Combinatorial Treatment with PARP-1 Inhibitors and Cisplatin Attenuates Cervical Cancer Growth through Fos-Driven Changes in Gene Expression

Cervical cancer continues to be the second leading cause of death in women aged 20 to 39 years despite progress made with vaccination against human papilloma virus, increased screening, and early diagnosis (1, 2). These statistics highlight the need for better therapeutic interventions for cervical cancers. Currently, the standard treatment for early-stage cervical cancer involves surgery, while patients with locally advanced disease are typically treated with a combination of chemotherapy and radiotherapy (3). Metastatic cervical cancer has traditionally been treated with chemotherapeutics, although recent advances have led to the use of more targeted therapies such as bevacizumab and pembrolizumab (4). Overall, most regimes that involve the use of chemotherapeutics use the drug cisplatin (CDDP) in combination with other drugs, such as paclitaxel (2, 5). However, development of chemoresistance is a major hindrance in the continued use of cisplatin (5) and underscores the need to explore alternative drugs with increased efficacy for combinatorial treatment with cisplatin.

Recently, small-molecule inhibitors of nuclear PARPs, primarily PARP-1, have been approved by the U.S. FDA for use as cancer therapeutics, particularly in gynecologic cancers, such as ovarian cancer (6). Four PARP inhibitors (PARPi) have been approved by the FDA for the treatment of ovarian and breast cancers with BRCA1 or BRCA2 mutations (7, 8), but the therapeutic potential of PARPi is likely to extend to other cancer types as well (7, 9). PARP-1 is a member of the PARP family of proteins that catalyze ADP-ribosylation (ADPRylation)—the transfer of ADP-ribose from NAD⁺ to target proteins (10). PARP-1 promotes the synthesis of chains of ADP-ribose called poly(ADP-ribose) or PAR. PARPi are thought to act by inhibiting PARP-1 activity, ultimately inducing synthetic lethality with BRCA1 or BRCA2 mutations in cancers that are deficient in homologous recombination (HR)-mediated DNA repair (11, 12). Recent studies, however, have reported that PARPi may have significant clinical benefits in patients in the absence of BRCA1/2 mutations or HR deficiency (HRD; ref. 7,13–16).

Indeed, a growing body of evidence points towards a critical role for PARP-1 beyond DNA damage repair, particularly in regulating gene transcription and RNA biology (17, 18). In this regard, we have shown that inhibition of PARP-1 catalytic activity alters gene expression programs in a number of biological contexts, including breast cancer cells, adipocytes, and macrophages (19–22). In each of these cases, PARP-1 regulates gene expression through distinct mechanisms (e.g., RNA polymerase II activity, rDNA transcription, chromatin modifications, and transcription factor activity). Together, these data make a strong argument for exploring the different mechanisms of action of PARP-1, and hence PARPi, in cancer, particularly in cervical cancer where relatively little is known about how PARP-1 regulates gene expression.

In this study, we observed that cervical adenocarcinoma cells treated with PARPi are sensitized to cisplatin treatment. Interestingly, this combination treatment promoted a dramatically altered transcriptome, including increased expression of the FOS gene, which encodes the transcription factor Fos. Enhanced expression of Fos in response to PARPi and cisplatin is sufficient to hinder growth in some cervical cancers.

Human cervical cancer cell lines HeLa, Caski, SiHa, ME180, and SW756, as well as primary cervical cells, were purchased from the ATCC. HeLa cells were maintained in DMEM (Sigma-Aldrich, D5796); SiHa in EMEM (Corning, 10009CV); and Caski, ME180, SW756 in RPMI1640 (Sigma-Aldrich, R8758). All media were supplemented with 10% FBS and 1% penicillin/streptomycin. The primary cervical cells were maintained in Cervical Epithelial Cell Basal Medium (ATCC, PCS-480–032) supplemented with the Cervical Epithelial Growth Kit (ATCC, PCS-480–042). Fresh cell stocks were regularly replenished from the original stocks and confirmed as Mycoplasma-free every 3 months using a commercial testing kit.

