Article Text

Original research
PARP-1 selectively impairs KRAS-driven phenotypic and molecular features in intrahepatic cholangiocarcinoma
  1. Friederike L Keggenhoff1,
  2. Darko Castven2,
  3. Diana Becker1,
  4. Stojan Stojkovic2,
  5. Jovana Castven2,
  6. Carolin Zimpel2,
  7. Beate K Straub3,4,
  8. Tiemo Gerber3,
  9. Harald Langer5,
  10. Patricia Hähnel6,
  11. Thomas Kindler6,
  12. Jörg Fahrer7,
  13. Colm J O'Rourke8,
  14. Ursula Ehmer9,
  15. Anna Saborowski10,
  16. Lichun Ma11,
  17. Xin Wei Wang11,12,
  18. Timo Gaiser13,
  19. Matthias S Matter14,
  20. Christian Sina15,
  21. Stefanie Derer15,
  22. Ju-Seog Lee16,
  23. Stephanie Roessler17,
  24. Bernd Kaina18,
  25. Jesper B Andersen8,
  26. Peter R Galle1,
  27. Jens U Marquardt2
  1. 1First Department of Internal Medicine, University Medical Center of the Johannes Gutenberg University, Mainz, Germany
  2. 2Department of Medicine I, University Medical Center Schleswig Holstein Campus Lübeck, Lübeck, Germany
  3. 3Institute of Pathology, University Medical Center of the Johannes Gutenberg University, Mainz, Germany
  4. 4Tissue Biobank of the University Medical Center of the Johannes Gutenberg University, Mainz, Germany
  5. 5Cardiology Angiology, University Medical Centre, Mannheim, Germany
  6. 6Department of Hematology, Medical Oncology and Pneumology, University Medical Center of the Johannes Gutenberg University, Mainz, Germany
  7. 7Department of Chemistry, RPTU Kaiserslautern-Landau, Kaiserslautern, Germany
  8. 8Department of Health and Medical Sciences, University of Copenhagen Biotech Research & Innovation Centre, Kobenhavn, Denmark
  9. 9Department of Internal Medicine II, Klinikum rechts der Isar der Technischen Universitat, München, Germany
  10. 10Department of Gastroenterology, Hepatology, and Endocrinology, Hannover Medical School, Hannover, Germany
  11. 11Laboratory of Human Carcinogenesis, National Cancer Institute, Center for Cancer Research, Bethesda, Maryland, USA
  12. 12Liver Cancer Program, Center for Cancer Research, National Cancer Institute, Bethesda, Maryland, USA
  13. 13Institute of Pathology, University Medical Centre Mannheim, University of Heidelberg, Mannheim, Germany
  14. 14Institute of Pathology, University Hospital Basel, Basel, Switzerland
  15. 15Institute of Nutritional Medicine, University Medical Center Schleswig Holstein Campus Lübeck, Lübeck, Germany
  16. 16Department of Systems Biology, The University of Texas MD Anderson Cancer Center, Houston, Texas, USA
  17. 17Institute of Pathology, University Hospital Heidelberg, Heidelberg, Germany
  18. 18Department of Toxicology, University Medical Center of the Johannes Gutenberg University, Mainz, Germany
  1. Correspondence to Professor Jens U Marquardt, Department of Medicine I, University Medical Center Schleswig Holstein Lübeck Campus, Lubeck, Schleswig-Holstein, Germany; Jens.Marquardt{at}


Objective Intrahepatic cholangiocarcinoma (iCCA) is the second most common primary liver cancer with limited therapeutic options. KRAS mutations are among the most abundant genetic alterations in iCCA associated with poor clinical outcome and treatment response. Recent findings indicate that Poly(ADP-ribose)polymerase1 (PARP-1) is implicated in KRAS-driven cancers, but its exact role in cholangiocarcinogenesis remains undefined.

Design PARP-1 inhibition was performed in patient-derived and established iCCA cells using RNAi, CRISPR/Cas9 and pharmacological inhibition in KRAS-mutant, non-mutant cells. In addition, Parp-1 knockout mice were combined with iCCA induction by hydrodynamic tail vein injection to evaluate an impact on phenotypic and molecular features of Kras-driven and Kras-wildtype iCCA. Clinical implications were confirmed in authentic human iCCA.

Results PARP-1 was significantly enhanced in KRAS-mutant human iCCA. PARP-1-based interventions preferentially impaired cell viability and tumourigenicity in human KRAS-mutant cell lines. Consistently, loss of Parp-1 provoked distinct phenotype in Kras/Tp53-induced versus Akt/Nicd-induced iCCA and abolished Kras-dependent cholangiocarcinogenesis. Transcriptome analyses confirmed preferential impairment of DNA damage response pathways and replicative stress response mediated by CHK1. Consistently, inhibition of CHK1 effectively reversed PARP-1 mediated effects. Finally, Parp-1 depletion induced molecular switch of KRAS-mutant iCCA recapitulating good prognostic human iCCA patients.

Conclusion Our findings identify the novel prognostic and therapeutic role of PARP-1 in iCCA patients with activation of oncogenic KRAS signalling.


Data availability statement

All data relevant to the study are included in the article or uploaded as online supplemental information.

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See:

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


  • PARP-1 is involved in multiple oncogenic pathways and processes, including DNA repair, genomic stability, chromatin modification, energy metabolism and apoptosis. Evidence suggests an association between PARP-1 overexpression and KRAS mutations in various cancers, such as acute myeloid leukaemia and colorectal cancer. However, there is limited information available on the impact of PARP-1 expression on therapeutic response in KRAS-mutant intrahepatic cholangiocarcinoma (iCCA).


  • Our results demonstrate that KRAS-mutant iCCA cells become highly sensitive to PARP-1 depletion and inhibition in vitro while Parp-1 deficient mice show reduced cholangiocarcinogenesis in vivo. Mechanistically, PARP-1 effects in KRAS-mutant iCCA are mediated through CHK1 activation, and this process can be reversed by chemical inhibition. Inhibiting PARP-1 in KRAS-mutant tumours leads to a favourable change in prognostic outcome for human iCCA.


  • These findings reveal a novel prognostic and therapeutic role of PARP-1 in iCCA patients with activated oncogenic KRAS signaling and poor prognosis. This knowledge may open new avenues for targeted therapies in this particular subset of iCCA patients.


