Synergistic PIM kinase and proteasome inhibition as a therapeutic strategy for MYC-overexpressing triple-negative breast cancer

SUMMARY

Triple-negative breast cancer (TNBC) is the BC subtype with the poorest clinical outcome. The PIM family of kinases has emerged as a factor that is both overexpressed in TNBC and associated with poor outcomes. Preclinical data suggest that TNBC with an elevated MYC expression is sensitive to PIM inhibition. However, clinical observations indicate that the efficacy of PIM inhibitors as single agents may be limited, suggesting the need for combination therapies. Our screening effort identifies PIM and the 20S proteasome inhibition as the most synergistic combination. PIM inhibitors, when combined with proteasome inhibitors, induce significant antitumor effects, including abnormal accumulation of poly-ubiquitinated proteins, increased proteotoxic stress, and the inability of NRF1 to counter loss in proteasome activity. Thus, the identified combination could represent a rational combination therapy against MYC-overexpressing TNBC that is readily translatable to clinical investigations.

eTOC Blurb:

Kunder et al. identifies a drug combination that targets PIM kinases and the proteasome as a potentially clinically viable tool to induce significant cytotoxicity in high-MYC TNBC cells. The findings encourage the proposed combination therapy to be further evaluated in other high-MYC cancer types that express PIM kinases.

INTRODUCTION

TNBC, accounting for ~20% of all BC cases, constitutes a collection of molecularly heterogeneous breast tumors that lack functional expression of the estrogen and progesterone receptors and do not exhibit overexpression of human epidermal growth factor receptor 2 (HER2) (Garrido-Castro et al., 2019). No molecular targets have been clinically established as efficacious targets for TNBC. This is in stark contrast to breast tumors positive for the expression of druggable hormone and HER2 receptors, for which less toxic, targeted therapies have been developed for clinical use (Waks and Winer, 2019). It has not been established what specific mutations might be required for TNBC to arise and progress to become clinically unmanageable; however, large-scale molecular profiling efforts in the past decade have reproducibly identified the MYC proto-oncoprotein as a factor significantly dysregulated in TNBC, compared with other BC subtypes, which is associated with significantly diminished progression-free survival rate in the context of neoadjuvant chemotherapy (Cancer Genome Atlas Network, 2012; Horiuchi et al., 2012).

MYC is a pleiotropic transcription factor that affects many pathways involved in tumorigenesis and tumor progression (Dang, 1999). MYC lacks readily identifiable small-molecule binding pockets and rationally designed small molecules that directly inhibit MYC have not been successfully developed for clinical use (McKeown and Bradner, 2014; Whitfield et al., 2017). Thus, numerous efforts have been directed at identifying immediately druggable vulnerabilities in MYC+ tumors. The PIM family of serine/threonine kinases, composed of three highly homologous members, PIM1, -2, and -3 (Nawijn et al., 2010) (hereafter, PIM), has emerged as such a clinically viable target. PIM possesses weak tumorigenic capabilities on its own (Van Lohuizen et al., 1989). However, PIM has been shown to function as a potent genetic enhancer of MYC in several mouse models (Jongchan Kim et al., 2010; Van Lohuizen et al., 1989; 1991; J. Wang et al., 2010). PIM phosphorylates several target proteins, including MYC, increasing its transcriptional activity (Pulverer et al., 1994; Sears, 2004; Y. Zhang et al., 2008), and 4EBP1 (Fox et al., 2003; Yang et al., 2013), negatively controlling its activity to promote the cap-dependent protein synthesis critical to the continued expression of otherwise short-lived pro-tumor factors. It has also recently been shown that PIM helps maintain the activity of NRF2, a master regulator of antioxidative response, in some cancer cells (Song et al., 2014; Warfel et al., 2016). Thus, PIM may cooperate with potent oncoproteins such as MYC by supporting the expression of NRF2-controlled antioxidant genes, thereby preventing the levels of tumor cell-intrinsic or therapy-induced oxidative stress from becoming lethal (Rojo de la Vega et al., 2018).

Four clinically relevant small-molecule PIM kinase inhibitors have been described to date: SGI-1776 (L. S. Chen et al., 2009), AZD1208 (Dakin et al., 2012; Keeton et al., 2014), PIM447 (Burger et al., 2015; Garcia et al., 2014), and INCB053914 (Koblish et al., 2018). These inhibitors are all pan-PIM kinase inhibitors designed to inhibit multiple PIM family members simultaneously. This strategy is based on the established genetic evidence that different PIM family members can functionally compensate for the loss of others (Mikkers et al., 2002; Nawijn et al., 2010; van der Lugt et al., 1995). Although the dose-escalating trials of the first-generation inhibitor SGI-1776 were terminated in 2010 due to an incidence of cardiac QTc prolongation, the past decade has seen the development of second and third-generation pan-PIM inhibitors (AZD1208, PIM447, and INCB053914). Some of these have entered clinical evaluations, and most recently, the next-generation pan-PIM inhibitors, including GDC-0339 (X. Wang et al., 2019) and inhibitor #28 (H.-L. Wang et al., 2019), have shown anti-tumor growth effects in animal models. Ongoing clinical observations indicate that the anticancer effects of PIM inhibitor monotherapy may be limited (Cortes et al., 2018). However, a recent screening approach showed that PIM447, a phase-1 inhibitor (Raab et al., 2014), could render PI3K inhibitor-resistant cells sensitive to alpelisib, a recently FDA-approved PI3K inhibitor, in some cancer cell lines (Le et al., 2016). PIM inhibition and the PI3K-AKT pathway inhibition could synergistically increase oxidative stress levels, resulting in increased cytotoxicity in tumor cells (Song et al., 2018). These observations suggest that PIM inhibition may be particularly effective in combination therapies. Indeed, a phase 1b/2 trial was initiated to investigate the safety and efficacy of combining PIM447 and alpelisib in multiple myeloma (MM) (ClinicalTrials.gov Identifier: NCT02144038).

It was recently shown that PIM expression (PIM1 and -2 in particular) was significantly elevated in clinical TNBC samples, compared with non-TNBC samples, and that it was associated with MYC expression and poor outcomes in TNBC patients (Brasó Maristany et al., 2016; Horiuchi et al., 2016). Single-agent treatment of various mouse models of TNBC with clinically relevant PIM inhibitors, used at the maximum tolerated dose (MTD), exerted a cytostatic effect in some models and resulted in modest inhibition of continued tumor growth in others (Brasó Maristany et al., 2016; Horiuchi et al., 2016). In the present study, we set out to identify a clinically viable targeted combination therapy based on PIM kinase inhibition and investigate any potential mechanisms of drug synergy. We find that PIM inhibition is distinctly synergistic with inhibition of the proteasome. Such drug synergy is achieved via the disruption of protein homeostasis and the inability of MYC+ TNBC cells to resist proteasome inhibition when PIM is also inhibited.

RESULTS

Drug combination screens.

