Dual inhibition of VEGF and PARP suppresses KRAS-mutant colorectal cancer.

The addition of bevacizumab to chemotherapy has prolonged overall and progression-free survival rates for metastatic colorectal cancer (mCRC). However, KRAS-mutant (KRAS-mut) CRC, lacking an ideal targeted agent, represents an inferior-response subgroup of patients. In the present study, we investigated a combination approach of bevacizumab + olaparib in KRAS-mut CRC in a preclinical setting. The combined therapy effectively prevented tumor growth in a KRAS-mut cancer cell-derived xenograft model, although this effect was not observed in vitro. Under bevacizumab treatment, we detected intratumor hypoxia and impaired homologous recombination repair (HRR), accompanied by vascular regression. We explored the underlying mechanism of this combined therapy by mimicking a hypoxic condition in vitro using cobalt chloride (CoCl2). The results showed that hypoxia impairs HRR and therefore sensitized KRAS-mut CRC cell lines HCT-116, SW620, and Lovo to olaparib. Furthermore, under this hypoxic condition, olaparib could arrest the cell cycle in the G2/M phase, increase DNA damage and dramatically induce cell apoptosis in KRAS-mut CRC cells. Taken together, these results indicated that the combination of bevacizumab + olaparib could be a potential therapeutic approach in a KRAS-mut CRC cohort.

Introduction

Colon or rectal (colorectal) cancer (CRC) is the third-leading cause of cancer-related mortality [1]. At the time of diagnosis, 30% of CRC patients present metastases [2], and there is an even higher metastasis rate of 55.9% in KRAS-mutant (KRAS-mut) CRC patients [3]. Metastatic CRC (mCRC) patients with KRAS have poorer progression-free survival (PFS) and overall survival (OS) rates than patients without KRAS mutations [4]. In addition, KRAS mutation is associated with resistance to anti–epidermal growth factor receptor (EGFR) therapy [5], [6], [7], [8], [9] and increases the risk of recurrence and death in CRC [10], [11].

KRAS mutations are frequently detected in pan-cancers, including 97.7% of pancreatic ductal adenocarcinomas, 30.9% of lung adenocarcinomas, and approximately 44.7% of colorectal adenocarcinomas [12]. Ras mutations render Ras persistently guanosine-5′-triphosphate (GTP)-bound and constitutively active, and thus permanent Ras signaling is activated [13]. However, due to the picomolar binding affinity of small GTPases for GTP, millimolar GTP cellular concentrations, and lack of amenable deep pockets in Ras, high-affinity small-molecule antagonists have not been feasible [14], [15]. Some attempts to target Ras downstream effectors—e.g., the rapidly accelerated fibrosarcoma (RAF)/mitogen-activated protein kinase kinase (MEK)/extracellular signal-regulated kinase (ERK) and phosphatidylinositol-4,5-bisphosphate 3-kinase (PI3K)/protein kinase B (Akt)/mammalian target of rapamycin (mTOR) signaling pathways [16], [17]—have been unsuccessful due to poor efficacy in KRAS-mut cancers [18], [19], [20], [21]. KRAS-mut malignancies therefore remain refractory, and no therapeutic agent directly targeting Ras has been approved for clinical use so far.

Bevacizumab, as a humanized monoclonal antibody that inhibits vascular endothelial growth factor A (VEGF-A), has been approved for the treatment of patients with mCRC [22]. In both first- and second-line settings, bevacizumab in combination with cytotoxic chemotherapy dramatically improved OS and PFS rates of mCRC patients [23], [24], [25], [26]. However, although it significantly inhibits tumor growth by preventing formation of new blood vessels, bevacizumab might lose its therapeutic effect due to drug resistance [27], [28], [29]. Intratumor hypoxia induced by anti-angiogenic treatment could be an important resistant mechanism to bevacizumab that further enhances tumor growth and invasion [30], [31], [32], [33]. On the other hand, it is well attested that hypoxia suppresses HR by downregulating HR repair (HRR) proteins, such as BRCA1 and RAD51, hence increases sensitivity to PARP inhibition [34], [35], [36]. Therefore, we hypothesized that bevacizumab induced-tumor hypoxia might increase sensitivity to PARP inhibitor, thus overcome bevacizumab resistance.