Cells were pretreated with 10 μmol/L BYK204165 (Santa Cruz Biotechnology, sc-214642) for 30 minutes prior to additional treatments. Cells were treated with the chemotherapeutic agent 2 μmol/L cisplatin (Sigma-Aldrich, P4394) as indicated.

The following antibodies were used for immunoblotting: Flag (Sigma-Aldrich, F3165); snRNP70 (Abcam, ab51266, RRID: AB_882630); GFP (Abcam, ab13970, RRID: AB_300798); Fos (Cell Signaling Technology, 4384, RRID: AB_2106617), and β-tubulin (Abcam, ab6046, RRID: AB_2210370). The custom rabbit polyclonal antiserum against PARP-1 used for immunoblotting was generated using an antigen comprising the amino-terminal half of PARP-1 (ref. 23; now available from Active Motif; 39559, RRID: AB_2793257). The custom recombinant antibody-like anti–ADP-ribose binding reagent (WWE-Fc) was generated and purified in-house (ref. 24; now available from Millipore, MABE103,1 RRID: AB_2665467). For detection of γH2AX lesions, FITC-conjugated phospho-histone H2A.X (Ser139) antibody was used (clone JBW301; Millipore, RRID: AB_568825). For Fos chromatin immunoprecipitation (ChIP)-qPCR, a Fos rabbit monoclonal antibody was used (Cell Signaling Technology, 2250, RRID: AB_2247211).

We used standard molecular cloning techniques to generate the following vectors for expressing or depleting proteins of interest. The human FOS cDNA was cloned into the pINDUCER20 lentiviral vector (Addgene, plasmid no. 44012, RRID: Addgene_44012) with the addition of a C-terminal Flag epitope tag and verified by sequencing. GFP was amplified from pEGFP-N3 (Clontech, RRID: Addgene_62043) and inserted into the pINDUCER20 vector using a Gibson Assembly kit (NEB, E2621).

HeLa cells were transduced with lentiviruses for ectopic expression. We generated lentiviruses by transfection of the pINDUCER20 constructs described above, together with: (i) an expression vector for the VSV-G envelope protein (pCMV-VSV-G, Addgene plasmid no. 8454, RRID: Addgene_8454); (ii) an expression vector for GAG-Pol-Rev (psPAX2, Addgene plasmid no. 12260, RRID: Addgene_12260); and (iii) a vector to aid with translation initiation (pAdVAntage, Promega), into 293T cells using Lipofectamine 3000 Reagent (Invitrogen, L3000015) according to the manufacturer's protocol. The resulting viruses were collected in the culture medium, concentrated using a Lenti-X concentrator (Clontech, 631231), and used to infect HeLa cells. Stably transduced cells were selected with G418 sulfate (Sigma, A1720; 0.4 mg/mL) in cell culture medium.

The following oligonucleotide primers were used for molecular cloning:

