Luminespib

BAG3-dependent expression of Mcl-1 confers resistance of mutant KRAS colon cancer cells to the HSP90 inhibitor AUY922

Chun Yan Wang, Su Tang Guo, Amanda Croft, Xu Guang Yan, Lei Jin, Xu Dong Zhang, Chen Chen Jiang
1 School of Biomedical Sciences and Pharmacy, The University of Newcastle, NSW, 2308, Australia
2 School of Medicine and Public Health, The University of Newcastle, NSW, 2308, Australia
3 Correspondence: Chen Chen Jiang or Xu Dong Zhang, LS3-49, Life Sciences Building, University of Newcastle, Callaghan, NSW, Australia.

Abstract
Past studies have shown that mutant KRAS colon cancer cells are susceptible to apoptosis induced by the HSP90 inhibitor AUY922. Nevertheless, intrinsic and acquired resistance remains an obstacle for the potential application of the inhibitor in the treatment of the disease. Here we report that Mcl-1 is important for survival of colon cancer cells in the presence of AUY922. Mcl-1 was upregulated in mutant KRAS colon cancer cells selected for resistance to AUY922-induced apoptosis. This was due to its increased stability mediated by Bcl-2-associated athanogene domain 3 (BAG3), which was also increased in resistant colon cancer cells by heat shock factor 1 (HSF1) as a result of chronic endoplasmic reticulum (ER) stress. Functional investigations demonstrated that inhibition of Mcl-1, BAG3, or HSF1 triggered apoptosis in resistant colon cancer cells, and rendered AUY922-naïve colon cancer cells more sensitive to the inhibitor. Together, these results identify that the HSF1-BAG3- Mcl-1 signal axis is critical for protection of mutant KRAS colon cancer cells from AUY922-induced apoptosis, with potential implications for targeting HSF1/BAG3/Mcl-1 to improve the efficacy of AUY922 in the treatment of colon cancer. This article is protected by copyright. All rights reserved

Introduction
Colon cancer is one of the most common and deadly malignancies [1]. Despite recent advances in early diagnosis and the development of novel treatment approaches, the overall survival of patients with metastatic colon cancers remains disappointing [1]. This is closely related to resistance of colon cancer cells to systemic therapies resulting from oncogenic mutations of KRAS that drive activation of multiple downstream signalling pathways important for cell survival and proliferation [2]. Indeed, active mutations of KRAS are found in up to 50% of colon cancers that are predictive of intrinsic resistance to many therapeutic drugs including antibodies against the epidermal growth factor receptor (EGFR) [2,3].
Heat shock protein 90 (HSP90) is the most abundant molecular chaperone and is essential for folding, stabilization, and activation of a large number of proteins [4,5]. In particular, many mutant and overexpressed oncoproteins such as EGFR, mutant BRAF, and Akt are clients of HSP90 [6]. As such, targeting HSP90 using small molecule inhibitors appears a promising strategy in the treatment of cancer [4]. A number of HSP90 inhibitors have entered pre-clinical and clinical studies [7]. Of note, oncogene-driven cancer cells, including mutant KRAS colon cancer cells, are susceptible to HSP90 inhibition [8-10]. However, only transient responses to HSP90 inhibitors have been observed in patients with mutant KRAS tumours [11]. This is suggestive of the development of acquired resistance to HSP90 inhibitors by cancer cells [12]. Various mechanisms have been reported to be involved in acquired resistance of cancer cells to HSP90 inhibition [12-15]. For example, reactivation of the ERK- p90RSK-mTOR signalling pathway mediates acquired resistance of KRAS mutant non-small cell lung cancer (NSCLC) to the HSP90 inhibitor ganetespib [14], whereas the secondary Y142N mutation in HSP90 confers acquired resistance of MDA-MB-231 breast cancer cells to the HSP90 inhibitor PU-H71 [13]. It seems that resistance mechanisms to HSP90 inhibition may vary considerably depending on cell types and inhibitors in question.
Induction of apoptosis is a common mechanism by which therapeutic drugs kill cancer cells [16]. This is frequently mediated by activation of the mitochondrial apoptotic pathway that is regulated by the balance between pro- and anti-apoptotic Bcl-2 family proteins [16]. Although multiple pro-apoptotic Bcl-2 family proteins are responsive to inhibition of HSP90, we have previously demonstrated that activation of the BH3-only protein Bim through the endoplasmic reticulum (ER) stress response is responsible for apoptosis of mutant KRAS colon cancer cells induced by the HSP90 inhibitor AUY922 [8]. In this study, we have examined the resistance mechanism of mutant KRAS colon cancer cells to AUY922-induced apoptosis. We report here that mutant KRAS colon cancer cells with acquired resistance to AUY922-induceded apoptosis express increased levels of the anti-apoptotic Bcl-2 family protein Mcl-1 that is critical for survival of the cells. Moreover, we show that Mcl-1 upregulation is due to its increased stability mediated by HSP70 co-chaperone Bcl-2-associated athanogene domain 3 (BAG3), which is also upregulated in mutant KRAS colon cancer cells with acquired resistance to AUY922 as a consequence of upregulation of heat shock factor 1 (HSF1) by chronic ER stress.

