Differential Effects of Clinically Relevant N- versus C-Terminal Truncating CDKN1A Mutations on Cisplatin Sensitivity in Bladder Cancer Free

Radical cystectomy is central to the standard of care for muscle-invasive bladder cancer (MIBC). Cisplatin-based neoadjuvant chemotherapy (NAC) is recommended prior to radical cystectomy with the goal of reduction/elimination of micrometastatic disease. In the absence of a clinical assay to measure the response of micrometastatic disease, pathologic complete response or pathologic downstaging is used as a surrogate. Between 15% and 40% of patients respond completely to NAC (1–3). Accurate prospective identification of patients who will benefit the most from NAC has numerous potential benefits including consideration of NAC avoidance in predicted nonresponders and the consideration of cystectomy avoidance in candidate complete responders. Indeed, several clinical trials are currently recruiting with the intent of cystectomy avoidance based on such biomarkers (NCT02710734 and NCT03609216). Recent characterization of mutational landscape of MIBC has shown that loss-of-function mutation in genes involved in DNA repair is strongly associated with chemoresponse (4–6). These findings may be a clinical biomarker that will allow clinicians to triage patients into cystectomy avoidance algorithms.

Somatic mutations that affect cellular pathways pertinent to chemoresponse may make valid clinical biomarkers (4–6). MIBCs frequently harbor mutations in the CDKN1A gene, which encodes the tumor suppressor and p53 target gene p21, with the majority of alterations truncating the peptide (7, 8). Although alteration in p21 expression levels has been observed different tumor types, mutations in the CDKN1A gene are rarely present in other common types of cancer (Table 1; ref. 9). Canonically, p21 executes its tumor suppressor function by inhibiting CDK–cyclin complexes, halting the cell cycle (10–13). Although in nonmalignant cells, this mechanism functions to protect the genome against unwanted aberrations, in cancer cells this mechanism may help to mitigate genotoxicity and protect the cell against chemotherapeutics (14). Given this line of evidence, we hypothesized that after DNA-damaging events, cells deficient in p21 would be unable to halt the cell cycle to efficiently repair the damaged DNA, and thus proceed down the apoptotic pathway, potentially supporting a role for CDKN1A loss of function as a predictive biomarker for chemoresponse.

Cells were harvested and lysed using ice-cold RIPA buffer supplemented with protease and phosphatase inhibitors and the lysates were clarified by centrifugation. Protein concentration was determined using BCA assay. Lysates were denatured by boiling in LDS sample buffer with DTT. Equal amounts of protein were loaded per well on a NuPAGE 4%–12% Bis-Tris gel and separated by electrophoresis. The resolved proteins were transferred onto PVDF membrane, which were blocked using 5% milk in TBST. The blocked membranes were probed with the desired primary antibody (antibody information found in Supplementary Tables S1 and S2), followed by the species-matched HRP-labeled secondary antibody. Proteins were visualized using ECL substrate.

Single-guide RNA (sgRNA) sequences were chosen using sgRNA Designer (https://portals.broadinstitute.org/gpp/public/analysis-tools/sgrna-design; Supplementary Table S3). Lentiviruses were produced in HEK293T cells. HEK293T cells (70% confluent) in 10-cm dishes were transfected with 4 μg psPAX2, 4 μg of pLentiCRISPR-E (Addgene, #78852) containing sgRNA directed to CDKN1A or GFP (which was used as a control), and 2 μg of CMV-VSVG using Lipofectamine 2000. pLentiCRISPR-E contains a cloning site for one sgRNA, S. pyogenes espCas9 v1.1 (15), and a puromycin-resistance cassette. Forty-eight hours later, conditioned media containing the lentiviral particles were collected and clarified by centrifugation.

SW780 and RT4 cells were infected by adding infectious media with 8 μg/mL of polybrene. After 48 hours of infection, cells were sparsely reseeded and treated with an empirically determined dose of puromycin for 7 to 10 days in parallel. Stably infected pooled cells were collected for indel quantitation, and a small fraction were replated at a low density (100 cells per 10 cm dish) for single-cell cloning/expansion.

For indel quantitation, DNA from pooled cells was PCR-amplified using primers flanking the expected indel sites. The purified PCR products were subjected to Sanger sequencing and analyzed with TIDE (16). The sequencing data were compared with the unedited DNA (from negative control cells) to confirm edits.

