Endoplasmic reticulum stress protein GRP78 modulates lipid metabolism to control drug sensitivity and anti-tumor immunity in breast cancer

Abstract

The unfolded protein response is an endoplasmic reticulum stress pathway mediated by the protein chaperone glucose regulated-protein 78 (GRP78). Metabolic analysis of breast cancer cells shows that GRP78 silencing increases the intracellular concentrations of essential polyunsaturated fats including linoleic acid. Accumulation of fatty acids is due to an inhibition of mitochondrial fatty acid transport, resulting in a reduction of fatty acid oxidation. These data suggest a novel role of GRP78 mediating cellular metabolism. We validated the effect of GRP78-regulated metabolite changes by treating tumor-bearing mice with tamoxifen and/or linoleic acid. Tumors treated with linoleic acid plus tamoxifen exhibited reduced tumor area and tumor weight. Inhibition of either GRP78 or linoleic acid treatment increased MCP-1 serum levels, decreased CD47 expression, and increased macrophage infiltration, suggesting a novel role for GRP78 in regulating innate immunity. GRP78 control of fatty acid oxidation may represent a new homeostatic function for GRP78.

Introduction

Glucose-regulated protein 78 (GRP78) is a protein chaperone that acts as a master regulator of the unfolded protein response (UPR) (1,2). In the absence of stress, GRP78 is primarily bound to the three protein effectors of each UPR arm, inositol requiring enzyme 1 (ERN1; IRE1), PKR-like endoplasmic reticulum kinase (EIF2AK3; PERK), and activating transcription factor 6 (ATF6). These heterodimers remain inactive in the endoplasmic reticulum membrane until released from GRP78. Release occurs following the accumulation of unfolded/misfolded proteins within the endoplasmic reticulum, allowing induction of the UPR. Stimulation of IRE1 results in the unconventional splicing of X-box binding protein 1 (XBP1), leading to production of the active transcription factor XBP1-S and its related signaling (3). Activated PERK can phosphorylate eIF2α, inhibiting cap-dependent protein translation and promoting the translation of activating transcription factor 4 (ATF4) and DNA damage-inducible transcript 3 (DDIT3; CHOP). The release of GRP78 from ATF6 enables ATF6 to translocate to the Golgi complex where it is cleaved by site 1 and 2 proteases (S1P and S2P) to form the activated ATF6 transcription factor. Activation of the UPR controls various cell signaling pathways including cap-dependent protein translation, cell cycle, apoptosis, autophagy, transcription of protein chaperones, antioxidant response, among other responses. While activation of the UPR is initially prosurvival, prolonged UPR activation can lead to cell death (1,2).

Breast cancers exhibit increased activation of several UPR signaling components (4-6). Furthermore, some breast cancer therapies, such as tamoxifen (TAM) and faslodex (fulvestrant, ICI) used in the management of estrogen receptor positive (ER+) breast cancers, stimulate UPR signaling to promote cell survival and drug resistance (7). Antiestrogen resistant breast cancer cell lines express elevated levels of both GRP78 and XBP1, suggesting UPR activation as a driver of endocrine therapy resistance (8,9). Treatment of ER+ breast cancer cells with antiestrogens can cause the accumulation of inactive ERα within the cell (10,11). Ablation of ERα through RNAi inhibited antiestrogen therapy-mediated UPR activation (7). Thus, accumulation of ERα can stimulate UPR signaling. Inhibiting GRP78 using RNAi can potentiate antiestrogen responses in sensitive cells and at least partly restore sensitivity in resistant cells. We also showed that inhibition of GRP78 prevented antiestrogen-mediated autophagy induction through regulation of AMPK (8,12), suggesting that targeting GRP78 may affect other AMPK regulated functions such as cellular metabolism (13).

Using a GRP78-targeting morpholino, for the first time we show that in vivo inhibition of GRP78 potentiates tamoxifen sensitivity in ER+ breast tumors and can restore sensitivity in resistant tumors. Diverging from GRP78’s canonical role in UPR signaling, metabolomics analysis shows a novel role of GRP78 in regulating lipid metabolism. For example, we now show that supplementation of the GRP78-regulated metabolite linoleic acid (LA), a polyunsaturated omega-6 fatty acid, restores endocrine therapy sensitivity in vivo. We further show that GRP78 inhibition prevents mitochondrial lipid transportation through a reduction of CPT1A that limits fatty acid oxidation and increases lipid accumulation, peroxidation, and ROS generation. Moreover, in vivo supplementation with LA in combination with tamoxifen produced a greater inhibition of tumor growth than does treatment with TAM alone. These data suggest that LA regulation by GRP78 mediates, at least partly, the anti-tumor activity of the GRP78 morpholino. We also show, for the first time, that GRP78 inhibition in BALB/c mice and in athymic tumor bearing mice treated with human-targeting GRP78 morpholino or the GRP78 regulated metabolite (LA) supplementation, regulates CD47 expression and stimulates an innate immune response, which includes increase macrophage infiltration, to reduce ER+ tumor growth.

Material and Methods

Materials

The following materials were obtained as indicated: Mouse and human specific targeting GRP78 morpholinos (GeneTools; Philomath, OR); Tamoxifen citrate diet (LabDiet) and 4-OH Tamoxifen (Tocris Bioscience, Ellisville, MO). Improved Minimal Essential Medium (IMEM; Gibco Invitrogen BRL, Carlsbad, CA); bovine calf charcoal stripped serum (CCS) (Equitech-Bio Inc, Kerrville, TX); oil-red-O stain and N-acetyl-cysteine (NAC; Sigma-Aldrich, St. Louis, MO); and crystal violet (Fisher Scientific, Fairlawn, NJ). GRP78 siRNA was obtained from Dharmacon (Lafayette, CO). GRP78 pcDNA was obtained from Origene (Rockville, MD). ACC inhibitor, TOFA was obtained from Santa Cruz Biotechnology. Antibodies were obtained from the following sources: p110 alpha, Akt, p-Akt, IRE1, CHOP, PERK, and XBP1-S GRP78, CPT1A, calreticulin, HMBG1, phospho-ACC, ACC, SCD1, FASN, and MCP-1 (Cell Signaling Technology; Beverly, MA); adipophillin (Abbiotec; San Diego, CA); β-tubulin (Sigma-Aldrich), GRP78 (for IHC), β-actin, and polyclonal and HRP-conjugated secondary antibodies (Santa Cruz Biotechnology, Santa Cruz, CA). CD47 antibody was obtained from eBioscience. Linoleic acid (Tocris) was used for the in vitro studies. Linoleic acid (Sigma Aldrich) and time release linoleic acid and estrogen pellets were obtained from Innovative Research of America (Sarasota, FL) for the in vivo studies. The fatty acid oxidation kit was from Abcam (Cambridge, MA), the lipid peroxidation kit was obtained from Invitrogen (Carlsbad, CA), and the ROS determination flow cytometry kit was from Enzo (Farmingdale, NY). The kit for the immunohistochemical determination of ROS was obtained from (Millipore; Billerica, MA).