Intrahepatic cholangiocarcinoma (iCCA) is the second most common primary liver cancer (PLC) with increasing incidence and rising mortality rates.1–3 A profound genetic heterogeneity and diverse spectra of prognostically distinct molecular subgroups render iCCA a prototype for precision oncological approaches.4 5 Several druggable alterations in oncogenic signalling pathways (eg, fibroblast growth factor receptor 2 gene fusions, isocitrate dehydrogenase 1 and 2 mutations) have been identified in iCCA. The application of specific inhibitors became a new mainstay of therapy and significantly improved the overall survival of patients. However, despite great progress in molecularly guided therapy, only 30%–40% of the iCCA patients present with the above-mentioned druggable alterations.6 Thus, therapeutic options remain severely limited for the majority of patients. Among the most prominent alterations identified in iCCA, KRAS mutations historically posed significant challenges for specific targeting.2 7 Activating KRAS mutations have been observed in around 10%–15% of iCCA patients. Importantly, KRAS alterations are characterised by poor response to commonly used chemotherapy and display a reduced overall survival.8 9 Besides direct protumourigenic properties and increased proliferative characteristics, activated RAS-signalling results in increased intracellular stress which renders RAS-driven tumours particularly dependent on non-oncogenic mechanisms that promote oxidative stress response, apoptosis, and, particularly, DNA damage response (DDR).10 11

Several lines of evidence suggest that inhibition of the DDR protein PARP-1 might selectively affect the survival of KRAS-mutant tumour cells of different entities, including colorectal cancer (CRC) as well as acute myeloid leukaemia (AML).12–14 PARP-1 is the most prominent member of the PARP family, involved in several cellular processes such as DDR, genomic stability, chromatin modification, transcription regulation, energy metabolism and programmed cell death. It catalyses poly(ADP-ribosyl)ation (PARylation) on metabolic, oxidative or genotoxic stress and influences the cellular metabolic status by enhancing NAD+ and ATP consumption.15 16 Inhibition of PARP-1 is closely linked to the concept of synthetic lethality, as pharmacologically inactivated PARP-1 is trapped onto the DNA, causing stalling of the replication fork and thereby sensitising DDR-deficient cells.17 Consequently, PARP-1 inhibitors have emerged as a therapeutic option in BRCA1/2-mutant ovarian and breast cancers, as well as in prostate and pancreatic cancer that depend on functional DNA repair mechanisms.18 19 Several clinical trials evaluate PARP-1 inhibition in CCA, however, the mechanistic and therapeutic relevance of PARP-1 and its inhibition in the context of KRAS-mutant iCCA is unknown.20 In the presented study, we aimed to dissect phenotypic and molecular characteristics of PARP-1 in iCCA with a particular focus on KRAS-mutant subgroups known to have a high cellular turnover (thus, constitutively activated DNA repair mechanisms).10 11


PARP-1 expression is abundant in KRAS-mutant iCCA

Given the relevance of deregulated DNA repair mechanisms in cholangiocarcinogenesis (particularly homologous recombination (HR) and non-homologous end-joining (NHEJ)), we sought to investigate the relationship between KRAS mutations and PARP-1 expression (figure 1A).12 13 First, we aimed to investigate a putative association between both molecules in iCCA tissue and analysed the correlation between PARP-1 and KRAS expression in the TCGA patient cohort.21 22 The median expression of PARP-1 and KRAS was significantly upregulated in cholangiocarcinoma (CHOL) (p<0.01) (online supplemental figure S1A-C). Further, pairwise correlation analysis showed a positive correlation (r=0.6, p<0.0001) (online supplemental figure S1D,E) between both genes. Interestingly, while PARP-1 expression was significantly upregulated in both iCCA and HCC, association with KRAS could only be confirmed in iCCA. To confirm these observations, we performed PARP-1 staining using a tissue microarray (TMA) comprising 194 iCCA as well as 54 specimens from normal bile ducts. We confirmed that PARP-1 levels were significantly increased in iCCA versus normal bile duct (online supplemental figure S1F, p<0.0001). In addition, we evaluated the expression of PARP-1 in an independent iCCA cohort including 151 iCCA, 143 surrounding liver and nine normal bile ducts. Consistently, a significant upregulation of the PARP-1 on a transcriptome level was observed in iCCA in comparison with the surrounding liver and normal bile ducts (online supplemental figure S1G). In addition, expression of PARP-1 was positively correlated with genes known to be associated with proliferative capacity (online supplemental figure S1H). Interestingly, expression of PARP-1 also showed a significant association with overall survival and recurrence-free survival based on KRAS status in iCCA (online supplemental figure S1I).

Supplemental material

Figure 1

PARP-1 expression in KRAS-mutant and KRAS-wildtype iCCA and effect of siRNA-mediated knockdown of PARP-1 on cell viability, colony and sphere formation capacity. (A) Graphical representation of the experimental design. (B) Relative PARP-1 expression normalised to normal liver tissue in KRAS-mutant (CCC16, HuCCT1, RBE; red) and KRAS-wildtype iCCA cell lines (CCC33, WITT, HuH28; blue). Mean±SD, n=3, *p<0.05, **p<0.01. (C) Representative Western blot and (D) densitometry analysis of basal PARP-1 protein expression in KRAS-mutant and KRAS-wildtype iCCA cell lines. Relative density of PARP-1 expression normalised to β-actin is shown. Mean±SD, n=3, *p<0.05. (E) Representative Western blot and densitometric analysis of siRNA-mediated knockdown of PARP-1 protein expression in KRAS-mutant (CCC16, HuCCT1, RBE; red) and KRAS-wildtype iCCA cell lines (CCC33, WITT; blue). Relative density of PARP-1 expression normalised to β-actin is shown. Mean±SD, n=3, *p<0.05, **p<0.01, ***p<0.001. (F) Representative images of colony and sphere formation assay as well as % of control after siRNA-mediated PARP-1 knockdown in KRAS-mutant and KRAS-wildtype iCCA cell lines. Mean±SD, n=3, *p<0.05, **p<0.01, ***p<0.001. iCCA, intrahepatic cholangiocarcinoma.

Next, we examined PARP-1 expression in a variety of different KRAS-mutant and KRAS-wildtype iCCA cell lines. Consistently, we found a significant overexpression of PARP-1 in KRAS-mutant iCCA cell lines versus non-mutant cell lines (CCC33 vs CCC16 p=0.0240; WITT vs HuCCT1 p=0.0142; WITT vs RBE p=0.0034; HuH28 vs HuCCT1 p=0.0113; HuH28 vs RBE p=0.0023) (figure 1B). Significant increase of PARP-1 could further be confirmed on protein level (CCC33 vs CCC16 p=0.0286; WITT vs HuCCT1 p=0.0105; WITT vs RBE p=0.0122; HuH28 vs HuCCT1 p=0.0162; HuH28 vs RBE p=0.0234) (figure 1C,D). Together, these investigations support the hypothesis that PARP-1 is preferentially upregulated on RNA and protein level in KRAS-mutant, but not KRAS-wildtype iCCA.

