Inhibition of HSP90 by AUY922 Preferentially Kills Mutant KRAS Colon Cancer Cells by Activating Bim through ER Stress

This article is featured in Highlights of This Issue, p. 345

Colon cancer is one of the most common and deadly malignancies (1). Despite recent advances in early diagnosis and the development of molecularly targeted therapy, the overall survival of patients with metastatic colon cancers remains disappointing (1). This is often associated with resistance of colon cancer cells to systemic therapies resulting from oncogenic mutations of KRAS which drive activation of multiple signaling pathways important for cell survival and proliferation (2). In fact, activating mutations of KRAS are found in up to 50% of colon cancers that forecast inherited resistance to antibodies against the EGFR (2, 3).

HSP90 is the most abundant molecular chaperone and is essential for folding, stabilization, and activation of a large number of proteins (4). In particular, many mutant and overexpressed oncoproteins such as EGFR, mutant BRAF, and Akt are clients of HSP90 (4). As such, targeting HSP90 appears a promising approach in the treatment of cancer (4). Intriguingly, mutant KRAS-driven cancer cells have been noted to be sensitive to HSP90 inhibition (5–7), but the molecular mechanisms involved remain poorly understood. Nevertheless, multiple HSP90 inhibitors are currently in clinical studies for the treatment of cancer (4).

Induction of apoptosis is a common mechanism by which therapeutic drugs kill cancer cells (8). This is frequently mediated by activation of the mitochondrial apoptotic pathway, which is regulated by the balance between proapoptotic and antiapoptotic Bcl-2 family proteins (8). Of note, multiple Bcl-2 family proteins are responsive to inhibition of HSP90 in a cell-type and context-dependent manner (9–12). In particular, p53-dependent induction of the BH3-only protein PUMA has been shown to be responsible for apoptosis of colon cancer cells induced by the HSP90 inhibitor 17-AAG (10). Moreover, the antiapoptotic Bcl-2 family proteins, Bcl-2, Mcl-1, and Bcl-XL, have all been reported to protect cells against HSP90 inhibition–induced apoptosis (9, 11–13).

Another frequently observed consequence of HSP90 inhibition is endoplasmic reticulum (ER) stress, which is characterized by accumulation of unfolded and/or misfolded proteins in the ER lumen. The ER responds to ER stress by activation of a range of signaling pathways, collectively called the ER stress response or the unfolded protein response (UPR; refs. 14, 15). The UPR is initiated by three ER transmembrane proteins, activating transcription factor 6 (ATF6), inositol-requiring enzyme 1 (IRE1), and double-stranded RNA-activated protein kinase-like ER kinase (PERK), and is essentially a cellular protective response. However, excessive UPR kills cells primarily by induction of apoptosis (14–16). Multiple mechanisms such as activation of the BH3-only proteins Bim, PUMA, and Noxa, and downregulation of antiapoptotic Bcl-2 family proteins including Bcl-2 and Mcl-1 have been implicated in ER stress–induced apoptosis (17–19).

We have examined the potency of the HSP90 inhibitor AUY922 in colon cancer cells with different mutational statuses of KRAS. We show here that AUY922 preferentially induces apoptosis in mutant compared with wild-type KRAS colon cancer cells, and that activation of Bim through ER stress is responsible for apoptosis triggered by AUY922. In addition, we demonstrate that AUY922 inhibits the growth of mutant KRAS colon cancer xenografts similarly through Bim-mediated apoptosis that is associated with ER stress.

Human colon cancer cell lines provided by Prof. Gordon Burns (Faculty of Health and Medicine, University of Newcastle, New South Wales, Australia) were described previously (20). The normal human colon epithelial cell line FHC (ATCC CRL-1831) was from ATCC, and cultured as described previously (20). Individual cell line authentication was regularly confirmed every 6 months using the AmpFISTR Identifier PCR Amplification Kit from Applied Biosystems and GeneMarker V1.91 software (SoftGenetics LLC). The last test was done in May 2015.

Three-dimensional (3D) culture was performed using the hanging drop technique as described previously (21). Briefly, 500 cells were seeded into the Perfecta3D hanging drop plate (3D Biomatrix), and monitored with the Axiovert and Axioplan microscope (Carl Zeiss) for at least 5 days. Cells were then stained with calcein AM and ethidium homodimer-1 (Life Technologies) for 24 hours followed by treatment. Spheroids were harvested onto slides and examined with a fluorescence microscope (Carl Zeiss).

