A Constitutive NADPH Oxidase-Like System Containing gp91phox Homologs in Human Keratinocytes

A Constitutive NADPH Oxidase-Like System Containing gp91phox Homologs in Human Keratinocytes Walee Chamulitrat, Wolfgang Stremmel, Tsukasa Kawahara,w Kazuhito Rokutan,w Hirotada Fujii,z Kirstin Wingler,y Harald H. H. W. Schmidt,y and Rainer Schmidtz

Department of Internal Medicine IV, University of Heidelberg, Heidelberg, Germany; wDepartment of Nutrition, Tokushima University, Tokushima, Japan;

zSchool of Health Sciences, Sapporo Medical University, Sapporo, Japan; yRudolf-Buchheim Institute for Pharmacology, University of Giessen, Giessen, Germany; zThe German Cancer Research Center, Applied Tumorvirology, Heidelberg, Germany

In non-phagocytic cells, superoxide has been implicated in physiological and pathological cellular functions in the skin and mucosa, such as, host defense, mitogenic responses, and malignant conversion. Here, we identify a constitutively expressed heme-flavoprotein NADPH oxidase (Nox) system as a source of superoxide in human skin (HaCaT) and gingival mucosal (GM16) keratinocyte cell lines. Western blot analysis showed that both cell lines expressed the phagocyte oxidase (phox) cytosolic proteins Rac1, p40phox, and p67phox. With respect to the catalytic flavoheme protein subunit, HaCaT membranes, which expressed p22phox, showed an absorbance peak at 558 nm indicative of a b-type cytochrome. At mRNA levels, both GM16 and HaCaT cells expressed gp91phox homologs Nox1, Nox2, and Nox4, however, HaCaT cells expressed very low levels of Nox1 mRNA. At protein levels, Nox1 was readily detected in HaCaT but was nearly undetectable in GM16 cells. Consistently, Nox activity of HaCaT membranes was demonstrated by electron paramagnetic resonance spin-trapping and cytochrome c reduction, and the activity was sensitive to the flavoprotein inhibitor diphenylene iodonium. Vmax values were 20-fold lower than those reported for phagocytic oxidase. In conclusion, keratinocytes expressed a Nox distinct from the phox isoform of phagocytes providing molecular evidence for a source of superoxide that may regulate cell proliferation and host defense in skin and oral mucosa.

Key words: human skin and oral epithelial cells/oxidoreductase/p67phox/spin trapping/superoxide radical J Invest Dermatol 122:1000 –1009, 2004 It has been recognized that low levels of superoxide and other reactive oxygen species (ROS) from non-phagocytic cells act as signal transducers in the control of mitogenic signaling (Finkel, 1998; Joneson and Bar-Sagi, 1998) and oxygen sensing (Babior, 1999; Jones et al, 2000). ROS produced by non-phagocytic cells have been associated with altered cell functions and cell injuries, such as senescence in fibroblasts (Lee et al, 1999), cell-shape changes in synovial fibroblasts (Kheradmand et al, 1998), shear-induced injury in endothelial cells (Yeh et al, 1999), balloon-injured arterial smooth muscle cells (Patterson et al, 1999), Helicobacter pylori infection in gastric mucosal cells (Teshima et al, 1998), as well as neoplastic transformation of keratinocytes (Nakamura et al, 1988; Gupta et al, 1999). Recently, nicotinamide adenine dinucleotide phosphate (reduced form) (NADPH) oxidases have been under intensive investigations as a source of superoxide in non-

Abbreviations: DEPMPO, 5-diethoxyphosphoryl-5-methyl-1-pyrroline N-oxide; DPI, diphenylene iodonium chloride; DTPA, dithylenetriaminepentaacetic acid; EPR, electron paramagnetic resonance; GM16, human gingival keratinocytes immortalized by human papillomavirus type 16; GTP-gS, guanosine 50 -O-(3thiotriphosphate); NADH, nicotinamide adenine dinucleotide (reduced form); NADPH, nicotinamide adenine dinucleotide phosphate (reduced form); Nox, NADPH oxidase; phox, phagocyte oxidase; ROS, reactive oxygen species; SDS, sodium dodecyl sulfate; SOD, superoxide dismutase

phagocytic cells (Finkel, 1998; Babior, 1999; Griendling et al, 2000; Jones et al, 2000). NADPH oxidases catalyze the one-electron reduction of molecular oxygen to superoxide using NADPH as an electron donor. The phagocyte NADPH oxidase consists of a membrane-bound flavocytochrome composed of p22phox (phox, phagocyte oxidase) and the catalytic flavo-heme protein gp91phox (Nox2), the cytosolic G protein Rac and subunits p40phox, p47phox, and p67phox (Babior, 1999; Jones et al, 2000). At least two Rac G proteins (Rac1 or Rac2) act as molecular switches in the control of the oxidase activity, alternating between binding guanosine diphosphate in the inactive state and guanosine triphosphate in the active state. Many, but not all of phagocyte NADPH oxidase components have been found in non-phagocytic cells. The cytochrome b of NADPH oxidase in fibroblasts was found to be structurally and genetically different from that of neutrophil NADPH oxidase (Meier et al, 1993). It has been recently reported that many non-inflammatory tissues express more than one homolog of gp91phox; this gp91phox homolog group is termed Nox (for NAD(P)H oxidase) consisting of Nox1, Nox3, Nox4, and Nox5 (Cheng et al, 2001). To date, Nox genes have been implicated to play a pathophysiological role, such as, tumorigenicity (Arnold et al, 2001) and cardiovascular diseases (Griendling et al, 2000; Wingler et al, 2001). In HaCaT skin keratinocyte cell line, stimuli such as epidermal growth factor, Ca2 þ -ionophore A23187 (Goldman

Copyright r 2004 by The Society for Investigative Dermatology, Inc.

1000

122 : 4 APRIL 2004

HUMAN KERATINOCYTE gp91phox HOMOLOGS

1001

et al, 1998), platelet-activating factor (Goldman et al, 1999), or lysophosphatidic acid (Sekharam et al, 2000) are capable of producing ROS in a calcium-dependent manner. This activation is diphenylene iodonium (DPI) sensitive indicating an involvement of flavoprotein(s). Recent data have implicated the physiological role of Nox in HaCaT during TGFb-induced activation of p38 (Chiu et al, 2001), and in induction of vascular endothelial growth factor and wound healing (Sen et al, 2002). An earlier study using a tumorigenic oral keratinocyte H357 cell line, however, reported that DPI was ineffective in inhibiting superoxide generation thus not supporting the presence of Nox (Turner et al, 1998). Here, we show that immortalized human skin keratinocyte (HaCaT) and gingival mucosal (GM16) cell lines constitutively express Nox components Rac1, p40phox, and p67phox proteins. GM16 cell line used in our study was different from H357 cell line (Turner et al, 1998) in that GM16 cells were generated from normal human gingiva, not from tumor origin. We found that HaCaT and GM16 cells differentially express Nox1 at mRNA and protein levels. The difference spectra of the heme in HaCaT membranes, which contain p22phox, were similar to that of the heme of neutrophil membranes. The DPI-sensitive activity of Nox was demonstrated in a cell-free system by using electron paramagnetic resonance (EPR) spin-trapping technique and cytochrome c reduction.

