Structural Studies on Pax-8 Prd Domain/DNA Complex

Journal of Biomolecular Structure & Dynamics, ISSN 0739-1102 Volume 24, Issue Number 5, (2007) ©Adenine Press (2007)

Structural Studies on Pax-8 Prd Domain/DNA Complex http://www.jbsdonline.com Abstract Pax-8 is a member of the Pax family of transcription factors and is essential in the development of thyroid follicular cells. Pax-8 has two DNA-binding domains: the paired domain and the homeo domain. In this study, a preliminary X-ray diffraction analysis of the mammalian Pax-8 paired domain in complex with the C-site of the thyroglobulin promoter was achieved. The Pax-8 paired domain was crystallized by the hanging-drop vapor-diffusion method in complex with both a blunt-ended 26 bp DNA fragment and with a sticky-ended 24 bp DNA fragment with two additional overhanging bases. Crystallization experiments make clear that the growth of transparent crystals with large dimensions and regular shape is particularly influenced by ionic strength. The crystals of Pax-8 complex with blunt-ended and sticky-ended DNA, diffracted synchrotron radiation to 6.0 and 8.0 Å resolution and belongs both to the C centered monoclinic system with cell dimensions: a = 89.88 Å, b = 80.05 Å, c = 67.73 Å, and β = 124.3º and a = 256.56, b = 69.07, c = 99.32 Å, and β = 98.1º, respectively. Fluorescence experiments suggest that the crystalline disorder, deduced by the poor diffraction, can be attributed to the low homogeneity of the protein-DNA sample. The theoretical comparative model of the Pax-8 paired domain complexed with the C-site of the thyroglobulin promoter shows the probable presence of some specific protein-DNA interactions already observed in other Pax proteins and the important role of the cysteine residues of PAI subdomain in the redox control of the DNA recognition.

Mara Campagnolo1 Alessandro Pesaresi1 Igor Zelezetsky1 Silvano Geremia1,* Lucio Randaccio1 Alessia Bisca2 Gianluca Tell2 Department of Chemical Sciences and

1

Centre of Excellence in Biocrystallography University of Trieste Via L. Giorgieri 1

34127 Trieste, Italy

Department of Biomedical

2

Sciences and Technologies University of Udine Piazzale Kolbe, 4

33100 Udine, Italy

Introduction Pax proteins form a family of transcriptional regulators that control a variety of cell fates during animal development (1). Their highly conserved function, across most of the animal kingdom, includes eye development and cephalization. Seven Pax genes exist in Drosophila, including paired protein (2), nine are known in mouse and human (Pax-1 to Pax-9), and other Pax genes are found in a variety of species from nematodes to vertebrates (3-6). The biological effects of Pax proteins depend on the intact function of a conserved DNA-binding domain, called paired (prd) domain (7, 8). The prd domain, composed of 128 amino acids, was first identified in the Drosophila prd and gooseberry genes (9). A number of murine and human developmental mutants are known to bear alterations in specific Pax genes, and several of these involve missense mutations on the prd domain (10). For example, mutations in human Pax-3 and Pax-6 genes cause Waardenburg’s syndrome (11) and aniridia (12), respectively. The crystal structure of the prototypical prd protein from Drosophila, bound to a 15 base paired DNA site (13) has shown the presence of two structurally independent subdomains (named PAI, residues 1-72 and RED, residues 77-128), each containing a helix-turn-helix motif joined by a linker region. The crystal structure of the human Pax-6 prd domain docked to its optimal 26 base paired site (14) gave the first detailed information about how the entire prd domain contacts DNA. The presence

*Email: [email protected]

429

430

of the distinct structural regions PAI and RED allows the prd domain to recognize DNA sequences by using different binding modes.

Campagnolo et al.