Materials and Methods Cell culture
Human colon cancer cell lines provided by Professor Gordon Burns (Faculty of Health and Medicine, University of Newcastle, Australia) were described previously [17]. Individual cell line authentication was regularly confirmed every 6 months using the AmpFISTR Identifier PCR Amplification Kit from Applied Biosystems (Mulgrave, VIC, Australia) and GeneMarker V1.91 software (SoftGenetics LLC, State College, PA). The last test was carried out in February 2017.

Annexin V and Propidium Iodide staining
Staining with Annexin V and Propidium Iodide was performed as described previously [8]. In brief, cells were pelleted, washed twice with cold PBS and re-suspended in 1× binding buffer. Cells were then stained with FITC-conjugated Annexin V and PE-conjugated Propidium Iodide (BD Pharmingen™ NSW, Australia) for 15 min in the dark followed by the addition of 400 µl of binding buffer. Cells were analysed by flow cytometry within 1 hour.

Preparation of mitochondrial and cytosolic fractions
Methods used for subcellular fraction were similar to the methods previously described [18,19]. Cell pellets were suspended in five volumes of buffer A (20 mM HEPES–KOH (pH 7.5), 10 mM KCI, 1 mM Na-EGTA, 1 mM DTT, and 0.1 mM phenylmethylsulfonyl fluoride containing 250 mM
sucrose) supplemented with protease inhibitor cocktail tablets. After incubation on ice for 15 min, the cells were disrupted by passing them 15 times through a 22-G needle. After centrifugation twice at 750 × g for 10 min at 4 °C, the supernatant was collected and centrifuged at 10 000 × g for 15 min at 4 °C, and the resulting mitochondrial pellets were resuspended in buffer A. The supernatants of the last spin were further centrifuged at 100 000 × g for 1 h at 4 °C, and the resulting supernatants were designated as the S-100 cytosolic fraction.

Three-dimensional (3D) culture
3-D culture was carried out using the hanging drop technique as previously described [8]. Briefly, 500 cells were seeded into a Perfecta3D® hanging drop plate (3D Biomatrix, Ann Arbor, MI) and allowed to grow for at least 5 days. Cells were then stained with calcein AM and ethidium homodimer-1 (Life Technologies, Scoresby, VIC, Australia) for 24 hours followed by treatment as indicated. Spheroids were harvested onto slides, and examined with the Axiovert and Axioplan microscopes (Carl Zeiss, North Ryde, NSW, Australia). Diameters of spheroids were quantified using ImageJ software (NIH, Bethesda, MD).

Plasmid vector and transfection
The pCMV6-AC-GFP-Mcl-1, pCMV6-AC-GFP-BAG3 and pCMV6-AC-GFP-HSF1 vectors were from Origene (Australian Biosearch, WA, Australia). Transfection of cDNA was carried out as described previously [8].

Immunoprecipitation
Immunoprecipitation was performed as described previously [17]. Briefly, whole cell lysates were incubated with the primary antibody or corresponding IgG for 2 hours at 4˚C, followed by incubation with 50 µl of protein A/G beads (Thermo Fisher Scientific, Scoresby, VIC, Australia). The bound proteins were eluted and processed for Western blot analysis.