Cell culture

LCC1 and LCC9 human breast carcinoma cells, previously derived in this laboratory (14,15), were grown in phenol-red free IMEM media containing 5% charcoal-treated calf serum (CCS) and defined as basal growth conditions. ZR-75-1 obtained from ATCC, were grown in RPMI containing 10% FBS. Cells were grown at 37°C in a humidified, 5% CO₂:95% air atmosphere.

Cell Proliferation

Human breast cancer cells (5 × 10⁴ cells/mL) in IMEM containing 5% CCS were plated in 24-well tissue culture plates. For some experiments cells were transfected with control (scrambled non-targeting) or GRP78 siRNA on day of plating. On day 1 after plating, cells were treated with varying doses of TAM (10 nM-1000 nM) and/or 100 nM-100 μM linoleic acid and/or 1 μM NAC (antioxidant). On day 3 or 6, media was aspirated and cells were stained with crystal violet, permeabilized in citrate buffer, and absorbance was read at 480 nm using a plate reader.

Metabolomics

Metabolite analysis was performed by Metabolon (Durham, NC); see Supplementary Experimental Procedures.

Inhibition of GRP78 in vivo xenograft mouse models

Five week old ovariectomized athymic nude mice (Harlan Laboratories, Fredrick, MD) were injected orthotopically into the mammary fat pads with a suspension of 1 × 10⁶ LCC1 or LCC9 cells in Matrigel. Mice were supplemented with s.c. implantation of a 17β-estradiol pellet (0.36 mg, 60-day release; Innovative Research of America). Once tumors obtained an area of 30-40 mm², mice were treated every three days with an i.p. injection of 250 μL of 30 μM human specific GRP78 targeting morpholino (antisense code: GAGAGCTTCATCTTGCCAGCCAGTT) or mouse-specific GRP78 targeting morpholino (antisense code: GCTCAGCAGTCAGGCAGGAGTCTTA) or a combination of both human- and mouse-targeting GRP78 morpholinos. Where appropriate, some mice were also placed on a 5053 PicoLab Rodent Diet 20 containing 400 ppm tamoxifen citrate. Tumors were measured weekly for 4-6 weeks. Mice were sacrificed and tumors were removed at necropsy, fixed in neutral buffered formalin, and processed using routine histological methods.

Systemic GRP78 inhibition by morpholino

Female, 4 week old, BALB/c mice were purchased from Harlan. Every three days, mice were injected i.p. with 250 μL of 30 μM mouse-specific GRP78 targeting morpholino for three weeks before being euthanized. At necropsy, serum was collected for cytokine analysis and mammary glands were harvested for protein and immunohistochemical analysis.

In vivo metabolite replacement model

Five week old ovariectomized athymic nude mice were injected orthotopically into the mammary fat pads with a suspension of 1 × 10⁶ LCC9 cells in Matrigel. Mice were supplemented with s.c. implantation of a 17β-estradiol pellet. Once tumors obtained an area of 30-40 mm², mice were treated with 0.25 mg/day linoleic acid, 2.5 mg/day linoleic acid and/or 400 ppm tamoxifen citrate diet. Tumors were measured weekly for 6 weeks. At the end of the study, mice were euthanized and serum, mammary glands, and tumors were obtained for analysis.

RT-PCR

RNA was extracted using Trizol by following the manufacturer’s protocol. cDNA was synthesized from 1 – 5 μg of total RNA using Superscript first strand RT-PCR reagents as described by the manufacturer. qRT-PCR was then performed using the SYBR green kit. The gene/primer sequence data is shown in the Supplementary Experimental Procedures.

Western blot hybridization

As previously described, cells, tumors, and mammary glands were harvested in RIPA lysis buffer, protein was measured using a standard BSA assay, and proteins were size fractionated by polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes. Membranes were incubated overnight with primary antibodies. Immunoreactive products were visualized by chemiluminescence and quantified by densitometry using the ImageJ digital densitometry software (http://imagej.nih.gov/ij/). Protein loading was visualized by incubation of stripped membranes with a monoclonal antibody to β-actin (1:1000).

Flow Cytometry

LCC9 cells were transfected with control (sequence-specific scrambled oligonucleotide) or GRP78 siRNA and treated with TAM (100 nM) for 3 days. To measure fatty acid oxidation, cells were stained as described in the Fatty Acid Oxidation Detection Kit (Abcam), and counted by flow cytometry (GUMC Flow Cytometry Shared Resource). To measure ROS generation or lipid peroxidation, cells were stained as described in the Total ROS Detection kit (Enzo Life Sciences, Ann Arbor, MI) or the lipid peroxidation kit (Invitrogen) and counted by flow cytometry (GUMC Flow Cytometry Shared Resource).

Oil-red-O staining

LCC9 cells were transfected with control siRNA, GRP78 siRNA, or treated with 10 μM linoleic acid for 72 hours. Cells were fixed using 4% PFA, then stained with oil-red-O to visualize lipid droplets.

