Tailoring Chemotherapy for the African-Centric S47 Variant of TP53

The p53 tumor suppressor acts as a central signaling hub in response to oncogenic, genotoxic, and metabolic stress (1–3). In turn, p53 inhibits tumorigenesis by engaging numerous pathways, including cell-cycle arrest, apoptosis, senescence, and ferroptosis (4–6). It is also a key determinant of the efficacy of numerous chemotherapeutic compounds (7, 8). Whereas most p53 pathways require the ability of this protein to function as a transcription factor, the apoptotic pathway by p53 also possesses a transcription-independent, direct mitochondrial role (9–11). Mutations in p53 in human tumors abrogate all of these tumor-suppressive pathways, and render tumors poorly responsive to therapy. In addition, in tumors with wild-type (WT) p53, there is strong evidence that single-nucleotide polymorphisms (SNP) in TP53 and p53 pathway genes can decrease tumor suppression, as well as the response to cancer therapy.

A nonsynonymous SNP at codon 47 in TP53 was reported over 20 years ago in African-descent populations and that has now been implicated in cancer (12). Approximately 6% of Africans and 1% of African Americans express a p53 allele with a serine residue at codon 47 (Pro47Ser, rs1800371). We created a mouse model for this variant, hereafter S47, and showed that mice expressing this variant in either heterozygous or homozygous form show increased risk for hepatocellular carcinoma and other cancers (13). We also showed that this variant of p53 demonstrates reduced phosphorylation on serine 46, as well as decreased ability to induce cell death by radiation and genotoxic stress (13, 14). These findings led to the possibility that individuals with the S47 variant might show reduced efficacy of cancer therapy. To test this hypothesis, we created transformed versions of WT and S47 cells using mouse embryonic fibroblasts from the WT and S47 mouse, following infection with the adenoviral E1A oncogene and activated RAS. Our goal was to use these reagents to determine whether S47 tumor cells showed enhanced tumorigenic properties, along with decreased sensitivity to commonly used chemotherapeutic drugs. We also sought to use these reagents to test the possibility that drugs might be discovered that preferentially targeted the S47 variant of p53. This might likewise reveal cell death pathways exploitable in S47 tumors. Herein, we show that this paradigm has been successful, and we have identified two chemotherapeutic compounds that are preferentially cytotoxic to S47 tumors, compared with WT p53. These studies serve as important proof of principle that chemotherapy can be tailored to p53 genotype, and they have important implications for individuals of African descent who are afflicted with cancer.

WT and S47 primary mouse embryonic fibroblasts (MEF) and E1A/RAS–transformed clones were generated previously, and MYC/RAS clones were generated as described previously (15). E1A/RAS MEFs and H1299 lung adenocarcinoma cells were grown in DMEM (Corning Cellgro) supplemented with 10% FBS (HyClone, GE Healthcare Life Sciences) and 1% penicillin/streptomycin (Corning Cellgro). WM278 cells were maintained in 80% MCDB153 (Sigma Aldrich)/ 20% Liebovitz L-15 (Corning Cellgro) media supplemented with 2% FBS and 1.68 mmol/L CaCl_2. Cell line authentication used STR profiling (ATCC). Cells were grown in a 5% CO₂ humidified incubator at 37°C. Cell lines were tested for Mycoplasma every 6 months. Cell viability, clonogenic survival, and soft agar assays were done as described previously (13, 16, 17).

