Dysregulation of the mevalonate pathway promotes transformation

Abstract

The importance of cancer metabolism has been appreciated for many years, but the intricacies of how metabolic pathways interconnect with oncogenic signaling are not fully understood. With a clear understanding of how metabolism contributes to tumorigenesis, we will be better able to integrate the targeting of these fundamental biochemical pathways into patient care. The mevalonate (MVA) pathway, paced by its rate-limiting enzyme, hydroxymethylglutaryl coenzyme A reductase (HMGCR), is required for the generation of several fundamental end-products including cholesterol and isoprenoids. Despite years of extensive research from the perspective of cardiovascular disease, the contribution of a dysregulated MVA pathway to human cancer remains largely unexplored. We address this issue directly by showing that dysregulation of the MVA pathway, achieved by ectopic expression of either full-length HMGCR or its novel splice variant, promotes transformation. Ectopic HMGCR accentuates growth of transformed and nontransformed cells under anchorage-independent conditions or as xenografts in immunocompromised mice and, importantly, cooperates with RAS to drive the transformation of primary mouse embryonic fibroblasts cells. We further explore whether the MVA pathway may play a role in the etiology of human cancers and show that high mRNA levels of HMGCR and additional MVA pathway genes correlate with poor prognosis in a meta-analysis of six microarray datasets of primary breast cancer. Taken together, our results suggest that HMGCR is a candidate metabolic oncogene and provide a molecular rationale for further exploring the statin family of HMGCR inhibitors as anticancer agents.

The classic hallmarks of cancer are intimately intertwined with an assortment of metabolic processes that a tumor cell effectively hijacks to enable malignant transformation (1–3). Recent work has highlighted the importance of fundamental metabolic genes and pathways in tumor development and progression (1, 4–6). A better understanding of their contributions to tumorigenesis, as well as how they can best be targeted for therapeutic intervention, will lead to improved cancer patient care.

As early as the 1920s (7), it has been known that many tumor cells rapidly metabolize glucose to supply their energetic requirements, even under conditions of high oxygen. For tumor cells, a key benefit of relying on glycolysis is that glucose breakdown also provides a carbon source for anabolic synthesis of critical biochemical precursors (8). In this manner, for example, acetyl-CoA is made available for the synthesis of several lipid building blocks, including mevalonate (MVA).

The MVA pathway is a complex biochemical pathway required for the generation of several fundamental end-products including cholesterol, isoprenoids, dolichol, ubiquinone, and isopentenyladenine (9). At the heart of this pathway is the rate-limiting enzyme, hydroxymethylglutaryl coenzyme A reductase (HMGCR). Both HMGCR and the MVA pathway received considerable attention decades ago, primarily through the Nobel Prize-winning efforts of Goldstein and Brown and the development of the cholesterol-lowering drugs known as statins (9, 10). Statin inhibition of HMGCR in normal cells triggers a robust homeostatic feedback response that ensures the cells upregulate and restore the MVA pathway (9), a mechanism that has been successfully exploited for over 20 years to control hypercholesterolemia (11).

A number of tumors have been reported to have either deficient feedback control of HMGCR, or increased HMGCR expression and activity (12, 13). Exogenous MVA administered to xenograft-bearing mice was also shown to promote tumor growth (14). Finally, recent epidemiological studies have shown that patients taking certain statins for cholesterol control displayed a decreased risk of developing some cancers (15–21). Taken together, these results suggest that HMGCR may play an important role in human malignancies. Indeed, recent transcriptional profiling demonstrated that cholesterol and lipid metabolism are linked to cellular transformation (22). However, it is not known whether dysregulation of HMGCR and the MVA pathway make a causal contribution to cancer etiology or whether their dysregulation occurs as a consequence of transformation.

