The CT26, HCT116, HT29, Colo205, SW48, RKO, SW480, SW620, LS513, LS1034, SW1116, LoVo, and LS174T cell lines were obtained from the American Type Culture Collection, whereas Calu-1 and H23 cell lines were purchased from the Korean Cell Line Bank. MC38 cell lines were obtained from Kerafast. RKO, MC38, and LS174T cell lines were cultured in DMEM-high glucose (Hyclone) supplemented with 10% fetal bovine serum (FBS, Gibco) and 1% antibiotics–antimycotics (Gibco). CT26, HCT116, HT29, Colo205, SW48, SW480, SW620, LS513, LS1034, SW1116, LoVo, Calu-1, and H23 cell lines were cultured in RPMI-1640 (Welgene) medium supplemented with 10% FBS (Gibco) and 1% antibiotics–antimycotics (Gibco). Patient-derived cancer cells (PDCs) were established, and the genetic identification and culture method were described previously.14 15 Bone marrow-derived macrophages or dendritic cells (BMDMs or BMDCs) differentiation method and cell viability and apoptosis analysis method were described in detail in the online supplemental material.

The Active Ras Pull-Down and Detection Kit (16117) was purchased from ThermoFisher Scientific. The negative control DsiRNA (51-01-14-03) and the siRNA targeting murine KRAS were obtained from Integrated DNA technologies. The sequences of the siRNA targeting murine KRAS were 5′-GUGCAAUGAGGGACCAGUA-3′ (sense), and its complementary anti-sense 5′-UACUGGUCCCUCAUUGCAC-3′. The transfections were conducted using Lipofectamine RNAiMAX transfection reagent (ThermoFisher) in accordance with the manufacturer’s protocol. The used antibody information was described in detail in the online supplemental material.

To evaluate the in vitro CRT expression, CT26 tumor cells were treated with simvastatin (10 µM) for 4 or 24 hours or KRAS siRNA (100 nM) for 24 hours. Calu-1 and H23 tumor cells were treated with simvastatin (10 µM) or AMG-510 (10 µM, HY-114277) for 24 hours. In the case of combination treatment, CT26 cancer cells are treated with simvastatin (1 µM); after 12 hours, oxaliplatin (300 µM) is added and then left for 12 hours. To evaluate the in vivo CRT expression, tumors from tumor-bearing mice treated with simvastatin were isolated and dissociated into single cells using a Tumor Dissociation Kit (Miltenyi Biotech). The tumor cells were collected by excluding the immune cells, endothelial cells, and fibroblasts using CD45 (130-052-301), CD31 (130-097-418), and CD90.2 microbeads (130-121-278). The detailed CRT staining protocol was described in the online supplemental material. To assess the release of HMGB1 and HSP70 from cancer cells, CT26 tumor cells were treated with simvastatin (10 µM) for 24 hours, and western blot was performed to detect HMGB1 and HSP70 in collected supernatant.

BMDMs were stained with green 5-chloromethylfluorescein diacetate (CMFDA) (1 µM, C2925) for 25 min at room temperature 24 hours before coculture with CT26 cells, whereas the BMDCs were stained similarly on the day of coculture. CT26 cells were treated with simvastain (10 µM) and/or FPP (13058-04-3, 5 µM) or GGPP (G6025, 5 µM) for 4 hours. Treated CT26 cells were stained with pHrodo Red, succinimidyl ester (pHrodo-SE) (P36600, 120 ng/mL) for 30 min and added to the BMDMs or BMDCs for 1 hour at a ratio of 1:2. The medium was refreshed with pH 10 phosphate-buffered saline. At least seven images per sample were randomly taken and analyzed under a fluorescence microscope (Nikon). Phagocytosis (%) was calculated by the following formula; the number of phagocytosed cancer cells (red)/total number of BMDMs or BMDCs (green) × 100.