Results Expression of Nox components in human keratinocytes In a previous report (Turner et al, 1998), while mRNAs of p67phox, p47phox, and p22phox were detected, none of the essential phox proteins was detected in the tumorigenic H357 oral keratinocyte cell line. Functional evidence, on the contrary, suggests the presence of a Nox-like source of superoxide in HaCaT skin keratinocyte cell line possibly involved in wound healing (Chiu et al, 2001; Sen et al, 2002). We aimed to measure Nox subunits and activity in immortalized human keratinocytes (i.e., at an early stage of cell transformation) of cutaneous (HaCaT) and gingival mucosal (GM16) origin. Western blot analysis showed the presence of Rac1 at 21 kDa in HaCaT, GM16, as well as in positive controls from mouse brain and human B MHH cells (Fig 1A). Rac2 protein was not detected in HaCaT and GM16 cell lysates, but in the lysates from positive control MHH cells (data not shown). The anti-p67phox monoclonal antibody recognized a protein at slightly lower molecular weight than expected in cell lysates of HaCaTs, GM16, as well as in positive controls from HL60 and MHH cells (Fig 1B). But we were not able to detect p47phox in cell lysates using two different antibodies and immunoprecipitates of HaCaT and GM16 cells whereas positive controls from HL60 and MHH cells expressed a 47 kDa immunoreactive protein (Fig 1C). p40phox was detected with the expected electrophoretic mobility by immunoprecipitation and immunodetection using p40phox antiserum (Bouin et al, 1998) in HaCaT, GM16 and, as positive control, in MHH cells (Fig 2A). A positive control of a recombinant histidine-tagged p40phox (treated the same way as samples) gave rise to a protein with the expected

Figure 1 Western blot analysis of the expression of NADPH oxidase cytosolic proteins in HaCaTs and GM16 cells. Western blotting was performed as described in Materials and Methods. (A) Rac1 protein was expressed in HaCaT, GM16, and positive controls brain and MHH cells. (B) p67phox protein was expressed in HaCaT, GM16, positive controls HL60, and MHH cells. (C) p47phox protein was expressed only in HL60 and MHH cells, but not in HaCaT and GM16 cells.

larger size at 45 kDa, which was detected just below the strong IgG signal of the antiserum at 55 kDa. The membrane component of phagocyte oxidase consists of subunits p22phox and gp91phox (Nox2). Consistently, p22phox protein was detected in the membranes of HaCaT and undifferentiated HL60 cells (Fig 2B). We further investigated whether mRNAs of gp91phox Nox homologs were present in these cells. Two independent RT-PCR assays were performed with a two-step PCR (Fig 3) and real-time RT-PCR using primers and TaqMan (Applied Biosystems, Darmistadt, Germany) probes listed in Table I. Both assays were positive for Nox1 mRNA which was detected as a 171 bp product in GM16 cells, and very weak (almost undetectable) product was observed in HaCaT cells (Fig 3A, and data analyzed by Nox1-TaqMan probes not shown). Caco-2 (Fig 3A) and HT-29 (not shown) cells known to express Nox1 mRNA were used as positive controls in our assays (Cheng et al, 2001; Perner et al, 2003). The primers for two-step nested PCRs were designed towards the 50 region of Nox1 sequence, hence excluding the detection of the short splicing-form of Nox1 (as described in Banfi et al, 2000). These primers, however, will detect both

1002 CHAMULITRAT ET AL

THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 2 Western blot analysis of p40phox in cell lysates of HaCaTs and GM16 cells, and of p22phox in membranes of HaCaT cells. For experimental details see Materials and Methods. (A) p40phox protein was expressed in HaCaT, GM16, and positive control MHH cells. An additional positive control was recombinant histidine-tagged p40phox protein, which showed a heavier protein at 45 kDa indicated with an arrow. (B) p22phox protein was expressed in membranes of HaCaT and positive control undifferentiated HL60 cells.

Nox1 and Nox1-long truncated variant (the latter was described in Banfi et al, 2000). Both cell lines expressed Nox2 mRNA (Fig 3B, and data analyzed by Nox2-TaqMan probes not shown). The product was present in positive control HT-29 cells (but not in Caco-2 cells), consistent with a recent report (Perner et al, 2003). By using TaqMan probes, we found that Nox4 mRNA was present (whereas Nox3 mRNA was absent) in a very little amount in HaCaT and a significant amount in GM16 cells (data not shown). RT-PCR of Nox4 mRNA using Nox4 primers listed in Table I showed the expected product in GM16 cells but not in HaCaT cells (Fig 3C). For Nox1 protein expression, anti-rabbit Nox1 antiserum (from Rokutan Laboratory) recognized a number of immunoreactive products, one with the strongest signal at 63 kDa in HaCaT (Fig 4A) and HT29 (Fig 4B) membranes. Very weak barely detectable 63 kDa protein was detected in GM16 membranes (Fig 4A). As shown in Fig 4B, membranes of HaCaT cells grown at proliferating 50% confluence phase (HaCaT, sub) expressed slightly more abundant 63 kDa protein than those membranes of cells grown at 85% confluence (HaCaT, con). This indicates that Nox1 expression is associated with cell proliferation. A relatively strong band at 40 kDa in Fig 4 was an unspecific staining of Nox1 antiserum as previously observed in Caco2 membranes (Yoshida et al, 2002). Due to the presence of many proteins detected in Western analysis, the specificity of anti-Nox1 antiserum was demonstrated in Fig 5. In Fig 5A, we show that anti-

Figure 3 mRNA of Nox1, Nox2 and Nox4 in HaCaT and GM16 cells. Primers are shown in Table I with PCR methods described in Materials and Methods. (A) The expected 171 bp product of Nox1 from nested PCR was detected at the annealing temperature of 611C in GM16 cells, and in positive control Caco-2 cells. Very faint almost undetectable 171 bp band was detected in HaCaT cells. Water control did not produce any PCR products. (B) The expected 550 bp product of Nox2 mRNA was detected at the annealing temperature of 621C in HaCaT, GM16, and positive control HT29 cells. (C) The expected product of Nox4 mRNA was detected in GM16cells at the annealing temperature of 581C.

