Identification of Retinoid-Modulated Proteins in Squamous Carcinoma Cells Using High-Throughput Immunoblotting Free

Retinoids, a group of natural and synthetic vitamin A analogues, including all-trans-retinoic acid (ATRA), are known to play a major role in regulating growth, differentiation, and apoptosis of many cell types both in vivo and in vitro. Most of the biological activities of ATRA and some synthetic retinoids are thought to be mediated by two classes of nuclear retinoids receptors, retinoic acid receptors (RARα, RARβ, and RARγ) and retinoid X receptors (RXRα, RXRβ, and RXRγ). These receptors form RXR-RAR heterodimers or RXR-RXR homodimers after binding to selective retinoids. These dimers then bind consensus DNA sequence or response elements called RAREs and RXREs and transactivate target genes, thereby altering the growth and differentiation of cells (1). Recently, certain synthetic retinoids have been demonstrated to exert their effects via receptor-independent mechanisms. For example, the synthetic retinoid 6-[3-(1-adamantyl)-4-hydroxyphenyl]-2-naphthalene carboxylic acid (CD437) that has been shown to transactivate RARγ receptors (2) can induce apoptosis in various cancer cell lines, including some that are resistant to ATRA. The apoptosis induced by CD437 in several cancer cells has been found to be independent of retinoid receptors. However, one study suggested a role for RARs in mediating the effect of CD437 in ovarian cancer cells (3). The receptor-independent mechanisms of apoptosis induction attributed to CD437 include an increase in c-Myc levels, an increase in activator protein 1 (AP-1), an increase in cell surface death receptors, and mitochondrial permeability transition (reviewed in Ref. 4).

N-(4-Hydroxyphenyl)retinamide (4HPR), a synthetic retinoid, which was examined in several clinical chemoprevention studies, can transactivate nuclear receptors. 4HPR also induces apoptosis in various cancer cell lines, including those resistant to ATRA. Currently, the mechanism of apoptosis induction by 4HPR is not fully elucidated. It may induce apoptosis by a receptor-independent mechanism (5), although in some cells receptors may be involved (6). Recently, generation of reactive oxygen species (ROS) and modulation of mitochondrial permeability transition and respiration have been found to mediate apoptosis induced by 4HPR (7). Increased ceramide production (8), increased GADD153, and activation of 12-lipoxygenase (9) have been also implicated in receptor-independent mechanisms of 4HPR-induced apoptosis.

Our group is interested in chemoprevention and therapy of head and neck cancers. We have found that retinoids are effective against head and neck squamous cell carcinoma (HNSCC) cells in vitro and in vivo. For example, ATRA suppresses the proliferation of HNSCC cells in monolayer culture, inhibits the formation of colonies in semisolid agar, and decreases the growth of HNSCC multicellular spheroids (10). In addition, ATRA suppresses the squamous differentiation of HNSCC cells (11). Several synthetic retinoids also showed antiproliferative, apoptotic, and differentiation modulatory effects on HNSCC cells (12). At low concentration, CD437 suppressed the expression of squamous differentiation markers and induced apoptosis in HNSCC cells through a retinoid receptor-dependent mechanism (13). 4HPR also induces apoptosis in specific HNSCC cells where ROS production level was high (14). However, in other cells, apoptosis induction by 4HPR partially involved retinoid receptor pathway (6).

Many of the effects of ATRA, CD437, and 4HPR, whether they are mediated by retinoid receptors or by other mechanisms, involve changes in gene expression and posttranscriptional changes (e.g., protein stability; Refs. 15, 16, 17, 18, 19). To improve our understanding of the mechanisms by which retinoids affect HNSCC cells, we tried to identify proteins that are differentially modulated after treatment with different retinoids. For this purpose, we used the Becton-Dickinson PowerBlot Western Array Screening system. This is a high-throughput Western blotting method, which uses carefully formulated mixtures of subsets of ∼800 monoclonal antibodies to evaluate differences in levels of cellular signaling proteins between total cell extracts from different cells or tissues. Treatments with ATRA, CD437, and 4HPR resulted in changes in many proteins that included transcription factors, signaling mediators, DNA synthesis and repair proteins, and tumor suppressor.

