|Home | About | Journals | Submit | Contact Us | Français|
Identification of the new molecular components of the DNA damage signaling cascade opens novel avenues to enhance efficacy of chemotherapeutic drugs. High mobility group protein 1 (HMGB1) is a DNA damage sensor responsive to incorporation of non-natural nucleosides into DNA; several nuclear and cytosolic proteins are functionally integrated with HMGB1 in the context of DNA damage response. The functional role of HMGB1 and HMGB1-associated proteins (high mobility group protein B2, HMGB2; glyceraldehyde-3-phosphate dehydrogenase, GAPDH; protein disulfide isomerase family A member 3, PDIA3; heat shock 70kDa protein 8, HSPA8) in DNA damage response was assessed in human carcinoma cells A549 and UO31 by transient knockdown with short interfering RNAs. Using the cell proliferation assay, we found that knockdown of HMGB1-associated proteins resulted in 8-50-fold decreased chemosensitivity of A549 cells to cytarabine. Western blot analysis and immunofluorescent microscopy were used to evaluate genotoxic stress markers in knocked down cancer cells after 24-72 hr incubation with 1 μM cytarabine. Our results dissect the roles of HMGB1-associated proteins in DNA damage response: HMGB1 and HMGB2 facilitate p53 phosphorylation after exposure to genotoxic stress; and PDIA3 has been found essential for H2AX phosphorylation (no γ-H2AX was accumulated after 24-72 hr incubation with 1 μM cytarabine in PDIA3 knockdown cells). We conclude that phosphorylation of p53, and phosphorylation of H2AX, occur in two distinct branches of the DNA damage response. These findings identify new molecular components of the DNA damage signaling cascade and provide novel promising targets for chemotherapeutic intervention.
Traditional chemotherapeutic agents targeted against DNA remain the core of anticancer therapy, and several million people worldwide receive the conventional chemotherapeutics yearly(1). The efficacy of these drugs is limited partly because of cellular mechanisms that diminish DNA damage by executing cell cycle arrest and DNA repair(2).
High mobility group protein B1 (HMGB1) is an architectural transcription factor, and a component of the early DNA damage sensor responsive to incorporation of non-natural nucleosides into DNA that occurs without compromising DNA integrity(3-5). HMGB1 binds flexible DNA (prone to acquire bent or kinked conformation), rather than DNA with single- or double-strand breaks, and recruits other proteins to the DNA-HMGB1 complex(3). In murine embryonic fibroblasts (MEFs), knockout of HMGB1 decreased sensitivity to antimetabolites 5-10- fold, and inhibited activation of p53-mediated response to genotoxic stress(5).
Initial identification of HMGB1 as a DNA damage-sensing protein revealed a group of proteins (high mobility group protein B2, HMGB2; glyceraldehyde-3-phosphate dehydrogenase, GAPDH; protein disulfide isomerase family A, member 3, PDIA3; heat shock 70kDa protein 8, HSPA8) which were physically associated with HMGB1(4). While HMGB2 is a structural analog of HMGB1 with amino acid sequence more than 85% identical to that of HMGB1, knockout of this protein in mice confers a phenotype quite distinct from that of Hmgb1-knockout mice(6). The role of HMGB2 in DNA damage stress response remained uncharacterized. Cytosolic proteins GAPDH and PDIA3 have been recently recognized to have intranuclear functions, and now are the focus of intensive studies(7;8). Intranuclear localization and molecular partners of PDIA3 (also known as ERp60, ERp57 or GRP58) suggests its participation in DNA repair processes(8;9). Finally, HSPA8 (also known as HSC70) is a molecular chaperone involved in intranuclear translocation of cytosolic proteins. In response to different types of stress, including heat shock and oxidative stress, HSPA8 is accumulated in the nucleolus though its functions still remain to be elucidated(10;11).
Recently, novel inhibitors of the cellular mechanisms responsive to DNA damage have been developed and entered the clinical trials(12;13). Within this approach, the cellular components involved in activation of chemotherapy-induced DNA damage response provide important targets, because inhibition of checkpoints that regulate cell division and limit DNA damage can potentiate the efficiency of chemotherapy. Especially this is true in respect with antimetabolite drugs known to incorporate into DNA without gross changes of the DNA structure(14;15).