The following antibodies were used: p53 (2524S), cleaved caspase-3 (9661S), cleaved Lamin A (2035S), Histone H3 (3638T), HSP90 (4877S), BAK (12105S), Ki-67 (12202S), PIN1 (3722S), c-MYC (5605S), and GAPDH (2118S; Cell Signaling Technology), RAS (610001), cytochrome C (556433; BD Biosciences), p21 (ab109199) and BRD4 (ab128874, Abcam), BRD2 (A302-583A, Bethyl Laboratories), H3 pan-Ac (39139, Active Motif), PCNA (sc-56), PIN1 (sc-15340), p53 (FL-393, sc-6243), and TOMM20 (sc-11415; Santa Cruz Biotechnology), and BAK-NT (06-536, Millipore Sigma). The following chemotherapeutic compounds were used: etoposide (E1383, Sigma Aldrich), cisplatin (HY-17394), and OTX-015 (HY-15743; MedChem Express), JQ-1 (11187), I-BET151 (11181), I-BET762 (10676), and carboplatin (13112, Cayman Chemical), and nedaplatin (S1826, Selleck Chemicals). For in vitro studies, OTX-015 was dissolved in DMSO (D8418, Sigma Aldrich). For in vivo studies, a stock of 100 mg/mL OTX-015 in DMSO was dissolved in 10% 2-hydroxypropyl-beta-cyclodextrin (HP-β-CD; CTD Holdings) in sterile water to give a final DMSO concentration of 6.25% (v/v). All compounds were dissolved in 0.9% saline solution (Sigma Aldrich, S8776). PIN1 SMARTPool siRNA was from Dharmacon.

The full-length human p53 (residues 1-393) and human p53 (residues 2-393) were cloned into the pET25 vector (Millipore Sigma) and the pGEX-4T-3 vector (GE Healthcare Life Sciences), respectively. Recombinant GST-tagged p53 proteins and N-terminal His₆-tag-p53 proteins were purified using protocols from the manufacturer and purified proteins were dialyzed against Assay Buffer (10 mmol/L HEPES/KOH, pH 7.4, 5 mmol/L KH₂PO₄, 5 mmol/L succinate, and 250 mmol/L sucrose) at 4°C. In vitro BAK oligomerization assays were performed as described previously (11, 18). Briefly, 25 μg of purified mitochondria were incubated with 100 pmol of recombinant p53 proteins. After 30-minute incubation at 4°C, the BAK oligomers were cross-linked with freshly-made 2.8 mmol/L bismaleimidohexane (BMH; Thermo Fisher Scientific) for 30 minutes at 30°C, prior to the addition of SDS loading buffer. Oligomerization products were size fractionated on 4% to 20% Tris-Glycine gradient gels (Lonza) and subjected to Western blotting using the BAK-specific antibody anti-BAK-NT (Millipore Sigma). For all other Western blot analyses, 50 to 100 μg of whole-cell lysate was resolved over SDS-PAGE gels using precast 10% NuPAGE Bis–Tris gels (Thermo Fisher Scientific) and transferred onto polyvinylidene difluoride (PVDF) membranes (IPVH00010, pore size: 0.45 μm; Millipore Sigma) prior to analysis.

Proximity ligation assay (PLA) was performed as described previously (19). Briefly, E1A/RAS cells were grown on Lab-Tek II 8-well chamber slides and fixed for endogenous protein detection as described above, using the Duolink Kit (Sigma Aldrich). As a negative control, one of the primary antibodies was omitted. For in vivo PLA, the Duolink PLA Brightfield Kit (Sigma Aldrich) was used according to the manufacturer's protocol. Quantification of PLA signals was performed using ImageJ Software (NIH, Rockville, MD).

All studies were carried out in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the NIH. All protocols were approved by The Wistar Institute Institutional Animal Care and Use Committee (IACUC). Both male and female mice were used, between 6 and 10 weeks of age. Mice were housed in plastic cages with ad libitum diet and maintained with a 12-hour dark/12-hour light cycle at 22°C. For tumor growth experiments, WT or S47 E1A/RAS MEFs were injected subcutaneously into the right flanks of 8-week-old male or female NSG (NOD.Cg-Prkdcscid II2rgtm1Wjl/SzJ) mice. Tumor volumes were measured using digital calipers and tumor volume was calculated using the formula: volume = (length × width²) × 0.52. All mice were monitored daily for signs of pain or distress. When tumors reached 10% of body weight, mice were euthanized. IHC of formalin-fixed tissues was performed as described previously (20).