The retinoid-modulated proteins identified here include some that have not been previously associated with retinoid response. Our results provide additionally support to the contention that retinoids modulate specific genes and proteins associated with cell growth and apoptosis.

ATRA, CD437, and 4HPR were obtained from Dr. Werner Bollag (F. Hoffmann-La Roche, Basel, Switzerland), Dr. Brahm Shroot (Galderma R+D, Sophia Antipolis, France), and Dr. Ronald Lubet (National Cancer Institute, Bethesda, MD), respectively. Retinoids were dissolved in DMSO at a concentration of 10 mm and stored under nitrogen gas at −80°C. Stock solutions were diluted to the appropriate final concentration with growth medium just before use.

Human HNSCC cell line UMSCC22B was provided by Dr. T. Carey (University of Michigan, Ann Arbor, MI). HNSCC cell line 886 was provided by Dr. Peter Sacks (New York University College of Dentistry, New York, NY), HNSCC SqCC/Y1 was provided by Dr. Michael Reiss (Yale University, New Haven, CT). The cutaneous SCC cell line COLO-16 was provided by Dr. Janet Price (University of Texas, M. D. Anderson Cancer Center, Houston, TX). The non-small cell lung cancer cell lines H460 and A549 were obtained from Dr. Adi Gazdar (University of Texas Southwestern, Dallas, TX). With one exception (COLO-16), the cells were maintained in monolayer culture in a 1:1 (v/v) mixture of DMEM and Ham’s F12 medium. The medium was supplemented with 5% fetal bovine serum and antibiotics (100 units/ml penicillin and 100 μg/ml streptomycin), and the cells were incubated at 37°C in an atmosphere consisting of humidified air with 5% CO₂. The COLO-16 cells were cultured in serum-free keratinocyte growth medium from Life Technologies, Inc. (Grand Island, NY). The cells were seeded in 10-cm diameter dishes at a density of ∼2 × 10⁶ cells/dish. After 24 h, cells were treated with CD437, 4HPR, or ATRA. The control cells were treated with the same volume of DMSO as the retinoid-treated cultures.

For growth inhibition assay, cells were seeded at a density 2 × 10⁵ cells/well in 96-well tissue culture plates. After 24 h, cells were treated with different concentrations of retinoids. Control cultures received the same amount of DMSO as did the treated cultures. Cell numbers were estimated by the Sulforhodamine B assay after 1, 2, and 3 days of treatment, and growth inhibition was calculated as by described previously (20).

Cells were grown in 10-cm diameter tissue culture plates and treated with retinoids for 24 h. Whole cell lysates were prepared as follows: after removing the media by aspiration, cells were rinsed with PBS. Boiling lysis buffer [10 mm Tris (pH 7.4), 1.0 mm sodium orthovanadate, and 1.0% SDS] was added to the plates to denature cellular proteins rapidly. The cell lysates were collected into a 50-ml conical tube and heated briefly in a microwave oven. The lysates were then sonicated for 10 s to shear genomic DNA. The protein concentration was determined with the Bio-Rad Protein assay kit, and the samples were frozen at −80°C and sent on dry ice to BD Biosciences/Transduction Laboratories (Lexington, KY).

The levels of proteins were determined by PowerBlot analysis developed as a custom service by BD Biosciences/PharMingen (San Diego, CA). Details are given at their website.¹

Briefly, samples containing 500 μg of protein in 500 μl of sample buffer were loaded in one big well on a top of a 7.5–15% gradient SDS-polyacrylamide slab gel (16 × 16 × 0.1 cm). The gel was run overnight at a constant current. The proteins were transferred to Immobilon-P nylon membrane (Millipore). After transfer, the membranes were incubated in a blocking buffer containing 5% milk. The membrane was clamped with a grid that isolates 25 lanes (chambers) oriented from the top to the bottom of the membrane. Mixtures containing four to five mouse monoclonal antibodies that recognize proteins of distinct molecular weights, and antigenicity were added in each lane and allowed to hybridize for one h. A total of ∼760 antibodies (identity of the antibodies is available on line) were applied to seven templates/each sample for this analysis.² The blot was removed from the manifold, washed, and hybridized with goat antimouse antibody conjugated with horseradish peroxidase. The blots were developed with chemiluminescence system (SuperSignal West Pico, Pierce, Rockford, IL). Each of the samples was analyzed in triplicate blots.