In addition to being tightly regulated at many levels (9), HMGCR has also been shown to be alternatively spliced such that there are two isoforms: full-length HMGCR (HMGCR-FL) and a version lacking exon 13 (HMGCR-D13) (23). Coding for a small region of the catalytic domain, exon 13 includes several residues important for binding both substrates and statins (Fig. 1A). A functional SNP has been identified (rs3846662) that regulates the splicing of HMGCR (24). Interestingly, overall risk of developing colorectal cancer was recently shown to correlate with genotype at rs12654264, a SNP in linkage disequilibrium with rs3846662, in statin-users but not nonusers, suggesting that patients whose cells expressed more HMGCR-D13 did not experience as much protective benefit from statin use (21). Transcript expression of HMGCR-D13 has also been associated with a decreased cholesterol-lowering response in lymphocytes exposed to simvastatin, suggesting it is refractory to inhibition by statins (25). Further work indicated that HMGCR-D13 is enzymatically inactive (24), prompting speculation that it may interfere with the regular activity of HMGCR-FL (26). Investigations into the transforming activities of either HMGCR-FL or -D13, however, have not been reported. Therefore, we have explored the role HMGCR plays in cancer and assessed whether dysregulation of the MVA pathway can be causal to transformation.

Fig. 1.

HMGCR-FL and HMGCR-D13 transcript levels are upregulated in response to lovastatin exposure. (A) A schematic representation of HMGCR-FL and its splice variant, HMGCR-D13, at the genomic, transcript, and protein levels with real-time PCR primers used to assess total endogenous HMGCR and each of HMGCR-FL and HMGCR-D13 specifically. (B) mRNA from HepG2 cells was harvested for real-time PCR analysis of basal transcript expression to show that HMGCR-FL is expressed at higher levels than HMGCR-D13. mRNA was also harvested from HepG2 cells exposed to either ethanol vehicle control or 20 μM lovastatin for the times indicated. Transcript expression of total endogenous HMGCR (C), HMGCR-FL (D) and HMGCR-D13 (E) all increased over time after lovastatin exposure. Data is expressed as log ₂ ratios of expression in lovastatin-exposed cells compared to ethanol-treated cells. * p < 0.05; one sample t-test comparing to 0, i.e., no induced expression. All experiments were performed a minimum of three times and data represent means and standard deviations.

Results

To detect both HMGCR transcripts independently and specifically, we devised a real-time PCR strategy using exon junction-spanning primers (Fig. 1A). Additionally, to study HMGCR in a human system with a well-regulated MVA pathway, we chose to work with HepG2 cells that have been characterized extensively as an ideal model for the study of MVA pathway homeostasis (27, 28). We first determined that HMGCR-FL is expressed at higher levels than HMGCR-D13 in HepG2 cells (Fig. 1B). Next, to address whether transcript expression of these HMGCR isoforms is regulated similarly or differently, we exploited statins as a tool to block HMGCR activity and trigger the feedback response that up-regulates its expression (9). HepG2 cells were exposed to 20 μM lovastatin for 4 to 48 h, and mRNA was harvested for real-time PCR analysis. Similar to total HMGCR transcript expression (Fig. 1C), we show that both HMGCR-FL (Fig. 1D) and HMGCR-D13 (Fig. 1E) are upregulated in response to lovastatin exposure with similar magnitudes and kinetics, suggesting that they are coregulated.

To determine whether dysregulation of the MVA pathway and HMGCR activity can drive transformation, expression constructs were generated to ectopically express each HMGCR isoform (Fig. 2A). To ensure that the constructs would be expressed in a deregulated state, only the catalytic domain was included (cHMGCR-FL and cHMGCR-D13), without their common transmembrane domain that harbors negative regulatory elements (29). Using real-time PCR we measured higher HMGCR transcript levels in cell lines expressing the ectopic constructs (Fig. 2B). Though ectopic expression of cHMGCR-FL increased total FL transcript levels about twofold, ectopic expression of cHMGCR-D13 increased total D13 transcript levels about 300-fold, to a roughly equal amount as FL. When exposed to lovastatin, only cells expressing HMGCR-FL showed a decrease in upregulation of total endogenous HMGCR transcript expression (Fig. 2C). HepG2 cells expressing the cHMGCR-D13 variant upregulated total endogenous HMGCR transcript expression to a similar level as the GFP control cells.