Male C57BL/6 mice and BALB/c (6–8 weeks old) were purchased from OrientBio, whereas the B6.129-Krastm3Bbd/J mice were purchased from the Jackson Laboratory. The Pdx1-Cre Ink4a/Arflox/+and LSL-KrasG12D Ink4a/Arflox/lox mice were kindly provided by Dr ANCA (INSERM, France). These mice were intercrossed to generate the KIC (p48Cre;LSL-KrasG12D;Cdkn2af/f) mice.16 All mice were maintained in a pathogen-free room at the Korea Institute of Science and Technology (KIST). We performed all animal studies under the Institutional Animal Care and Use Committee (IACUC) guidelines of the KIST. The method of in vivo tumor model experiments and flow cytometry analysis were described in detail in the online supplemental material.

DCs were sorted from tumors of CT26 or MC38 tumor-bearing mice via CD11c microbeads (130-125-835). The QIAZOL reagent (Qiagen Ltd) was used to extract the total RNA from DCs. The quantification of samples was assessed by Qubit Fluorimetric Quantitation (Thermo Scientific) and Nanodrop (Thermo Scientific). To analyze the gene expression profiles of six samples, the NanoString nCounter System was used by evaluating 100 ng total RNA from the sample with nCounter mouse myeloid innate immunity panel (LBL-10 398-02), which is optimized for evaluation of genes representing the overall myeloid innate immunity pathways. The correction of background was done by subtracting the geometric mean of eight negative-control probes. The values of gene expression were normalized with six positive-control probes and 32 housekeeping genes by the nSolver program, followed by log2-transformation and standardization within each sample. The fold changes and p values were displayed via volcano plots.

DCs were extracted from the TDLNs of CT26 or MC38 tumor-bearing mice with CD11c microbeads (130-125-835). CD8+ T cells were sorted from the spleen of control mice by using the CD8α+ T-Cell Isolation Kit (130-104-075). DCs (5×10⁴) and T cells (2.5×10⁵) were coincubated using the RPMI-1640 medium containing GM-CSF (20 ng/mL) for 72 hours. The supernatant was collected, and interferon-γ (IFN-γ) was analyzed with an IFN-γ ELISA kit (R&D Systems) following the protocol from provider. The expression of CD107α on CD8+ T cells was analyzed using flow cytometry (Accuri C6) with APC antimouse CD107α and FITC (Fluorescein isothiocyanate) antimouse CD3 antibodies.

Paraffin blocks of tumor tissues were sliced into 4 µm thick slides and heated at 60°C for 1 hour. The slides were dewaxed with xylene and stained with Leica Bond Rx Automated Stainer (Leica Biosystems). Afterwards, the slides were baked for 30 min, dewaxed, placed in Bond Epitope Retrieval 2 (Leica Biosystems) and Bond Epitope Retrieval 1 (Leica Biosystems) sequentially. The slides were incubated with the PD-L1 primary antibodies (13 684S) and CD8 (98 941S) and detected using the corresponding polymer horseradish peroxidase. Visualization was performed with using the Opal TSA Plus dye 690 and 720, respectively (Akoya Biosciences). Anti-CD8 antibody was incubated after incubation with the Opal TSA-DIG reagent (Perkin-Elmer). The nuclei were stained using DAPI (4′,6-diamidino-2-phenylindole) before the slides were covered using HIGHDEF IHC fluoromount (Enzo). The images were gained with PerkinElmer Vectra V.3.0 Automated Quantitative Pathology Imaging System (Perkin-Elmer) and evaluated with the InForm software V.2.2 and TIBCO Spotfire (Perkin-Elmer).

All statistical analyses were performed on Prism 8 (GraphPad). Student’s t-test was used to analyze the comparisons between two groups, whereas one-way analysis of variance was used to analyze those among at least three groups, followed by Tukey’s post-hoc test. Tumor-free frequency and survival benefit were evaluated using Kaplan–Meier analysis with log-rank test. The error bars are shown as SE of the mean (SEM). P values<0.05 were considered statistically significant.

