to drive the redox-cycle, but it may be possible, that the hydroperoxyl radical HO2 is responsible for that (de Grey, 2002). The protonated form of superoxide is found only in very low amounts at pH 7.4 in the cytosol, but in larger amounts in the mitochondrial matrix at pH 4. In contrast to superoxide, hydroperoxyl as an uncharged molecule can diffuse easily across cell membranes and thus reach the lysosomal compartment, naturally only in low amounts, but able to reduce the transition metals presented in the surface of lipofuscin into their active form (Prasad et al., 2002). This continuous generation of OH within the lysosomal system may be the reason for the extreme amounts of a-tocopherol in the lysosomal membranes (Wang and Quinn, 1999) (up to 37-fold increased compared to any other membrane) (Rupar et al., 1992). Studies revealed connections between vitamin E deficiency and a raised lipofuscin formation (Fattoretti et al., 2002) as well as an increased amount of damaged mitochondria (Bertoni-Freddari et al., 2002). As mentioned, the mitochondria are the main sources of free radicals in mammalian cells and at the same time organelles very susceptible to oxidative damage. Oxidatively damaged mitochondria increase their free radical generation over time. Since oxidatively damaged mitochondria usually are degraded by the lysosomal pathway, the lysosomal system is with time slowly overloaded with indegradable lipofuscin, the amount of damaged mitochondria in the cytosol increases, causing increased oxidative stress in the cell, lipofuscin formation and further mitochondrial damage. This increased stress also enhances lipid peroxidation, releasing MDA and HNE, while both have the ability to enhance the formation of cross-linked proteins and the latter is able to directly damage the proteasome, reducing the degradation of damaged proteins (Friguet and Szweda, 1997; Friguet et al., 1994a,b; Okada et al., 1999). (ii) The ability of lipofuscin to directly inhibit the proteasome (Grune et al., 2004; Powell et al., 2005; Sitte et al., 2000b). How exactly this mechanism works is still unclear, but the fact that the proteasome recognized hydrophobic structures as substrate suggest that the lipophilic surface of lipofuscin is recognized as substrate, too. Another hypothesis is the binding of the proteasome to small protein fragments on the surface of lipofuscin that are long enough to bind the a-rings, but too short to reach the inner proteolytic chamber to be degraded. Thus proteasomal activity is detracted in futile attempts of degradation, resulting in an increased amount of oxidized proteins in the cytosol that are not degraded before reaching the third stage of oxidation (Terman and Sandberg, 2002) (Fig. 27).
Fig. 27. Model of proteasomal inhibition by lipofuscin. This model suggests the binding of the proteasome to short remnants of proteins that are covalently bound and cross-linked to the surface of the lipofuscin aggregate. These residues are recognized as substrate and thus proteasomal activity is competitively detracted in futile attempts of substrate degradation (A) or by binding of metals lipofuscin might be able to produce ROS in turn able to inhibit the proteasome (B).
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
231
Thus the formation of lipofuscin is a stochastic process, dependent on the amount of oxidized proteins that are not degraded before aggregating. Normally, the intracellular accumulation of lipofuscin is a very slow process, taking a whole life to fill the cells until a threatening amount is reached, endangering the functionality of the cell. In the cells of aged animals lipofuscin can occupy up to 40% of the cytosolic volume, in the motor neurons of centenarians up to 75% have been found filled (Yin, 1996), finally causing apoptotic cell death. Inefficiency of the proteasome or the proteasomal system might have devastating consequences in mammalian cells. Inhibition of the 26S proteasomal system resulted in the formation of Lewy-like inclusions bodies as found in Parkinson’s disease and neurodegeneration in mice (Bedford et al., 2008). Proteasomal inhibition turned out to result in the formation of aggresomes, mainly by decreased proteolysis of slightly oxidized proteins (Johnston et al., 1998; Bardag-Gorce et al., 2004; Catley et al., 2006; Fratta et al., 2005). Knockout of single proteolytic active subunits resulted in reduced proteasomal activity and an increased accumulation of oxidatively modified proteins (Ding et al., 2006). Another interesting model is the formation of ‘‘Mallory bodies’’ (MB), as found in alcoholic hepatitis, non-alcoholic steatohepatitis and Wilson’s disease. Affected cells showed a (partial) proteasome depletion and at the same time cytosolic aggregates of hyperphosphorylated and ubiquitylated proteins (in MB), mostly build of cytokeratin (Bardag-Gorce et al., 2001). The same effects are caused by dysfunctions of the lysosomal system. Neuronal ceroid lipofuscinoses, also termed as ‘‘Batten’s disease’’, including different inherited lysosomal storage disease, resulting in progressive and finally lethal neuronal disorders (Dawson and Cho, 2000; Weleber et al., 2004; Cooper, 2003). Until now eight different forms of the disease are known, affecting proteases that are involved in lysosomal degradation, like palmitoyl-protein thioesterase 1 (CLN1), the tripeptidylpeptidase 1 (CLN2), and cathepsin D (CLN8). These pathologies are accompanied by a rapid intralysosomal accumulation of a lipophilic, autofluorescent pigment. 4.4.3. Proteasomal degradation of oxidized proteins Following oxidation of proteins one can detect changes in the proteolytic susceptibility of globular proteins. Such changes in proteolytic susceptibility show a biphasic response. At moderate oxidant concentrations proteolytic susceptibility increases slowly (Fig. 28, phase 1), whereas continuous oxidation leads to a dramatic increase in proteolytic susceptibility (Fig. 28, phase 2). Further oxidation leads to a decline in the proteolytic susceptibility, sometimes even below the basal degradation level (Fig. 28, phase 3). Such behavior seems to be a common feature of all globular, soluble proteins with defined secondary and tertiary structures, independent of the origin of the proteins. On the other hands proteins with no folded structure, such as tau (Jeganathan et al., 2008; Mukrasch et al., 2007), a-synuclein (Dorval and Fraser, 2006) or casein, are inherently good substrates for proteolysis and their susceptibility does not increase due to oxidation; although it can decrease by heavy oxidation. Since a number of different sources for the proteasomal system have been employed by various investigators, using diverse sources as crude cell lysates, purified cell lysates or isolated proteasomes and a large number of substrate proteins often without a species mach of the proteasomal source and the substrate proteins one can assume species overlapping mechanisms for the recognition of oxidized proteins (Grune et al., 1996; Sitte et al., 1998; Pacifici and Davies, 1990; Salo et al., 1990; Pacifici et al., 1993; Rivett, 1985a,b; Giulivi and Davies, 1993; Grune et al., 1995, 1997, 1998; Mehlhase et al., 2000; Gieche et al., 2001; Davies and Lin, 1988a; Sitte et al., 2000a,b; Shringarpure et al., 2001). Therefore, it can be concluded that the removal of ‘‘minimally’’ oxidized proteins is an essential function of the proteasome in the maintaining of cellular homeostasis by preventing the accumulation of highly oxidized and cross-linked proteins. This
Degradation of oxidized substrates
I.
II.
III.
Maximal degradation
Proteolytic stimulation
Basal degradation Optimal oxidant concentration
0
Proteolytic inhibition Oxidant concentration / oxidative protein modification
Fig. 28. The proteolytic susceptibility of oxidatively modified proteins. In this figure the different stages of oxidative modifications/damage of a protein are shown and it is susceptibility to proteasomal degradation. Proteins in the modification Stage I have perhaps a slightly increased proteolytic susceptibility until a maximum is reached (at the optimal oxidant concentration) in Stage II. Stage III is not longer optimal degradable by the proteasome and moreover able to inhibit proteasomal degradation, thus decreasing the proteolytic stimulation to a level beyond the normal ‘‘base line’’ of proteolysis. (For structural features of proteins in oxidation Stages I–III see also Fig. 25).
232
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
Relative amount of protein carbonyls [%]
250 200 150
7.5•10-5 µm H 2O2•cell-1
untreated cells H2O2-treated cells
+LC
+LC
*
-LC
*
con
100 50 10 µm
0
0h 24h con
24h H2O2
24h 0h 24h Time after H2O2treatment [h]
Fig. 29. Role of the proteasome in removing of oxidized proteins. This figure shows the important role of the proteasome in the removing of oxidatively modified proteins in cells after oxidative stress. The graph shows the intracellular amount of protein carbonyls, representing the amount of oxidatively damaged proteins, found in cells (HT22 cells, a mouse hippocampal tumor line, immunocytochemistry) with and without applied oxidative stress (by H2O2). The left part of the graph shows the untreated control, whereas 0 h indicates no incubation time after hydrogen peroxide treatment. Incubation after treatment were performed either in the presence (+LC) or absence (LC) of the proteasomal inhibitor lactacystin. The two fluorescence microscopically images on the right show representative cells without (Con) and after H2O2 treatment without incubation time after treatment. Statistical significance is indicated by asterisks (Student’s t-test, p < 0.05). (Jung and Grune, unpublished data, for details see Jung et al. (2005, 2006, 2007a)).
assumption suggests the existence of a species overlapping recognition motif of oxidized proteins by the proteasome. Numerous examples show the correlation of proteasomal degradation with single amino acid oxidation, e.g. methionine (Levine et al., 1996) or tyrosine (Lasch et al., 2001). But one has to be aware that the oxidation of many amino acids occurs during protein oxidation in parallel. In addition to that the native folding of the protein get lost. Several years ago the group of Davies proposed that oxidized proteins are partially unfolded due to the loss of regular secondary and tertiary structures within the domain of the oxidative impact and this unfolding is the signal for the proteasomal recognition (Pacifici et al., 1993; Giulivi et al., 1994). Hemoglobin (Hb) exposed to OH radicals increased its surface hydrophobicity at the same time as the susceptibility for proteasomal degradation (Pacifici et al., 1993; Giulivi et al., 1994; Cervera and Levine, 1988). That increased susceptibility was shown in different systems for albumin and superoxide dismutase, too (Davies and Lin, 1988a,b; Davies et al., 1987). For the proteasome the peptide bonds between two hydrophobic residues have turned out to be a preferred substrate (Pacifici et al., 1993). Exposed hydrophobic sequences would be an ATP-independent way of recognition and degradation by the 20S proteasome during oxidative stress. Lasch et al. (2001) were able to demonstrate an up to 50% unfolding of RNase A under conditions allowing an increase in proteolytic susceptibility. Unfolding of soluble globular proteins is clearly accompanied by an exposure of hydrophobic patches from the interior of the protein globule to the outside. This might be a reason for the initial aggregation of oxidized proteins through hydrophobic interactions. A correlation between the increase in proteolytic susceptibility towards the proteasome and the increase of substrate surface hydrophobicity was demonstrated by means of several techniques, e.g. separation of substrates, according to their hydrophobicity, by hydrophobic interaction chromatography (Pacifici et al., 1993; Giulivi et al., 1994) or by using fluorescence labels detecting the surface hydrophobicity of proteins (Levine et al., 1996, 2000). As earlier mentioned the proteasome has a preference to bind hydrophobic and aromatic amino acids (Hough et al., 1987). Therefore, a recognition of these hydrophobic ‘‘unfolded’’ patches by the proteasome seems likely. Numerous studies have been performed using the isolated 20S ‘‘core’’ proteasome to degrade oxidized proteins. Today it is accepted that the 20S proteasome is able to degrade such oxidized proteins (Grune et al., 1996, 1998, 2003, 2004; Friguet et al., 2000; Pacifici et al., 1989, 1993; Pacifici and Davies, 1990; Salo et al., 1990; Stadtman, 1993; Rivett, 1985a,b; Giulivi and Davies, 1993; Peters, 1994; Mehlhase et al., 2000; Giulivi et al., 1994; Davies and Goldberg, 1987a,b; Davies et al., 1987; Davies, 1987; Davies and Delsignore, 1987; Stadtman and Oliver, 1991; Giulivi and Davies, 1994; Levine et al., 1994; Shang and Taylor, 1995; Jahngen-Hodge et al., 1997; Obin et al., 1998; Ullrich et al., 1999a,b; Shringarpure et al., 2000; Chao et al., 1997; Mehlhase et al., 2005; Chondrogianni et al., 2003; Farout and Friguet, 2006; Bulteau et al., 2001; Friguet, 2002; Szweda et al., 2002; Stadtman et al., 2003). Also in cells the proteasome is responsible for the degradation of oxidized proteins, most evidently shown by means of anti-sense oligodeoxynucleotides directed against the (essential) a6 (or C2) proteasomal subunit. Anti-sense treated cells are essentially depleted of the proteasome and are no longer able to increase protein turnover after oxidative stress and to degrade oxidatively modified proteins (Grune et al., 1996, 1995, 1998). In addition to that selective proteasomal inhibitors prevent the degradation of oxidized proteins (Sitte et al., 1998; Davies, 2001; Grune et al., 2002). So, the amount of oxidatively damaged proteins was even after 24 h still as high as in cells immediately after exposure to oxidative stress, if cells were treated with the irreversible proteasome inhibitor lactacystin. In contrast, control cells without proteasomal inhibition were able to reduce the amount of protein carbonyls 24 h after exposure to an oxidizing agent (Fig. 29). As already described the proteasome is regulated by numerous regulators. Especially well investigated is the 19S/PA700 regulator forming the 26S proteasome if binding to the core proteasome (see Chapter 3.2) for a possible involvement in the breakdown of oxidized proteins. More and more facts accumulate that the 26S proteasome does not play any or only a minor role in the degradation of oxidized proteins (Shringarpure et al., 2003). The lack in ubiquitination of oxidized proteins may be
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
233