Nox1 antiserum recognized the GST-Nox1 protein, but not GST alone, as demonstrated by Western blotting (upper panel) and protein staining (lower panel). With an immunoabsorption test, we show that GST-Nox1 expression in Western analysis was completely abolished by adding 50 molar excess amount of antigen peptide to the antiserum (Fig 5B). Similar to HaCaT membranes, Nox1-immunoblotting of membranes from T84 cells also showed multiple protein bands including the 63 kDa protein (Fig 5B). In T84 membranes, the antigen peptide was able to significantly decrease the 63 kDa protein as well as the two upper bands (Fig 5B, left-hand panel). In HaCaT membranes, the antigen peptide was also able to completely abolish the 63 kDa protein band (Fig 5B, right-hand panel). Taken together, we verified that the antiserum recognized Nox1 protein, and that Nox1 was indeed the 63 kDa protein detected in membrane samples. To measure the flavocytochrome subunits, we analyzed detergent extracts of HaCaT membranes by using reducedminus-oxidized difference spectroscopy. As shown in Fig 6,

122 : 4 APRIL 2004

HUMAN KERATINOCYTE gp91phox HOMOLOGS

1003

Table I. Primers for the detection of human NADPH oxidase (Nox) isoforms for RT-PCR reactions shown in Fig 3, and primers for real-time RT-PCR using TaqMan using probes Target

Forward primer sequence (50 ! 30 )

Reverse primer sequence (30 ! 50 )

TaqMan probe sequence

–

RT-PCR in Fig 3 Nox1

CAGGGAGACAGGTG CCTTTTCC

GAACCAGAGCAGTCCA AACTCG

Nested Nox1

CATCGTGACAGGTCT

CAGTGGCTATAAGGTCA

GAAACAG

GACCT

GGAGTTTCAAGATGC

TCGTAGAGGTTGAGAC

GTGGAAACTA

TCAGACCG

TGTCCTGCTTTTCTG GAAAACC

GTAGTCAGAATTGGCT TGGTCG

AAGCCGACAGGCCA

GTCACATACTCCACTGTC

CCTTGCCTCCATTCTC

CAGAT

GTGTTTC

TCCAGCCTATCTC

GCAGCCTGCCTGAA

TGAGCAGCACGCAC

TTGCCAGTCTGTCGA

TTTCA

TGGA

AATCTGCTGTCCTT

CTGGACCTTTGTGCC

ACCATTCGGATTTCCA

AGCAATAAGCCAGTC

TGTACTG

TGACA

ACCATCATTTCGGTC

Nox2

Nox4

– –

TaqMan RT-PCR Nox1

Nox2

Nox4

HaCaT membranes contained a b-type cytochrome with spectral characteristics similar to those of human neutrophil flavocytochrome b558. Both peaks of the a-band at 558–559 nm and the Soret g-band at 426–427 nm were detected. The a-band of HaCaT membranes appeared to be broader than that of neutrophil membranes, which are known to contain Nox2, but not Nox1 (Perner et al, 2003). Nox activity in a cell-free system from HaCat cells To further corroborate these findings we analyzed Nox activity in a reconstituted cell-free system (Clark et al, 1987; Peveri et al, 1992). Membranes and cytosol from unstimulated HaCaT cells were prepared. NADPH-dependent superoxide generation by HaCaT membranes was determined by EPR spin-trapping (Fig 7) and superoxide dismutase (SOD)inhibitable reduction of ferricytochrome c (Fig 8). In spintrapping protocol, superoxide radical reacts to spin trap DEPMPO to form radical adduct designated as DEPMPO/ OOH (Frejaville et al, 1995). DEPMPO/OOOH may be reduced enzymatically in the samples to form DEPMPO/ OH (Finkelstein et al, 1979). A complete inhibition by SOD is the proof that DEPMPO/OH is due to the trapping of superoxide radical. This renders specificity of spin-trapping technique for superoxide detection. In Fig 7, HaCaT membranes and cytosol mixed with NADPH produced signals specific for the DEPMPO/OH radical adduct (Fig 7A) (Frejaville et al, 1995). HaCaT membranes alone produced 70% of superoxide produced by the complete system (Fig 7B). But membranes from GM16 cells did not produce any radical adducts in the same experiment as Fig 7B. DEPMPO/OH adduct formation from HaCaT membranes was completely inhibited by SOD (Fig 7C) indicating that this radical adduct was dependent on the generation of superoxide, and not from the trapping of hydroxyl radical

which may have been produced via Fenton reactions. Omission of membranes resulted in a very low rate of superoxide indicating that the membranes were necessary for the Nox activity (Fig 7D). DPI pre-treatment completely inhibited superoxide generation both in the complete system (Fig 7E) and in reactions with membrane alone (data not shown). In this cell-free system, NADH also produced DEPMPO/OH adduct with slightly weaker signal intensity than NADPH (data not shown). Thus, keratinocyte Nox also functions as an NAD(P)H oxidase. The EPR spin-trapping method is specific for superoxide as a measure of NAD(P)H oxidase activity in a cell-free assay (Fig 7). But to confirm this activity, measurement with a more widely used ferricytochrome c reduction was also performed in order to allow for comparison of HaCaT oxidase activity with those values reported for neutrophils (Clark et al, 1987; Peveri et al, 1992). The rate of SOD-inhibitable superoxide generation (nmol per min) was determined in the presence of 200 mM NADPH. Vmax rates of up to 0.25 nmol per min were observed when cytosol was mixed with membranes, whereas membranes alone produced only about 56% of Vmax (Fig 8). Consistent with our EPR results (Fig 7), the flavoprotein inhibitor DPI abolished superoxide generation. As expected, cytosolic fractions only produced very low oxidase activity (Fig 8). When superoxide rates were corrected for the contribution of membrane fraction (see Materials and Methods), upon stimulation with AA or SDS, specific activities were
9 and
6 nmol per mg per min, respectively (Table II). Km and Vmax values for SDS-activated NADPH-driven superoxide generation for HaCaT membranes were 87 mM and 9.5 nmol per mg per min, respectively. For neutrophil membranes, these values were 45 mM and 200 nmol per mg per min, respectively (Peveri et al, 1992). This indicates that the NAD(P)H oxidase activity of

1004 CHAMULITRAT ET AL

Figure 4 Western blot analysis of Nox1 protein (63 kDa) in membranes of HaCaT, GM16, and positive control HT29 cells. (A) Nox1 protein (indicated at 63 kDa) was detected in membranes from HaCaT and GM16 cells (30 mg protein loaded). There were some weak protein bands due to unspecific reactivity of Nox1 antiserum. (B) The 63 kDa Nox1 protein was expressed in membranes of proliferating cells (HaCaT, sub, at
50% confluence) in higher amounts than those membranes of non-proliferating cells (HaCaT, con, at
85% confluence). The amount of protein loaded was 30 and 20 mg in HaCaT membranes in the left and right pair (HaCaT, con, and HaCaT, sub), respectively. HT29 membranes were loaded at 50 mg protein.

THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Discussion

Figure 5 Specificity of anti-Nox1 antiserum demonstrated by GST-Nox1 fusion protein and the use of antigen peptide. (A) A recombinant GST protein of the cytosolic domain of human Nox1 (290–569 amino acids) was prepared. Nox1 antiserum recognized this recombinant protein, but not GST, as shown by western analysis (upper panel) and the protein staining (lower panel). (B) On the left-hand panel, the antiNox1 antiserum recognized the GST-Nox1 (loaded at 0.5 mg protein, left lane) as well as the 63 kDa Nox1 protein in T84 membranes (loaded at 20 mg protein, right lane). Fifty molar excess amounts of the antigen peptide completely absorbed the immunoreactivity of the Nox1antiserum to the GST-Nox1 (left lane), and significantly decreased the expression of 63 kDa Nox1 protein in T84 membranes (right lane). On the right-hand panel, similar experiments were carried out using HaCaT membranes (loaded at 50 mg protein). The antigen peptide completely abolished the 63 kDa Nox1 protein signal detected in HaCaT membranes.