Bevacizumab + olaparib has been clinically confirmed as efficacious and safe in advanced solid tumors [37]. A recent phase 3 trial confirmed that the addition of maintenance olaparib to bevacizumab provided significant PFS benefit in patients with advanced ovarian cancer, regardless of breast cancer gene (BRCA) mutation status [38]. Olaparib is one of the typical poly (adenosine diphosphate [ADP]–ribose) polymerase (PARP) inhibitors that have been approved in advanced epithelial ovarian or metastatic breast cancer patients with deleterious or suspected deleterious germline or somatic BRCA mutations (gBRCAm or sBRCAm, respectively) [39]. PARP repairs DNA single-strand breaks (SSBs) through the base excision repair (BER) pathway [40]; PARP inhibitors compromise this pathway by blocking PARP enzymatic activity, resulting in the conversion of SSBs to double-strand breaks (DSBs) during DNA replication. Normally, DSBs are repaired via the homologous-recombination (HR) pathway, which involves proteins such as BRCA1, BRCA2, and RAD51 [40]. However, whether such a combination could be applied in KRAS-mut CRC, and the underlying mechanism by which it would work, remain unclear.

Based on all of the above, we proposed that a PARP inhibitor combined with an anti-angiogenic agent might have a synergistic effect in KRAS-mut CRC. Our results confirmed that bevacizumab treatment induced vascular regression and intratumor hypoxia, which led to HR deficiency and thus sensitized cancer cells to the PARP inhibitor olaparib.

Materials and methods

Cell cultures and reagents

We purchased 5 human KRAS-mut cell lines (HCT116, SW620, Lovo, HT29, and SW480) from the American Type Culture Collection (ATCC; Manassas, Virginia, US). Tumor cell culture was maintained in Roswell Park Memorial Institute (RPMI)-1640 medium supplemented with 10% fetal bovine serum (FBS), penicillin (100 U/mL), and streptomycin (50 mg/mL) in a humidified CO2 incubator at 37 °C. All of the cells were passaged for <3 months before renewal from frozen, early-passage stocks. We obtained olaparib and bevacizumab from Selleck Chemicals (Houston, Texas, US) and cobalt chloride (CoCl2) from Sigma-Aldrich (Saint Louis, Missouri, US). Olaparib was dissolved in dimethyl sulfoxide (DMSO) to a final concentration of 50 mmol/L; CoCl2 was dissolved in double-distilled water to a concentration of 100 mmol/L and stock. Both were then stored at −20 °C.

Cell viability assay

Using the 3-(4,5-dimethylthiazol-2yl)-2,5-diphenyltetrazolium bromide (MTT) dye reduction method [41], we measured cell proliferation. Briefly, tumor cells were plated at a density of 2 × 103 cells in 100 µL RPMI-1640 plus 10% FBS per well in 96-well plates and incubated for 24 h. Next, we added bevacizumab, olaparib, or CoCl2 to each well and continued incubation for another 72 h. Then, 50 μL stock MTT solution (2 mg/ml; Sigma-Aldrich) was added to all wells, and the cells were incubated for 2 h at 37 °C. We then removed the media containing the MTT solution and dissolved the dark-blue crystals by adding 100 μL DMSO. Absorbance was measured with an MTP-120 microplate reader (Corona Electric, Corona, New York, US) at test and reference wavelengths of 490 and 550 nm, respectively.

Antibodies and Western blot analysis

We separated protein aliquots (20 μg) on sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) gel (Bio-Rad, Hercules, California, US) and transferred them to polyvinylidene difluoride (PVDF) membranes (Bio-Rad). The membranes were washed 3 times and incubated with blocking solution for 1 h at room temperature (RT). After washing, we incubated the membranes overnight at 4 °C with primary antibodies against Poly (ADP-Ribose) Polymer (PAR) antibody (ab14460; Abcam, Cambridge, UK), hypoxia-inducible factor 1 alpha (HIF-1α; 20960-1-AP; Proteintech, Wuhan, China), and β-actin (Cell Signaling Technology [CST], Danvers, Massachusetts, US). Next, we washed and incubated the membranes for 1 h at RT with species-specific horseradish peroxide (HRP)-conjugated secondary antibodies. Finally, using an enhanced-chemiluminescent (ECL) substrate (Thermo Fisher Scientific, Waltham, Massachusetts, US), we visualized immunoreactive bands. All of the data are representative of 3 independent experiments.