We first evaluated the association between the antiproliferative effects of statin treatment and various KRAS mutations using CRC cell lines (figure 1A,B). Simvastatin treatment significantly decreased the proliferation and promoted the apoptosis of KRAS^mut cancer cells in comparison with that its effects on KRAS wild type (KRAS^wt) cancer cells (figure 1C–E). The simvastatin-induced apoptosis in KRAS^mut CT26 cancer cells was partially abrogated by GGPP but completely abrogated by FPP, indicating that the tumor cell death by statins is more dependent on FPP depletion in CT26 cells (figure 1D).

To determine whether statin can suppress KRAS signaling by blocking isoprenylation, we observed the level of phosphorylated-ERK1/2 (p-ERK1/2), a protein located downstream in the RAS signaling pathway. We found that p-ERK1/2 levels were significantly reduced by simvastatin in KRAS^mut cancer cells (CT26, HCT116, and SW480) but not in KRAS^wt cancer cells (HT29 and Colo205) (figure 1F and online supplemental figure 1A,B). In addition, the inhibitory effect of simvastatin on KRAS activity in CT26 and HCT116 cancer cells was impaired by FPP but not by GGPP (figure 1F and online supplemental figure 1A). These findings imply that FPP depletion by simvastatin can inhibit the RAS signaling pathway in the KRAS^mut cancer cells.

To assess potential activity of statin in more clinically relevant models, we used PDCs with various oncogenic driver mutations (figure 1G). Consistent with the results obtained using cell lines, simvastatin substantially decreased the viability of KRAS^mut PDCs (141T, 150T, 166T, and 122T) compared with those of KRAS^wt PDCs (146T and 149T) (figure 1H). Interestingly, 139T PDCs (KRAS^G12D) harboring the BRAF (v-raf murine sarcoma viral oncogene homolog B1) mutation did not respond to simvastatin (figure 1H). Likewise, simvastatin failed to induce cytotoxicity in LS174T cell lines that carried both KRAS and BRAF mutations (figure 1E). These results suggest that genetic activation of the signaling pathways downstream of KRAS may also affect the susceptibility of cancer cells to statins.

To investigate whether statin could provoke ICD in KRAS^mut cancer cells, we analyzed the levels of ICD markers following simvastatin treatment. One of the most characteristic ICD events is the transfer of CRT, an “eat-me” signal, from the ER lumen to the plasma membrane in the preapoptotic cell.17 18 The phosphorylation of PERK-elf2-α, an ER stress response, has been reported to trigger CRT/ERp57 exposure on the plasma membrane.19 We demonstrated that simvastatin increased the signaling of the elf2-α phosphorylation (p-elf2-α) pathway including ATF4 and CHOP in KRAS^mut cancer cells (figure 2A,B), which was not observed in KRAS^wt cells (online supplemental figure 1C), subsequently leading to a significant increase in the expression of CRT and ERP57 (figure 2C,D and online supplemental figure 1D) on the plasma membrane of tumor cells. However, FPP blocked this translocation, whereas GGPP did not (figure 2C,D and online supplemental figure 1D). Moreover, simvastatin evoked the release of HMGB1 and HSP70 proteins from CT26 tumor cells, which serve as damage-associated molecular patterns signals to activate the functions of APCs (figure 2E).10 Likewise, treatment with simvastatin effectively decreased p-ERK1/2 levels in CT26 tumor cells in vivo and increased expression of CRT on membranes (figure 2F–H).

A previous study has demonstrated that RAS signaling negatively regulates ER stress responses, including the phosphorylation of elf2-α, thereby preventing ER stress-associated cell death.12 To determine whether KRAS inhibition could induce ER stress responses, we treated KRAS^G12D (CT26) and KRAS^G12C cancer cells (Calu-1 and H23) with the specific KRAS inhibitors; KRAS^G12D siRNA and AMG-510 (KRAS^G12C inhibitors), respectively. These KRAS inhibitors, like simvastatin, effectively induced the apoptosis of KRAS^mut cancer cells and exposed CRT and ERP57 on plasma membranes in preapoptotic states (figure 2I–L and online supplemental figure 1E). These findings indicate that KRAS acts as a negative regulator of ER stress responses in KRAS^mut tumors.