Antibodies and reagents used are listed in Supplementary Tables S1 and S2.

qPCR was performed as described previously (17). The relative abundance of target mRNA in control group was arbitrarily designated as 1. The primes are forward primers for Bim (GCCCCTACCTCCCTACAGAC), Bcl-2 (CTGCACCTGACGCCCTTCACC), spliced XBP-1 (GCACCTGAGCCCCGAGGAGA), XBP-1 (AGCCAAGGGGAATGAAGTGAG), or β-actin (GGCACCCAGCACAATGAAG); reverse primers for Bim (ATGGTGGTGGCCATACAAAT), Bcl-2 (CACATGACCCCACCGAACTCAAAGA). Spliced XBP-1 (TCATTCCCCTTGGCTTCCGCC), XBP-1 (CTGCAGAGGTGCACGTAGTC), or β-actin (GCCGATCCACACGGAGTACT).

The pCMV6-AC-GFP-Bcl-2 vector was from Origene (Australian Biosearch). GRP78 cDNA cloned into pcDNA3.1 was provided by Dr. Richard C. Austin (McMaster University and the Hamilton Civic Hospitals Research Centre, Hamilton, Ontario, Canada) and described elsewhere (22).

siRNA constructs are listed in Supplementary Table S3. siRNA transfection was carried out as described previously (17). MISSION human short hairpin RNA (shRNA) lentiviral transduction particles Bim (SHCLNV-NM_009754) and the control particles were from Sigma-Aldrich.

Colon cancer cells were injected subcutaneously into flanks of male athymic nude mice (Model Animal Research Centre of Nanjing University, China). Ten days after injection, when xenografts were approximately 100 mm³, mice were randomly assigned into different groups (n = 8). Mice were treated daily with AUY922 (50 mg/kg/day in sterile PBS via i.p. injection), or equivalent volumes of the vehicle (DMSO) for 10 days. Mice were examined as described previously and sacrificed at 28 days after tumor cell transplantation (23). Studies on animals were approved by the Animal Research Ethics Committee of Shanxi Cancer Hospital, China.

Statistical analysis was performed using JMP Statistics Made Visual software. Student t test was used to assess differences between different groups. A P value less than 0.05 was considered statistically significant.

HCT116 (KRAS^G13D), SW620 (KRAS^G12V), and WiDr and Caco-2 (wild-type KRAS) cells were treated with AUY922 at increasing concentrations for 48 hours. Strikingly, HCT116 and SW620 cells appeared more sensitive than WiDr and Caco-2 cells to AUY922-induced reduction in viability as measured using CellTiter-Glo assays (Fig. 1A), with the half-maximum inhibitory concentration (IC₅₀) values ranging from 76.5 to 82.2 nmol/L in HCT116 and SW620 cells, respectively, to >400 nmol/L in WiDr and Caco-2 cells (Fig. 1A and B). Of importance, AUY922 did not significantly affect survival of normal colon epithelial cells even when used at 400 nmol/L (Supplementary Fig. S1).

The different responses of the colon cancer cells to AUY922 were also reflected in clonogenic assays and in cells grown in 3D cultures when the inhibitor was used at 80 nmol/L (the approximate IC₅₀ value in HCT116 and SW620 cells; Fig. 1C and D). Nevertheless, AUY922 at this concentration inhibited HSP90 activity similarly in sensitive and resistant cells, as it induced comparable degrees of downregulation of the HSP90 clients CRAF and S-phase kinase-associated protein 2 (SKP2), and upregulation of HSP70 (Fig. 1E). Consistent with previous findings (24), AUY922 inhibited activation of ERK and Akt, which are downstream of multiple clients of HSP90, in colon cancer cells irrespective of their KRAS mutational status (Fig. 1E).

The increased susceptibility of mutant KRAS colon cancer cells to AUY922 was confirmed in additional two mutant KRAS cell lines [SW480 (KRAS^G12V) and EB (KRAS^G12D)] and three wild-type KRAS colon cancer lines (Colo205, Lim1863, and Lim1215; Fig. 1B). Of note, sensitivity of colon cancer cells to AUY922 was not associated with the expression levels of HSP90 (Supplementary Fig. S2). Nevertheless, knockdown of HSP90α and β, two major isoforms of HSP90 that are inhibited by AUY922 (25, 26), similarly inhibited survival of HCT116 and SW620 but not WiDr and Caco-2 cells (Fig. 1F). Collectively, these results indicate that, in contrast to resistance to many therapeutic drugs (2, 3), mutant KRAS colon cancer cells are vulnerable to HSP90 inhibition.