To identify and preclinically evaluate PIM inhibitor (PIMi)-based targeted combination therapies that have the potential to advance into clinical studies, we conducted two independent drug screens in the MDA-MB-436 cell line. MDA-MB-436 abundantly expresses MYC and PIM (Fig. 1A), belongs to the poor outcome-associated mesenchymal stem-like TNBC subtype (Kalimutho et al., 2015; Lehmann et al., 2011), and has been identified as one of the TNBC cell lines most chemo-resistant to the FDA-approved drugs, including paclitaxel and doxorubicin (Heiser et al., 2012; Wali et al., 2016). First, we conducted a single-dose combination screen to identify those drugs among the 600 drugs in the FDA-approved Drug Library (MedChem Express, not restricted to anticancer drugs) whose cytotoxic effects could be increased by > 50% when combined with PIM447 at 5μM (Fig. 1B), at which concentration PIM447 does not affect cell viability by itself (Fig. 2C). Thirty-four drugs were identified as hits (Fig. 2A and Table. S1). PIM447 was chosen because of its ability to potently inhibit PIM2, in addition to PIM1 and -3, which is considered more challenging to inhibit sufficiently (Burger et al., 2015; Garcia et al., 2014). Second, we conducted an independent combination screen in which we performed 10-dose × 10-dose drug combination studies between PIM447 and a pool of 28 anticancer drugs that target a wide range of pathways, including metabolism, cell cycle, cell survival/death, epigenetics, DNA repair, and protein homeostasis (Fig. 1B). These agents are either FDA approved or currently under clinical evaluation (phase 1–3) and have shown single-agent activity in preclinical BC models or MYC-driven cancer. We used the established Bliss Independence model and Chou-Talalay method (BLISS, 1939; Chou, 2010; Chou and Talalay, 1984; Foucquier and Guedj, 2015) to quantify the extent of synergy between the drugs. Carfilzomib (Demo et al., 2007), an FDA-approved, second-generation, irreversible 20S-proteasome inhibitor (20Si), was identified as the most potent combination partner for PIM447 (Fig. 2B and Table S2). Carfilzomib was recently shown to achieve more complete and longer-lasting proteasome inhibition than other FDA-approved proteasome inhibitors (Besse et al., 2019).

Figure 1. Drug combination screens performed in a chemo-resistant human TNBC cell line, MDA-MB-436.

(A) Protein expression levels of MYC and the PIM family of kinases (PIM1, -2, -3) in a panel of human TNBC cell lines and non-immortalized, non-tumorigenic human mammary epithelial cells (HMEC). Actin serves as a loading control.

(B) Schematic representation of the drug combination screens performed in this study.

Figure 2. Identification of the PIMi/20Si combination as the most synergistic drug combination.

(A) Results from Screen 1 (Fig. 1B) performed in the MDA-MB-436 cell line. The top 20 out of 34 hits identified (out of 600 tested) are listed. Carfilzomib, which was pursued in this study, is highlighted in red.

(B) Results from Screen 2 (Fig. 1B) performed in the MDA-MB-436 cell line. Carfilzomib, which was pursued in this study, is highlighted in red (n = 3).

(C) Left and middle columns: Representative 10 × 10 dose-response matrices showing percent viability and Bliss analysis of predicted versus observed cell viability in the indicated TNBC and non-tumorigenic HMEC cells treated with PIM447 and carfilzomib. Bliss scores > 0 = synergy, close to zero = additivity, and < 0 denote antagonism. Right column: Dose-response curves, determined by CellTiter-Glo assay (Promega), showing the effect of PIM447 alone, carfilzomib alone, or both combined on the indicated TNBC and HMEC cells (n = 3). Combination index (CI) was calculated according to the Chou-Talalay method using the CompuSYN software, where CI < 1 denotes synergy, CI = 1 denotes additivity, and CI > 1 denotes antagonism.

(D) Representative western blots showing expression of PIM1, -2, and -3 in MDA-MB-436 cells transiently transfected with Cas9 proteins and the indicated PIM-specific, synthetic multi-guide sgRNA or a nonspecific control sgRNA. Actin serves as a loading control. The resulting cells represent heterogenous KO pools and not single-cell clones.

(E) The effects of carfilzomib (CFZ) on relative cell viability in the cells shown in (D) (n = 3). Cas9/PIM-specific sgRNA-transfected cells were treated with CFZ for 72 h before subjected to CellTiter-Glo assay. Error bars represent means +/− SEM. P-values were calculated by a two-tailed t-test. **p < 0.01, ***p < 0.001, ****p < 0.0001.

Among the synergistic combinations identified, we pursued the combination that targets PIM and the 20S proteasome (the PIMi/20Si combination). This combination was chosen because it exhibited the highest Bliss excess scores and the Combination Index (CI) values indicative of strong synergism (Figs. 2B and 2C). Such synergism was also attainable in several MYC+ TNBC lines (Fig. S1) and with another PIMi GDC-0339 (Fig. S2), which potently inhibits all PIM family members, or with ixazomib (Kupperman et al., 2010) (Fig. S3), an FDA-approved, reversible 20Si. Importantly, neither cell proliferation nor cell viability was affected by the PIMi/20Si combination in non-immortalized, non-tumorigenic human mammary epithelial cells (HMECs) (Figs. 2C, S1B, S2, and S3). On the other hand, carfilzomib alone had a considerable antiproliferative effect on the more commonly used, hTERT-immortalized version of human mammary epithelial cells (hTERT-HME1), while no drug synergy between PIM447 and carfilzomib was observed in them (Fig. S1C). We next sought to determine whether the PIMi/20Si combination-induced cytotoxicity in MYC+ TNBC cells is dependent on their MYC status. We subjected two low-MYC TNBC lines (HCC38 and HCC1937) to the PIMi/20Si combination and observed no significant synergy in these lines (Fig. S4A). We next lentivirally overexpressed MYC in otherwise low-MYC TNBC cells (HCC38 and HCC1937) to examine whether elevating MYC levels alone is sufficient for the combination therapy to induce drug synergy. We observed moderate to substantial synergies in these TNBC cells engineered to overexpress MYC (Fig. S4B). We also subjected two low-MYC receptor-positive cell lines (BT-483 and HCC1419) to the combination therapy and found no synergy in them (Fig. S4A). These observations suggest that elevated MYC expression is needed for this combination therapy to induce robust drug synergy. Finally, we found that the synergistic effect of simultaneously inhibiting PIM and the proteasome was observable using a CRISPR-based genetic approach (Figs. 2D and 2E).

What sets the PIMi/20Si combination apart from other combinations identified is its ability to induce significant cytotoxicity in TNBC cells with both drugs used at doses significantly lower than their typical effective single-agent doses: Typical antiproliferative IC50 values for PIM447 in TNBC cell lines are 10–16 μM, whereas those for carfilzomib are 5–20 nM (Table S3). Carfilzomib is often used at 10–50 nM in cell-based cytotoxicity assays (Demo et al., 2007; Zang et al., 2012).

Pathways modulated by the PIMi/20Si combination.

The mechanisms of cell death induced by proteasome inhibitors have been under extensive investigation. One of the prevailing mechanisms is that proteasome inhibitors disrupt protein homeostasis by altering the balance between protein synthesis and protein degradation, which controls the abundance of essential proteins and amino acids recycled to make new proteins. The removal of misfolded/damaged proteins is also a critical role that the proteasome executes. Disrupting proteostasis can result in cell death mediated by a phenomenon known as a proteotoxic crisis, which involves the terminal activation of stress-responsive cell death mechanisms (Deshaies, 2014; Manasanch and Orlowski, 2017; Schneider and Bertolotti, 2015). Preclinically, it was previously shown that TNBC cells, but not non-tumor cells, were sensitive to loss of proteasome function (Petrocca et al., 2013); however, it was also observed clinically that the single-agent efficacy of proteasome inhibitors in solid tumors is limited (Manasanch and Orlowski, 2017). Thus, our combination screens suggest that the PIMi/20Si combination may prove particularly effective in inducing robust antitumor effects when either drug alone is unable (e.g., Fig. 2C).