Reverse-transcription and quantitative real-time PCR (qRT-PCR)

We isolated total ribonucleic acid (RNA) from cells or tumor tissues using RNAiso Plus (TaKaRa, Dalian, China) and reverse-transcribed it into complementary DNA (cDNA) using PrimeScript RT Master Mix (TaKaRa) per manufacturer’s instructions. Quantitative real-time PCR (qRT-PCR) was performed using a Roche LightCycler 480 system (Roche Diagnostics, Meylan, France) with SYBR Premix Ex Taq (Tli RNase H Plus; TaKaRa). Oligonucleotide primers used for RAD51 and glyceraldehyde 3-phosphate dehydrogenase (GAPDH; internal control) were as follows: RAD51: 5′-CAACCCATTTCACGGTTAGAGC-3′ (sense); 5′-TTCTTTGGCGATAGGCAACA −3′ (antisense); GAPDH: 5′-AGAAGGTGGGGCTCATTTG-3′ (sense); 5′-AGGGGCCATCCACAGTCTTC-3′ (antisense). All reactions were performed in triplicate for each sample. Cycle threshold (Ct) values of RAD51 cDNA were normalized to GAPDH using the −2ΔΔCt method.

Apoptosis and cell cycle analyses

We seeded cells at a density of 2 × 105 cells/well into 6-well plates in RPMI-1640 medium with 10% FBS. After incubation for 24 h, we added various reagents to each well and continued incubation for another 72 h, after which we harvested cells and washed them once with phosphate-buffered saline (PBS). Apoptosis was measured with an Annexin V-Fluorescein Isothiocyanate (FITC)/Propidium Iodide (PI) Cell Apoptosis Detection Kit (TransGen Biotech Co., Ltd., Beijing, China) per manufacturer’s protocols. Cell cycle arrest was measured with a Cell Cycle Staining Kit (Hangzhou Multi-Sciences Biotech Co., Ltd., Hangzhou, China) per manufacturer’s protocols. We performed both analyses using a FACSCalibur using CellQuest software (BD FACS Aria; BD Biosciences, Franklin Lakes, New Jersey, US). All of the experiments were performed at least 3 times.

Subcutaneous xenografts in BALB/c-nu/nu nude mice

We injected suspensions of 5 × 106 HCT116 cells subcutaneously into the right hind limbs of 5- to 7-week-old female BALB/c-nu/nu nude mice, which we purchased from the Experimental Animal Center of Southern Medical University (Guangzhou, China; N = 24). When tumor volume reached approximately 150 mm3, calculated as:we randomized the mice into 4 groups that received oral olaparib (50 mg/kg/d; n = 6), intraperitoneal bevacizumab (5 mg/kg/twice weekly; n = 6), olaparib + bevacizumab (n = 6), or vehicle control (n = 6). Tumor volume and mouse body weights were measured twice weekly. All of the animal experiments were carried out in strict accordance with the principles and procedures approved by the Committee on the Ethics of Animal Experiments of Southern Medical University (Guangzhou, China).

Immunohistochemistry (IHC)

We performed IHC staining on tissues (5 mm thick) harvested from the xenografts. Tissue sections were fixed overnight in 4% formaldehyde and then embedded in paraffin for IHC. After deparaffinization and hydration, we pretreated the sections for 20 min with 10 mmol/L sodium citrate buffer in a microwave for antigen retrieval. Next, the tissue sections were incubated overnight at 4 °C with antibodies against Ki-67 (CST), Cluster of Differentiation 31 (CD31; Abcam), and HIF-1α (Proteintech). After incubation with a secondary antibody and treatment with the Histofine Simple Stain MAX-PO Kit (Nichirei, Tokyo, Japan), we visualized peroxidase activity via a 3,3′-Diaminobenzidine (DAB) reaction. The sections were counterstained with hematoxylin. Finally, all of the sections were stained with hematoxylin and eosin (H&E) for routine histological examinations.