due to oxidative side chain modification of lysine residues, which are the binding site for ubiquitin. Our studies of protein oxidation and proteolysis have consistently shown a lack of ATP or ubiquitin involvement in the degradation of oxidized proteins in mammalian cells and cell extracts (Grune et al., 1996; Takeuchi and Toh-e, 1997; Davies, 2001; Pacifici et al., 1989, 1993; Pacifici and Davies, 1990; Salo et al., 1990; Grune et al., 2002, 1995, 1998, 2002; Giulivi and Davies, 1993; Giulivi et al., 1994; Davies et al., 1987; Davies, 1987; Davies and Delsignore, 1987; Davies and Goldberg, 1987b; Giulivi and Davies, 1994; Ullrich et al., 1999a,b; Shringarpure et al., 2000, 2003; Davies, 1986; Reinheckel et al., 1998). Extensive studies have been performed to investigate the compartment of protein oxidation and removal of oxidized proteins (Jung et al., 2007a, 2005, 2006; Powell et al., 2005) by means of immunocytochemistry. Although the highest proteasome concentration was found in the nucleus, the most extensive load of proteasome with oxidized proteins was found in the distant cytosol (Jung et al., 2007a,b,c, 2005, 2006), where extensive degradation of oxidized proteins occurs. Experiments demonstrated no statistical significant removal of oxidized proteins in any compartment of the cells if treated with proteasome inhibitors (Jung et al., 2007a, 2006). 4.5. The proteasomal system and aging Aging is characterized by a fundamental change in cellular functionality, characterized by a decline of adequate responses to the changing environment. Importantly, this process, passed through by every living cell and thus by every organism, is not considered to be pathologic. About 300 theories and hypotheses try to explain the process of aging or the cellular changes during aging (Merker et al., 2001). The causes of aging are still not clear, but it is known today that a complex network, involving several interacting factors, is involved. The most promising and currently accepted model of aging is a refined version of the ‘‘oxidative theory of aging’’, proposed by Harman (2001), published in 1956 (Harman, 1956) and in 1972/3 refined to the ‘‘free radical theory of aging’’ (FRTA) (Harman, 1972a,b, 1973) by the same author, now involving a major role of the mitochondria. The FRTA can be united without contradiction with newer theories like the ‘‘the lysosomal-mitochondrial axis theory of postmitotic aging and cell death’’ (Brunk and Terman, 2002; Terman et al., 2006), theories of ‘‘stress induced premature senescence’’ (SIPS) (Toussaint et al., 2000) or these of ‘‘biological garbage accumulation’’ (Terman and Brunk, 2006; Stroikin et al., 2005). In all theses theories the primary cause of aging are free radicals alone or in combination with other damaging agents. The damage induced by these compounds accumulates over time, finally resulting in a functional breakdown of the cell and cell death. An important fact, underpinning the FRTA is the widely accepted increase of the formation of free radicals with age; another one is the positive correlation of life span with the resistance of an organism to oxidative stress and a negative correlation with the amount of free radicals that are produced by its cellular metabolism over time. Another fact underlining the FRTA is a decreasing antioxidative capacity of the cell with age; this decrease is represented by a reduced amount of free antioxidants, the diminution of the efficiency of antioxidative proteins and finally of systems that are able to recognize and to degrade damaged cellular structures (in detail explained in Chapter 4.4). Thus aging could be considered as a chronic redox dysregulation (Humphries et al., 2006; Valko et al., 2007). Although, the exact cause or the fundamental mechanism inducing this shift are still unclear (Gilca et al., 2007). But one of the systems declining with age for an unknown reason is the proteasome and the UPS. The generally accepted main sources of free radicals in the cell are mitochondria, at the same time these organelles are most susceptible to oxidative damage. During aging an increasing number of damaged and malfunctioning mitochondria accumulate. The mitochondrial malfunction might reduce the cellular availability of ATP, needed for UPS-mediated proteolysis, de novo synthesis of both proteins and for the maintenance of the cellular redox-state (also important for proper protein folding) and thus stress-resistance. Interestingly, an aging phenotype can be induced in young cells by exposure to chronic oxidative stress (Jung et al., 2008; Muller, 2006; Ma et al., 2002), as described by the group of Toussaint (Toussaint et al., 2000; Zdanov et al., 2006; Pascal et al., 2005; de Magalhaes et al., 2002; Dierick et al., 2002, 2000). As mentioned, the proteasomal activity decreases in aging cells and in reverse proteasomal inhibition in young cells is able to enhance the formation of (polyubiquitinated) protein-aggregates and lipofuscin (Powell et al., 2005; Terman and Sandberg, 2002). A decrease in the proteasomal activity has been shown in human tissue for muscle (Ferrington et al., 2005; Husom et al., 2004), lens (Viteri et al., 2004) (but not in the nucleus of the lens, except for cases of cataract-formation (Zetterberg et al., 2003)), lymphocytes and epidermal cells like keratinocytes (Petropoulos et al., 2000) and fibroblasts (Sitte et al., 2000c,d; Hwang et al., 2007). An intracellular increase of oxidatively modified but not degraded proteins was found, too (Stadtman, 1992). Decreases in the proteasomal activity have been found in the hearts (Bulteau et al., 2002), spinal cords (Keller et al., 2000) and the brain (with differences in the single regions) of rats (Zeng et al., 2005), too. The decline was about 1.5- to 2-fold, while the activity of the b1-subunit (peptidyl-glutamyl-peptide-hydrolysing-activity) was mostly affected (Chondrogianni and Gonos, 2005). The proven ability of lipofuscin to inhibit the proteasome (see Chapter 4.4.2) is only one factor of the age-dependent decrease of the proteasomal activity. To investigate endogenous factors, both expression and functionality of the single proteasomal subunits have been investigated, and the results revealed that the subunits were differently affected. The changes in the substructure of the proteasomal system during the aging process turned out to be manifold. The expression of the a-subunits was not down-regulated, and the loss of function seems to be due to a reduced availability of the b-subunits (Chondrogianni et al., 2003). Experimental results according to the changes in the inducible proteasomal subunits could not be more inconsistent. In lymphocytes no change in b1i and b5i was detected (Carrard et al., 2003), in rat heart both a decrease in b5i and the 11S proteasome activator was found (Lee et al., 2002), while in aged
234
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
muscle the expression seemed to be increased (Ferrington et al., 2005; Husom et al., 2004). If that shift is due to a mechanism compensating the decreased expression of the constitutive b-subunits or to an inflammatory response, respectively increased levels if IFN-c is still unclear. Another aspect of aging is the mentioned increase of ROS, resulting not only in protein oxidation, but also in a significant increase of lipid peroxidation. Typical products of lipid peroxidation are HNE and MDA (see Chapter 4.4.2), both able to modify and inhibit the proteasome that can be oxidatively damaged in the same way as every other cellular protein (Friguet and Szweda, 1997; Friguet et al., 1994a,b; Farout et al., 2006). As the fully assembled proteasome, the non-assembled subunits can be modified, too. That was shown experimentally by Bulteau, revealing HNE-adducts in the proteasomal subunits a1, a3, a4 and other oxidative modifications in a3-, a4-, a5- and b4-subunits (Bulteau et al., 2000, 2001). Keller et al. (2000) instead detected HNE-adducts mainly in b-subunits in the spinal cord of Fisher rats after exposure to oxidative agents and suggested a relation of decreasing proteasomal activity to neuronal cell death. In human lymphocytes glycation of the single subunits was detected, too (Carrard et al., 2003). But not only the constitutive proteasomal subunits were increasingly modified in aged cells, the b5i-subunit is a target of HNE modification, too (Chondrogianni and Gonos, 2005). In contrast to the 20S proteasomal complex that is mostly degraded via the lysosomal pathway, oxidatively damaged or misfolded single subunits are polyubiquitinated and degraded by the UPS (Chondrogianni and Gonos, 2005). In aged cells the amount of subunits, both proteasomal ones and components of the 19S regulator cap, that are found in a polyubiquitinated state is increased compared to young control cells. The a2-, a5-, b5- and b5i-subunits are mostly affected by ubiquitination in aged cells (Chondrogianni and Gonos, 2005). Glycation as age-related oxidative modification was detected in the subunits Rpt2 and Rpt1 in the 19S components (Chondrogianni and Gonos, 2005), but, interestingly, HNE-adducts were not detected (Carrard et al., 2003). Recent studies showed a significant decrease of the 19S regulator in senescent fibroblasts from Wistar rats, that may explain the accumulation of polyubiquitinated proteins with age (Chondrogianni et al., 2003). Despite of these findings, the formation of the 20S ‘‘core’’ proteasome seems not to be interfered: investigations of peripheral blood lymphocytes from donors in the age of 20–63, showed a reduced activity in aged donor cells, probably due to posttranslational modifications, nevertheless the amount of intracellular proteasomes was not decreased compared to cells from young donors (Carrard et al., 2003). One of the most important questions is related to the age-related changes of the proteasome/UPS in post mitotic cells, as neurons. Several studies revealed that proteasomal inhibition is able (even if not in all cases) to induce cell death in neuronal cells (Keller and Markesbery, 2000), mediated by the p53-pathway (Lopes et al., 1997) and caspase activation (Pasquini et al., 2000; Qiu et al., 2000). Although the evolutionary ‘‘invention’’ of static postmitotic neuronal cells may bring the advantage of a long-term memory; on the other hand the process of aging shows a significant increased impact in these cells and thus a functional proteasomal system to maintain protein homeostasis is essential in prevention of lipofuscin accumulation and neurodegenerative diseases (Terman and Brunk, 2005). Very interesting are findings that in healthy centenarians both the amounts of proteasomal activity and oxidatively modified proteins were comparable to those found in a much younger control group, but not comparable to a ‘‘normal’’ aged one (Chondrogianni et al., 2000). This brought up the idea of an artificial activation of the proteasomal system as an anti-aging strategy (Chondrogianni and Gonos, 2008): after a transfection of HeLa cells with the b5-proteasomal subunit, at the same time the b1- and b2-subunits were upregulated, suggesting a mutual activation (Chondrogianni et al., 2005). Cell lines that were ‘‘activated’’ in this way, showed a significantly increased proteolysis. The same transfection in IMR90 cells (human embryonic fibroblasts) extended the life-span of the cells by 15–20% (Chondrogianni et al., 2005). Similar effects were achieved by overexpression of the POMP protein (compare Chapter 2.3) in fibroblasts (Chondrogianni and Gonos, 2007): the cells showed increased levels of the 20S proteasome and as the result a higher turnover of damaged/modified proteins and enhanced recovery after externally applied oxidative stress. These results suggest a strong connection between proteasomal activity, cellular protein homeostasis and longevity.
4.6. The proteasome in neurodegeneration and stroke The brain has several metabolic and anatomic characteristics leading to special role of protein turnover and proteasome in this organ. Since the brain is protected by the blood–brain-barrier the glial cells, microglia and the neurons are adapting to special extracellular conditions. Any disruption of this barrier has, therefore, consequences. Furthermore, neurons of the central nerve system (CNS) are postmitotic cells. Considering the high amount of radical production and the relatively low antioxidative capacity, it is obvious that oxidative stress, particularly if it becomes chronic, quickly shows devastating effects (Merker and Grune, 2000; Grune, 2000). The brain is the mammalian organ showing high oxygen consumption. The human brain uses about 25% of the oxygen uptaken by the body, even if this organ represents only about 5% of the body weight. It consumes 3.5 ml of oxygen per 100 g of tissue per minute (Kish et al., 1992). It is possible that about 2% of the consumed oxygen are turned into O 2 (Boveris and Chance, 1973). Moreover the brain has a high iron content (Gerlach et al., 1994) and shows an excess of (poly)unsaturated fatty acids, that are susceptible targets to oxidative damage (Halliwell, 1992). On the other hand the brain shows only low activities of antioxidative enzymes like superoxide dismutases, catalase and glutathione peroxidases (GPx) and glutathione reductase (Dringen et al., 2000). Since the activity of catalase in the brain is low, glutathione peroxidase becomes the most important H2O2 and peroxide removing enzyme (Bharath et al., 2002), while the necessary glutathione is found in mil-
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
235
Fig. 30. The possible pathways in the intracellular accumulation of modified proteins. Sources of misfolded protein can be a genomic mutation (A), a direct modification of the expressed protein (may be due to oxidation or other posttranslational modifications) (B) or a shifted balance between expression and proteolytic degradation of a cellular protein (C).
40 25-35% Prevalence for AD [%]
35 30 25 20 15
3-12%
10 5 0
1-3% 60-70
70-80 Age [years]
>85
Fig. 31. Age dependence of the prevalence for Alzheimer’s disease (AD). The prevalence of AD rises quickly with age. The graph shows data of the American population (Evans et al., 1989). The white part of each column indicates the minimal, the grey part the maximal prevalence in different stages of age.
limolar concentrations in the brain (Dringen et al., 2000). Interestingly, there exist several high-activity systems producing free radicals in the brain like mitochondria, nNOS, xanthine oxidase, or MAO. Proteolysis, proteasome and protein turnover in the brain and in brain cells were intensively investigated, especially in front of the background of the extensive formation of protein deposits in neurodegenerative diseases. This accumulation of protein aggregates can occur both extracellularly, and within various cellular compartments. Differences in the effects of protein aggregates on brain functions might be expected, depending on the rate of formation and the exact location of such aggregates. Extracellular protein deposits might be phagocytosed, whereas intracellular protein aggregates often undergo autophagocytosis, resulting in an accumulation of such material in lysosomes (Brunk and Terman, 2002; Terman et al., 2006; Terman and Brunk, 1998). The accumulation of proteins can be the result of malfunctions of the cellular metabolism. As demonstrated in Fig. 30 the accumulation of protein aggregates can result from a genetical modification resulting in a lower proteolytic susceptibility of the protein or a proneness to aggregation (Fig. 30, option A). Metabolism or environment induced modification of proteins might result in posttranslational modifications, which either lead to a direct crosslinking and aggregation or to a reduced proteolytic susceptibility (Fig. 30, option B). Additionally, an excessive high formation rate of posttranslationally modified proteins might simply overwhelm the capacity of the proteolytic systems. Examples for posttranslational modifications changing the proteolytic susceptibility of substrates include oxidative modifications (Grune et al., 1996, 1995, 1998; Takeuchi and Toh-e, 1997; Pacifici et al., 1989, 1993; Pacifici and Davies, 1990; Salo et al., 1990; Stadtman, 1993; Rivett, 1985a,b; Giulivi and Davies, 1993; Mehlhase et al., 2000; Gieche et al., 2001; Giulivi et al., 1994; Davies and Goldberg, 1987a,b; Davies et al., 1987; Davies, 1987; Davies and Delsignore, 1987; Stadtman and Oliver,
236
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
1991; Giulivi and Davies, 1994; Levine et al., 1994; Shang and Taylor, 1995; Jahngen-Hodge et al., 1997; Obin et al., 1998; Ullrich et al., 1999a,b; Shringarpure et al., 2000; Chao et al., 1997; Stadtman and Levine, 2000) and phosphorylation of the tau protein (Mack et al., 2001). Furthermore, due to malfunction of the proteolytic machinery an imbalance between protein synthesis and proteolysis might be the reason for an accumulation of cross-linked proteins (Fig. 30, option C). Examples are the ceroid lipofuscinoses with their decline in proteolytic capacity (Weimer et al., 2002; Holopainen et al., 2001). Due to the prolonged half-life and the exposure to external influences such primary protein aggregates are transferred into a highly cross-linked and reactive material, referred to as lipofuscin, ceroid or AGE-pigment-like fluorophores (Yin, 1996; Yin, 1995). During this process such material might be several times taken up into lysosomes and released by disrupting them. Postmitotic cells, as neurons, are especially prone to the accumulation of such material, since the formed protein aggregates are not divided into two daughter cells upon cell division. During lifetime a number of cells, e.g. neurons (Glees and Hasan, 1976; Terman and Brunk, 1998; Wisniewski and Wen, 1988) and RPE cells (Wihlmark et al., 1997; Wolf, 1993) accumulate tremendous amounts of aggregated proteins, sometimes using up to 75% of the cellular volume (Yin, 1996). The accumulation of such tremendous amounts of protein debris might lead to cell death by apoptosis (Brunk and Terman, 2002; Powell et al., 2005). 4.6.1. The role of the proteasome in Alzheimer’s disease (AD) Alzheimer’s disease (AD) was first described by Alois Alzheimer. AD patients are about 60% of all cases of the dozens existing neurodegenerative diseases. Thus AD is becoming the most common one. In the year 2008 about 30 million people are affected worldwide. The number of affected people is expected to increase to 80 millions in 2040 (Ferri et al., 2005). In Fig. 31 the age dependence of the prevalence for Alzheimer’s disease is depicted for the American population.
Fig. 32. Two ways of amyloid precursor protein (APP) processing. The leftmost image shows the unprocessed APP, a transmembrane protein. The transmembrane domain of APP is labeled TmD, the cell membrane is symbolized by a stylized lipid bilayer and the ‘‘amyloid beta-fragment’’ is marked red. The upper line shows the non-amyloidogenic pathway of APP processing, while the protein is cleaved by the a-secretase, releasing the sAPPa-fragment (‘‘s’’ for ‘‘soluble’’), and afterwards the additional c-secretase cleaving is forming the P3 fragment. The line below shows the amyloidogenic pathway, while the APP is first cleaved by b-secretase, releasing sAPPb (‘‘s’’ for ‘‘soluble’’) and then by c-secretase, releasing the amyloid beta fragment with 40 (Ab40)or 42 (Ab42) amino acids, whereas the latter shows strong aggregating capacity the Ab40 is less able to form an aggregation core (Crouch et al., 2008). (For interpretation of the references in colour in this figure legend, the reader is referred to the web version of this article).