Cells were seeded into black 96-well plates at a density of 5,000 cells per well and allowed to attach overnight. Plated cells were treated with increasing doses of cisplatin for a 3-hour pulse dose. Cell viability was assessed using CellTiter-Glo after 72 hours of recovery.

J82, HT1376, 5637, SW780, RT4, TCCSUP, UMUC3, and HEK293T cells were obtained from ATCC. RT112 cells were obtained from Cell Line Services. UMUC13 cells were generously provided by Dr. Evan T. Keller (University of Michigan). J82 and HT1376 cells were cultured in MEM supplemented with 10% heat-inactivated fetal bovine serum (FBS), nonessential amino acids (NEAA), l-glutamine, and sodium pyruvate. SW780 and 5637 cells were cultured in RPMI-1640 supplemented with 10% FBS, HEPES, l-glutamine, and sodium pyruvate. RT4 cells were cultured in McCoy's 5A medium supplemented with 10% FBS. TCCSUP and UMUC3 were cultured in EMEM supplemented with 10% FBS, NEAA, and sodium pyruvate. HEK293T and UMUC13 cells were cultured in DMEM supplemented with 10% FBS. RT112 cells were cultured in RPMI-1640 supplemented with 10% FBS, l-glutamine, and sodium pyruvate. All cell lines were grown in a humidified incubator at 37°C and 5% carbon dioxide. All cells were regularly tested for Mycoplasma using a previously described degenerate PCR-based approach (17).

For all cisplatin treatments, cells were seeded, allowed to attach overnight, and incubated in media containing the appropriate dose of cisplatin for a 3-hour pulse dose because this is the most physiologic model. Cisplatin serum half-life is approximately 3 hours (18, 19). Treated cells were allowed to recover for the designated period of time and were harvested according to the requirements of the downstream assay.

xCELLigence assay measures cell growth or death by measuring the impedance of a small electrical current that passes through a surface onto which cells are attached. As more cells attach, the impedance increases. This assay was used to measure the cell proliferation of SW780 and RT4 cells with different p21 truncations after optimizing assay conditions. 8,000 SW780 sgGFP, sg109A, and sg12A cells, or 12,000 RT4 sgGFP, sg109A, and sg12A cells were plated on 16-well E-plates in quadruplicates and allowed to settle for 1 hour. A baseline measurement of growth was recorded for 20 hours. SW780 cells were treated with 2.5 μmol/L, 5 μmol/L, or 10 μmol/L cisplatin, and RT4 cells were treated with 5 μmol/L, 10 μmol/L, or 15 μmol/L cisplatin. After 3 hours, media were replaced by 200 μL of fresh media, and the cell index was recorded for 1 week using 15-minute sweeps.

Co-IP was performed using the Crosslink Magnetic IP/Co-IP kit (Pierce, #88805). RT4 cells were plated in 10-cm dishes overnight. Cells were harvested 9 hours after the end of treatment and lysed using 1 mL of Pierce IP Lysis/Wash Buffer (provided with the kit) for 10 minutes at 4°C with constant agitation. The samples were briefly sonicated and clarified by centrifugation at 13,000 rpm for 10 minutes. Protein concentration was measured using BCA assay, and 1 mg of protein in 0.5 mL was used for each IP.

p21 was immunoprecipitated overnight at 4°C using 2 μg of p21 antibody crosslinked to protein A/G magnetic beads according to the manufacturer's protocol. The beads were separated using a magnetic stand and the supernatant was collected and saved for Western blot analysis. The beads were washed using Pierce IP Lysis/Wash buffer and eluted using the low pH elution buffer provided with the kit. Following elution, the samples were neutralized using the neutralization buffer provided with the kit.

The eluted immune complex, along with the supernatant, was denatured by boiling with DTT and LDS sample buffer and resolved by SDS-PAGE. The resolved proteins were transferred onto a PVDF membrane and immunoblotted for p21 and PCNA.

All fluorescence microscopy was performed using the Leica SP8 confocal microscope at 630X. All slides were mounted using Vectashield mounting medium (Vector Laboratories).