KRAS-mutant iCCAs show preferential sensitivity towards PARP-1 inhibition in vitro

Given the potential association of KRAS mutations and PARP-1 overexpression in iCCA, we next employed selective RNAi knockdown of PARP-1 in the KRAS-mutant and non-mutant iCCA cell lines to further characterise the potential of PARP-1 as a putative therapeutic target in this subgroup of patients. Successful knockdown of PARP-1 protein expression was confirmed by Western blotting overall achieving a 66.6%–83.8% reduction compared with control cells (CCC33 78.5%, p=0.0091; CCC16 71.0%, p=0.0040; WITT 66.6%, p=0.0358; HuCCT1 82.4%, p<0.0001; RBE 83.8%, p=0.0006) (figure 1E). Importantly, PARP-2 protein levels were not affected by the PARP-1 knockdown. Notably, transfection of the KRAS-wildtype iCCA cell line HuH28 was not successful due to the low proliferation rate of this cell line. The impact of PARP-1 knock-down on proliferation was subsequently demonstrated using CFU and SFU. A significant reduction of colony (CCC16 40.5%, p=0.0043; HuCCT1 38.8%, p=0.0243; RBE 40.4%, p=0.0002) and spheroid (CCC16 40.8%, p=0.0041; HuCCT1 41.5%, p<0.0001; RBE 46.7%, p<0.0001) formation capacity ranging from 35.6% to 45.8% was observed in KRAS-mutant versus control cells. In contrast, neither colony nor spheroid formation capacity was affected in KRAS-wildtype iCCA cell lines (figure 1F).

Next, we used the FDA-approved PARP-1 inhibitor olaparib to confirm the association of PARP-1 and KRAS in the context of chemical inhibition. In concordance with the finding from RNAi experiments, KRAS-mutant iCCA cell lines showed a significantly reduced viability in response to PARP-1 inhibition compared with KRAS-wildtype cell lines (CCC33 vs CCC16 p=0.0068; WITT vs HuCCT1 p=0.0078; WITT vs RBE p=0.0026; HuH28 vs HuCCT1 p=0.0725; HuH28 vs RBE p=0.0114) (online supplemental figure S2A). Moreover, olaparib caused a selective G1 cell cycle arrest in KRAS-mutated primary cells (online supplemental figure S2B).

In addition, PARP-1 inhibition also led to a decrease in the number of CFU and SFU in KRAS-mutant iCCA cell lines (CFU: CCC16 50.3%, p=0.0003; HuCCT1 51.6%, p<0.0001; RBE 49.6%, p<0.0001; SFU: CCC16 36.6%, p<0.0001; HuCCT1 38.5%, p=0.0002; RBE 42.1%, p<0.0001). Although olaparib exerted slight effects on colony formation in KRAS-wildtype iCCA cell lines CCC33 (27.6%, p=0.0003) and WITT (8.4%, p=0.0171), the spheroid forming capacity of KRAS-wildtype iCCA cell lines remained unaffected in all cell lines on olaparib treatment (online supplemental figure S2C). Notably, reduction in CFU was considerably less pronounced in KRAS-wildtype versus KRAS-mutant cells, which is presumably due to unspecific toxic effects of olaparib unrelated to PARP-1 inhibition.

To unveil potential synergistic effects between olaparib and cytotoxic compounds used for iCCA therapy, we treated KRAS-wildtype and mutated primary cell lines (CCC16 and CCC33) with olaparib in combination with cisplatin and gemcitabine. Combination therapy showed a significantly higher level of synergism for both tested drugs, gemcitabine and cisplatin, in the KRAS-mutated cell line (online supplemental figure S2D). To further address whether inhibition of other components of the DDR would induce a similar selective response in KRAS-mutated cancers, we tested the efficacy of an additional drug involved in DDR, namely the DNA-PKcs inhibitor KU57788, leading to inhibition of NHEJ. Consistent with a selective effect of PARP-1 inhibition in KRAS-mutated iCCA, effect of KU57788 was independent of the mutational status and no significant differences between KRAS WT and KO cell lines were observed (online supplemental figure S2E). Taken together, these investigations confirm the preferential antitumourigenic effects of PARP-1-based interventions in KRAS-mutant iCCA cell lines and validate the potential utility of combination therapy for the treatment of this subtype of iCCA.

Impact of PARP-1 depletion on transcriptomic profile, DDR and ROS in KRAS-mutant iCCA cell lines

Our functional analyses showed pronounced KRAS-dependent differences in iCCA cell lines on PARP-1-based interventions. To further define molecular alterations influenced by PARP-1 inhibition, we generated stable PARP-1 knockout (PARP-1 KO) clones of the different iCCA cell lines HuCCT1, RBE, WITT and CCC33 using CRISPR/Cas9 followed by RNA sequencing (online supplemental figure S3A).

First, we explored whole transcriptome differences in PARP-1 KO in KRAS-mutant iCCA cell lines compared with their scrambled control clones using Wald’s statistics and revealed a total of 1171 (660 down, 511 up) differentially expressed genes (p<0.05; online supplemental table S1). Accordingly, unsupervised hierarchical cluster analyses and principal component analysis plot confirmed that PARP-1 KO clones of both KRAS-mutant iCCA cell lines display distinct molecular profiles (figure 2A).

Supplemental material

Figure 2

Differential gene expression in KRAS-mutant iCCA cell lines on PARP-1 KO. (A) Unsupervised cluster and PCA of significant genes (p<0.05) of KRAS-mutant iCCA cell lines (HuCCT1, RBE) on PARP-1 KO versus control. (B) Canonical pathways significantly regulated in KRAS-mutant iCCA cell lines on PARP-1 KO versus control identified by IPA. The dashed line indicates the significance threshold of –log (p value >1.3). Shown are z-scores of respective canonical pathways (positive z-score=red/activated, negative z-score=blue/inhibited). iCCA, intrahepatic cholangiocarcinoma; IPA, ingenuity pathway analysis; PCA, principal component analysis.