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
237
First clinical symptoms in AD are malfunctions of both the short-term memory and the sense of direction, later the longterm memory is affected until even close relatives are no more recognized. The patients become unaware of time and the location, loosing their orientation. Further an impaired motor coordination might occur, caused by AD-affected motor neurons. The median survival rate is about 9 years after clinical diagnosis, while death is mainly caused by bedridden induced pneumonia, not by the pathology of AD itself. Physical manifestations are brain atrophy and loss of neurons and synapses, mostly found in the temporal, frontal and parietal cortex, as well as in hippocampus and amygdala. The massive decrease of brain volume occurs by cell loss, primarily of neurons. Typical physiologic manifestations are both an extracellular accumulation of b-amyloid aggregates and an intracellular formation of so-called neurofibrillary tangles (NFTs) consisting of the tau protein in a hyperphosphorylated form (Lee et al., 2001). The cytotoxic b-amyloid fragments are formed by cleaving of the amyloid precursor protein (APP) by the enzymes b – (also termed ‘‘BACE1’’) and c-secretase. APP is first cleaved by b-secretase into sAPPb and a smaller fragment termed as C99, while the latter is cleaved by c-secretase, both generating Ab40 or Ab42 (Fig. 32). The c-secretase activity is located in a protein complex of various proteins (see below). Native APP shows two isoforms containing different carboxyl endings, finally causing the release of either Ab40 or Ab42, the major types of amyloid that are accumulating the AD-brain. The fundamental role if Ab-formation was first discovered in the heritable forms of the AD, revealing mutations in the cleaving sites of APP (Hardy and Allsop, 1991). Cytotoxic effects of b-amyloid are the result of many different factors. Among them a direct influence on the mitochondria, decreasing their amount (Hirai et al., 2001) in the mainly affected areas of AD-brains and changing their morphology (Baloyannis et al., 2004) in affected neurons. The cellular functions of the APP family are still unknown, but potentially APP may be involved in calcium homeostasis (Mattson et al., 1993; Milward et al., 1992), cell growth and adhesion (Rossjohn et al., 1999; Mattson, 1997), axional vesicle transport (Gunawardena et al., 2003) and regulations of free metal ions (Smith et al., 2007; Adlard and Bush, 2006). APP shows a strong affinity to Cu2+ (Kd 10 nM) (Barnham et al., 2004) ions and the ability to reduce them to Cu+, able to drive the Fenton-reaction. Knock out (APP and APLP2) mice show increased levels of free copper in brain and liver, while overexpression induces increased reduction of metal ions (Bayer et al., 2003). Electron transfer from Ab resulting from b-/c-secretase cleavage to Cu2+ or Fe3+ releases Ab radicals (Ab) that are able to oxidize proteins and lipids in further reactions, as well as reduced copper enables the Fenton-reaction. The lipid peroxidation by-product HNE has been found elevated in AD brain tissue (Markesbery and Lovell, 1998; Sayre et al., 1997) and protein carbonyls (frontal pole, hippocampus and superior middle temporal gyrus) (Pamplona et al., 2005; Hensley et al., 1995), both indicating a strong involvement of oxidative processes in the progression of the disease. Moreover 3-nitrotyrosin formation, an indicator for peroxynitrite induced protein oxidation, was found to be 8-fold increased in the hippocampus and neocortical regions (Hensley et al., 1998; Smith et al., 1997). Oxidative stress and its effects seem to play early and considerably roles in the progression of AD (Perry et al., 2000; Nunomura et al., 2001; Christen, 2000). One of the pathological highlight of AD is the intracellular accumulation of insoluble and indegradable aggregates of hyperphosphorylated tau proteins, found sometimes in a highly ubiquitinated form. Tau can be phosphorylated by various kinases, including GSK3, cdk5, p38 or JNK (proline directed), respectively PKA, PKC, CaMKII, MARK or CKII (non-proline directed) (Sengupta et al., 1998; Yamamoto et al., 2005; Kyoung et al., 2004; Robertson et al., 1993; Taniguchi et al., 2001; Singh et al., 1995; Avila et al., 2006). In the tau-441, the longest human tau isoform, 79 residues are found that are susceptible to serine/threonine phosphorylation and about 30 of them can be processed by the mentioned kinases. In a non-phosphorylated or low phosphorylated state the tau protein is associated to microtubules, providing stability of these structures. The phosphorylated form of tau dissociated from microtubules causing their depolymerization and thus a decreasing axonal transport in neuronal cells (Cuchillo-Ibanez et al., 2008). By unknown reasons tau might be hyperphosphorylated. Those free, hyperphosphorylated tau proteins tend to form larger structures like paired helical filaments (PHFs) (Avila, 2006; Mandelkow et al., 2007; Hernandez et al., 2005) and neurofibrillary tangles (NFTs) (Brion et al., 2001; Layfield et al., 1996). If these aggregates are a stress response of the cell or toxic byproducts is still under discussion. It was shown that dementia in AD correlates with the formation of NFTs (Caughey and Lansbury, 2003), but in other neurodegenerative disorders like Huntington’s disease these tau aggregates revealed protective functions reducing neurodegeneration (Arrasate et al., 2004). In transgenic mouse models defects in memory were not associated with tau filament association, even though memory functions improved after suppression of transgenic tau (SantaCruz et al., 2005). In drosophila the pathology (early onset, neurodegeneration and death) progressed without the formation of neurofibrillary tangles (Wittmann et al., 2001), supporting the results from transgenic mice. Other results suggest a toxicity dependent of the state of tau aggregation and different toxicity of some tau isoforms (Gomez-Ramos et al., 2006; Avila, 2006). Although the tau aggregates in AD are heavily ubiquitinated, they are not degraded by the UPS. Moreover hyperphosphorylated tau has been shown to form cross-linked aggregates in the presence of HNE, suggesting a state of chronic increased cellular oxidative stress, that are able to inhibit the proteasome (Hernandez and Avila, 2007). It is worth mentioning that such aggregates can be found also in other neurodegenerative diseases, too, and even in the brains of people without any symptoms (Lee et al., 2001). This might point at other factors that are needed to turn these normal accumulations into pathologic structures. Two other main players in the formation of senile plaques of Ab-peptides are the transmembrane proteins presenilin-1 and presenilin-2 (PS1 and PS2). Both proteins play a major role in the maturation of APP and have been identified as a part of the c-secretase complex. An overexpression of PS1 and PS2 in HEK293 cells resulted in an increase of APPa and Ab40, while an inhibition dramatically decreased the c-secretase activity. The early onset form of AD seems to result from mutations in the APP protein or in the PS1 or PS2, all resulting in an overproduction of Ab and reduce significantly the ‘‘normal’’ product of
238
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
Fig. 33. A model of the c-secretase structure. This image displays a model of the, until now, identified components of c-secretase, the main protein, responsible for proteolytic cleavage of the amyloid precursor protein (APP) into an extracellular (Ab) and an intracellular (AICD) domain. Image according to Selkoe and Wolfe (2007).
APP-a-secretase cleavage APPa. PS1 and PS2 are cleaved by presenilinase (Xia, 2008; da Costa, 2005) into a 18 kDa C-terminal (CTF) and an 28 kDa N-terminal (NTF) fragment, while overexpression of the C-terminus induced an increase of APPb (in HEK 293 cells). CTF and NTF remain associated as a heterodimeric structure. This CTF–NTF-complex turned out in co-immunoprecipitating experiments to be part of a large complex containing the glycoprotein nicastrin (plays a role in substrate recognition), anterior pharynx-defective 1 (Aph-1, assembling factor of the whole complex, binding first nicastrin and after this presenilin and Pen-2), presenilin enhancer 2 (Pen-2, triggers the cleavage of presenilin to its active NTF-CTF heterodimer). NTF and CTF, while the latter forms the active site of the c-secretase, is a complex essential for the life of multicellular organisms (Selkoe and Wolfe, 2007) and responsible for the cleavage of many type 1 membrane proteins (Fig. 33). An overexpression of all four components resulted in an increased c-secretase activity. Mutations of PS1 or PS2 might cause minor but continuous accumulation and finally devastating changes of APP-cleavage. Until now about 160 of such mutations are known, causing a positive shift of the ratio Ab42:Ab40, that induces accumulation of Ab and thus the pathology of AD. Even if the four mentioned proteins are required and sufficient for c-secretase functionality, further proteins have been found, that might fulfill a regulator function and seem to be constitutive parts of the c-secretase complex: CD147 (Zhou et al., 2006, 2005), TMP21 (Liu et al., 2008; Chen et al., 2006) and Rer1p (Spasic et al., 2007). Removal of CD147 from c-secretase results in an increased formation of Ab-peptides, while overexpression leads to a quick degradation of Ab. Despite this, CD147 seems to be involved as a stimulator in the expression of matrix metalloproteinases (MMPs) and thus indirectly regulate the amount of extracellular Ab that can be degraded by MMPs (Vetrivel et al., 2008). TMP21 reduction significantly increases the amounts of Ab40 and Ab42. ‘‘Retention in endoplasmic reticulum 1’’ (Rer1p) is a so-called ‘‘retrieval receptor’’ and an interaction partner of nicastrin’s transmembrane domain. This domain is needed for the interaction of nicastrin with Aph-1 and thus Rer1p competes with the binding of nicastrin to the c-secretase complex, and therefore, Rer1p has turned out to be a limiting factor of c-secretase activity. Other studies even revealed an interaction of Rer1 with free (not assembled to c-secretase) Pen-2 (Kaether et al., 2007); Rer1 overexpression stabilized Pen-2 in its unassembled state, while knockdown increases the amount of Pen-2 in the cell’s surface. All mutations that are known until now leading to the AD pathology are located in either the substrate protein (APP) or the presenilins, releasing different Ab-isoforms. After this short introduction into the biochemical network of proteins involved in the AD pathology, the role of the proteasome and the UPS will be highlighted. Considering the fact that the UPS is responsible for virtually the half-life regulation of every cellular protein, the role of the proteasomal system may be manifold. Looking at the abnormal accumulation of protein that is found in a polyubiquitinated state but still not degraded suggests a problem with the UPS. For the involvement of the UPS in the pathology of AD both genetic and biochemical evidence is available (Oddo, 2008). In AD Ab-plaques and hyperphosphorylated tau-filaments are both found polyubiquitinated but not degraded. Also it has been revealed that these formation contain a mutant form of ubiquitin (ubiquitin-B mutant, UBB+1), that shows a 19 amino acid extension at its Cterminus and block Ub-mediated protein degradation in neuronal cells (Lindsten et al., 2002). It is a hypothetic contributor to the neurotoxicity of Ab (Song et al., 2003). Moreover a de-ubiquitinating protein, the ubiquitin carboxy-terminal hydrolase
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
239
L1 (UCH-L1) (Setsuie and Wada, 2007; Luchansky et al., 2006) is found oxidatively modified and downregulated in early stages of AD (Castegna et al., 2002a,b). One characteristic of the Ab protein is its ability to inhibit (ubiquitin-dependent) proteasomal degradation. While the N-terminal sequences Ab1–11 and Ab1–16 as well as Ab25–35 showed proteasomal inhibition, the most effective inhibitor was still the holoprotein Ab40. A more recent study has shown a proteasomal inhibition by Ab42, too (Oh et al., 2005) suggesting a possible role of Ab as a proteasomal inhibitor. The inhibitory effect results from binding of Ab to the inner chamber of the proteasome as revealed by electron microscopic investigations (Gregori et al., 1997), while all three proteasomal activities were reduced by both oligomers in a dose dependent manner (Lopez et al., 2003). However, Shringarpure et al. (2000) demonstrated that the inhibitory effect of Ab depends clearly on the cross-linking status of the peptide. Exposure of cultured cells to Ab42 induced the cellular expression of E2-25K (also termed as Hip-2), an E2 ubiquitin-conjugating enzyme, that is found in AD colocalized with the ER-resident caspase-12 (Song et al., 2008). Despite of caspase-12 stabilization by proteasomal inhibition, E2-25K activated the enzyme, inducing apoptotic cell death. Cortical neurons lacking E2-25K turned out to be remarkably resistant to Ab-mediated cytotoxicity and caspase-12 triggered cell death. A similar function of E2-25K is found in the pathology of PD, directly interacting with huntingtin and inducing aggregate-formation (see Chapter 4.6.5) (de et al., 2007). It might be possible that the decreased degradation of Ab is in part due to the known age-dependent reduced proteasomal activity (Petropoulos et al., 2000; Breusing and Grune, 2008). But not only Ab is proteasomal degraded: PS1 and PS2 (Marambaud et al., 1998; Kim et al., 1997), as well as other proteins found in the c-secretase complex like nicastrin (He et al., 2007), Aph-1 (He et al., 2006), Pen-2 (Bergman et al., 2004) and TMP21 (Liu et al., 2008) are all proteasomal degraded or their cellular amounts are regulated by the UPS. The proteasomal pathway is not only involved in the Ab processing but also in the tau protein turnover. PHFs showed the ability to bind and inhibit the proteasome, resulting in significant decrease of proteasomal activity in key-regions of AD pathology: Compared to age-matched control patients, the proteasome activity was reduced to 56% in the straight gyrus, tough the amount of proteasomes found was not affected (Keck et al., 2003). Hippocampus and parahippocampal gyrus showed a remaining activity of 48%, the superior and middle temporal gyri 38%, the inferior parietal lobule 28% (Keller et al., 2000), both occipital lobe or cerebellum (a less AD-susceptible region) were not affected. The inhibition of the proteasome alone was sufficient to induce neurodegeneration and suggests to play a major role in AD-related neuronal damage (Keck et al., 2003). Mishto et al. (2006) showed an increased expression of immunoproteasomes in AD-affected brain regions (hippocampus and cerebellum in both neurons and astrocytes) and in non-demented elderly patients, while the expression of immunoproteasomes in younger brains is very low, but this study excluded an influence of the investigated b1i-polymorphism with the onset of AD, even if the proteasomal activity in the brain was influenced. However, the proteasome-mediated turnover of tau was inhibited by the hyperphosphorylation of the protein and the following formation of insoluble aggregates. While recombinant tau was quickly degraded by the proteasome, independent of its state of oxidative modification, the protein phosphatase 1 (PP1A) and 2 (PP2A) inhibitor okadaic acid (OA) instead almost completely blocked tau degradation in HT22 cells (a mouse hippocampal tumor cell line) (Poppek et al., 2006). The use of CFP- and GFP-labeled tau protein showed a significant decomposition of the fibrillar structure formed by tau in the cytosol, when hyperphosphorylation was induced (Chohan et al., 2005; Alonso et al., 2001). Tau seems to be polyubiquitinated by ‘‘carboxy-terminus of heat shock protein70-interacting protein’’ (CHIP) (Hatakeyama et al., 2004), a specific E3 protein, for UPS-mediated degradation. Soluble and hyperphosphorylated tau tends to form insoluble aggregates in the brains of CHIP-knockout mice (Dickey et al., 2006). The current hypothesis explaining the inability of CHIP to mediate tau degradation is a possible modification of the CHIPbinding site on tau that disables polyubiquitin-labeling to mark it for degradation by UPS (Keck et al., 2003). Although, this might not be the single pathway of tau degradation, since the tau protein has not a completely folded structure and might be therefore, a substrate of the 20S proteasome without prior unfolding (Dorval and Fraser, 2006). Delayed degradation of tau, promoting its hyperphosphorylation, may also be due to detracted proteasomal activity by Ab, since already aggregated tau cannot be degraded after restoring or increasing the proteasomal activity (Oddo et al., 2004). Considering the diminished proteasomal activity during aging (Stadtman and Levine, 2000; Stadtman, 2006), the scenario of an enhanced accumulation of tau and Ab in higher age groups, seems to likely. 4.6.2. Huntington’s disease (HD) Huntington’s disease is a progressive neurodegenerative disease, mainly characterized by motoric dysfunctions, cognitive decline and psychotic symptoms. In humans psychological symptoms and personality disorders occur about three years (averaged) before the motoric ones appear (Ortega et al., 2007). The psychological symptoms can comprise subcortical dementia, depression, mania, affective and personality changes as well as memory defects (Haddad and Cummings, 1997; Rosenblatt and Leroi, 2000). While the progression of HD, severe neuronal loss occurs in thalamus, hippocampus, spinal cord and other brain regions (Heinsen et al., 1994; Vonsattel et al., 1985). Until now no prevention or treatment for HD is known, only the symptoms can be eased. The typical histological markers of HD are the so-called inclusion bodies (IBs), aggregates formed of polyubiquitinated proteins, the main protein of the aggregates is mutant huntingtin protein that shows an extended polyglutamine (polyQ) sequence at its N-terminus, resulting from an insertion of multiple CAG-triplets in the gene coding the huntingtin protein (autosomal dominant inherited). Additional to HD eight further neurodegenerative diseases are known, that are caused by the insertion of polyQ-sequences in the according expressed proteins: the spinocerebellar ataxias (SCA) 1 (Netravathi et al., 2009; Schmitz-Hubsch et al., 2008), 2 (Mutesa et al., 2008; Lastres-Becker et al., 2008), 3 (Koyama et al., 2008; Socal