To understand what molecular changes might underlie the mechanisms of drug synergy between PIM inhibitors and 20S-proteasome inhibitors, we employed unbiased label-free quantitative proteomics analysis (Drabik and Silberring, 2016), which revealed that the PIM447/carfilzomib combination, when used at a dose combination with robust synergy (Fig. 2C), rapidly increased the levels of proteotoxic stress, as indicated by a significant increase in the abundance of HSP70 (HSPA1B; HSPA1A) (Fernández-Fernández and Valpuesta, 2018) in MYC+ TNBC cells (Fig. 3A and Table S4). We found that the protein abundance of HSP70 could be increased 2- to 4-fold in the combination-treated MYC+ TNBC cells, compared with the DMSO or single-agent treated cells, but not in HMEC cells, and these results were reproduced with GDC-0339 (Fig. 3B). An increase in HSP70 levels was associated with cleavage of PARP, indicative of early-stage apoptosis (Kaufmann et al., 1993), in the combination-treated TNBC cells, whereas no such PARP cleavage was observed in HMEC cells (Fig. 3B). Interestingly, these non-immortalized HMEC cells, unlike the hTERT-immortalized counterpart (Fig. S1C), were highly resistant to carfilzomib (Fig. S5A). In addition to HSP70, BAG3, another proteotoxic stress-responsive chaperone protein (Meriin et al., 2018), was also increased in the combination-treated TNBC cells (Fig. S5B and Table S4).

Figure 3. Biological pathways affected by the PIM447/carfilzomib combination.

(A) Left: Venn diagram illustrating the effects of PIM447 alone, carfilzomib alone, or both combined on the proteome in MDA-MB-436 cells as determined by label-free quantitative proteomics analysis. Right: Volcano plots showing the distribution of >1600 proteins with their relative protein abundance in the indicated comparison groups. The HSPA1B; HSPA1A (HSP70) were determined as statistically significant proteins by a t-test using a permutation-based FDR method for multiple hypothesis corrections using LFQ intensity derived by MaxQuant. Significance lines: FDR<0.05.

(B) The effects of PIM447 or GDC-0339 alone, carfilzomib (CFZ) alone, or both combined on the abundance of HSP70 in the indicated TNBC lines and HMEC (non-tumor) cells. PARP cleavage indicates apoptosis. Actin serves as a loading control. Numbers in red indicate the relative protein expression of HSP70.

(C) Summaries of the cellular pathways dysregulated by PIM447 alone, carfilzomib alone, or both combined as determined by the Hallmark Gene Enrichment Analysis (top) or the Reactome pathway analysis (bottom). FDR: False discovery rate (FDR)-adjusted p-values.

The acute increase in the abundance of HSP70 and BAG3 indicates increased levels of proteotoxic stress induced by the combination therapy. However, an increase in the levels of these proteins is indicative of a cytoprotective mechanism and does not cause cytotoxicity. To understand the mechanisms by which the PIMi/20Si combination acutely induces cell death in MYC+ TNBC cells, we sought to determine what cellular pathways are affected by the combination. Subjecting our proteomics data to the Hallmark Gene Set Enrichment Analysis (Liberzon et al., 2015) in the Molecular Signature Database (MSigDB) and Reactome pathway analysis (Jassal et al., 2020) revealed that PIM447 at 5 μM negatively and prominently affected MYC signature genes and protein synthesis and moderately affected the unfolded protein response (UPR) pathway, whereas carfilzomib at 3 nM did not significantly affect many pathways. On the other hand, the PIM447/carfilzomib combination negatively and significantly affected MYC signature genes and those factors mediating overall mitochondrial health, the UPR pathway, protein quality control, and protein synthesis (Fig. 3C). The inhibitory effect of PIM inhibitor on MYC activity was also observable at the phosphorylation levels of MYC S62 (Fig. S5C), one of the PIM-controlled transcriptional activation sites (Y. Zhang et al., 2008). Our observations that the UPR pathway was downregulated by PIM447 alone or by the combination appear to be consistent with a recent report that MYC can directly maintain the activity of a cytoprotective component (i.e., IRE1/XBP1) of the UPR pathway in TNBC cells (Zhao et al., 2018).

The antitumor effects of PIMi-induced increase in reactive oxygen species (ROS).

Our observation that proteostasis is disrupted in the PIMi/20Si combination-treated cells may be, in part, consistent with recent reports that PIM controls the activities of antioxidant proteins regulated by NRF2 and that PIM inhibition results in a significant increase in ROS levels in cancer cells (Song et al., 2018; 2014; Warfel et al., 2016). A PIMi-mediated rapid and significant increase in ROS levels was achievable in MYC+ TNBC cells, whereas no such increase was observed in HMEC cells (Fig. 4A). The detectable ROS levels in the combination-treated cells were comparable to those found in the cells treated with the PIMi alone (Fig. 4A). Interestingly, our dose-response experiment showed that PIM447 at as low as 2.5 μM, at which concentration PIM447 does not exert cytotoxic effects in MYC+ TNBC cells (Figs. 2C and S1), achieved the same extent of increase in ROS levels that PIM447 at 10 μM achieved (Fig. 4B), indicating that a PIMi-induced increase in ROS levels alone does not pose significant antitumor effects in these cells. Based on these observations, we hypothesized that the origin of the observed proteotoxic stress induced by the PIMi/20Si combination might be a pool of oxidatively damaged proteins that continue to accumulate abnormally in MYC+ TNBC cells in the absence of proteasome function, leading to a proteotoxic crisis.

Figure 4. The effects of the PIM447/carfilzomib combination on the abundance of general reactive oxygen species (ROS) and accumulation of poly-ubiquitinated proteins.

(A) Representative graphs showing the effects of PIM447 alone, carfilzomib (CFZ) alone, or both combined on the abundance of ROS as determined by the CellRox Flow Kit (ThermoFisher) in the indicated TNBC lines and HMEC cells (left) and quantification of changes in ROS levels (right) (n = 3).

(B) The effects of PIM447 at indicated concentrations on ROS levels.

(C) Representative western blots showing the effects of PIM447 or GDC-0339 alone (5 μM), carfilzomib alone (3 nM), or both combined on the abundance of poly-ubiquitinated proteins (top) and quantification of western signals (bottom) (n = 3). Actin serves as loading as well as the normalization control.

(D) Western blots showing the effects of the PIM447 (5 μM)/carfilzomib (3 nM) combination on the abundance of poly-ubiquitinated proteins in the absence or presence of NAC at the indicated concentrations (top) and quantification of western signals (bottom) (n = 3). The cells were pretreated with NAC for 45 min before treatment with the combination. Actin serves as loading as well as the normalization control.

Error bars represent means +/− SEM. P-values were calculated by a two-tailed t-test. *p < 0.05 **p < 0.01, ***p < 0.001, ****p < 0.0001; N.S, not significant.

Consistent with our hypothesis, we found that the PIM447/carfilzomib combination uniquely increased the abundance of poly-ubiquitinated proteins in MYC+ TNBC cells, and the results were reproduced with GDC-0339 (Fig. 4C). Next, we sought to determine if the abnormal accumulation of poly-ubiquitinated proteins was due to increased ROS levels. To this end, we treated the cells with the PIM447/carfilzomib combination in the presence or absence of N-acetylcysteine (NAC), one of the most commonly used antioxidants, to determine if the presence of NAC could prevent the combination from causing abnormal accumulation of poly-ubiquitinated proteins. When used at a conventional 2 mM, NAC reduced the amount of protein aggregation by approximately 0–15%, whereas it did so by 20–35% when used at 5 mM (Fig. 4D). We found that NAC treatment, either at 2 mM or 5 mM, did not have measurable effects on the basal levels of poly-ubiquitinated proteins (Fig. S6A). Finally, we asked if the modest levels of rescue achieved with NAC could be reproduced using a genetic method. We found that reducing KEAP1 expression, an established method to activate NRF2-controlled antioxidant genes (Devling et al., 2005; D. D. Zhang, 2006), rendered the cells modestly but significantly less sensitive to the combination (Fig. S6B and S6C). These observations suggest that abnormal protein accumulation occurs in a manner that is partly dependent on increased ROS levels.