Fig. 2.

Ectopic expression of cHMGCR-FL and cHMGCR-D13 in HepG2 cells. (A) Schematic of the ectopic HMGCR constructs. (B) Confirmation of cHMGCR-FL and cHMGCR-D13 ectopic expression by real-time PCR assessing both HMGCR-FL (left) and HMGCR-D13 (right) transcript expression. * p < 0.05; student’s t-test comparing ectopic construct expressing cell lines to the GFP control. Data represent means and standard deviations. (C) Real-time PCR analysis of endogenous HMGCR transcript levels in HepG2 cells expressing the indicated constructs and exposed to vehicle control or 20 μM lovastatin for 16 hours. * p < 0.05; student’s t-test comparing lovastatin-exposed cells to ethanol-exposed cells. Data represent means and standard deviations. (D) Immunoblot using Upstate/Millipore antibody #07-572 detecting HMGCR protein expression in the indicated HepG2 cell lines exposed to vehicle control or 20 μM lovastatin for 48 hours prior to being harvested for protein lysates as described in the materials and methods. 1—High molecular weight signal corresponding to oligomerized HMGCR; 2—Approximately 100 kDa band corresponding to endogenous HMGCR; 3—Approximately 60 kDa band corresponding to ectopic cHMGCR-FL; 4—Approximately 55 kDa band corresponding to ectopic cHMGCR-D13. Tubulin was assayed as a loading control. All experiments were performed a minimum of three times and one representative example is shown.

Antibodies to human HMGCR are not well established because most previous work on the protein was conducted in nonhuman systems. While a few are now available, their specificity has not been thoroughly validated. We therefore characterized two commercially available antibodies and showed one to be specific (Fig. S1). This antibody was used to confirm ectopic expression of our constructs (Fig. 2D).

To assess the impact of HMGCR-FL and -D13 on cancer cells, we assayed whether ectopic expression of either isoform could alter the growth and proliferation of HepG2 cells. Using both proliferation (Fig. 3A) and cell cycle analyses (Fig. 3B) we demonstrated that ectopic expression of cHMGCR-FL or cHMGCR-D13 does not affect the rate of cellular proliferation. Furthermore, when cells were exposed to common chemotherapy drugs, doxorubicin and taxol, we also showed that cHMGCR-FL and cHMGCR-D13 are not antiapoptotic in the same manner as BCL2 (Fig. 3C).

Fig. 3.

Dysregulation of the MVA pathway by ectopic cHMGCR-FL and cHMGCR-D13 expression increases transformation of HepG2 cells. HepG2 cells expressing the indicated constructs were seeded in proliferation assays (A) or for cell cycle analysis by fixed PI (B) but no difference in growth rate or cell cycle was observed. (C) Similarly, cells exposed to 2 μM doxorubicin or 10 nM taxol were not protected from cell death by expression of either cHMGCR-FL or cHMGCR-D13, whereas BCL2 inhibited cell death significantly. Cell death was assessed by measurement of subdiploid DNA content (% pre-G1) in fixed PI assays combined with flow cytometry. * p < 0.05; student’s t-test comparing ectopic construct expressing cell lines to the GFP control. Experiments were performed a minimum of three times and data represent means and standard deviations. (D) When seeded in soft agar assays, HepG2 cells expressing either cHMGCR-FL or cHMGCR-D13 formed more colonies than cells expressing the empty GFP vector control, which was normalized to 1 in each experiment. * p < 0.05; one sample t-test comparing ectopic construct expressing cell lines to the GFP control. Experiments were performed a minimum of three times and data represent means and standard deviations. To address transformation in vivo, the flanks of sublethally irradiated SCID mice were injected with HepG2 cells ectopically expressing the empty GFP vector and either cHMGCR-FL (E) or cHMGCR-D13 (F). In each case, the cHMGCR construct-expressing cells grew larger tumors more quickly than the GFP cells. * p < 0.05; student’s t-test comparing ectopic construct expressing cell lines to the GFP control. Data represent means and standard errors of the mean.