Quantification of IHC results

For Ki-67 and CD31 IHC staining, we selected the 5 areas per section showing the highest levels of staining intensity for histological quantification by light microscopy at 200 × magnification. The IHC score of HIF-1α was the product of the staining intensity and the percentage of positive cells. Staining intensity was graded on a scale of 0 to 3+ (0 for colorless, 1+ for pale yellow, 2+ for brown–yellow, and 3+ for saddle brown). Percentage of positive cells was scored on a scale of 0 to 4+ (0 for negative, 1+ for <10% positive cells, 2+ for 10–50% positive cells, 3+ for 51–75% positive cells, and 4+ for >75% positive cells). All of the results were independently evaluated by 2 authors (Longhui Zhong and Rong Wang).

Immunofluorescence staining

HCT116, Lovo and SW620 cells were plated overnight in chambered slides (1 × 104 cells/chamber) (Idibi, Germany). Cells were treated with or without CoCl2 (100 µM) and/or olaparib (30 µM) for 24 h. Cells were washed with PBS and fixed in 4% paraformaldehyde at room temperature for 20 min. Then, Cells were permeabilized (0.3% Triton X-100 in PBS, 10 min) and blocked in 10% goat serum at room temperature for 1 h. Cells were incubated overnight at 4 °C with primary antibody for RAD51 (1:200 dilution) (Proteintech, Wuhan, China) or γ-H2AX -Ser139 (1:200 dilution) (CST, Danvers, MA). After three times washing with PBS, cells were incubated with Alexa Fluor 488-labeled goat antirabbit IgG (H + L) antibody (1:500 dilution) for 1 h at RT. Nuclear was stained with DAPI (Beyotime, Shanghai, China) for 5 min at RT.

For tumor tissue, tissue sections were fixed overnight in 4% formaldehyde and then embedded in paraffin. After deparaffinization and hydration of 4-μm-sliced sections, heat-induced antigen retrieval was performed. The primary antibody (anti-RAD51) was applied and incubated overnight at 4 °C. After Alexa Fluor 488-labeled goat antirabbit IgG (H + L) antibody (1:500 dilution) (Beyotime, Shanghai, China) was applied, slides were stained with DAPI (Beyotime, Shanghai, China) for 5 min.

Images from random fields were acquired with an OLYMPUS BX63 fluorescence microscope (OLYMPUS, Tokyo, Japan). For quantification of RAD51 foci and γ-H2AX foci, at least 100 cells from each group were visually scored. Cells showing more than 3 foci were counted as positive for γ-H2AX or RAD51.

Statistical analysis

Data from the tumor progression of xenografts are expressed as means ± standard error (SE), and data from other experiments are expressed as means ± standard deviation (SD). Comparisons between 2 groups were analyzed using Student’s t test, those between >2 groups using 1-way analysis of variance (ANOVA). P < 0.05 (2-sided) was considered statistically significant. All of the statistical analyses were performed using SPSS software version 20 (IBM Corp., Armonk, New York, US).

Results

No combination benefit was observed in the in vitro culture system of KRAS-mut colon cancer cells

Because it blocks VEGF-related angiogenesis, bevacizumab in combination with chemotherapy was approved by the US Food and Drug Administration (FDA) for the treatment of mCRC [42]. However, in our study, bevacizumab did not affect the viability of KRAS-mut colon cancer cells, even at a high concentration in an in vitro culture system (Supp. Fig. 1A). This result was consistent with that from a previous study in which bevacizumab blocked the binding of VEGF-A to endothelial cells via VEGF receptors (VEGFRs) during the process of pathological angiogenesis in the tumor microenvironment but did not directly inhibit the survival of tumor cells [43].

We next examined the effect of olaparib on the viability of KRAS-mut colon cancer cells. Under our experimental conditions, olaparib inhibited cell viability in a drug concentration–dependent manner (Supp. Fig. 1B). However, the presence of bevacizumab for 72 h did not influence sensitivity to olaparib in the cell lines HCT116, SW620, and Lovo (Supp. Fig. 1C).

Taken together, these findings suggested that olaparib had a dose-dependent effect on KRAS-mut colon cancer cells and that no additional inhibition could be obtained by combining it with bevacizumab in vitro. Therefore, combinational benefits should be further evaluated in vivo.