Analysis of cellular proliferation was performed using the Click-iT Plus EdU imaging kit (Thermo, #C10637). Approximately 40,000 cells were seeded per well in 8-well chamber slides and allowed to attach overnight. Cells were treated with cisplatin and allowed to recover for the designated period of time in fresh drug-free media. An hour before harvest, the cells were switched to fresh media containing 20 μmol/L EdU and incubated. Cells were washed with PBS and fixed in 4% paraformaldehyde for 15 minutes at room temperature. Cells were permeabilized in 0.5% Triton X-100 in PBS, and Click-iT reaction to detect EdU was performed according to the manufacturer's protocol. EdU incorporation was visualized using Alexa Fluor 488, and the nuclei were counterstained with Hoescht 33258.

Approximately 40,000 cells were seeded per well in 8-well chamber slides and allowed to attach overnight. Cells were treated with cisplatin and allowed to recover for the designated period of time in fresh drug-free media. Cells were washed in PBS, fixed in 4% paraformaldehyde for 10 minutes, and permeabilized in 0.2% Triton X-100. The permeabilized cells were blocked using 10% goat serum in PBS and were incubated overnight at 4°C with anti-RPA70 antibody (1:500) diluted in the blocking solution. Cells were incubated in goat anti-rabbit secondary antibody conjugated to Alexa Fluor 594 for 1 hour at room temperature and then counterstained with Hoescht 33258. RPA70 foci-positive cells were counted using Fiji.

Cells were seeded, treated, fixed, permeabilized, and blocked as described above. Cells were incubated overnight at 4°C with anti-gamma-H2AX antibody (1:400) in blocking solution. Cells were incubated in goat anti-rabbit secondary antibody conjugated to Alexa Fluor 488 for 1 hour at room temperature and then counterstained with Hoescht 33258.

Cells were seeded, treated, fixed, permeabilized, and blocked as described above. Cells were incubated overnight at 4°C with anti-p21 antibody (1:400) in blocking solution. Cells were incubated in goat anti-rabbit secondary antibody conjugated to Alexa Fluor 488 for 1 hour at room temperature and then counterstained with Hoescht 33258.

Cells were seeded in 6-well plates and treated with 20 μmol/L cisplatin for 3 hours. Both floating debris and adherent cells were collected after 72 hours and fixed in ice-cold 70% ethanol. The fixed cells were incubated in 50 μg/mL propidium iodide and 250 μg/mL RNase and analyzed by flow cytometry using an EPICS elite flow cytometer (Beckman-Coulter). Sub-G₁ fraction were analyzed using FlowJo software (FlowJo, LLC).

Cells were plated in 6-cm dishes and allowed to attach overnight. Cells were treated with cisplatin for 3 hours and recovered for the designated period of time. DNA was isolated using phenol-chloroform extraction. The concentration of DNA per sample was measured by OD₂₆₀. In order to determine the amount of DNA-bound platinum (Pt) per sample, the isolated DNA was suspended in an acid matrix comprised of 2% HNO₃ and 0.5% HCl (both v/v), and then Pt analysis was completed using an inductively coupled plasma mass spectrometer (ICP-MS; Agilent 7900 with high matrix introduction system set to “general purpose”). The amount of platinum adducts per μg of DNA in each sample was then calculated using the two values using a calibration curve.

Cisplatin was obtained from the Fox Chase Cancer Center infusion pharmacy. Supplementary Tables S3 and S4 contain a list of the all primary and secondary antibodies that were used.

All statistical analyses were performed using GraphPad Prism 7. Data are presented as means ± standard deviation from at least three replicates. One-way or two-way ANOVA was performed to determine statistical significance. Statistically significant changes are indicated in the figures with P values (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001).

To test our hypothesis that truncating CDKN1A mutations might affect platinum sensitivity, we wanted to identify cells that were (i) cisplatin resistant and (ii) harbored WT CDKN1A in order to make isogenic cell line pairs. The cisplatin sensitivity of nine commonly used bladder cancer cell lines was determined by assessing PARP cleavage 24 hours after treatment. Next, we examined CDKN1A and TP53 mutation status using the Cancer Cell Line Encyclopedia (20). RT4 and SW780 cell lines were chosen due to their known wild-type CDNK1A and TP53 status and relative resistance to cisplatin (Fig. 1B and C). p21 and p53 are both upregulated in these cell lines after cisplatin treatment in a time- and dose-dependent manner, suggesting that p21 and p53 could play a role in cisplatin-induced DNA damage response (Fig. 1D).