The primary focus of our study is to test the hypothesis that HMGB1 and associated proteins are involved in cellular response to antimetabolite drugs. Multiple publications demonstrated that antimetabolite treatment induced p53-mediated apoptotic death of cells via genotoxic stress, though the beginning of this pathway remains obscure. In the present study, we set out to characterize the functional role of HMGB1-associated proteins in the early steps of DNA damage response to chemotherapeutic agents, at the cellular level. To this end, we used a model system based on human carcinoma cells where individual proteins including HMGB1, HMGB2, GAPDH, PDIA3, and HSPA8 were knocked down by short interfering RNA treatment. Our studies demonstrated that the homologous proteins HMGB1 and HMGB2 participate in DNA damage response by modulating p53 phosphorylation. We also found that, after chemotherapy-induced stress, the cytosolic proteins PDIA3 and HSPA8 accumulate in the nucleus; importantly, our data indicate that intranuclear PDIA3 modulates phosphorylation of H2AX histone.
Therefore, for the first time we demonstrated the functional role of HMGB1-associated proteins HMGB1, HMGB2 and PDIA3 in chemotherapy-induced DNA damage stress response in human cancer cells. Together, these findings identify molecular components of the DNA damage signaling cascade and provide novel promising targets for chemotherapeutic intervention.
Lung carcinoma cells A549 were obtained from the ATCC collection (ATCC, Manassas, VA), and renal carcinoma UO31 cells were obtained from the Tumor Cell Line Repository, NCI-Frederick. Cells were treated with drugs dissolved in DMSO (fluorouracil, FU), 0.1 N NaOH (mercaptopurine, MP), or water (cytarabine, araC) as 500-1000X stock solutions; drug concentrations were determined spectrophotometrically (FU, ε265=7,010; MP, ε320=19,600; araC, ε272=9,259)(16). [5-3H]-cytosine-β-D-arabinofuranoside (14.9 Ci/mmol) (Moravek Biochemicals, CA) was used in DNA incorporation experiments. The IC50 values were calculated by fitting a sigmoid Emax model to the cell viability versus drug concentration data, determined in triplicate from three independent experiments.
Cell viability and cell count were determined by flow cytometry using ViaCount reagent with Guava Personal Cell Analyzer (Guava Technologies, CA). Clonogenic assay was performed after cell treatment for 3 days, cell growth for 10 days, and staining with methylene blue(17). For the MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium) assay (CellTiter 96 cell proliferation kit, Promega, WI), A549 and UO31 cells (250 cells per well) were plated into 96-well plates, and cultured for 3-5 days in varying concentrations of the following drugs: 0-100 μM MP, 0-50 μM araC, and 0-100 μM FU.
RNAi experiments were performed using the pre-designed Stealth RNA (Invitrogene, Carlsbad, CA) (HMGB1 - HSS142453, HSS142454, HSS142455; HMGB2 - HSS104854, HSS104855, HSS104856; GAPDH Validated Stealth RNAi DuoPak duplexes 1 and 2; PDIA3 - HSS142315, HSS142316, HSS142317; and HSPA8 - HSS105082, HSS105083, HSS105084). Effective siRNA were selected using a lac-Z reporter system (BLOCK-iT RNAi Target Screening System, Invitrogen). Scrambled Negative Stealth RNAi control (Invitrogen) was used as negative control in all siRNA experiments.
Total cellular RNA was extracted with TriReagent (GIBCO BRL/Invitrogen, Carlsbad, CA) from A549 and UO31 cells (about 5×106 cells per experiment, 3 replicates), reverse transcribed using the TaqMan Reverse Transcription kit (Applied Biosystems, CA) according to manufacturer’s instructions. The level of mRNA was evaluated using Relative Quantification protocol with human β-actin as a normalization standard on ABI 7300 Real Time PCR instrument (Applied Biosystems, CA) according to manufacturer’s instructions. Data were collected from 3 independent experiments for each sample.