Immunohistochemistry (IHC)

Tumors were fixed in 10% formalin for at least 24 h prior to embedding in paraffin. Embedded tumors were cut into 5 μm thick sections and immunostaining was performed with an antibody to CD68 (1:100), GRP78 (1:100), adipophillin (1:100), CD47 (1:100) or a non-specific antibody (negative control) using the DAB method. Stained sections were visualized and photographed. ROS IHC was performed using OxyIHC^TM Oxidative Stress Detection kit.

Cytokine analysis

Serum from mice was collected and snap frozen at the time of necropsy. Quansys Biosciences (Logan, UT) Q-Plex Array^TM kits were used to measure MCP-1 mouse cytokine levels as described previously (16).

Statistics

Data are presented as the mean ± standard error of the mean (SEM). Statistical differences were evaluated by Student’s t test or one way analysis of variance (ANOVA) followed by Bonferoni post hoc tests. Criterion for statistical significance was set at p < 0.05.

Results

In vitro inhibition of GRP78

We first confirmed that inhibition of GRP78 by RNAi restores endocrine therapy sensitivity in LCC9 breast cancer cells (estrogen independent and antiestrogen cross-resistant) and also potentiates antiestrogen therapy responsiveness in antiestrogen sensitive ER+ ZR-75-1 breast cancer cells in vitro. Breast cancer cells were transfected with control (untargeted) siRNA, GRP78 siRNA, or GRP78 cDNA for 24 hours then plated (1×10⁴) in an Acea E-plate to measure cell index by electrical impedence for 72 hours in the presence of 100 nM TAM (Figure 1A and Supplemental Figure 1A+1B). Inhibition of GRP78 potentiated endocrine therapy responsiveness, while overexpression of GRP78 conferred resistance.

Figure 1.

Effects of targeting GRP78. A. LCC9 or ZR-75-1 ER+ breast cancer cells were transfected with control siRNA, GRP78 siRNA, or GRP78 cDNA for 24 hours, plated in an ACEA E-plate and treated with vehicle or 4-OHT for 72 hours and cell index was then measured by electrical impedance. n=3; *p<0.002 B. LCC9 orthotopic tumors were untreated (control) or treated with TAM, human and/or mouse GRP78-targeting morpholino (GRP78M), or human and/or mouse GRP78M+TAM for 4 weeks. Tumors were measured weekly with calipers and tumor area calculated. C. Average wet weight of LCC9 upon sacrifice. n=8-10, *p<0.03. D. Protein lysates from LCC9 treated tumors were isolated and Western blot hybridization was used to confirm levels of GRP78 and other UPR signaling components. E. Protein lysates from untreated (control) or mammary glands treated with human or mouse targeting GRP78M were isolated and Western blot hybridization was used to confirm expression levels of GRP78 and other UPR signaling components. F. LCC1 orthotopic tumors were grown to 25-30 mm² before treatment with TAM, GRP78M, or GRP78M+TAM for 6 weeks. Tumors were measured weekly with calipers and tumor area calculated. G. LCC1 tumor weight upon completion of study. n=6-10, *p<0.001.

In vivo inhibition of GRP78

We then determined whether inhibiting GRP78 in vivo would be a successful therapeutic strategy for the treatment of ER+ breast cancer. In the LCC9 (TAM-resistant) xenografts, a combination of human targeting GRP78 morpholino+TAM or human+mouse targeting GRP78 morpholino+TAM significantly reduced tumor area (Figure 1B) and tumor wet weight (Figure 1C) when compared with either control, TAM-only, or GRP78 morpholino-only treated animals, or mouse targeting GRP78 morpholino+TAM treated animals. These data suggest that inhibition of GRP78 in the tumor epithelial cells, not in the microenvironment, results in the re-sensitization of tumors to endocrine targeted therapies. Successful targeting of GRP78 by the morpholino was evident in the LCC9 tumor bearing mice treated with human targeting GRP78 morpholino-only and TAM+human targeting GRP78 (Figure 1D) and resulted in the increase of other UPR signaling component protein levels (PERK, CHOP, IRE1, and XBP1-S). Furthermore, specificity of the mouse-targeting GRP78 morpholino was confirmed in mammary glands from mice treated with human targeting GRP78 morpholino or mouse targeting GRP78 morpholino (Figure 1E) confirming the specificity of GRP78 targeting morpholinos. Tumor sections from treated LCC9 xenografts were immunostained with a GRP78 antibody. Successful targeting of GRP78 by the morpholino was evident in the GRP78 morpholino-only and TAM+GRP78 morpholino treated animals (Supplemental Figure 1C)

Tumor area was decreased in the LCC1 (TAM-sensitive) xenografts treated either with TAM-only or human targeting GRP78 morpholino+TAM when compared with the untreated and GRP78 morpholino only treated control mice (Figure 1F). However, tumor area was significantly smaller in mice treated with TAM+human targetingGRP78 morpholino when compared with TAM-only treated mice. Mice treated with TAM-only and TAM+human targeting GRP78 morpholino also showed decreased tumor weight when compared with their respective control tumors (Figure 1G). Moreover, a combination of TAM+human targeting GRP78 morpholino significantly reduced tumor weight when compared with tumors from the TAM-only treated mice. Thus, inhibiting GRP78 in combination with antiestrogen treatment potentiated endocrine therapy sensitivity.

Cell surface GRP78 localization was previously shown to activate PI3K/Akt signaling (17,18). We determined the protein levels of p110α, phosphorylated Akt (Ser473 and Thr308), and total Akt (Supplemental Figure 1D) in LCC9 xenograft tumors. Targeting of GRP78 by human GRP78 morpholino had no overall effect on PI3K/Akt signaling.

Metabolomic profile of GRP78 inhibition

To identify the molecular mechanism of GRP78-mediated potentiation of the TAM responsiveness, given our previous work suggesting GRP78 regulation of AMPK (8), metabolomics analysis of over 330 validated metabolites was performed on LCC1 and LCC9 breast cancer cells treated with TAM and/or transfected with GRP78 siRNA. Inhibition of GRP78 in the LCC1 cells significantly up-regulated over 14 metabolites and down-regulated 39 metabolites. GRP78 knockdown in the antiestrogen resistant LCC9 cells significantly up-regulated over 30 metabolites and down-regulated 13 metabolites. Principal component analysis (PCA) revealed a distinct separation between LCC1 and LCC9 samples implying significant differences in basal metabolism between cell types (Figure 2A). The effects of GRP78 knockdown were more subtle when compared within, rather than between, cell lines. Desmosterol was elevated in TAM treated samples in agreement with earlier studies (19,20), thereby serving as an internal control of drug efficacy (Figure 2B).