We also evaluated functional signalling pathways related to the differential response to PARP-1 deficiency and identified that gene sets related to cell cycle control, for example, G1/S checkpoint regulation as well as BRCA1-mediated DNA damage, among others, were significantly dysregulated after PARP-1 KO (figure 2B). In addition, oxidative stress response (NRF-2 mediated oxidative stress response, HIF1α signalling) as well as apoptosis-related pathways (death receptor signalling, apoptosis signalling, Necroptosis signalling, and Myc-mediated apoptosis) were affected by selective depletion of PARP-1 in KRAS-mutant iCCA cells (figure 2B). Further, the cell lines also showed disruption in known oncogenic signalling resembling ERK/MAPK, p53, HIF1α and NF-κB signalling after PARP-1 depletion. Moreover, PARP-1 KO enhanced adverse processes in KRAS-mutant iCCA cell lines (online supplemental figure S4A), known to promote hepatobiliary carcinogenesis. Consistently, gene sets related to DNA repair mechanisms, such as the G2/M DNA damage checkpoint and BER, apoptosis and TGFβ signalling were significantly enriched in KRAS-mutant PARP1 KO cells (online supplemental figure S4B). Overall, transcriptomic analyses indicated a dependency of KRAS-mutant iCCA cells on functional DNA repair mechanisms to compensate for increased oxidative stress, apoptotic stimuli and replicative stress induced by a high cellular turnover and proliferative capacity of KRAS-mutant iCCA cell lines on PARP-1 KO, which might confer the distinct effects in KRAS-mutant iCCAs.

We further evaluated the impact of irradiation and oxidative stress on KRAS-mutant and KRAS-wildtype iCCA cell lines on PARP-1 KO. First, we investigated the impact of irradiation-induced DNA damage in PARP-1 KO clones by measuring phospho-γH2AX foci as a marker for DNA double-strand breaks. Irradiation led to a significantly increased number of DNA double-strand breaks in KRAS-mutant iCCA cell line RBE only in PARP-1 KO clones compared with control conditions (online supplemental figure S5A). In contrast, KRAS-wildtype iCCA cell line CCC33 showed a significantly increased number of DNA double-strand breaks independent of PARP-1 KO status. Assessment of ROS level showed that in KRAS-mutant iCCA cell line RBE the basal ROS levels were significantly higher in the PARP-1 KO clones when compared with their respective scrambled control clones. In contrast, the basal status of ROS did not differ in PARP-1 KO clone of KRAS-wildtype iCCA cell line CCC33. Consistently, H2O2 administration led to a significant increase in oxidative stress in KRAS-mutant iCCA cell lines, whereas levels of oxidative stress in KRAS-wildtype iCCA cell line CCC33 remained unaffected (online supplemental figure S5B).

Mechanism of PARP-1 regulation was further explored in authentic iCCA tumours harbouring KRAS mutation. KRAS-mutated iCCA showed upregulation of genes associated with DNA double-strand break repair mechanisms. In addition, genes associated with HR (eg, BARD1, EXO1, RAD54L, CHEK1, UIMC1, RAD51, BRCA2) and both canonical and alternative NHEJ (c-NHEJ; eg, XRCC4, DCLRE1C; alt-NHEJ; eg, POLQ, LIG1, FEN1, XRCC1) were activated (online supplemental figure S5C). We further tested enrichment of gene sets in KRAS-mutant tumours to test association with PARP-1 and DNA damage control. Analyses revealed enrichment of BRCA1 and HR in the tumours carrying KRAS mutation (online supplemental figure S5D). To substantiate these observations, we performed RPPA analyses of two primary cell lines representing the mutational KRAS status and confirmed that proteins involved in HR are significantly increased in KRAS-mutated cells (CCC16) (CtIP, RAD50 and RAD51) (online supplemental figure S5E). Thus, these investigations underline that major mechanisms of DNA damage repair, including PARP-1, are closely associated with KRAS mutation in iCCA.

In addition, we preformed single-cell analysis of tumours from 11 iCCA patients. These analyses confirmed differential expression of DDR genes in malignant cells from patients with KRAS mutations compared with those cells derived from patients without the mutations. This was not observed in non-malignant cells, suggesting that the phenomenon is tumour cell specific (online supplemental figure S6A-C).

Hepatobiliary carcinogenesis is selectively impaired in Kras/Tp53-driven tumours with genetic Parp-1 deficiency

To dissect the relevance of PARP-1 in KRAS-driven and non-KRAS-mutant iCCA development, we employed HDTV-induced hepatobiliary carcinogenesis in genetically modified Parp-1 proficient and deficient mice. Herein, Kras induction together with Tp53 knockdown (Kras/Tp53) led to preferential occurrence of distinct solid tumours within 10 weeks after injection in Parp-1 proficient mice (online supplemental figure S7A). Histopathological analyses of an expert hepatopathologist revealed tumours with gland-like structures in accordance with the diagnosis of iCCA, most often well differentiated, in one example also with signs of dedifferentiation with sarcomatoid tumour cells including tumour giant cells (figure 3A, online supplemental figure S8A). No mucin production was observed, that is, the tumours resemble iCCA of small-duct type. In addition, Parp-1 proficient mice also displayed multiple small-cell dysplastic foci and nodules with development of early hepatocellular carcinoma and focal tumourous vein invasion (online supplemental figure S8B).

Figure 3

Histology and immunohistochemistry (IHC) of liver sections injected with Kras/Tp53 via HDTV and Akt/Nicd via HDTV and quantification of in vivo tumour growth. (A) Representative images of livers with tumour induction via HDTV (Kras/Tp53; Akt/Nicd) in Parp-1+/+ (n=5/6) and Parp-1−/− mice (n=6). H&E and IHC staining of selected proteins (Sox9, Parp-1, Ki67 and γH2ax) of representative paraffin-embedded tumour sections are shown (3.5 µm). Scale bars indicate 500 µm (×10, H&E) and 1000 µm (×5, IHC). (B) Liver weight/body weight ratio (%) of Parp-1+/+ mice (blue) and Parp-1−/− mice (red) after HDTV of empty vector (EV), Akt/Nicd or Kras/Tp53 plasmid combinations with HSB2. Mean±SD, n=5, **p<0.01, ***p<0.001. (C) Quantification of tumour growth shown as scores: 0=no tumour, 1=small foci/nodules, 2=distinct solid tumour. EV n=5, Akt/Nicd n=5, Kras/Tp53 n=5/6, **p<0.01. (D) Percentage of different lesions in Parp-1+/+ and Parp-1−/− mice with Kras/Tp53 plasmid combination. HDTV, hydrodynamic tail vein.

In striking contrast, animals with Parp-1 deficient genotype were characterised by a complete absence of CCA features. Histopathological assessment revealed multiple, diffuse hepatocellular carcinomas as well as dysplastic nodules with focal tumourous vein invasion (figure 3A). Further, moderate macrovesicular steatosis (11%–50%) and disrupted architecture of liver parenchyma was also observed (online supplemental figure S8A,B). These investigations suggest that the Parp-1 deficiency preferentially inhibits cholangiocarcinogenesis.