Pax-8 protein is the only member of the Pax family expressed in the thyroid tissue. Inactivation of the Pax-8 gene causes absence of follicular cells, and therefore absence of thyroid hormone (15). In a recent work (16) it was demonstrated that thyroid-specific gene expression may be regulated by synergistic action of the Pax8 prd domain with TTF-1, a homeodomain-containing factor. Genetic and biochemical experiments demonstrated the specific Pax-8 prd domain recognition of the natural binding site (C site) of the thyroglobulin promoter (17). One reason why the C sequence does not completely conform to the Pax-8 consensus obtained by in vitro selection is the necessity for TTF-1 binding. It has been demonstrated that C site is recognized through a cooperative action between PAI and RED subdomains of Pax-8 protein. In particular, these data suggest that the RED subdomain-C sequence interaction is required for correct DNA recognition by the PAI subdomain (17). Protein-DNA crosslinking experiments indicate that the RED subdomain binds to the C site as a monomer. Since the RED subdomain is much more variable than the PAI subdomain among Pax proteins, these results could explain how distinct Pax proteins may select different target genes (17). The binding activity of the Prd domain of Pax-8 is regulated through the oxidation/ reduction of cysteine residues located at positions 37 and 49 of the PAI subdomain, and 109 of the RED subdomain (18, 19). These residues are highly conserved in the whole family of Pax proteins. CD experiments have established that the oxidized state of the cysteine residues induces structural variations leading to an interference with the proper conformation useful for DNA binding. Mass spectrometry analysis demonstrated that Cys-49 is the most reactive cysteine residue of the Prd domain of Pax-8 establishing the existence, in the oxidized state, of a disulfide bridge with either Cys-37 or Cys-109. Since Cys-49 is believed to establish specific contacts with base pairs of DNA, the DNA binding activity of the PAI subdomain should be extremely sensitive to redox conditions, whereas the DNA binding activity of the RED subdomain should be poorly sensitive to redox control. The importance of the two cysteine residues of the PAI subdomain in the regulation of the DNA binding activity of Pax-8 has been recently confirmed by glutathionylation experiments (20). In order to investigate the Prd domain-DNA interactions and the specific function of the PAI and RED subdomains, we present an experimental and theoretical structural study of a protein/oligonucleotide complex containing the human Pax-8 Prd domain with its natural DNA-binding site, the C-site of the thyroglobulin promoter. Material and Methods Expression, Purification, and Sample Preparation of the Pax-8 Prd Domain The Pax-8 Prd domain consists of 128-amino acids. The plasmid pT7.7Pax8-Prd was obtained by subcloning the DNA fragment encoding residues 6-143 of the human Pax-8 protein into plasmid pT7.7 [for more details see (18)]. The plasmid was used to transform the BL21 bacterial strain. Transformed cells were grown at 37 ºC to OD600 0.6-0.7 and then induced by 1 mM isopropyl-1-thio-β-D-galactopyranoside (IPTG) for three hours. Cells were harvested by centrifugation and resuspended in lysis buffer [20 mM tris (hydroxymethyl) aminomethane (TRIS)-HCl pH 8.0, 0.25 M NaCl, 1 mM EDTA, 0.1% TWEEN20, 5 mM dithiothreitol (DTT), 10 μg/ml lysozyme, 1 mM phenylmethylsulfonyl fluoride (PMSF)] in a volume of 10 ml/g of bacterial pellet. The presence of DTT is important from the first step of extraction of the peptide in order to control the redox state of the cysteine residues. After cell lysis by sonication, the cell debris was removed by centrifugation. DNA

was removed by addition of 0.3 mg/ml protamine sulphate to the supernatant and the precipitate removed by centrifugation at 10000g at 4 ºC for 12 minutes. The supernatant was purified in two chromatographic steps using two cation exchange columns (theoretical pI = 10.45), a HiTrap SP HP and a Mono S 5/50 GL column (Amersham Biosciences) pre-equilibrated with binding buffer (10 mM TRIS pH 8.0 and 1 mM DTT). The Pax-8 Prd domain was purified using a linear gradient of 0.251.0 M NaCl in 10 mM TRIS buffer at pH 8.0 containing 1 mM DTT. The protein is eluted at 0.65 M NaCl (Figure 1aS Supplementary Material). At this stage, the affinity-purified protein gave a single band on an overloaded SDS (15%) polyacrylamide gel (PAGE) (Figure 1bS Supplementary Material). Mass spectrometry analysis confirmed the identity of the protein bearing a mass of 18,775 Da, as expected. Following the crystallization procedure reported in the Pax-6 Prd domain work (14) the sample was dialyzed against a buffer containing HEPES 40 mM pH 7.5, 10 mM spermine, 10 mM DTT, and 5 mM EDTA. Preliminary protein-DNA crystallization experiments revealed that spermine gives crystals of spermine phosphate hexahydrate, probably due to the presence of phosphate ions in DNA solution. In order to avoid this interference, the protein sample was successively dialyzed against TRIS/ HCl 10.0 mM pH 8.0 and 1 mM DTT, concentrated to 15.56 mg/ml (0.83 mM), and stored at -20 ºC. Recent studies (18, 20) have shown that Pax-8 polypeptide is sensitive to oxidative agents and that the disulfide bonds formation disrupts the DNAprotein interactions. To avoid the oxidation of peptide, DTT reagent has been added during protein preparation and also in the final crystallization sample (1 mM). DNA Sample Preparation Two double strand DNA containing the C-site of the Tg promoter, a 24-mer whose upper strand is 5ʹ-CACTGCCCAGTCAAGTGTTCTTGA-3ʹ, were prepared: a bluntended DNA having two additional 3ʹ T bases and a sticky-ended DNA having two T and two A bases 5ʹ and 3ʹ overhanging, respectively. The three samples of high purity oligonucleotides single strands were provided by MWG-Biotech AG, with composition verified with MALDI mass spectrometry. Double strand DNA samples were obtained according to the following annealing procedure. Two complementary lyophilized strands were resuspended in 500 μl 10 mM of TEAB (triethylamine bicarbonate), pH 7.0, in stoichiometric amounts. Solution was heated for five minutes in a 70 ºC water bath, and cooled slowly to room temperature. The purity of the double strands oligonucleotides was observed by an Electrophoretic Mobility Shift Assay (EMSA) analysis, and the concentration of the annealed DNA was determined spectrophotometrically using a molar absorption coefficient of 598.6 M-1 cm-1 at λ=260 nm. The DNA was aliquoted, lyophilized, and stored at -20 ºC. A DNA sample preparation was also performed by using the protocol reported in the structural study on Pax-6 Prd domain-DNA complex (14). This method makes use of annealing by simply dissolving DNA in water. The quality of samples obtained in both protocols were tested by Exclusion Size HPLC chromatography (Amersham Superdex 200 column). The chromatograms, of both blunt-ended and sticky-ended samples annealed in TEAB buffer show a single peak corresponding to a double strand species (Figures 2aS and 2bS Supplementary Material). The chromatogram of the sample obtained from desalted water is much more complex. The elution profile is formed by convolution of several peaks corresponding to different species as results of non-specific couplings (Figure 2cS Supplementary Material). These experiments confirmed the necessity of an ionic strength provided by a small amount of salts for an efficient DNA double strand coupling. Preparation of the Protein-DNA Complex The protein-DNA complex was prepared by adding the concentrated protein solution directly to the lyophilized DNA. To test the protein-DNA complex formation,