For lung colonization studies, 5 × 10⁵ cells from WT or S47 E1A/RAS MEFs were injected into the tail vein of 8-week-old NSG mice. Mice were monitored daily for 2 weeks; at day 14 mice were euthanized, and the lungs were formalin fixed for further analysis. For therapy experiments, WT and S47 E1A/RAS MEFs were subcutaneously injected into 6- to 8-week-old NSG mice as described above. Once mice obtained palpable tumors (∼100 mm³) mice were randomly assigned into separate groups (8 mice per group): dilution vehicle, cisplatin, etoposide, or OTX-015. Cisplatin and etoposide were administered at a concentration of 4 mg/kg and 10 mg/kg, respectively, by intraperitoneal injection every 4 days. OTX-015 was administered daily by oral gavage at a concentration of 50 mg/kg. Tumor volumes were measured every other day until the control groups contained tumors that were 10% of body weight, at which point all mice were euthanized.

Unless otherwise stated all experiments were carried out in triplicate (n = 3) on both independent clones of E1A/RAS MEFs of each genotype (WT and S47). All mouse experiments had a minimum of n = 8 per experimental group. For in vitro studies, the two-tailed unpaired Student t test was performed. All in vitro data are reported as the mean ± SD unless stated otherwise, and in vivo data are reported as the mean ± SE. The tumor volume data are longitudinal; because the growth trends are observed in piecewise linear manner during the follow-up period, we applied a mixed effects model with linear spline regression to estimate and compare the tumor growth velocities between experimental groups, as described previously (21, 22). Statistical analyses were performed using GraphPad Prism. P values are as indicated: *, P < 0.05; **, P < 0.01; and ***, P < 0.001.

Our data in a mouse model indicated that the S47 mouse was markedly susceptible to hepatocellular cancer and histiocytic sarcoma (13). However, the possibility remained that the S47 variant was an intrinsically poorer tumor suppressor compared with WT p53 in other tissues and cell types as well. To address this question, we generated WT and S47 transformed cells using MEF that were retrovirally infected and transformed with the oncogenes E1A and RAS (Supplementary Fig. S1A). To ensure reproducibility we generated two clones of each and analyzed and pooled all data from both clones for each experiment described herein. We sequenced the entire p53-coding region from these clones and ensured there were no second-site mutations.

We tested our WT and S47 E1A/RAS cell lines for several markers of transformation, including growth in low serum, growth in soft agar, and growth as xenograft tumors. Although WT and S47 transformed cells proliferated similarly in normal culture conditions, WT cells were growth inhibited in low serum; in contrast S47 cells continued proliferating (Fig. 1A and B). Both WT and S47 clones showed equivalent numbers of colonies in soft agar, but we noted a consistent increase in colony size for the transformed S47 clones (Fig. 1C and D). Notably, we found a consistently increased ability of S47 transformed cells to develop as xenograft tumors, compared with the WT clones; this was reflected in more rapid kinetics of growth and increased S47 tumor weights at the conclusion of the study (Fig. 1E and F). IHC analysis of WT and S47 xenograft tumors of approximately similar size revealed equivalent p53 staining but markedly increased Ki-67 staining in S47 tumors (Fig. 1G).

We next sought to test the possibility that S47 might be a poorer suppressor of lung colonization in a tail vein assay, relative to WT p53. This question was of interest because a fraction of S47 knock-in mice develop tumor metastases (13). Toward this goal, we injected equal numbers of WT and S47 E1A/RAS cells into the tail vein of immunocompromised mice. Two weeks following injection, we assessed lung histology and metastatic burden, and found significantly increased lung weights and tumor burden for S47 cells (Supplementary Fig. S1B and S1C) as well as increased Ki-67 immunostaining in S47 tumors (Supplementary Fig. S1D) and tumor burden (Supplementary Fig. S1E). The combined data indicate that the S47 protein is an intrinsically poorer tumor suppressor compared with WT p53, at least for the oncogene combinations and assays used here.

