Background
Approximately 20 % of all breast cancers have increased expression of the ERBB2 (HER2/neu) oncogene [
1]. This overexpression is often due to chromosomal amplification and is associated with resistance to chemotherapy, increased recurrence and worse prognosis [
2]. Several studies have shown that a number of genes are frequently co-overexpressed or co-amplified along with HER2/neu [
1‐
5]. Several of these co-overexpressed genes have been shown in functional genomics studies to be required for HER2/neu-positive breast cancer cell survival. Since many of these critical genes have roles in fat metabolism and adipogenesis, HER2/neu-positive breast cancer cells possess higher levels of stored triacylglycerides (TAGs) as well as higher levels of saturated fatty acids compared to other cell types [
6,
7].
The lipogenic metabolism of HER2/neu-positive breast cancer cells may represent a therapeutic opportunity and have consequences that impact treatment (reviewed recently [
8,
9]). The pro-adipogenic transcriptional regulators, PPARγ binding protein (PBP) and RevERBα/NR1D1 cooperatively contribute to increased expression of pro-lipogenic enzymes and a unique, Warburg-like metabolism in HER2/neu-positive breast cancer cells [
6]. Overexpression of HER2 itself has also previously been shown to have pro-lipogenic effects translationally increasing protein production of acetyl-CoA carboxylase alpha (ACACA) and fatty acid synthase (FASN) [
10]. The result of these genetic changes is the constant production of fatty acids as a means to regenerate reducing equivalents for glycolysis. This occurs through the concerted action of malic enzyme (ME1) and malate dehydrogenase (MDH1), while PBP, NR1D1 and PPARγ orchestrate the sequestration of fatty acids in neutral lipids to avoid lipotoxicity [
6,
7]. Since the addition of physiological concentrations of exogenous saturated fatty acids, such as palmitate, induces cell death in HER2/neu-positive breast cancer cells at significantly lower concentrations than in other breast cancer cells or normal human mammary epithelial cells (HMECs) [
7], the lipogenic pathway is likely operating at maximum capacity in these cells. This sensitivity of HER2/neu-positive breast cancer cells to palmitate may be related to new epidemiological data that shows that a diet rich in saturated fatty acids is positively associated with the development of HER2/neu-negative disease, but not HER2/neu-positive disease [
11]. By modeling the lipotoxicity associated consequences of this physiology we sought to identify signaling pathways that are regulated by physiological concentrations of exogenous palmitate in HER2/neu-positive breast cancer cells and gain insights into the molecular mechanism and its relevance to disease prevention and treatment.
Methods
Cell culture and chemicals
Breast cancer cell lines SKBR3, BT474, HCC1569, MCF7 and MCF10A were obtained from the American Type Culture Collection (Manassas, VA, USA) in 2011. Cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Hyclone, Logan, UT) supplemented with 10 % fetal bovine serum (Hyclone) at 37 °C and 5 % CO2. MCF10A cells were cultured and infected with pLXSN-neu or vector control retroviruses as previously described [
6]. All cell lines were authenticated in March 2016 by the SUNY-Albany Center for Functional Genomics Molecular Core Facility using a short tandem repeat method (Promega GenePrint 10 system). The IRE1 inhibitor STF803010 was obtained from EMD Millipore (Billerica, MA). Trastuzumab was a generous gift from Genentech (San Francisco, CA). Sodium palmitate, MG132 and 4-phenyl butyrate (4-PBA) were obtained from Sigma-Aldrich (St. Louis, MO). Palmitate was solubilized in DMSO or ethanol and diluted in full growth medium (DMEM, 10 % FBS) prior to the treatment of cells. The palmitate concentration was chosen based on the sensitivity profiles of the different HER2/neu-positive breast cancer cell lines. 250 μM palmitate leads to a 70–80 % reduction in viability in SKBR3 cells. According to the literature, fasting FFA concentrations in plasma/serum are in the range of 300–600 μM [
12‐
14] with palmitate representing about one quarter of the total FFAs [
15,
16]. However, these concentrations vary extensively based on the analysis method used for quantification [
16]. Based on these studies, 250 μM palmitate was deemed to be in the physiological range. All solutions were prepared immediately before usage. For nuclei counts, cells were plated in 96-well plates and allowed to adhere overnight. After treatment and incubation cells were fixed with 2.5 % formaldehyde and nuclei were stained with 1 μg/mL Hoechst 33342 (Life Technologies, Grand Island, NY). Images of cells were acquired using the INCell Analyzer 2200 high-content imaging system and images were analyzed using the INCell Investigator software (GE Healthcare, Piscataway, NJ). For anchorage-independent growth cells were seeded in ultra-low attachment plates in the presence of 250
μM palmitate or vehicle control and incubated for 11 days. Viable cells were assessed using the Alamar Blue cell health indicator assay (Life Technologies, Grand Island, NY) [
17].