Bevacizumab + olaparib shrank KRAS-mut tumors in vivo

To verify whether olaparib affected the efficacy of bevacizumab against KRAS-mut colon cancer cells in vivo, we established subcutaneous-tumor models using HCT116 cells and the mice with olaparib, bevacizumab, or a combination thereof. As shown in Fig. 1A, continuous bevacizumab + olaparib treatment resulted in markedly greater tumor regression than treatment with either therapy alone. We observed no significant weight loss in any group of mice during the 20 days of treatment (Supp. Fig. 2). These data clearly indicated that bevacizumab improved the response of KRAS-mut tumors to olaparib treatment without significant toxicity.

Fig. 1

Effects of bevacizumab and olaparib on human KRAS-mut CRC in vivo. (A) HCT116 cells (5 × 106 cells) were inoculated subcutaneously into nude mice, which then received oral olaparib (50 mg/kg/d; n = 6), intraperitoneal bevacizumab (5 mg/kg/twice weekly; n = 6), olaparib + bevacizumab (n = 6), or vehicle control (n = 6). Tumor size was measured twice weekly, and tumor volume was calculated as described in the “Materials and Methods” section. Error bars = SE. (B) Representative images of H&E staining and HIF-1α, Ki-67, and CD31 immunostaining of subcutaneous tumors established by injection of HCT116 cells (day 19). Scale bar = 100 μm. (C) Quantification of hypoxic cells in subcutaneous tumors, shown as IHC scores for HIF-1α. The data shown represent the mean ± SD of 5 areas. (D) Left: representative images of immunofluorescence staining showing the RAD51 foci in subcutaneous tumors. RAD51 and nuclei are shown in green (Alexa Fluor 488) and blue (DAPI), respectively, in the merged images. Scale bar = 20 μm. Right: quantification of the RAD51 foci formation. Data shown represent the mean ± SD of 3 areas. (E) qRT-PCR analysis of RAD51 expression levels in the isolated tumor tissues; n = 3. Relative mRNA expression was normalized to GAPDH. (F) Quantification of proliferating cells in subcutaneous tumors, shown as the percentage of Ki-67-positive cells. (G) Quantification of tumor-associated endothelial cells in subcutaneous tumors, shown as the number of CD31-positive vessel fields. Data shown represent the mean ± SD of 5 areas. N.S. = non-significant. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 for the indicated comparisons by 1-way ANOVA. Cont = Control; Olap = Olaparib; Beva = Bevacizumab.

Bevacizumab damaged HRR via hypoxia, which increased the efficacy of olaparib in a subcutaneous-tumor model of KRAS-mut colon cancer cells

To explore whether bevacizumab could induce hypoxia and damage to HR, we confirmed our hypothesis using in vivo tumors. Via IHC staining for HIF-1α, we observed an obvious hypoxic area in the subcutaneous tumors treated continuously with bevacizumab (Fig. 1B and C). We next examined HR ability after bevacizumab or combination therapy in tumors using RAD51 focus formation experiment, since RAD51 foci that are microscopically visible are believed to represent sites of recombinational DNA repair[44], [45]. As shown in Fig. 1D, RAD51 focus positive cells were decreased significantly under the hypoxia situation by bevacizumab treatment (Fig. 1D). Collectively, bevacizumab inhibited the experiment of RAD51 mRNA in both groups, and the phenomena was relatively obvious with bevacizumab + olaparib treatment (Fig. 1E). These results suggested that bevacizumab induced hypoxia, thereby increasing HRR defection, which might have resulted in an elevated sensitivity to olaparib.

To examine the efficacy of bevacizumab + olaparib in vivo, we next assessed tumor cell proliferation and microvessel density via IHC staining of subcutaneous tumors after treatment. Compared with vehicle and treatment with either bevacizumab or olaparib alone, combination treatment reduced the number of Ki-67-positive proliferating tumor cells and CD31-positive endothelial cells in KRAS-mut subcutaneous tumors (Fig. 1B, F, and G). These results indicated that olaparib + bevacizumab increased sensitivity to olaparib in a KRAS-mut colon cancer mouse model, presumably via damage to HRR induced by bevacizumab-caused hypoxia.