240
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
et al., 2008; Kieling et al., 2008), 6 (Craig et al., 2008; Howell et al., 2006; Khan et al., 2005), 7 (Lin et al., 2008; Rub et al., 2008), and 17 (Stevanin and Brice, 2008; Mariotti et al., 2007; Hubner et al., 2007), the dentatorubro-pallidoluysian atrophy (DRPLA) (Tsuji, 2007; Hashi et al., 2007; Yazawa et al., 1999) and finally the spinobular muscular atrophy (SBMA) (Tokui et al., 2008; Suzuki et al., 2008; Sobue, 2003; Zoghbi and Orr, 2000; Ross, 2002). Normally huntingtin shows a polyQ extension of about 6–35 glutamine residues, while clinical symptoms occur, when more than 40 residues are attached (The Huntington’s Disease Collaborative Research Group, 1993; Andrew et al., 1993). Both the onset of HD-pathology and the gravity of the symptoms depend directly of the length of the attached polyQ, at the same time with an increase in the tendency of aggregate formation. The mentioned IBs show a fibrillary or ribbon like structure (Scherzinger et al., 1997), that can be immunolabeled using antibodies specific for the N-terminal end of huntingtin (DiFiglia et al., 1997; Gutekunst et al., 1999). Usually found in the cytosol as neutropil inclusions, low amounts of aggregated huntingtin are found in the pericaryon or the nucleus: about 1–4% of all neurons in the striatum, the most affected brain region, show nuclear inclusions (Gutekunst et al., 1999). Since the loss of function in mutant huntingtin is only partial, the toxicity seems to be due to its aggregation, but the toxic gain of the protein is still unknown, as well as the exact function of huntingtin. Possible functions of huntingtin may be an involvement in transcription (Sugars et al., 2004), mitochondrial damage (Cevedo-Torres et al., 2009, 2000), regulation of axional transport (Gunawardena et al., 2003), vesicle transport and endocytosis (Sugars and Rubinsztein, 2003), apoptosis (Friedlander, 2003; Hickey and Chesselet, 2003; Rangone et al., 2004) and even the UPS (Diaz-Hernandez et al., 2006; Valera et al., 2005). The function and causes of the aggregate are still under discussion. Poly-glutamine aggregate in vitro were not able to inhibit proteasomal function, while fibrillary huntingtin from mice brains were (Diaz-Hernandez et al., 2006; Jana et al., 2001; Goswami et al., 2006; Rangone et al., 2005). Large aggregates of huntingtin instead were not able to inhibit the proteasome. This suggests that the surface of the aggregates exposing potential proteasome-binding structures may play a role, according to the proteasomal inhibition model of lipofuscin (see Chapter 4.4.2). Thus the formation of large aggregates may show protective effects decreasing the amount of free huntingtin and at the same time the accessibility of potential inhibiting structures. That may result in a slower decrease of proteasomal activity that has been shown to reveal devastating effects. As mentioned the aggregates found in HD are polyubiquitinated, suggesting an impairment of the UPS. Furthermore additional inhibition of the proteasome increases the rate of aggregate formation (Waelter et al., 2001). As found in many different neurodegenerative diseases, like PD, AD, ALS (see the corresponding Chapters in 4.6) the activity of the UPS is significantly decreased combined with an intracellular formation of polyubiquitinated protein aggregates, that can be immunolabeled with ubiquitin specific antibodies (Goedert et al., 1998; Sherman and Goldberg, 2001). Ortega et al. (2007) hypothesized that this might be explained with an interaction of the additional polyQ-sequence at huntingtin with the gating a-rings of the 20S ‘‘core’’ proteasome that are responsible for substrate recognition (see Chapter 2.1), an interaction with the 19S regulator cap, blocking the formation of the 26S proteasome or substrate recognition and defolding by the already assembled 26S. The inclusion bodies in HD have been shown to interact with proteins of the UPS like 20S, 19S and ubiquitin (Johnston et al., 1998; DiFiglia et al., 1997; Davies et al., 1997; Cummings et al., 1998; Kopito, 2000), possibly causing a decrease of the UPS functionality, again increasing the amount of cellular huntingtin and its aggregates. Thus the formation of IBs would increase, reducing the amount of free mutant huntingtin and restore the functionality of the UPS, at least partially (Ortega et al., 2007). The huntingtin aggregates have been shown not to inhibit the UPS and thus the formation of aggregates may inherit a cytoprotecting function (Diaz-Hernandez et al., 2006). How this idea has been proven in large parts experimentally will be described in the following. As mentioned the proteasome has three different specificities, for acidic, basic and hydrophobic amino acid residues in a substrate protein. Glutamine instead shows more neutral characteristics. Thus the group of Venkatraman (Venkatraman et al., 2004) constructed polyQ sequences with length between 10 and 30 residues, flanked by a lysine residue at C- and N-terminus in order to increase solubility of the fragment. Both the mammalian 26S and 20S proteasome turned out to be unable to cut the oligopeptides containing 9-29 glutamine residues; only the flanking lysine residues were cleft. The archaea proteasome from T. acidophilum in contrast was able to degrade the polyQfragments due to its 14 active sites with lower specificity. Since the so-called polyQ diseases are caused by glutaminesequences in the range from >40 to 300 residues, the normal fragment length produced by the mammalian proteasome (2–25 amino acids) is exceeded, that may result in mechanical problems to release that peptide sequences resistant to degradation. Thus exactly the polyQ peptides are most prone to aggregate-formation, the proteasomal degradation may produce and release peptides that are more toxic to the cell than the polyQ-containing protein itself. Further experiments with recombinant polyQ-containing proteins in vitro showed no inactivation of the 26S proteasome (Bennett et al., 2005; Chen and Wetzel, 2001; Poirier et al., 2002). Though the in vitro applied substrates were not polyubiquitinated, what may contribute to the inhibiting characteristics of the peptides and especially with their ability to interact with the 19S regulator cap. To investigate this Diaz-Hernandez et al. (2006) tested the ability of polyubiquitinated IBs and huntingtin-filaments in a polyubiquitinated state extracted from mutant mice (Tet/HD94 mice expressing exon 1 of huntingtin with polyQ chain of 94 residues with a tetracycline-controlled promotor (Diaz-Hernandez et al., 2003, 2005; Yamamoto et al., 2000)). The remaining 26S proteasomal activities were determined using fluorogenic substrate proteins, specific for the single proteolytic active subunits: suc-LLVY-AMC for b5, suc-LSTR-AMC for b2 and z-LLE-bNAP for b1. In fact, the degradation-rate of the polyubiquitinated test substrate (Ub)n-IjBa was reduced by the polyubiquitinated huntingtin-filaments, but not by the ‘‘massive’’ IBs. Those results suggest an interaction of the polyubiquitinated huntingtin with the 19S regulator cap of the 26S proteasome that was confirmed by electron microscopy (Diaz-Hernandez et al., 2006). Thus polyubiquitination of the mutant huntingtin, respectively polyQ-fragments seems to be ‘‘needed’’ for UPS inhibition.
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
241
Similar results were gained in experiments with N2a cell-homogenates, neural mouse cells, stable transfected to express huntingtin with different length (16, 60 and 150 residues) of attached polyQ-sequences: after a centrifugation step only the supernatant soluble fraction was able to inhibit the 26S proteasome (Jana et al., 2001), the precipitated fraction was unable to. The amount of 20S ‘‘core’’ proteasome was not changed in the disease, although the activity of b1 and b5 was changed, suggesting an exchange of the constitutive subunits for the interferon-c inducible ones (in Tet/HD94 mice), maybe as a consequence of neuroinflammatory processes during HD-pathology (Diaz-Hernandez et al., 2004). Although Seo et al. (2004) published a study showing contrary results, i.e. a decreased activity of the 20S proteasome (both in b1- and b5-activity). The overall results show a clear decrease and inhibition of the UPS both in mutant mice models and human HD-patients, but whether the activity of the 20S ‘‘core’’ proteasome is actually changed, is still under discussion (Ortega et al., 2007). 4.6.3. Amyotrophic lateral sclerosis (ALS) As already described, the recognition and removing of (posttranslational) modified proteins is essential for functionality and surviving of every cell, especially in postmitotic aging tissues. One essential aspect of preventing the accumulation of modified proteins and the resulting formation of insoluble protein aggregates (though many of them may be polyubiquitinated) (Watanabe et al., 2001; Ciechanover and Brundin, 2003) is the functionality of the cellular antioxidative system. Like many other neurodegenerative diseases amyotrophic lateral sclerosis (ALS) is accompanied by the formation of protein aggregates, mainly resulting from mutant Cu,Zn superoxide dismutase 1 (SOD1) (Cookson et al., 2002; Jaiswal and Keller, 2008; Gaudette et al., 2000; Jaarsma et al., 2001), an important antioxidative enzyme, converting the superoxide anion ðO 2 Þ to hydrogen peroxide (H2O2). Naturally both composition and rate of aggregate formation are dependent of the corresponding pathology. In ALS the mutation of an essential part of the antioxidative defense system both detracts proteasomal activity and increases the amount of free radicals by reducing the conversion of the primary radical superoxide to hydrogen peroxide. The lumbal spinal cord motor neurons are particularly susceptible for the aggregation of SOD1, while in these cells protein-chaperoning and proteasomal degradation is reduced at the same time (Kabashi and Durham, 2006). SOD1-aggregates are found both in motor neurons and neighboring astrocytes (Kato et al., 2000; Shibata et al., 1996) in humans as well as in rodent models (Durham et al., 1997). Whether the observed aggregates are protecting cells from misfolded proteins is still under discussion; despite of being not inherently toxic, they detract ‘‘functional volume’’ of the cell and even if this is a slow process they finally cause cell death by filling gradually the whole cytosol. The fact that the amount of intracellular mutant SOD1-protein aggregates negatively correlates with cell functionality has been experimentally proven in many cases (Matsumoto et al., 2005; Stieber et al., 2000; Sasaki et al., 2005). If the formation of those aggregates was reduced, the cell viability increased at the same time (Bruening et al., 1999; Takeuchi et al., 2002). Furthermore, a typical protein aggregation both decreases proteolytic (proteasomal) activity of the cell and is a direct result of that decrease. To the current model of ALS contributes the toxicity of mutant SOD1: the functional homodimer SOD1 is destabilized and polymers are formed by SOD1-monomers, promoted by one of the more than hundred disease causing mutations (Kabashi and Durham, 2006). The main factors seem to be demetallation, disulphide reduction and oxidative modifications of SOD1 (Cardoso et al., 2002; Tiwari and Hayward, 2003; Urushitani et al., 2002). If the mutation is responsible as single factor or accompanied by further posttranslational modifications is still unknown. The most important modifications found in ALS patients are oxidation, nitration, and HNE-adduction (Poon et al., 2005; Perluigi et al., 2005). Cells that retain their ability of both chaperoning and thus support of native proteins for proper folding and proteasomal degradation of mutant SOD1 are less susceptible for the effects of ALS (Kabashi and Durham, 2006). As mentioned the main task of chaperones is the support of nascent proteins in reaching their proper state of folding, respectively if this is not (longer) possible in passing the misfolded protein to the UPS for terminal degradation (see Chapter 3.2), while passing is performed by the heat shock proteins Hsp40 and Hsp70, whereas the latter one hands over the proteins to the UPS in a CHIP mediated way (Kumar et al., 2007; Meacham et al., 2001). The UPS-mediated degradation of the mutant SOD1 in ALS is realized via the Hsp70-CHIP pathway (Choi et al., 2004; Urushitani et al., 2004; Ishigaki et al., 2007). Experiments showed that only mutant SOD1, but not the wild type protein induces the formation of aggregates during proteasome inhibition (Kabashi and Durham, 2006), as well as only the mutant form seems to be polyubiquitinated in neuronal cells (Basso et al., 2006; Ganesan et al., 2008), mediated by the E3 ligases dorfin, CHIP and NEDL1 (Miyazaki et al., 2004; Li et al., 2008). These results suggest that even the susceptible motor neurons are capable of degrading mutant SOD1, if the efficiency of the UPS is ensured. But within the normal aging process or increased oxidative stress the UPS capacity might decrease. Experiments with SOD1G93A transgenic mice showed a decrease in every single proteasomal activity, with a decrease of about 50% of the chymotrypsin-like activity (b5-subunit) after 75 days of life (Kabashi et al., 2004). Furthermore, not only the amount of b5 was reduced, but also the levels of the proteasomal a-subunits, surprisingly not due to reduced mRNA or an increased amount of unassembled subunits (Kabashi and Durham, 2006), pointing to the possibility of posttranslational modification that prevents the incorporation of a-subunits, that initiate the formation of the proteasome assembly. Thus the amount of cellular available 20S proteasome seems to be reduced during the pathologic progression of ALS. Moreover the experimental results suggest that different cells in different regions of the spinal cord are differently affected. Most affected are astrocytes and glial cells of the spinal cord, as found both in transgenic mice (SOD1G93A) and human ALS patients, expressing increased amounts of the IFN-c inducible form of the proteasomal subunit b5i and the 11S proteasomal regulator particle (Kabashi et al., 2004; Puttaparthi and Elliott, 2005), while the loss of chymotrypsin-activity seems to be due to the