Mice with tumours induced by Kras/Tp53 combination showed significant differences in their average body weight between Parp-1+/+ and Parp-1−/− animals with 27.8 g and 29.7 g, respectively (p=0.0234). Further, the average liver weight significantly differed depending on the Parp-1 status with 2.75 g for Parp-1+/+ and 6.10 g for Parp-1−/− mice (p<0.0001), likely due to the presence of tumour nodules (online supplemental figure S7B). Consistently, we observed pronounced differences in liver to body weight (L/B)-ratio and scoring on Kras/Tp53 injection between Parp-1+/+ and Parp-1−/− experimental groups (L/B-ratio p<0.0001, tumour scoring p=0.003) (figure 3B-D).

Interestingly, classic biliary marker (Sox9) confirmed cholangiocarcinoma in tumour sections of Parp-1+/+ mice injected with Kras/Tp53, whereas dysplastic nodules in surrounding liver sections of Parp-1+/+ mice as well as of Parp-1−/− mice were expectedly negative for Sox9 expression, confirming the hepatocellular lineage of these cells (figure 3A). Nuclear Parp-1 expression was detectable in all sections of Parp-1+/+ mice, whereas no Parp-1 expression was determined in Parp-1 deficient liver sections (Parp-1+/+ tumour vs Parp-1−/− dysplastic foci p<0.0001; Parp-1+/+ dysplastic foci vs Parp-1−/− dysplastic foci p<0.0001) (online supplemental figure S9A). As expected, significantly higher proliferation (Ki67) was detected in tumour tissue of Parp-1 proficient mice compared with dysplastic nodules (Parp-1+/+ tumour vs Parp-1+/+ dysplastic foci p<0.0001; Parp-1+/+ tumour vs Parp-1−/− dysplastic foci p<0.0001) (online supplemental figure S9A). Further, DNA damage sites represented by γH2ax positive foci were randomly distributed in both tumour and surrounding liver sections of Parp-1+/+ mice and significantly more pronounced than in Parp-1−/− mice (Parp-1+/+ tumour vs Parp-1−/− dysplastic foci p<0.0001; Parp-1+/+ dysplastic foci vs Parp-1−/− dysplastic foci p=0.009) (online supplemental figure S9A). Taken together, the histopathological assessment of the Kras/Tp53-induced tumours showed differences in tumour entity from the evolution of cholangiocellular carcinoma instead of hepatocellular carcinoma, confirming that iCCA development in Kras-mutant cancers is highly dependent on proficient Parp-1 signalling.

To confirm the relevance of activated Kras in a Parp-1 deficient background, non-Kras-driven iCCAs were induced by myrAkt/myc-tagged Nicd (combination referred to as Akt/Nicd). Average body and liver weight did not differ significantly dependent on the Parp-1 genotype in the Akt/Nicd model (online supplemental figure S7B). Further, animals injected with Akt/Nicd developed tumours and cystic alterations 7 weeks after HDTV with pronounced hepatomegaly as well as steatohepatitis independent of the Parp-1 genotype (figure 3A, online supplemental figure S8A). An expert pathologist classified the tumours as iCCA with well to moderate differentiation grade (G1–2), severe macrovesicular steatosis (>50%) and disrupted architecture of liver parenchyma due to the occurrence of multiple tumour foci (online supplemental figure S8B). Sporadic or unspecific effects of HDTV were excluded in livers of mice injected with empty vector (EV) control (results not shown). In contrast to Kras/Tp53-induced hepatobiliary tumour growth, EV-injected and Akt/Nicd-injected animals showed no differences in morphology, histological classification and quantification (L/B-ratio, tumour scoring) of tumours/livers (figure 3B–D, online supplemental figures S8A,B and S9B). Thus, a selective effect of Parp-1 depletion in Kras/Tp53-driven tumours was confirmed.

Impact of Parp-1 deficiency on the transcriptome of Kras/Tp53-induced cholangiocarcinogenesis via HDTV

To further define the molecular features underlying distinct morphological and histopathological differences in the groups, we performed RNA sequencing of HDTV-induced tumours driven by Kras/Tp53 or Akt/Nicd.

We identified a total of 7661 differentially expressed genes (4577 downregulated, 3084 upregulated) between liver/tumour tissue samples of Kras/Tp53 Parp-1+/+ and Parp-1−/− animals (p<0.05) (figure 4A; online supplemental table S2). Pathways related to cell cycle control, like G1/S checkpoint regulation and G2/M DNA damage checkpoint regulation and ATM signalling were among the prominent molecular changes associated with the phenotype (figure 4B, online supplemental figure S10). Further, the DNA double-strand break repair pathway associated with BRCA1 was predicted to be negatively regulated in Parp-1 deficient animals versus Parp-1 proficient animals. These findings suggest that DDR mechanisms are dysregulated in Kras/Tp53-injected animals in a Parp-1-dependent manner. Further functional networks involved inhibition of oxidative stress pathways (NRF-2-mediated oxidative stress response, HIF1α signalling), whereas apoptosis-related pathways were activated (death receptor signalling) in Parp-1 deficient animals. Kras/Tp53 induced tumours under Parp-1 KO showed enrichment of LXR/RXR, p53, PTEN and HIPPO signalling. Inhibition of pathways was shown in key oncogenic pathways like TGFβ, NF-κB, Notch and ERK/MAPK signalling (figure 4B). Interestingly, several DNA repair pathways were significantly enriched in Parp-1+/+ mice injected with Kras/Tp53. Besides DNA single-strand break repair mechanisms, also double-strand break repair pathways (NHEJ, HR) were enriched suggesting that Parp-1 proficiency is important for DDR and repair mechanisms in Kras-driven hepatobiliary tumourigenesis (online supplemental figure S11A). Activation of key pathways identified at the transcriptomic level was validated by IHC in Kras/Tp53 Parp-1−/− animals (online supplemental figure S10C). These findings are in concordance with our in vitro findings (figure 2, online supplemental figure S4).

Supplemental material

Figure 4

Differential expressed genes after HDTV with Kras/Tp53 in Parp-1−/− versus Parp-1+/+ mice. (A) Unsupervised cluster and PCA plot of significant genes (p<0.05) after HDTV with Kras/Tp53 in Parp-1−/− versus Parp-1+/+ mice. (B) Canonical pathways significantly regulated in tumours induced with Kras/Tp53 in Parp-1−/ versus Parp-1+/+ mice identified by IPA. Dashed line indicated significance threshold of –log (p value >1.3). Shown are z-scores of respective canonical pathways (positive z-score=red/activated, negative z-score=blue/inhibited). HDTV, hydrodynamic tail vein; IPA, ingenuity pathway analysis; PCA, principal component analysis.