431 Pax-8 Prd Domain/ DNA Complex

432 Campagnolo et al.

two samples were prepared with an excess of DNA (50%, 20%) over the amount needed for 1:1 protein-DNA ratio. The complete formation of the complex was examined through EMSA analysis (Figure 3S Supplementary Material). The complex was dissolved in TRIS buffer (10 mM, pH 8.0) and 1 mM DTT with a final concentration of 29.6 mg/ml (0.83 mM). Crystallization Experiments Initial crystallization screening was performed using blunt-ended DNA with the hanging-drop vapor-diffusion method. These were performed following the successful crystallization conditions of Pax6-Prd domain-DNA complex (14). Polyethylene glycol with average molecular weight of 200 (PEG200) at concentration range 16-28% v/v with intervals of 2% v/v was used as precipitant. The reservoir solution also contained 10 mM DTT reducing agent, 10 mM spermine additive salt, and 5 mM EDTA used as complexing agent. Ammonium acetate was added to promote the complex stabilization. However, these attempts resulted unsuccessful, since only crystals of spermine phosphate hexahydrate were obtained, under different pH conditions. In a second series of crystallization trials the protein was dialyzed in a different buffer consisting of 10 mM TRIS pH 8.0 and 1 mM DTT. In this new buffer the Prd domain-DNA complex is completely soluble, and no ammonium acetate was necessary. The crystallization screening was performed, adding in the well solution 1 mM DTT and varying the concentration of PEG4000 from 5% to 30% w/v in intervals of 5% w/v. PEG4000 was used instead of PEG200 to reduce the nucleation rate. Hanging drops were prepared mixing 1 μl of the protein-DNA solution (0.83 mM, 29.6 mg/ml) and 1 μl of reservoir solution and were placed to equilibrate with 1.0 ml of well solution at 293 K. The crystals grew in a narrow range of 18-23% of PEG4000 with an irregular plate-like morphology with a maximum size of 200 × 50 × 10 μm. A crystal sample was analysed by EMSA together with the starting sample solution and the drop solution. The pattern from the recovered crystals showed that there are at least two components in the crystal: one corresponded to the complex (34 kDa), and the other to the DNA double strands only (16 kDa) (Figure 1). However, the crystals obtained by co-crystallization of Prd domain with this 26-bp sequence diffracted very poorly (see below). In order to improve the quality of crystals, crystallization with DNA double strand of sticky type were performed. The obtained crystals grew at lower concentration range of PEG4000 (10-16% w/v), probably due to the higher propensity of sticky DNA double strands to form stronger intermolecular interactions. The morphology of the new crystals was quite different from that of those obtained with blunt-ended DNA (Figure 2). (a)

Figure 1: The electrophoretic mobility shift assay (EMSA) of dissolved crystals confirmed the presence of the protein/oligonucleotide complex in the crystals (lane 2). The excess double strand in lane 2 comes from the solution that wets the crystals.

(b)

Figure 2: (a) Plate-like irregular shaped crystals of the complex formed by Pax-8 Prd domain and DNA blunt-ended sequence. The maximum size of crystals measures 200 × 50 × 10 μm. (b) Plate like shape crystals with typical dimension of 200 × 100 × 50 μm of the complex formed by Pax-8 Prd domain and DNA sticky-ended sequence.