However, when cells were seeded in soft agar assays to assess their capacity for anchorage-independent growth we demonstrated that both the FL and D13 constructs increase transformation (Fig. 3D). These results were confirmed in vivo when HepG2 cells expressing either the empty GFP vector or one of the two cHMGCR constructs were injected into opposing flanks of sublethally irradiated severe combined immuno-deficient (SCID) mice. In both cases, we observed larger and faster-growing tumors in the flanks of mice that had been injected with HepG2 cells expressing one of the two cHMGCR isoforms compared to the GFP control (Fig. 3 E and F) and observed approximately equal levels of HMGCR ectopic protein expression in tumor extracts (Fig. S2).

To address the impact of HMGCR on cells of different origin, we assessed whether dysregulation of the MVA pathway could increase transformation in additional cell types. cHMGCR-FL and cHMGCR-D13 were introduced into transformed MCF7 breast carcinoma cells (Fig. S3), which were then seeded in soft agar to assess their ability to grow in an anchorage-independent manner (Fig. 4A). Ectopic expression of both cHMGCR-FL and cHMGCR-D13 increased the number of colonies formed. Interestingly, when cHMGCR-FL and cHMGCR-D13 were introduced into MCF10A cells (Fig. S3), we demonstrated that dysregulation of the MVA pathway could also drive transformation of immortalized, nontransformed breast cells (Fig. 4B). As was observed in HepG2 cells, we confirmed that ectopic expression of cHMGCR did not inhibit cell death in breast cells (Fig. S4). To test the capacity of HMGCR to potentiate transformation in an entirely different model we introduced cHMGCR-FL and cHMGCR-D13 into normal murine bone marrow or fetal liver cells. In this case, only cHMGCR-FL increased myeloid colony formation in methylcellulose, whereas cHMGCR-D13 did not (Fig. 4C). Finally, we further evaluated the oncogenic potential of HMGCR by determining if cHMGCR-FL or cHMGCR-D13 could drive transformation and cooperate with conventional oncogenes. To this end we performed a series of classic transformation experiments in primary wild-type MEF cells. Interestingly, both cHMGCR-FL and -D13 were able to cooperate with RAS, but not E1A, to promote foci formation (Fig. 4D). Together, these results clearly argue that HMGCR has oncogenic potential and support a causal role for HMGCR in tumorigenesis.

Fig. 4.

Dysregulation of the MVA pathway demonstrates the oncogenic potential of HMGCR. (A) Ectopic expression of either cHMGCR-FL or cHMGCR-D13 in transformed MCF7 cells potentiated anchorage-independent growth in soft agar. * p < 0.05; one sample t-test compared to normalized control. These experiments were performed a minimum of three times and data represent means and standard deviations. Representative images of plate quadrants are shown (magnification 1.6×). (B) Nontransformed MCF10A cells formed significantly more colonies in soft agar when ectopically expressing either cHMGCR-FL or cHMGCR-D13. * p < 0.05; one sample t-test compared to normalized control. These experiments were performed a minimum of three times and data represent means and standard deviations. Representative images of plate quadrants are shown (magnification 1.6×). (C) cHMGCR-FL and cHMGCR-D13 constructs in MSCV-YFP vectors were introduced into either normal murine bone marrow or fetal liver cells. Cells were plated in methylcellulose media containing myeloid cytokines. Three independent experiments (two with bone marrow and one with fetal liver) yielded similar results in which cHMGCR-FL increased, and cHMGCR-D13 decreased, myeloid colony formation. One representative experiment is shown; data represent means and standard deviations between duplicate plates. (D) Primary MEFs were infected with retroviral constructs carrying the indicated genes and drug-selected. Transformed foci formed after approximately three weeks and were counted and imaged. Two to four plates of each gene combination were scored and representative images are shown.