• Forward: 5′- GCGGCTAGCATGATGTTCTCGGGCTTCAACGCA-3′

• Reverse: 5′-GCCCTCGAGTCACTTGTCATCGTCATCCTTATAATCCAGGGCCAGCAGCGTG-3′

• Forward: 5′-CTAGCTAGCATGGTGAGCAAGGGCGAGGAGCT-3′

• Reverse: 5′-GGGGAGCTCTTACTTGTACAGCTCGTCCATGCC-3′

The cells were cultured and treated as described above. The cells were then washed with ice cold PBS, collected with ice cold PBS, and pelleted by centrifugation at 1,000 RCF in a microcentrifuge. After collecting the cells, nuclear extracts were prepared according to the Sigma CelLytic NuCLEAR Extraction Kit protocol. Briefly, the cell pellets were resuspended in isotonic buffer (10 mmol/L Tris-HCl pH 7.5, 2 mmol/L MgCl₂, 3 mmol/L CaCl₂, 0.3 mol/L sucrose) supplemented with 1 mmol/L DTT, 250 nmol/L ADP-HPD (Sigma, A0627; a PARG inhibitor), 20 μmol/L PJ34 (a PARPi), and 1x complete protease inhibitor cocktail (Roche, 11697498001), incubated on ice for 15 minutes, and lysed by the addition of 0.6% NP-40 detergent with gentle vortexing. The nuclei from the lysed cells were collected by centrifugation in a microfuge at 11,000 RCF for 30 seconds. The pelleted nuclei were resuspended in Extraction Buffer C (20 mmol/L HEPES pH 7.9, 1.5 mmol/L MgCl₂, 0.42 M NaCl, 0.2 mmol/L EDTA, 25% v/v glycerol, 1 mmol/L DTT, 250 nmol/L ADP-HPD, 10 μmol/L PJ34, and 1x complete protease inhibitor cocktail), and incubated for 20 minutes at 4°C with intermittent vigorous vortexing. The resuspended nuclear material was then clarified by centrifugation at 21,000 RCF in a microfuge for 15 minutes at 4°C. The supernatant was collected as nuclear extract.

Cells were cultured, treated, and collected as described above. The cell pellets were lysed in Lysis Buffer [10 mmol/L HEPES pH 8.0, 2 mmol/L MgCl₂, 1% SDS, 250 units of Universal Nuclease (Pierce), and 1x protease inhibitor cocktail (Roche)] and incubated for 5 minutes at room temperature while being mixed to generate whole cell protein extracts. The extracts were clarified by centrifugation at 21,000 RCF in a microfuge for 15 minutes at 4°C. The supernatant was collected as whole cell extract.

Protein concentrations in the lysates were determined using Bradford reagent (Bio-Rad, 50000006) (for nuclear extracts) and a BCA protein assay (Pierce; for whole cell extracts). The extracts were run on polyacrylamide-SDS gels and transferred to nitrocellulose membranes. The membranes were blocked with 5% nonfat milk in TBST and incubated with the primary antibodies in 1% nonfat milk made in TBST followed by anti-rabbit HRP-conjugated IgG (1:5,000) or anti-mouse HRP-conjugated IgG (1:5,000). Immunoblot signals were detected using an ECL detection reagent (ThermoFisher, 34077, 34095).

Cell survival was quantified using a crystal violet dye-based assay. Cells were seeded as indicated in 6-well plates, grown overnight to ∼50% confluence, and treated with PARPi, cisplatin, or both. The plates were incubated for several days as specified for each experiment. After collection, the cells were washed with PBS and then fixed with 10% formaldehyde for 10 minutes. The fixed cells were stored in PBS at 4°C until all the time points were collected. The fixed cells were stained with 0.1% crystal violet in 20% methanol solution for 30 minutes at room temperature. After washing to remove the excess stain, the crystal violet dye was extracted using 10% acetic acid and the absorbance measured at 595 nm using a spectrophotometer. Surviving fraction was calculated by dividing absorbance values from treated cells with absorbance values from vehicle (DMSO)-treated cells. All experiments were done a minimum of three times with independent biological replicates to ensure reproducibility.

Double stranded DNA breaks were measured by detection of γH2AX lesions using anti–phospho-Histone H2A.X (Ser139) antibody and costaining the cells for DNA content using propidium iodide. For cell-cycle analysis, the cells were stained with propidium iodide. After staining, cells were analyzed using a BD FACSCalibur flow cytometer (BD Biosciences). Data analysis was performed using FloJo data analysis software (Ashland, OR).