Hypoxia increased damage to HRR and sensitized KRAS-mut colon cancer cells to olaparib

Noting the synergistic effect of olaparib + bevacizumab in vivo, we chose CoCl2 to mimic a hypoxic environment in vitro to explore the role of bevacizumab-induced hypoxia on olaparib (Fig. 2A). We selected 100 μmol/L CoCl2, a dose reported to be able to induce molecular responses similar to those found in low-oxygen conditions in mammalian systems [46], after confirming the expression of induced-hypoxia protein HIF-1α in cell lines HCT116, SW620, and Lovo (Fig. 2B, Supp. Fig. 3A). CoCl2 remarkably induced HIF-1α overexpression and sustained hypoxic conditions for at least 72 h (Fig. 2C, Supp. Fig. 3B). Meanwhile, we evaluated the biological activity of olaparib. Western blot results showed that olaparib could quickly inhibit PAR activity and sustain this effect for at least 72 h (Fig. 2D, Supp. Fig. 3C).

Fig. 2

CoCl2-induced hypoxia in KRAS-mut colon cancer cells and sensitized cells to olaparib. (A) Working model for the effects of bevacizumab + olaparib combination therapy in replicating cells. Normally, SSBs activate PARP, and resulting SSB repair occurs through PAR of histones and recruitment of additional PARP-dependent repair proteins; DNA is repaired, and cells survive. Olaparib, a PARP inhibitor, inactivated the PARP-dependent repair system and consequently trapped PARP on DNA repair intermediates, obstructing replication forks, which created DSBs during DNA replication. DSBs can be repaired through HR, resulting in cell survival. Bevacizumab-induced hypoxia led to downregulation of HRR gene RAD51, impairing HRR; DSBs accumulated, causing cell death. To detect HIF-1α, we treated HCT116 and SW620 cells with CoCl2 at the indicated (B) concentrations and (C) time points. (D) To detect PAR, we treated HCT116 and SW620 cells with olaparib (30 µM) at the indicated time points. Cells were lysed, and the indicated proteins were detected by Western blot. (E) We treated HCT116 and SW620 cells with or without CoCl2 (100 µM) and/or olaparib (30 µM) for 72 h. Cell growth was measured via MTT assay. (F) We treated HCT116 and SW620 cells with or without CoCl2 (100 µM) and/or olaparib (30 µM) for 72 h to examine RAD51 expression using qRT-PCR. Relative mRNA expression was normalized to GAPDH. Each experiment was performed at least 3 times independently. Bars indicate SD. N.S. = non-significant. *P < 0.05, ***P < 0.001, and ****P < 0.0001 for the indicated comparisons by 1-way ANOVA. Olap = Olaparib.

We next sought to evaluate the correlation between hypoxia and susceptibility to olaparib in KRAS-mut colon cancer cell lines. Cell viability of the above mentioned 3 tested cell lines was significantly inhibited by CoCl2 + olaparib treatment (Fig. 2E, Supp. Fig. 3D).

To confirm the role of hypoxia on HRR, we analyzed the RAD51 expression and RAD51 focus formation in KRAS-mut CRC cells in vitro using KRAS-mut cancer cell lines. Results shown that CoCl2-induced hypoxia decreased RAD51 expression (Fig. 2F, Supp. Fig. 3E). Meanwhile, the percentage of the RAD51 focus positive cells after CoCl2 treatment, was significantly lower compared with control group, indicating an impaired HRR by hypoxia (Fig. 3A and B).

Fig. 3

CoCl2 impaired HRR in KRAS-mut colon cancer cells. (A) Representative images of immunofluorescence staining showing the RAD51 foci in different groups for different cell lines. RAD51 and nuclei are shown in green (Alexa Fluor 488) and blue (DAPI), respectively, in the merged images. Scale bar = 20 μm. (B) Quantification of the RAD51 foci formation. Data shown represent the mean ± SD of 3 areas. *P < 0.05 and **P < 0.01 for the indicated comparisons by 1-way ANOVA. Olap = Olaparib.

Taken together, these results clearly indicated that sensitivity to olaparib could be increased by damaging HR via hypoxia, which reduced the viability of KRAS-mut colon cancer cells.