PIMi as a tool to counter the resistance mechanism associated with proteasome inhibition.

One of the proposed mechanisms by which cancer cells become resistant to proteasome inhibitors is via a compensatory upregulation of proteasome subunits by a stress-responsive transcription factor NRF1 (also designated as TCF11). In this scenario, resistance is driven by an evolutionarily established feedback mechanism by which cells transcriptionally induce proteasome subunits in an NRF1-dependent manner in order to ensure that proteasome activity meets demand, thereby limiting the therapeutic potential of proteasome inhibitors (Motosugi and Murata, 2019; Radhakrishnan et al., 2010). Currently, there is no clinically viable strategy to overcome this resistance mechanism. Our proteomics analysis shows that 20S inhibition, as expected, increased the abundance of many proteasome subunits, whereas, surprisingly, PIM inhibition markedly decreased that of several subunits (Fig. 5A). Interestingly, some of those that became downregulated upon PIM inhibition have been identified as MYC-controlled genes (X. Chen et al., 2008; Jonghwan Kim et al., 2008; Zeller et al., 2003) (Fig. 5A). Furthermore, PSMB4, one of the MYC-controlled subunits that became downregulated upon PIM inhibition (Fig. 5A), was identified previously as one of the essential proteasome subunits in oncogene-transformed cells, but not in non-transformed cells (Petrocca et al., 2013). PSMB4 was also identified as a proteasome subunit that was upregulated in a large number of tumor samples and could promote tumorigenesis in vivo (Lee et al., 2014). These observations raised the possibility that PIM inhibition becomes synergistic with proteasome inhibition in MYC+ TNBC cells, in part by lowering MYC activity (e.g., Figs. 3C and S5C), resulting in decreased expression of some of the most critical proteasome subunits that NRF1 may not control. If this is the case, PIMi might contribute to sensitizing MYC+ TNBC cells to proteasome inhibition.

Figure 5. The effects of the PIM447/carfilzomib combination on proteasome activity.

(A) Heat map representation of the effects of PIM447 alone (5 μM), carfilzomib alone (3 nM), or both combined on the relative abundance of proteasome subunits in MDA-MB-436 cells as determined by label-free quantitative proteomics analysis. MYC target gene set #1 is from (Zeller et al., 2003), #2 is from (X. Chen et al., 2008), and #3 is from (Jonghwan Kim et al., 2008). The list of the top 23 TNBC vulnerability genes is from (Petrocca et al., 2013).

(B) Western analysis of the effects of PIM47 alone (5 μM), carfilzomib (CFZ) alone (3 nM), and both combined on NRF1 expression in the indicated TNBC lines. Actin serves as a loading control.

(C) The effects of PIM447 or GDC-0339 alone, carfilzomib alone, or both combined at the indicated concentrations on proteasome activity in the indicated TNBC lines and HMEC cells as measured by Proteasome 20S Activity Assay Kit (Sigma). The assay was run in duplicate (n = 3), and each data point corresponds to each assay well in a 96-well plate.

(D) Western blots showing the effects of carfilzomib in the presence or absence of MYC siRNAs (two siRNA sequences) on the abundance of poly-ubiquitinated proteins in the indicated TNBC lines (left) and quantification of western signals (poly-ubiquitin) (right). Actin serves as loading as well as the normalization control.

(E) The effects of carfilzomib in the presence or absence of MYC siRNAs on proteasome activity. The assay was run in duplicate (n = 4), and each data point corresponds to each assay well in a 96-well plate.

(F) The effects of carfilzomib on proteasome activity in the presence or absence of exogenous MYC activation in HMEC-MYC-ER cells. The cells were pretreated with 4OH-TAM for 24 h before treatment with carfilzomib.

Error bars represent means +/− SEM. P-values were calculated by a two-tailed t-test. *p < 0.05 **p < 0.01, ***p < 0.001, ****p < 0.0001; N.S, not significant.

To test this idea, we first determined how NRF1 responds to PIMi alone, 20Si alone, or both combined in MYC+ TNBC cells. We found that PIM447 alone did not induce NRF1 expression, whereas carfilzomib, as expected, did so markedly (Fig. 5B), suggesting that 20Si significantly lowered the overall proteasome activity and activated the resistance mechanism. We found that the PIM447/carfilzomib combination induced NRF1 expression noticeably more than did carfilzomib alone (Fig. 5B). Consistent with these observations, PIM447 significantly increased the ability of carfilzomib to lower proteasome activity in MYC+ TNBC cells (e.g., carfilzomib 2.5 nM vs. carfilzomib 2.5 nM + PIM447 5 μM, Fig. 5C), and the results were reproduced with GDC-0339 (Fig. 5C). These observations indicate that PIMi could help 20Si to counter the NRF1-mediated resistance mechanism associated with proteasome inhibition. While carfilzomib treatment resulted in a significant decrease in proteasome activity in HMEC cells, PIM447 did not increase the extent of such a decrease (Fig. 5C). Notably, in the presence of PIMi, carfilzomib 2.5 nM, which as a single agent does not cause substantial toxicity in MYC+ TNBC cells (Fig. 2C), was as effective as carfilzomib 5.0 nM, which as a single agent can cause significant toxicity (Fig. 2C), in inhibiting proteasome activity (Figs. 5C and Table S5). These observations suggest that drug synergy mechanisms involve a further decrease in proteasome activity achieved by the PIMi/20Si combination. Next, we sought to explore the possibility that the mechanism of drug synergy involves inhibition of MYC via PIM inhibition. To this end, we used MYC siRNA to determine whether MYC inhibition phenocopies PIM inhibition. We found that carfilzomib treatment in the presence of MYC siRNA caused abnormal accumulation of poly-ubiquitinated proteins in MYC+ TNBC cells (Fig. 5D), which was associated with a further decrease in proteasome activity (Fig. 5E). Furthermore, we found that exogenous MYC activation in non-immortalized HMEC cells containing a 4-hydroxytamoxifen (TAM)-activatable MYC-ER transgene (HMEC-MYC-ER cells) (Horiuchi et al., 2016) resulted in increased proteasome activity and significantly protected it from being decreased upon carfilzomib treatment (Fig. 5F). These observations reveal a previously unappreciated significance of MYC in contributing to maintaining proteasome activity and the potential utility of PIMi in sensitizing MYC+ TNBC cells to small-molecule proteasome inhibition.

In vivo efficacy of the PIM447/carfilzomib combination in orthotopic xenograft models of human TNBC.