Microarray analysis
After 24 h of treatment with 250 μM palmitate or vehicle control, cells were harvested by trypsinization and total RNA was extracted using the RNeasy mini kit (Qiagen, Valencia, CA). The quality and the concentrations of total RNA were assessed using the Agilent Bioanalyzer (Agilent Technologies, Santa Clara, CA). Total RNA (100 ng) deemed to be of good quality (RNA integrity number (RIN) greater than 8) was processed according to the standard Affymetrix Whole Transcript Sense Target labeling protocol (Affymetrix, Santa Clara, CA). The fragmented biotin-labeled cDNA from three independent biological replicates was hybridized over 16 h to Affymetrix Gene 1.0 ST arrays and scanned on an Affymetrix Scanner 3000 7G using AGCC software. The resulting CEL files were analyzed for quality using Affymetrix Expression Console software and were imported into GeneSpring GX11.5 (Agilent Technologies) where the data was quantile normalized using PLIER and baseline transformed to the median of the control samples. The probe sets were further filtered to exclude the bottom 20th percentile across all samples as well as probe sets with expression levels with CV > 20 % across all replicates in a condition. The resulting entity list was subjected to a t-test with Benjamini-Hochberg FDR correction. The data files have been deposited in Array Express, Accession number: E-MTAB-2601.
For detection of neutral fat stores, cells were stained with 1 μg/ml 4,4-difluoro1,3,5,7,8-pentamethyl-4-bora-3a,4a-diaza-s-indacence (BODIPY 493/503; Life Technologies, Grand Island, NY). Cells were grown in 96-well plates, fixed with 2.5 % formaldehyde, stained with 1 μg/ml BODIPY 493/503 and counterstained for nuclei with 1 μg/mL Hoechst 33342. Cells were imaged using the INCell Analyzer 2200 (GE Healthcare, Piscataway, NJ) high-content imaging system. Images were analyzed using the INCell Investigator software.
Apoptosis assays
Induction of apoptosis was assessed as described previously using Apo-BrdU staining of fragmented DNA [
18]. Treated and control cells were harvested by trypsinization and fixed with 4 % formaldehyde in PBS for 20 min on ice, followed by permeabilization with 70 % EtOH overnight at -20 °C. The 3′-OH ends of fragmented DNA were enzymatically labeled with bromodeoxyuridine triphosphate (Br-dUTP, Phoenix Flow Systems, San Diego, CA), using terminal deoxynucleotidyl transferase (Roche Applied Science) in TdT reaction buffer containing 2.5 mM cobalt chloride (Roche Applied Science) for 1 h at 37 °C. Br-dUTP labeled fragmented DNA was detected using a FITC-conjugated anti-BrdU monoclonal antibody (1:20,BD Pharmingen, San Jose, CA). Cells were counterstained with 5 μg/mL propidium iodide (Sigma-Aldrich) in the presence of 0.015 U/mL RNase (Roche Applied Science) in PBS with 0.1 % Triton X-100 for 30 min at room temperature. Samples were analyzed on a BD LSR II Flow Cytometer (BD Biosciences, San Jose, CA). Percentage of dead cells was assessed by propidium iodide uptake. Cells were trypsinized and incubated in PBS containing 5 μg/mL propidium iodide (Sigma-Aldrich) for 15 min. PI content was determined by flow cytometry. Three independent biological replicates and a minimum of 10,000 events were acquired for each experimental condition. The data were analyzed using the FlowJo software package (Treestar Inc., Ashland, OR).