CDKN1A most often harbors indels (presumably truncating) or nonsense (definitely truncating) mutations that are concentrated in the N-terminal half of the p21 peptide but can be found throughout the coding region in bladder cancer. In order to simulate patient-derived CDKN1A mutations in bladder cancer cell lines, we targeted three CRISPR sgRNAs along the length of the CDKN1A gene (Fig. 1A). As a negative control, we generated sgRNA-targeting GFP. sgRNAs were then cloned into pLentiCRISPR-E and assessed for indel generation. sgRNAs edited the CDKN1A gene with high efficiency after puromycin selection, and analysis of selected single-cell clones showed largely homogeneous populations (Fig. S1). The consequent CRISPR edits were assessed at the protein level using immunoblot. There was no detectable p21 protein in the sg12 (N-terminal indel) or sg59 (mid-protein indel) single-cell clones. Interestingly, truncated peptides were observed in the sg109 (C-terminal indel) single-cell clones (Fig. 1F). The sg109 single-cell clones behaved similarly to the parental cells and upregulated p53 and the truncated p21 peptide following cisplatin treatment, which was expected given that p21 is transcriptionally controlled by p53. Given that the truncated p21 peptide was expressed, retains its CDK-inhibiting domain, and is upregulated after DNA damage, the truncated peptide may play important functions in the cell after DNA-damaging events.

In order to test the hypothesis that CDKN1A frame-shifting mutations sensitize cells to cisplatin, several orthogonal assays were performed to evaluate cisplatin sensitivity in the CRISPR-edited clones. Cisplatin dose response was measured in CRISPR-edited and control clones. Cell viability was significantly reduced by cisplatin in both SW780 sg12A and RT4 sg12A clones, where p21 is completely abrogated, as compared with unedited sgGFP control (Fig. 2A). In order to determine if this reduction in cell viability was not only a product of hindered cell proliferation but actual cell death/apoptosis, sub-G₁ DNA fragmentation assay was used. SW780 sg12A and RT4 sg12A had significantly greater amount of sub-G₁ fraction 72 hours after being treated with a 3-hour pulse dose of cisplatin, suggesting apoptosis was occurring (Fig. 2D and E). In confirmation, a dose- and time-dependent increase in PARP cleavage in both SW780 sg12A and RT4 sg12A cells was observed. PARP cleavage was also assessed in several other SW780 sg12 clones. Independent clones were observed to behave similarly, indicating sensitization to cisplatin after p21 loss occurs due to increased cell death (Fig. S2). Therefore, SW780 sg12A was carried forward for the rest of our experiments. Interestingly, PARP cleavage was delayed and was only induced at higher doses in both SW780 sg109A and RT4 sg109A cells when compared with their unedited sgGFP counterparts (Fig. 2B and C). In addition, Xcelligence cell proliferation assay confirmed that RT4 sg12A and SW780 sg12A clones were more sensitive than their respective sgGFP control cells (Fig. S3). The SW780 sg12A clone was also much more sensitive to cisplatin than the sg109A clone, but the RT4 sg12A and sg109A clones had similar cisplatin sensitivity. Taken together these results support the notion that complete loss of p21 sensitizes cells to cisplatin, whereas a distal truncation results in a retained and truncated peptide that may enhance resistance to cisplatin.

Cisplatin forms bulky platinum–DNA adducts that are predominantly removed and repaired by nucleotide excision repair (NER; ref. 21). Previous studies have shown that p21 deficiency is associated with impaired DNA repair, and given our previous observation that p21 loss sensitized bladder cancer cells to cisplatin, we asked whether cisplatin sensitization was a result of impaired DNA repair (22, 23). We measured DNA deplatination using ICP-MS over the course of 24 hours and found that the rate of removal of platinum adducts was significantly lower in SW780 sg12A cells compared with SW780 sgGFP cells (Fig. 3A).

During NER, repair synthesis is initiated to fill in the gap following the excision of the DNA lesion (24). The repair synthesis capacity of the NER pathway in the p21 WT and p21-deficient cells was evaluated using EdU incorporation assay. EdU is a thymidine analogue and is incorporated into the repair site after the damaged section has been excised (25–27). We hypothesized that loss of p21 would result in loss of EdU incorporation during DNA repair. Cells were treated with a 3-hour pulse of 30 μmol/L cisplatin, then allowed to recover for 16 hours, and then incubated in media containing 20 μmol/L EdU for 1 hour and harvested. Using this method, we observed greater EdU incorporation in SW780 sgGFP cells than SW780 sg12A cells after cisplatin treatment, suggesting that proximal truncation of the p21 peptide prevents DNA repair (Fig. 3B).