We previously showed that nontransformed S47 cells are resistant to conventional agents of genotoxic stress, including adriamycin and etoposide (13). In contrast, we found that transformed S47 tumor cells show increased sensitivity to cisplatin in vitro compared with WT p53 (15). The increased sensitivity of transformed S47 cells held true for multiple concentrations of cisplatin (Fig. 2A) and was consistent when tumor cells were grown in soft agar (Fig. 2B and C). This increased sensitivity of transformed S47 cells was also evident for other platinum-based compounds such as nedaplatin and carboplatin (Fig. 2D). We next transformed WT and S47 cells with another oncogene combination, MYC plus oncogenic RAS. These transformed S47 cells showed 2.5-fold enhanced sensitivity to cisplatin compared with WT p53 (Supplementary Fig. S2A and S2B). It is important to note that primary S47 MEFs do not show increased sensitivity to cisplatin, in fact they are resistant to this drug (13), so this is a property only of transformed S47 cells. Finally, we used CRISPR/Cas9 gene editing to convert the WT p53 in the WM278 (BRAF V600E mutant) melanoma cell line to S47; we chose this cell line because it is homozygous for the proline 72 variant of p53, which is the relevant background upon which the S47 SNP arose (14). We generated two S47 knock-in clones, one with a heterozygous knock-in (WT/S47) and one with a homozygous knock-in (S47/S47). Because the S47/S47 clone showed markedly reduced steady-state levels of p53 protein, we chose to focus on the WT/S47 line. Despite showing slightly reduced level of p53 protein compared with the parental clone, the WT/S47 line showed markedly increased cell death following cisplatin treatment (Supplementary Fig. S2C). These WT/S47 cells also showed reduced IC₅₀, compared with control cells, to cisplatin, nedaplatin, and carboplatin (Supplementary Fig. S2D). The combined data are consistent with the premise that the S47 variant confers increased sensitivity to platinum-based compounds, independent of tumor background.

Our findings with cisplatin were proof of principle that drugs could be discovered that could effectively treat tumors containing the S47 variant. Therefore, we next performed a cell viability screen of a library of over 200 conventional and experimental chemotherapeutic drugs, using our WT and S47 E1A/RAS clones. The majority of tested compounds showed either reduced efficacy in S47-transformed cells, or no differences in drug sensitivity between WT and S47-transformed lines. Notably, however, a small subset of epigenetic compounds showed increased sensitivity in S47 cells: these were from the class of BET protein (bromodomain and extra-terminal motif) inhibitors (Supplementary Table S1). Analysis of cell viability of WT and S47 cells treated with several BET protein inhibitors, including the well-characterized tool compound JQ-1 (23) and the BRD2/4 inhibitor OTX-015 (currently in clinical trials; refs. 24, 25) revealed that S47 cells show a 4- to 5-fold increased sensitivity to BET inhibitors, compared with tumor cells with WT p53 (Fig. 2E; Supplementary Fig. S2E). In contrast, there was no difference in cell viability in response to the broad-spectrum bromodomain inhibitor bromosporine (Supplementary Table S1). Clonogenic survival assays with WT and S47 E1A/RAS MEFs and the BET inhibitor OTX-015 confirmed that S47-transformed cells showed significantly reduced colony-forming ability following treatment with OTX-015 compared with WT p53 (Fig. 2F and G).