Western analysis was performed as described earlier (Krynetski et al., 2003). The subcellular fractionation into cytosolic and nuclear fractions was performed using NE-PER Extraction reagent (Pierce Biotechnology, Rockford, IL) according to manufacturer’s instructions. The protein concentration was determined in cellular extracts using PlusOne2D Quant kit (Amersham Biosciences, NJ). Electrophoretic separation was performed using 16% PAGE gels for analysis of γ-H2AX, HMGB1, and HMGB2; 12% PAGE gels for analysis of phosphorylated p53; and gradient 4-12% PAGE gels for analysis of PARP, PDIA3, and HSPA8 (PageGel, San Diego, CA). Membranes were developed with rabbit polyclonal antibodies specific to HMGB1 at 1:1000 dilution and HMGB2 at 1:500 dilution (Abcam, Cambridge, MA); rat anti-HSPA8 monoclonal Ab at 1:5000 dilution (Stressgen, Ann Arbor, MI); rabbit anti-GAPDH polyclonal Ab at 1:10000 dilution (Santa Cruz, Santa Cruz, CA); rabbit anti-PDIA3 polyclonal Ab (Rockland, Gilbertsville, PA) were generated as described earlier and used at 1:5000 dilution(4); rabbit anti-Ser15-phosphorylated p53 polyclonal Ab at 1:1000 dilution and rabbit anti-γ-H2AX (H2AX phosphorylated at Ser139) polyclonal Ab at 1:500 dilution (Calbiochem, La Jolla, CA); rabbit anti-PARP polyclonal Ab at 1:1000 dilution (Cell Signaling Technology, Danvers, MA); rabbit anti-85 kDa PARP fragment polyclonal Ab at 1:500 dilution (Abcam, Cambridge, MA) and mouse anti-β-Actin monoclonal Ab (loading control) at dilution 1:10000 (Sigma, St. Louis, MO). Bands were visualized with secondary antibody - IRDye680 donkey anti-mouse antibody and IRDye680 goat anti-rabbit antibody; or IRDye 800CW donkey anti-rabbit antibody (LI-COR Biosciences, Lincoln, NE) and IRDye 800CW goat anti-rat antibody (Rockland, Gilbertsville, PA) at 1:10000 dilution and quantified by Odyssey Infrared Imaging system (LI-COR Biosciences, Lincoln, NE) using two-color fluorescence detection at 700 and 800 nm.
Cells were grown on BD BioCoat poly-L-Lysine precoated glass coverslips (Thermo Fisher Scientific, MA) in 12-well plates at density 50000 cells/well, treated with 10 uM araC for 48 hr and fixed in 4% formaldehyde in PBS for 15 min. Cells were then washed in ice-cold methanol, blocked in 5% normal rabbit serum, and labeled with anti-γ-H2AX antibody (phospho-histone H2A.X rabbit mAb conjugated with Alexa Fluor 488, Cell Signaling Technology, MA). The slides were incubated with anti-phospho-histone H2A.X antibody (1:10 dilution) overnight at 4°C. Nuclei were counterstained with 4’, 6’-diamidino-2-phenylindole (DAPI) and mounted with Vectashield hard set mounting medium (Vector Laboratories, CA). Fluorescence images were recorded with a Nikon Eclipse 50 fluorescent microscope and analyzed by ImageJ 1.37v software (NIH, USA). All experiments were repeated at least 4 times. At least 100 cells were analyzed in each experiment.
A549 and UO31 cells were grown in 96-well plates in Ham’s F12K or RPMI1640 medium and treated with 10 μM FU, 10 μM MP, or 0.5 μM araC for 24-48 hr. Actinomycin D (0.5 μg/ml) was used as a positive control. Caspase activity was assessed using fluorogenic substrates for caspases 3, and 7 using Apo-ONE Homogenous Caspase 3/7 Assay, per manufacturer’s instructions (Promega, Madison, WI). Caspase activity was normalized per mg of total protein, and compared with activity in non-treated cells. Each experiment was performed in triplicate and repeated at least three times.