Our studies provide evidence for the presence of a constitutive NAD(P)H oxidase-type system in human skin (HaCaT) and gingival mucosal (GM16) keratinocytes. We identified the presence of the cytosolic NAD(P)H oxidase protein components Rac1, p67phox, and p40phox. Characterization of the heme by difference spectral analyses showed the presence of an atypical broad b-type cytochrome in HaCaT membranes, which expressed p22phox. HaCaT membranes readily expressed Nox1 protein but little Nox1 mRNA, alternatively, GM16 membranes expressed Nox1 mRNA but little Nox1 protein. This indicates differential expression of Nox1 at the mRNA and protein levels in HaCaT and GM16 cells. Superoxide-generating activity of

the oxidase was demonstrated in HaCaT cell-free system, and was 20 times lower than in the neutrophil system. Km for NADPH of keratinocyte membranes was by a factor of two higher than that of neutrophil membranes (Peveri et al, 1992). These kinetic values bolster the notion that keratinocyte NAD(P)H oxidase generates low levels of superoxide, but in a constitutive manner. We identified cytosolic NAD(P)H oxidase components in keratinocytes. Rac1, but not Rac2, was expressed in both HaCaT and GM16 keratinocytes. Hepatocytes also express Rac1 (Cool et al, 1998). Although Rac2 is primarily present in granulocytes (Didsbury et al, 1989) and vascular smooth

HaCaT membranes is a low-output form with a specific activity 20 times less than that of neutrophil membranes.

122 : 4 APRIL 2004

Figure 6 The reduced-minus-oxidized absorption difference spectra of solubilized neutrophil and HaCaT membranes. Solid dithionite was added prior to measuring difference spectra. Similar to neutrophil membranes, HaCaT membranes showed the Soret g-band (at 426 nm) and a-band (at 558 nm).

muscle cells (Patterson et al, 1999). Rac1 may play a physiological role by inducing vascular endothelial growth factor expression in HaCaT cells by a Nox-dependent mechanism (Sen et al, 2002). Since we could not detect p47phox in Western blot and immunoprecipitation assays, it is possible that p47phox was present at levels lower than detection limit, or was absent in keratinocytes. Interestingly, in a reconstituted system, p47phox is not absolutely required for Nox activity when p67phox and Rac protein are present in sufficient amounts (Koshkin et al, 1997), so that p47phox is required only for optimal superoxide production (Koshkin et al, 1996). HaCaT and GM16 cells lacking p47phox may be able to sufficiently constitute an oxidase-like system capable of producing superoxide. A Nox-like system with p67phox, but lacking p47phox, has been described in human melanocytes of the skin (Brar et al, 2002), and in Caco-2 cells (Yoshida et al, 2002). It has been recently shown that novel p47phox-like (so-called p41nox) and p67phox-like (so-called p51nox) proteins can activate Nox1 to produce superoxide (Banfi et al, 2003; Geiszt et al, 2003; Takeya et al, 2003). This may be the case for Nox1-expressing keratinocytes where p41nox and p51nox (or p67phox) proteins may be required for a stimulus-dependent activation. In contrast to p47phox, p40phox was detectable in keratinocytes. Both, inhibitory and stimulatory functions of p40phox in neutrophils have been suggested for this subunit (De Mendez et al, 1995). To account for low-output activity in keratinocytes, it is possible that p40phox may be an inhibitory oxidase subunit. p22phox, a protein subunit of phox and Nox, which is widely present in many tissues (Cheng et al, 2001) was also present in HaCaT membranes. The detection of cytochrome b-type in HaCaT membranes suggests that p22phox may be associated with gp91phox homolog Nox2, and/or not excluding Nox1. Recent data have shown that p22phox interacts with Nox1 (Takeya et al, 2003). Cytosol did not

HUMAN KERATINOCYTE gp91phox HOMOLOGS

1005

produce superoxide but was required only for the maximal activation from HaCaT membranes treated with NADPH. This effect may be due to complementary p41nox and p51nox proteins, which may be present in HaCaT cytosol. Although GM16 membranes did not show any oxidase activity, HaCaT membranes (which readily expressed abundant Nox1 protein) exhibited oxidase activity 20 times less than neutrophil membranes. This low-output oxidase activity is consistent with constitutive signaling events elicited by ROS, such as, wound healing, and migration of keratinocytes. Nox1 activation may be downstream of receptor ligation resulting in large bursts of ROS that may mediate the killing of pathogens, such as, H. pylori (Kawahara et al, 2001). Accordingly, specific ligation of keratinocyte receptor has been implicated in the cutaneous innate immunity to bacterial infections of the skin (Song et al, 2002). On the mRNA levels, Nox1, Nox2, and Nox4 (but not Nox3) were expressed in HaCaT and GM16 cells. In our hands, protein expression in these cells was studied using currently available antibodies against Nox1 (from Rokutan laboratory), and Nox4 (from H. W. W. Schmidt laboratory). Nox4 protein was, however, absent in HaCaT and GM16 cells (but present in positive control A7r5 cells) (data not shown). Although Nox2 mRNA was expressed in HaCaT and GM16 cells, we still lack Nox2 protein data in keratinocytes. As Nox1 overexpression confers tumorigenic conversion of mouse fibroblasts (Arnold et al, 2001), we directed our focus on characterization of Nox1 in human keratinocytes. This is because keratinocytes are the type of cells where a majority of human tumors is derived from. Our data show differential expression of Nox1 at mRNA and protein levels in HaCaT and GM16 cells (Figs 3 and 4), i.e., with little mRNA but abundant protein in HaCaT cells, and with little protein but abundant mRNA in GM16 cells. This indicates differential rates of Nox1 RNA translation or protein stability when comparing these two cell lines. Translation to Nox1 protein appears to be more efficient in HaCaT cells than in GM16 cells. A protein at 65–67 kDa is predicted for gp91phox homologs (Lambeth et al, 2000), our anti-Nox1 antiserum detected a 63 kDa protein, which was verified to be indeed Nox1 (Fig 5). Nox1 protein was present in membranes of HaCaT, HT29, and T84 cells. All of these cells grow in DMEM/FCS containing 1.8 mM calcium. Nox1 was expressed very little in GM16 cells, which grew in KGM containing 0.05 mM calcium. GM16 cells are regarded to be closer to primary keratinocytes because of their appearance and slow growths in KGM. We recently reported that the cell lines derived from GM16 cells, that were capable of growing in DMEM, did express Nox1 protein (Chamulitrat et al, 2003). The more advanced transformed cell line showed higher Nox1 mRNA and protein levels than the nontransformed cell line. We however found no difference in Nox2 mRNA levels in these cell lines (Chamulitrat W., unpublished data), indicating that Nox2 may not be associated with keratinocyte transformation. It is speculated that increased rates of translation of Nox1 RNA to protein may be associated with the ability of cells to resist calcium-induced differentiation and subsequently to grow in high-calcium medium. It has been recognized that resis-