To verify that simvastatin can increase the phagocytosis of cancer cells, we performed in vitro phagocytosis assays using pHrodo-SE that emits higher signals in the low-pH conditions in the phagosome.20 We found that CT26 cells pretreated with simvastatin were readily engulfed by BMDMs and BMDCs (figure 3A,B), and that FPP impaired such effects (figure 3A,B), indicating that statin-induced-FPP depletion in KRAS^mut tumor cells boosted cancer cell clearance. These results strongly suggest that statins can efficiently elicit ICD in KRAS^mut tumors via suppressing the isoprenylation of KRAS.

ICD of cancer cells allows APCs to recognize tumor cells as danger signals, thereby triggering tumor-specific T-cell immunity.10 To elucidate the sequence of the immunological events induced by simvastatin treatment, we analyzed its effect on DCs in CT26 tumor tissues. To test whether statin can activate tumor DCs in vivo, we analyzed the gene signatures of DCs isolated from tumor tissues (figure 3C). DCs from CT26 tumors (KRAS^mut) of simvastatin-treated mice had a higher expression of genes closely related to antigen uptake, antigen processing, peptide loading, DC activation, and DC maturation compared with the controls (figure 3C and online supplemental figure 2) and exhibited enhanced expression of cross-presentation related genes, such as Tap1, Tap2, Tapbp, Cybb, Ncf2, and H2-K1, (figure 3D and online supplemental figure 3) and H2K^d proteins (figure 4A) in DCs of CT26 tumors. In contrast, in the MC38 tumor (KRAS^wt) model, simvastatin did not substantially change the expression of these genes (figure 3C,D and online supplemental figure 2). These results suggest that the statin-induced KRAS^mut cancer cell death is immunogenic to efficiently activate DCs in vivo.

To evaluate the ability of the dying cells to initiate adaptive immunity in vivo, we assessed whether simvastatin-based DC vaccines could promote a prophylactic vaccine effect against KRAS^mut cancers. BMDCs coincubated with statin treated-cells or control CT26 cells were subcutaneously injected into the left flank of immunocompetent, BALB/C mice. After 7 days, the immunized and control mice were challenged with live CT26 cells in the right flank. Remarkably, approximately 70% of the mice vaccinated with simvastatin-based DC vaccines were protected from the tumor growth of CT26 cells, whereas other groups were not (figure 3E). These results demonstrate that simvastatin has the potential to generate DC-mediated T-cell immunity to prevent tumor growth.

DCs carrying tumor antigens migrate to the tumor-draining lymph nodes (TDLNs), where they induce T cells to recognize and eradicate tumors.21 22 To assess whether statin therapy potentiates the functions of DCs that are necessary to achieve potent T-cell immunity against cancer, we evaluated the immunological benefits of simvastatin on DC maturation in TDLNs. Although simvastatin did not directly induce DC maturation in vitro (online supplemental figure 4A), systemic administration of simvastatin significantly increased the levels of DC maturation markers, such as CD40, CD80, and CD86, in the CT26 tumor model whereas did not in MC38 (figure 4B). Previous studies have demonstrated that lipophilic statins mediated GGPP depletion, including simvastatin, promoted antigen presentation in APCs by suppressing Rab5 signaling.23 Likewise, we found that simvastatin could directly enhance cross-presentation by DCs via GGPP depletion when challenged with OVA peptides (online supplemental figure 4B). These results indicate that statins can activate the functions of DCs effectively in KRAS^mut tumors.