242
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
increased formation of immunoproteasomes (Kabashi et al., 2004). Though substitution of constitutive subunits seems not to be sufficient to explain the whole decrease of proteasomal activity (Kabashi et al., 2004). Under discussion are posttranslational modifications of single proteasomal subunits (Kabashi and Durham, 2006). Kabashi et al. (2004) pointing out that often the proteasomal activity is determined in homogenates of the whole spinal cord and has not been determined for single cell types involved in ALS-pathology. In vitro studies underlay the significant role of the proteasome, since its inhibition, as mentioned, increases the amount of mutant SOD1-aggregates formed over time (Johnston et al., 2000; Niwa et al., 2002; Puttaparthi et al., 2003) in motor neurons and interferes with mitochondrial homeostasis (Ling et al., 2003; Sullivan et al., 2004) as found both in human ALS patients and two mutant mice models (SOD1G93A and SOD1G37R) (Hervias et al., 2006). The exact mechanism of the latter effect, the links between the proteasome, respectively the UPS, and mitochondria (Torres and Perez, 2008; Radke et al., 2008; Papa and Rockwell, 2008), is still investigated. Until now it has been shown that proteins in the outer mitochondrial membrane are degraded by the UPS (Neutzner et al., 2007) in a way reminding to the already described ERAD (see Chapter 4.2), thus the term ‘‘OMMAD’’ has been suggested, while recognition and degradation of these proteins seems likely to be realized via certain E3 ubiquitin ligases (see Chapter 3.2). One of these proteins identified is Fzo1p (Neutzner and Youle, 2005), a transmembrane GTPase, that regulates the fusion of outer mitochondrial membranes (Cohen et al., 2008). Furthermore apoptotic cell death is connected to the proteasomal activity since many apoptotic factors are regulated via the proteasome and accumulate in sufficient amounts to trigger cell death (Almond and Cohen, 2002). Though it is unclear why the decrease of proteasomal activity is mainly found in motor neurons and less in other cells. The answer may be found in the functionality of these specialized neurons: if the AMPA glutamate receptors are blocked, the viability of the cells increases and the amount of mutant SOD1 inclusions formed is reduced (Yin et al., 2007; Tortarolo et al., 2006; Tateno et al., 2004; Rembach et al., 2004). Despite of the causes for reduced proteasomal activity, it appeared obvious to test whether an increase of the proteasomal activity is improving the clinical conditions: the human ciliary neurotrophic factor (CNTF) (Giess et al., 2002; Bongioanni et al., 2004) has been used in clinical trials, since it has been revealed the ability to increase the amounts of 20S proteasomal mRNA. Although recent clinical trials revealed no correlation between CNTF treatment and the progression of ALS, while high doses showed some side effects (Jho et al., 2004). Another experimental way was the stimulation of the Keap1–Nrf2 pathway (one of the most important antioxidative defense mechanisms) (Tanigawa et al., 2007; Zhang, 2006; Kobayashi and Tong, 2006), upregulating both antioxidative enzymes (via antioxidant-responsive element/electrophile-responsive element, termed as ‘‘ARE/EpRE’’) and proteasomal subunits (Philbert et al., 1991), since Nrf2 mRNA was downregulated in ALS affected motor neurons (Kirby et al., 2005). 4.6.4. Friedreich ataxia (FA) Friedreich ataxia is the most common autosomal recessive hereditary ataxia, caused by a mutation in the key gene FRDA (Sarsero et al., 2005). That neurodegenerative disease is caused by a repeated insertion of the trinucleotide GAA of the first intron of the FRDA gene, resulting in a reduced (5–35% (Coppola et al., 2006)) intramitochondrial expression of frataxin. The exact function of frataxin is still unknown, experimental results and hypotheses suggest involvement in the formation of the mitochondrial iron–sulfur clusters (Guillon et al., 2009), the mitochondrial iron storage (Gakh et al., 2002) or antioxidative defense (in part via activation of glutathione peroxidase) (Auchere et al., 2008; Napoli et al., 2006; Shoichet et al., 2002). An overexpression in drosophila significantly increased oxidative stress resistance (Runko et al., 2008). The involvement of frataxin in the formation of iron–sulfur clusters was experimentally verified by Zhang et al. (2006). Further results suggest an involvement of frataxin in the respiratory chain (Gonzalez-Cabo et al., 2005). Usually the first symptoms, increased oxidative stress, hypertrophic cardiomyopathy (one of the main causes of FA-caused death (Van Driest et al., 2005; Cuda et al., 2002)), progressive ataxia, lack of tendon reflexes in the legs, dysarthria, areflexia, neurodegeneration and sensory loss (Ponka, 2004), manifest before the age of 25. FA-models of Saccharomyces cerevisiae, expressing mutant frataxin (the yeast YFH1 is the orthologue of human FRDA), revealed about 10-fold increases of iron content found in mitochondria compared to wild type cells (Babcock et al., 1997). Recent investigations, using the yeast FA-model, too, underline the mayor role of free radicals and oxidative stress in the disease (Calabrese et al., 2005). Bulteau et al. (2007) showed in the yeast model (DYFH1) that increased oxidative stress in FA is directly dependent on the amount of available oxygen and that oxidative stress is not only limited to the mitochondria, but found throughout the cytosol, where it causes the largest part of the phenotypical metabolic aberration in FApathology. Furthermore, the amount of oxidative stress seems to be not directly linked to the increase of iron accumulated in the mitochondria, even though a Fenton-mediated mechanism has been assumed by several groups (Calabrese et al., 2005). Nevertheless, one of the initial events in the disease might be oxidative stress that seems to be caused by leaks in the respiratory chain and insufficient formation of iron–sulfur clusters (Bulteau et al., 2007). Another argument for this hypothesis is the fact that tissue containing large amounts of mitochondria is particularly affected, including heart, liver and skeletal muscle cells (Becker and Richardson, 2001). In mitochondria the main proteolytic machinery for the removal of damaged proteins is the Lon protease (Bayot et al., 2008; Ngo and Davies, 2007) (in S. cerevisia the ATP-dependent Pim1 (Major et al., 2006)) and not the proteasome. In FA-pathology the activity of Pim1 is significantly increased, while the proteasomal activity decreases, nevertheless the protein amounts of both proteases remain unchanged. Considering the fact, that oxidized proteins although accumulate in the mitochondria, is suggesting that the amount of protein oxidation overwhelms the proteolytic Pim1-capacity. Since the subunit composition of the proteasome is not changed in FA-progression (Bulteau et al.,
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
243
2007), the decreased proteasomal activity seems to be due to oxidative damage of the proteasome. The cause may be the already described inhibition of the proteasome by heavily oxidized and covalently cross linked proteins (see Chapter 4.4) as found in conditions of severe oxidative stress as well as the direct inhibition of the proteasome by HNE. The increased intracellular accumulation of oxidized proteins, HNE and HNE-modified proteins has been shown, too (Bulteau et al., 2007). In contrast to the yeast FA-model in mammals (mice-model) no increased intramitochondrial accumulation of oxidatively modified proteins was found by Seznec et al. (2005), a fact that is explained with a more efficient Lon protease-mediated degradation in mammals than in yeast by Bulteau et al. (2007). This murine FA-model (complete frataxin-deficiency) even showed both no increased oxidative stress and no effects of a Cu, ZnSOD-overexpression or of application of the MnSOD mimetic MnTBAP (Cuzzocrea et al., 1999a,b) on the development of murine FRDA cardiomyopathy; in neuronal cells no oxidative stress was induced and the intramitochondrial MnSOD level remained unchanged. Thus Seznec summarized that despite of the iron accumulation in mitochondria FA is not associated with oxidative damage (Seznec et al., 2005). One explanation may be the chemical properties of iron to be precipitated in an amorphous form as ferric phosphate (Fe(III)PO4), and thus in a state that is less accessible to biological processes, since iron has to be in its reduced state (Fe2+) to drive Fenton chemistry (Lesuisse et al., 2003). 4.6.5. Parkinson’s disease (PD) Parkinson’s disease is the second most common neurodegenerative disorder after AD (see Chapter 4.6.1). It affects about 1% of the population >65 years of age, while 95% of all patients suffer the late onset sporadic form, while the early onset form is mostly found in the familiar form of PD (some 5% of all cases) (Chinta and Andersen, 2008). The common clinical symptoms are rigidity, resting tremor and bradykinesia (Chinta and Andersen, 2008). These symptoms are mainly caused by the specific degeneration of the so-called nigrostriatal doparminergic neurons found in the substantia nigra pars compacta (SNpc). The first symptoms occur, when the dopamine levels are reduced to 40% of the normal amount (Bernheimer et al., 1973). The main histological marker is the intracellular cytosolic formation of Lewy bodies (LB). The main protein forming LBs is a-synuclein (Kawahara et al., 2008; Lee, 2003), as in most of all neurodegenerative diseases in a polyubiquitinated state. A missense mutation of a-synuclein is responsible for some early-onset forms of familial PD. The exact cellular function of a-synuclein has still to be figured out yet, but there are hypotheses about involvement in neuronal plasticity or neurotransmitter release (Di et al., 2003), further studies have shown that mutations in this protein strongly increase the susceptibility of neuronal cells to oxidative stress (Jiang et al., 2007). Another protein, that forms aggregates in PD is parkin (containing 465 amino acids, summarizing to an overall mass of about 52 kDa) (Moore, 2006; Dawson, 2006), normally functioning as an E3-ligase of the UPS, interacting with the RPN10(or S5a)-subunit of the 19S regulator particle (Sakata et al., 2003). Mutations in the parkin-coding gene are a widespread cause for the autosomal recessive hereditary form of PD. Finally the third main protein found attached to aggregates is the ubiquitin carboxy-terminal hydrolase L1 (UCHL1, 230 amino-residues and a mass of about 26 kDa). This deubiquitinating enzyme is exclusively found in neurons and hydrolyzes poly-Ub-sequences into single monomers (Wilkinson et al., 1989). Until now 11 genes have been identified that are responsible for the familial form of PD: SNCA, PARK2, -3, -4, -5, -6, -7, -8, -9, -10, and NR4A2. One of the most common forms the familiar forms of PD seem to be due to a mutation in the parkin gene (PARK2) (Kawahara et al., 2008). Despite of the protein aggregation in PD oxidative stress seems to be one of the main actors, even if it is still unclear whether the increased formation of ROS is the initiating event or just resulting from cellular changes due to the PD-pathology (Chinta and Andersen, 2008); nevertheless this and a reduction of complex I of the mitochondrial respiratory chain in cells of the substantia nigra (SN) point into the direction of a mitochondrial involvement in PD (Parker et al., 1989; Schapira et al., 1989; Mann et al., 1992; Mizuno et al., 1989). Damage of complex I may result both in decreased levels of cellular ATP, with consequences in ubiquitination and protein degradation via the UPS, de novo synthesis of proteins, and energy supply of the cellular antioxidative systems; furthermore increased formation of ROS is worsening the functioning of the cell (Beal, 2005). ROS found in PD are superoxide, peroxynitrite (resulting from the reaction of superoxide and nitric oxide) and hydrogen peroxide. As ‘‘markers’’ of oxidative damage to biological structures protein carbonyls occur, at the same time as 3-nitrotyrosine modifications (mostly a result of peroxynitrite-mediated protein modification), products of lipid peroxidation like MDA and HNE, and an oxidative DNA modification, 8-hydroxy-20 -deoxyguanosine. In dopaminergic cells dopamine itself plays an important role, too. Due to its chemical instability the molecule can decompose, releasing superoxide anions, a reaction that can be catalyzed by transition metals (Jenner and Olanow, 1996; Youdim and Riederer, 1997): in the substantia nigra the amount of iron is increased (Berg et al., 2004). This can be induced by the release of iron from iron containing proteins. Iron can be released from ferritin by superoxide, from cytochrome oxidase by peroxides and from the intramitochondrial Fe–S-clusters by peroxynitrite (Kaur and Andersen, 2004). At the same time, the GSH-levels in affected cells are significantly reduced. All this findings indicate an increase of oxidative stress (Jenner, 2003; Jenner, 1996). As mentioned the mitochondria seem to play a major role in PD. One possibility is the influence of mitochondrial inhibition at complex I, a complex that seems to be damaged in PD, using the neurotoxin 1-methyl-4-phenylpyridinium (MPP+). The result was a change in proteasomal activity after 24 h of MPP+ incubation, shown by a decrease of the b5- and b1-activity (CanedaFerron et al., 2008, 2007). In dopaminergic neurons of the substantia nigra the amount of proteasomal a-subunits, but not b-subunits, has been shown to be decreased compared to an age-matched control group (McNaught et al., 2002). Since the a-subunits are essential for proteasomal assembly (see Chapter 2.3) this may cause a decrease in the available level of
244
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