For diffraction studies by synchrotron radiation is necessary to froze the crystal by nitrogen stream. Since the Pax-8-DNA crystals were grown in solution without cryo-protected conditions the crystals were soaked in higher concentrated PEG solution and than mounted in the diffractometer. Several different cryoprotectant agents (ethylene glycol 20% v/v, PEG400 20% v/v, MPD 20%, and glycerol 20%) varying the concentration of the primary precipitant PEG4000 over the range of 4-24% w/v in intervals of 4% were explored. The reservoir was constituted of 4-24% w/v PEG4000, 1 mM DTT, and 20% v/v cryo-protectant. Both twined and single crystals were obtained under condition of 20% ethylene glycol and 12% PEG4000 in three days. Diffraction Studies Diffraction studies were performed at the Elettra synchrotron X-ray beam-line (Trieste, Italy), using a MARCCD detector after preliminary study on rotating anode Enraf-Nonius kappa-CCD detector. The crystals of the complex obtained with the blunt-ended DNA experiment were quickly passed through a solution of 22 % PEG4000 and 20% PEG200 and flash-cooled in a nitrogen stream at 100 K from an Oxford Instruments Cryostream. The diffraction patterns collected with 1.5 Å wavelength showed spots up to 6.0 Å of maximum resolution (Figure 4aS Supplementary Material). The collected images were processed using MOSFLM (21). These diffraction experiments suggested the presence in this crystal form of large internal disorder limiting the scattering at high resolution. A possible solution to improve the crystal order was to increase the intermolecular contacts through the use of sticky-end DNA. The crystallization of the Pax-8 Prd domain with the sticky-ended DNA having the 24 base pair of the C-site and two T and two A bases 5ʹ and 3ʹ overhanging, respectively, was quite promising because the obtained crystals were less soluble compared to the previous crystal form and also their dimension was generally larger. These new crystals were quickly passed through a solution of 30-40% PEG4000 and mother liquid then flash cooled in a stream of nitrogen gas at 100 K. However, no data with resolution better than 8 Å were collected (Figure 4bS Supplementary Material). Also the diffraction experiments with crystals grown in solution having cryo-protected conditions shown any significant improvement to the limit resolution of diffraction pattern. Fluorescence Experiments Fluorescence measurements were performed utilizing excitation wavelength at 280 nm (10.5 nm bandpass). Maximum emission intensity was optimized at 344 nm. The equilibrium binding titrations were performed at 37 ºC in a solution containing 10 mM TRIS pH 8.0, 10 μM DTT, and MgCl2 and at low protein (0.1 μM) and DNA concentration. Fluorescence measurements were corrected for inner filter effect. Molecular Modeling The comparative modeling of Pax-8/DNA complex is based on the structures of Pax-6, Pax-5, and Prd determined by X-ray crystallography (PDB accession codes 6PAX, 1K78, and 1PDN, respectively). Mutations on the Pax-6 structure and rotamer searching were performed within the INSIGHTII framework (Biosym Technologies) on a Silicon Graphics Octane. The initial model of Pax-8/DNA complex was submitted to restrained minimization by means of DISCOVER (Biosym Technologies) with the AMBER force field (22). A distance-dependent dielectric constant (0.4 x, where x is the distance in nm) and an infinite cut-off distance for non-bonded interactions were employed to

433 Pax-8 Prd Domain/ DNA Complex

434

compensate partially for the lack of explicit solvent. The van der Waals’ term for 1-4 interactions was scaled by 0.5 and explicit charges were employed.

Campagnolo et al.

The minimization iterations were performed by means of the steepest-descent algorithm (100 cycles), followed by 1900 steps with conjugate gradients. Hydrogen bonds and hydrophobic interactions were determined by LIGPLOT program (23). Results and Discussion Pax-8 Prd Domain and DNA Sequence As a result of the cloning strategy used, the expressed protein contains 138 amino acids of the Pax-8 sequence plus 8 and 24 extra amino acids at the N and C termini, respectively (see Figure 3).

ARIRARGSIR SGHGGLNQLG GAFVNGRPLP EVVRQRIVDL AHQGVRPCDI SRQLRVSHGC VSKILGRYYE TGSIRPGVIG GSKPKVATPK VVEKIGDYKR QNPTMFAWEI RDRLLAEGVC DNDTVPSVSS INRIIRTKVQ QPFNLPGSSR VDLQPKLIDD KLSNMRIKSI ARIRARGS

β1

β2

α1

α2

α3

Figure 3: Primary sequence of the Pax-8 Prd domain with representation of the secondary structure. Boxes indicate α-helical regions, ovals indicate β-sheets. The 138 amino acids residues constituting the Prd domain are in bold, whereas the 8 and 24 extra residues at the N and C terminus coming from cloning strategy are reported in normal characters.

Figure 4: DNA 26 base pairs sequence and schematic representation of the double helices of the blunt-ended double strand (a) and of the sticky-ended double strand (b) used for crystallization experiments. The second model (b) puts in evidence the overhanging bases at each terminus of the nucleic acid sequence.

α4

α5

α6

GSSRVDLQPKLIDDKLSNMRIKSI

The choice of the DNA fragment plays a significant role in the success of a crystallization experiment. In some successful crystallization trials, it has been shown (24) that it is essential not only to have high DNA binding affinity but, since periodicity of the crystal arrangements, it is very important to use DNA sequence of lengths corresponding to multiples of integral or half integral turns of the DNA (11, 15, 16, 20, 21, 26 base pairs). Pax-8 Prd domain binding affinity for the C site of the Thyroglobulin promoter has been previously studied and characterized in great detail (17). Another important variable for successful crystallization is the mode of pairing the 3ʹ and 5ʹ DNA ends. Single or double complimentary overhanging bases at either the 3ʹ and 5ʹ end (sticky ends) are frequently included to promote end-to-end stacking of oligonucleotides in the crystal, although the complete pairing of oligonucleotides (blunt-ended) has yielded high-quality crystals in a number of cases. In a first series of experiments the blunt-ended double