To confirm the effects of statin on T-cell priming, we isolated DCs from the TDLNs of simvastatin-treated CT26 and MC38 tumor-bearing mice and coincubated them with CD8+ T cells from the lymphoid organs of CT26 and MC38 tumor models, respectively. We observed that DCs isolated from KRAS^mut TDLNs, but not from KRAS^wt TDLNs, provoked efficient CD8+ T-cell priming as evidenced by IFN-γ expression following simvastatin treatment (figure 4C). Simvastatin treatment also elevated the CD8+ T-cell proportion in the TDLNs of CT26 tumors with increased CD44 expression, an effect related to the antigen-experienced status of the T cells (figure 4D,E). These findings suggest that the induction of ICD in KRAS^mut tumors by statin can promote the cross-priming abilities of DCs, thereby producing tumor-specific CD8+ T cell immunity.

We next investigated the dependence of the antitumorigenic effect of statins on T-cell immunity. We discovered that simvastatin treatment substantially elevated the fractions of CD8+ T cells in CT26 but not in MC38 tumors (figure 4F). Because the properties of the TME can impair the cytotoxic T-cell immunity, the activation status of CD8+ T cells is important for tumor control.24 25 We observed that simvastatin treatment increased the levels of CD69 (an early activation marker for CD8+ T cells) and CD107α (an activation marker for the degranulation of CD8+ T cells) in the CD8+ T cells of CT26 tumors (figure 4G,H). These results suggest that simvastatin increases not only the infiltration of CD8+ T cells but also their activity in CT26 tumors but not in MC38 tumors.

Consistent with our immunological data, simvastatin treatment significantly suppressed the growth and tumor weight in the CT26 model but not in the MC38 model (figure 4I and online supplemental figure 5A,D). There was no substantial change in body weight between the two groups (online supplemental figure 5B,C). We also assessed the correlation between the antitumor effects of simvastatin and the depletion of CD4+ or CD8+ T-cell immunity in the CT26 tumor model. Antibody-mediated CD8+ T-cell depletion abrogated the antitumor efficacy of simvastatin, whereas depletion of CD4+ T cells did not (figure 4J and online supplemental figure 5E–H). Overall, these results imply that CD8+ T-cell immunity is necessary for simvastatin-mediated tumor regression.

Oxaliplatin is a well-known ICD inducer that can increase both the preapoptotic expression of CRT and the postapoptotic release of HMGB1,26 thus promoting the activation of DCs and CD8+ effector T -cell function.27 Hence, this drug has been widely used in combination therapy in oncoimmunology related studies and has exhibited excellent results when combined with ICBs.28 Oxaliplatin is a first-line chemotherapeutic agent for colorectal, non-small-cell lung, and pancreatic cancers, which are tumors frequently exhibiting KRAS mutations.29 Accordingly, we investigated whether a combined therapy of oxaliplatin and simvastatin could better provoke the ICD of KRAS^mut cancer cells. We found that the combination treatment significantly enhanced CRT expression in CT26 cells compared with monotherapy using either simvastatin or oxaliplatin alone (figure 5A). Notably, the combination of simvastatin and oxaliplatin caused a substantial increase in BMDC-mediated phagocytosis of immunogenic dying CT26 cells (figure 5B).

Furthermore, we determined the effect of the combined therapy on tumor growth and found that it effectively suppressed the tumor without increased toxicity compared with monotherapy (figure 5C,D and online supplemental figure 6). We next assessed its potential immunological effect against cancer. DCs from the TDLNs of combination therapy-treated mice exhibited higher levels of costimulatory markers, such as CD40, CD80, and CD86, which reflected DC maturation compared with other groups (online supplemental figure 7). In addition, the combination therapy effectively increased the capability of DCs to prime CD8+ T cells, as measured by enhanced IFN- γ production and CD107α expression of primed CD8+ T cells (figure 5E,F and online supplemental figure 8A), as well as elevation of the number of CD8+ T cells in the TME and the proportion of IFN-γ+ in CD8+ T cells (figure 5G,H and online supplemental figure 8B,C). However, combined therapy did not show any effects on other immune cells (online supplemental figure 8D–H).