After exposure to X-ray films, the developed films were scanned, and the “trace” of each band was processed for densitometric analysis. Briefly, trace includes a measure of both band intensity and band area. Standard average is the average of Trace for all bands on a blot or bands from the standards lane. The individual band intensity was expressed as a percentage of the corresponding trace or standards average and designated as “percent lane average” to normalize for different exposure and protein loading between gels. The percent lane average for treated bands was expressed as a percentage of the percent lane average for the corresponding control bands. The “percent of control” was determined to express increases or decreases in protein expression. Changes are expressed as fold increase or decrease between control percent lane average and treated percent lane average. A summary file was provided, which listed all protein expression changes detected, in order of confidence, 1 through 5, with 5 being the highest confidence. The confidence level was based on fold change, reproducibility, and signal intensity. Level 5: changes > 2-fold in triplicate from good quality signals; level 4: changes 1.50–1.99-fold in triplicate from good quality signals; level 3: changes 1.25–1.49-fold in triplicate from good quality signals; level 2: changes > 1.25-fold in triplicate from low signals; and level 1: changes > 2-fold in duplicate from good quality signals. In this article, we only present changes from level ≥ 2.

Cell lysates prepared as described above or using cold lysis buffer [25 mm HEPES (pH 7.7), 400 mm NaCl, 1.5 mm MgCl₂, 2 mm EDTA, 0.5% Triton X-100, 3 mm DTT, 20 mm β-glycerophosphate, 1 mm sodium orthovanadate, and 25 mm paranitrophenylphosphate and protease inhibitor mixture (Roche)] were analyzed by conventional Western blotting. Protein (50 μg) was electrophoresed through a SDS-polyacrylamide slab gel and transferred to polyvinylidene difluoride membrane (Millipore) by electroblotting. The membranes were incubated in blocking buffer [5% nonfat dried milk, 10 mm Tris (pH 7.5), 100 mm NaCl, and 0.1% Tween 20]. Immunoblotting for protein expression was performed using mouse monoclonal antibodies (BD Biosciences/Transduction Laboratories) against 43 proteins identified by the PowerBlot analysis. As secondary antibodies, we used horseradish peroxidase-conjugated goat antimouse (Amersham). The blots were developed using the enhanced chemiluminescence system (Amersham). The Western blotting was repeated at least three times. In ∼80% of the proteins tested, there was an agreement between the findings of independent experiments. For those proteins, which showed different results in two experiments, an additional one was performed, and a conclusion about a change was made based on agreement of at least two of three experiments.

The HNSCC 22B cell line has been chosen for this study because these cells respond well to retinoids both in vitro (12, 13) and in vivo (21). The three retinoids selected for this study differ in their effects on the growth and apoptosis of HNSCC 22B cells (Fig. 1). Growth inhibition by ATRA did not exceed 45% even at 1 μm after a 3-day treatment, whereas CD437 exhibited a nearly linear increase in growth inhibition over the 3 days of treatment reaching >90% at 1 μm. 4HPR was less potent than ATRA when used at 1 μm for the first 2 days but surpassed the effect of ATRA when used at 1 μm for 3 days. At 5 μm 4HPR was quite effective in growth inhibition (Fig. 1).

Previous studies by our group have demonstrated that CD437 and 4HPR induced apoptosis when used at 1 and 5 μm for 24 h, respectively, whereas ATRA failed to induce apoptosis even when used at 10 μm for 24 h (13, 14). Because peak plasma levels of ∼1 μm ATRA have been measured in patients after a single oral dose of 45 mg/m² (22), we decided to use this dose for our studies. CD437 was used at 1 μm because it was an effective proapoptotic dose (13). 4HPR was used at 5 μm because it failed to induce apoptosis at 1 μm, and recent dose escalation studies have shown that plasma levels of ∼5 μm can be achieved in humans with minimal side effects (23).