strand was used, engineered, respectively, with two bases 5ʹ A and two bases 3ʹ T (Figure 4a). The crystals obtained by co-crystallization of Prd domain with this 26-bp sequence gave very poor diffraction. The low diffraction property was attributed to the disorder inside the crystal for the absence of end-to-end stacking of the oligonucleotides. A second double strand with sticky ends was prepared with two bases 5ʹ overhanging T (instead of 3’) (Figure 4b). Comparison Between Two Crystal Forms of Pax-8 prd Domain-DNA Complex Analysis of the diffraction pattern of crystals of Pax-8 prd domain in complex with blunt-ended DNA indicated that these crystals belong to the C centred monoclinic system with cell dimension: a = 89.88 Å, b = 80.05 Å, c = 67.73 Å, and β = 124.3º, while crystals obtained with sticky-ended DNA belonged to the same crystal system, but with a larger unit-cell: a = 256.56 Å, b = 69.07 Å, c = 99.32 Å, and β = 98.1º. The calculation of the crystal volume per protein mass (VM) (25) in the crystals obtained with DNA blunt-ended and in the crystals with double strand sticky-ended DNA gave in the first case the Matthews coefficient is 2.96 Å3 Da-1, assuming one crystallographically independent unit of the complex made up of a protein chain and one double strand DNA fragment, while in the second case it is necessary to assume four molecules of the protein-DNA complex to obtain an acceptable solvent content and a Matthews value of 3.20 Å3 Da-1 (Table I). Table I Unit cell parameters, estimation of number of protein/DNA complex molecules in the asymmetric unit, and solvent content (assuming a protein density of 1.35 g cm-3) of the two different crystal forms with blunt and with sticky ended oligonucleotides. Data and parameters Unit cell a, (Å) b, (Å) c, (Å) , (°) Crystal system Bravais lattice Unit cell volume (Å3) Molecular Weight (kDalton) No. of complex in asymmetric unit VM (Å3 Da -1) Solvent content

Pax8-Prd domain and DNA blunt-ended complex

Pax8-Prd domain and DNA sticky-ended complex

89.88 80.05 67.73 124.3 Monoclinic C 402768.5 34 1 2.96 58.4%

256.56 69.07 99.32 98.1 Monoclinic C 1742712.9 34 4 3.20 61.5%

Salt Effect on DNA-protein Interaction and Crystal Formation Several crystallization trials were performed to improve crystal quality. Addition of MgCl2 solution evidently improved crystal aspect. From 10 to 30 mM MgCl2 solutions were tested in presence of different precipitating agent (isopropanol, MPD, PEG8000). Higher salt concentration corresponded to bigger and more transparent crystals (Figure 5S Supplementary Material). The main impression was that the presence of ions increasing ionic strength might contrast the de-hydratating action of precipitating agent, favoring the slow growth of crystals. The benefits of salt effects in the crystallization experiments evidenced that a certain ionic strength is necessary to regulate specific interaction between DNA consensus sequence and Pax-8 Prd domain peptide. In particular, the formation of better quality crystals depend on the presence of a significant amount of MgCl2. The reason may be correlated to the nature of the protein-DNA interaction. The magnitude of the ionic strength effects is highly depen-

435 Pax-8 Prd Domain/ DNA Complex

436 Campagnolo et al.

Figure 5: Schematic representation of the Pax-8 Prd domain-DNA complex. The formation of the complex can promote an hydrogen bond between tryptophan 97 and DNA (5ʹ hydroxyl of the modeled structure and phosphate group in the double strand DNA used in the experiments) quenching the fluorescence signal of the aromatic residue.

Figure 6: The normalized emission spectra (F/F0, fluorescence intensity of the sample after complex formation normalized with fluorescence of the peptide solution) is reported in function of the ratio (added-DNA):(Prd domain) concentrations. The addition of 200 mM MgCl2 (gray curve) in the sample solution respect to the addition of 20 mM MgCl2 (black curve) diminishes the quenching effect due to the complex formation.

Figure 7: Overview of the Pax-8 Prd domain-DNA complex. The protein (cartoon representation) interacts with DNA surface by the PAI (right side) and RED (left side) subdomains.

437 Pax-8 Prd Domain/ DNA Complex

a)

H47 H47 H47 H47 H47 K52

N14 N14 N14 N14 R16 R16 I68 R16 R16

S71 S71 S71 S71 R74 K74 R122 R122 R122 R122 S118 S118 S119 R122 S118 S118

Pax-8 Hydrogen bond

Pax-6

TAGA CTGA ATAC ATGG GTAT ACTC AAGA CCGC AGGG CCCG TATT TTCT GGCG AAAA CGGC TTT GGG GCG GAC CCC AA GG

CTA CAG GTAT CCAT ATAC GAGT TCTT GCGG CCCT CGGG AATA AGAA CGCC TTTT GCCG TAAA TCCC CGC GTC GGG ATT ACC

Pax-5

Prd

S46 C49 K52 K52 K52

I68 I68 I68

K74

R125 N121 N121

Water mediated hydrogen bond van der Waalʼs interactions

Figure 8: Comparison of specific protein-DNA interactions between amino acids side chains and DNA bases. (a) H-bonds (yellow highlighted), water mediated H-bonds (in box), and van der Waal’s contacts (in bold) in Pax-8 (red), Pax-6 (blue), Pax-5 (green), and Prd (black) complexes; (b) sequence alignment of Pax8, Pax-6, Pax-5, and Prd with evidenced the H-bonds (yellow highlighted), the water mediated H-bonds (red), and van der Waal’s contacts (in bold).