To determine whether impaired repair and increased level of DNA-platinum adduct resulted in more DNA DSBs, gamma-H2AX, a commonly used marker of DSBs, was measured in two ways. First, immunofluorescence microscopy demonstrated that gamma-H2AX was induced more strongly in p21-deficient SW780 cells (Fig. 3C). This was further corroborated using immunoblotting (Fig. 3D). Together, these findings indicate that p21-deficient cells accumulate more cisplatin–DNA adducts, and this is associated with more DSBs. This could be explained by an absence or reduction in NER after exposure to cisplatin, which is the main mechanism of cisplatin adduct removal.

Cisplatin forms bulky platinum-DNA adducts that impede the DNA replication machinery (28). Given that p21-deficient cells suffer from impaired DNA repair, resulting in persistence of DNA–platinum adducts, we hypothesized that these cells would experience a greater degree of stalled replication forks and activate the cell-cycle checkpoint proteins. To test this hypothesis, RPA foci formation was used first to assess the amount of stalled replication forks, which would label ssDNA found at replication forks. Surprisingly, we found that a larger percentage of SW780 sgGFP cells formed RPA foci than SW780 sg12A cells after 8 and 16 hours of recovery after a 3-hour pulse treatment with cisplatin (Fig. 4A and B). This indicated that there were more SW780 sgGFP cells with stalled replication forks than SW780 sg12A cells.

To explore this finding in more detail, cell-cycle checkpoint proteins were measured in isogenic cell line pairs after cisplatin treatment. Phospho-Chk1 (S317 and S345), which is activated by ATR predominantly in response to tracts of single-stranded DNA (ssDNA) at stalled replication forks (29), was strongly induced in SW780 sgGFP and SW780 sg12A after cisplatin treatment (Fig. 4C). Once Chk1 is activated by phosphorylation, it transduces distal signaling cascades that regulate a variety of cellular functions including cell-cycle arrest, apoptosis, and replication fork stabilization (29). CDK1, which plays a role in G₂–M transition, is inhibited by the actions of downstream effectors of activated Chk1. CDK1 phosphorylation at Y15 inhibits its function and leads to cell-cycle arrest in G₂–M. We found that phospho-CDK1 (Y15) decreased gradually in SW780 sg12A and increased modestly in SW780 sgGFP after cisplatin treatment (Fig. 4C and D). This suggests that the loss of p21 results in the inability for DNA damage to arrest cells due to a loss of the maintenance of CDK1 Y15 phosphorylation.

Although Chk1 is equally activated in both cell types, p21-deficient SW780 sg12A cannot maintain G₂–M checkpoint arrest due to pressure to progress through the cell cycle from the release of inhibition of CDK1. This phenomenon may also explain the increased gamma-H2AX seen in SW780 sg12A cells (Fig. 3C), as these cells would be ineffective at sustaining the stalled replication forks, resulting in increased DSBs. p21 is an important modulator of CDK1 activation (10) and given these findings, it is reasonable to deduce that in the absence of p21, cells are unable to effectively inhibit CDK1 and arrest in the G₂–M phase of the cell cycle, resulting in increased cisplatin sensitivity, possibly through increased replication fork collapse.

p21 function is entirely mediated through its interactions with other proteins (10). It has also been shown that the functions that p21 plays in the cell can be affected by its subcellular localization (10, 30). Given that the nuclear localization sequence of p21 is located in the C-terminus, we sought to establish how distal CDKN1A frameshift resulting in C-terminal neopeptide would affect its subcellular localization (11). This was determined using immunofluorescence microscopy. Whereas full-length p21 in SW780 sgGFP and RT4 sgGFP localized to the nucleus after cisplatin treatment, truncated p21 peptide in SW780 sg109A and RT4 sg109A remained in the cytoplasm and failed to localize to the nucleus (Fig. 5A). Lack of nuclear localization after cisplatin adduction would affect its ability to halt the cell cycle because it would not be able to interact with its target cyclins and CDKs (30). Therefore, although wild-type full-length p21 and truncated p21 are both similarly induced post cisplatin treatment, their differential subcellular localization may affect what functions it plays in the cell.