Figure 2.

Metabolomic profile of GRP78 inhibition. A. PCA analysis shows a distinct separation between LCC1 and LCC9 samples. B. Desmosterol levels in LCC1 and LCC9 cells treated with vehicle, TAM, GRP78 siRNA or GRP78 siRNA+TAM that serves as an internal control for TAM efficacy. Relative levels of linoleate (C), linolenate (D), dihomo-linoleate (E), dihomo-linolenate (F), arachidate (G), and arachidonate (H) in LCC1 and LCC9 cells treated with vehicle, TAM, GRP78 siRNA or GRP78 siRNA+TAM. n=6 independent experiments, *p<0.05.

Metabolomic profiling when GRP78 was inhibited identified a change in lipid metabolism common to both LCC1 and LCC9 breast cancer cells. The heat map for lipid metabolite levels is shown in Supplemental Figure 2A. Six significant lipid metabolites were regulated by GRP78 silencing and GRP78 knockdown + TAM in both LCC1 and LCC9 cell lines. GRP78 silencing alone and in combination with TAM treatment was accompanied by the accumulation of cellular linoleate (18:2 n6), linolenate (18:3 n3 or n6), dihomo-linoleate (20:2 n6), dihomo-linolenate (20:3 n3 or n6), and arachidonate (20:4 n6) (Figure 2C-H).

Targeting GRP78 reduces fatty acid oxidation

The increase in cellular fatty acids could reflect perturbations in fatty acid uptake, lipid biosynthesis, or fatty acid β-oxidation. We investigated the impact of GRP78 silencing on expression of the lipid/cholesterol metabolism modulator genes sterol-regulatory element binding factor-1 and -2 (SREBP1, SREBP2). Inhibition of GRP78 significantly reduced SREBP1 (Figure 3A) and SREBP2 (Figure 3B) when compared with control transfected cells. TAM treatment resulted in a significant increase in SREBP1 in LCC1 cells and SREBP2 in LCC9 cells, suggesting a differential regulation of these genes by TAM in antiestrogen sensitive and resistant cells.

Figure 3.

Targeting GRP78 reduces fatty acid oxidation. Relative gene expression of SREBP1 (A) or SREBP2 (B) in LCC1 and LCC9 cells treated with vehicle, TAM, GRP78 siRNA or GRP78 siRNA+TAM normalized with HRPT housekeeping gene expression. n=4 independent experiments in triplicate, *p<0.05. C. Analysis of downstream SREBP1 regulated proteins SCD1, FASN, ACC (total), p-ACC (inactive) in LCC1, LCC9, and ZR-75-1 cells transfected with control or GRP78 siRNA. D. Relative protein levels of GRP78, FASN, SCD1, ACC, CPT1A in BALB/c mice livers either control (untreated) or treated with GRP78-targeting morpholino or in 129S wildtype or GRP78 heterozygous mice. Equivalence of protein loading on gels was confirmed by measuring )-actin expression. n=4; *p<0.02. LCC9 (E) or ZR-75-1. (F) cells were transfected with control siRNA, GRP78 siRNA, GRP78 cDNA, or GRP78 siRNA+ GRP78 cDNA. Relative protein levels of GRP78, ACC, CPT1A, and p-AMPK were measured by Western blot hybridization. G. LCC9 cells transfected with control or GRP78 siRNA were treated with TOFA (5-(tetradecyloxy)-2-furoic acid), an ACC inhibitor, for 72 hours. Protein levels of GRP78, ACC, CPT1A, and )-actin were measured by Western blot hybridization. H. LCC9 cells were transfected with control siRNA, GRP78 siRNA, PERK siRNA, or XBP1 siRNA for 72 hours. Relative protein levels of GRP78, PERK, XBP1-S, ACC, CPT1A, and )-actin were measured by Western blot hybridization. I. Levels of fatty acid oxidation enzymes ACADVL, ACADM, and HADHA were measured by flow cytometry in LCC1 and LCC9 cells transfected with control or GRP78 siRNA. n=3, *p<0.04. J. Butyrylcarnitine and propionylcarnitine metabolite levels in LCC1 and LCC9 cells transfected with control or GRP78 siRNA +/−TAM. n=6; *p<0.05.

Several genes are controlled by SREBP1 including stearoyl-CoA desaturase (SCD; SCD1), fatty acid synthase (FASN), and acetyl-CoA carboxylase (ACACA; ACC) (21). Western blot hybridization of protein lysates from control siRNA or GRP78 siRNA transfected LCC1, LCC9, and ZR-75-1 cells shows that GRP78 silencing inhibits SCD1 and FASN protein expression, while modestly increasing ACC protein levels (Figure 3C). ACC is phosphorylated by AMP-activated protein kinase (PRKAA1; AMPK) at Ser79, inhibiting ACC activity. GRP78 knockdown reduces pACC Ser79 protein levels (Figure 3C). ACC can inhibit CPT1A through malonyl-CoA synthesis. Inhibition of GRP78 results in decreased CPT1A expression (Figure 3C), suggesting that GRP78 silencing may inhibit fatty-acid transport into the mitochondria.

We measured the effect of GRP78 inhibition in liver tissue of female wild type, B6.129(Cg)-Hspa5tm1.1Alee/J (GRP78 heterozygous), BALB/c mice treated with GRP78-targeting morpholino, and in BALB/c untreated controls. Inhibition of GRP78 in normal, non-cancerous tissue had no effect on SCD1 or FASN protein levels. However, we observed an increase in ACC protein and reduced CPT1A protein expression (Figure 3D). These data suggest a differential effect of GRP78 inhibition in tumors versus non-cancerous tissue, where de novo lipogenesis proteins are only inhibited in breast cancer cells and CPT1A/fatty acid mitochondrial transport may be reduced systemically by GRP78 targeting.