The general caspase inhibitor z-VAD-fmk efficiently inhibited AUY922-induced HCT116 and SW620 cell death, indicating that AUY922 kills these cells by apoptosis (Fig. 2A and Supplementary Fig. S3; ref. 27). In support, AUY922 caused accumulation of sub-G₁ DNA content, exposure of phosphatidylserine onto the cell surface, activation of caspase-3, and cleavage of its substrate PARP (Fig. 2B and Supplementary Fig. S4). Overexpression of Bcl-2 inhibited apoptosis induced by AUY922 (Fig. 2C), indicating that the mitochondrial apoptotic pathway plays an essential role. Consistently, knockdown of Bax also diminished apoptosis induced by AUY922 (Fig. 2D; ref. 10). Moreover, AUY922 triggered reduction in the mitochondrial membrane potential, release of cytochrome c and Smac/DIABLO, and activation of caspase-9 in HCT116 and SW620 cells (Fig. 2B and Supplementary Fig. S5).

As a client protein of HSP90, Bcl-2 was reduced rapidly in HCT116 and SW620 cells, but not in wild-type KRAS colon cancer cells, upon treatment with AUY922 (Fig. 2E; refs. 9, 28). On the other hand, the expression of the other antiapoptosis Bcl-2 family proteins, Mcl-1 and Bcl-XL, was not changed by AUY922 even in HCT116 and SW620 cells (Fig. 2E). Remarkably, AUY922 upregulated multiple BH3-only proteins, including Bim, Bik, and PUMA in sensitive but not in resistant colon cancer cells (Fig. 2E). There was no change in the expression of Bax and Bak, nor was there any alteration in phosphorylation (deactivation) of the BH3-only protein Bad, upon AUY922 treatment in both mutant and wild-type KRAS colon cancer cells (Fig. 2E).

We knocked down Bim, Bik, and PUMA using two individual siRNAs in HCT116 and SW620 cells (Fig. 3A–C). Strikingly, knockdown of Bim abolished cell death induced by AUY922 (Fig. 3D), whereas knockdown of Bik also inhibited, albeit partially, AUY922-induced cell death (Fig. 3D). Consistent with the lack of upregulation of Bim in wild-type KRAS colon cancer cells by AUY922 (Fig. 2E), knockdown of Bim did not alter sensitivity of Caco-2 cells to AUY922 (Supplementary Fig. S6). Of interest, although PUMA and p53 have been reported to mediate killing of colon cancer cells by the other HSP90 inhibitor 17-AAG (10), knockdown of PUMA or p53 did not impinge on killing of HCT116 and SW620 cells by AUY922 (Fig. 3C and D and Supplementary Fig. S7). Nevertheless, PUMA or p53 knockdown inhibited cell death induced by the MDM2 inhibitor nutlin-3 (Supplementary Fig. S7; ref. 29), verifying the functional efficacy of the PUMA and p53 siRNAs.

Upregulation of Bim by AUY922 was associated with elevation in its mRNA levels (Supplementary Fig. S8A). This was due to a transcriptional increase rather than changes in its stability as shown in actinomycin D–chasing assays (Supplementary Fig. S8B and S8C; ref. 17). As Bim transcription is responsive to ER stress (19), we tested whether ER stress was involved in upregulation of Bim by AUY922. Indeed, treatment with AUY922 activated the ER stress response in mutant but not wild-type KRAS colon cancer cells (Fig. 4A and B). Noticeably, the expressions of IRE1 and PERK that have been reported to be HSP90 clients remain unchanged after treatment with AUY922 (Fig. 4A).

As CCAAT-enhancer-binding protein homologous protein (CHOP) is responsible for transcriptional upregulation of Bim by ER stress (19), we examined the potential role of CHOP in Bim upregulation by AUY922 in colon cancer cells. As shown in Fig. 4C, siRNA knockdown of CHOP inhibited upregulation of Bim, activation of caspase-3, cleavage of PARP, and cell death induced by the inhibitor in HCT116 and SW620 cells (Supplementary Fig. S9). In support, the chemical chaperon BGP-15 or 4-PBA attenuated killing by AUY922 (Fig. 4D; refs. 30, 31). This was associated with reduced induction of GRP78, spliced XBP-1, CHOP, and Bim (Fig. 4E and Supplementary Fig. S10). Similarly, overexpression of GRP78 inhibited AUY922-induced upregulation of CHOP and Bim and apoptosis in HCT116 and SW620 cells (Fig. 4F and Supplementary Fig. S11).