Having seen that ectopic HMGCR expression accentuated transformation of human breast cells (MCF-7 and MCF-10A), we next addressed the impact of HMGCR on breast cancer patient survival by performing a meta-analysis of six large primary patient microarray datasets, encompassing 865 patients (30–35). Patients were dichotomized on the basis of their HMGCR abundance and Kaplan–Meier survival curves were plotted with differences in outcome evaluated using a Cox proportional hazards model. This analysis complemented our cell line data by showing that high HMGCR levels were associated with poor patient prognosis and reduced survival, with a hazard ratio of 1.5 (95% CI 1.15 to 1.95; p = 0.0029) (Fig. 5 Upper Left).

Fig. 5.

High mRNA levels of MVA pathway genes correlate with poor prognosis and reduced survival of breast cancer patients. A meta-analysis of six primary breast cancer datasets encompassing 865 patients was performed to evaluate a relationship between patient prognosis and the mRNA expression of HMGCR, hydroxymethylglutaryl coenzyme A synthase 1 (HMGCS1), mevalonate diphosphate decarboxylase (MVD), farnesyl pyrophosphate synthase (FDPS), acetoacetyl-CoA thiolase 2 (ACAT2), and mevalonate kinase (MVK). Kaplan–Meier analysis demonstrates that higher mRNA levels of five out of six gene products correlates with poor prognosis and decreased patient survival.

To further assess if mRNA levels of other MVA pathway genes could be correlated to patient survival we selected five additional representative genes [hydroxymethylglutaryl coenzyme A synthase 1 (HMGCS1), mevalonate diphosphate decarboxylase (MVD), farnesyl pyrophosphate synthase (FDPS), acetoacetyl-CoA thiolase 2 (ACAT2), and mevalonate kinase (MVK)]. Remarkably, high levels of four out of five of these genes, HMGCS1, MVD, FDPS, and ACAT2, correlated with poor patient prognosis and reduced survival (Fig. 5). Whereas some of the genes are weakly correlated with each other (R between 0.2 and 0.4; Fig. S5), a multivariate model containing all genes (HR = 1.22) is a weaker predictor of patient survival.

In summary, we have shown that deregulated HMGCR accentuates transformation in tumor cells of different origin; promotes colony formation in nontransformed breast cells and normal, diploid hematopoietic cells; and cooperates with RAS to drive transformation of primary MEFs. Furthermore, we showed that high mRNA levels of MVA pathway genes negatively impacted patient prognosis. Taken together, our results demonstrate that dysregulation of HMGCR and the MVA pathway plays a key role in transformation and suggests that HMGCR is a candidate metabolic oncogene.

Discussion

Our preliminary analysis of HMGCR isoforms has shown that HMGCR-FL and HMGCR-D13 are coregulated at the mRNA level in response to lovastatin exposure (Fig. 1). Furthermore, HepG2 cells expressing ectopic cHMGCR-FL or cHMGCR-D13 showed decreased upregulation of basal, endogenous HMGCR protein (Fig. 2D). However, only the FL construct reduced statin-induced upregulation of HMGCR transcript (Fig. 2C), suggesting the levels of regulation of these two isoforms are distinct and complex. The level of ectopic cHMGCR-D13 transcript expression was comparable to the level of the ectopic FL transcript expression in the cell (Fig. 2B), suggesting that a mechanism may be in place to ensure a certain maximal capacity of HMGCR expression is not exceeded in cells with intact regulation of the MVA pathway. Additionally, because transcript levels were roughly equal (Fig. 2B) and cHMGCR-FL protein was more highly expressed than cHMGCR-D13 (Fig. 2D), it is possible that a D13 protein is relatively unstable and more readily degraded. Thus, the regulation and interplay of HMGCR-FL and -D13 may influence MVA pathway function and statin sensitivity.

To further elucidate the role of HMGCR in cellular transformation we conducted anchorage-independent growth assays in tissue culture systems and xenograft models in mice. Both cHMGCR-FL and cHMGCR-D13 increased colony formation (Fig. 3D) and tumor growth in vivo (Fig. 3 E and F) using transformed HepG2 cells. We also showed that both deregulated HMGCR isoforms increased transformation in transformed MCF7 cells (Fig. 4A) and nontransformed MCF10A breast cells (Fig. 4B), suggesting that HMGCR may have oncogenic potential. In contrast, cHMGCR-FL, but not cHMGCR-D13, increased myeloid colony formation of cells derived from murine bone marrow or fetal liver (Fig. 4C), indicating that some difference may exist in their transformative potential or tissue-specific activities. Most remarkably, however, both cHMGCR-FL and -D13 cooperated with RAS, but not E1A, to transform primary MEFs in traditional focus-formation assays.