To determine whether GRP78 specifically regulated the ACC/CPT1A signaling axis, LCC9 (Figure 3E) and ZR-75-1 cells (Figure 3F) were transfected with control or GRP78 siRNA +/− GRP78 cDNA to ‘rescue’ the GRP78 protein levels. Overexpression of GRP78 in GRP78-silenced breast cancer cells prevented the ability of GRP78-silencing to induce ACC and inhibit CPT1A. To confirm the specificity of GRP78 for mediating the observed differences in lipid metabolism, LCC9 cells were transfected with GRP78 siRNA, PERK siRNA, or XBP1 siRNA to inhibit each of the three UPR signaling arms. Expression of ACC and CPT1A was then measured by Western hybridization (Figure 3H). Knockdown of GRP78, but neither PERK nor XBP1, increased ACC protein levels and reduced CPT1A. LCC9 cells were transfected with control or GRP78 siRNA and treated with various doses of 5-(tetradecyloxy)-2-furoic acid (TOFA), an ACC inhibitor (Figure 3G). Inhibition of ACC activity by TOFA prevented the GRP78-mediated reduction of CPT1A expression.

Fatty acid β-oxidation enzymes acyl-Coenzyme A dehydrogenase very-long chain (ACADVL), acyl-CoA dehydrogenase, C-4 to C-12 straight chain (ACADM), and hydroxyacyl-CoA dehydrogenase/3-ketoacyl-CoA thiolase/enoyl-CoA hydratase trifunctional protein alpha subunit (HADHA) were measured by flow cytometry. LCC9 cells expressed higher levels of ACADVL, ACADM, and HADHA when compared with their antiestrogen sensitive parental cells (LCC1). Knockdown of GRP78 inhibited ACADVL, ACADM, and HADHA in both LCC1 and LCC9 cells (Figure 3I). Reduced levels of the carnitine conjugates butyrylcarnitine and propionylcarnitine, generated by the oxidation of even and odd chain fatty acids respectively, were observed predominantly in GRP78 silenced/TAM-treated LCC1 and LCC9 cells, suggesting that loss of GRP78 may sensitize resistant cells by disrupting lipid metabolism (Figure 3J).

Inhibition of GRP78 increases cellular lipid content

To confirm GRP78-mediated regulation of cellular lipid content, LCC9 cells were transfected with control siRNA or GRP78 siRNA or treated with 10 μM linoleic acid and stained with oil-red-O (Figure 4A). Inhibition of GRP78 or treatment of cells with linoleic acid resulted in an overall increase in cellular lipid content. LCC9 xenograft sections were immunostained with Adipose differentiation related-protein (adipophilin), a marker of intracellular lipid droplets in metabolically active cells (Figure 4B). Adipophilin expression was increased in human-targeting GRP78 morpholino-only and TAM+human targeting GRP78 morpholino treated mice. The elevated levels of lipid droplets and cellular lipid content are consistent with GRP78 knockdown increasing cellular fatty acid concentrations. When LCC9 (Figure 4C) or ZR-75-1 (Figure 4D) were transfected with control or GRP78 siRNA in vitro,knockdown of GRP78 in both ER+ breast cancer cell lines increased adipophilin protein levels.

Figure 4.

Inhibition of GRP78 increases cellular lipid content. A. Oil-red-O staining of LCC9 cells transfected with control or GRP78 siRNA or treated with 10 μM LA. Cells were permeabilized and stained; absorbance was read at 490 nm to measure cellular lipid content. n=3, *p<0.02. B. LCC9 tumor sections were stained using an adipophilin antibody and visualized at 40x. LCC9 (C.) or ZR-75-1 (D.) human breast cancer cells were transfected with control or GRP78 siRNA for 96 hours and adipophilin was measured by Western blot hybridization.

Combination of GRP78 silencing and endocrine therapy increases ROS and cell death

Knockdown of GRP78+TAM in LCC9 cells significantly increased lipid oxidation (Figure 5A). Interestingly, only a combination of GRP78 knockdown and endocrine-targeted therapy increased ROS in both LCC9 cells (Figure 5B) and xenografts (Figure 5C). LCC9 xenografts were stained with TUNEL to measure apoptosis. As we previously observed in vitro in LCC9 cultured cells (8), only simultaneous knockdown of GRP78 with TAM treatment resulted in a stimulation of cell death in vivo in LCC9 xenograft tumors (Figure 5D). Thus, the increased cellular lipid content resulting from GRP78 inhibition in the presence of endocrine-targeted therapy induces lipid oxidation and ROS production. Increased ROS generated by GRP78-mediated lipid oxidation promotes TAM-mediated cell death. We have previously shown increased cellular concentrations of ROS promotes cell death in these cell lines (7). We then treated LCC9 and ZR-75-1 cells that were transfected with control or GRP78 siRNA with 1μM N-Acetylcysteine (NAC) and/or TAM for 72 hours and measured relative cell density by crystal violet assay (Figure 5E). Inhibition of ROS by NAC partially rescued GRP78 knockdown re-sensitization (LCC9) or potentiation (ZR-75-1) endocrine therapy responsiveness, suggesting a key role of ROS in mediating this phenotype. Furthermore, inhibition of ROS by NAC treatment prevented GRP78-mediated MCP-1 induction in GRP78-silenced breast cancer cells (Figure 5F), suggesting that targeting GRP78 may affect the tumor microenvironment through ROS regulation.

Figure 5.

Combination of GRP78 silencing and endocrine therapy increases ROS and cell death. A. Lipid peroxidation was measure by flow cytometry in LCC9 cells treated with cumene hydroperoxide (positive control), vehicle, TAM, GRP78 siRNA or GRP78 siRNA+TAM n=4, *p<0.001. B. ROS was measure by flow cytometry in LCC9 cells treated with 100 μM pyocyanin (positive control), vehicle, TAM, GRP78 siRNA or GRP78 siRNA+TAM n=3, *p<0.001. C. LCC9 tumor sections were stained using OxyIHC kit to measure ROS and visualized at 10x. D. LCC9 tumor sections were stained TUNEL to measure apoptosis and visualized at 20x. E. LCC9 and ZR-75-1 human breast cancer cells were transfected with control or GRP78 siRNA and treated with 1 μM N-acetylcysteine (NAC) and various doses of 4-OHT (TAM) for 72 hours; relative cell density was measured by crystal violet. n=3; *p<0.05 F. LCC9 and ZR-75-1 breast cancer cells were transfected with control or GRP78 siRNA and treated with 1 μM NAC for 48 hours and MCP-1 protein levels were measured by Western blot hybridization.