b) Pax-8 Pax-6 Pax-5 Prd

------------------MDLEKNYPTP ----------

-----GHGGL -----SHSGV RTSRTGHGGV -----GQGRV

NQLGGAFVNG NQLGGVFVNG NQLGGVFVNG NQLGGVFING

RPLPEVVRQR RPLPDSTRQR RPLPDVVRQR RPLPNNIRLK

IVDLAHQGVR IVELAHSGAR IVELAHQGVR IVEMAADGIR

Pax-8 Pax-6 Pax-5 Prd

PCDISRQLRV PCDISRILQV PCDISRQLRV PCVISRQLRV

SHGCVSKILG SNGCVSKILG SHGCVSKILG SHGCVSKILN

RYYETGSIRP RYYATGSIRP RYYETGSIKP RYQETGSIRP

GVIGGSKPKV RAIGGSKPRV GVIGGSKPKV GVIGGSKPRI

ATPKVVEKIG ATPEVVSKIA ATPKVVEKIA ATPEIENRIE

Pax-8 Pax-6 Pax-5 Prd

DYKRQNPTMF QYKQECPSIF EYKRQNPTMF EYKRSSPGMF

AWEIRDRLLA AWEIRDRLLS AWEIRDRLLA SWEIREKLIR

EGVCDNDTVP EGVCTNDNIP ERVCDNDTVP EGVCDRSTAP

SVSSINRIIR SVSSINRVLR SVSSINRIIR SVSAISRLVR

TKVQQPFNNLASEKQQTKVQQPPNQ GRD------

Hydrogen bond Water mediated hydrogen bond van der Waal's contacts

Figure 9: Stereoview of Pax-8 Prd domain-DNA complex. Both cysteine residues of the PAI domain (C37, C49) interact with DNA forming two hydrogen bonds between the thiol side chains and the phosphate groups of DNA. The cysteine of the RED domain (C109) is exposed to the solvent.

438 Campagnolo et al.

dent on electrostatic interactions between negatively charged phosphates of DNA and positively charged groups of protein. Recently, Norberg (26) has calculated the salt-dependent part of the free energy for several protein-DNA complexes. In most cases, the protein has an overall positive charge and in general the salt-dependent electrostatic free energy increases with higher ionic concentrations and, therefore, complex association is stronger opposed at higher ionic concentrations. In the case of regulatory proteins, the major driving force is constituted by release of both ions and water molecules upon binding of a protein to specific DNA sequence (27-29). The release of small species is thus expected to provide the significant entropic driving force for formation of the protein-DNA complexes (30). On the other hand, the release of solvent molecules upon protein binding to non-specific DNA sequences is less significant and the formation of non-specific protein-DNA complexes is driven primarily only by ions release (31). It has been suggested that the water molecules released in the proper complexes may occur because of direct, specific hydrogen bonding of protein side-chains with the polar atoms in the major and minor grooves of DNA, which does not occur in the non-specific complexes (27-29). The increase of ionic strength reduces the importance of electrostatic interactions, both specific and non-specific, and consequently increases the role of specific intermolecular hydrogen bond formation. Then, the relatively large electrostatic component of the DNA-protein interaction suggests that the specificity of the DNA-protein interaction increases as the salt concentration is increased. Fluorescence Experiments In order to quantify the affinity of protein-DNA complex and its dependence on salt concentration a fluorescence experiment was set up. A common way to determine protein affinity for DNA is by measuring the fluorescence quenching of tryptophan residues, often at a single wavelength (32). Tryptophan fluorescence has proven to be a very useful spectroscopic marker for major state transition since it is sensitive to possible changes in structure and dynamics that might occur in proteins upon interactions with DNA duplex (33). Trp97 is the only tryptophan residue present in the Pax-8 Prd domain sequence, and it is highly conserved in all Pax species. The structural model of Pax-8 Prd domain-DNA complex shows that upon peptideDNA interaction, Trp97 contacts the DNA through Nε1 (Figure 5). Therefore, the formation of new hydrogen bonds may quench tryptophan fluorescence. Pax-8 Prd domain-DNA consensus site binding affinity through gel-retardation assays can be evaluated below 50 nM (17). However, these gel-retardation measurements were performed in non-equilibrium conditions. Tryptophan fluorescence spectroscopy is a valid technique to calculate binding affinity in complex system like protein/DNA maintaining complex formation equilibrium. Fluorescence measurements were performed utilizing excitation wavelength at 280 nm (10.5 nm bandpass). Maximum emission intensity was optimized at 344 nm (Figure 6S Supplementary Material). The normalized emission spectra are reported against the ratio [added-DNA]:[Prddomain] (Figure 6). The fast quenching of fluorescence signal, soon after the first addition of DNA duplex, far away from the 1:1 stoichiometric ratio between peptide and consensus sequence, demonstrate the large effect of non-specific recognition between the molecules even at elevated salt concentration. The irregular response of the system to DNA addition did not allow a calculation of the binding affinity. The non-specific interactions were demonstrated also by the fluorescence quenching in an additional experiment using a non-specific double strand sequence (polidIdC). Molecular Modeling The overall structure of the Pax-8/DNA complex obtained by minimization energy (Figure 7) is very similar to that seen with other crystal structure Pax-6, Pax-5, and