functional proteasomes and thus a reduction of overall proteasomal activity. Furthermore, the expression of the regulator caps PA700 and PA28 were both reduced (McNaught et al., 2003), even if the exact mechanism of this phenomenon is still unknown. Other experiments pointing into the direction of a proteasome/UPS failure in PD have been performed by treating rats with the proteasome inhibitor lactacystin; the result was a quick progression of the somatic symptoms in PD and an increased loss of dopaminergic neurons after animal treatment with lactacystin; at the same time the formation of a-synuclein aggregates was induced (Lev et al., 2006). A recent study by Yuan et al. (2008) suggest an interaction between a-synuclein and nuclear factor jB (NFjB), a protein with antiapoptotic function and involved in neuronal inflammatory response (see Chapters 4.6.6 and 4.9). a-Synuclein turned out to be an effective inhibitor of the NF-jB cascade that suppresses apoptosis via expression of Bcl-1 and Bcl-2. The physical mechanism may be based on the fact that a-synuclein shares about 40% of structural homology with the chaperone 14-3-3, that is known to bind NF-jB. Another function of a-synuclein could be the triggering of the GSK3b expression, a protein with pro-apoptotic activity, that might interact with several anti-apoptotic transcription factors like b-catenin, c-jun, STAT, NFAT and CREB and by inhibition of the phosphorylation of the NF-jB binding protein IjBa (see Chapter 4.6.6). Proteasome inhibition followed by apomorphine treatment reversed behavioral changes (McNaught et al., 2002). This result suggests an important role of proteasomal impairment in pathology and progression of PD. Promising PD treatment attempts have been performed with the application of iron chelators like desferoxamine or V-28, a newer chelator able to pass the blood brain barrier. Another approach is the donation of GSH precursors like glutamyl cysteine ethyl ester (GCEE) or glutathione ethyl ester (GEE) in order to restore the normal GSH levels and thus the redox state of the cell. In the same way the thiol containing antioxidant a-lipoic acid was able both in vitro and in vivo to increase cellular GSH-levels (Bharat et al., 2002; Karunakaran et al., 2007). Another strategy is the pharmaceutical inhibition of the monoaminoxidases A and B (MAO-A and MAO-B), the major hydrogenperoxide generators in the substantia nigra (Youdim and Lavie, 1994). MAO-B inhibition by deprenyl has become a widespread approach in therapy, like the more effective rasagiline (Fernandez and Chen, 2007). 4.6.6. Stroke Stroke is considered to be one of the major causes of death and disability in the United States and European countries. The main events in stroke are the ischemic brain injury, primary generating oxidative stress, releasing cytokines activating the nuclear factor jB (NF-jB), followed by the secondary reactions in affected cells, the activation of inflammatory genes inducing the excretion of cytokines and adhesion molecules. Adhesion molecules (like the endothelial VCAM-1 (Norman et al., 2008; Zerfaoui et al., 2008), ICAM-1 (Montecucco et al., 2009; Marino et al., 2008) and E-selectin (Formigli et al., 1995; Takeda et al., 2002; Piconi et al., 2004)) fulfill a messenger function directing monocytes and leukocytes to cells and tissues that express inflammatory response. These secondary cellular reactions can induce further damage in a cascade that involves the UPS as a triggering and regulating system. Normally brain ischemia starts with an interruption of the cerebral blood flow, causing the inhibition of oxidative phosphorylation, and disturbances of the ATP-dependent ionic homeostasis of the affected cells (Lipton, 1999). After about 3 h several secondary reactions of cells are provoked, the most important one is a massive neuro-inflammatory response (Danton and Dietrich, 2003; Stanimirovic and Satoh, 2000). These inflammatory reactions are both detrimental and useful for survival of the affected cells and the surrounding tissue (Danton and Dietrich, 2003; Wang and Shuaib, 2002). Recent studies have investigated the pharmacological inhibition of the secondary inflammatory response by targeting the inflammatory mediators. Special attention has been paid to the transcriptional factor NF-jB. NF-jB activation lead to a polyubiquitination of the tumor necrosis factor receptor-associated factor 6 (TRAF6) (Deng et al., 2000; Wojcik and Di, 2004). TRAF6 (a RING finger-containing protein, acting as E3 ubiquitin ligase (Geetha et al., 2005)) is a signal transducer that is fit out with a lysine-63-linked chain of polyubiquitin by Ubc13 (also termed as Uev1A), a ubiquitin conjugating enzyme complex (Wooff et al., 2004). After that modification TRAF6 is able to activate IjB kinase (IKK) that phosphorylates the inhibitor protein IjBa that binds and masks NF-jB. The phosphorylated form of IjBa is polyubiquitinated by the E3 ligase SCFbTRCP and thus labeled for UPS-mediated degradation. The degradation of IjBa releases NFjB, which is translocated into the nucleus and binding to specific promoters (compare Chapter 4.1 and Fig. 20). Thus the UPS is involved in the inflammatory response, rendering the UPS to an effective target in the suppression of that cellular response. As mentioned the activation of the NF-jB dependent cascade starts at about 3 h after and lasts for about 72 h in the brains of rats with a transient middle cerebral artery occlusion (Nijboer et al., 2008). Despite of the inflammatory response in brain ischemia, NF-jB plays roles in rheumatoid arthritis (see Chapter 4.9), inflammatory bowel disease, multiple sclerosis, psoriasis (see Chapter 4.15.3) and asthma. Since NF-jB triggers the release of both inflammatory and anti-apoptotic cellular responses it is still under discussion whether the inhibition of NF-jB is useful or damage-promoting. Though pharmaceutical inhibition of the proteasome (and thus NF-jB) using the proteasomal inhibitor MLN519 (1.0 mg per kg of body weight via an intravenous injection) has shown cell protecting effects (Williams et al., 2006, 2004, 2003). Another important consequence of both oxidative stress (ROS) and pro-inflammatory cytokines is an infiltration of the damaged brain regions by leukocytes (neutrophiles and macrophages). Especially the leukocytes, attracted by released adhesion molecules (Wong and Crack, 2008) are associated with secondary tissue damage in stroke. This was proven in mice lacking NF-jB triggered adhesion factors (Wong and Crack, 2008), while the animals showed increased neuroprotection after ischemia and reperfusion. Furthermore, NF-jB is able to induce iNOS, a protein that produces in the highly diffusible NO-radical, at the same time when large amounts of superoxide are released by mitochondria with a damaged respiratory chain, due to the oxygen-deficiency while ischemia. After reperfusion the mitochondria resume oxidative phosphorylation
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
245
and the resulting superoxide (an overproduction of superoxide has been shown to induce NO production by both eNOS and iNOS, Hishikawa et al. (1995)) reacts with NO to the highly oxidative peroxynitrite (ONOO), causing further damage even to cells that were not affects by the previous ischemia. In mice lacking nNOS (Panahian et al., 1996; Huang et al., 1994; Shimizu-Sasamata et al., 1998) and iNOS (Iadecola et al., 1997), the infarct volume was significantly reduced after focal cerebral ischemia in comparison to wild type animals. Later ischemic tissue responses are less pronounced, if the proteasome is inhibited. Several groups have performed experiments with the proteasome inhibitor MLN-519 in brain postischemic reperfusion and gained promising results, as an increase of neuroprotection of about 60% and a decrease in neutrophil infiltration of 63–70% (Phillips et al., 2000), a reduced infarct volume and improved neurological recovery of the animals (Zhang et al., 2001), and a decrease in both neuronal and astrocytic degeneration, combined with a reduced infarct volume (48% compared to an untreated control), a better neurologic recovery (+51%), reduced tissue infiltration by neutrophiles (38%) and finally a reduction of free NF-jB (45%) (Williams et al., 2003). In those studies MLN-519 was applied 4–8 h after ischemia. Due to the complex activation cascades of NF-jB even a complete proteasomal inhibition is not sufficient for total suppression of NF-jB mediated post-ischemic cellular response. 4.7. Cystic fibrosis (CF) Cystic fibrosis (CF) is a hereditary autosomal recessive disease causing the expression of a misfolded cystic fibrosis transmembrane conductance regulator (CFTR) protein. The symptoms of CF, a result of premature UPS-mediated degradation of the functional FTR protein, respectively, the expression of a non-functional CFTR-mutant protein, are pancreatic failure and lung disease the main reason of patient mortality (Strong et al., 2005; Nash et al., 2008; Vender, 2008; Dodge et al., 2007; Bellis et al., 2007; Della et al., 2008; Banjar, 2003). CFTR is a chloride ion channel protein that is found on the surface of epithelial cells in lungs and gut working in an ATP-consuming manner, regulating Cl-transport (Ciechanover and Brundin, 2003; Lu et al., 2006; Zhou et al., 2006). The protein is build of 1480 amino acids containing two main ‘‘membrane-spanning domains’’ (MSD1 and MSD2), each of them formed of six ‘‘transmembrane domains’’ (TMD). Moreover it contains two cytoplasmic ‘‘nucleotide binding domains’’ (NBD1 and NDB2, responsible for binding and hydrolyzing of ATP) and a regulatory (R) region (regulated by cAMP) (Watanabe et al., 2001). The ion transport channel is formed by NBD 1 and 2 and the R domain. The proper folding of the nascent amino acid chain (A-form of CFTR) is slow and very inefficient, resulting in misfolding of 55–80% of the protein expressed in human cells (Shibata et al., 1996). Degradation of the misfolded protein is mediated via polyubiquitination (by an E1–E2–E3 catalyzed ubiquitin targeting) followed by degradation via the UPS. The correctly folded CFTR-protein (B-form) is incorporated into the ER-membrane and transported into the Golgi-apparatus by ‘‘coat protein complex II’’ (COPII)-coated vesicles (Kabashi et al., 2004; Yoo et al., 2002). Terminal posttranslational processing of the protein in the Golgi-apparatus results in the C-form of CFTR (Gelman and Kopito, 2002), that is transported to the cell surface. The CFTR-levels are regulated by incorporation in sub-apical vesicles that are either traced back to the cell membrane or degraded by the lysosomal system (Gelman and Kopito, 2002). About 600 different mutations are identified until now. The most common one is the DF508 CFTR found in about 70% of all patients (Kopito, 1999). An incidence of about 1:2900 births in the UK results in a large amount of patients (Raskin et al., 1997). The different mutants are subdivided into six different classes (classes I–VI): frameshifts, deletions and non-sense mutations that produced truncated CFTR-proteins (I); defects in intracellular trafficking (even if it shows channel-activity) (II); full-length proteins showing reduced or no activity (III); proteins with a slightly reduced activity that generate only minor symptoms (IV); functional proteins that are expressed in decreased amounts (V); and finally proteins that are expressed in normal levels but showing significantly reduced stability in the cell membrane (VI). The severe CF-phenotypes are associated with class I-III mutations, characterized by (almost) complete lack of CFTR-activity (Matsumoto et al., 2005; Rubenstein, 2005). The DF508 CFTR class II mutation (temperature sensitive) (Zielenski and Tsui, 1995) shows the normal channel function but is retained in the ER and retro-translocated into the cytosol for UPS-mediated degradation. Misfolded structures are recognized in the cytosolic part (Meacham et al., 2001; Cyr, 2005; Younger et al., 2004) as well as in the misfolded part located in the ER membrane (Tiwari and Hayward, 2003). The cytosolic part of CFTR, the three dimensional folded structure formed of NBD1/2 and the R domain is, if misfolded, recognized by Hsp70 and kept in a soluble state. After binding to CHIP, an E3-ubiquitin-ligase, the Hsp70–CHIP–CFTR complex promotes UPS-mediated degradation of CFTR. In this case polyubiquitination is carried out by the cytosolic E2 enzyme UbcH5a (Meacham et al., 2001; Younger et al., 2004; Saxena et al., 2005; Knutson et al., 2004). On the other hand, Hsp70 has been shown to play a role in the folding pathway of CFTR, since overexpression of Hsp70 increased the transport of the DF508 CFTR mutant to the cells surface (Choo-Kang and Zeitlin, 2001). The mechanism deciding between further maturation on the one side and CHIP binding of the Hsp70-bound CFTR, leading to UPS-mediated degradation, on the other side is still unknown. Two Hsp70-co-chaperones, BAG-2 and HspBP1, are able to prevent Hsp70 induced degradation of CFTR (Alberti et al., 2004; Dai et al., 2005). Misfolding of the ER-membrane located structure of CFTR is controlled by Derlin-1 (Sun et al., 2006; Younger et al., 2006). In human cells Derlin-1 overexpression caused retention of CTFR in the ER, while knockdown increased the amounts of CTFR-mutants (Sun et al., 2006). Derlin-1 has been identified as an important part of the already described ERAD (see Chapter 4.2). Both in yeast and human cells Derlin-1 bound CFTR has been found associated with the retro-translocation factors p97 (also termed as ‘‘Cdc48’’) (Meyer and Popp, 2008; Schuberth and Buchberger, 2008) or VIMP, the E2-ubiquitin ligase Ubc6e and the E3-ligases RMA1, HRD1, gp78 and finally the peptide N-glycanase (Poon et al., 2005; Perluigi et al., 2005; Casoni et al., 2005; Ye et al., 2005, 2004; Lilley and Ploegh, 2005), a deglycosylating enzyme. Derlin-1 knockout
246
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
experiments in yeast resulted in an accumulation of CFTR as an ERAD substrate in the ER lumen, making it obvious that Derlin-1 plays a role in retro-translocation. Experiments showed that a recovery of only 5% of the endothelial CTFR-activity significantly improves lung and gut function of CF-patients (Heink et al., 2005). Therapeutic approaches are obtained via the application of so-called ‘‘potentiators’’ or ‘‘correctors’’. The ‘‘potentiators’’ are able to increase the functionality of misfolded CFTR-channels like genistein that binds to the NBDs and stabilized their dimerization; another one is CFpot-532, though its molecular mechanism is still unknown. The ‘‘correctors’’ support proper CFTR-folding, resulting in an increased amount of functional transporter proteins in the cell surface. Therapeutic used ‘‘correctors’’ are VRT-325, Corr-3a and Corr-4a. These molecules enable the transport of the functional DF508 CFTR mutant to the cell membrane by reducing their ERAD degradation (Van Goor et al., 2006). Corr-3a and Corr-4a are proven to be very specific for the DF508 CFTR mutant but not for P547H or N1303K in human bronchial epithelial cells (Pedemonte et al., 2005), while VRT-325 increases the levels of the G601S mutant of the cardiac potassium channel hERG, causing human long-QT syndrome type 2 (Van Goor et al., 2006). Further therapeutic attempts may include application of combined ‘‘correctors’’ and ‘‘potentiators’’ or targeting of the UPS (especially the ubiquitinating-machinery) itself (Kabashi and Durham, 2006). 4.8. Atherosclerosis (AS) The process of development and progression of Atherosclerosis (AS) is very complex and again it is unclear if oxidative stress or a dysfunction of the UPS is the initial step. This might be due to the hypothesized first events, the transport of oxidized low-density lipoproteins (ox-LDL) across the thin layer of endothelial cell into the artery wall. Damage to the endothelium can be caused by ox-LDL itself, whereupon these cells excrete chemical factors attracting monocytes and promote mitosis (monocyte chemoattractant protein-1 and the macrophage colony stimulating factor (CSF)). Leukocytes and monocytes migrate into the subendothelian space, where the monocytes incorporate the lipoproteins and become macrophages, cells generating large amounts of ROS. ROS further oxidize ox-LDL, which converts macrophages into foam cells, if phagocytosed. Under participation of smooth muscle cells, macrophages and foam cells fibrous plaques are formed, growing into the vascular lumen, in later stages causing vessel occlusions or rupture. The plaques are built of a thin fibrous layer covering a massive lipid core; the main characteristics in the environment of these lesions are infiltration of inflammatory cells, necroses, reduced collagen production combined with an increase in collagen proteolysis (Naghavi et al., 2003; Naghavi et al., 2003). Collagen degradation is mediated by matrix metalloproteinase 9 (MMP-9) (Knipp et al., 2004; Su et al., 2006; Chen et al., 2006), an enzyme that is activated by a NF-jB triggered cascade (Lu and Wahl, 2005). In this scenario the main sources of vascular stress are due to intracellular proteins like NADH/NAD(P)H oxidases, xanthine oxidase, lipoxygenases, myeloperoxidase, transition metals and (damaged) mitochondria (Madamanchi et al., 2005). In the beginning of the atherosclerotic plaque formation, the UPS seems to be not impaired. It is considered that an increased formation of ROS is one of the primary steps in the pathology of AS. One indicator for this hypothesis in an increased expression of the heat shock proteins Hsp60 and Hsp70, those are related to atherosclerosis and stress-inducible. The amount of Hsp60 in affected regions (i.e. the vascular wall) was found to be directly correlated with the progression of AS (Kleindienst et al., 1993). Later the expression of Hsp70 increases in stages of advanced AS. Experiments with a mice knockout mutant (apolipoprotein E/) resulted in hypercholesterolaemia and an early onset of atherosclerotic