To the best of our knowledge, no genes traditionally central to fundamental metabolic pathways have been shown to directly promote transformation. Importantly, we provide evidence demonstrating that dysregulation of the MVA pathway may have sufficient oncogenic potential to drive tumorigenesis as HMGCR itself appears to be capable of promoting the transformation of transformed, nontransformed, and normal cells alike. How the MVA pathway and HMGCR become dysregulated is not yet well defined but could occur at many levels (9). The MVA pathway as a whole is highly regulated in nontransformed cells, and loss of any one of those regulatory mechanisms may contribute to dysregulation of this critical pathway and ultimately drive tumorigenesis. Recently, for example, it was determined that HMGCR can be regulated through the transcriptional activity of hypoxia-inducible factor 1 alpha (36), a transcription factor that plays a fundamental role in the response to hypoxia and in tumor cell metabolism. This suggests that activation of upstream signaling cascades may deregulate HMGCR and the MVA pathway. Moreover, three different point mutations (generated by mutagenesis) have been shown to disrupt sterol-mediated degradation of HMGCR and thereby increase its total enzymatic activity (37). Regardless of the mechanism by which it occurs, our results demonstrate that dysregulation of the MVA pathway can make a causal contribution to cancer etiology.

As our work was in progress, three tissue microarray studies reported that higher expression of HMGCR correlates with favorable breast and ovarian cancer patient prognosis (38–40). However, the antibody used in their studies is one that we showed is unable to detect human HMGCR (Fig. S1). Therefore, to assess the impact of HMGCR expression on breast cancer patient prognosis in an independent manner, we performed a meta-analysis encompassing 865 patients from six primary patient microarray datasets that examined mRNA expression in breast cancers (30–35). Interestingly, high HMGCR mRNA levels correlated with poor patient prognosis and reduced survival (Fig. 5), supporting our cell work. The levels of four out of five additional MVA pathway genes were also significantly correlated with poor prognosis, suggesting the entire pathway may be dysregulated and contributing to tumor etiology.

These data are particularly provocative when considered with recent epidemiological studies that controlled for specific covariates and showed the use of certain statins may be associated with decreased risk of breast cancer (15–18). For example, use of lipophilic statins was associated with lower breast cancer incidence in a large group of over 150,000 postmenopausal women (15), and the risk of breast cancer recurrence decreased with increasing durations of postdiagnosis statin use (18). These studies largely show that statin use reduces risk and/or severity of breast cancer and support our own data linking high HMGCR levels and poor patient prognosis.

Finally, we also provide molecular rationale for the significant tumor-selective therapeutic index observed for statin-induced apoptosis, arguing for further research toward exploiting the statin family of inhibitors as anticancer agents. Our lab and others’ have shown that a subset of tumor cells undergo apoptosis when treated with physiologically achievable levels of statins (41–44), and we have recently shown that dysregulation of HMGCR and the MVA pathway can predict if multiple myeloma cells are sensitive to statin-induced apoptosis (44). With a better understanding of the mechanisms by which HMGCR and the MVA pathway are dysregulated, we will be poised to rationally and effectively combine statins with other anticancer agents to the benefit of patient care.

Although HMGCR and the MVA pathway have been the subjects of extensive research, their function and regulation in human settings relevant to cancer have, up to now, remained unknown. Here we provide evidence that dysregulation of the MVA pathway can promote transformation, suggesting that HMGCR is a candidate oncogene. Ultimately, we have also put HMGCR and the MVA pathway into the broader context of tumor cell metabolism and have shown how they may contribute to the array of metabolic changes a normal or precancerous cell undergoes to promote transformation.