The statistical analyses were carried out using Student’s t test with Statistica software program (StatSoft, OK), and non-linear regression analysis with GraphPad Prizm 4.0 software (GraphPad software, CA). A P value <0.05 was considered statistically significant. Data are presented as the mean ± SE.
Two human carcinoma cell lines (lung carcinoma A549 and renal carcinoma UO31) proficient in p53-dependent DNA damage response pathway were selected as a model system for the analysis of cellular response to antimetabolite drugs(18). Both cell lines were easily transfected with short interfering RNA, and revealed effective knockdown of targeted proteins within 24-48 hr (see below). To assess chemosensitivity of the model cell lines, we used clonogenic and MTT assays. While clonogenic potential of A549 cells was readily measured after drug treatment, UO31 cells did not form colonies and were therefore excluded from clonogenic assays. Clonogenic survival of A549 was decreased following treatment with 0-50 μM araC, and 0-100 μM FU but not after 0-100 μM MP treatment; araC was the most potent cytotoxic agent for A549 cells (Fig. 1A and Table S1).
Further, sensitivity of A549 and UO31 to antimetabolites was evaluated by the MTT assay indicative both of cytotoxic and cytostatic effects(19). After 24-72 hr treatment with 0-100 μM FU, A549 and UO31 cells demonstrated decreased MTT reducing activity; both cells were most sensitive to araC treatment (Fig. 1C and Table S1). Similar to the results from clonogenic assay, A549 was found to be resistant to MP treatment in the range 1-100 μM (data not shown); UO31 was sensitive to this drug (IC50=2.3±0.55 μM) indicating that metabolic activation of MP in UO31 cells is functional.
Activation of apoptosis in the A549 and UO31 cell lines was tested after treatment with three antimetabolite drugs (MP, FU, and araC) at concentrations five to ten times higher than IC50. After treatment with 10-50 μM FU we did not observe caspase activation or PARP proteolysis neither in A549 nor UO31 cells (Fig. 1B, D), though FU treatment had a growth-inhibiting effect both on A549 and UO31 (results not shown). 10μM MP was a weak inducer of apoptosis in UO31 (Fig. 1D). Treatment with araC at concentrations 5-10 times above IC50 values induced apoptotic death in both cell lines after 48 hr incubation, as revealed by activation of apoptotic markers (proteolysis of caspase 3 substrate PARP (Fig. 1B), and 2-4-fold induction of caspases 3/7 activity) (Fig. 1D). Because incorporation in DNA is a major mechanism of araC cytotoxicity, we verified ability of both cell lines to use araC as a DNA precursor. Incorporation of araC into DNA was evaluated using [5-3H]-araC; these experiments confirmed that araC was incorporated into DNA of both cell lines at comparable level (0.1-0.5%) and was detectable already after 24 hr incubation (Fig. S1). Therefore, in further experiments we used araC treatment to assess cytotoxic effects in A549 and UO31 cells with knockdown proteins.
Western blot analysis demonstrated that both cell lines were proficient in expressing HMGB1, HMGB2, GAPDH, PDIA3, and HSPA8 proteins (Fig. 2A). To assess the role of HMGB1-associated proteins in genotoxic stress response, we established a model system based on short interfering RNA technology. This transient knockdown system made it possible to rapidly test the functional role of five HMGB1-associated proteins in response to genotoxic stress caused by nucleoside analogs. Selection of efficient siRNA targeted against mRNAs coding for HMGB1-associated proteins was performed in A549 cell line using a siRNA screening system; the efficiency of siRNA-mediated mRNA knockdown estimated by Real-Time PCR was typically greater than 90%. The changes in protein level were monitored by Western blot analysis. siRNA that reduced the protein level more than 70% were selected for further analysis. The following siRNA were chosen for knockdown: HMGB1 - HSS142453; HMGB2 - HSS104854; GAPDH - validated Stealth RNAi DuoPak duplex 1; PDIA3 - HSS142316, and HSPA8 - HSS105082. Knockdown effect was evident during the entire experiment (24-72 hr; see Fig. S2). Interestingly, siRNA knockdown was consistently more effective in A549 cells (10-25% of residual protein) compared with UO31 cells (25-40% of residual protein), as evidenced by the lower levels of residual proteins (Fig. 2A, B). Knockdown of HMGB1/2, PDIA, and HSPA8 proteins in A549 and UO31 cells did not significantly change the rate of cell growth compared to negative control (p>0.05), while cell growth arrest was observed in cells with knocked down GAPDH (p<0.004, Table S2).