Cell cycle analysis
Cell cycle analysis was performed as described previously [
19], briefly, the cells were treated for 24 h with palmitate or vehicle control and harvested by trypsinization, followed by 90 % ethanol permeabilization overnight at -20 °C. Permeabilized cells were stained with 5 μg/mL propidium iodide (Sigma-Aldrich) in the presence of 0.015 U/mL RNase (Roche Applied Science) in PBS for 20 min at room temperature. Data were acquired on a BD LSR II Flow Cytometer (BD Biosciences, San Jose, CA) with a minimum of 10,000 events collected for each experimental condition and analyzed using the FlowJo software package (Treestar Inc., Ashland, OR).
Immunoblotting
Immunoblots were performed using standard protocols. Cells were lysed directly in Laemmli loading buffer, proteins were resolved by SDS-PAGE and transferred to PVDF membranes (EMD Millipore, Billerica, MA). Membranes were blocked in Tris buffered saline (TBS) containing 0.1 % Tween and 5 % non-fat powdered milk (TBS-T, 5 % milk). Primary antibody incubation was performed in TBS-T, 5 % milk overnight at 4 °C. Proteins were visualized using a species-specific HRP-conjugated secondary antibody and the ECL Plus chemifluorescent detection system (Thermo Fisher Scientific Inc., Waltham, MA) on a STORM scanner ((GE Healthcare, Piscataway, NJ). Signal was quantified using ImageJ software ((
https://imagej.nih.gov/). The following primary antibodies were used: PERK (#5683, Cell Signaling Technology, Inc., Danvers, MA)(1:1000), phospho-PERK (Thr980, #3179, Cell Signaling) (1:1000), PDI (#3501, Cell Signaling) (1:1000), BiP (#3177, Cell Signaling) (1:1000), eIF2
α (#5324, Cell Signaling) (1:1000), phospho-eIF2α (Ser51, #3398, Cell Signaling) (1:1000), DDIT3/CHOP (#2895, Cell Signaling) (1:500), HER2 (#4290, Cell Signaling), phospho-HER2 (Tyr1221/1222, #2243, Cell Signaling) (1:1000), HER3 (#4754, Cell Signaling) (1:1000), phospho-HER3 (Tyr1289, #4791, Cell Signaling) (1:1000), EGFR (#4267, Cell Signaling) (1:1000), phospho-EGFR (Tyr1068, #3777, Cell Signaling) (1:1000), GAPDH (#5174, Cell Signaling) (1:15000),
α-tubulin (MCA78G, AbD Serotec, Oxford, UK) (1:15000). Secondary antibodies: goat-anti-rabbit-HRP (#7074, Cell Signaling) (1:5000), horse-anti-mouse-HRP (#7076, Cell Signaling) (1:5000), goat-anti-rat-HRP (sc-2303, Santa Cruz Biotechnology, Inc., Santa Cruz, CA) (1:15000).
Statistical and computational analyses
The impact of molecular perturbations (inhibitors, shRNAs) on breast cancer cell phenotypes including cell survival, proliferation, apoptosis resistance, was measured in three independent experiment, from replicate cultures. Means and standard deviations were calculated and results were analyzed for statistical significance using Student’s t-tests for pairwise comparisons or one-way ANOVA with Bonferroni post test, when more than two experimental groups were compared. P values < 0.05 were considered statistically significant.
Gene ontology (GO) enrichment analysis was carried out using the DAVID Bioinformatics Resource [
20,
21]. Logistic regression analysis for transcription factor motif enrichment was performed using the free web service LRPath (
http://lrpath.ncibi.org/). LRpath functionally relates the odds of gene set membership (dependent variable) with the statistical significance of differential expression (independent variable) using logistic regression, and calculates q-values using the FDR method as a measure of statistical significance [
22]. The False discovery rate (FDR) is a statistical method when performing multiple comparisons used to control the expected proportion of rejected null hypotheses that were incorrect rejections (“false discoveries”) [
23]. The network neighborhood of enriched transcription factors was obtained by querying the STRING database [
24].
Transfections and reporters
For pCAX-XBP1-ΔDBD-venus reporter construct assays [
25], cells were seeded in 96-well plates and allowed to adhere overnight before they were transfected using XtremeGene HP (Roche), according to the manufacturer’s instructions. Cells were treated as indicated in the individual experiments, 24 h post-transfection. Expression of the fluorescent protein is indicative of IRE1-mediated XBP1 splicing. The pCAX-XBP1-ΔDBD-venus reporter construct was a generous gift from Dr. Masayuki Miura, University of Tokyo. shRNA-mediated knockdown experiments were carried out using pSM2 constructs obtained from Open Biosystems. Transfections were performed as described for pCAX-XBP1-ΔDBD-venus experiments and cells were treated 48 h post-transfection. Proteasome activity was tested using the ZsProSensor-1 reporter construct which utilizes the ZsGreen1 fluorescent protein coupled to a proteasome-targeting sequence (Clontech, Mountain View, CA). Accumulation of fluorescence in the cell is indicative of decreased proteasome activity.