1006 CHAMULITRAT ET AL

THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 7 Production of superoxide radicals detected by spin-trapping technique in a HaCaT cellfree system. The buffer used was 65 mM sodium phosphate, pH 7.0, containing 2 mM azide, 1 mM EGTA, and 65 mM DTPA. HaCaT membranes (90 mg per mL) and/or cytosol (98 mg per mL) were mixed in the presence of 65 mM DEPMPO. After addition of NADPH (500 mM), the incubation mixture was measured immediately. (A) HaCaT membranes were mixed with HaCaT cytosol in buffer containing DEPMPO and NADPH. (B) The same incubation as in (A) except that the cytosol was omitted. (C) The same incubation as in (A) but with addition of SOD (100 mg per mL). (D) The same incubation as in (A) except that membranes were omitted. (E) The same incubation as in (A) except that DPI (15 mM) was pre-incubated with membranes and cytosol 30 min prior to addition of NADPH. Spectrometer conditions were modulation amplitude, 1.2 G; microwave power, 20 mW; time constant, 20ms; scan rate, 107 G per min; 5 spectra with four-accumulated scans in each spectrum were subsequently added.

tance against differentiation is a character of preneoplastic keratinocytes (Rheinwald and Beckett, 1980). Calcium in the medium can increase telomerase activity (AlfonsoDeMatte et al, 2002), and hence producing immortalized cell lines capable of growing in DMEM. Our findings for Nox1 protein expression in immortalized cells growing in DMEM/FCS, i.e., HaCaT and tumorigenic intestinal T84 and HT29 cell lines, are consistent with these views. In summary, this study demonstrates that human keratinocyte HaCaT and GM16 cells expressed Rac1, p67phox, p40phox, and Nox1 proteins. Furthermore, HaCaT membranes contained p22phox protein, exhibited a heme spectrum similar to flavocytochrome b558, and were capable of producing low-ouput superoxide. Nox1, Nox2, and Nox4 mRNAs were detected, whereas Nox1 but not Nox4 protein was detected in HaCaT and GM16 cells. In human keratinocytes, Nox1 may play a vital role in redox-mediated signaling which is associated with functional changes that may lead to (patho)physiological and diseased states.

Further investigations are warranted to study the mechanisms of Nox1 activation and regulation in human keratinocytes.

Materials and Methods Materials Superoxide dismutase (SOD), catalase, guanosine 50 O-(3-thiotriphosphate) (GTP-gS), NADPH, nicotinamide adenine dinucleotide (reduced form) (NADH), NADP þ, arachidonic acid (AA), Ca2 þ ionophore A23187, PMSF, and dithylenetriaminepentaacetic acid (DTPA) were obtained from Sigma (Deisenhofen, Germany). DPI, pepstatin, and leupeptin were obtained from Calbiochem (Schwalbach, Germany). 5-diethoxyphosphoryl-5methyl-1-pyrroline N-oxide (DEPMPO) produced by Oxis International was obtained from Rolf Greiner Biochemica (Flacht, Germany). Monoclonal antibodies against p47phox (P33720), p67phox (P69820), and Rac1 (R56220) were obtained from BD Transduction laboratories (Heidelberg, Germany), and polyclonal Rac2 (SC-96) and p47phox (SC7660) antibodies were from Santa Cruz Biotechnology (Heidelberg, Germany). Horseradish peroxidase-

122 : 4 APRIL 2004

HUMAN KERATINOCYTE gp91phox HOMOLOGS

1007

Table II. NADPH oxidase activity in a cell-free system of HaCaT keratinocytes

Concentration (mM)

Specific activitya (nmol per mg per min)

Kmb (NADPH) (mM)

Vmax (nmol per mg per min)

AA

40

8.7  1.0 (3)

–

–

AA

100

8.9 (2)

–

–

SDS

100

5.7  0.7 (3)

–

–

SDS

200

6.1  1.3 (3)

87

9.5

Activator

a

Figure 8 NADPH oxidase activity determined as rates of SOD-inhibitable cytochrome c reduction in a HaCaT cell-free system. HaCaT membranes (60 mg per mL), cytosol (80 mg per mL) were mixed with 40 mM AA in 65 mM phosphate buffer, pH 7.0, containing 170 mM sucrose, 2 mM azide, 1 mM EGTA, 20 mM GTP-gS and 80 mM cytochrome c. NADPH (200 mM) was added following a three-minute pre-incubation. Reference incubation was prepared in the presence of SOD (30 mg per mL). The absorbance at 550 nm of the sample and reference was measured in the first 5 min for initial rates. Membranes þ cytosol showed increased activities compared with cytosol or membranes alone (po0.05, memb þ cytosol vs cytosol or membranes alone). DPI inhibited oxidase activity in the memb þ cytosol mixtures (y, po0.05, memb þ cytosol þ DPI vs memb þ cytosol).

conjugated anti-mouse or -rabbit antibodies were obtained from Promega (Mannheim, Germany). Cell cultures HaCaT cell line (gift from N. Fusenig, The German Cancer Research Center, Heidelberg, Germany) was derived from human keratinocytes spontaneously immortalized during long-term culture under selected condition. HaCaT cells were grown in DMEM supplemented with 10% (v/v) calf serum, 100 U per mL penicillin and 100 mg per mL streptomycin. Human gingival mucosal keratinocytes (GM16 cells; gift from P. Tomakidi of University of Heidelberg) were produced by immortalization of human primary gingival keratinocytes with human papillomavirus type 16 E6/E7 (Halbert et al, 1992). GM16 cells were grown in serum-free Clonetics KGM-2 medium supplemented with KGM-2 singlequots from Cell Systems (St Katharinen, Germany). MHH B, HL60, Caco-2, T84, and HT-29 cells were gifts from F. Reuss, V. Lehmann, and W. Franke of the German Cancer Research Center (Heidelberg, Germany). Detection of Nox isoform mRNA For Nox1 mRNA detection, we performed two independent RT-PCR assays. The analysis of Nox1 mRNA was performed by two-step RT-PCR. Gene-specific priming reverse transcription was performed (using the reverse primer of first-round PCRs, see Table I), and cDNA from 750 ng total RNA was used for each PCR reaction. Nox1 primers for the first-round and nested PCRs listed in Table I were designed from GeneBank accession # NM_007052. The nested PCR products appeared at the expected 171 bp (data shown in Fig 3A). For Nox2 and Nox4 mRNA analyses, poly dT-primed cDNA from 500 ng total RNA was used in one-step PCRs using primers listed in Table I. PCR products at 550 and 580 bp were indicative for Nox2 and Nox4