Olaparib induced cell cycle arrest, apoptosis and DNA damage of KRAS-mut colon cancer cells under hypoxia situation

To investigate the underlying mechanism by which hypoxia enhanced cancer cells’ sensitivity to olaparib, we performed both cell cycle and cell apoptosis assays on KRAS-mut CRC cells exposed to CoCl2 and/or olaparib. Compared with either CoCl2 or olaparib treatment alone, treatment with CoCl2 + olaparib treatment led to G2/M arrest (Fig. 4A and B, Supp. Fig. 3F and G). In addition, while treatment with either CoCl2 or olaparib alone induced moderate apoptosis in HCT116 and SW620 cells, treatment with CoCl2 + olaparib induced apoptosis to a striking degree (Fig. 4C and D). This indicated that hypoxia-mediated sensitivity to olaparib might be associated with increased cell apoptosis and cell cycle arrest.

Fig. 4

Olaparib induced cell cycle arrest and apoptosis of KRAS-mut colon cancer cells under hypoxic conditions. (A) Flow cytometry (FCM) images with PI staining for analysis of cell cycle distribution in HCT116 and SW620 cells after treatment with or without CoCl2 (100 µM) and/or olaparib (30 µM) for 72 h. (B) Proportions of HCT116 and SW620 cells in G1, S, and G2/M phases, presented as bar graphs. (C) We performed FCM analysis with Annexin V–PI staining to evaluate the percentage of HCT116 and SW620 cells that were apoptotic after treatment with or without CoCl2 (100 µM) and/or olaparib (30 µM) for 72 h. (D) Total proportion of cells undergoing apoptosis, represented as bar graphs. Each experiment was performed at least 3 times independently. Bars indicate SD. N.S. = non-significant. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 for the indicated comparisons by 1-way ANOVA. Olap = Olaparib.

An increased level of γ-H2AX is considered to be a marker of DNA damage [47]. Hence, we analyzed the γ-H2AX foci in cells under CoCl2 treatment with or without olaparib. As shown in Fig. 5A and B, CoCl2 treatment induced moderate DNA damage in cancer cells, while olaparib further enhanced the degree of DNA damage caused by CoCl2 in KRAS-mut CRC cells.

Fig. 5

Olaparib enhanced the degree of DNA damage caused by CoCl2 in KRAS-mut colon cancer cells. (A) Representative images of immunofluorescence staining of γ-H2AX foci in different groups for different cell lines. γ-H2AX foci and nuclei are shown in green (Alexa Fluor 488) and blue (DAPI), respectively, in the merged images. Scale bar = 20 μm. (B) Quantification of the γH2AX foci formation. Data shown represent the mean ± SD of 3 areas. N.S. = non-significant. ***P < 0.001, and ****P < 0.0001 for the indicated comparisons by 1-way ANOVA. Cont = Control; Olap = Olaparib; Beva = Bevacizumab.

Taken together, these results demonstrated that bevacizumab impaired HRR via hypoxia, providing additional benefit to olaparib in our subcutaneous-tumor model of KRAS-mut CRC. This antitumor effect might have been associated with cell cycle arrest, apoptosis and DNA damage.

Discussion

CRC is one of the major causes of cancer mortality worldwide [1]. Multiple monoclonal antibodies, such as cetuximab and bevacizumab—directed against human epidermal growth factor receptor 1 (HER1)/EGFR and VEGF, respectively—significantly improve the outcomes of patients with mCRC [23], [48]. However, these drugs are less effective in the KRAS-mut subset of colon cancer [9], [49], [50], [51]. Therefore, several clinical trials of combinational therapies have been conducted in this subset. In the present study, we confirmed the benefit of bevacizumab + olaparib in KRAS-mut colon cancer and further demonstrated that the underlying mechanism of this drug combination was impairment of HRR caused by bevacizumab-induced hypoxia.