To determine whether the drug synergy between PIMi and 20Si can be reproduced in vivo, we evaluated the effects of PIM447 alone, carfilzomib alone, or both combined on the continued growth of mammary tumors generated with TNBC cells. Two TNBC cell lines, MDA-MB-436 and BT-549, were percutaneously injected into the 4^th mammary glands of female non-obese diabetic-severe combined immunodeficient gamma (NSG) mice for tumor formation to occur (approximately 125–300 mm³ for MDA-MB-436 tumors, and 75–135 mm³ for BT-549 tumors, Fig. S7). The MTD of PIM447 is 100 mg/kg in mice (Burger et al., 2015; Garcia et al., 2014), and that of carfilzomib is 10 mg/kg (Caenepeel et al., 2018; Demo et al., 2007; Rosebeck et al., 2016). We followed the “30% rule” conventionally adopted in studying a new combination therapy and treated tumor-bearing mice with PIM447 at 30 mg/kg (daily, 6 days a week) and carfilzomib at 3 mg/kg (twice a week/two consecutive days) for 3 weeks. At these relatively lower doses, we found that the PIM447/carfilzomib combination almost completely inhibited the continued growth of fast-growing MDA-MB-436 tumors, whereas it induced modest tumor regression in slow-growing BT-549 tumors (Fig. 6A). Next, we tested if the cellular phenotypes induced by the PIMi/20Si combination in vitro, namely the accumulation of poly-ubiquitinated proteins and increased levels of HSP70 and NRF1, could be observed in drug-treated tumors. An appreciable increase in the levels of poly-ubiquitinated proteins and HSP70 was observable in MDA-MB-436 tumors isolated from mice that received 3 doses of PIM447 and 2 doses of carfilzomib (day 4 samples) (Fig. 6B). A marked increase in NRF1 expression was also seen in these tumors (Fig. 6B). In the tumor samples collected at the end of the 3-week drug treatment, various levels of increase in the amount of poly-ubiquitinated proteins, HSP70, and NRF1 could be observed (day 22 samples) (Figs. 6B and 6C). These results indicate that the mechanisms of drug synergy uncovered in our in vitro experiments hold true in vivo.

Figure 6. Antitumor effects of the PIM447/carfilzomib combination in vivo.

(A) In vivo growth of MDA-MB-436 and BT-549 orthotopic xenograft tumors in NSG mice treated with PIM447 alone, carfilzomib (CFZ) alone, or both combined at indicted doses for 3 weeks (MDA-MB-436, n = 5 mice in each treatment group; BT-549, n = 3~5 in each treatment group). Error bars represent means +/− SEM. P-values were calculated by linear regression analysis comparing the slopes.

(B) Western blots showing the effects of PIM447 alone, carfilzomib alone, or both combined on the abundance of poly-ubiquitinated proteins, HSP70, and NRF1 in the indicated tumor samples collected on day 4 (MDA-MB-436) or after 3 weeks of treatment (day 22) (MDA-MB-436 and BT-549). For day 4 samples, two tumor samples from two independent mice are represented for each treatment group. Numbers in red indicate relative protein expression. For day 22 samples, three samples from three independent mice are represented for each treatment group. Actin serves as a loading control.

(C) Quantification of western signals corresponding to poly-ubiquitinated proteins, HSP70, and NRF1 in day 22 samples in (B), where actin serves as a normalization control. Error bars represent means +/− SEM. P-values were calculated by a two-tailed t-test. *p < 0.05 **p < 0.01, ***p < 0.001; N.S, not significant.

DISCUSSION

PIM kinases have become recognized as readily druggable potential anticancer targets in several solid and liquid cancer types. Genetic interaction between PIM and a potent proto-oncoprotein MYC has been established in multiple animal models. Both PIM and MYC were shown to be overexpressed in clinical TNBC samples, making MYC overexpression a potential predictive biomarker of response to small-molecule PIM inhibitors. Despite growing excitement around the potential clinical utility of pan-PIM inhibitors, both preclinical and clinical observations thus far do not strongly support the idea that PIMi as a single agent could induce significant clinical benefit. Thus, the identification of PIM inhibition as a method to render PI3K-resistant cancer cells sensitive to alpelisib, a recently FDA-approved PI3K alpha-subunit specific inhibitor, was of considerable significance, and the outcomes from the phase 1b/2 study evaluating the safety and efficacy of the PIM447/alpelisib combination in MM (NCT02144038) will be informative. The present study identifies the PIMi/20Si combination as an exceptionally synergistic drug combination in MYC+ TNBC cells. The data presented in this study also suggest that this combination can be significantly more synergistic than the one targeting PIM and PI3K. However, the observations made in this study are not to be construed to discount the potential clinical utility of the PIM447/alpelisib combination, particularly in specific genetic contexts, because the present study did not take into consideration the mutational status of PI3K or the activation levels of the PI3K/AKT pathway in the cell models used.

Proteasome inhibitors represent some of the most successfully used clinical agents in hematological malignancies. Proteasome inhibitors are currently under active clinical investigations in combination with the standard of care agents in various solid tumor types. However, treatment strategies not based on specific pharmacodynamic or predictive biomarkers of response might not yield the desired antitumor effects at the doses tolerated in patients. The present study identifies two distinct but synergistic mechanisms by which the PIMi/20Si combination can induce significant antitumor effects, specifically in MYC+ TNBC. We find that the PIMi/20Si combination induces toxic levels of proteotoxic stress in MYC+ TNBC cells, which is associated with abnormal accumulation of poly-ubiquitinated proteins. It is probable that this is, at least partly, due to an increase in the amount of oxidatively damaged proteins upon PIM inhibition, which cannot be efficiently degraded when the proteasome is also inhibited. We also find that PIM inhibition becomes uniquely synergistic with proteasome inhibition in MYC+ tumors by lowering MYC activity, resulting in diminished proteasome activity due to decreased expression of a subset of proteasome subunits controlled by MYC (Fig. 7). Considerable synergy achieved by the PIMi/20Si combination likely depends on both of these mechanisms.

Figure 7.

Schematic representation of the proposed mechanisms of drug synergy between PIMi and 20Si in MYC+ TNBC cells.

The present study raises a number of additional questions. Our study is limited to a relatively small panel of BC cell lines. It is not established whether elevated MYC expression can be universally expected to render cancer cells resistant to proteasome inhibition. One line of reasoning found in MM literature is that MYC overexpression could make cancer cells more sensitive to proteasome inhibition because MYC can accelerate the rates of cell proliferation and protein synthesis while controlling the abundance of some of the pro-apoptotic proteins (McConkey and Zhu, 2008). The other line of rationale is that MYC overexpression renders cancer cells resistant to proteasome inhibition because it can transcriptionally upregulate some of the subunits, thereby lowering the effectiveness of proteasome inhibitors (Farrell and Reagan, 2018). A recent, sophisticated single-cell RNAseq analysis of clinical samples has shown that the MYC pathway activation was one of the molecular features most significantly associated with MM resistance to bortezomib-based therapies (Cohen et al., 2021). Thus, additional studies using clinical samples will increase our understanding of the significance of MYC in cancer cell resistance to proteasome inhibition. Secondly, our study does not address the issue of heterogeneity in tumor response to the proposed combination therapy. It is unknown whether the PIM447/CFZ combination allows some populations of tumor cells to continue to proliferate while inducing cell death in others. It is also unknown whether this combination therapy could be efficacious in more advanced in vivo tumor models such as PDX models, which better represent tumor heterogeneity found in human primary tumors. Finally, our study leaves a number of mechanistic questions unanswered. For example, it has not been determined how PIM controls NRF2 activity, nor has it been determined what proteins might be selectively damaged oxidatively upon PIM inhibition. How precisely MYC/PIM and NRF1 might cooperatively control the abundance of proteasome in cancer cells is also not understood and requires detailed, mechanistic investigations.

In summary, the present study identifies the PIMi/20Si combination as a mechanism-based drug combination that has the potential to be clinically investigated using the existing agents that have been tolerated in humans as single agents (PIMi) or FDA approved (20Si). Our observations also reveal a functional interaction between MYC and the proteasome, which has not been described previously, and raise the possibility that PIMi can serve as a clinically viable tool to help overcome one of the resistance mechanisms associated with proteasome inhibitors. In this respect, although our study was focused on MYC+ TNBC, the proposed therapeutic strategy could also hold promise against other MYC+ cancers that abundantly express PIM kinases.