Mass Spectrometry
Whole cell lysates were subject to immunoprecipitation with the anti-Mcl-1 antibody. Eluted proteins were identified using a gel-based liquid chromatography-tandem mass spectrometry (LC/MS) approach [20]. A MASCOT database search was used to visualize and validate results.

Statistical analysis and data presentation
Statistical analysis was carried out using JMP Statistics Made Visual software. Differences between different groups were assessed using Student’s t-test. A P value less than 0.05 was considered statistically significant.

Results
Mcl-1 is upregulated and is important for survival of mutant KRAS colon cancer cells with acquired resistance to AUY922
Past studies have shown that the mutant KRAS colon cancer cell lines HCT116 (KRASG13D) and SW480 (KRASG12V) are relatively sensitive to apoptosis induced by the HSP90 inhibitor AUY922 with IC50 being 76.5nM and 126nM, respectively [8]. Nevertheless, prolonged exposure of HCT116 and SW480 cells to AUY922 at their individual IC50 concentrations resulted in stable cell populations that could be passaged in the presence of the inhibitor (Fig. 1A&B). The resistant cells could also grow in medium containing AUY922 in three-dimentional (3-D) cell culture in order to imitate tumour growth in vivo (Fig. 1C; Fig. S1). These data indicate the successful generation of HCT116 and SW480 sublines (HCT116.R and SW480.R, respectively) resistant to AUY922-induced apoptosis. Of note, HCT116.R and SW480.R cells were cross-resistant to apoptosis induced by the structurally unrelated HSP90 inhibitor, XL888 (Fig. 1D) [21].
We examined whether anti-apoptotic Bcl-2 family proteins are involved in promoting survival of HCT116.R and SW480.R cells. Remarkably, Mcl-1 was upregulated in HCT116.R and SW480.R cells compared with their corresponding parental counterparts (Fig. 1E). In contrast, Bcl-XL expression remained largely unaltered, whereas Bcl-2 that is a client protein of HSP90 was downregulated in HCT116.R and SW480.R cells (Fig. 1E) [20] . Silencing of Mcl-1 by shRNA triggered apoptosis that was associated with activation of caspase-3 and mitochondrial release of cytochrome C and Smac/DIABLO in HCT116.R and SW480.R cells (Figs. 1F&G; Fig. S2).
Moreover, it triggered cell death in HCT116.R and SW480.R cells grown in 3-D cultures (Fig. 1H; Fig. S3). Therefore, Mcl-1 plays an important role in survival of colon cancer cells with acquired resistance to AUY922. In support, treatment with the small molecule Mcl-1 inhibitor UMI-77 similarly induced cell death in HCT116.R and SW480.R cells (Figs. 1I&J; Fig. S4) [22]. Of note, Mcl- 1 silencing or treatment with UMI-77 rendered parental HCT116 and SW480 cells more sensitive to AUY922-induced apoptosis (Figs. 1K-1M). On the other hand, Overexpression of Mcl-1 protects treatment naïve HCT116 and SW480 cells from AUY922-induced apoptosis in short-term cell viability assays and long-term clonogenic assays (Fig. S5). These data suggest that Mcl-1 also plays a role in intrinsic resistance of colon cancer cells to HSP90 inhibitors.

Improved stability of the Mcl-1 protein in colon cancer cells with acquired resistance to AUY922-induced apoptosis
Having demonstrated the functional significance of Mcl-1 in survival of colon cancer cells with acquired resistance to AUY922, we focused on investigation of the mechanism responsible for upregulation of Mcl-1 in HCT116.R and SW480.R cells. The expression levels of Mcl-1 mRNA remained comparable in HCT116.R and SW480.R cells relative to HCT116 and SW480 cells, respectively (Fig. 2A), suggesting that upregulation of Mcl-1 in HCT116.R and SW480.R cells is due to a posttranscriptional increase. Indeed, the turnover rate of Mcl-1 in HCT116.R and SW480.R cells was decelerated as shown in cycloheximide (CHX)-chase assays (Figs. 2B & 2C). Moreover, the translational rate of Mcl-1 was comparable in both parental and resistant cells (Figs. 2D & 2E) [23].
Therefore, the increase in Mcl-1 is mediated by improved stability of the protein [24]. In accordance, polyubiquitination of Mcl-1, which is required for its proteosomal degradation [24], was reduced in HCT116.R and SW480.R cells compared with their individual parental counterparts (Fig. 2F)