GRP78-regulated metabolite supplementation inhibits tumor growth

Antiestrogen resistant LCC9 breast cancer cells were treated with escalating doses of linoleic acid and TAM in vitro. Co-treatment with the GRP78-regulated metabolite (linoleic acid; LA) and TAM resulted in a modest resensitization of the resistant LCC9 breast cancer cells to TAM (Figure 6A). LA concentrations that successfully potentiated antiestrogen sensitivity significantly reduced GRP78 protein expression (Figure 6B), suggesting a reciprocal relationship between GRP78 and GRP78-regulated metabolites.

Figure 6.

Supplementation with linoleic acid (LA), a GRP78-regulated cellular metabolite, inhibits tumor growth. A. LCC9 cells were treated with 1-100 (M LA, and/or various concentrations of TAM (vehicle, 10 nM, 100 nM, 1000 nM) for 6 days. Relative cell density was determined by crystal violet assay. n=4, *p<0.02. B. LCC9 cells were treated with 1-100 (M LA for 72 hrs. Western blot hybridization was used to confirm levels of GRP78. n=4, *p<0.003. C. LCC9 orthotopic tumors were untreated (control) or treated with TAM, 0.25 mg/day LA, 2.5 mg/day LA, 0.25 mg/day LA +TAM, or 2.5 mg/day LA+TAM for 6 weeks. Tumors were measured weekly with calipers and tumor areas were calculated from the lengths on the two longest perpendicular measurements. D. Average wet weight of LCC9 tumors upon sacrifice. n=6-9, *p<0.001. E. LCC9 tumor sections were stained using GRP78 antibodies and visualized at 40x.

LCC9 xenografts were grown in the mammary fat pad regions of female athymic mice and mice treated with various doses of LA and/or TAM for 6 weeks. Combination treatment with LA+TAM reduced both tumor area (Figure 6C) and tumor weight (Figure 6D). Importantly, this combination was as effective as the combination of GRP78 targeting and TAM treatment, suggesting that the GRP78-regulated metabolite (LA) could mediate GRP78 morpholino anti-tumor activity. Furthermore, independent of TAM exposure, LA treatment inhibited GRP78 protein expression in LCC9 breast tumors (Figure 6E), supporting the feedback interaction observed in vitro. In vivo supplementation with LA (Figure 6C) resulted in a more effective inhibition of growth than was observed in vitro (Figure 6A). These data suggest an important role for the tumor microenvironment in mediating GRP78-regulated metabolite breast tumor growth inhibition.

GRP78 inhibition and GRP78-mediated metabolites affect innate immunity

Serum was extracted from LCC9 xenograft bearing mice treated with vehicle (control), TAM, LA, or LA+TAM and was used to determine circulating levels of MCP-1 by ELISA (Figure 7A). Supplementation with the GRP78-regulated metabolite LA significantly increased serum levels of monocyte chemotactic protein 1 (CCL2; MCP-1) when compared with vehicle (control) or TAM treated mice. Female BALB/c mice (no mammary tumors) were injected with control or a mouse targeting GRP78 morpholino for 3 weeks to systemically reduce overall GRP78 protein levels. Serum from BALB/c mice expressed higher levels of circulating MCP-1 when compared with serum isolated from control injected BALB/c mice (Figure 7A). Thus, systemic inhibition of GRP78 or supplementation with LA, a GRP78-regulated metabolite, produce similar effects on circulating MCP-1 levels.

Figure 7.

Inhibition of GRP78 and GRP78-regulated cellular metabolites affect innate immunity. A. Serum from LCC9 xenograft-bearing female aythmic mice untreated (control) and treated TAM, LA, LA+TAM and serum from untreated or GRP78 morpholino treated female BALB/c mice were analyzed by ELISA for circulating systemic levels of MCP-1. n=4, *p<0.001. B. Protein lysates from LCC9 tumors (control, TAM, LA, or LA+TAM) were analyzed using Western blot hybridization for calreticulin, HMGB1, and self-recognition identifier CD47. C. Protein lysates from LCC9 tumors (control, TAM, human targeting GRP78M, or human targeting GRP78M+TAM) were analyzed using Western blot hybridization for calreticulin, HMGB1, and self-recognition identifier CD47. E. Protein lysates from mammary glands extracted from untreated or GRP78M treated female BALB/c mice were analyzed for calreticulin and CD47. Gel loading was confirmed by measuring actin expression. Treated LCC9 tumor sections (D) or mammary glands from untreated or GRP78 morpholino (GRP78M) treated BALB/c mice (F) were stained using CD68 antibody to determine macrophage infiltration and visualized at 40x.

Western blot hybridization of protein lysates isolated from LCC9 xenograft control or treated tumors with TAM, LA, or LA+TAM (Figure 7B) or human targeting GRP78 morpholino +/−TAM (Figure 7C) were used to measure relative protein levels of “eat me/don’t eat me” protein signals including calreticulin (CALR), high mobility group box 1 (HMGB1), and CD47. Only GRP78M+/−TAM or LA+TAM treated tumors have increased multiple “eat me” signaling proteins including HMGB1 and calreticulin. LCC9 tumors treated with either GRP78M +/−TAM or LA+/−TAM have decreased CD47 expression, a potent “don’t eat me” signal. These data suggest that either supplementation with LA or targeting GRP78 stimulates a switch in immunosurveillance signaling.