We examined whether inhibition of GRP78 and XBP-1, two major prosurvival effectors of the ER stress response (14, 15), affected sensitivity of colon cancer cells to AUY922. Knockdown of either GRP78 or XBP-1 increased apoptosis induced by AUY922 in HCT116 and SW620 cells (Fig. 5A–C). Of note, it also rendered WiDr cells sensitive, albeit moderately, to AUY922-induced killing (Fig. 5A–C), suggesting that resistance of wild-type KRAS colon cancer cells to AUY922 is, at least in part, due to better adaptation to ER stress. The effect of XBP-1 was further confirmed by treatment with salicylaldehyde that inhibits the IRE1 endonuclease activity thus blocking splicing (activation) of XBP-1 (Fig. 5D; ref. 32).

To test the effect of AUY922 on the growth of colon cancers in vivo, we transplanted HCT116 and WiDr cells into nu/nu mice. Administration of AUY922 inhibited established HCT116 tumor growth, but did not impinge on the growth of WiDr tumors (Fig. 6A–C). Inhibition of HCT116 tumor growth by AUY922 was associated with activation of caspase-3, upregulation of Bim, and induction of ER stress (Fig. 6D and E). However, the inhibitory effect was diminished in xenografts of HCT116 cells with Bim stably knocked down, although AUY922 triggered ER stress in the tumors (Fig. 6F–H).

We have shown in this study that mutant KRAS colon cancer cells are more susceptible to apoptosis induced by the HSP90 inhibitor AUY922, consistent with the observations in other types of cancers that HSP90 inhibition has the greatest effect in tumors addicted to particular driver oncogene products, such as HER2 in breast cancer (33), mutant EGFR or ALK translocations in non–small cell lung carcinoma (NSCLC; refs. 34, 35), and mutant BRAF in melanoma (36). In particular, inhibition of HSP90 has been shown to have potent antitumor activity in mutant KRAS NSCLC models in vivo (37, 38). Moreover, we have demonstrated that activation of Bim though ER stress plays an essential role in AUY922-induced apoptosis of mutant KRAS colon cancer cells (Supplementary Fig. S12).

Although HSP90 inhibition has diverse effects on cancer cells (4), AUY922 primarily induced caspase-dependent mitochondrion-mediated apoptosis in mutant KRAS colon cancer cells. This bears important implications, in that the ultimate goal of treatment of cancer is to eradicate cancer cells. While our results showing that knockdown of Bax blocked induction of cell death consolidated the importance of the mitochondrial apoptotic pathway in AUY922-triggered killing, Bax-deficient colon cancer cells have been shown to commit to necrosis upon treatment with the other HSP90 inhibitor 17-AAG (10, 13, 39). This difference may be due to the different HSP90 inhibitors used in the individual studies. Indeed, comparative analyses have shown that there are profound differences in biologic consequences of treatment of colon cancer cells with 17-AAG and AUY922 (40). In support, although p53 and PUMA are required for 17-AAG–induced apoptosis of colon cancer cells (10), they were indispensable for killing by AUY922. Regardless, our results clearly demonstrated that Bim, but not PUMA or the other BH3-only protein Bik, played a fundamental role in AUY922-induced apoptosis of KRAS-mutant colon cancer cells.

An important finding of this study was that mutant KRAS colon cancer cells were more prone to AUY922-induced ER stress. Although ER stress has been found previously to contribute to apoptosis induced by HSP90 inhibition (41), the role of ER stress–induced Bim has not been documented. Consistent with the previous finding that ER stress activates Bim transcription through CHOP (19), we found that knockdown of CHOP inhibited AUY922-triggered upregulation of Bim. Indeed, chemical chaperones or overexpression of GRP78, which alleviates ER stress (30), attenuated upregulation of CHOP and Bim by AUY922. On the other hand, siRNA inhibition of GRP78 or XBP-1, two well-established prosurvival effectors of the UPR (14, 15), enhanced upregulation of Bim and apoptosis caused by AUY922. It seems therefore that other agents that induce ER stress may cooperate with HSP90 inhibition to induce apoptosis in mutant KRAS colon cancer cells. In support with our findings in mutant KRAS colon cancer, deletion of CHOP in a mouse model of mutant KRAS-induced lung cancer increases tumor incidence, suggesting that CHOP activity is a barrier to mutant KRAS-driven malignancy (42).