NADPH oxidase activity was measured by SOD-inhibitable cytochrome c reduction. Cell fractionation was performed as described in Materials and Methods. Rates of SOD-inhibitable cytochrome c reduction from membrane þ cytosol were subtracted with those rates from cytosol alone. These corrected rates as shown in the table were then calculated as nmol per min per mg prot. Number in brackets is number of experiments. b Km and Vmax were calculated from a plot of 1/NADPH and 1/V. The corrected rates (V) of SOD-inhibitable cytochrome c reduction in a cellfree system activated by 200 mM SDS were measured at NADPH concentrations between 20 and 1000 mM. NADPH, nicotinamide adenine dinucleotide phosphate (reduced form); AA, arachidonic acid; SDS, sodium dodecyl sulfate; SOD, superoxide dismutase.

mRNA, respectively. Another independent RT-PCR assay was performed using real-time RT-PCRs in the Giessen laboratory where total RNA was isolated using the SV total RNA isolation system (Promega, Mannheim, Germany). mRNAs of Nox1, Nox2, Nox3, and Nox4 were determined with a real-time RT-PCR assay using the fluorescent TaqMan technology as previously described (Wingler et al, 2001). TaqMan assay oligonucleotide primers and probes (Applied Biosystems) are listed in Table I. Cell fractionation 50–100  106 HaCaT, HL60, or HT29 cells were resuspended at 20  106 cells per mL in sucrose buffer (10 mM sodium phosphate buffer, pH 7.4, 140 mM NaCl, 0.34 M sucrose, 2 mM NaN3, 1 mM EGTA, 1 mM PMSF, 50 mg per mL pepstatin A, and 50 mg per mL leupeptin). Cells were subjected to sonic disruption by three 15 s bursts with 30% duty cycle. The sonicates were centrifuged at 10,000  g at 41C for 10 min, and the postnuclear/mitochondria supernatant was separated into a membrane pellet and a cytosol by ultracentrifugation at 100,000  g at 41C for 60 min. Cytosol and membrane fractions were used in a cell-free oxidase activity assay. Membrane fractions were used for immunoblotting of p22phox and Nox1, and for the analysis of heme spectra. In the case for GM16 cells (which grew slowly in low-calcium Clonetics medium with population doubling time of 8 d), membrane fractions were prepared from 10  106 cells sufficient for immunoblotting of Nox1. We did not obtain enough GM16 membranes to perform complete analysis of heme spectra and oxidase activity. Spectral analyses of HaCaT membranes The post-nuclear/ mitochondria supernatant (post 10,000  g spin at 41C for 10 min) was separated into a membrane pellet and a cytosol by ultracentrifugation at 100,000  g at 41C for 60 min. The membrane pellets were then homogenized with 50 mM sodium phosphate buffer (pH 7.4) containing 50 mM NaCl, 1 mM EGTA, 1 mM PMSF, and 20% (v/v) glycerol; and were then extracted with n-heptyl thioglucoside (HTG, Dojindo Laboratories, Kumamoto, Japan) (1% final concentration) at 01C for 30 min. The solubilized membranes were then separated again by ultracentrifugation. Absorption spectra of the solubilized membranes were measured with a Hitachi U-2010 spectrophotometer at 251C. A difference (reduced-minus-oxidized) spectrum was obtained by subtracting

1008 CHAMULITRAT ET AL the spectra of the oxidized form from that of the reduced form, which was obtained by adding a few grains of sodium dithionite. Western blot analyses HaCaT, GM16, or MHH cells were lysed in lysis buffer (50 mM Tris-HCl, 1% Triton X-100, 40 mM b-glyceraldehyde, 100 mM NaCl, 50 mM NaF, 2 mM EDTA, 200 mM sodium orthovanadate, and 0.2 mM PMSF). Thirty microgram protein from cell lysates were solubilized in Laemmli buffer and subjected to sodium dodecyl sulfate (SDS)/PAGE on a 10%–15% acrylamide gel followed by electrotransfer onto a polyvinylidene difluoride membrane (Schleicher & Schuell, Dassel, Germany). After blocking with 5% milk overnight at 41C, the blots were washed and incubated with human antibody specific for p47phox (1:250 dilution), p67phox (1:500 dilution), Rac1 (1:1000 dilution), or Rac2 (1:1000 dilution) for 1 h at room temperature. Following thorough washings, the blots were incubated for 1 h at room temperature with secondary anti-mouse or anti-rabbit antibody (1:7000 dilution). For p22phox detection, 25 mg HaCaT or undifferentiated HL60 membranes prepared as described above were separated in a SDS-Tricine PAGE gel (Schagger and Jagow, 1987), and transferred onto PVDF membrane by electroblotting. The blots were blocked with 5% milk in Tns balanced saline containing 0.1% Triton X-100 (TBST) overnight at 41C with 1:1000 p22phox antiserum (kindly provided by J. Kreuzer of University of Heidelberg) and probed with anti-chicken antibody (1:5000 dilution). For Nox1 detection, 20–50 mg HaCaT, GM16, HT29, or T84 membranes were separated in an SDS/PAGE gel. After blot transfer onto PVDF membrane and blocking, the blot was treated with 1:1000 anti-Nox1 antiserum (obtained from Dr Rokutan laboratory) overnight at 41C. The anti-Nox1 antiserum did not detect gp91phox in human neutrophils and mononuclear cells (data not shown). For immuno-absorption experiment to verify specificity of anti-Nox1 antiserum, 50 molar excess amount of the antigen peptide was incubated with the serum for 20 min at room temperature prior to immunoblotting. Detection of immune complexes was performed using anti-rabbit IgG or mouse-IgG coupled to HRP (1:2000, 1 h, at room temperature). For all Western analyses, the blots were treated with ECL reagent (Amersham, Freiburg, Germany) and exposed to a Biomax ML film. Western blot analyses of each protein were carried out at least three times. For GST fusion protein of Nox1 shown in Fig 5, we isolated a DNA fragment encoding the cytosolic domain of human Nox1 (amino acid 290–569, GeneBank accession # NM_007052) from total RNA from T84 cells by RT-PCR using primers 50 -AAA GAA TTC CGC TCC CAG CAG AAG GTT GTG A-30 and 50 -AAA CTC GAG TCA AAA ATT TTC TTT GTT GAA GTA GAA TTG AAC CT-30 that contain EcoR1 and Xho1 sites (underlined, respectively). The PCR product was subcloned into pGEX-4T (Amersham Pharmacia, Piscataway, NJ), and subjected to DNA sequencing for identifying the correct product. The GST fusion protein was expressed in Escherichia coli JM109 cells, and purified by glutathione–sepharose-4B. Immunoprecipitation of p40phox and p47phox HaCaT or GM16 monolayers grown in 100 mm culture dishes were washed with icecold PBS, and lysed with 0.6 mL immunoprecipitation buffer (1% Triton X-100, 150 mM NaCl, 10 mM Tris-HCl, pH 7.4, 1 mM EDTA, 1 mM EGTA, and 0.5% Nonidet P40, 0.2 mM orthovanadate, 0.2 mM PMSF, 10 mg per mL leupeptin, 10 mg per mL pepstatin, and 50 KU per mL aprotinin). Total cell lysate was obtained following centrifugation at 16,000  g at 41C for 15 min. One milligram protein of the cell lysate was incubated with 2 mg of p40phox antiserum (kindly provided by Marie-Claire Dagher of Laboratory BBSI/DBMS, CEA Grenoble, France) or p47phox antibody for 1 h at 41C. The mixture was added to 30 mL protein A–sepharose beads (Amersham, Freiburg, Germany), and further incubated with rotation at 41C for 1 h. The beads were then centrifuged and washed three times with immunoprecipitation buffer. The pellet was then resuspended in 30 mL SDS sample buffer, and the mixture was boiled for 5 min followed by centrifugation for 5 min at 41C.

