Introduction
Cholesterol is a fundamental structural components of lipid rafts, which are crucial for cholesterol to stimulate tumor cell growth and survival [1, 2]. Cellular cholesterol levels are regulated by a complex network, mainly including cholesterol biosynthesis, uptake and esterification [3]. It was reported that accelerating de novo synthesis of cholesterol can meet the needs of rapid proliferation in tumor cells [4]. Although intracellular cholesterol biosynthesis could be modulated by transcriptional factors such as sterol regulatory element-binding protein 1 and 2 (SREBP1 and 2) and nuclear receptor (NR) liver X receptors (LXRs) under physiological and pathological conditions [5], the concrete regulatory mechanisms of cholesterol biosynthesis in cancers are intricate, which still remain elusive and warrant in-depth elaboration. Therefore, it is of significant importance to identify the regulatory factors involved in cholesterol biosynthesis during the progression of malignancies.
Triple negative breast cancer (TNBC) is defined as lack of canonical therapeutic targets such as estrogen/progesterone receptor and human epidermal growth factor receptor 2 (HER2) [6]. It is characterized with high invasion capability, metastatic potential, recurrence rate and poor clinical outcome [7]. The 10-year recurrence free survival (RFS) of early TNBC is less than 70%. Furthermore, for advanced TNBC patients, the duration of response for first-line chemotherapy is less than 6 months and the average lifespan is only 1–2 years [8]. In recent years, the functions of cholesterol in breast cancer have attracted great concern and cholesterol has been considered as a risk factor for survival and recurrence of breast cancer patients [9]. Previous studies reported that cholesterol biosynthesis pathway is much hyper-activated in TNBC than in receptor status positive breast cancer [10]. It was reported that obesity is positively associated with higher risk in premenopausal TNBC patients and negatively related to estrogen receptor status [11, 12]. Usually, we know that obese patients often exhibit high cholesterol levels. Hence, abnormal cholesterol biosynthesis pathway might be the specific feature of TNBC and the molecular mechanisms of cholesterol biosynthesis in TNBC deserve in-depth study.
It was reported that accelerating de novo synthesis of cholesterol can meet the needs of rapid proliferation in tumor cells [4]. The de novo synthesis of intracellular cholesterol requires over 20 enzymatic reactions [4]. In the process of malignant tumor progression, the enzyme regulating cholesterol changes, which can increase the cholesterol pool in the cells [13]. Therefore, key molecules regulating cholesterol synthesis are crucial in the progression of tumors. Farnesyl diphosphate farnesyltransferase 1 (FDFT1), a squalene synthase, located on the endoplasmic reticulum membrane can catalyze the formation of squalene from two farnesyl diphosphate molecules and is an important enzyme molecule in the endogenous synthesis pathway of cholesterol [14]. Interestingly, FDFT1 could exert double-sided role in cancers, both oncogenic and suppressive role [2, 15]. FDFT1 can negatively regulated aerobic glycolysis to suppress colorectal cancer growth [16]. On the contrary, high FDFT1 expression endows chemoresistance in bladder cancer [17] and overexpression of FDFT1 in breast cancer cells could mitigate drug induced cell apoptosis [18]. All these effects of FDFT1 on cancers are all independent on its cholesterol synthesis function.
Members of the endoplasmic reticulum membrane protein complex (EMC) family can affect various cellular processes, including protein transport, endoplasmic reticulum stress, and lipid homeostasis [19]. EMC in yeast can recruit protein maturation factors in the ER lumen and participate in the insertion of transmembrane domains through the synergistic translocation of Sec61 [20] and can cooperated with molecular chaperones to assist the correct folding of newly generated peptides [21]. EMC2, a member of EMC, can bind to the molecular chaperone heat shock protein HSP90 to promote the correct insertion and folding of the protein into the endoplasmic reticulum [22]. At present, the researches on the function and molecular mechanism of EMC2 in tumors are still very limited, only bioinformatics analysis reported that high expression of EMC2 is closely related to the poor prognosis of bladder urothelial carcinoma, breast cancer and uveal melanoma [23]. In this study, we uncovered a unique biological mechanism which orchestrates intracellular cholesterol biosynthesis. We unraveled the role of EMC2 in cholesterol synthetic dysfunction and ferroptosis suppression with bioinformatic analysis of TCGA dataset, protein mass spectrum (MS) and further accurate experiments. Mechanistically, we clarified that EMC2 could prevent FDFT1 from endoplasmic reticulum related degradation (ERAD) by controlling its protein quality in the endoplasmic reticulum. All these findings may offer a novel theoretical basis for exploiting novel anti-tumor drugs targeting EMC2 for TNBC.