According to our hypothesis, HMGB1 and associated proteins act at the beginning of the genotoxic stress pathway. We used MTT assay to test chemosensitivity of cancer cell lines following knockdown of nuclear proteins associated with HMGB1, to assess short-term effects generated by the DNA damage sensor (Fig. 2C, and Table 1). In these experiments, a five orders of magnitude range of concentrations of araC was used to evaluate its inhibitory effect on cell growth and viability. Importantly, knockdown of HMGB1 and HMGB2 resulted in 8-13-fold increase of IC50 in A549 cells treated with araC, and 3-8-fold increase in UO31. Similarly to effects of HMGB1/2, knockdown of PDIA3 and HSPA8 in A549 resulted in 8-10-fold increase of IC50 (Table 1). In contrast, PDIA3 and HSPA8 knockdown in UO31 did not alter cell viability (Table 1). While the levels of these proteins were reduced to 10-20% in A549 cells, in UO31 the residual levels were about 30-40% (Fig. 2B), providing an explanation to a smaller difference in cytotoxic effects. The strongest effect was achieved by knockdown of GAPDH (>50-fold increase of IC50, see Fig. S3). Because GAPDH knockdown causes cell growth arrest (Table S2), we excluded siGAPDH treatment from further cytotoxicity experiments. Therefore, in experiments on molecular markers of cytotoxicity we used A549 and UO31 cells with knockdown proteins HMGB1, HMGB2, PDIA3, and HSPA8 after treatment with araC.
The levels of phosphorylated S15-p53 (the marker of p53 activation), γ-H2AX (the marker of DSB formation), and accumulation of 85kDa PARP fragment (the marker of caspase 3 activation) were monitored to assess the effects of knockdown of HMGB1-associated proteins. After transfection with siRNA targeted against HMGB1, HMGB2, PDIA3, and HSPA8, both A549 and UO31 cells were treated with 1 μM araC for 24-72 hr, and cell lysates were analyzed by Western (Fig. 3). Accumulation of Ser15-phosphorylated form of p53 was notable already after 24 hr incubation in control cells (transfected with scrambled siRNA), but it was delayed till 48-72 hr in cells with knockdown HMGB1 and HMGB2, corroborating the functional role of these proteins in cellular response. Knockdown of HMGB1/2, which delayed accumulation of the S15-p53 level, also slightly affected formation of γ-H2AX (Fig. 3). Importantly, accumulation of Ser15-phosphorylated p53 (Figs. 3A, C) paralleled accumulation of 85 kDa PARP fragment indicative of caspase 3 activation (Fig.4A, B). The most pronounced drop in accumulation of 85 kDa PARP fragment was observed in the cells with reduced HSPA8 protein, the finding still awaiting explanation (Figs. 4A, B).
Western blot analysis of cytosolic and nuclear extracts isolated from untreated A549 cells demonstrated that PDIA3 and HSPA8 reside in cytosol. These proteins relocate into the nuclei after araC treatment (Figs. 5, and S4). Fig. 5B shows accumulation of PDIA3 in the nucleus of cells following incubation with 1 μM araC for 24-72 hr. In A549/siPDIA3 knockdown cells, the total level of PDIA3 was reduced by 90% (Fig. 2B), and its accumulation in the nucleus of siPDIA3-treated A549 cells in response to drug treatment was notable after 72 hr incubation (Fig. 5B).