plaque formation (Kanwar et al., 2001), while both Hsp60 and Hsp70 are increased. The affected cells show increased occurrence of polyubiquitinated proteins as found in carotid artery cells of patients with symptoms of a focal brain ischemia compared to asymptomatic patients (Herrmann et al., 2008). The amount of polyubiquitinated proteins found is increased from early stages of AS to later ones (Hermann et al., 2003; Tan et al., 2006), combined with an increase in proteasomal activity (Tan et al., 2006; Herrmann et al., 2007; Xu et al., 2007). Further triggered events are the unfolded protein response (see Chapter 4.2) and further oxidative stress, involving the generation of peroxynitrite, suggesting inflammatory processes (compare Chapter 4.9) in both macrophages and endothelial cells, the main actors in AS (Feng et al., 2003; Dickhout et al., 2005; Gargalovic et al., 2006). In later stages of AS these counteractions are not sufficient to balance the effects and the cell gets into a state of chronic oxidative stress, while the AS pathology worsens. In later stages of AS as found in post mortem in human samples, in advanced cellular and collagenous plaques the expression of heat shock proteins decreases, initiating the development of complications (Herrmann et al., 2008). The amount of protein polyubiquitination remains unchanged, detection of the proteasomal activity shows inconsistent results (Marfella et al., 2006; Versari et al., 2006), although inhibition of the proteasome increases AS progression (Herrmann et al., 2007), accompanied by intracellular accumulations of carbonylated, nitrated or glycated (especially in diabetes) proteins (Herrmann et al., 2008). The detrimental effects of decreased proteasome activity (by aging, pathology or inhibition) are discussed in the Chapters 3.9, 4. and 4.5. The activation of the UPS induces inflammatory processes by activation of the NF-jB pathway (see Chapters 4.1 and 4.6.6) resulting in further oxidative stress (Marfella et al., 2007a). In later stages of AS ubiquitinated, not degraded proteins aggregate and are further oxidized and chemically cross-linked by products of lipid peroxidation (for detailed description see Chapter 4.4), now decreasing proteasomal activity. In a case study about 54% off all patients showed inclusions of amyloid, both in the intima and in atherosclerotic plaques (Rocken et al., 2006), while apolipoproteins are considered to be the main extracellular source for amyloid fibrils (Herrmann et al., 2008). Experiments infusing mouse with solubilised b-amyloid (Ab40) resulted in a cerebral hypoperfusion (Suo et al., 1998), while the cortex was especially affected; further studies proved that b-amyloid damages endothel cells and is capable to induce an inflammatory response (Thomas et al., 1997a,b). This process is termed as ‘‘cerebral amyloid angiopathy’’ and
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
247
can often be found in progressed Alzheimer’s disease (Revesz et al., 2003). Mental prostration is better correlated to the amount of amyloid plaques in cerebral vasculature that to the intraneuronal aggregate formation (Herrmann et al., 2008). Diabetes turned out to be a special factor in development and progression of AS, too, intimately associated with the UPS (Marfella et al., 2007a,b). The UPS is responsible for regulation of the insulin receptor (a receptor-associated tyrosine kinase) and the amounts of the insulin receptor substrates 1 and 2. Insulin resistance has been revealed to play a crucial role in the generation of AS and the tumor necrosis factor alpha (TNF-a) induced form of insulin resistance is dependent of the activity of the UPS-mediated proteolysis (Pallares-Trujillo et al., 2000). Insulin resistance downregulates the phosphatidylinositol 3kinase (PI3K) (Wu et al., 2007) pathway that decreases antiatherogenic responses (Marfella et al., 2007a). Hyperglycemia is able to induce endothelial dysfunction (Ceriello, 2005; Giugliano et al., 1997) and an increase of mitochondrial superoxide release (Brownlee, 2001), triggering again NO production by eNOS and iNOS, resulting in peroxynitrite-mediated cell stress. Other experiments showed a hyperglycemia-induced degradation of both eNOS and tetrahydrobiopterin (BH4), an essential cofactor of eNOS, that may be mediated by peroxynitrite and its common product of protein oxidation, the 3-nitrotyrosine, both able to trigger eNOS degradation via an activation of the UPS (Wei and Xia, 2006). In the case of chronic massive oxidative stress, the amount of eNOS is reduced over time (Wei and Xia, 2006). Thus chronic hyperglycemia is able to damage endothelial cells by oxidative stress. Marfella et al. (2007a) suggest activation of the UPS as the binding link between hyperglycemia and endothelial activation (secretion of cytokines and adhesion molecules in an NF-jB mediated pathway). Long term results of this chronic inflammatory process in AS are plaque rupture and cardiovascular events (Libby et al., 2002). Cytokines, growth factors, T-cells, monocytes, macrophages, foam cells are the major components of atherosclerotic lesions (Eckel et al., 2002). The effects of AS is more severe in patient with type 2 diabetes, even tough the exact mechanism is still unclear. Marfella et al. (2006) presented data, showing that the amount of macrophages, T lymphocytes and HDLA-DR+ inflammatory cells are found at higher concentration in plaques of diabetic patients in comparison to those of non-diabetic ones, suggesting a massive inflammatory process. Furthermore, the amount of superoxide and the NF-jB levels are significantly increased in diabetic lesions (Marfella et al., 2006), possibly both accelerating and deteriorating the pathologic processes. In contrast to the view of Herrmann et al. (2008), interpreting AS as a disease resulting from the accumulation of damaged proteins as link to the increased activity of the proteasomal system, Di et al. (2008) distinguish between early and late stages of AS. The early stage might be due to the constant increase of the UPS throughput by oxidatively damaged proteins that are formed by stress. The progression of the atherosclerotic plaques in later stages occurs, if the antioxidative systems of the cell and the UPS are overwhelmed by the amount of the applied stress. This chronic state turns AS into an inflammatory process. Therefore, a filigree balance between activation of the UPS, activation of the NF-jB pathway and accumulation of polyubiquitinated proteins seems to be important. The interactions of the involved systems and the resulting shifts of the steady state might result over time in a destabilization of the formed plaques, vessel necrosis and rupture. To determine the exact mechanism, the clinical data are still not sufficient and have to be provided in further studies (Di et al., 2008). Similar to the treatment in stroke (see Chapter 4.6.6), an inhibition of the proteasomal system, due to the similarity of the basic pathologic mechanisms, seems to be a promising approach in slowing down the progress of AS, if applied in early stages (Stangl et al., 2004; Herrmann et al., 2004). Another approach is to restore the insulin-sensitivity of the adipose tissue; this can be achieved by using peroxisome proliferator-activated receptor-c (PPARc) agonists (Takano et al., 2004). One of them is rosiglitazone, capable of suppressing inflammatory response (and the consequences from it) and reducing proteasomal activity (Marfella et al., 2006), thus reducing plaque growing (Marfella et al., 2006). 4.9. Rheumatism and autoimmune disease As atherosclerosis, rheumatism/rheumatoid arthritis (RA) is a typical disease, resulting from excessive inflammatory processes. The same mechanisms result in pathologies like reperfusion injury after ischemia (see Chapter 4.6.6), asthma, multiple sclerosis or psoriasis (see Chapter 4.15.3). The base mechanism is explicit explained in the Chapters 4.1 and 4.6.6, and again pharmacological inhibition of the proteasome seems to be a promising approach. The main symptoms of RA are pain, swelling and finally grave damage of tissue accompanied by inflammation, tightly connected with an activation of NF-jB. In response, the affected cells release proinflammatory mediators like the tumor necrosis factor alpha (TNF-a), the interleukines IL-1 and IL-6, activate iNOS and excrete cell adhesion molecules, attracting leukocytes. A widespread animal model is intraperitoneal injection of a streptococcal cell wall into female Lewis rats (Palombella et al., 1998). After about 14 days these animals develop a chronic progressive synovitis that shows pathologic events on joints that are similar to that found in RA (Cromartie et al., 1977). Using the proteasome inhibitor bortezomib (PS-341, see Chapter 3.9) it is possible to reduce the clinical symptoms and to slow down the progression and to massively reduce inflammatory response (Elliott et al., 2003). The mentioned markers (IL-1, IL-6 and NF-jB) were found in significantly lower amounts than in untreated control animals. The result of the massive chronic inflammation of the synovial lining is destruction of the joint (Handel et al., 1995). The tissue is invaded by immunocompetent cells and the proliferation of synovial fibroblasts results in the formation of pannus tissue, a tumor like structure that slowly induces destruction of the articular cartilage and the subchondral bone by invading and eroding. If the NF-jB pathway is inhibited, at the same time the anti-apoptotic effects of NF-jB are disabled, resulting in the inhibition of cell proliferation. This decreases growth of the pannus tissue (Calzado et al., 2007). Similar effects can be achieved with other known inhibitors of the NFjB pathway like salicin and aspirin (at a concentration of about 50 lM reducing the
248
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
IKKb kinase by about 50%, Calzado et al. (2007)) that inhibit IKKb thus preventing phosphorylation of the NF-jB inhibitor IjBa (as described in Chapter 4.1). Another therapeutic target might be the IjBa-ubiquitination in order to inhibit the NF-jB pathway; this furthermore prevents activation of pathways induced by SCFbTRCP (see Chapter 4.1). 4.10. The proteasome and viral infections Viral infections are changing the cellular pattern of presented antigens (see Chapter 4.3). Since the proteasome and the UPS are key players in this process, in all viral infections the proteasome/UPS system is playing a key role. Several viruses developed strategies to evade antigen presentation catalyzed by the proteasome/UPS pathway, interestingly often by interfering very specific with E3-mediated ubiquitination. The strategy of specific E3-inhibition or the expression of own proteins showing E3 functions is a strategy used by different viruses and may be a product of co-evolution with the UPS. Here the strategy of the HIV-I virus has to be mentioned, expressing a protein (HIV-I Vpu, a 16 kDa integral membrane protein containing 81 amino acids), that is able to target CD4 to ERAD-like degradation. Vpu is constitutively phosphorylated and shows a sequence in its cytosolic part that mimics a substrate for the SCFbTrCP (Gallegos et al., 2008; Asada et al., 2008; Peschiaroli et al., 2006) complex, an E3 ubiquitin ligase and part of the ubiquitin proteasome system. CD4 is a membrane protein expressed on the surface of a T lymphocyte subclass that recognizes the presented major histocompatibility complexes II (MHC II). One of the first steps to CD4 degradation is the binding of Vpu to the cytoplasmic part of the protein and the binding to the human F-box protein b-TrCP (Schubert et al., 1998). b-TrCP is responsible for Ub-transfer to CD4 mediated by SCFbTrCP. Polyubiquitination of CD4 is the signal for 26S proteasomal degradation, whereas Vpu is not degraded. The 26S mediated proteolysis of CD4 can be prevented by proteasomal inhibitors, in cells expressing a temperature sensitive E1, the Ub-activating enzyme of the UPS, or by expression of an Ub-mutant (Ub K48/R), that is unable to form of poly-Ub chains as degradation signal (Schubert et al., 1998). Another example is the human cytomegalovirus (HCMV) (Barel et al., 2006; Lee et al., 2005), expressing two membrane bound proteins, US2 and US11, that bind MHC I molecules in the ER and induce their dislocation into the cytosol where they are degraded by the UPS. 4.11. The proteasome and age related macular degeneration (AMD) Age related macular degeneration (AMD) (Jager et al., 2008; Pauleikhoff and Holz, 1996; Nowak and Bienias, 2007; Nowak, 2006) is a chronic progressive disease, manifesting a severe impairment of vision induced by loss of retinal cells. AMD is considered to be the most common cause of vision loss and blindness in people at the age above 65 (Klaver et al., 1998), affecting about 8 million people worldwide (Hooper et al., 2008). Two main forms of AMD are known, the dry (nonexudative, accounting for about 80% of all cases) form and the wet (exudative form), accompanied by a choroidal neovascularization. Oxidative stress and the formation of free radicals have turned out to be a major factor in AMD pathogenesis, though their initial source is still unknown. Light is assumed to be one of the driving forces, inducing the so-called ‘‘blue light phototoxicity’’ of AMD (Lamb et al., 2001). Additionally, the effects of a defective iron homeostasis (Chowers et al., 2006; He et al., 2007; Dunaief, 2006; Wong et al., 2007) are discussed, since mice, showing a mutation of ceruloplasmin (also involved in iron transport), developed an age-dependent retinal iron overload and indications of AMD. The retinas of human post mortem AMD-patients showed both an increase in the iron content and the iron transport protein transferrin, compared to an age-matched control group. Another characteristic of AMD is the intracellular accumulation of oxidized proteins and the formation of a lipofuscin-like material, a process playing a limiting role in the life expectancy of postmitotic aging cells (see Chapter 4.5). Matching this, clinical trials showed significant effects of antioxidant and zinc supplementation, slowing down the progression of the pathology (Evans, 2008; Evans and Henshaw, 2000; Evans and Henshaw, 2008). In this special case lipofuscin is the precursor of the so called ‘‘drusen’’ (Pauleikhoff et al., 1990; Johnson et al., 2003; Luibl et al., 2006), a material considered as the hallmark of the AMD pathology. Drusen are aggregates resulting from an extracellular accumulation of debris and waste products between retinal pigment epithel (RPE) and Bruchs membrane. This material is the target of an immune reaction, causing a state of chronic inflammation (Johnson et al., 2001; Hageman et al., 2001; Anderson et al., 2002). Since the two of the major causes of AMD are oxidative stress and inflammation (inducing the release of cytokines), factors known to affect the proteasomal system and its regulators, it was to postulate a possible role of the proteasome within the disease. In a recent study the proteasomal chymotrypsin-like activity (localized on the b5 subunit) was shown to be increased with AMD progression. To investigate the influence of an exchange of the constitutive against the inducible subunits, the amounts of b1, b1i, b5 and b5i have been determined. The results indicated a clear rise of the inducible subunits (Ethen et al., 2007). A change of the total amount of cellular proteasome was not found, indicating that an increase of the inducible proteasomal subunits must be accompanied by a reduction of the constitutive proteasomes. Furthermore, the amount of inducible subunits increased with the stage of AMD, both in the macular and its periphery. Even though no infiltration of the retina by CD45 (immune cells containing large amounts of the immunoproteasome) could be detected, the amount of b5i was significantly increased. In addition to that it was described that the proteasomal regulators PA28 and PA700, PA700 showed no change, but PA28 seems to be punctual increased in the macular during the latest stage of AMD. Hsp90, another proteasomal regulator, was not correlated with the stage of pathology (Decanini et al., 2007).