SIGNIFICANCE

The current standard of care for TNBC patients is limited to a combination of conventional chemotherapy, radiation, and surgery, and 30–50% of patients experience early relapse and a significantly diminished 5-year disease-free survival. Many potential targeted therapies have been proposed to treat patients with TNBC and other difficult-to-treat cancer types; however, a lack of potency as a single agent has been a common obstacle, particularly in clinical settings. The present work identifies a clinically viable drug combination that targets PIM kinases and the proteasome as a highly synergistic, mechanism-based combination effective against a poor outcome-associated subset of TNBC. This work encourages the initiation of clinical studies evaluating the efficacy of this combination therapy in TNBC.

STAR METHODS text

RESOURCES AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and fulfilled by the Lead Contact, Dai Horiuchi (dai.horiuchi@northwestern.edu).

Materials availability

Antibodies, reagents and cell lines used in this study were obtained from the commercial or academic sources described in the attached Key resources table.

Data and code availability

The original proteomics data files have been deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org/) via the MassIVE partner repository (PXD026291). The raw data generated by Bliss analysis and the uncropped versions of western images are available from the corresponding author upon reasonable request. Software programs used in this study were obtained from the commercial or academic sources described in the attached Key resources table. This paper does not report original code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Human breast cancer cell lines and mammary epithelial cells

A panel of established human TNBC cell lines and non-immortalized HMEC-MYC-ER cells and their culture conditions were previously described (Horiuchi et al., 2016; Neve et al., 2006). HMEC-MYC-ER cells were treated with 500 nM 4-Hydroxytamoxifen (TAM) to induce MYC activation, as previously described (Horiuchi et al., 2016). Non-immortalized HMEC cells (without MYC-ER transgene) were purchased from Lonza (CC-2551) at passage 5 and maintained in Mammary Epithelial Basal Medium (MEBM), Phenol Red-free (Lonza, CC-3153), containing epidermal growth factor (5 ng/mL), hydrocortisone (500 ng/mL), insulin (5 μg/mL), bovine pituitary extract (70 μg/mL), transferrin (5 μg/mL), isoproterenol (10 μM), and GlutaMAX (ThermoFisher, 35050061). hTERT-immortalized human mammary epithelial cells (referred to as hTERT-HME1 [ME16C]) were purchased from ATCC (CRL-4010) and maintained in the complete MEBM. The MYC-overexpressing versions of HCC38 and HCC1937 cell lines were established by infecting the cells with the lentivirus prepared using the previously described pLVX-MYC plasmid (Horiuchi et al., 2012), Lipofectamine 2000 transfection reagent (Thermo Fisher, 11668019), ViraPower Lentiviral Packaging Mix (Thermo Fisher, K497500), and Lenti-X 293T cell line (Takara, 632180). The Universal Mycoplasma Detection Kit (ATCC, 30–1012K) was used to ensure that cells were not infected with mycoplasma.

Animal experiments

The animal efficacy experiments described in this study were approved by the Northwestern University Institutional Animal Care and Use Committee and were executed by the Preclinical Rodent Services team at the Center for Comparative Medicine of Northwestern University. MDA-MB-436 and BT-549 cell lines were tested by Charles River Research Animal Diagnostics Services for rodent infectious agents. To grow MDA-MB-436 and BT-549 tumors, the cells (7.5 × 10⁵ cells for MDA-MB-436, and 5 × 10⁶ cells for BT-549), resuspended in a mixture of PBS/Matrigel (Corning, 356231) at 1:1, were percutaneously injected into the fourth mammary glands of female NSG mice aged 6–7 weeks. The tumors were allowed to reach approximately 125–300 mm³ in volume for MDA-MB-436 tumors and 75–135 mm³ in volume for BT-549 tumors, at which time drug treatment was initiated. The animals were treated with the following inhibitors: PIM447, a pan-PIM kinase inhibitor obtained from Novartis Oncology, reconstituted in 50 mM sodium acetate buffer, pH 3.0; carfilzomib (PR-171), an irreversible 20S proteasome inhibitor purchased from TargetMol (T1795), reconstituted in a 50 mM sodium citrate solution, pH 3.0, containing 20% (v/v) (2-Hydroxypropyl)-β-cyclodextrin (Sigma, H5784-10ML). PIM447 was administered via oral gavage at 30mg/kg, daily, 6 days/week, for 3 weeks, and carfilzomib was administered via intravenous injection at 3mg/kg, daily, twice (two consecutive days)/week, for 3 weeks.

METHOD DETAILS

Western blot analysis

Tumor or cell lysates were prepared as previously described (Horiuchi et al., 2012). Cultured cells were washed with ice-cold PBS and lysed in radioimmunoprecipitation assay (RIPA) buffer (50 mM Tris, 150 mM NaCl, 0.5% sodium-deoxycholate, 1% NP-40, 0.1% SDS, 2 mM EDTA, pH 7.5) containing protease inhibitor cocktail (Thermo Fisher, A32955) and phosphatase inhibitors (Thermo Fisher, A32957). Isolated tumor tissues were first washed in ice-cold PBS and homogenized on ice using a powered tissue homogenizer (OMNI Tissue Master 125) in RIPA buffer containing protease inhibitors and phosphatase inhibitors. Protein concentrations were determined using the DC Protein Assay (Bio-Rad, 5000112) with BSA as the standard. The antibodies used in this study are listed in Key resources table. The ECL reaction was done using the Clarity Western ECL Substrates (Bio-Rad, 170561), and chemiluminescent signals were acquired with the Bio-Rad ChemiDoc Touch Imaging system equipped with a supersensitive CCD camera. Where indicated, unsaturated band intensities were quantified using Bio-Rad Image Lab software (Version 6.0.1 build 34). Actin was used as the loading control and for the quantification of western bands.

siRNA experiments

siRNA products were purchased from Thermo Fisher (Silencer Select siRNA). The specific products used in this study were MYC #1 (s9129), MYC #2 (s9130), Negative Control (#2, 439084), and GAPDH (7390850). Negative Control and GAPDH siRNA sequences were mixed at 1:1 and used as the sole nonspecific control. siRNA transfection was performed using Lipofectamine RNAiMAX Transfection Reagent (Thermo Fisher, 13778150) via the reverse transfection method according to the manufacture’s protocol. The reverse-transfection method requires that cells be added to assay plates that contain a mixture of siRNA molecules and transfection reagent. MYC siRNAs were used at 60 pmol per well in 6-well culture dishes or 2.4 pmol per well in 96-well plates. MYC siRNA transfected cells were treated with carfilzomib approximately 20 h after transfection for 6 h before being subjected to western analysis or proteasome activity assay.

CRISPR experiments

CRISPR experiments were performed using Gene Knockout Kits v2 (Synthego) for human PIM1, -2, -3, and KEAP1 (with each Kit containing a mixture of three, target-specific, synthetic multi-guide sgRNA sequences and purified Cas9 nuclease) and Lipofectamine CRISPRMAX Cas9 Plus transfection reagent (ThermoFisher, CMAX00008) according to the protocol (entitled “CRISPR editing of immortalized cell lines with RNPs using lipofection”) provided by Synthego. Non-targeting sgRNAs (Negative control scrambling sgRNA #1 and #2) were also purchased from Synthego and mixed at 1:1 before use. sgRNA/Cas9 reverse-transfection was performed in 24-well plates (3.9 pmol sgRNA + 3 pmol Cas9 per well), and the resulting heterogeneous populations of cells were expanded in 6-well plates before used for cell-based assays carried out in 384-well plates. The sgRNA/Cas9-treated cells were used in downstream assays within two weeks of transfection to avoid the expansion of undesired cell populations. Each biological replicate was initiated by sgRNA/Cas9 transfection.

Detection of general ROS

CellRox Green Flow Cytometry Assay Kit (ThermoFisher, C10492) was used to determine ROS levels in human TNBC cell lines and HMEC cells according to the manufacturer’s instructions. The reagent-stained cells were analyzed on the CytoFLEX Flow Cytometer (Beckman Coulter), and FlowJo (version 10.7) was used to visualize the acquired data.