Interestingly, knockdown of PDIA3 protein in A549 cells inhibited accumulation of γ-H2AX following araC treatment (Fig. 3B, D). Under the same conditions, phosphorylation of Ser15 in p53 was observed already after 24 hr incubation, indicating that phosphorylation of p53 and H2AX are independent events (Fig. 3). To corroborate these data, we evaluated formation of γ-H2AX foci in A549/siPDIA3 cells treated with araC by immunofluorescent microscopy (Fig. 5A). Cells treated with hydroxyurea were used as positive control (Fig. S5)(20). Significantly lower number of cells treated with araC was stained with anti-γ-H2AX antibody when PDIA3 protein was knocked down (Fig. 5A), consistent with the results of Western blot analysis (Figs. 3B, D). In parallel to inhibition of H2AX phosphorylation, accumulation of 85 kDa PARP fragment in the cells with knockdown PDIA3 protein occurred with delay compared with control cells (Fig. 4A, B). The same effects were observed in UO31 cells after knockdown of HMGB1-associated proteins (data not shown).
The role of HMGB1 in recognizing aberrant or damaged DNA has been shown in multiple in vitro experiments. A recent study directly demonstrated accumulation of HMGB1 at sites of oxidative DNA damage in live cells thus defining HMGB1 as a component of an early DNA damage response(21). Reduced histone acetylation after DNA damage in HmgB1-deficient cells indicates a role of HMGB1 in DNA damage-induced chromatin remodeling(22). In the present study, we established a model system based on human carcinoma cells A549 and UO31 where two groups of proteins - nuclear proteins (HMGB1/2), and proteins transiently present in the nucleus after the drug stress (GAPDH, PDIA3, and HSPA8) were knocked down with short interfering RNA (siRNA). Because araC was the strongest inducer of apoptosis in both cell lines, we used it to challenge cells following siRNA treatment. Cytotoxic effect of araC is due to incorporation into DNA and is mediated by p53(12;23). The araC concentration for cell treatment (1 μM araC) was selected to maximize the effect of the drug at the relatively short exposure period (24-72 hr). This concentration lies within the range of araC plasma concentrations achieved in some protocols for treatment of hematopoietic malignancies(24).
siRNA treatment effectively decreased the level of proteins in both cell lines, though the residual levels of four proteins (HMGB1, GAPDH, PDIA3, and HSPA8) were consistently higher in UO31 cells compared with A549 cells (Fig. 2). The rate of proliferation of knockdown cells did not change significantly, with the exception of cells treated with siGAPDH where this treatment arrested cell growth. This was not unexpected, as GAPDH is involved in cell cycle regulation via interaction with cyclin B(25), and/or S phase-inducible H2B transcription activator OCA-S(26). Depletion of GAPDH in A549 and UO31 cell lines dramatically altered cell chemosensitivity (Table 1 and Fig. S3). Because antimetabolite drugs exert their cytotoxic effects during the S phase of the cell cycle, the unusually strong effect of GAPDH knockdown on cell viability is probably the consequence of cell cycle arrest, a hypothesis under investigation in our lab.
Abrogation of HMGB1 and HMGB2 increased chemoresistance both in A549 and UO31 cell lines (Table 1 and Fig. 2C). Importantly, the effect of HMGB2 knockdown was similar to that of HMGB1 knockdown, despite the fact that HMGB2 is a much less abundant protein than HMGB1 (Fig. 2A). The results received after transient knockdown of HMGB1 in human cancer cells are in line with our data obtained in Hmgb1-knockout MEFs(5). The present study supports the role of chromatin-associated proteins HMGB1/2 in early steps of drug-induced activation of p53 response by demonstrating delay or abrogation of the stress marker (phosphorylated form of p53) in HMGB1/2-depleted cells (Fig. 3A, C). Our results indicate that HMGB1 (and probably HMGB2) is a sensor of DNA damage which induces p53-mediated DNA damage response. In the cells with depleted HMGB1, manifestation of apoptosis markers (e.g., caspase-mediated PARP proteolysis) is also delayed (Fig. 4). Active scanning for flexible points within DNA as a mechanism of DNA damage detection provides an attractive explanation of how HMGB1 contributes to several seemingly unrelated DNA repair pathways, including mismatch repair (MMR), base excision repair (BER), and nucleotide excision repair (NER)(21;22;27). Of note, DNA substrates of MMR and BER reveal increased flexibility(28-30). While stimulation of DNA repair by HMGB1 is demonstrated reasonably well, the consequences of HMGB1 depletion for the cells are less clear: no change, decreased, and increased sensitivity to the a panel of DNA damaging agents was reported in HMGB1 KO MEFs(4;5;21;22;31). Most interestingly, HMGB1 physically and functionally interacts with p53 indicating a link with p53-dependent signaling network(32;33). The intracellular level of HMGB1 is controlled by developmental and tissue-specific factors, and variations in HMGB1 level may explain different sensitivity to chemotherapy in certain cell populations(34). In line with this notion, analysis of gene expression in ovarian tumors indicated a potential association between HMGB1 expression level and resistance to chemotherapy(35).