To confirm the selective effects of Kras in Parp-1 deficient animals, we analysed transcriptomic profiles of Parp-1 deficient and proficient mice injected with Akt/Nicd. The number of significantly altered genes was considerably less, comprising only 158 (95 down, 63 up) differentially expressed genes (p<0.05) vs 7661 in Kras/Tp53 (online supplemental figures S12 and S13; online supplemental tables S2 and S3). These findings are in concordance with the equivalent tumour growth and histopathological features of Akt/Nicd-induced cholangiocarcinogenesis independent of the Parp-1 genotype (figures 3 and 4, online supplemental figure S12). However, several components of DNA repair pathways of DSB repair (NHEJ, HR) showed upregulated expression in Parp-1 proficient mice injected with Akt/Nicd confirming that the role of Parp-1, although important for DNA damage control, is independent of cholangiocarcinogenesis in Akt/Nicd experimental group.

Supplemental material

Mechanisms of PARP-1 activation in KRAS-mutated iCCA

Transcriptome analyses confirmed preferential impairment of cell cycle regulation, DDR pathways and replicative stress response in human iCCA cells, mouse model, as well as patient samples harbouring KRAS mutation. To elucidate the molecular mechanisms underlying the selective impairment of these pathways in KRAS-mutant iCCA, we first examined the expression of DDR genes in Kras-mutant and wild-type animals with functional Parp-1. Genes associated with HR, c-NHEJ and alt-NHEJ were significantly upregulated. In addition, a dominant activation of Chk1 expression was observed. Interestingly, on Parp-1 inhibition in Kras-mutant animals, the expression of Chk1 was also downregulated, suggesting an important relevance for Chk1 kinase in Kras-mutated iCCA (figure 5A). We further used GSEA in our different datasets to investigate molecular differences in KRAS-mutated and wildtype iCCAs. Commonly enriched set of genes across iCCA models encompassed cell cycle regulation and activation of E2F targets, G2M and spindle checkpoint activation, as well as DNA-damage repair via HR. Importantly, we have consistently observed activation of gene sets associated with CHK1/CHK2, key regulators of the cell cycle and cell survival (figure 5B). We further recognised that CHK1 was highly expressed in iCCA but not normal liver tissue and showed a significant upregulation predominantly in KRAS-mutated cancers (figure 5C). Accordingly, public data also show a significant correlation of CHK1 and PARP-1 suggesting a regulatory network in KRAS-mutated iCCA that might be induced by the high replicative stress in this subgroup of tumours. Significant upregulation of CHK1 was further demonstrated in our KRAS-mutated cell lines and Kras/Tp53 Parp-1+/+ mouse model (figure 5D). As already demonstrated for PARP-1 inhibition, KRAS-mutated iCCA cells were significantly more sensitive to specific inhibition of CHK1 by the selective CHK1 inhibitor Rabusertib (figure 5E). Inhibition of CHK1 also led to a significant downregulation of PARP-1 while combined treatment with olaparib did not induce synergistic effects (figure 5F,G). Gene expression analysis following CHK1 inhibition showed upregulation of CDC25C and downregulation of RAD51 and XRCC2, which could be actors of PARP-1 regulation in KRAS-mutated iCCAs. Interestingly, protein–protein interaction confirmed an association of these proteins with PARP-1 (online supplemental figure S14). Lastly, we tested the effect of Olaparib alone and in combination with rabusertib using an ex vivo model of primary cell lines derived from Parp-1−/− Kras/Tp53 and KrasG12D;Rbdel;Tp53del. As expected, cell lines with Kras mutation and functional Parp-1 showed significantly higher sensitivity to Parp-1 inhibition while the observed synergistic effects in both cell lines were negative. Colony formation analysis in Kras/Tp53 Parp-1−/− after exposure to the IC50 concentration of olaparib unveiled that the number of colonies in the treatment was not significantly different from the control (figure 5H,I). The observed findings confirm that CHK1 has a profound effect on PARP-1 levels in this subtype of highly replicative cancers and contributes to the dependence of KRAS-mutated iCCA on functional PARP-1 signalling.

Figure 5

Molecular mechanisms of PARP-1 regulation in KRAS-mutated iCCA. (A) Expression of DNA damage response genes including Chk1 in Parp-1−/− and Parp-1+/+ mice after Kras/Tp53 injection. Volcano plots are depicted with the log (fold change) of each gene and the –log (p adjusted) was calculated by performing Wald test. Selected genes associated with HR, c-NHEJ and alt-NHEJ are coloured and gene names are displayed. Expression of Chk1 is depicted with red ellipse (B) Gene set enrichment analysis (GSEA) in KRAS-mutated cell lines, Kras/Tp53 Parp-1+/+ mouse model and patient samples. The selection of gene sets was based on statistical significance calculated by nominal p<0.05 and FDR<0.25. NES indicates the degree of overexpression for each group at the peak of the entire gene set. (C) CHK1 gene expression profiles (TCGA) in cholangiocarcinoma versus normal liver tissue (CHOL) presented in box plots and correlation between PARP-1 and CHK1. Values of *p<0.05 were considered as of significant difference. (D) Upper graphs depict gene expression level of CHK1 in KRAS-mutated cell lines and Kras/Tp53 Parp-1+/+/wildtype mouse model, Lower graphs show expression of Chk1 in Kras/Tp53 and Akt/Nicd mouse model with Parp-1+/+ and Parp-1−/− genotype. (E) Shown are IC50 concentrations of Rabusertib for KRAS-mutated and non-mutated primary iCCA cell lines. Mean±SD, n=3, *p<0.05. (F) Representative Western blot and densitometric analysis of CHK1 and PARP-1 protein expression on treatment with IC25 concentration of Rabusertib in KRAS-mutant iCCA cell line. Relative density of PARP-1 expression normalised to β-actin is shown. Mean±SD, n=3, *p<0.05, **p<0.01, ***p<0.001. (G) Evaluation of synergistic or antagonistic effects between Rabusertib/Olaparib in primary human cell line. Plots indicate level of synergism between investigated drugs, where red colour represents synergism and green colour antagonism. (H) Dose-response curves of Kras-mutant mouse cell lines with and without functional Parp-1 treated with increasing concentrations of olaparib (left) and their respective IC50 values (middle). Right is shown total number of colonies with and without olaparib treatment. Mean±SD, n=3, *p<0.05. (I) Evaluation of synergistic or antagonistic effects between rabusertib/olaparib in mouse cell lines. Plots indicate level of synergism between investigated drugs, where red colour represents synergism and green colour antagonism. On the right, graph shows average synergy score (ZIP) for both cell lines. iCCA, intrahepatic cholangiocarcinoma; NHEJ, non-homologous end-joining.