In contrast with subcutaneous syngeneic tumor models, genetically engineered mouse models (GEMMs) form de novo orthotopic tumors that display genetic heterogeneity resembling the tumor-immune microenvironments in humans.30 To confirm the efficacy of the combination of simvastatin and oxaliplatin in GEMMs, we used the B6.129-Kras^tm3Bbd/J lung and the KIC pancreatic tumor models, which harbor KRAS mutations (online supplemental figure 8I).16 31 We found that combined therapy efficiently inhibited the number and fraction of tumor nodules in the B6.129-Kras^tm3Bbd/J lung tumor models (figure 5I,J). Likewise, this treatment improved the survival rate in KIC pancreatic tumor model compared with other treatment groups (figure 5K). Furthermore, immunofluorescence analysis revealed that the CD8+ T-cell infiltration to tumors was robustly increased in the combination therapy groups (figure 5L). These results clearly indicate that the combination of simvastatin and oxaliplatin produces substantial therapeutic effects even in GEMMs carrying a KRAS mutation, thereby reducing tumor growth and enhancing CD8+ T cell immunity.

To better understand the complicated tumor-immune microenvironment, we performed a multiplex IHC on the treated tumor tissues from the CT26 syngeneic tumor model. Consistent with our Fluorescence-activated cell sorting (FACS) results, the combination therapy significantly increased CD8+ T cells in the TME (figure 6A). In addition, we found that increased CD8+ T-cell accumulation by combined therapy was correlated with enhanced tumor cell death (online supplemental figure 9A). Although CD8+ T-cell infiltration is instrumental for tumor control, cancer cells can evade the CD8+ T-cell-mediated-immune destruction by expressing inhibitory signals, such as PD‐L1.24 PD-L1 on tumor cells is upregulated by IFN-γ and oncogenic signaling including RAS.32 Given that statin could inhibit RAS activity by blocking farnesylation, we hypothesized that simvastatin may suppress PD-L1 expression on KRAS^mut cancer cells. We thus explored whether simvastatin treatment affects PD-L1 expression in CT26 tumors. The overall PD-L1 expression in tumor tissues was decreased following combination therapy (figure 6A). Moreover, FACS analysis revealed that the combination or treatment with simvastatin alone efficiently inhibited PD-L1 expression in tumors (CD45⁻ cells) but not in immune cells (CD45+ cells) (figure 6B,C). These results suggest that the combination therapy can markedly suppress PD-L1 expression of KRAS^mut tumor cells even in the presence of activated CD8+ T cells.

However, heterogenous PD-L1 expression restrains CD8+ T-cell infiltration into tumors following combination therapy (figure 6D). The area with high PD-L1 expression (right white circle) showed little CD8+ T-cell infiltration, whereas the region with relatively low PD-L1 expression (left white circle) showed higher infiltration (figure 6D and online supplemental figure 9B). These findings indicate that PD-L1 blockade may be required to enhance the therapeutic index of the combination therapy using statin and oxaliplatin. To investigate the effects of ternary combination therapy of simvastatin, oxaliplatin, and anti-PD-1 antibodies, we used CT26 syngeneic tumor model, resistant to PD-1 blockade (figure 6E).33 Consistent with previous reports, anti-PD-1 antibody monotherapy did not affect CT26 tumor growth, whereas combination therapy with statins and oxaliplatin exhibited a modest tumor-suppressing effect (figure 6F). Importantly, the ternary combination treatment of simvastatin, oxaliplatin, and anti-PD-1 antibodies dramatically suppressed tumor growth and improved the survival rate of CT26 tumor-bearing mice (figure 6F,G). Together, these results demonstrate that combination therapy with simvastatin and oxaliplatin sensitizes the tumors lacking CD8+ T cells to PD-1 blockade, thereby eliciting remarkable tumor-suppressing effects.