Extracts of 22B cells grown for 24 h in control medium or in media supplemented with one of the three retinoids (ATRA, 1 μm; CD437, 1 μm; or 4HPR, 5 μm) were subjected to PowerBlot analysis. A representative blot of proteins extracted from untreated cells is shown in Fig. 2,A. Enlarged parts of this control blot compared with corresponding regions from blots derived from retinoid-treated cells demonstrating changes (increases or decreases) in several proteins (circled or enclosed within a rectangle) are presented in Fig. 2, B–I).

The analyses of the antibody array data indicated that all three retinoids induced both increases and decreases in the level of specific proteins. Thus, treatment of the 22B cells with ATRA, CD437, and 4HPR has led to up-regulation of 14, 22, and 22 proteins and down-regulation of 5, 10, and 7 proteins, respectively (Table 1). Included in this table are proteins involved in cell signaling, cell adhesion, cell growth, gene transcription, DNA repair, reactive oxygen generation, apoptosis, and other fundamental cellular processes. The changes induced in protein levels by the three retinoids could be categorized into several types: (a) changes that were similar for the three retinoids (e.g., AF6, MSH3, and Rb2/p130); (b) changes that were induced by two of the three retinoids (e.g., CDC42, p140mDia, SIN, FNK, JAK1, ILK, RAF, PKA_RIIα, MEK5, JUN, ESE-1, phospho-caveolin, procaspase-3, cathepsin L, p43/EMAPII precursor, and AP50); (c) changes that were induced by only one of the retinoids (e.g., ATRA altered HspBP1, HSF4, MEF2D, plakophilin 2a, RIG-G, Pex-1, and mEPHX, whereas 4HPR altered Gβ, FYN, GRB2, JNK1, ERK3, O⁶-methylguanine-DNA methyltransferase (MGMT), Ercc-1, XPA, Na+, K+ ATPase β2, Mena, p67phox, and exportin-1, and CD437 altered rabaptin 5, β-arrestin 2, CSK, IRAK, SH2-B, Smad 4/DPC4, Mre-11, HAX-1, ALDH, Trax, GS27, and tomosyn). The levels of some proteins were modulated uniquely by each retinoid (e.g., SP1 was increased by CD437, not changed by ATRA, and decreased by 4HPR; topoisomerase (topo) IIα was increased by ATRA, not changed by 4HPR, and decreased by CD437).

An analysis of all of the proteins listed in Table 1 by conventional Western blotting (one antibody at a time) was performed using monoclonal antibodies purchased from the same source as the ones used for the PowerBlot array analysis. The conventional analysis confirmed the differential expression of 20 proteins, which are shown in Fig. 3, A and B, and listed in Table 2. These proteins represent 27% of those listed in Table 1.

Some proteins were modulated similarly by the three retinoids. These include ELF3 and Rb2/p130. Some proteins were modulated by two of the retinoids. These include c-Jun (4HPR and CD437), topo IIα (4HPR and CD437), and cathepsin L (ATRA and CD437). Interestingly, topo IIα was decreased in cells treated with 4HPR and CD437 but increased in cells treated with ATRA. Cathepsin L was decreased by ATRA and CD437, yet it was increased by 4HPR. Finally, some proteins were modulated by only one of the retinoids. Proteins in this category included for ATRA: RIG-G, EMAPII, MEF2D, Mena, ALDH, and ERK3; for 4HPR: JAK1, Grb2, MGMT, Ercc-1, and p67phox; and for CD437: Sp1, Sin, tomosyn, and Mre11.

In addition to quantitative changes, some of the retinoids also induced qualitative changes. For example, a slower migrating protein band, possibly representing a phosphorylated form of Jun, was apparent in cells treated with 4HPR and CD437. Likewise, new bands appeared in Mre11 (CD437), Sp1 (CD437), and Rb2 (4HPR; Fig. 3 B).

PowerBlot analysis of changes caused by treatment of the non-small cell lung cancer H460 cells with CD437 revealed increases in the following proteins: topo IIα, IRAK, Mre11, MGMT, Grb2, Ese-1, HSF4, and Pex-1; and decreases in Sin, AF6, and procaspase-3. Some of these changes were similar to those found in the analysis of CD437 effects on HNSCC 22B cells (e.g., IRAK, Mre11, and procaspase-3), and others were different from those found in the 22B cells (e.g., topo IIα, MGMT, Grb2, Ese-1, HSF4, Pex-1, Sin, and AF6).