Depletion of nuclear matrix-associated protein PDIA3 significantly changed chemosensitivity to araC treatment in A549 cell line. Western blot analysis of nuclear extracts from A549 and UO31 cells and immunofluorescent microscopy revealed that knockdown of PDIA3 abrogated phosphorylation of H2AX (Figs. 3, and and5).5). Importantly, phosphorylation of the genotoxic marker H2AX occurred independently of p53 phosphorylation: in cells with knocked down PDIA3, phosphorylation of H2AX was significantly inhibited at 24, 48, and 72 hr while phosphorylation of Ser15-p53 was not affected; it was observed already after 24 hr (Fig. 3). Phosphorylation of p53 and phosphorylation of H2AX are not mutually dependent, because phosphorylation of p53 at Ser 18 (corresponding to Ser 15 in human p53) and p53 stabilization were normal in H2AX-/- MEFs and thymocytes after DNA damage, and phosphorylation of H2AX was normal in the human fibroblast cells with compromised p53(36;37). Taken together with our observation that abrogation of PDIA3 by RNA interference inhibits phosphorylation of Ser 139 in H2AX but not Ser 15 in p53, these facts indicate that phosphorylation of p53 and H2AX occur via independent mechanisms. Though PDIA3 is a member of the protein disulfide isomerase family of proteins mainly present in endoplasmic reticulum, this protein was found to be associated with the internal nuclear matrix, and its DNA-binding properties have been demonstrated by DNA-protein cross-linking, chromatin immunoprecipitation and cloning of PDIA3-bound DNA(8;38). Moreover, PDIA3 was shown to cooperate with Ref-1, a protein involved in DNA repair which is a potent activator of p53(9;39). The abrogated phosphorylation of H2AX in PDIA3-depleted cells made us to assume that PDIA3 participates either in DSB formation, or H2AX phosphorylation, the hypotheses being investigated in our lab. Accumulation of γ-H2AX is important in establishing a full DNA damage response(40); therefore, PDIA3 could be an important molecular target for enhancing efficacy of conventional antimetabolite therapy. Additional experiments are needed to assess the exact roles of PDIA3 and HSPA8 in the response to genotoxic stress.
In conclusion, our results for the first time demonstrate that chromatin-associated HMGB1, HMGB2, and matrix-associated protein PDIA3 in cancer cells are important determinants of cellular response to antimetabolite drugs. Depletion of these proteins in human carcinoma cell lines resulted in alteration of cellular sensitivity to antimetabolite drugs, and changed manifestation of the cellular stress markers. In this process, proteins have different functions: HMGB1 and HMGB2 facilitate p53 phosphorylation after exposure to genotoxic stress; and PDIA3 has been found essential for H2AX phosphorylation. Hence, phosphorylation of p53, and H2AX, occurs in two distinct branches of the DNA damage response. We hypothesize that HMGB1/2 proteins act as a sensor of DNA modification, and their interaction with chemically altered DNA changes chromatin structure thus inducing DNA damage response. Elucidation of molecular components of the cell responsive to chemotherapeutic agents will rationalize strategy of anticancer chemotherapy, and will define molecular markers discriminating between therapeutic failure and success.
We are grateful to Nguyen Ngoc for excellent technical assistance. We thank Temple University students Olubunmi Fiki, Robert Tetteh, Hiren Patel, Boris Gankin, Justin Kolb, and Rostislav Meyerzon for their contribution to this project.
Grant support: This work is supported by the National Cancer Institute grant R01-CA104729 (to Evgeny Krynetskiy).
Potential conflict of interests: None