Two biological replicates of HeLa cells from each treatment condition were used to isolate total RNA with the RNeasy Plus kit (QIAGEN). Total RNA samples were enriched for polyA+ RNA using Dynabeads Oligo(dT)25 (Invitrogen). Strand-specific RNA sequencing (RNA-seq) libraries were prepared from the polyA+ RNA as described previously (22). The RNA-seq libraries were subjected to quality control analyses (i.e., number of PCR cycles required to amplify each library, the final library yield, and the size distribution of the final library DNA fragments) and sequenced using an Illumina HiSeq 2000. The RNA-seq reads were aligned to reference human genome (hg19) using TopHat (version 1.4.0; ref. 25) and used to determine the steady state levels of the transcripts. We determined significantly regulated genes at 6 hours for the treatment groups: “BYK,” “cisplatin,” and “BYK + cisplatin”. A list of unique oncogenes and tumor suppressors was compiled from Walker and colleagues (26) and Zhao and colleagues (27). Expression values for the “BYK + cisplatin” condition for this unique set of oncogenes and tumor suppressors was obtained from the RNA-seq data to create a cumulative distribution plot using the empirical cumulative distribution function and a custom R script. The RNA-seq datasets generated for this study can be accessed from the NCBI's Gene Expression Omnibus (GEO) repository (www.ncbi.nlm.nih.gov/geo/) using accession number GSE176326.

Gene ontology analyses were performed using the DAVID (Database for Annotation, Visualization, and Integrated Discovery) tool (28). DAVID returns clusters of related ontological terms that are ranked according to an enrichment score. The inputs into the analyses were the sets of BYK + CDDP upregulated and BYK + CDDP downregulated genes from the RNA-seq analyses.

De novo motif analyses were performed on a 2,000 bp region surrounding the gene promoter using the command-line version of MEME (29). The following parameters were used for motif prediction: (i) zero or one occurrence per sequence (-mod zoops); (ii) number of motifs (-nmotifs 12); (iii) minimum, maximum width of the motif (-minw 8, -maxw 15); and (iv) search for motif in given strand and reverse complement strand (-revcomp). The predicted motifs from MEME were matched to known motifs using TOMTOM.

HeLa cells were cultured and treated with DMSO or CDDP + BYK for 12 hours. ChIP was performed as described previously (22) with slight modifications. Briefly, the cells were cross-linked with 1% formaldehyde in PBS for 10 minutes at 37°C and quenched in 125 mmol/L glycine in PBS for 5 minutes at 4°C. Crude nuclear pellets from these cells were sonicated to generate chromatin fragments of ∼300 bp in length. The sonicated samples were incubated overnight at 4°C with a Fos antibody. The samples were washed and the immunoprecipitated genomic DNA was eluted in Elution Buffer (100 mmol/L NaHCO₃, 1% SDS), digested with proteinase K and RNase H to remove protein and RNA, respectively, extracted with phenol:chloroform:isoamyl alcohol, and precipitated with isopropanol. The precipitated ChIPed DNA along with input DNA was collected by centrifugation, air dried, and dissolved in nuclease-free water. Fos enrichment on the ChIPed DNA was then analyzed by qPCR using the primer sets listed below and a LightCycler 480 real-time PCR thermocycler (Roche) for 45 cycles.

• Forward: 5′- CGTGGAAGATCCAGCAGTCC-3′

• Reverse: 5′- GAGAGTTAGAAGGGGCCGGA-3′.

• Forward: 5′- CAACGGCAAGTTAAGAGCCC-3′

• Reverse: 5′- AGAACCACAGCCATCAGCG-3′.

The siRNA oligos used for the knockdown experiments were purchased from Sigma: FOS (SASI_Hs01_00184573) and siRNA Universal Negative Control #1 (SIC001). The siRNA oligos were transfected at a final concentration of 30 nmol/L using Lipofectamine RNAiMAX reagent (Invitrogen, 13778150) according to the manufacturer's instructions. The cells were used for various assays 48 hours after siRNA transfection as indicated.

The RNA-seq datasets generated for this study can be accessed from the NCBI's GEO repository (www.ncbi.nlm.nih.gov/geo/) using accession number GSE176326. All other data generated in this study are available within the article and its supplementary data files. Unique materials used in this study are available upon request from the corresponding author.