Conventional Western blotting analysis of extracts from several human cancer cell lines using antibodies against selected proteins that were modulated in HNSCC 22B revealed that the changes induced by CD437 in Mre11 and Sp1 in non-small cell lung cancer H460 and A549 (Fig. 4,A) were similar to those found in HNSCC 22B (Fig. 3,A). Likewise, the ATRA induced increase in the level of RIG-G observed in 22B cells was also noted in H460, A549, and SqCC/Y1 cells (Fig. 4,B). 4HPR increased cathepsin L and JAK1 levels in HNSCC 886 and SqCC/Y1, whereas only JAK1 increased in COLO-16 cells. Similarly, an increase in the level of hypophosphorylated Rb2/p130 was detected in all three cell lines after 4HPR treatment (Fig. 4 C). Thus, some of the effects of these retinoids are not limited to only one cell type.

It is thought that many of the effects of retinoids on cell growth, differentiation, and apoptosis are mediated by changes in gene transcription. Indeed, changes in gene expression have been identified in cancer cells treated with natural and synthetic retinoids using differential display (15, 24, 25, 26, 27, 28) and cDNA array (28, 29, 30) techniques, both of which are focused on mRNA analysis. Changes at the mRNA level may not correspond to the protein level or to its activity, especially when it depends on posttranslational modifications and protein stability. Therefore, it is important to investigate the effects of retinoids on protein levels. Thus far, only a few studies have analyzed the effects of retinoic acid on the modulation of proteins in leukemia and mouse embryonic stem cells using large-scale proteomics (28, 31, 32). We designed this study to compare the effects of retinoids that cause distinct biological effects in a given cell line (i.e., 22B). Specifically, ATRA is cytostatic, whereas CD437 and 4HPR are proapoptotic. This article provides the first description of modulation of proteins by synthetic retinoids in addition to the natural retinoic acid.

We identified changes in protein levels induced by three retinoids ATRA, 4HPR, and CD437 by using the PowerBlot Western Array Screening system. The analysis has demonstrated differences in the levels of a variety of proteins after retinoid exposure (Table 1). These included proteins associated with G protein-mediated signaling, both Ser/Thr and Tyr protein kinases, heat shock response proteins, transcription factors, DNA repair enzymes, and proteins engaged in vesicular transport, apoptosis, cell cycle, oxidative stress, and cell adhesion. Additional studies are required to determine whether some or any of the changes in the levels of proteins are because of transcriptional regulation via retinoid receptors and whether they are the cause or consequence of the biological effects. It is noteworthy that we have analyzed changes that occurred after 24 h of treatment, so some of the changes may be earlier than changes in cell cycle or cell viability.

Proteins showing at least a 1.25-fold difference in expression by the PowerBlot analyses were selected for validation by conventional Western blotting. The differential expression of 27% of these proteins was thus confirmed (Fig. 3 and Table 2). This low concordance rate may be because of the selection of a relatively nonstringent (1.25-fold threshold) or the small number of replicate analyses (not more than 3) of the samples by the PowerBlot method, which may have resulted in inclusion of false-positive or false-negative data. Therefore, we focused the remaining part of the discussion on the group of proteins that was validated by conventional Western blotting.

We identified in the HNSCC 22B cells retinoid-regulated proteins (e.g., Rb2, RIG-G, topo IIα, c-Jun, and ELF3) that have been previously implicated in retinoid effects in other cell types, including leukemia, ovarian cancer, lung cancer, hepatoma, melanoma, and embryonal carcinoma, indicating that some retinoids may have similar effects in different cell contexts (33, 34, 35, 36, 37, 38). More importantly, we also identified novel changes in proteins (e.g., Jak1, p67phox, ERK3, Mre11, and MGMT) not known to be modulated by retinoids. Some of the changes in protein levels were induced by the same retinoids in several other human cancer cell lines, including non-small cell lung cancer, HNSCC, and cutaneous SCCs. Additional studies using retinoid-sensitive and retinoid-resistant cells and manipulation of gene expression using SiRNA or gene transfection are required to determine whether these novel changes are involved causally in the effects of the retinoids on cell growth and apoptosis.