PARP-1 expression as prognostic factor in KRAS-mutant iCCA

To evaluate a potential prognostic impact of our molecular profiles, we integrated our identified in vitro and in vivo transcriptomic profiles with different established prognostic subgroups of PLC (poor and good prognosis).8 Consistently, KRAS-mutant CRISPR/Cas9-mediated PARP-1 KO clones grouped with good prognosis CCA patients, whereas KRAS-mutant control clones recapitulated transcriptomic features of poor prognosis CCA patients (figure 6A). Similarly, integration of in vivo data showed that Parp-1 deficient mice with Kras/Tp53 injection clustered with a good prognosis while Parp-1 proficient mice grouped with poor prognosis (figure 6B). Importantly, mice injected with Akt/Nicd revealed no distinct clustering dependent on the Parp-1 genotype (figure 6B). Overall, these results suggest that PARP-1 depletion in iCCA with activated KRAS mutations leads to a shift from poor to good prognosis.

Figure 6

Integration of in vitro and in vivo transcriptomic data with prognostic subgroups of CCA patients. (A) The graph shows the integration of KRAS-mutant CRISPR/Cas9 PARP-1 KO clones (dark blue) and respective control clones (light blue) with a previously published dataset of 45 CCA patients with good (pink) and poor (yellow) prognosis. (B) Upper graph shows the integration of Parp-1−/ mice (dark blue) and Parp-1+/+ mice (light blue) with Kras/Tp53-induced carcinogenesis with a previously published dataset of 45 CCA patients with good (pink) and poor (yellow) prognosis. Lower graph shows integration of Parp-1−/− mice (dark green) and Parp-1+/+ mice (light green) with Akt/Nicd -induced carcinogenesis with a previously published dataset of 45 CCA patients with good (pink) and poor (yellow) prognosis. CCA, cholangiocarcinoma.


Recently, the PARP family has gained attention in cancer research due to its involvement in various oncogenic pathways and processes (DNA repair, genomic stability, chromatin modification, energy metabolism, apoptosis) mediated by the family members.23 Elevated PARP-1 expression was consistently observed in a variety of solid tumours.24–28 Further, recent evidence suggests an association between PARP-1 overexpression and KRAS mutations in different tumours, including in AML and CRC models.12 13 29 However, until now, only very limited information is available as to the role of PARP-1 expression in the therapeutic response of KRAS-mutant iCCA. Here, we demonstrated an upregulation of PARP-1 expression in KRAS-mutant iCCA tissue and cancer-derived cell lines in comparison to normal intrahepatic bile duct tissue and KRAS-wildtype cell lines (figure 1). We were able to show that this patient subgroup has significant association with decreased overall and recurrence-free survival (online supplemental figure S1). In concordance with recent studies, a significantly higher expression, and positive correlation of KRAS and PARP-1 in cholangiocarcinoma (CHOL) tumour samples in comparison to normal tissue samples (online supplemental figure S1).9 30 31

Our data confirm that knockdown of PARP-1 by RNAi as well as treatment with Olaparib, an effective PARP-1/2 inhibitor, exerted preferential effects in KRAS-mutant cell lines compared with KRAS-wildtype cell lines (figure 1, online supplemental figure S2). A comprehensive meta-analysis of Ku et al recently confirmed reduced proliferation on PARP inhibition in KRAS-mutant MCF10a cells (transfected with mutant KRAS constructs) compared with KRAS-wildtype. The study also implies that KRAS-mutant cells are highly dependent on DDR pathways confirming a synthetic lethality in KRAS-mutant cancers, which aligns with our hypothesis for KRAS-driven iCCA.14

Molecular analyses of iCCA cell lines on PARP-1 KO confirmed that the lack of PARP-1 results in dysregulation of several DNA repair pathways (BER, DSB repair) predominantly in KRAS-mutant iCCA cell lines (figure 1, online supplemental figure S4).12 14 32 This observation supports the hypothesis that KRAS-mutant tumours are more dependent on functioning DDR pathways in general and show preferential impairment of the PARP-1-associated DNA repair pathway alt-NHEJ.13 14 However, stable PARP-1 depletion lacks the effect of PARP-1-DNA trapping caused by PARP-1 inhibition,33 which limits therapeutic predictions in this model. Nevertheless, dysregulation of DDR processes on PARP-1 depletion was explained on the basis of multiple interactions of PARP-1 and PARylation with other DDR factors.34 35 PARP-1 activation was associated with enhanced activation of the error-prone alt-NHEJ pathway over c-NHEJ by competing with Ku70 protein on the site of DNA damage.36 37

In addition to changes in DDR pathways, our transcriptomic data indicate an involvement of apoptotic signalling, inflammatory response and oxidative stress (figure 2, online supplemental figure S4). Further analyses are required to clarify the role of these pathways in the context of tumour initiation and progression and the proposed pro-apoptotic and anti-inflammatory effects of PARP-1 deficiency and PARPi.13 32 38 Hähnel et al showed that KRAS-mutant AML cells are sensitised towards apoptosis on PARP-1 inhibition in combination with the DNA damaging anticancer drug daunorubicin.13 Further, PARP-1 deficiency or PARPi was associated with decreased gene expression of proinflammatory cytokines in previous studies.12 39 Other studies hypothesised that PARP-1 activity might act as a double-edged sword during carcinogenesis, affecting apoptotic and oxidative processes and proinflammatory signalling, dependent on the cellular metabolic status and state of tumourigenesis.12 16 23 40 41 Overall, our in vitro experiments exploring diverse modes of PARP-1 inhibition, including RNAi, and CRISPR/Cas9-mediated PARP-1 KO confirmed that hypersensitivity of KRAS-mutant iCCA cell lines is related to altered tumourigenic properties. These findings propose a novel therapeutic strategy for a previously difficult-to-treat subgroup of progressed iCCA patients harbouring KRAS mutations. Of note, synergistic effects observed between Olaparib and commonly used chemotherapeutic compounds in KRAS-mutant cells indicate that a combination treatment might be of further therapeutic interest and could be pursued in future investigations (online supplemental figure S2D).

To further determine the role of PARP-1 in tumour initiation and cancer development, we employed a model of KRAS-driven cholangiocarcinogenesis in Parp-1 deficient animals.11 42–44 In consistence with previous reports, Parp-1 proficient mice display predominant development of solid iCCA (80%) accompanied by dysplastic hepatocellular foci and nodules as well as early hepatocellular carcinoma in the surrounding liver tissue 10 weeks after HDTV with Kras/Tp53 (figure 3).11 In stark contrast, in Parp-1-deficient mice, Kras-induced carcinogenesis was characterised by the absence of cholangiocarcinoma and the presence of multiple dysplastic foci and nodules as well as small HCC with tumourous vein invasion. Importantly, transcriptomic analyses of our in vivo data suggest a pronounced inhibition of NOTCH signalling (figure 4, online supplemental figures S8 and S12), a key oncogenic driver of iCCA and predicted factor for transdifferentiation of hepatocytes towards biliary traits, is present in our PARP-1-deficient lesions.45 46 Furthermore, Ikenoue et al demonstrated that activated Kras in combination with active Pten results predominantly in HCC, whereas activated Kras in combination with homozygous inactive Pten leads to iCCA development in vivo.47 In line with this, Parp-1 deficiency in our study confirms activation of Pten. Taken together, activation or inactivation of the above-mentioned pathways, although not being exclusively specific to cholangiocarcinogenesis, might explain the absence of iCCA and shift towards HCC development in our Kras/Tp53-induced tumour model under Parp-1 deficiency. Further studies are clearly warranted to define the underling molecular mechanisms responsible for the phenotypic shift.