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
249
4.12. The proteasome in the eye lens – involvement in cataract formation The physiological function of the eye lens is the focusing of light on the retina. The mammalian lens is a very specialized part of the eye, mainly composed of a gradient of differentiated epithelial cells, which accumulate high amounts of crystallins. These cells differentiate to fiber cells, exit cell cycle and lose virtually all intracellular organelles including their nuclei (Kuszak, 1995). The nucleus of the eye lens is containing fully transformed fiber cells. Despite of the UPS a large amount of other proteases is involved in differentiation and pathologic processes in the eye lens, like calpains, caspases, matrix metalloproteinases (MMPs), tissue inhibitors of metalloproteinases (TIMPs), ADAMS, and ADAMTS. The differentiation and massive morphological transformation seems to be necessary to afford transparency of the lens. Aberrations in this process, like an incomplete degradation of cell organelles, result in diverse forms of cataract formation (Pan and Griep, 1994; Nakamura et al., 1995). Some studies compare that process of complete organelle removing with a special form of apoptosis (Dahm et al., 1999; Wride, 2000; Bassnett, 2008). This idea is supported by results showing that the morphological changes in the nuclei (chromatin condensation and DNA fragmentation) (Bassnett and Mataic, 1997) during degradation is comparable to the one found in apoptotic cells and that parts of the ‘‘apoptotic machinery’’ (proteases and caspases) are involved in loss of cellular organelles. The expression of the UPS has been found upregulated during the differentiation phase of the lens fiber cells (Guo et al., 2004, 2006). In epithelial cells in the phase of transformation, the amount of polyubiquitinated substrate proteins is found to be 5–10-fold increased compared to ‘‘resting’’ cells (Shang et al., 1999), while the cell cycle itself and the differentiation are UPS-regulated, too (Liu et al., 2004). It was shown that lens epithelial cells, and even the fiber cells in the lens-nucleus contain most and possibly all components of a fully functional UPS (Pereira et al., 2003). As in most somatic cells the lens UPS responds to oxidative stress with increased activity and degrade damaged proteins in a highly selective way (Shang et al., 2001a,b). The protein density and the activity of the lens UPS nevertheless decreases while approximating the centre of the lens. The performance of the ubiquitinating system is in this case limited by Ubc4 and Ubc5 (Lorick et al., 2005; Mulder et al., 2007), two E2-ubiquitin-conjugating enzymes. Artificial increase of these E2-proteins enhances the performance of the lens UPS in the inner regions about 10-fold (Pereira et al., 2003). At the same time the amount of polyubiquitinated substrates are higher in the outer regions of the lens, representing the amount of substrate proteins and the activity of the ubiquitinating cascade. In the nucleus of the lens the E2-proteins Ubc4 and Ubc5 seem to be the rate limiting steps in polyubiquitination of substrates. Thus even in the transformed fiber cells of the lens nucleus a functional UPS can be found. Moreover isopeptidases are in the lens nucleus available (Pereira et al., 2003), enzymes that deubiquitinate target substrates for the 26S proteasome. Pereira et al. (2003) suggest that the (26S) proteasomal activity in the nuclear region of the eye lens and especially the enzymes Ubc4 and Ubc5 are necessary to ensure the transparency of the lens. During the lens formation, the adhesion of the fibre cells is granted by gap junctions, connections between cells, that are necessary to provide transparence of the mature eye lens (Yin et al., 2001). Mutation or degradation of the connexins, the proteins of the gap junctions, result in different pathologies including the formation of cataracts (Wei et al., 2004). During differentiation the connexins are cleft by caspases and an overactivity of these enzymes may result in opaque structures of the eye lens. At the other hand underexpression of caspase-3 has been shown to result in cataract formation due to aberrant differentiation (Andersson et al., 2003). Another important role is played by calpains. It should be mentioned that a cause of cataract formation is the insolubilization of crystalline proteins that are needed for transparency of the lens. This insolubilization can be the result of an unregulated degradation of crystallins by calcium-dependent calpains (Biswas et al., 2004, 2005). Therefore, a potential treatment approach might be the application of calpain inhibitors (Biswas et al., 2004). Recently the effectiveness of the calpain inhibitor SJA6017 has been shown in reducing the cataract formation in lens by preventing the emerge of opaque structures (Biswas et al., 2004; Tamada et al., 2001). In porcine lens a reduction of about 40% in cataract formation was achieved (Biswas et al., 2004). The key calpains may be the lens-specific Lp82 and the widespread m-calpain, both playing a role in differentiation of the lens and the pathology of cataract formation (Wride et al., 2006; Shih et al., 2006). Apart from the mentioned proteases the role of a strictly regulated proteasomal activity may be more important. In an age-dependent manner the activity of ubiquitin conjugating enzymes has been found to decrease in the lens (Wride et al., 2006), while even structural changes in the ubiquitin itself have been revealed (Stiuso et al., 2002). Whether the activity of the proteasome does change is unclear, since according to some studies it remained unchanged (Zetterberg et al., 2003), whereas according to others it did decline (Viteri et al., 2004). However, in cases of cataract formation it was found to be reduced (Zetterberg et al., 2003; Andersson et al., 1999). Moreover extensive oxidative stress is capable of reducing the activity of the whole UPS, due to indegradable protein modifications or even damage to the UPS itself (see Chapter 4.4). Studies of the age-dependent changes in lens proteins and the decreasing proteasomal activity have been performed by Viteri et al. (2004). During age, the lens shows a little browning and increased amounts of oxidized protein aggregates that may be mainly due to Maillard reaction (protein glycation) (Viteri et al., 2004). If these glycated protein are further oxidized and chemically cross-linked, the proteasome is not longer able to degrade them (Bulteau et al., 2001), despite of that glycated proteins might be able to inhibit proteasomal activity (Portero-Otin et al., 1999). Looking at specific protein modifications in the lens, the amount of water soluble protein decreases with age from 95.1% in children (0–10 years of age) to 69.9% (>60 years), while at the same time the amount of both advanced glycation end products (AGEs) and proteins containing a carboxymethyl-lysine modification was significantly increased (Viteri et al., 2004). In this study a clear decrease of proteasomal activity has been detected in the soluble fraction of lens proteins, mostly affecting the peptidyl–glutamyl-peptide hydrolase activity on the b1-subunit, until incommensurability at age >60. The other two activities were significantly decreased in the
250
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
‘‘oldest’’ compared to the ‘‘youngest’’ group of samples, but the activities of b2 and b5 were still detectable. In experiments with bovine lenses an age-related decrease in the activities of b1 and b2 was detected, while b5-activity was almost not affected (Wagner and Margolis, 1995). This may be due to a higher sensitivity of the human lens to oxidative stress and stress induced proteasomal damage (Conconi et al., 1996). Thus the formation of cataract and opaque structures in the eye lens may be due to a decrease in proteasomal activity, reducing or even disabling the degradation of oxidatively modified proteins in advance to further oxidation (see Chapter 4.4). 4.13. The proteasome and ethanol-induced liver injury Since the liver is the main organ of ethanol metabolism, in hepatocytes the main proteins involved in ethanol (EtOH) degradation are expressed: alcohol dehydrogenase (ADH) and cytochrome P450E1 (CYP2E1) (Niemela et al., 1998; Caro and Cederbaum, 2004; Donohue et al., 2006; Gouillon et al., 1999), catalyzing the formation of acetaldehyde and ROS. The proteolytic degradation of CYP2E1 is due to the proteasomal degradation, since proteasomal inhibition results in an increased amount of intracellular CYP2E1 apoprotein (Bardag-Gorce et al., 2006). On the other hand EtOH is able to stabilize CYP2E1. This stabilization might be mediated by the heat shock proteins Hsp70 and Hsp90 (Morishima et al., 2005). The products of EtOH degradation by CYP2E1 are able to reduce proteasomal activity as shown in HepG2 cells using concentrations of about 100 mmol EtOH, an effect that was not observed in CYP2E1 knockout mice (Osna et al., 2003). Blood EtOH concentrations P40 mmol already show a reduction of proteasomal activity, the trypsin-like activity (b2-subunit) is most affected (Koll et al., 2002). Since CYP2E1 is forming ROS, it induces lipid peroxidation resulting also in the formation of 4hydroxy-2-nonenal (HNE), able to covalently cross-link proteins (Friguet and Szweda, 1997; Friguet et al., 1994a) and to inhibit the proteasome (Farout et al., 2006). Purified proteasomes from liver treated with HNE showed rapid inhibition of the chymotrypsin-like (b5 subunit) activity, accompanied by a modification of the a6 subunit (Ferrington and Kapphahn, 2004). As mentioned above, the proteolytic activity of the single 20S particle are in the order of b5 b2 > b1, indicating almost complete proteasomal inactivation by inhibiting both b5 and b2. Another result of the ROS formation by CYP2E1 is the formation of peroxynitrite (ONOO) inducing nitration of the 20S proteasome, resulting in a decrease of the chymotrypsin-like activity (Osna et al., 2007). Peroxynitrite has been found to induce 3-nitrotyosine and tryptophan oxidation in the constitutive and inducible subunits of the proteasome, whereas the constitutive form (20S) seems to be more resistant to oxidative modification (quantified by proteasomal degradation of b-casein) than the immunoproteasome (i20S) (Amici et al., 2003). Despite of ROS, ethanol metabolism generates acetaldehyde. Acetaldehyde inhibits the proteasome in a dose dependent manner both added to purified proteasome and in the form of acetaldehyde–protein adduct (Rouach et al., 2005). Malondialdehyde, another product of lipid peroxidation, and acetaldehyde are able to react with proteins in a synergistic manner. The rate of protein modification is increased up to 32-fold in the presence of both aldehydes, compared to single agent treatment (Tuma, 2002). The product is a malondialdehyde–acetaldehyde–protein-adduct, termed as MAA adducts. These products and the presence of acetaldehyde, MDA and HNE have been found in the liver of EtOH-exposed humans, rats and micropigs (Niemela et al., 1995, 2000; Worrall et al., 1991). MAA adducts are able to induce strong immune responses, both against the adduct epitope and the unmodified protein (Thiele et al., 1998). This triggers a pro-inflammatory and fibrinogenic response of the liver, playing an important role in the pathogenesis of alcoholic liver disease. Moreover these modified proteins are less susceptible to proteasomal degradation. Especially the formation of insoluble protein aggregates enhanced by proteasome inhibition in line with further oxidation by ROS and cross-linking by HNE results in the accumulation of indegradable material again able to reduce proteasomal activity. These aggregates are the Mallory bodies, respectively alcoholic hyaline, mainly found in patients with alcoholic liver cirrhosis. Mallory bodies are mainly formed by the cytokeratins 7 and 19 (Bardag-Gorce et al., 2006). Cell death induced by EtOH might be facilitated due to proteasomal inhibition, too. Several pro- and anti-apoptotic factors are regulated by UPS degradation. If the UPS-activity is reduced or even inhibited, this may result in the intracellular accumulation of pro-apoptotic factors, affecting the mitochondria and finally causing cell death. A second factor for the induction of both apoptotic and necrotic cell death is the effect of EtOH on the mitochondria, since metabolism of EtOH damages mtDNA, decreases the b-oxidation of fatty acids and the ATP-production by the oxidative phosphorylation, causing cellular ATP-depletion. The result can be the release of cytochrome c by opening of the mitochondrial pore, starting the apoptotic cascade. Furthermore, EtOH induced inhibition of the proteasome, can result in an accumulation of pro-apoptotic factors like Bid and Bax (Korsmeyer et al., 1999; Bhattacharjee et al., 2008; Crompton, 2000), that are associated with the mitochondrial membrane and normally degraded by the proteasome. If the cell undergoes apoptosis or necrosis depends on the level of available ATP, so EtOH is able to induce both kinds of cell death in the liver. By interfering with the IFN-c induced expression (caused by a ROS-induced stabilization of the suppressor of cytokine signaling 1 (=SOCS1) by the proteasome) of the i20S proteasome, the MHC class I presentation of antigens is decreased, causing a decreased immune defense. 4.14. The proteasome and diabetes The number of cases of type 2 diabetes is constantly rising. In 2006 about 6% of the world population was affected. Unfortunately, until now no treatment is available. Diabetes is classified in two main groups: Type 1a (immune mediated, autoimmune disease), 1b (idiopathic) and Type 2, having perhaps various causes. Diabetes induces several severe complications
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
251
in its pathologic progression: increased blood pressure, diabetic retinopathy, diabetic neuropathy, myocardial infarction, atherosclerosis, stroke, nephropathy, blindness, diabetic foot syndrome and the majority of nontraumatic lower extremity amputations. One of the most important factors in the pathology of diabetes is the increased amount of hyperglycemia-induced oxidative stress, especially in the diabetic neuropathies (DN) and atherosclerosis. Hyperglycemia is able to induce oxidative damage by the autooxidation of glucose (Hunt et al., 1990; Wolff, 1993; Wolff et al., 1991), enhanced activity of aldose reductase (AR), the formation of advanced glycation end products (AGEs), that are able to induce in a receptor-mediated oxidative stress (Mullarkey et al., 1990), an increased activity of protein kinase C (PKC) and a mitochondrial overproduction of superoxide anions. Even in conditions of low glucose concentration, increased oxidative stress occurs, mediated by vascular cells and in particular the endothelium (Bouloumie et al., 1997; Dohi et al., 1990). Several studies in humans showed increased lipid peroxidation (via ROOH), oxidative DNA-damage (via 8-OH-dG) and oxidized low-density lipoprotein (oxLDL) in types 1 and 2 diabetes (Dandona et al., 1996; Sundaram et al., 1996; Rabini et al., 1994; Hussein et al., 2007) compared to an age-matched control. Protein oxidation was increased, too, detected by the use of protein carbonyls and nitrotyrosine, both in plasma and intracellularly (Telci et al., 2000). Besides protein carbonyls, nitrotyrosine and a decrease of sulfhydryl groups was found (Telci et al., 2000). Additionally a chronic decrease of the cellular antioxidative capacity in diabetes has been shown: the important antioxidant glutathione (Likidlilid et al., 2007) and the vitamins C and E are reduced (Vincent et al., 2004) and a cellular depletion of NADPH (Feldman et al., 1997; Greene et al., 1999) has been reported. This may be due to an increasing oxidative stress during the progression of diabetes (Baynes and Thorpe, 1999). It was discussed, that the proteasomal activity and, therefore, the ability to degrade oxidatively damaged proteins declined (Xu et al., 2007; Portero-Otin et al., 1999; Bennett et al., 2003; Ahmed et al., 2005). Recent evidence shows a role of the UPS in diabetes by an inappropriate degradation of insulin signaling molecules like the insulin receptor substrates 1 and 2 (IRS-1 and -2) compared with an up-regulation of suppressors of cytokine signaling (SOCS) (He and Stephens, 2006; del Rincon et al., 2004; Pederson et al., 2001). 4.15. Proteasome and skin The skin, an important barrier that protects the body from damage of the environment, is composed of three main layers: epidermis, dermis and subcutis (Fig. 34). The dermis lies below the epidermis and in conjunction provides mechanical support for the outer protective layers of the epidermis. While epidermis and dermis contain several cells like keratinocytes, Langerhans cells, melanocytes, fibroblasts and leukocytes, subcutis contains only fat and connective tissue. Distinct disorders and environmental factors show different effects in these skin parts and cell types. The half-life of nearly 80% of cellular proteins is under proteasomal control, also in various skin cells. In skin diseases, as in the other diseases, proteasome is often
Fig. 34. UV penetration into the skin layers. The penetration of various components of sun irradiation into the skin is different. While UVC (200–280 nm) in general is not reaching the surface of the earth, skin is exposed to UVB (280–320 nm) and UVA (320–400 nm) penetrating into different layers of the skin. While UVA penetrates the epidermis and is absorbed in the dermis beyond, UVB only reaches a small dermal volume close to the epidermis.
252
T. Jung et al. / Molecular Aspects of Medicine 30 (2009) 191–296
Table 5 Important cancer related regulatory proteins which are degraded by the UPS. Cell cycle regulatory proteins Oncogenic products and tumor suppressors
Transcriptional regulators
Enzymes
CDK inhibitors (p27, p21 and others) Cyclins (mitotic cyclins, G1 cyclins) p53 and Mdm2 c-jun/c-fos; c-myc E2A proteins IjB and NFjB (p105) b catenins HIF1 (hypoxia-inducible factor-1) ATF2 (activating transcription factor-2) cdc25 phosphatase (CDK1/cyclin B phosphatase) Tyrosine amino transferase
CDK, cyclin dependent kinases.
involved and might be used for therapeutic approaches. As described in Chapter 3.9, many structurally different proteasome inhibitors might be used in these approaches with different chemical mechanisms and they provide a window for therapeutic intervention and may offer hope in several cases. 4.15.1. Skin carcinogenesis Skin cancer might have several reasons, such as environmental factors – mainly UV irradiation – or might also be chemically induced. Skin cancer might form from several cell types, including melanocytes. A vast majority of studies concerning the involvement of the proteasomal pathway in skin carcinogenesis were performed to highlight the mechanisms of proteasomal involvement in skin cancer carcinogenesis. Malignant melanoma is the most aggressive form of the skin cancer. It is generally difficult to control the progression of the disease due to the development of resistance to conventional chemotherapy and radiotherapy (Sorolla et al., 2008). The best-studied single agent chemotherapies for melanomas have objective response rates of