We next sought to test the efficacy of cisplatin and BET inhibitors against WT and S47 tumors in vivo. Toward this goal, we injected equal numbers of both clones of WT and S47 E1A/RAS MEFs subcutaneously into immunocompromised mice. Once tumors reached 100 mm³, mice were randomized and treated by daily oral gavage with OTX-015 or by intraperitoneal injections every 4 days with 4 mg/kg cisplatin. As expected, untreated S47 tumors grew significantly faster than WT tumors (Fig. 3A). Notably, however, both OTX-015 and to a greater extent cisplatin showed markedly increased ability to cause tumor regression of S47 tumors compared with WT (Fig. 3A and B). We realized that the increased proliferative rate of S47 tumors relative to WT might contribute to the increased efficacy of these drugs in S47 tumors. Therefore, we repeated this study using another genotoxic agent, etoposide, which does not show preferential toxicity in S47 transformed cells in vitro. Unlike OTX-015 and cisplatin, etoposide did not show evidence for increased efficacy in S47 tumors compared with WT (Fig. 3C and D). We analyzed the kinetics of tumor regression following treatment using a spline mixed model; this analysis assesses the reduction in tumor volume per measurement window. A spline mixed model analysis confirmed that cisplatin, and to a lesser extent OTX-015, led to significantly greater decreases in tumor volume per treatment window in S47 compared with WT tumors (P < 0.0001 between S47 and WT for cisplatin; P = 0.0058 between S47 and WT for OTX-015; Fig. 3E and F). In contrast, the efficacy of etoposide between WT and S47 did not show statistically significant differences (Fig. 3G). IHC analysis of cisplatin-treated WT and S47 tumors revealed that this treatment led to decreased Ki-67 immunostaining in both WT and S47 tumors. In addition, in S47 tumors there was a significant increase in staining for cleaved caspase-3 compared with WT (Fig. 3H). These data suggested that the increased efficacy of cisplatin in S47 tumors was caused by increased apoptosis.

We next sought to identify potential underlying mechanisms for increased sensitivity of S47 tumors to cisplatin and BET inhibitors. BET inhibitors like JQ-1 and OTX-015 bind to the BET proteins BRD2, -3, and -4, and prevent these proteins from binding to chromatin; in some cases, BET inhibitors also cause downregulation of BET proteins (26). In some cases, the cytotoxicity of BET inhibitors can be correlated with their ability to cause downregulation of the MYC oncoprotein (27, 28). We treated WT and S47 E1A/RAS cells with JQ1 or OTX-015 and monitored the level of BRD-2 and -4, as well as MYC. We found that there were no differences in the level of BRD-2 and -4 in untreated WT and S47 cells, and that JQ-1 and OTX-015 caused a modest downregulation of BRD2 in S47-transformed cells, but no change in BRD4 level (Supplementary Fig. S2F). In contrast, we found that OTX-015 led to a marked downregulation of MYC in E1A/RAS-S47–transformed cells, much greater than that in WT cells (Supplementary Fig. S2G). This profound decrease in c-MYC might be a cause for the increased cytotoxicity of OTX-015 in S47 cells. Interestingly, we also saw a marked downregulation of MYC in E1A/RAS S47 cells treated with cisplatin (Supplementary Fig. S2H), suggesting a potentially broad role for MYC in cisplatin sensitivity; this is consistent with the findings of others (29). We also observed increased acetylated histone H3 in S47 cells, which may explain increased sensitivity to BET inhibitors (Supplementary Fig. S2I). Because our data indicated that cisplatin outperformed BET inhibitors in S47 tumors, we focused further on understanding the mechanistic basis for increased sensitivity to cisplatin.