Prd. Both subdomains have three alpha helices with an additional β-hairpin in the PAI subdomain. The last helix of the two subdomains (helices 3 and 6) lies in the major groove of DNA while the extended polypeptide linker (residues 61-76) lies in the minor groove and make extensive contact with DNA. Most of the protein-DNA interactions occur with the DNA backbone or involve the protein main chain and only a small part of the protein-DNA interactions involves a direct side chain-base bonding. However, the specificity of protein-DNA interaction should be ascribable to these specific contacts between amino acids side chains and DNA bases. A comparison of these specific interactions between the minimized structure of Pax-8 with the crystal structures of Pax-6, Pax-5, and Prd complexes is shown in Figure 8. Both ionic (hydrogen bonds and salt bridges) and van der Waals’s interactions are schematically reported. Noticeably, most of the conserved contacts occur within a single DNA strand. The overall most conserved interaction involves residue Ser 71. It forms in all four structures a direct hydrogen bond with adenine 14. Another important specific interaction involves Asn 14. This Asn residue forms hydrogen bonds with guanine 9 in Prd, Pax-5, and Pax-5, while in Pax-8 interacts with an adenine base thought a water molecule. In Pax-6 an additional interaction of Asn 14 with the DNA base is mediated by a water molecule. Another specific polar contact involves Ser 118, that interacts with a cytosine or a guanine, directly in Pax-6 and Pax-5, respectively. Furthermore, in Pax-5 it forms together with Ser 119 and Arg 122 a cluster of H-bonds via water molecules with the preceding guanidine base. The contact that involves the residues found in position 47 is also very recurrent, even though here both polar and van der Waal’s interactions are present. The side chain of the His found at this position in Prd and in Pax-5 forms a hydrogen bond directly with guanine 4. In Pax-6, His 47 is replaced by Asn 47 that is in van der Waal’s contact with thymine 4. This finding prompted to hypothesize that the His in this position specifically recognizes a guanine (13). In fact, in our model of Pax-8 His 47 is simply in van der Waal’s contact with adenine 3 and does not make any polar interaction with the adenine base in position 4. Lys 52 side chain makes an hydrogen bond mediated by water with a guanine of the other DNA strand in Pax-6, Pax-8, and Prd. Interestingly, this contact is lost in Pax-5 where cytosine replaces the guanidine in this position and the lysine side chain forms a hydrogen bond, also mediated by solvent, with the thymine 6 of the other DNA filament. On average, van der Waal’s interactions involving a direct contact between residue side chains and DNA bases seem to be less conserved than polar ones. Three recurring van der Waal’s interactions are observed involving His 47, Ile 68, and Arg 122. In particular, for Ile 68 and Arg 122, these interactions are quite conserved and the side chains contact adenine and thymine bases, respectively. CD experiments have established that the binding activity of the Prd domain of Pax-8 is regulated through the oxidation/reduction of cysteine residues (18, 19). The comparative modeling of Pax-8 Prd domain-DNA complex shows that two of these residues located at positions 37 and 49 of the PAI subdomain make H-bond interaction with the phosphate groups of two DNA strands, while the 109 cysteine of the RED domain is exposed to the solvent (Figure 9). It is interesting to note that the cysteine residues of the PAI subdomain delimitate the region of the protein that interacts with the DNA major groove. This is consistent with the observation that oxidation of the cysteine residues of PAI subdoman induces structural variations leading to an interference with the proper conformation useful for DNA binding, while the RED subdomain should be poorly sensitive to redox control (18). Conclusions Pax-8 is a transcriptional factor that activates the promoters of the thyroglobulin (Tg) and thyroperoxidase (TPO) genes in the follicular thyroid cell, by interacting through its Prd domain with sites located near the transcriptional start sites. The knowledge of the structural details of the interactions between the Prd domain of