To assess the effect of PARP inhibition on cervical cancer cells, we treated cell lines representing the different histologic types of cervical cancer, including adenocarcinoma (HeLa cells), epidermoid (Caski and ME180 cells), and squamous (SiHa and SW756 cells; Fig.1A), as well as primary cervical cells, with the PARP-1-selective inhibitor BYK204165 (BYK; ref. 30; Fig. 1B). Each cell line had different levels of basal ADPRylation, with HeLa, ME180 and primary cells having low levels, and Caski, SiHa, and SW756 cells having high levels (Fig. 1B). The basal ADPRylation in each cell lines was effectively inhibited by treatment with BYK (Fig. 1B), suggesting that PARP-1 is the primary mediator of ADPRylation in these cells.

In cell growth and survival assays, BYK treatment had dramatically different effects on the different cell lines, with both HeLa and Caski cells exhibiting greater sensitivity to high doses of BYK (100 μmol/L) compared with the other cell lines (Fig. 1C). Moreover, BYK treatment inhibited cell-cycle progression in S-phase in HeLa cells (Supplementary Fig. S1A and S1B). In previous studies with cervical and ovarian cancer cells, high ADPRylation levels were associated with greater sensitivity to PARPi (31), an effect not universally observed with the cervical cancer cell lines that we used herein. On the basis of the different sensitivities to PARPi observed in the different cell lines, we decided to explore the therapeutic potential of PARPi in the “BYK sensitive” (HeLa and Caski) versus “BYK resistant” (SiHa, ME180, SW756) cell lines.

We assessed the survival of the cervical cancer cells in response to treatment with BYK, cisplatin, or both in combination (Fig. 2; Supplementary Fig. S2). While cisplatin alone resulted in decreased survival in both HeLa and Caski, cotreating the cells with BYK and cisplatin led to even greater cytotoxicity (Fig. 2,A and B). Interestingly, cotreating the “BYK resistant” cells (i.e., SiHa, ME180, and SW756) with BYK and cisplatin did not enhance the sensitivity to cisplatin (Supplementary Fig. S2A–S2C). These results demonstrate the different sensitivities of the cells to therapeutic mechanisms.

The therapeutic mechanism of action of PARPi has historically been associated with impaired DNA repair responses leading to cell death (18). To assess the effect of the combination of BYK and cisplatin on DNA damage, we treated HeLa cells with either BYK alone, cisplatin alone, or both agents in combination. The accumulation of γH2AX foci was used as a readout for the amount of double-strand DNA breaks (32). We observed that while BYK treatment led to no substantial changes in γH2AX accumulation, treating the cells with cisplatin dramatically increased the number of γH2AX foci (Supplementary Fig. S3A and S3B). Cotreatment with BYK and cisplatin had no additional effect on double-strand break accumulation compared with cisplatin alone (Supplementary Fig. S3A and S3B). These data suggest that while the combination of PARPi and cisplatin induces greater cell death, the effects appear to be independent of enhanced DNA damage. Taken together, these results suggest the possibility of a DNA damage-independent mechanism of action for PARPi in cervical cancer cells.

A number of recent studies have explored the role of PARP-1 in regulating gene expression (19–22, 33, 34). Given these findings, we hypothesized that PARP-1–dependent modulation of transcription could play a critical role in the cytotoxic response of cisplatin-treated HeLa cells to PARP-1 inhibition. To test this, we performed RNA-seq in HeLa cells to determine the genome-wide changes in the transcriptome in the presence of BYK, cisplatin, or BYK + cisplatin (Fig. 3A). We observed altered expression of ∼1,200 genes in response to either BYK or cisplatin alone after 6 hours of treatment, relative to the vehicle control (Fig. 3B). In contrast, we observed a dramatic expansion of the transcriptome, with altered expression of an additional ∼1,500 genes in response to BYK + cisplatin (Fig. 3B). These results indicate that the combined treatment induces a distinct, expanded gene expression program.