To begin to elucidate the potential mechanism for enhanced cell death of S47 cells by cisplatin, we first assessed the transcriptional pathway of p53-mediated cell death and performed qPCR of a dozen p53 target genes; we did not see increases in the proapoptotic genes PUMA, NOXA, and BAX in S47-transformed cells, compared with WT, following cisplatin treatment. Therefore, we next sought to assess the transcription-independent cell death pathway of p53, which uses a direct mitochondrial mechanism of cell death involving a direct interaction of p53 with BCL2 family members (9–11). To do this, we first tested the ability of cycloheximide, which blocks the transcriptional pathway by inhibiting new protein synthesis (30), to block cisplatin-mediated apoptosis. We also purified mitochondria from WT and S47 E1A/RAS cells, and used Western blotting to assess the level of p53 colocalized with this organelle. Finally, we performed proximity ligation assays (PLA) to assess the p53–BAK interaction, which is important in transcription-independent apoptosis (12). We found that cycloheximide treatment inhibited the induction of the p53-target protein p21 (CDKN1A), but had no effect on apoptosis in S47 cells, as assessed by the accumulation of cleaved caspase-3 (Fig. 4A, compare CC3 lanes 6 and 8). We next isolated purified mitochondria from untreated and cisplatin-treated WT and S47 E1A/RAS cells, and found markedly increased amount of the S47 protein at the mitochondria of treated cells compared with WT p53 (arrows, Fig. 4B). Consistent with this, PLA analysis revealed increased association of S47 protein with BAK at the mitochondria of S47 cells following cisplatin treatment compared with WT cells (Supplementary Fig. S3A). These combined findings support the conclusion that the increased cell death of transformed S47 cells by cisplatin is through increased usage of the direct mitochondrial, transcription-independent cell death pathway of p53.

The peptidyl-prolyl isomerase PIN1 interacts directly with p53 and is required for the localization of p53 to mitochondria (31). Although serine 46 phosphorylation is important for the PIN1 interaction, there are many PIN1-binding sites in p53 protein (32). To determine whether PIN1 plays a role in the increased trafficking of S47 to the mitochondria following cisplatin treatment, we performed PLA to assess the interaction of WT p53 and S47 with PIN1. PLA analyses showed that the association between S47 and PIN1 was markedly increased compared with WT p53 (Fig. 4C), although there were equivalent levels and immunostaining of p53 and PIN1 in WT and S47 cells (Supplementary Fig. S3B). We next silenced PIN1 using siRNA in WT and S47 E1A/RAS cells and assessed cell death following cisplatin treatment. As predicted from published data (31), silencing of PIN1 reduced cell death by the S47 variant, thus supporting a role for PIN1 in this cell death pathway (Supplementary Fig. S4A). In addition, we found that silencing PIN1 markedly reduced the ability of the S47 variant to localize to the mitochondria, as assessed by PLA of S47 with TOMM20, which binds to mitochondrial p53 (Fig. 4D; refs. 33, 34).

A key step in the mitochondrial cell death pathway by p53 involves interaction with proapoptotic BAK; p53 induces conformational changes in BAK, leading to BAK oligomerization, cytochrome c release, and caspase activation (35). Whereas our data indicate that S47 traffics efficiently to mitochondria and binds to BAK, it remained formally possible that this variant was impaired for the ability to drive BAK oligomerization. To address this, we performed BAK oligomerization assays using purified mitochondria and bacterially expressed and purified WT and S47 proteins, isolated using either GST or 6-Histidine tags. Incubation of purified mitochondria with either GST-WT p53 or GST-S47 followed by incubation with the cross-linking agent BMH and SDS-PAGE led to the appearance of oligomerized BAK (76 kDa band; Fig. 5). Notably, S47 protein could induce BAK oligomerization in these assays slightly better than WT p53, purified with either tag (Fig. 5). These data support the conclusion that once at the mitochondria S47 protein is capable of inducing BAK oligomerization and cell death via the mitochondrial pathway.

To confirm and extend our findings, we next sought to test the premise that S47 tumors from cisplatin-treated mice also show enhanced usage of the mitochondrial cell death pathway. To do this, we performed PLA assays on formalin-fixed tumor material from WT and S47 tumors following cisplatin treatment and monitored p53 localization at mitochondria by assessing its interaction with TOMM20 and BAK. Not surprisingly, in untreated tumors there was no evidence for complexes between p53 and TOMM20, or p53 and BAK. However, there were marked increases in p53–TOMM20 and p53–BAK complexes in S47 cisplatin-treated tumors compared with WT (Fig. 6A and B; Supplementary Fig. S4B–S4D). These data confirm that the S47 variant shows increased engagement of the mitochondrial pathway in E1A/RAS-induced tumors.

