PARP inhibitor Olaparib overcomes Sorafenib resistance through reshaping the pluripotent transcriptome in hepatocellular carcinoma

Hepatocellular carcinoma is one of the most common human malignancies worldwide with poor prognosis [1]. Currently, the multi-kinase inhibitors Sorafenib was approved by FDA as first line treatment for unresectable advanced HCC. However, the benefit of the patients from the therapy is very limited, with a prolonged median overall survival rate less than 3 months [2, 3]. Thus, further investigation of the molecular mechanisms in drug resistance and development of novel therapeutic strategy is urgently needed.

Increasing evidences suggested that the hierarchy of cancer stem cells and their differentiated progenies contributed substantially to the heterogeneous tumor and therapeutic failure [4, 5]. To better understand the dynamic transcriptomic change during liver development, we have recently established a hepatocyte differentiation model, which specifically induced human embryonic stem cells (hESCs) to differentiate into mature hepatocytes along hepatic lineages [6]. Meanwhile, liver tumors were inoculated into immune deficient mice and treated with Sorafenib to establish a drug resistant model. Combining the transcriptomic data from the two model, PARP1 was identified to be the critical gene which actively expressed in embryonic stem cells and residual tumors after Sorafenib treatment, but progressively decreased along hepatic differentiation. PARP inhibitors have promising effects in inducing synthetic lethality in homozygous recombination deficient tumors in the clinic [7]. Our current study suggested that PARP1 was required for HCC tumor lineage plasticity and residual tumor survival potentially through CHD1L, a chromatin remodeling protein frequently amplified in HCC [8, 9]. Our study revealed a novel mechanism of PARP inhibitors in cancer treatment and further supported their extension to non-HR deficient tumors including HCC.

hESCs were differentiated into human hepatocytes along hepatic lineages. The whole differentiation process was defined with four stages, cells from the four different developmental stages were collected for transcriptomic profiling. (Additional file 1: Fig. S1A). Meanwhile, BALB/c nude mice were subcutaneous injected with PLC-8024 cells and treated with Sorafenib or vehicle control (Additional file 1: Fig. S1B, Additional file 2: file S1). Combining the two set of data together, PARP1 was identified to be significantly up-regulated both in embryonic stem cells and the residual tumors after Sorafenib treatment (Fig. 1a, Additional file 1: Fig. S1C). The expression of PARP1 remains low in distant normal liver tissues, but progressively increased from para-tumor liver tissues to the tumor tissues (Fig. 1b). High expression of PARP1 was detected in 53/196 (27%) HCC patients from a tissue microarray (Additional file 3: Tab S1, Additional file 1: Fig. S1D). Kaplan-Meier analysis indicated that the PARP1 staining score significantly stratified the overall survival (Fig. 1c) and disease-free survival (Fig. 1d) of HCC patients. Furthermore, univariate and multivariate cox regression analysis also proposed the high expression of PARP1 as an independent prognostic factor in HCC (Additional file 4: Tab S2).

PARP inhibitor Olaparib has already been approved by FDA in treatment of BRCA-mutated ovarian cancer and showed promising clinical benefit [10]. Functional assays proved that Olaparib could inhibit HCC cell growth and colony formation ability (Additional file 5: Fig. S2A, S2B). Sphere formation assay indicated a significant decrease in sphere number after Olaparib exposure (Additional file 5: Fig. S2C). A significant reduction of tumor size was also observed after exposes of Olaparib at the dose of 50 mg/kg (Additional file 5: Fig. S2D). Treatment of Olaparib alone or in combination with Sorafenib significantly inhibited cell colony formation in HCC cells with high pluripotency (Hep3B and Huh7) both under normal culture conditions or in the spheres which mimic stem cell microenvironment (Additional file 5: Fig. S2E, S2F, S2G). Similar results were found in cells treated with another PARP inhibitor Niraparib (Additional file 5: Fig. S2H, S2I). Olaparib significantly potentiated Sorafenib drug efficacy (Additional file 6: Fig. S3A and S3B). Western blot analysis further indicated that Sorafenib and Olaparib could induce apoptotic signaling (Additional file 6: Fig. S3C). A significant decrease in tumor size was observed when the mice were treated with Olaparib or Sorafenib alone, and enhanced tumor regression was observed when the mice were treated with both drugs (Fig. 1e, Additional file 6: Fig. S3D, and S3E). In situ TUNEL assay of the xenograft tumors confirmed the elevated level of apoptosis (Fig. 1f). Similar results were also found in HepG2 cells (Additional file 6: Fig. S3F, S3G, S3H, S3I, and S3J). Treatment of Olaparib also significantly potentiated the tumor inhibitory effects of Sorafenib in patient-derived HCC organoids (Fig. 1g).