To assess the changes in gene expression in greater detail, we focused on alterations in the relative expression of individual genes that are regulated in any one of the treatment conditions compared with the vehicle-treated cells (Fig. 3C). BYK + cisplatin had the most dramatic effect on gene expression, with a distinct set of genes being either significantly up- or downregulated in the cotreated cells (Fig. 3,C and D). On the basis of this observation, we hypothesized that this combination of treatments stimulated the expression of genes whose protein products mediated the increased cytotoxicity in HeLa cells. To test this, we evaluated the ontologies of the genes that were up- or downregulated by BYK + cisplatin (Fig. 3E). We observed that ontological terms related to transcription and cell death were enriched in the upregulated gene set (Fig. 3E, top). In contrast, we observed that ontological terms related to cellular maintenance, including protein localization and transport, were enriched in the downregulated gene set (Fig. 3E, bottom). These results provide additional support for our hypothesis that BYK + cisplatin drives cell cytotoxicity by modulating transcriptional responses in cervical cancer cells.

To identify candidate genes in the upregulated gene set that may drive the alterations in the transcriptional program and the downstream inhibition of cell proliferation, we ranked all the genes upregulated by BYK + cisplatin treatment based on their expression levels. Interestingly, in the set of top-ranked genes, we identified a number of genes associated with cancer-related phenotypes (both tumor suppressors and oncogenes; Fig. 4A). One of the highest ranked genes was FOS, a known oncogene that encodes Fos, a member of the AP-1 transcription factor family (35). We observed that FOS expression is upregulated specifically in the BYK + cisplatin treatment (Fig. 4B). Importantly, Fos protein levels are elevated in the “BYK sensitive” cell lines (HeLa and Caski; Fig. 4,C and D), but not in the “BYK resistant” (SiHa, ME180, SW756, and primary cells; Fig. 4E; Supplementary Fig. S4). On the basis of this, we surmised that inhibition of PARP-1 catalytic activity in the sensitive cell lines in the presence of cisplatin might cause increased cell death by inducing the expression of FOS, leading to an accumulation of Fos protein, that could function as key effector in determined cellular outcomes.

We observed an enrichment of the Fos DNA-binding motif in the gene body or upstream regulatory region (± 2 kb) of the genes upregulated upon cotreatment with BYK + cisplatin (Fig. 5A). Conversely, when we selected for genes that had the Fos-binding motif in the gene body or upstream regulatory region (± 2 kb), we observed greater expression of these genes in the BYK + cisplatin treated condition (Fig. 5,B and C). We validated this by assessing Fos binding to the upstream regulatory region using ChIP-qPCR. Indeed, the Fos protein was enriched at these regions upon BYK + cisplatin treatment as compared with vehicle alone (Fig. 5D). These results suggest that Fos can act as a downstream mediator of responses to BYK + cisplatin treatment by serving as a regulatory transcription factor that acts at the promoters of key target genes to drive a distinct gene expression program.

To determine if elevated Fos protein levels might reduce the survival of cisplatin-treated HeLa cells, we ectopically expressed Fos protein from a FOS transgene in a doxycycline (Dox)-inducible manner (Fig. 6A). Dox-inducible GFP-expressing cells were used as a negative control. The cotreatment of the GFP expressing cells with BYK and cisplatin resulted in increased cell cytotoxicity, thus confirming our previous results. We then evaluated the effect of Fos expression on HeLa cell growth. Ectopic expression of Fos in HeLa cells resulted in decreased cell growth relative to the GFP-expressing control cells across a period of 6 days (Fig. 6B), thus mimicking the phenotype observed with BYK + cisplatin treatments, which act upstream, of FOS to induce its expression. However, we did not observe a significant decrease in cell growth after treating the Fos overexpressing cells with BYK + cisplatin. This corroborates our hypothesis that Fos expression alone is sufficient to drive the cytotoxicity induced by BYK and cisplatin cotreatment. Conversely, upon siRNA-mediated knockdown of Fos, we observed an increase in cell growth upon BYK + cisplatin treatment (Fig. 6C and D). This confirms that Fos is necessary for mediating the BYK + cisplatin-driven cytotoxic effects in cervical cancer.