The induction of changes in protein levels by retinoids may result from different mechanisms. First, retinoids may alter gene expression directly through retinoic acid response elements by activating nuclear retinoid receptors that function as ligand-dependent transcription factors (1, 39); second, retinoids may induce or suppress the expression of certain genes indirectly by directly regulating transcription factors (e.g., AP-2, Hox; 40, 41); and third, retinoids may antagonize transcription factors by trans-repression (e.g., AP-1; Ref. 42). In addition to these changes in gene expression, retinoids have also been reported to exert posttranscriptional effects on proteins (16, 17), alter protein glycosylation and stability (43), alter translation of proteins (44), and alter protein degradation rate via modulation of ubiquitination (45).

Some of the changes in protein levels were induced by all three retinoids, suggesting a common mechanism of action. Indeed, all these retinoids can activate retinoid receptor- mediated transcriptional activation of RARE-driven reporter genes and exert similar effects on cell differentiation (2, 13, 39, 45, 46, 47). Changes unique to ATRA may be expected because this retinoid does not induce apoptosis in the 22B cells during the short treatment period used here but can suppress cell proliferation. Changes common to CD437 and 4HPR may be related to induction of retinoid receptor-independent apoptosis because these two retinoids share the ability to induce cell death. In the case of two proteins (topo IIα and cathepsin L), we found that different retinoids may have opposite effects. The reason for these differences is not clear at this time, but we presume that they may be related to different gene expression programs associated with cell growth inhibition compared with differentiation or apoptosis.

We have selected to highlight several of the modulated proteins that may be related to the mechanisms underpinning the effects of retinoids on cell growth, differentiation, and apoptosis.

Several transcription factors were altered by at least one of the retinoids. Both alternative splice products (39 and 41 kDa, respectively) of ELF3 (also known as ESE-1/ESX/jen), which belongs to the Ets transcription factor family (48), were increased by all three retinoids (Table 2). ELF3 up-regulates the expression of several genes associated with terminal squamous cell differentiation (49). It has been shown to act also as a negative regulator of the expression of certain differentiation markers (e.g., keratin 4) in esophageal and cervical cancer cells (50). This function is consistent with the report that retinoic acids suppressed keratin 4 in HNSCC1483 xenografts in vivo (51).

c-Jun, an essential component of all AP-1 complexes, is required for apoptosis induced by various stress stimuli (52). Interestingly, it was increased only by the proapoptotic retinoids CD437 and 4HPR (Table 2). The ability of CD437 and 4HPR to increase c-Jun may be important for the apoptogenic activity (33, 53). The mechanism by which 4HPR increases c-Jun level may be through activation of c-Jun NH₂-terminal kinase (54), which can increase c-Jun’s stability by phosphorylating its serine 63 and 73 residues (55).

The phosphorylated form of Sp1 was increased in cells treated with CD437. Several genes regulated by phosphorylated Sp1 are involved in apoptosis, including Bcl-2 family members and death ligands (56). Interestingly, ELF3 and Sp1 cooperate in the regulation of the transcription of certain genes. Although Sp1 was found to be cleaved by caspases during CD437-induced apoptosis of T cells (57), we did not detect such an effect in CD437-treated HNSCC cells.

ATRA decreased the level of MEF2D, a member of the myocyte enhancer factor 2 family of transcription factors, which is activated after growth factor stimulation (e.g., EGF; 58). Down-regulation of MEF2D may contribute to ATRA’s growth inhibitory effects in HNSCC cells.

All three retinoids used in this study increased the hypophosphorylated form of Rb2/p130. Rb2/p130, a member of the Rb family (59), acts as a negative regulator of cell cycle progression in its hypophosphorylated form and as a suppressor of tumorigenicity of cancer cells in vitro and in vivo (60). Zhang et al. (34) showed an increase in Rb2 hypophosphorylation in ovarian cancer cells after ATRA treatment. Therefore, Rb2 appears to participate in growth inhibition in different cell context and stimuli and may be involved causally in the inhibitory effects of the three retinoids on our HNSCC cells.