Additionally, the PCNA-binding motif is also located at the C-terminus of p21 (amino acids 143–160; ref. 11). PCNA is vital in the NER process as it binds different DNA polymerases and orchestrates repair synthesis (24, 31). p21 binds and sequesters PCNA following DNA damage, mediating tightly regulated PCNA–DNA polymerase interactions (31, 32). With this in mind, we hypothesized that the truncated p21 peptide would be unable to bind to PCNA. This was tested using coimmunoprecipitations to explore the p21–PCNA interaction. Full-length p21 was in a complex with PCNA; however, the truncated p21 did not pulldown PCNA, indicating lack of interaction between the two peptides (Fig. 5B).

In this study, we have shown that proximal CDKN1A truncating mutations result in complete loss of p21-sensitized bladder cancer cells to cisplatin, whereas a more distal truncating mutation results in a stable peptide and enhances resistance. Apoptosis was strongly induced by cisplatin in p21-deficient SW780 and RT4 cells. p21-deficient SW780 cells were less proficient at removing bulky DNA–platinum adducts, accumulated more platinum adducts, and sustained more DNA DSBs. In the absence of p21, SW780 cells were unable to prevent the activation of CDK1. This inability to effectively inhibit CDK1, which is integral for cell-cycle progression and a crucial p21 target, may lead to ineffective maintenance of G₂–M cell-cycle arrest and hinder stabilization of stalled replication forks. Together, these findings help to explain the increased sensitivity seen in the p21-deficient SW780 cells (Fig. 6).

p21 lacks a structure until it binds to other proteins (33), and this lack of structure may explain why a frame-shifting indel does not result in nonsense-mediated decay. Interestingly, although the C-terminal truncated p21 peptide was induced in a similar manner to the full-length wild-type p21 after genotoxic cisplatin treatment, it behaved differently in the cell. The C-terminally truncated p21 peptide failed to localize to the nucleus and perhaps more importantly was unable to interact with PCNA. As a loading-clamp for DNA polymerases, PCNA is integral in facilitating DNA repair and DNA damage tolerance through translesion synthesis. Normally, full-length wild-type p21 binds PCNA, outcompeting DNA polymerases and tightly regulating these processes (10, 31). However, in the absence of a p21–PCNA interaction, these processes may run unchecked, contributing to cellular resistance to cisplatin. Differential cellular localization and lack of interaction with PCNA may play a crucial role for conferring greater degree of resistance observed in the SW780 sg109A and RT4 sg109A cells.

p21 is transcriptionally activated by p53 after DNA-damaging events and brings about cell-cycle arrest by inhibiting CDK–cyclin complex needed for proper cell-cycle progression and inhibiting cell proliferation by binding PCNA to obstruct DNA polymerase loading (10, 34). p21 carries out its cellular functions primarily by binding its target proteins (35). Therefore, truncating mutations resulting in loss of C-terminal domains but preservation of the CDK inhibitor domain can significantly affect what roles it plays in the cell. Loss of p21 impairs DNA damage–induced checkpoint activation and maintenance of G₂–M cell-cycle arrest. When p21 is present, it prevents the activation of CDK1 to maintain G₂–M cell-cycle arrest to afford the damaged cells time to repair the damaged DNA. This is not possible in the absence of p21, as CDK1 is activated and the cells progress through the cell cycle without sufficient time for DNA repair. Additionally, as the cells continue to progress through the cell cycle with unrepaired bulky DNA–platinum adducts, the cells encounter additional replication stress, leading to destabilization and eventual collapse of the stalled replication fork. We showed that collapsed replication forks resulting in DSBs are highly toxic and probably lead to the increased apoptosis seen in the p21-deficient cells.

The diverse functions p21 plays in the cell create an added layer of complexity in deciphering the consequence of distal truncating mutations (10). We found that truncated p21, lacking functional PCNA-binding domain (PIP-box) and nuclear localization sequence, is associated with a greater degree of resistance to cisplatin. This finding warrants further exploration as some tumor-associated mutations could cause such effects and thus could be a biomarker of cisplatin resistance. The PCNA–p21 interaction is integral to numerous processes such as cell proliferation, DNA repair, and adduct bypass by translesion synthesis (32). Exploring these processes may elucidate the underlying mechanisms contributing to the observed enhancement in resistance. Most importantly, it would provide the basis for taking into consideration not only the presence of the mutation but also the phenotypic consequence of the location of the mutation.