Proteasome activity assay

The Proteasome 20S Activity Assay Kit (Sigma, MAK172-1KT) was used in 96-well plates to measure proteasome activity in human TNBC cells and HMEC cells according to the manufacturer’s instructions. Fluorescence signals were acquired using the Cytation 5 Cell Imaging Multi-Mode Reader (BioTek).

Small-molecule inhibitors and drug combination screens

The FDA-approved Drug Library (MedChem Express, HY-L022) was made available to this study through the High-Throughput Analysis Laboratory of Northwestern University. Novartis Oncology provided PIM447, and GDC-0339 was purchased from MedChem Express (HY-16976). The other 28 anticancer-targeted agents used in this study were purchased from Selleck Chemicals and TargetMol. For Screen 1 (Fig 1B, left column), 600 compounds were randomly chosen from the FDA-approved Drug Library and tested for their ability at 5 μM to promote drug synergy with PIM447 at 5 μM in MDA-MB-436 cells in 384-well plates. For Screen 2 (Fig 1B, right column), 28 anticancer agents were combined with PIM447 in 10 × 10 dose-response matrices in 384-well plates. Drugs were dispensed to 384-well plates using the Echo 550 acoustic liquid-transfer system (LABCYTE) or D300e Direct Digital Dispenser (TECAN). In both screens, relative cell viability was determined using the CellTiter-Glo 2.0 Cell Viability Assay (Promega, G9242) after 72 h of drug treatment, and luminescence signals were acquired by the Synergy Neo2 Hybrid Multi-Mode Reader (BioTek) or the Cytation 5 Cell Imaging Multi-Mode Reader (BioTek). For Screen 1, the drugs whose cytotoxic effects could be increased by > 50% when combined with PIM447 were identified as hits. For Screen 2, the Bliss independence model and Chou-Talalay method (BLISS, 1939; Chou, 2010; Chou and Talalay, 1984; Foucquier and Guedj, 2015) were employed to rank the drugs based on their ability to be synergistic with PIM447. In the Bliss independence model, percent viability values were converted to fraction affected (Fa). The predicted fractional growth inhibition of the drug combination is calculated using the equation F_A + F_B − (F_A × F_B), where F_A and F_B are the fractional growth inhibitions of the drugs A and B at a given dose. Bliss excess is the difference between the expected growth inhibition and the observed inhibition. Bliss excess > 0 = synergy, close to zero = additivity, and < 0 denote antagonism. Bliss sum is the sum of individual Bliss excess scores in the matrix of drug doses. The Chao-Talalay CI was calculated using the CompuSyn software (ComboSyn Inc), where CI < 1 denotes synergy, CI = 1 denotes additivity, and CI > 1 denotes antagonism.

Label-free proteomics

Sample preparation for liquid chromatography tandem mass spectrometry (LC-MS/MS) analysis was performed as follows. The cell pellets were lysed in the lysis buffer (0.5% SDS, 50 mM ammonium bicarbonate, 50 mM NaCl, Halt protease inhibitor), and the supernatants containing proteins were used for the BCA assay (Pierce BCA Protein Assay Kit, Thermo Fisher) to determine protein concentration. For each sample, 100 μg of proteins were concentrated and purified by acetone/TCA precipitation with 8 volumes of ice-cold acetone and one volume of trichloroacetic acid overnight at −20 °C. After washing with ice-cold acetone, resulting protein pellets were resuspended in 50 μl of 8 M urea in 400 mM ammonium bicarbonate, pH 7.8, reduced with 4 mM dithiothreitol at 50 °C for 30 min, and cysteines were alkylated with 18 mM iodoacetamide in the dark for 30 min. The solution was then diluted to < 2 M urea (final concentration), and trypsin was added at the final trypsin/protein ratio of 1:50 before overnight incubation at 37 °C with shaking. The resulting peptides were desalted using solid-phase extraction on a Pierce C18 Spin column and eluted in 80 μl of 80% acetonitrile in 0.1% formic acid. After lyophilization, peptides were reconstituted with 5% acetonitrile in 0.1% formic acid.

The obtained peptides were analyzed by LC-MS/MS using a Dionex UltiMate 3000 Rapid Separation nanoLC and a Q Exactive^™ HF Hybrid Quadrupole-Orbitrap^™ Mass Spectrometer (Thermo Fisher). Approximately 1 μg of peptide samples were loaded onto the trap column, 150 μm × 3 cm in-house packed with 3 μm C18 beads. The analytical column was a 75 μm × 10.5 cm PicoChip column packed with 3 μm C18 beads (New Objective, Inc). The flow rate was kept at 300 nL/min. Solvent A was 0.1% FA in water, and Solvent B was 0.1% FA in ACN. The peptides were separated on a 120-min analytical gradient from 5% ACN/0.1% FA to 40% ACN/0.1% FA. The mass spectrometer was operated in a data-dependent mode. The source voltage was 2.10 kV, and the capillary temperature was 320 °C. MS¹ scans were acquired from 300–2000 m/z at 60,000 resolving power and automatic gain control (AGC) set to 3×10⁶. The top 20 most abundant precursor ions in each MS¹ scan were selected for fragmentation. Precursors were selected with an isolation width of 2 Da and fragmented by higher-energy collisional dissociation (HCD) at 30% normalized collision energy in the HCD cell. Previously selected ions were dynamically excluded from re-selection for 20 seconds. The MS² AGC was set to 1×10⁵. All samples were run in duplicate.

Protein tandem MS data were queried for protein identification and label-free quantification (LFQ) against the SwissProt Homo sapiens database using MaxQuant v1.6.0.16 (Cox et al., 2014; Cox and Mann, 2008). The following modifications were set as search parameters: peptide mass tolerance at 6 ppm, trypsin digestion cleavage after K or R (except when followed by P), 2 allowed missed cleavage sites, carbamidomethylated cysteine (static modification), and oxidized methionine, protein N-term acetylation (variable modification). Search results were validated with peptide and protein FDR, both at 0.01. Proteins that were identified with >1 unique peptide (1,885 proteins) were subjected to further statistical analysis using Perseus software (version 1.6.13) (Tyanova et al., 2016). The LFQ intensity was log2 transformed, normalized by subtracting the median, grouped into 4 treatment groups (vehicle control, PIM447 alone, carfilzomib alone, and PIM447+carfilzomib), and filtered to retain proteins with at least 60% LFQ values in at least one group, which yielded 1,643 proteins. ANOVA and the t-test were used to identify proteins that were differentially expressed among the treatment groups.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical analysis methods used in this study are indicated in the figure legends. Unless otherwise indicated, statistical analyses were performed using Prism 9 (version 9.1.2) from GraphPad Software, Inc., taking into consideration the assumptions required for the respective tests. P < 0.05 was considered to indicate statistical significance throughout the study. All cell-based in vitro experiments were independently repeated at least three times in duplicate or triplicate depending on each assay. No statistical method was used to predetermine the sample size throughout this study. For animal experiments, efforts were made to achieve this study’s scientific goals with the minimum number of animals. With respect to randomization, for animal experiments, tumor-bearing mice of similar tumor burden were equally divided into the control and experimental groups for subsequent drug treatment. No experimental samples were excluded throughout this study, except for animals that experienced unexpected, acute illness or injury, per the veterinarian’s order.