Tumor hypoxia commonly occurs during the anti-angiogenic period and might play an important role in anti-angiogenic drug resistance [30], [31], [32]. Indeed, there is still some controversy over whether anti-angiogenic therapy, due to its heterogeneous effect, induces intratumor hypoxia in the tumor microenvironment. On the one hand, the traditional concept of anti-angiogenic therapy is to inhibit the formation of tumor blood vessels, thus aggravating tumor hypoxia. On the other hand, several studies have reported that anti-angiogenic therapy can normalize tumor vasculature, and thus improve tumor blood supply and ameliorate hypoxia [52], [53], [54], [55]. In cancer cell-derived xenograft models of cervical cancer, malignant glioma, and CRC, increased hypoxic areas and elevated hypoxia markers were found after treatment with diverse doses of bevacizumab, including 10 mg/kg, 3 doses over 8 days [55]; 6 mg/kg, 3 times weekly [56]; and 0.5 or 5 mg/kg, twice weekly [57]. However, in other cancers, such as squamous-cell carcinoma of the head and neck, researchers using [(61)Cu]Cu-diacetyl-bis(N4-methylthiosemicarbazone) positron emission tomography (Cu-ATSM PET) detected significant reductions in tumor hypoxia after bevacizumab monotherapy [58]. In our setting, at a twice-weekly dose of 5 mg/kg bevacizumab, hypoxic areas and hypoxia markers of HIF-1α in the tumor microenvironment increased; this was accompanied by tumor vascular regression, which was marked by CD31-positive endothelial cells. Taken together, these findings showed that bevacizumab treatment increased oxygenation in some experimental tumors and induced hypoxia in others. This might have been due to differences in tumor type, duration of administration, and dose.

Olaparib was approved by the FDA in December 2014 for advanced ovarian-cancer patients with a deleterious germline mutation of BRCA1 or BRCA2 [59]. Similar to BRCA1, RAD51 is another famous HRR gene that plays a central role in DSB repair [60], [61]. In our study, bevacizumab-induced hypoxia reduced the expression of RAD51 gene and the formation of RAD51 foci, suggesting damage to HRR; this observation was consistent with previous reports [34], [35], [36]. Disruption of such HR protein conferred sensitivity to PARP inhibitors and displayed marked antitumor effects in an HCT116 CRC xenograft model. Meanwhile, bevacizumab + olaparib was well tolerated in the mouse model, which was consistent with clinical practice for advanced solid tumors [37]. olaparib + cediranib, a VEGFR antagonist, is also reported to be tolerable and to improve PFS compared with olaparib alone in patients with recurrent platinum-sensitive high-grade serous or endometrioid ovarian cancer [62], [63]. Additionally, in the phase 3 Platine, Avastin, and OLAparib in First-line Therapy (PAOLA-1) trial, the addition of maintenance therapy to olaparib led to a statistically significant improvement in PFS in patients with advanced ovarian cancer receiving first-line standard therapy including bevacizumab, with a substantial benefit in patients with homologous recombination deficiency (HRD)-positive tumors [38].

Encouraged by the above-mentioned good tolerance and outcomes, in our present study, we sought further insight into the potential mechanism of bevacizumab + olaparib. We used CoCl2 to mimic hypoxia [64], [65] in vitro and to clarify the potential role of combined therapy. CoCl2 induced hypoxic response, which was confirmed by elevated HIF-1α protein expression in CoCl2-treated cells in a time- and concentration-dependent manner. In addition, consistent with our in vivo findings, we found that CoCl2 markedly downregulated expression levels of RAD51 mRNA and the percentage of the RAD51 focus positive cells, indicating an impaired HRR by hypoxia. Moreover, we confirmed that CoCl2-induced hypoxia could sensitize KRAS-mut CRC cells to olaparib by inducing additional DNA damage, cell apoptosis and cell cycle arrest in the G2/M phase. Our present data demonstrated the promising efficacy of olaparib + bevacizumab against KRAS-mut CRC. It will be interesting to evaluate this combination in other KRAS-mut tumor types, such as lung and pancreatic cancers. It will also be necessary to further clarify the potential relationship between KRAS mutations and HR, as well as the underlying mechanism of bevacizumab-induced hypoxia.

Conclusion

In conclusion, our findings suggested that bevacizumab led to hypoxia in the tumor microenvironment and induced HR deficiency in KRAS-mut cells. Therefore, it increased the effectiveness of olaparib, which might have occurred via G2/M phase arrest and DNA damage. These findings provide a rationale for this combination treatment in KRAS-mut CRC patients.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.