Materials and Methods

Please refer to SI Text for detailed Materials and Methods.

Cell Culture, Vectors, and Reagents.

HepG2 cells were cultured in DMEM H21 and MCF7 cells in α-MEM; both supplemented with 10% FBS. MCF10A cells were cultured in DMEM/F12 supplemented with 5% HS, 100 μg/mL EGF, 5 mg/mL hydrocortisone, 1 mg/mL cholera toxin, and 10 mg/mL insulin. Retroviral particles were produced, and target cells infected, as described previously (41). Approximately equal levels of GFP positive cells were obtained after infection with all viral constructs as determined by flow cytometry. In all cases, pools of infected cells were used. To assess proliferation, 5 × 10³ cells were seeded subconfluently in 12-well plates and counted in triplicate wells over 4 d. For cell cycle, 5 × 10⁵ cells were grown subconfluently in 100 mm tissue culture plates, harvested, washed in cold PBS, and fixed in cold 80% ethanol prior to propidium iodide staining and FACS analysis. Lovastatin powder was a gift of Apotex Corp. (Mississauga, ON, Canada) and was activated as described previously (43).

Real-Time PCR.

Approximately 5 × 10⁵ cells were treated as indicated and harvested. RNA was extracted using TRIZOL Reagent (Invitrogen, Carlsbad, CA) and cDNA was synthesized with SuperScript II (Invitrogen). Real-time PCR was conducted using indicated primers (Table S1) and SYBR Green master mix (Applied Biosystems, Foster City, CA) in triplicate. Expression was determined relative to GAPDH.

Immunoblotting.

Approximately 5 × 10⁵ cells were treated as indicated. Cells were pelleted, washed in cold PBS, lysed, and isolated from precipitated cellular debris. DTT was added to a final concentration of 1M. 6× Laemmli’s loading dye was added at room temperature and samples were never boiled to limit aggregation of membrane proteins. Blots were probed with anti-HMGCR (Cat. #07-572; Millipore/Upstate, Billerica, MA; 1∶1,000 or Cat. #07-457; Millipore/Upstate; 1∶1,000; refer to Fig. S1 for characterization), antitubulin (Santa Cruz, Santa Cruz, CA; 1∶2,000), or anti-BCL2 (kindly provided by David Andrews, Hamilton, ON, Canada; 1∶2,000).

Colony and Foci Formation Assays.

To assess anchorage-independent growth, cells were plated in their respective media + 0.3% noble agar in 30-mm petri dishes containing an agar plug. Images were taken and colonies were quantified using ImageJ software (NIH, Bethesda, MD). To assess the colony formation of murine myeloid cells, hematopoietic progenitor cells (HPCs) were isolated from the bone marrow, followed by lysis of nonnucleated cells, or fetal liver, enriched for the CD24low population. HPCs were infected twice with concentrated retroviral supernatant in a prestimulation cocktail and plated in methylcellulose containing myeloid cytokines for 7 d. To assess the transformation of wild-type MEFs, cells were infected with retroviral particles three times prior to selection with either puromycin (2 μg/mL) or hygromycin (140 μg/mL) until no more death was observed. After approximately three weeks, plates were washed with PBS twice on ice and fixed with ice cold methanol for 10 min on ice. Foci were scored and imaged using a Typhoon 9410 scanner (GE Healthcare, Uppsala, Sweden).

Xenografts.

Seven-week-old male SCID mice (Ontario Cancer Institute) were sublethally irradiated (3 GY) and injected subcutaneously with 5 × 10⁶ cells suspended in 50% matrigel (BD Bioscience) to a final volume of 0.2 mL. Each mouse was injected with GFP cells in one flank and either ectopic cHMGCR-FL or cHMGCR-D13 cells in the other flank. Resulting tumors were measured using digital calipers.

Primary Patient Expression Analysis.

A meta-analysis of six breast cancer datasets (30–35) was performed, the details of which will be provided elsewhere (Boutros PC et al., in preparation). Briefly, the patients were dichotomized and a multivariate, meta-analytic Cox proportional hazards model was fit.