Supplementary Material

3

Table S4. Differential expression analysis of quantitative mass spectrometry [Related to Figure 3]. Differential expression analysis was performed between the indicated treatment groups.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
βActin (clone C4)	Santa Cruz Biotechnology	Cat# sc-47778 HRP, RRID:AB_2714189
c-MYC (clone Y69)	Abcam	Cat# ab32072, RRID:AB_731658
c-MYC phospho-S62 (clone 33A12E10)	Abcam	Cat# ab78318, RRID:AB_1566069
PIM1 (clone ZP003)	Abcam	Cat# ab54503, RRID:AB_882031
PIM2 (clone D1D2)	Cell Signaling Technology	Cat# 4730, RRID:AB_2163921
PIM3 (clone D17C9)	Cell Signaling Technology	Cat# 4165, RRID:AB_1904094
HSP70 (clone 3A3)	Santa Cruz Biotechnology	Cat# sc-32239, RRID:AB_627759
Ubiquitin (clone P4D1)	Cell Signaling Technology	Cat# 14049, RRID:AB_2798376
NRF1/TCF11 (clone D5B10)	Cell Signaling Technology	Cat# 8052, RRID:AB_11178947
KEAP1	Proteintech	Cat# 10503-2-AP, RRID:AB_2892676
BAG3	Proteintech	Cat# 10599-1-AP, RRID:AB_2062602
PARP	Cell Signaling Technology	Cat# 9542, RRID:AB_2160739
anti-mouse IgG-HRP	Santa Cruz Biotechnology	Cat# sc-2055, RRID:AB_631738
anti-rabbit IgG-HRP	Santa Cruz Biotechnology	Cat# sc-2054, RRID:AB_631748
Chemicals, peptides, and recombinant proteins
PIM447	Novartis Oncology	N/A
GDC-0339	MedChemExpress	Cat# HY-16976
Carfilzomib	TargetMol	Cat# T1795
Ixazomib	Selleck Chemicals	Cat# S2181
Bazedoxifene	Selleck Chemicals	Cat# S2128
Crizotinib	TargetMol	Cat# T1661
Tazemetostat	Selleck Chemicals	Cat# S7128
Afuresertib	Selleck Chemicals	Cat# S7521
Pazopanib	Selleck Chemicals	Cat# S1035
Cabozantinib	TargetMol	Cat# T2586
Abemaciclib	TargetMol	Cat# T3111
Trametinib	TargetMol	Cat# T2125
Panobinostat	Selleck Chemicals	Cat# S1030
Venetoclax	Selleck Chemicals	Cat# S8048
Alpelisib	Selleck Chemicals	Cat# S2814
Dasatinib	Selleck Chemicals	Cat# S1021
Plerixaflor	Selleck Chemicals	Cat# S8030
Capmatinib	Selleck Chemicals	Cat# S2788
Ganetespib	Selleck Chemicals	Cat# S1159
CB839	Selleck Chemicals	Cat# S7655
Erlotinib	TargetMol	Cat# T0373
Alisertib	Selleck Chemicals	Cat# S1133
I-BET-762	Selleck Chemicals	Cat# S7189
Dabrafenib	TargetMol	Cat# T1903
Marimastat	Selleck Chemicals	Cat# S7156
GSK3326595	Selleck Chemicals	Cat# S8664
AZD6738	Selleck Chemicals	Cat# S7693
Everolimus	Selleck Chemicals	Cat# S1120
Prexasertib	Selleck Chemicals	Cat# S7178
Olaparib	Selleck Chemicals	Cat# S1060
YM155	TargetMol	Cat# T2111
Doxorubicin Hydrochloride	Sigma	Cat# D1515-10MG
FDA-Approved Drug Library	MedChemExpress	Cat# HY-L022
N-Acetyl-L-cysteine	ThermoFisher Scientific	Cat# A15409
ViraPower Lentiviral Packaging Mix	ThermoFisher Scientific	Cat# K497500
Cas9	Synthego	Cat# 0000000033
Lipofectamine CRISPRMAX Cas9 Plus transfection reagent	ThermoFisher Scientific	Cat# CMAX00008
(2-Hydroxypropyl)-β-cyclodextrin	Sigma	Cat# H5784
Matrigel Growth Factor Reduced Basement Membrane Matrix, Phenol Red-Free, LDEV-free	Corning	Cat# 356231
Critical commercial assays
CellRox Green Flow Cytometry Assay Kit	ThermoFisher Scientific	Cat# C10492
Proteasome 20S Activity Assay Kit	Sigma	Cat# MAK172-1KT
CellTiter-Glo 2.0 Cell Viability Assay	Promega	Cat# G9242
Gene Knockout Kits v2, human PIM1	Synthego	Cat# PIM1
Gene Knockout Kits v2, human PIM2	Synthego	Cat# PIM2
Gene Knockout Kits v2, human PIM3	Synthego	Cat# PIM3
Gene Knockout Kits v2, human KEP1	Synthego	Cat# KEAP1
Deposited data
Proteomics data (raw data)	This paper	PXD026291
Experimental models: Cell lines
Lenti-X 293T cell line	Takara	Cat# 632180
Human mammary epithelial cells	Lonza	Cat# CC-2551
hTERT-HME1 [ME16C]	ATCC	Cat# CRL-4010
MDA-MB-231	ATCC	Cat# HTB-26
MDA-MB-436	ATCC	Cat# HTB-130
MDA-MB-468	ATCC	Cat# HTB-132
HCC1143	ATCC	Cat# CRL-2321
HCC1806	ATCC	Cat# CRL-2335
HCC3153	Neve et al., 2006	N/A
BT-549	ATCC	Cat# HTB-122
HCC38	ATCC	Cat# CRL-2314
HCC1937	ATCC	Cat# CRL-2336
BT-483	ATCC	Cat# HTB-121
HCC1419	ATCC	Cat# CRL-2326
HMEC-MYC-ER	Horiuchi et al., 2016	N/A
Experimental models: Organisms/strains
Mouse: NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ	Jackson Lab	Cat# 005557
Oligonucleotides
Silencer Select siRNA, MYC #1	ThermoFisher Scientific	Cat# s9129
Silencer Select siRNA, MYC #2	ThermoFisher Scientific	Cat# s9130
Silencer Select siRNA, Negative Control #2	ThermoFisher Scientific	Cat# 439084
Silencer Select siRNA, GAPDH	ThermoFisher Scientific	Cat# 7390850
Negative control scrambling sgRNA #1	Synthego	Cat# 063-110-000-000
Negative control scrambling sgRNA #2	Synthego	Cat# 063-111-000-000
Recombinant DNA
pLVX-MYC	Horiuchi et al., 2016	N/A
Software and algorithms
Prism 9 (version 9.1.2)	GraphPad	https://www.graphpad.com/scientific-software/prism/
FlowJo (version 10.7)	BD Biosciences	https://www.flowjo.com/solutions/flowjo
CompuSyn	ComboSyn, Inc.	https://www.combosyn.com
Image Lab (version 6.0.1)	Bio-Rad	https://www.bio-rad.com/en-us/product/image-lab-software?ID=KRE6P5E8Z&source_wt=imagelabsoftware_surl
MaxQuant	Max-Planck-Institute of Biochemistry	https://www.maxquant.org/maxquant/
Perseus (version 1.6.13)	Max-Planck-Institute of Biochemistry	https://maxquant.net/perseus/
Molecular Signature Database (MSigDB)	Broad Institute, Massachusetts Institute of Technology, and Regents of the University of California	https://www.gsea-msigdb.org/gsea/msigdb/index.jsp
BioRender	BioRender	https://biorender.com

Highlights:

Combining PIM kinase and proteasome inhibitors induces synergies in high-MYC TNBC

PIM inhibition-induced ROS level elevation alone might be well-tolerated in TNBC cells

Both NRF1 and MYC can contribute to maintaining proteasome activity

The mechanisms of drug synergy involve increased proteotoxic stress