Transcriptome RNA sequencing was performed in residual tumors after treatment with Olaparib, Sorafenib, and vehicle control. Hierarchical clustering analysis revealed that the tumors treated with Olaparib showed the lowest global gene expression level and was segregated with other subgroups (Fig. 2a). Gene ontology analysis revealed that genes associated with DNA damage repair was up-regulated in HCC tumors treated with Sorafenib. Interestingly, absolute reverse effect was found in HCC tumors treated with Olaparib. This indicated that Olaparib might suppress the DNA damage repair signaling and counteract with the effects of Sorafenib on HCC tumors (Fig. 2b).

The DNA damage repair signaling molecules were significantly up-regulated after Sorafenib treatment but significantly suppressed by Olaparib (Fig. 2c). Similar results were found in Hep3B and Huh7 cells, as well as in PLC-8024 cells treated with another PARP inhibitor Niraparib (Additional file 7: Fig. S4A, S4B, S4C). CHD1L was dynamically expressed during hepatocyte differentiation (Additional file 7: Fig. S4D). Doxorubicin (Dox)-induced CHD1L knockout PLC-8024 cells were established using CRISPR/Cas9 gene editing methods (Additional file 7: Fig. S4E). The DNA damage repair genes were examined in wildtype PLC-8024 cells and cells with CHD1L knock out (sgCHD1L) treated with or without Olaparib. We found the suppression of DNA damage repair genes by Olaparib was totally abolished when CHD1L was deleted (Fig. 2d). Furthermore, both PARP1 and CHD1L was found to occupy the promoter regions of the DNA damage repair genes and associated with open chromatin histone methylation marker H3K4me3. Deletion of CHD1L abolished the binding of PARP1 to the promoters of DNA damage repair genes (Fig. 2e). These findings suggested that PARP1 might form a complex with CHD1L and maintain an “open chromatin” status at the promoter regions of critical DNA damage repair genes.

The residual tumors after Sorafenib might be eliminated with treatment of Olaparib (Additional file 7: Fig. S4F). Genes up-regulated after Sorafenib treatment, but suppressed by Olaparib showed high expression in hESCs but significantly decreased in the differentiated hepatocytes, which was in accordance with the expression pattern of DNA damage repair genes and key pluripotent transcriptional factors (Additional file 7: Fig. S4G, Fig. S4H, and S4I). This indicated that embryonic stem cells might rely on active DNA damage repair to maintain pluripotency, and they might contribute substantially to Sorafenib resistance in HCC treatment. Deletion of CHD1L also decreased the binding of PARP1 and active histone marker H3K4me3 at promoters of pluripotent transcriptional factors (Additional file 7: Fig. S4J). A dose-dependent decrease of pluripotent transcriptional factors was observed after Olaparib treatment (Fig. 2f). The inhibitory effects were further confirmed at protein level in a dose- and time- dependent manner (Fig. 2g, h and Additional file 8: Fig. S5A). Immunohistochemical staining (IHC) also confirmed the inhibition of pluripotency transcriptional factors in xenograft tumors after Olaparib treatment (Additional file 8: Fig. S5B).

The open chromatin architecture could present more nucleosome-free promoter region, which was in turn readily digested by micrococcal nuclease (MNase). Inhibition of PARP1 by Olaparib could reverse such process, and thus exhibited a “protection” effect, leading to reduced digestion and consequently more uncut DNA retained for qPCR detection. After MNase treatment, the number of uncut DNA fragments near the promoter region of target genes were determined by qPCR in cells with or without Olaparib exposure (Fig. 2i). The tumor cells were also treated with different concentrations of MNase to monitor the dynamic chromatin structure change and sensitivity to MNase digestion. The results indicated that PARP1 could sustain the open chromatin structure for SOX2, OCT4, c-MYC. Pharmaceutical interruption of PARP1 by Olaparib could reverse the process and might account for its role in suppressing the DNA damage repair genes, as well as the whole HCC pluripotent transcriptome (Fig. 2j, k, Additional file 8: Fig. S5C).

In conclusion, we found Sorafenib treatment could retain resistant tumor cells characterized with elevated cancer stemness and activation of DNA damage repair signaling. PARP1, which is highly activated in embryonic stem cells and Sorafenib resistant cancer cells, might be responsible for the active transcription of the pluripotent transcriptional factors and DNA damage repair signaling through maintaining an “open chromatin” structure. PARP inhibitor Olaparib extensively suppressed the pluripotent transcriptome through condensation of the chromatin structure and might greatly reinforce Sorafenib in eliminating HCC further in the clinic (Fig. 2l).