In conclusion, we have shown that PARP inhibition together with cisplatin treatment induces FOS expression in cervical cancer cells, which in turn causes transcriptional changes that result in diminished cell proliferation (Fig. 6E). These preclinical results provide a compelling argument for the use of a combination therapy of PARPi and cisplatin for treating cervical cancer patients.

To explore potential therapeutic mechanisms that might be useful for treating cervical cancers, we performed a variety of experiments combining a PARPi with cisplatin. We used a number of cell lines that represent different cervical cancer subtypes, as well as primary cervical cells. Surprisingly, we observed that only a few of cell lines were sensitive to PARP inhibition treatment, irrespective of the basal ADPRylation levels. The factors that contribute to these differences remain to be understood. We observed that the PARPi-sensitive cervical cancer cells exhibit greater sensitivity when cotreated with cisplatin. Interestingly, this combination treatment promoted a dramatically altered transcriptome, including increased expression of the transcription factor Fos and subsequent binding of Fos to its target genes. Enhanced expression of Fos in response to PARPi and cisplatin is sufficient to hinder cervical cancer growth. Conversely, the loss of Fos correlated with greater cell growth. These results have revealed a previously unknown mechanism for the regulation of Fos-dependent transcription by PARP-1.

The therapeutic mechanisms of action of drugs are tied to the biological functions of the proteins that they target. The original therapeutic mechanism of action defined for PARPi was the induction of synthetic lethality in cancers that are deficient in HR-mediated DNA repair (11, 12). But, PARP-1 also plays important roles beyond DNA-damage repair, particularly in regulating gene transcription and RNA biology (17, 18). This opens that possibility of therapeutic mechanisms of action for PARPi that extend beyond DNA repair. Indeed, recent studies have reported that PARPi may have significant clinical benefits in patients in the absence of BRCA1/2 mutations or HRD (7, 9, 13–16, 36).

Previous work has shown that in breast cancers, PARP-1 regulates transcriptional outcomes by modifying the activity of (1) RNA polymerase II and its associated factors (37) (2), transcription factors (21, 22), and (3) chromatin (20), and regulating rDNA transcription through the RNA helicase DDX21 (19). Together these examples emphasize the diversity in regulatory mechanisms impacted by PARP-1 and underscore the need to identify the distinct therapeutic mechanisms of action of PARPi in different cellular contexts.

In this study, we discovered a new potential mechanism of action of for PARPi administered in combination with cisplatin, namely enhanced FOS gene expression leading to a dramatic increase in Fos protein levels. We found that a large number of genes whose expression is altered in response to BYK + cisplatin cotreatment are potential targets of Fos. The upregulated genes in this set exhibit an enrichment of the Fos-binding motif, leading us to postulate that Fos might be acting by directly regulating the expression of these genes. The mechanism of Fos-dependent activation of these genes, however, remains to be determined. Together, these results extend our understanding of PARP-1–dependent transcriptional regulation.

Given that FOS is a proto-oncogene, one might expect increased Fos protein levels to be growth-promoting (38). But, Fos and related members of the Fos family are known to inhibit cancer cell growth or promote apoptosis in certain contexts (38), dependent on the relative ratios of the different family members (38). Fos promotes cell-cycle inhibition and cell death in hepatocellular carcinomas (39) and suppresses ovarian cancer proliferation by altering cell adhesion (40). Finally, analysis of data from The Cancer Genome Atlas datasets showed that higher FOS expression levels, along with higher FOS target gene expression, is associated with an increased survival in patients with breast cancer (41). In the context of these previous studies, our results demonstrate how combination therapies can alter signaling pathways that change gene expression patterns to enhance therapeutic benefits.

A number of studies in different cancer types including non–small cell lung cancer, liver cancer, oral squamous cell carcinoma, osteosarcoma, and melanoma (42–46) have assessed the effect of PARP inhibition along with cisplatin treatment. Similar to our observations, in these cancers, PARP inhibition augmented the cytotoxic effects of cisplatin. While some of the studies attributed the increased cytotoxicity to greater DNA damage with PARP inhibition (42, 46), none of them show a DNA damage-independent, transcription-dependent effect of the combinatorial treatment. These data suggest that while different cancers can exhibit similar phenotypic effects to the combination of PARPi and cisplatin, the mechanisms of cell death are variable and can be specific to different cancer types.