BAG3 is upregulated and is responsible for improved stability of Mcl-1 in HCT116.R and SW480.R cells
We carried out mass spectrometry analysis of immunoprecipitates generated with an anti-Mcl-1 antibody in total protein extracts of HCT116 and HCT116.R cells (Fig. 3A). One of the proteins with increased amount co-precipitated with Mcl-1 in HCT116.R compared with HCT116 cells was BAG3 (Figs. 3A & 3B; Fig. S6), which is a co-chaperone of HSP70 that inhibits proteasomal degradation of HSP70 client proteins including Mcl-1 [25]. Indeed, HSP70 was also one of the proteins that were increasingly co-precipitated with Mcl-1 in HCT116.R cells in the mass spectrometry analysis (Figs. 3A & 3C; Fig. S6). The increased association between Mcl-1 and BAG3 or HSP70 in HCT116.R compared with HCT116 cells was confirmed by reciprocal co-precipitation with antibodies against BAG3 or HSP70 (Figs. 3D & 3E; Fig. S7). Similarly, there was increased association between Mcl-1 and BAG3 or HSP70 in SW480.R compared with SW480 cells (Figs. 3B-3E; Fig. S6 & S7).
We compared the expression levels of BAG3 in HCT116.R and SW480.R cells and their corresponding parental counterparts. As shown in Figure 3F, BAG3 was markedly upregulated in HCT116.R and SW480.R cells compared with HCT116 and SW480 cells, respectively, at both the protein and mRNA levels. The increase in BAG3 mRNA expression was due to a transcriptional mechanism rather than changes in its stability as its turnover rate remained comparable between resistant and parental cells as shown in actinomycin D-chasing assays (Fig. S8).
To investigate the functional significance of BAG3 in the increased stability of Mcl-1, we silenced BAG3 using two individual shRNAs in HCT116.R and SW480.R cells (Fig. 3G). Indeed, BAG3 silencing resulted in marked downregulation of Mcl-1 (Fig. 3G). This was associated with reduced stability of Mcl-1, as silencing of BAG3 shortened half-life time of the protein and increased Mcl-1 polyubiquitination (Figs. 3H & 3I; Fig. S9). In contrast, it did not alter the expression levels of Mcl-1 mRNA (Fig. 3J). Therefore, stabilization of the Mcl-1 protein by BAG3 is responsible for the increased expression of Mcl-1 in colon cancer cells resistant to AUY922. Of note, although both Mcl-1 and BAG3 were expressed at relatively low levels in HCT116 and SW480 cells, silencing of BAG3 similarly resulted in downregulation of Mcl-1 (Fig. 3K). In contrast, Overexpression of BAG3 protects treatment naïve HCT116 and SW480 cells from AUY922-triggered apoptosis in short-term cell viability assays and long-term clonogenic assays (Figs. 3L & 3M; Fig. S10). Collectively, these data indicate that BAG3 is also involved in maintaining de novo expression of Mcl-1 in colon cancer cells.