Personalized medicine is a concept that has been gaining appreciation over the years, yet there are few examples of truly personalized medicine approaches for cancer. Children with nonfunctional SNPs in TPMT do not get treated with 6-mercaptopurine or 6-thioguanine, to prevent highly toxic side effects. Individuals with melanoma are sequenced for mutations in the BRAF gene before being treated with BRAF inhibitors. And women with breast cancer are analyzed for ER positivity and HER2 level before being treated with SERMs (selective estrogen receptor modifiers) or Herceptin. It is important to note that these cases are all relatively clear cut: these compounds either work or do not work based upon the genetics of the individuals (TPMT) or the tumors (SERMs, BRAF inhibitors, and Herceptin). What we are proposing here is slightly different: in the case of S47 individuals with cancer, general cytotoxic chemotherapeutic drugs are likely to show some efficacy, but cisplatin and BET inhibitors should work better. This improved efficacy could have dramatic consequences on cancer survival in S47 individuals; greater efficacy means fewer treatments, and this leads to reduced chance that drug-resistant cells can emerge. We posit that African-descent individuals diagnosed with tumors where cisplatin is one therapeutic possibility (such as for example lung cancer, bladder cancer, or triple-negative breast cancer) should be encouraged to undergo genotyping for the S47 SNP; we propose that S47-positive individuals should be treated with cisplatin or a BRD inhibitor, where the choice is a possibility.

Since its discovery in 1979, p53 has been a cornerstone of cancer biology, and mutational inactivation of this tumor suppressor is a hallmark of human cancer (36–38). Our group and others have obtained data that in addition to mutations, SNPs in p53 and p53 pathway genes can impact the signaling potential of this pathway, and likewise can adversely impact cancer risk and therapy. We modeled the S47 SNP in mice and showed that S47 mice have high propensity for developing spontaneous liver and other cancers (13). One question remaining from our study was whether S47 was a poorer tumor suppressor in other cell types; herein, we show that this is the case. It remains to be determined why this is the case. One possibility is that the ferroptotic defect in S47 mice leads to significant iron accumulation in these cells (J. Leu, unpublished manuscript). It is well known that iron is required for cell proliferation because it is required for ribonucleotide reductase and other enzymes in cell division (39, 40). This may explain the increase in Ki-67 staining in S47 tumors compared with WT.

It is well established that p53 can promote apoptosis using a variety of different mechanisms (41, 42). Although the canonical role of p53 as a transcriptional activator of proapoptotic genes is well characterized, it is also clear that p53 possesses a transcription-independent role in apoptosis, by binding to BCL2 family members (9, 43). Multiple reports have shown that this occurs in part by the ability to p53 to traffic to the mitochondria of transformed cells, in turn causing the oligomerization of the proapoptotic effectors BAK and BAX, leading to cytochrome c release and caspase activation (9–11). We found that S47-transformed MEFs, which are more sensitive to cisplatin, show increased trafficking of p53 to the mitochondria and increased ability to oligomerize BAK, suggesting that S47 cells show enhanced apoptosis by activating the mitochondrial cell death program. We also show that PIN1, which is required for mitochondrial localization of p53 (31), shows enhanced binding to S47 protein, thus providing a mechanistic explanation for our findings (see model in Fig. 6C). Some of these findings were quite surprising, given that phosphorylated serine 46 is a major docking site for PIN1 (31). However, PIN1 binding to p53 occurs on several other phosphorylated residues (32); in addition, other posttranslational modifications on p53 are implicated in mitochondrial trafficking (44, 45). One possibility is that serine 47 is a site for phosphorylation by casein kinase II, and this may influence mitochondrial trafficking or cell death by p53. We have been unable to detect the phospho-S47 species in human or mouse cells using polyclonal antisera raised against this phosphopeptide (M. Murphy, unpublished results); however, we cannot rule out this formal possibility.