We observed an increase in topo IIα level after ATRA treatment but down-regulation by CD437 and 4HPR. Topo II is a nuclear enzyme that decatenates DNA through an ATP-dependent double-strand break in DNA replication, transcription, chromosome segregation, cell cycle progression, and DNA repair (61). In HL-60 cells, the level and phosphorylation of topo IIα was increased during ATRA-induced differentiation, and increased protein level was due to slower degradation (37). The down-regulation of topo IIα by CD437 may be attributed to changes in p53 because CD437 increased p53 protein in the 22B cells (data not shown), and topo IIα is a target for p53-mediated transcriptional repression (62).

We found the down-regulation of cathepsin L, an abundant lysosomal cystein protease (63) in CD437- and ATRA-treated cells. In contrast, it was up-regulated in 4HPR-treated cells. The expression of cathepsin L is elevated in a variety of human tumors, including head and neck cancer (64). Recent studies have demonstrated that inhibition of the cathepsin L overexpression reduced tumor growth and invasion (65). Therefore, down-regulation of cathepsin L may contribute to the growth inhibitory effect by ATRA and CD437. 4HPR increased cathepsin L, and this may be related to its mechanism of apoptosis induction because recent studies have demonstrated that cathepsin L can be involved in apoptosis after released from lysosomes to cytosol through direct activation of caspases (66).

p67 was up-regulated in cells treated with 4HPR. This novel observation may eventually lead to clues on the mechanism by which 4HPR induces reactive oxygen species because p67phox is a subunit of NADPH oxidase, a multicomponent electron-transfer complex (67). The NADPH oxidase catalyzes the reduction of molecular oxygen to superoxide anion in acute inflammatory reaction. Recent evidence suggests that NADPH oxidase participates in apoptosis as a major source of ROS in different cell systems (68). p67 is regulated by AP-1, therefore, the ability of 4HPR to increase p67 may be related to its ability to increase c-Jun.

MGMT was increased in cells treated with 4HPR. MGMT, a ubiquitous DNA repair protein, is responsible for removal of O⁶-alkylguanine from DNA (69). The potential function of cis-elements on the MGMT promoter (e.g., AP-1 sites) in activation of MGMT has been reported previously (70). As mentioned above, the level of c-Jun was increased in 4HPR-treated cells, and it may play a role in the regulation of MGMT expression.

Mre11 is known to function in DNA maintenance and repair in S-phase DNA-damage response (71). Although the level of Mre11 decreased in CD437-treated cells, a slowly migrating protein band, presumed to be a phosphorylated form, has appeared in the treated cells. It has been reported that Mre11 was phosphorylated upon exposure of cells to various DNA damage-inducing agents, including γ-radiation, UV radiation, Adriamycin, and cis-platinum (72). In this context, it is noteworthy that CD437 has been shown recently to induce DNA damage, leading to apoptosis in melanoma cells (73). Therefore, modification of Mre11 in CD437-treated cells may represent a response to DNA damage induced by CD437.

The level of Ercc1 was increased in 4HPR-treated cells. Ercc1 is one of the essential proteins in nucleotide excision repair pathway (74). It was indicated that the level of Ercc-1 expression after DNA damage and the activity of excision repair are strongly associated (75). 12-O-tetradecanoylphorbol-13-acetate, a known AP-1 activator, was also shown to induce Ercc-1 expression (76). Thus, the up-regulation of Ercc-1 may be caused by increase of AP-1 activity led by increase in Jun level after treatment with 4HPR.

4HPR treatment up-regulated Jak1 a cytoplasmic tyrosine kinase that mediates cytokine/IFN and growth hormone signaling pathways (77). Although the mechanism of this up-regulation is not known, it may be the consequence of 4HPR-induced ROS because Jak/Stat pathway can be activated by ROS (78). Jak1 may act as a second messenger to mediate 4HPR-induced apoptosis in these cells if its up-regulation also involves an increased activity by analogy with the effects of IFNs via the Jak/STAT pathway (79, 80).

In conclusion, the large-scale screening of proteins modulated after treatment of 22B cells with different retinoids has enabled us to identify proteins that are implicated in the effect of retinoids in HNSCC cells. Among these proteins, there were some that have not been previously associated with retinoid effects. Additional studies are required to determine whether there is a causal relationship between the changes in the levels of these proteins and growth inhibition or apoptosis induction potential of the retinoids studied.

We thank Dr. Margaret Spitz for the support of this work through the Early Detection Research Network (EDRN) program.