Discussion
Our previous work has shown that HER2/neu-positive breast cancer cells contain high levels of endogenous saturated fatty acids and neutral lipids and generally exhibit a pro-lipogenic phenotype [
6‐
9,
19]. This Warburg-like physiology relies on active fatty acid synthesis for survival and aggressive behavior. Since small amounts of exogenously supplied palmitate are toxic to HER2/neu-positive breast cancer cells, however, this phenotype might represent an Achilles heel [
7]. In this study, we used physiological concentrations of exogenous palmitate both to investigate molecular mechanisms of lipotoxicity associated with this Warburg-like physiology and to model high levels of dietary saturated fat. We identified pathways that are modulated by exogenous palmitate in the HER2/neu-positive SKBR3 breast cancer cell line and compared this response to that of HER2-normal MCF7 breast cancer cells, which have previously been shown to respond to exogenous palmitate in a way that is comparable to non-tumorigenic MCF10A mammary epithelial cells or normal human mammary epithelial cells (HMECs) [
7,
26]. This analysis was extended to additional HER2/neu-positive breast cancer cell lines and publicly available molecular profiling data of human breast tumors.
Exogenous palmitate induces distinct transcriptional responses in SKBR3 and MCF7 cells, which are congruent with the severity of the observed toxicity (Fig.
2 and [
6]). Most notably in HER2/neu-positive SKBR3 cells, exogenous palmitate induces a partial ER stress response through the activation of the IRE1-XBP1 and ATF6 axes, but not the PERK-eIF2α axis, which ultimately leads to CHOP-dependent cell death (Fig.
5). Increased levels of ER stress response markers are also found in HER2/neu-positive breast cancer cell lines and human breast tumors. The general trend of the three HER2/neu-positive breast cancer cell lines tested was that they possessed higher basal levels of DDIT3/CHOP, ATF6 and spliced XBP1 and that these levels increased in response to palmitate treatment (Additional file
1: Figure S7). These effects were not observed in the other breast cells tested. Additionally, analysis of molecular profiling data sets of invasive breast carcinomas generated by the TCGA Research Network [
1] revealed an association between the ER stress markers and the HER2/neu, NR1D1 and PBP genes linked to lipogenesis [
6‐
9,
19] in human breast tumors (Additional file
1: Figure S8). Together, these data point to ER stress as a potential consequence of increased palmitate levels in HER2/neu-positive breast cancer cells, although the precise pathway may not be clear at present. Several studies have reported that palmitate induces ER stress responses in liver and pancreatic beta-cells. Cao et al. report that palmitate induces ER stress and apoptosis in liver cells through the activation of PERK/ATF4/CHOP [
36]. Sommerweis et al. obtained similar results in pancreatic beta-cells, where palmitate induced PERK activation and eIF2α phosphorylation but no upregulation of ATF6 was detected [
37]. These studies have identified palmitate induced ER stress pathways that are somewhat different from what we have found in the HER2/neu-positive breast cancer cells. However, the overexpression of HER2, which has been linked to increased sensitivity towards ER stress-inducing agents [
38], may complicate the interpretation. Similar to our results, Young et al. report saturated fatty acid-induced ER stress and CHOP-dependent cell death in mouse embryonic fibroblasts (MEFs) with consitutively active mTOR signaling when the cells are cultured under hypoxic conditions [
39]. mTOR signaling is frequently activated in HER2/neu-positive breast cancer cells and tumors and may contribute to the lipotoxicity observed in HER2/neu-positive SKBR3 cells (reviewed in [
9]).