It is also important to note that although we explored the effects of p21 mutations in WT p53 systems, almost 50% of MIBCs harbor p53 mutations (7). In tumors harboring p53 mutations, p21-deficient cells might behave differently. There are p53-independent pathways of p21 induction (10), and p53-mutant cell lines have highly variable p21 induction after exposure to DNA-damaging agents (Supplementary Fig. S4). However, simultaneous mutations p53 and p21 may offer other avenues of therapeutic interventions. Indeed, Liu and colleagues have shown that combination of gemcitabine and Chk inhibitors is highly effective in killing bladder cancer cells with p53 and p21 mutations (36). Further exploration into this combination of mutations could elucidate other avenues of effective interventions for MIBCs.

Our data would suggest that tumors harboring proximal CDKN1A truncations might be more likely to respond to chemotherapy because their tumor may not arrest after cisplatin-induced damage. To determine if CDKN1A mutation was associated with response, we evaluated the publicly available Van Allen and colleagues data set in which 50 patients underwent neoadjuvant chemotherapy followed by radical cystectomy (6). However, only four patients harbored a CDKN1A mutation (one patient had two separate frame-shifting mutations). No patients had proximal truncating mutations, so no inferences can be made on this data set. We also evaluated the association between CDKN1A mutation and chemoresponse in 48 patients who underwent whole-exome sequencing as part of a validation study for ERCC2 with the response to chemotherapy (37). This data set did not contain CDKN1A mutations.

Larger data sets of patients treated with chemotherapy whose response status is known would be needed to determine if proximal CDKN1Amutations are clinically relevant. If the relationship between CDKN1A mutation and cisplatin resistance is validated, this candidate biomarker could be used to determine the likelihood of chemoresponse. In the case of predicted chemosensitivity, patients might undergo chemotherapy followed by observation to avoid the morbidity of a life-altering procedure. Importantly, identification of a distal truncation which might generate a retained “hypomorphic” allele might preclude the administration of chemotherapy. Indeed, such biomarker-driven clinical trials are already recruiting (NCT02710734 and NCT03609216).

In light of these findings, future studies could explore correlations between CDKN1A mutations (and their location) and clinical responses to cisplatin. Retrospective analysis of the cisplatin response rate in patients harboring CDKN1A mutation could guide further preclinical, translational, and clinical studies. This would require large data sets considering the frequency of CDKN1A mutation.

W.S. El-Diery reports other from Oncoceutics, Inc. and other from p53 Therapeutics outside the submitted work; in addition, Dr. El-Diery has patents for US5871968A and US8211644B2 issued. P.H. Abbosh reports grants from Natera, DoD, NCI, and BCAN, personal fees from AstraZeneca, Janssen, and ArTara outside the submitted work; in addition, Dr. Abbosh has a patent for urinary DNA detection for urothelial cancer pending. No disclosures were reported by the other authors.

R.K. Sikder: Conceptualization, data curation, formal analysis, investigation, visualization, methodology, writing–original draft, writing–review and editing. M. Ellithi: Formal analysis, investigation, visualization, and methodology. R.N. Uzzo: Data curation, formal analysis, investigation, and methodology. D. Weader: Data curation, validation, investigation, visualization, and methodology. A.L. Metz: Investigation and methodology. A. Behbahani: Investigation and methodology. E.R. McKenzie: Investigation and methodology. W.S. El-Deiry: Conceptualization and supervision. P.H. Abbosh: Conceptualization, resources, formal analysis, supervision, funding acquisition, investigation, methodology, writing–review and editing.

This work has been supported by BCAN Young Investigator Award (P.H. Abbosh), NCI CA218976 (P.H. Abbosh), DoD CA181178 (P.H. Abbosh), Avery Postdoctoral Fellowship (US), the Fox Chase Center Core Grant P30 CA006927 (all authors), NCI CA176289/CA173453 (W.S. El-Deiry), and the Fox Chase Cancer Center Summer Undergraduate Fellowship (D. Weader). We are grateful for generous gifts from Peter and Alice Kriendler, Bucks County Board of Associates, Uplifting Athletes, and Newport Foundation. We acknowledge the administrative assistance of Linda Cathay, Denise Dieter, and Alison Conn, and technical assistance from Michael Slifker, Jim Oesterling, John Krais, and Andrey Efimov.

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.