As a HSP90 client protein (9), Bcl-2 is reduced, as anticipated, by AUY922. However, this was observed only in mutant KRAS colon cancer cells, suggesting that the chaperone effect of HSP90 on Bcl-2 is highly selective and is associated with the mutational status of KRAS in colon cancer cells. Many cochaperones are involved in determining the selectivity of HSP90 on its clients (43–45). It is likely that signaling driven by mutant KRAS affects the activity of the cofactor needed for HSP90 to chaperone Bcl-2, thus making Bcl-2 more vulnerable to HSP90 inhibition.

How does AUY922 preferentially induce ER stress in mutant KRAS colon cancer cells? It has been reported that killing of mutant KRAS cancer cells by the HSP90 inhibitor 17-AGG or PU-H71 involves degradation of the serine/threonine kinase STK33 (6). However, degradation of STK33 appeared less likely to be involved in AUY922-induced ER stress–mediated apoptosis of mutant KRAS colon cancer cells, as treatment with AUY922 did not cause any significant change in the expression of STK33 in the cells (Supplementary Fig. S13). On the other hand, oncogenic activation of KRAS may promote protein synthesis as does activating mutations of BRAF (46), which represents an underlying mechanism of chronic ER stress in cancer cells by uncoupling the ER protein folding load with the ER protein folding capacity (46). In this case, inhibition of HSP90 conceivably increases the content of proteins that cannot be properly folded and thus exacerbates the ER stress condition more markedly in mutant compared with wild-type KRAS colon cancer cells.

Previous studies have shown that HSP90 inhibition reduces activation of Akt and ERK, which are downstream of multiple clients of HSP90, contributes to induction of apoptosis (24, 47, 48). In particular, Akt itself is known to be a HSP90 client protein (49). However, we found that activation of Akt and ERK was reduced by AUY922 in both mutant and wild-type KRAS colon cancer cells, whereas the latter were markedly less sensitive to apoptosis induced by the inhibitor, suggesting that inhibition of Akt and ERK may not play a major role in AUY922-induced apoptosis in mutant KRAS colon cancer cells. Moreover, our finding that colon cancer cells that harbor mutations in both BRAF and PI3K that is responsible for activation of Akt (WiDr cells) are resistant to apoptosis induced by AUY922 further highlights the selectivity of AUY922 for mutant KRAS colon cancer cells.

The cytotoxic effect of AUY922 on mutant KRAS colon cancer cells was mirrored by inhibition of the growth of mutant KRAS colon cancer xenografts, which, in accordance with the findings in colon cancer cell lines, was mediated by Bim and associated with ER stress. This suggests that clinical evaluation of the efficacy of AUY922 in patients with metastatic colon cancers carrying activating mutations of KRAS is warranted, and that agents that potentiate the apoptosis-inducing effect of Bim, such as inhibitors of the prosurvival Bcl-2 family proteins, may be considered for combination with AUY922 in the treatment colon cancer (50). As a precedent, although the HSP90 inhibitor ganetespib showed only modest clinical activity in patients with mutant KRAS NSCLC, combinations of the inhibitor and other agents have been proposed (37, 38).

No potential conflicts of interest were disclosed.

Conception and design: C.Y. Wang, X.D. Zhang, C.C. Jiang

Development of methodology: C.Y. Wang, S.T. Guo, X.G. Yan, L. Jin

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C.Y. Wang, S.T. Guo, J.Y. Wang, F. Liu, Y.Y. Zhang, X.G. Yan

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C.Y. Wang, J.Y. Wang, H. Yari, X.G. Yan, L. Jin, X.D. Zhang

Writing, review, and/or revision of the manuscript: C.Y. Wang, H. Yari, X.D. Zhang, C.C. Jiang

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C.Y. Wang, X.G. Yan

Study supervision: X.D. Zhang, C.C. Jiang

This study was supported by Cancer Council NSW, Australia (RG 15-08), which was awarded to X.D. Zhang. C.C. Jiang and L. Jin are recipients of Cancer Institute NSW Fellowships. X.D. Zhang is supported by a Senior Research Fellowship of NHMRC.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.