Determination of the molecular mechanism of palmitate in inducing ER stress and reducing HER2/HER3 levels will require additional study. Previous studies have suggested that exogenous palmitate exerts its toxic effects through the dysregulation of protein palmitoylation. Baldwin et al. found that palmitate-induced ER stress and apoptosis in beta-cells could be ameliorated by adding the palmitoylation inhibitor 2-bromopalmitate [
40] and the ER-resident chaperone, calnexin, has been shown to be functionally regulated by palmitoylation. Upon ER stress induction, palmitoylation of calnexin decreases, which promotes its chaperone function [
41]. However, this does not appear to be a relevant mechanism in HER2/neu-positive breast cancer cells. 2-bromopalmitate appears to be even more toxic to these cells than palmitate (Baumann et al., unpublished observations). These results are consistent with a metabolic effect as 2-bromopalmitate inhibits a variety of enzymes, some of which are required for TAG formation, a process that has been shown to be critical for the survival of HER2/neu-positive breast cancer cells [
6,
7,
42].
Given the altered metabolic phenotype of HER2/neu-positive breast cancer cells, which rely on active fatty acid synthesis for survival [
6,
7], it is possible that exogenous palmitate induces ER stress and decreases HER2/HER3 protein levels indirectly by interfering with FA synthesis. Upon uptake into the cell, palmitate is converted into palmitoyl-CoA which is a major allosteric feedback inhibitor of ACACA, the rate-limiting enzyme in fatty acid synthesis [
43]. The inhibition of fatty acid synthesis caused by palmitate (Baumann et al. unpublished) may have effects on growth factor signal transduction. Several lines of investigation have linked HER2 and fatty acid synthase (FASN) previously. ACACA and FASN are upregulated in HER2/neu-positive breast cancers at the transcriptional [
6] and the translational level [
10]. Overexpression of FASN in immortalized, non-tumorigenic mammary epithelial cells has been shown to induce HER2 overexpression and activation [
34] and there are reports that FASN may be directly phosphorylated by HER2 [
35]. Additionally, pharmacological inhibition of FASN induces ER stress [
44], decreases HER2 protein levels [
45,
46], sensitizes cells to trastuzumab treatment [
46] and reverses acquired autoresistance to trastuzumab [
47,
48]. All of these findings are consistent with our observations that exogenous palmitate induces ER stress, decreases HER2 and HER3 protein levels and sensitizes the cells to trastuzumab-mediated growth inhibition. These data suggest that exogenous palmitate may function, at least in part, as a feedback inhibitor of fatty acid synthesis in HER2/neu-positive breast cancer cells, as well as an ER stress inducer. Since prolonged activation of the ER stress response has previously been linked to increased sensitivity to various chemotherapeutic agents, including trastuzumab [
49] as we see (Fig.
7), levels of palmitate in the breast cancer microenvironment may have positive impacts on treatment. Along these lines, high levels of dietary saturated fatty acids may be capable of interfering with HER2 expression and signaling during disease development. While many studies have shown a protective effect of polyunsaturated fatty acids on breast cancer development in general, elevated levels of saturated fat may impose a metabolic constraint that specifically decreases the development of HER2/neu-positive breast cancer - a situation consistent with recent epidemiological studies [
11]. Although consumption of high levels of saturated fat is not an envisioned therapy, further investigation of this phenomenon may lead to an improved understanding of breast cancer cell physiology. In any event, this study provides further evidence that HER2 signaling, fatty acid metabolism and ER stress signaling are highly integrated processes that may be important for disease development and progression.
Abbreviations
4-PBA, 4-phenylbutyric acid; ACACA, acetyl-CoA carboxylase alpha; ACLY, ATP-citrate lyase; C16, palmitate; Ctrl, control; DPBS, Dulbecco’s phosphate buffered saline; ES, enrichment score; FA, fatty acid; FASN, fatty acid synthase; FBS, fetal bovine serum; GSEA, gene set enrichment analysis; HMECs, human mammary epithelial cells; HRP, horseradisch peroxidase; luc, luciferase; PBP, PPARγ binding protein; PERK, PKR-like endoplasmic reticulum kinase; RIN, RNA integrity number; SDS-PAGE, sodium dodecylsulfate polyacrylamide gel electrophoresis; TAG, triacylglycerides; TBS, Tris buffered saline; UPR, unfolded protein response
Acknowledgements
We thank M. Miura for the pCAX-XBP1-ΔDBD-venus reporter construct as well as Genentech for providing us with trastuzumab. We also thank A.-C. Gaupel and C. Sevinsky for critically reading the manuscript as well as helpful suggestions.