439 Pax-8 Prd Domain/ DNA Complex

440 Campagnolo et al.

Pax-8 and its DNA consensus sequence is particularly important to reveal consequences of specific Pax mutations and in general the biological functions of the Prd domain. Crystals of the complex, constituted by the Pax 8 Prd-domain and the DNA sequence of the C site of the Tg promoter, have been obtained. Preliminary X-ray diffraction analysis of two crystalline forms, grown respectively in the presence of blunt-ended and of sticky-ended DNA sequences, allowed the determination of unit cell dimensions and the assignment of the space group. However, the low resolution of the diffraction patterns, mainly attributed to an intrinsic disorder of the crystalline lattice, did not allow the determination of the crystal structure. Many attempts to improve the crystal quality were unsuccessful. In spite of this, some interesting information about the parameters that have a significant effect on crystallization of the protein-DNA complex was obtained. Growth of transparent crystals with large dimensions and regular shape is particularly influenced by addition of salt. In order to investigate the effects of salt addition on the complex Pax-8 Prd domain-DNA constitution equilibrium, fluorescence spectroscopy experiments were performed based on the tryptophan emission signal. These experiments showed in our case the formation of multimeric species with high DNA/ protein ratio over a wide range of ionic strength. This suggests that non-specific interactions between the protein containing the Prd domain and the DNA C-site prevail over the formation of the specific binary complex. Consequently, estimation of the binding affinity constant of the investigated binary complex through fluorescence experiments is hampered. Fluorescence experiments suggested that the crystalline disorder can be attributed to the non-homogeneity of the proteinDNA sample. An important role could be played by the 30 extra amino acids present in the expressed protein. The extra amino acids are distributed, 9 at the N- and 21 at the C- termini of Prd domain. These additional sequences contain numerous positive charged residues that may increase the interaction with the negative DNA double strands through non-specific contacts. Furthermore, these highly hydrophilic extra peptidic chains probably form loops structurally disordered that may affect the construction of an ordered three-dimensional arrangement, compromising the growth of crystals suitable for structure determination. In absence of an experimental model of Pax-8 Prd domain-DNA complex the theoretical comparative model shows the probable conservation of some specific protein DNA interactions observed in the other Pax proteins and the important role of the cysteine residues of PAI subdomain in the redox control of the DNA recognition. Acknowledgements This work was supported by the Ministero dell’Istruzione dell’Università e della Ricerca, Rome (FIRB-RBNE0155LB). References and Footnotes 1. A. Mansouri, M. Hallonet, and P. Gruss. Curr Opin Cell Biol 8, 851-857 (1996). 2. G. Frigerio, M. Burri, D. Bopp, S. Baumgartner, and M. Noll. Cell 47, 735-746 (1986). 3. G. Chalepakis, A. Stoykova, J. Wijnholds, P. Tremblay, and P. Gruss. J Neurobiol 24, 13671384 (1993) 4. P. Stapleton, A. Weith, P. Urbanek, Z. Kozmik, and M. Busslinger. Nat Genet 3, 292-298 (1993). 5. J. Wallin, Y. Mizutani, K. Imai, N. Miyashita, K. Moriwaki, M. Taniguchi, H. Koseki, and R. Balling. Mamm Genome 4, 354-358 (1993). 6. C. Walther, J. L. Guenet, D. Simon, U. Deutsch, B. Jostes, M. D. Goulding, D. Plachov, R. Balling, and P. Gruss. Genomics 11, 424-434 (1991). 7. G. Chalepakis, R. Fritsch, H. Fickenscher, U. Deutsch, M. Goulding, and P. Gruss. Cell 66, 873-884 (1991). 8. J. Treisman, E. Harris, and C. Desplan. Genes Dev 5, 594-604 (1991). 9. D. Bopp, M. Burri, S. Baumgartner, G. Frigerio, and M. Noll. Cell 47, 1033-1040 (1986). 10. T. Strachan and A. P. Read. Curr Opin Genet Dev 4, 427-438 (1994). 11. M. Tassabehji, A. P. Read, V. E. Newton, M. Patton, P. Gruss, R. Harris, and T. Strachan. Nat Genet 3, 26-30 (1993). 12. T. Glaser, D. S. Walton, and R. L. Maas. Nat Genet 2, 232-239 (1992). 13. W. Xu, M. A. Rould, S. Jun, C. Desplan, and C. O. Pabo. Cell 80, 639-650 (1995). 14. H. E. Xu, M. A. Rould, W. Xu, J. A. Epstein, R. L. Maas, and C. O. Pabo. Genes Dev 13, 1263-1275 (1999).

15. M. Zannini, H. Francis-Lang, D. Plachov, and R. Di Lauro. Mol Cell Biol 12, 42304241 (1992). 16. T. Di Palma, R. Nitsch, A. Mascia, L. Nitsch, R. Di Lauro, and M. Zannini. J Biol Chem 278, 3395-3402 (2003). 17. L. Pellizzari, G. Tell, and G. Damante. Biochem J 337 (Pt 2), 253-262 (1999). 18. G. Tell, A. Scaloni, L. Pellizzari, S. Formisano, C. Pucillo, and G. Damante. J Biol Chem 273, 25062-25072 (1998). 19. G. Tell, L. Pellizzari, D. Cimarosti, C. Pucillo, and G. Damante. Biochem Biophys Res Commun 252, 178-183 (1998). 20. X. Cao, F. Kambe, X. Lu, N. Kobayashi, S. Ohmori, and H. Seo. J Biol Chem 280, 2590125906 (2005). 21. A. Leslie. Crystallographic Computing (1990). 22. P. A. Kollman, S. Weiner, G. Seibel, T. Lybrand, U. C. Singh, J. Caldwell, and S. N. Rao. Ann NY Acad Sci 482, 234-244 (1986). 23. A. C. Wallace, R. A. Laskowski, and J. M. Thornton. Protein Eng 8, 127-134 (1995). 24. A. Batchelor, D. Piper, and C. Wolberger. Crystallization of Protein-nucleic Acid Complexes (2000). 25. B. W. Matthews. J Mol Biol 33, 491-497 (1968). 26. J. Norberg. Arch Biochem Biophys 410, 48-68 (2003). 27. M. M. Garner and D. C. Rau. Embo J 14, 1257-63 (1995). 28. J. H. Ha, M. W. Capp, M. D. Hohenwalter, M. Baskerville, and M. T. Record, Jr. J Mol Biol 228, 252-64 (1992). 29. M. G. Fried, D. F. Stickle, K. V. Smirnakis, C. Adams, D. MacDonald, and P. Lu. J Biol Chem 277, 50676-50682 (2002). 30. S. Khrapunov and M. Brenowitz. Biophys J 86, 371-383 (2004). 31. P. L. deHaseth, T. M. Lohman, and M. T. Record, Jr. Biochemistry 16, 4783-4790 (1977). 32. T. Lohman and D. Mascotti. Thermodynamics of Ligand-Nucleic Acid Interactions 212, 400424 (1992). 33. J. M. Beechem and L. Brand. Annu Rev Biochem 54, 43-71 (1985).

Date Received: October 11, 2006

Communicated by the Editor Valery Ivanov

441 Pax-8 Prd Domain/ DNA Complex