THE JOURNAL OF INVESTIGATIVE DERMATOLOGY The supernatant was loaded on a 10% SDS/PAGE. The gel was transferred by electroblotting to PVDF membrane and probed with p40phox antiserum (1:1000 dilution) or p47phox antibody (1:250 dilution) followed by incubation with the secondary antibody. The protein was visualized with ECL detection system and exposure to a Biomax ML film. Immunoprecipitation for each protein was carried out at least three times.

Nox activity in a cell-free system NADPH-dependent superoxide-generating activity in a cell-free assay system was determined by EPR spin-trapping as well as SOD-inhibitable rate of cytochrome c reduction. The assay was carried out in buffer containing 65 mM sodium phosphate, pH 7.0, containing 170 mM sucrose, 2 mM NaN3, 1 mM EGTA, 1 mM MgCl2, 20 mM GTP-gS (Clark et al, 1987; Peveri et al, 1992). For spin trapping, DEPMPO (65 mM) was used to react with superoxide. For an optical assay, cytochrome c (80 mM) was used. Membranes (60–90 mg) and cytosol (80–100 mg) in 1 mL-reaction buffer were mixed with AA (40 or 100 mM), or SDS (100 or 200 mM). For spin-trapping assay, NADPH was added to the mixture that was then pipetted into an EPR aqueous quartz flat cell for an EPR spectral scan. For cytochrome c assay, initial rate of SOD-dependent reduction was obtained by recording an absorbance at 550 nm 3–4 min after NADPH addition. Superoxide concentrations were calculated based on extinction coefficient 21 per mM per cm. Cytochrome c reduction rates from the cytosol, membrane, or membrane þ cytosol were determined as nmol per min. The Nox activity of the composite membrane þ cytosol was determined with a correction by subtracting rate of the cytosol from that of the composite membrane þ cytosol. Nox activity of the membrane was then calculated as nmol per min per mg protein shown in Table II. For Km and Vmax determination, seven NADPH concentrations between 20 and 1000 mM were used in cell-free assay activated by 200 mM SDS.

We thank Dr M.-C. Dagher (Laboratory BBSI/DBMS, CEA Grenoble, France) for p40phox antisera and p40phox positive control, and Dr J. Kreuzer (Internal Medicine, University of Heidelberg) for p22phox antibody. We are grateful to Dr D. Noethe (Chemistry Institute, University of Heidelberg) for the use of an EPR spectrometer. We thank Petra Kronich and Warangkana Chunglok for technical assistance. This study was in part supported by the Deutsche Forschungsgemeinschaft (SFB547/C7) (to Harald H. H. W. Schmidt), Grants-in-Aid Scientific Research from Japan Society for the Promotion of Science (No. 14370184) (to K. Rokutan), and the Dietmar–Hopp Foundation (to W. Stremmel). DOI: 10.1111/j.0022-202X.2004.22410.x Manuscript received June 26, 2003; revised October 15, 2003; accepted for publication October 27, 2003 Address correspondence to: Walee Chamulitrat, Department of Internal Medicine IV, University Hospital Heidelberg, Bergheimer Strasse 58, D-69115, Heidelberg, Germany. Email: Walee_Chamulitrat@ med.uni-heidelberg.de

References Alfonso-De Matte MY, Moses-Soto H, Kruk PA: Calcium-mediated telomerase activity in ovarian epithelial cells. Arch Biochem Biophys 399:239–244, 2002 Arnold RS, Shi J, Murad E, et al: Hydrogen peroxide mediates the cell growth and transformation caused by the mitogenic oxidase Nox1. Proc Natl Acad Sci USA 98:5550–5555, 2001 Babior BM: NADPH oxidase: An update. Blood 93:1464–1476, 1999 Banfi B, Clark RA, Steger K, Krause KH: Two novel proteins activate superoxide generation by the NADPH oxidase NOX1. J Biol Chem 278:3510–3513, 2003