Recent work from other groups has also sought to understand the implications of using PARPi in cervical cancer. Treating HeLa cells with PARPi has been shown to induce apoptosis specifically through the modulation of PARP-1 activity (47). HeLa and SiHa cells with hyperthermia also show greater sensitivity when treated with PARPi and cisplatin together. Moreover, treating HeLa cells with PARPi and cisplatin stimulates β-catenin signaling and increases cisplatin-induced cytotoxicity (48). Although the phenotype is similar to the one that we have observed, the mechanism is different. This could be due to the use of different PARPi and their respective specificities. Alternatively, these pathways could be activated concurrently; this possibility needs to be explored further.

A number of clinical trials have explored the use of PARPi in combination with chemotherapeutic agents (49). Indeed, a recently clinical trial tested a regimen of the PARPi veliparib with cisplatin in cervical cancer patients (50). They showed that adding veliparib to the chemotherapeutic regimens was safe and correlated with increased progression-free survival (50). However, this work involved few patients and needs to be corroborated with a larger cohort. Collectively, our data strengthen the rationale for using PARPi with cisplatin as a therapy for cervical cancer.

While we have sought to understand the biological effects and mechanism of action of BYK and cisplatin on cervical cancer cell lines, we do not yet know the effect of this combination treatment in more complex biological settings of cervical cancers, such as xenografts in mice or patient samples. These additional data would provide valuable insight into the efficacy of the cotreatment as a therapeutic approach. We have identified cervical cancer cell types that are both sensitive and resistant to PARPi. Previous studies have shown a direct association between PARPi sensitivity and cellular ADPRylation levels (31). However, in this case, no such correlation is observed and the reason for the differential sensitivities remains to be understood. We have used BYK, a PARP-1 selective inhibitor; however, FDA-approved PARPi, such as Olaparib and Niraparib, are used clinically. Thus, verifying our results in the future with other PARPi will be required to translate these studies to a clinical setting.

W.L. Kraus reports grants from NIH/NIDDK; and grants from Cecil H. and Ida Green Center for Reproductive Biology Sciences Endowment during the conduct of the study; other support from Ribon Therapeutics, Inc.; and grants, personal fees, and other support from ARase Therapeutics, Inc. outside the submitted work; in addition, W.L. Kraus has a patent for U.S. Patent 9,599,606 issued and licensed to EMD Millipore; and W.L. Kraus is a founder, consultant, and member of the scientific advisory board for Ribon Therapeutics, Inc. and ARase Therapeutics, Inc. No disclosures were reported by the other authors.

The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the article; or in the decision to publish the results.

R. Gupte: Conceptualization, data curation, formal analysis, investigation, visualization, methodology, writing–original draft, writing–review and editing. K.Y. Lin: Conceptualization, data curation, formal analysis, investigation, visualization, methodology, writing–original draft. T. Nandu: Software, formal analysis, investigation, visualization, methodology. J.S. Lea: Conceptualization, supervision, funding acquisition, project administration. W.L. Kraus: Conceptualization, resources, supervision, funding acquisition, visualization, project administration, writing–review and editing.

We thank Drs. Sridevi Challa and Cristel Camacho for critical comments on this manuscript. We thank Dr. Shrikanth Gadad for his help with the RNA-seq library preparation. We acknowledge and thank the UT Southwestern Next-Generation Sequencing Core for deep sequencing services (Vanessa Schmid) and the UT Southwestern Flow Cytometry Core for the FACS analysis. This work was supported by a grant from the NIH/NICHD Reproductive Scientist Development Program to K.Y. Lin, and a grant from the NIH/NIDDK (R01 DK058110) and funds from the Cecil H. and Ida Green Center for Reproductive Biology Sciences Endowment to W.L. Kraus.

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.