BAG3 protects colon cancer cells from apoptosis induced AUY922 through Mcl-1
Similar to silencing of Mcl-1, silencing of BAG3 induced apoptosis in HCT116.R and SW480.R cells grown in monolayer and in 3-D cultures (Figs. 4A & 4B; Fig. S11). This was due to downregulation of Mcl-1, as overexpression of Mcl-1 in HCT116.R and SW480.R cells abolished apoptosis induced by BAG3 silencing (Fig. 4C). Moreover, silencing of Mcl-1 overrides silencing of BAG3 in induction of apoptosis in the cells (Fig. 4D). These results suggest that BAG3 upregulation plays an important role in promoting survival of colon cancer cells with acquired resistance to HSP90 inhibitors.
Furthermore, silencing of BAG3, similar to silencing of Mcl-1, rendered HCT116 and SW480 cells more sensitive to AUY922-induced apoptosis (Fig. 4E), indicating that BAG3 also plays a role in intrinsic resistance of colon cancer cells to HSP90 inhibitors.
Chronic activation of the ER stress response is responsible for BAG3 upregulation through HSF1 in colon cancer cells with acquired resistance to AUY922
We have previously shown that treatment with AUY922 induces ER stress in sensitive colon cancer cells [8]. We examined whether HCT116.R and SW480.R cells have increased activation of the ER stress response. As shown in Figure 5A, the expression of GRP78 and cleaved (active) form of XBP- 1 mRNA, two commonly used indicators of activation of the ER stress response [26,27], were increased in HCT116.R and SW480.R cells compared with HCT116 and SW480 cells, respectively (Fig. 5A), suggesting that colon cancer cells with acquired resistance to AUY922 have adapted to chronic ER stress.
The XBP-1 branch of the ER stress response induces the expression of the transcription factor HSF1 that is known to regulate the expression of BAG3 [28]. We therefore tested whether activation of the ER stress response is involved in upregulation of BAG3 in HCT116.R and SW480.R cells. Indeed, HSF1 expression was increased at both the protein and mRNA levels in HCT116.R and SW480.R compared with HCT116 and SW480 cells, respectively (Fig. 5B; Fig. S12), which appeared to be critical for upregulation of BAG3 and Mcl-1, as silencing of HSF1 reduced the expression of BAG3 and Mcl-1 in HCT116.R and SW480.R cells (Fig. 5C). Moreover, silencing of HSF1 triggered apoptosis (Fig. 5D). On the other hand, overexpression of HSF1 resulted in increased expression of BAG3 and Mcl-1, and protected against apoptosis induced by AUY922 in HCT116 and SW480 cells (Figs. 5E & 5F). Silencing of XBP-1 reduced the expression of HSF1 at both the protein and mRNA levels in HCT116.R and SW480.R cells, suggesting that XBP-1 is responsible for upregulation of HSF1 (Figs. 5G & 5H).