Expression of the calreticulin and CD47 proteins was measured in protein lysates from mammary glands from BALB/c mice treated with control or mouse-targeting GRP78 morpholino by Western hybridization (Figure 7E). Reduction of GRP78 in the mammary glands of BALB/c mice increased calreticulin levels, similar to the effects of human targeting GRP78M treatment in xenografts. Unlike breast tumor tissue, inhibiting GRP78 in normal mammary tissue increased expression of the CD47 “don’t eat me” signal. These data suggest a differential role of GRP78 in regulating CD47 signaling in neoplastic versus normal tissues. LCC9 xenografts (Figure 7D) and BALB/c mammary glands (Figure 7F) were stained for CD68 to determine macrophage infiltration. Both knockdown of GRP78 by morpholino and supplementation with its regulated metabolite LA, increased macrophage infiltration in target tissues. Thus, either inhibition of GRP78 or treatment with LA can potentiate an anti-tumoral immune response.

Discussion

Breast cancer is the most frequently diagnosed cancer among women. Over 230,000 new cases of invasive breast cancer are diagnosed annually, with 70% of all breast cancers expressing the ERα (22). These cancers are often treated with ERα targeted therapies such as receptor antagonists (antiestrogens) including tamoxifen (TAM). However, many initially responsive tumors develop resistance to these endocrine therapies and, overall, more women die from ER+ breast cancer than from any other subtype of breast cancer (23). Our previous work reported elevated GRP78 protein levels in all molecular subtypes of breast cancer when compared with the normal surrounding breast tissue (8). Furthermore, endocrine therapy resistant breast cancer cell lines overexpress GRP78, suggesting GRP78 as a mediator of breast cancer resistance (8,24). While others have proposed GRP78 as a general target for therapy (25), our current study highlights the importance of targeting GRP78 as a specific therapeutic strategy for ER+ breast cancer (26). We show, for the first time, that antisense morpholino can successful target GRP78 protein expression in vivo to potentiate endocrine sensitivity in ER+ breast tumors. Combining TAM with the GRP78-targeting morpholino restored sensitivity in resistant ER+ tumors (Figure 1B-C) and increased TAM responsiveness in antiestrogen sensitive ER+ breast tumors (Figure 1F-G).

These studies expand the role of GRP78 from a protein chaperone controlling the unfolded protein response to include an important function in regulating lipid metabolism. Knockdown of GRP78 resulted in the accumulation of cellular essential fatty acids (Figure 2), suggesting that GRP78 regulates their uptake and/or catabolism. We show that inhibiting GRP78 reduces SREBP-1 transcript levels and decreases the expression of some SREBP-1 target genes (Figure 3A, 3C). SREBP1 is a basic helix-loop-helix-leucine zipper transcription factor that is maintained as an inactive precursor when located within the endoplasmic reticulum lumen (21). Activation of SREBP1 involves its translocation to the Golgi complex by SREBP cleavage-activating protein (SCAP) and proteolytic cleavage by site 1 (S1) and site 2 (S2) proteases. It is not surprising that GRP78 regulates SREBP, given the similarity in activation of SREBP to the activation of ATF6 UPR arm. Integration of UPR and SREBP regulation is intuitively rational. Accumulation of unusable unfolded proteins within the endoplasmic reticulum should trigger a mechanism to inhibit concurrently protein production to help relieve endoplasmic reticulum stress. Integration of UPR signaling and lipid metabolism is likely needed to address the problems with the unfolded protein components of lipoproteins being managed by the UPR in the presence of endoplasmic reticulum stress.

Previous studies have suggested a possible role of UPR signaling in lipid metabolism. GRP78 heterozygous mice are resistant to obesity when placed on a high fat diet, suggesting a role of GRP78 in modulating lipid metabolism (27). Other studies in a hepatic steatosis model showed increased UPR signaling components and endoplasmic reticulum stress in the liver of obese mice (28,29). Overexpression of GRP78 reduced UPR signaling and prevented insulin-mediated SREBP1c cleavage (30). The authors proposed that GRP78 binds to the SREBP complex preventing SCAP translocation to the Golgi complex and activation. While these data seemingly contradict some of our findings, we observed that GRP78 knockdown reduced cellular SCAP protein levels, thereby preventing SREBP translocation and activation by S1P/S2P (Supplemental Figure 3E). Therefore, both GRP78 overexpression and GRP78 depletion may inhibit SREBP through two distinct mechanism: SREBP1 binding and SCAP inhibition.

Other UPR signaling arms are implicated in promoting lipogenesis (28). In an ATF6 knock-out mouse model, endoplasmic reticulum stress led to liver steatosis resulting from impaired β-oxidation mediated by reduced C/EBP transcriptional activity (29). IRE1 knockout mice also developed liver steatosis and lipid accumulation (31). The mechanism mediating hepatic lipid accumulation in IRE1 deficient mice is unclear but may be mediated in part by loss of XBP1 activity (31,32). PERK inhibition can reduce SCD1 and FASN expression, also implicating this UPR signaling arm in lipogenesis (33). In support of these data, we show that inhibition of either XBP1 or PERK in LCC9 breast cancer cells has no overall effect on ACC and CPT1A signaling (Figure 3H). Since we previously showed that GRP78 silencing results in a significant increase in all three arms of UPR, our observed results may represent novel actions of GRP78 that occur independent of some UPR signaling components.

GRP78 inhibition led to a modest increase in ACC, suggesting differential regulation of SREBP1 controlled genes. We also observed a significant decrease in the inactivated phosphorylated-ACC Ser79; hence, GRP78 knockdown results in ACC activation. ACC is inactivated when phosphorylated on Ser79 by AMPK. We have previously shown that GRP78 overexpression increases autophagic signaling by stimulating AMPK (8,12). Knockdown of GRP78 prevents TSC2/AMPK signaling activation (8). Activation of ACC leads to increased malonyl-CoA synthesis resulting in the inhibition of CPT1A (34). CPT1A is localized in the outer mitochondrial membrane and catalyzes the primary regulated step in overall mitochondrial fatty acid oxidation (35). We observed decreased levels of both the mitochondrial fatty acid transporter protein, CPT1A, (Figure 3C-D) and β-oxidation enzyme activities and byproducts (Figure 3I-J). Furthermore, treatment of LCC9 cells transfected with GRP78 siRNA with TOFA (an ACC inhibitor), prevented GRP78-targeting reduction of CPT1A (Figure 3G). Taken together, these data suggest that the increased cellular concentration of lipids observed with GRP78 silencing is due to an overall inhibition of fatty acid oxidation.