We find that S47 protein can directly cause BAK oligomerization; these data contrast with the findings of Kriwacki and colleagues, who show that the S47 variant of p53 cannot oligomerize BAX (46). Although the results of that study indicate that the S47 variant might be impaired for certain aspects of the mitochondrial pathway, we clearly find that this is not the case. The reasons for this discrepancy are unclear: it could reflect the proapoptotic target protein analyzed in our study versus theirs (BAK vs. BAX), the form of p53 used (they used an N-terminal fragment of p53, whereas we used full-length) or other assay-specific differences. We were unable to detect a p53–BAX complex or oligomerized BAX in our cells so we were unable to address this discrepancy directly.

Our data indicate that cisplatin, but not several other DNA-damaging agents tested, preferentially induces the mitochondrial pathway of p53-mediated apoptosis. The reasons for this activity of cisplatin are unclear, but our data are consistent with the findings of others. Cisplatin is known to use the protein TAB1 to activate p38α, which activates p53 to engage the direct mitochondrial pathway of cell death (47). In addition, cisplatin-mediated apoptosis requires p53-mediated generation of reactive oxygen species (ROS); ROS then facilitates mitochondrial localization of p53 (48). We consistently find that the ROS levels of S47 cells are higher than WT cells (S. Basu, unpublished manuscript). Finally, unlike most other genotoxic agents, cisplatin forms a variety of structural adducts on DNA that inhibit transcription, thus dampening the p53-dependent transcriptional pathway (49, 50). These combined factors likely explain the preferential use of the direct mitochondrial pathway of cell death by S47 in response to cisplatin.

Collectively, our data suggest that personalized medicine approaches to chemotherapy based upon a p53 SNP are possible and promising. It is important to note that although we see increased cytotoxicity by cisplatin in transformed S47 cells, we see the opposite for nontransformed S47 cells and tissues, which show reduced cytotoxicity by cisplatin, compared with cells or tissues with WT p53 (13). These data augur well for the use of cisplatin in S47 individuals with cancer. It is our hope that this research will spur the genotyping of African Americans with cancer, to improve the therapeutic success and survival of these individuals.

No potential conflicts of interest were disclosed by the authors.

Conception and design: T. Barnoud, R. Zhang, M.E. Murphy

Development of methodology: T. Barnoud, A. Budina-Kolomets

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): T. Barnoud, A. Budina-Kolomets, S. Basu, J.I.-J. Leu, M. Good, C.-P. Kung, J. Liu, D.L. George, M.E. Murphy

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): T. Barnoud, A. Budina-Kolomets, S. Basu, J.I.-J. Leu, M. Good, Q. Liu, D.L. George, M.E. Murphy

Writing, review, and/or revision of the manuscript: T. Barnoud, A. Budina-Kolomets, S. Basu, J.I.-J. Leu, M. Good, Q. Liu, J. Villanueva, R. Zhang, D.L. George, M.E. Murphy

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): T. Barnoud, J. Villanueva, M.E. Murphy

Study supervision: J.I.-J. Leu, M.E. Murphy

This work was supported by R01 CA201430 and R01 CA102184 (to M.E. Murphy); R01 CA139319 and P01 CA114046 (to M.E. Murphy, D.L. George, and J.I. Leu); TL1TR002344 (to C.P. Kung); and F32 CA220972 (to T. Barnoud). The Molecular Pathology and Imaging Core at the University of Pennsylvania (Philadelphia, PA) is funded by the Center for Molecular Studies in Digestive and Liver Diseases (NIH P30 DK050306), P01 CA098101, and P01 DK049210. Support for the Core Facilities used in this study was provided by Cancer Center Support Grant CA010815 to The Wistar Institute. The authors would like to acknowledge the Histotechnology, Laboratory Animal, Molecular Screening and Imaging facilities at The Wistar Institute. The authors would also like to thank Joshua Parris and Keerthana Gnanapradeepan in the Murphy Laboratory for assistance and thoughtful discussions, and Ileabett Echeverria-Vargas for help with administration of OTX-015.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.