Discussion
Although a number of mechanisms have been shown to contribute to acquired resistance of mutant KRAS cancer cells to HSP90 inhibitors, such as reactivation of the ERK-p90RSK-mTOR signalling pathway and secondary Y142N mutation in HSP90 [13,14], the results in this study highlight the importance of resistance to apoptosis in determining long-term responses of mutant KRAS colon cancer cells to HSP90 inhibitors. Conceivably, upstream resistance mechanisms identified previously may converge on blockade of apoptosis. For example, activation of ERK is known to promote the expression of Mcl-1 [29], which we found in this study to be important not only for survival of colon cancer cells with acquired resistance, but also for intrinsic resistance of colon cancer cells to AUY922 [30]. Indeed, past studies have shown that AUY922 induces apoptosis in colon cancer cells through activating the BH3-only protein Bim [8], which can be bound to and antagonized by all anti-apoptotic Bcl-2 family proteins including Mcl-1 [31].
The significance of Mcl-1 in survival of colon cancer cells in the presence of AUY922 was supported by its upregulation in the cells. This made Mcl-1 distinct from the other anti-apoptotic Bcl-2 proteins, Bcl-XL and Bcl-2. While the expression of Bcl-XL remained unaltered, as a HSP90 client protein, Bcl- 2 expression was reduced in the presence of AUY922. Intriguingly, resistance of mesothelioma and lung adenocarcinoma cells to apoptosis induced by the other HSP90 inhibitor ganetespib involves loss of Mcl-1 addiction, where inhibition of Bcl-2 reverses the resistance [32]. The relative importance of individual anti-apoptotic Bcl-2 family proteins in protection of cancer cells from HSP90 inhibition seems therefore to be cell type-dependent and is also related to the different inhibitors in question.
Regardless, our results showed that colon cancer cells selected for resistance to AUY922 were cross- resistant to the structurally unrelated HSP90 inhibitor XL888 [21], emphasizing the significance of resistance to apoptosis in protection of cells from HSP90 inhibition.
Although the expression of Mcl-1 can be regulated at multiple levels [24], our results pointed to the importance of posttranslational modifications of Mcl-1 in its upregulation in colon cancer cells with acquired resistance to AUY922. Strikingly, BAG3 that has been previously shown to be important in stabilization of Mcl-1 was co-precipitated with Mcl-1 [25], suggesting the potential involvement of BAG3 in upregulation of Mcl-1. In support, silencing of BAG3 resulted in downregulation of Mcl-1, which was associated with increased Mcl-1 polyubiquitination, shortened half-life time of the protein, and induction of apoptosis, in colon cancer cells with acquired resistance to AUY922. Moreover, silencing of BAG3 also downregulated Mcl-1 and rendered AUY922-naïve colon cancer cells more sensitive to killing by the inhibitor. Of note, although it has been reported that BAG3 inhibition similarly caused reduction in the expression of Bcl-XL and Bcl-2 [33], we did not observe any change in the levels of Bcl-XL and Bcl-2 expression. This was probably associated with slower rates of Bcl- XL and Bcl-2 protein turnover in comparison with Mcl-1 [34].
Colon cancer cells carrying mutant FBW7 are reported to be resistant to inhibition of HSP90, in which mutant FBW7 fails to degrade Mcl-1 leading to Mcl-1’s stabilization [35]. HCT116 and SW480 cells used in this manuscript harbour wild-type FBW7 [36]. Therefore, we believe that FBW7 may not play an important role in acquired resistance of colon cancer cells to AUY922.
Unlike other BAG family members such as BAG1 that increases HSP70 client protein degradation [37], BAG3 inhibits proteasomal degradation of HSP70 clients, which has been shown to require the interaction between BAG3 and HSP70 [38]. In accordance, BAG3 was also found to bind to HSP70, which was increased in expression, probably as a consequence of compensation for inhibition of HSP90 [39], in colon cancer cells with acquired resistance to AUY922. Of importance, BAG3 was also upregulated in resistant colon cancer cells compared with those naïve to HSP90 inhibition, further substantiating the role of BAG3 in protection of colon cancer cells against HSP90 inhibitors. In agreement with the previous finding that BAG3 is responsive to cellular stress through HSF1 [28], we found that HSF1 was upregulated, and silencing HSF1 reduced the expression of BAG3 and Mcl-1, and triggered apoptosis in colon cancer cells with acquired resistance to AUY922. Therefore, a signal axis of HSF1-BAG3-Mcl-1 is activated in colon cancer cells with prolonged exposure to AUY922 that is critical for survival of the cells.
Although HSF1 is known to be induced by inhibition of HSP90 [40], the mechanism involved remains elusive. On one hand, our past results have shown that AUY922 induces ER stress in mutant KRAS colon cancer cells [8]. On the other, induction of ER stress triggers the expression of HSF1 [41]. Indeed, colon cancer cells with acquired resistance to AUY922 displayed increased expression of GRP78 and the active form of XBP-1 mRNA, characteristic of activation of the ER stress response [26,27]. As a transcription factor, XBP-1 appeared necessary for HSF1 upregulation in resistant colon cancer cells, in that silencing of XBP-1 reduced the expression of HSF1 at both the mRNA and protein levels, in agreement with previous results on regulation of HSF1 by the ER stress response in other experimental systems [41]. Collectively, these results indicate that chronic activation of the ER stress response drives the signal pathway of HSF1-BAG3-Mcl-1 that promotes survival of colon cancer cells with acquired resistance to AUY922 (Fig. S13).

A practical implication of this study is targeting Mcl-1 to enhance the therapeutic efficacy of HSP90 inhibitors in the treatment of mutant KRAS colon cancer. Silencing of Mcl-1 kills colon cancer cells with acquired resistance to Luminespib grown in 3-D cultures, further highlighting the potential of inhibition of Mcl-1 in combination with HSP90 inhibitors in vivo [42]. Various Mcl-1 inhibitors are under development for the treatment of solid cancers [43,44]. Our results indicate that pre-clinical and clinical evaluations of combinations of Mcl-1 and HSP90 inhibitors are warranted. Moreover, the results also suggest that inhibition of BAG3 may be an alternative approach to improve the treatment effect of HSP90 inhibitors.