Consistent with our in vitro data and previous studies with Parp-1 deficient mice, our transcriptomic analyses of in vivo data showed enrichment of DNA repair pathways (alt-NHEJ, HR, BER) in all experimental groups of Parp-1 proficient mice while in Parp-1−/ mice DNA repair pathways were not enriched.32 48 Decreased gene expression of proinflammatory cytokines was already observed in Parp-1−/ mice and on PARPi and might explain the predicted inhibition of proinflammatory pathways (NF-κB, TGFβ) in our data (figure 4, online supplemental figures S10 and S12).12 39 Importantly, neither histopathological assessment nor quantification of tumour burden or molecular analyses revealed a dependency on the Parp-1 genotype in a non-Kras-driven iCCA model (Akt/Nicd) (figures 3 and 4; online supplemental figures S12 and S13).45 The results clearly underscore the selective relevance of oncogenic Kras in Parp-1 deficient background. Taken together, our in vivo studies confirm the association of Parp-1 expression in Kras/Tp53-driven cholangiocarcinogenesis. Reduced cholangiocarcinoma development under Parp-1 deficiency could be mechanistically explained by dysregulation of apoptotic and inflammatory processes as well as DDR.

It is well established that KRAS-mutated cancers exhibit enhanced replicative stress and constitutively activated stress response pathways.49 Here, CHK1-mediated cell cycle checkpoints maintain genomic integrity and, thus, CHK1 activation might confer key protective mechanisms that protect KRAS-driven cancers from this therapeutic liability. Consistently, we could provide evidence that selective induction of PARP-1 in iCCA could be induced by activation of CHK1 in KRAS-mutant iCCA (figure 5). We observed a significant upregulation of CHK1 in KRAS-mutated iCCA cell lines and Kras/Tp53 Parp-1+/+ mice while inhibiting Parp-1 in Kras/Tp53 Parp-1−/− mice resulted in Chk1 inhibition, as observed in human patients. Furthermore, KRAS-mutated iCCA cells were significantly more sensitive to CHK1 inhibition by Rabusertib, a selective CHK1 inhibitor, than KRAS-wildtype cells or mouse cells deficient in Parp-1 (figure 5). Inhibition of CHK1 significantly influenced components of HR (online supplemental figure S14). Importantly, inhibition of CHK1 also led to a significant reduction of PARP-1 levels in KRAS-mutant. When rabusertib was combined with Olaparib, there was no induction of synergistic effects, further implying a potential regulatory role of CHK1 (figure 5). These investigations are in line with several other recent investigations that established a formal association between CHK1 and PARP-1 in highly replicative cancers that rely on a functional DDR, particularly one mediated by HR.50 51 Studies have also described a direct interaction between CHK1 and PARP-1 at the DNA damage site and subsequent promotion of HR.52

Thus, we propose a regulatory axis involving CHK1-mediated PARP-1 activation induced by replicative stress in KRAS-mutant iCCA that maintains cellular survival in this subtype of iCCA.

To validate our findings in authentic human tumours, we integrated our in vitro and in vivo findings with a publicly available dataset of well-characterised CCA patients.8 Consistently, transcriptome profiles of the PARP-1-deficient cell lines as well as control cell lines with different established prognostic subgroups of PLC (poor and good prognosis) revealed a shift of KRAS-mutant iCCA cell lines from poor to good prognosis on PARP-1 KO (figure 6). In concordance, Parp-1 deficient samples from mice injected with Kras/Tp53 clustered with good prognosis iCCA patient samples, whereas Parp-1 proficient samples grouped with poor prognosis. Interestingly, this prognostic shift was not visible in Akt/Nicd-induced tumour samples, which clustered randomly together with poor prognostic patients. Hence, the obtained data clearly confirm a clear synthetic vulnerability and provide a rational for therapeutic targeting of PARP-1 in KRAS-mutant iCCA patients.


In conclusion, the study presented confirms upregulation of PARP-1 and hypersensitivity towards PARP-1-based intervention preferentially detectable in KRAS-mutant iCCA in vitro and in vivo. Mechanistically, our data revealed that PARP-1 inhibition provoked downregulation of DSB repair pathways and inhibition of oxidative and inflammatory processes.13 Thus, our data open novel therapeutic options for this difficult-to-treat iCCA subgroup that warrant further clinical investigations (online supplemental visual abstract).

Supplemental material

Data availability statement

All data relevant to the study are included in the article or uploaded as online supplemental information.

Ethics statements

Patient consent for publication


JUM and FLK thank Monika Herr for excellent technical support. Sorting of CRISPR/Cas9 clones was performed with the support of the Core Facility Flow Cytometry (CFFC) of the PaulKlein Center for Immune Intervention (PZKI), University Medical Center, Mainz. Tissue staining were provided by the tissue bank of the University Medical Center, Mainz in accordance with the regulations of the tissue biobank and the approval of the ethics committee of University Medical Center, Mainz. Aspects of this article are part of the doctoral thesis of FLK.


Supplementary materials


  • X @jaboeje

  • FLK and DC contributed equally.

  • Contributors Conceptualisation: JUM and FLK; Data curation: FLK, DC, DB, SS, JC, CZ, BKS, TG, CJO'R, CS, SD, UE, AS, LM, XWW, SR, JBA, PRG and JUM; Resources: PH, TK, JF, UE, TG, MSM and BK; Formal analysis: JUM, FLK, DC, DB and BKS; Writing: JUM, FLK and DC. Guarantor: JUM; All authors discussed the results and critically commented on the manuscript.

  • Funding JUM is supported by grants from the Wilhelm Sander Foundation (2021.089.1) and the Volkswagen Foundation (Lichtenberg program). SR was supported by Deutsche Forschungsgemeinschaft (DFG, Project-ID 314905040) within SFB/TRR 209 Liver Cancer, by Eurostars (grant E! 113707, LiverQR) and by the German Cancer Aid (no. 70113922). XWW is supported by grants (ZIA BC010313 and ZIA BC011870) from the Intramural Research Program of the Center for Cancer Research of the National Cancer Institute.

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.