Metabolomic analysis also identified elevated levels of oleic and palmitoyl ethanolamide in GRP78 inhibited cells (Supplemental Figure 2B). Co-treatment with oleic ethanolamide and TAM restores TAM sensitivity in antiestrogen resistant LCC9 breast cancer cells (Supplemental Figure 2C). Thus, the lipid accumulation that accompanies GRP78 inhibition increases the production of other anti-growth lipid metabolite bioproducts. Oleic and palmitoyl ethanolamide can also inhibit the expression of fatty acid hydrolase, reflecting their key roles in lipid metabolism (36). Furthermore, mice treated with oleic ethanolamide had reduced body weight gain when fed a high fat diet, perhaps explaining the obesity resistant phenotype observed in GRP78 heterozygous mice (37).

Inhibition of GRP78 led to increased cellular lipid content (Figure 2; Figure 4). Increases in fatty acid accumulation within non-adipose tissue can lead to cellular dysfunction and death, a phenomenon called lipid toxicity (38). Lipid toxicity may promote the cell death mediated by GRP78 targeting; others have reported a role for PUFAs in cancer cell death in vitro(39). For example, PUFA-induced cytotoxicity can be mediated by lipid peroxidation (40). Inhibition of GRP78 in LCC9 breast cancer cells increased lipid peroxidation (Figure 5A). Furthermore, morpholino-mediated GRP78 silencing in combination with TAM in vivo resulted in a significant increase in ROS generation and cell death (Figure 5C-D). Blockade of ROS generation by treatment with NAC partially rescued GRP78-silencing re-sensitization of LCC9 cells to endocrine therapy (Figure 5E), further supporting the critical role of UPR-mediated ROS generation. Treatment of LCC9 breast cancer cells with increasing concentrations of the GRP78-regulated metabolite LA showed a modest increase in TAM sensitivity in vitro (Figure 6A). However, in vivo supplementation of LA resulted in a clear resensitization of ER+ resistant breast tumors to TAM (Figure 6C). These data suggest that, while the effect of LA may partly reflect stimulation of lipid peroxidation in the tumor epithelial cells, the tumor microenvironment plays a vital role in mediating GRP78-regulated metabolite tumor cytotoxicity.

Since UPR can be prosurvival or prodeath, UPR driven signaling must be able to regulate the recognition and elimination of cells by the immune system. Systemic reduction of GRP78, or supplementation with LA, increased both circulating levels of MCP-1 (Figure 7A) and macrophage recruitment in the mammary tumors and glands (Figure 7D+F). Treatment of mice bearing LCC9 xenografts with human GRP78-targeting morpholino increased macrophage infiltration as indicated by CD68 immunoreactivity (Figure 7D). Previous reports showed that endoplasmic reticulum stress induced MCP-1 expression in the kidneys of db/db mice through induction of XBP1, linking UPR signaling to MCP-1 chemokine regulation (6). Moreover, LA treatment of endothelial cells also stimulated MCP-1 production through an oxidative mechanism (41). Treatment with NAC (a ROS inhibitor) prevented GRP78-silencing mediated MCP-1 induction (Figure 5F). Thus, the primary effect of lipid peroxidation and ROS generation mediated by either GRP78 inhibition or dosing with LA appears to be the stimulation of MCP-1 expression and macrophage recruitment.

CD47 is a widely expressed cell surface receptor that serves to regulate innate and adaptive immune system recognition (42). Clinical data indicates that CD47 is often upregulated in breast cancer and is associated with poor survival (43). We show, for the first time, that GRP78 is a modulator of “self recognition” through differential regulation of CD47 (Figure 7B, C, E). Supplementing mice bearing LCC9 xenografts with LA significantly reduced CD47 expression in tumor cells, an effect expected to increase their immune recognition and susceptibility to T cell and/or macrophage-mediated cytotoxicity. Treatment of mice bearing LCC9 xenografts with a human GRP78-targeting morpholino ±TAM decreased CD47 expression (Figure 7C and Supplemental Figure 4). Interestingly, TAM treatment significantly increased CD47 protein levels in both breast cancer cell lines and xenograft tumors; CD47 may be a critical component mediating endocrine resistance. Moreover, tamoxifen was shown to significantly upregulate CD47 expression in the human endometrium, suggesting a role of CD47 in promoting tamoxfien-induced endometrial cancer (44). In normal mammary tissue, systemic GRP78 protein reduction increased macrophage infiltration (Figure 7F). While GRP78-inhibition may negatively impact normal tissue, the GRP78-targeting morpholino significantly increased mammary gland expression of CD47 (Figure 7E), thereby protecting normal tissue from the cytolytic activities of macrophages.

Many normal tissues increase cellular fatty acid levels through uptake of circulating fatty acids (28). Unlike normal tissue, to meet their increased metabolic needs malignant cancer cells often up-regulate de novo fatty acid synthesis enzymes, perhaps to supplement reduced access to exogenous fatty acids from poor perfusion within the tumor microenvironment. This phenomenon is characteristic of the metabolic-switch often observed in cancer cells (45). We show, for the first time, that targeting GRP78 specifically inhibits de novo fatty acid synthesis proteins in breast cancer cells and reduces mitochondrial β-oxidation through CPT1A inhibition. Thus, GRP78 inhibition increased cellular lipid content, and promoted both lipid peroxidation and ROS generation. GRP78 knockdown increased lipid toxicity and breast cancer cell death. Increased ROS generation, by GRP78 knockdown, increases circulating chemokine MCP-1 and recruits macrophages into the tumor microenvironment. GRP78 differentially regulates CD47-dependent immune surveillance signaling in mammary tumor and normal cells to protect normal tissue and sensitize breast tumors to macrophage cytolytic activities. Taken together, these data establish a novel role for GRP78 in mediating lipid metabolism and explain why targeting GRP78 could be an effective therapeutic option for the treatment of breast cancer, and particularly endocrine resistant disease.