122 : 4 APRIL 2004 Banfi B, Maturana A, Jaconi S, et al: A mammalian H þ channel generated through alternative splicing of the NADPH oxidase homolog NOH-1. Science 287:138–142, 2000 Bouin A-P, Grandvaux N, Vignais PV, Fuchs A: p40phox is phosphorylated on threonine 154 and serine 315 during activation of the phagocyte NADPH oxidase. Implication of a protein C-type kinase in the phosphorylation process. J Biol Chem 273:30097–30103, 1998 Brar SS, Kennedy TP, Sturrock AB, Huecksteadt TP, Quinn MT, Whorton AR, Hoidal JR: An NAD(P)H oxidase regulates growth and transcription in melanoma cells. Am J Physiol 282:C1212–C1224, 2002 Chamulitrat W, Schmidt R, Tomakidi P, Stremmel W, Chunglok W, Kawahara T, Rokutan K: Association of gp91phox homolog Nox1 with anchorage independent growth and MAP kinase-activation of transformed human keratinocytes. Oncogene 22:6045–6053, 2003 Cheng G, Cao Z, Xu X, Van Meir EG, Lambeth JD: Homologs of gp91phox: Cloning and tissue expression of Nox3, Nox4, and Nox5. Gene 269:131– 140, 2001 Chiu C, Maddock DA, Zhang Q, Souza KP, Townsend AR, Wan Y: TGF-b-induced p38 activation is mediated by Rac1-regulated generation of reactive oxygen species in cultured human keratinocytes. Int J Mol Med 8:251– 255, 2001 Clark RA, Leidal KG, Pearson DW, Nauseef WM: NADPH oxidase of human neutrophils. Subcellular localization and characterization of an arachidonate-activatable superoxide-generating system. J Biol Chem 262:4065– 4074, 1987 Cool RH, Merten E, Theiss C, Acker H: Rac 1, and not Rac2, is involved in the regulation of the intracellular hydrogen peroxide level in HepG2 cells. Biochem J 332:5–8, 1998 De Mendez I, Leto TL: Functional reconstitution of the phagocyte NADPH oxidase by transfection of its multicomponents in a heterogenous system. Blood 85:1104–1110, 1995 Didsbury J, Weber RF, Bokoch GM, Evans T, Snyderman R: Rac, a novel rasrelated family of proteins that are botulinum toxin substrates. J Biol Chem 264:16378–16382, 1989 Finkel T: Oxygen radicals and signaling. Curr Opin Cell Biol 10:248–253, 1998 Finkelstein E, Rosen GM, Rauckman EJ, Paxton J: Spin trapping of superoxide. Mol Pharmacol 16:676–685, 1979 Frejaville C, Karoui H, Tuccio B, et al: 5-(Diethoxyphosphoryl)-5-methyl-1pyrroline N-oxide: A new efficient phosphorylated nitrone for the in vitro and in vivo spin trapping of oxygen-centered radicals. J Med Chem 38:258–265, 1995 Geiszt M, Lekstrom K, Witta J, Leto TL: Proteins homologous to p47phox and p67phox support superoxide production by NAD(P)H oxidase 1 in colon epithelial cells. J Biol Chem 278:20006–20012, 2003 Goldman R, Moshonov S, Zor U: Generation of reactive oxygen species in a human keratinocyte cell line: Role of calcium. Arch Biochem Biophys 350:10–18, 1998 Goldman R, Moshonov S, Zor U: Calcium-dependent PAF-stimulated generation of reactive oxygen species in a human keratinocyte cell line. Biochim Biophys Acta 1438:349–358, 1999 Griendling KK, Sorescu D, Ushio-Fukai M: NAD(P)H oxidase: Role in cardiovascular biology and disease. Circ Res 86:494–501, 2000 Gupta A, Rosenberger SF, Bowden GT: Increased ROS levels contribute to elevated transcription factor and MAP kinase activities in malignantly progressed mouse keratinocyte cell lines. Carcinogenesis 20:2063–2073, 1999 Halbert CL, Demers GW, Galloway DA: The E6 and E7 genes of human papillomavirus type 16 have weak immortalizing activity in human epithelial cells. J Virol 66:2125–2134, 1992 Jones RD, Hancock JT, Morice AH: NADPH oxidase: A universal oxygen sensor. Free Radic Biol Med 29:416–424, 2000 Joneson T, Bar-Sagi DA: Rac1 effector site controlling mitogenesis through superoxide production. J Biol Chem 273:17991–17994, 1998 Kawahara T, Kuwano Y, Teshima-Kondo S, Kawai T, Nikawa T, Kishi K, Rokutan K: Toll-like receptor 4 regulates gastric pit cell responses to Helicobacter pylori infection. J Med Invest 48:190–197, 2001

HUMAN KERATINOCYTE gp91phox HOMOLOGS

1009

Kheradmand F, Werner E, Tremble P, Symons M, Werb Z: Role of Rac1 and oxygen radicals in collagenase-1 expression induced by cell shape change. Science 280:898–902, 1998 Koshkin V, Lotan O, Pick E: Electron transfer in the superoxide-generating NADPH oxidase complex reconstituted in vitro. Biochim Biophys Acta 1319:139–146, 1997 Koshkin V, Latan O, Pick E: The cytosolic component p47phox is not a sine qua non participant in the activation of NADPH oxidase but is required for optimal superoxide production. J Biol Chem 271:30326–30329, 1996 Lambeth JD, Cheng G, Arnold RS, Edens WA: Novel homologs of gp91phox. Trends in Bio Sci 25:459–461, 2000 Lee AC, Fenster BE, Ito H, et al: Ras proteins induce senescence by altering the intracellular levels of reactive oxygen species. J Biol Chem 274:7936– 7940, 1999 Meier B, Jesaitis AJ, Emmendorffer A, Roesler J, Quinn MT: The cytochrome b558 molecules involved in the fibroblast and polymorphonuclear leucocyte superoxide-generating NADPH oxidase systems are structurally and genetically distinct. Biochem J 289:481–486, 1993 Nakamura Y, Ginhart T, Winterstein D, Tomita I, Seed JL, Colburn NH: Early superoxide dismutase-sensitive event promotes neoplastic transformation in mouse epidermal JB6 cells. Carcinogenesis 9:203–207, 1988 Patterson C, Ruef J, Madamanchi NR, et al: Stimulation of a vascular smooth muscle cell NAD(P)H oxidase by thrombin. Evidence that p47phox may participate in forming the oxidase in vitro and in vivo. J Biol Chem 274:19814–19822, 1999 Perner A, Andresen L, Pedersen G, Rask-Madsen J: Superoxide production and expression of NAD(P)H oxidases by transformed and primary human colonic epithelial cells. Gut 52:231–236, 2003 Peveri P, Heyworth PG, Curnette JT: Absolute requirement for GTP in activation of human neutrophil NADPH oxidase in a cell-free system: Role of ATP in regenerating GTP. Proc Natl Acad Sci USA 89:2494–2498, 1992 Rheinwald JG, Beckett MA: Defective terminal differentiation in culture as a consistent and selectable character of malignant human keratinocytes. Cell 22:629–632, 1980 Schagger H, von Jagow G: Tricine-sodium dodecyl sulfate-polyacrylamide gel electrophoresis for the separation of proteins in the range from 1 to 100 KDa. Anal Biochem 166:368–379, 1987 Sekharam M, Cunnick JM, Wu J: Involvement of lipoxygenase in lysophosphatidic acid-stimulated hydrogen peroxide release in human HaCaT keratinocytes. Biochem J 346:751–758, 2000 Sen CK, Khanna S, Babior BM, Hunt TK, Ellison EC, Roy S: Oxidant induced vascular endothelial growth factor expression in human keratinocytes and cutaneous wound healing. J Biol Chem 277:33284–33290, 2002 Song PI, Park YM, Abraham T, et al: Human keratinocytes express functional CD14 and toll-like receptor 4. J Invest Dermatol 119:424–432, 2002 Takeya R, Ueno N, Kami K, et al: Novel human homologues of p47phox and p67phox participate in activation of superoxide-producing NADPH oxidase. J. Biol Chem 278:25234–25246, 2003 Teshima S, Rokutan K, Nikawa T, Kishi K: Guinea pig gastric mucosal cells produce abundant superoxide anion through an NADPH oxidase-like system. Gastroenterology 115:1186–1196, 1998 Turner CP, Toye AM, Jones OTG: Keratinocyte superoxide generation. Free Radic Biol Med 24:401–407, 1998 Wingler K, Wunsch S, Kreutz R, Rothermund L, Paul M, Schmidt HH: Upregulation of the vascular NAD(P)H-oxidase isoforms Nox1 and Nox4 by the renin-angiotensin system in vitro and in vivo. Free Radic Biol Med 31:1456–1464, 2001 Yeh LH, Park YJ, Hansalia RJ, et al: Shear-induced tyrosine phosphorylation in endothelial cells requires Rac-1 dependent production of ROS. Am J Physiol 276:C838–C847, 1999 Yoshida L, Nishida S, Shimoyama T, Kawahara T, Rokutan K, Tsunawaki S: Expression of a p67phox homolog in Caco-2 cells giving O2— reconstituting ability to cytochrome b558 together with recombinant p47phox. Biochem Biophys Res Commun 296:1322–1328, 2002