Methods
Cell lines
The TNBC cell lines, MDA-MB-231, BT-20, and MDA-MB-468 used in this study were obtained from the Shanghai Jiaotong University affiliated Renji Hospital and tested for mycoplasma contamination. MDA-MB-231 and MDA-MB-468 were cultured in Leibovitz Medium (L15, Gibco) containing 10% Fetal Bovine Serum (FBS, Gibco) and 1% penicillin and streptomycin (Gibco). BT-20 was cultured in Dulbecco’s Modified Eagle Medium (DMEM, Gibco) containing 10% Fetal Bovine Serum (FBS, Gibco) and 1% penicillin and streptomycin (Gibco). All the cells were maintained at 37 °C in a humidified atmosphere containing 5% CO2.
Reagents and antibodies
Antibodies and corresponding applications used in this study were shown in Supplementary Table 1. Matrigel was purchased from Corning. Polybrene and puromycin from Sigma-Aldrich. Cholesterol (C3045) from Sigma-Aldrich. MG132(S1748) from Beyotime. CHX (HY-12320), MCD (HY-101461), Eer I (HY-110078) from Med Chem Express.
Plasmids, virus and constructs
The small interfering RNA (siRNA), knockdown and control lentiviruses (designated as shEMC2 and shNC) were purchased from Genepharma (Shanghai, China). The target sequences were listed in Supplementary Table 2. The overexpression or control lentiviruses (designated as oeEMC2, oeFDFT1, oeNC, and HA-Ub) were purchased from Obio Technology Co. (Shanghai, China). The construction reports were affiliated with the supplementary information.
The siRNA and DNA transfection
The siRNAs and plasmids were transfected into the TNBC cells at a final concentration of 20 nM with the jetPRIME DNA & siRNA transfection reagent (Polyplus, #114-15) according to the manufacturer’s instructions. The cells were harvested for the following experiment 24–72 h after transfection.
The stable cell line construction
The stable cell lines were obtained by infection with the lentivirus and puromycin selection (10 μg/ml). The efficiency of stable cell line selection was examined by real-time PCR and western blotting.
Cell Counting Kit-8 (CCK8) proliferation assay
Cell suspensions (3–5 × 103/well) were seeded in the 96-well culture plates and incubated or treated with the reagent. 10% CCK8 solution (Shanghai Coryea Biotechnology Co, KS301) was added into each well and cells were incubated for another 1 h and then measured by a microplate reader (BioTek) at 450 nm.
Cell colony-formation proliferation assay
Cell suspensions (2–5 × 103/well) were seeded in the 6-well culture plates and incubated or treated with the reagent. After 2 weeks and the single colonies reached more than 50 cell, the cells were fixed with 4% paraformaldehyde (PFA) for 20 min at room temperature. Then cells were stained with 0.1% crystal violet for 1 h and washed. The colonies were counted under a microscope (Leica).
Transwell migration and invasion assay
In the transwell migration assay, 2 × 105 cells were seeded into the upper compartment of the transwell chamber (Corning) in serum-free DMEM. For cell invasion assay, 2 × 105 cells were seeded into the matrigel upper chamber in serum-free DMEM. DMEM with 10% FBS was added in the lower chamber. After 24 h of incubation, the cells were fixed with 4%PFA and stained with 0.1% crystal violet. The cells were counted under a microscope (Leica).
Clinical samples
Tumors and adjacent normal tissues of RENJI cohort 1,2, and 3 were obtained from patients with breast cancer who performed surgery at Shanghai Renji Hospital. RENJI cohort 1 contains 37 tumor tissues and 26 adjacent normal tissues for RNA extraction. RENJI cohort 2 contains 49 formalin-fixed paraffin-embedded human triple negative breast cancer specimens with survival information. RENJI cohort 3 contains 57 formalin-fixed paraffin-embedded human breast cancer specimens. None of the patients in these cohorts received chemotherapy before surgery.
Immunohistochemical (IHC) staining
The IHC staining of paraffin-embedded tissues was performed and scored as standard procedure. In brief, tumor samples were fixed in formalin, embedded in paraffin, and sectioned at 4 μm thickness. These sections were then subjected to IHC staining. Representative images were randomly taken at the mentioned magnification using a Leica DM2500 microscope. The IHC score was calculated according to the positive cell percentage calculated by the average calculation of two pathologists. The antibody application concentration of IHC was listed in the supplementary Table 1.
RNA isolation and quantitative real-time PCR
Total RNA was isolated using Simply P total RNA extraction kit (BioFlux, BSC52S1) following the manufacturer’s instructions. The purity and concentration of RNA were measured with a UV spectrophotometer (Thermo Scientific NanoDrop, 2000C). The RNA was reverse transcripted with the HiScript-TS 2 × PCR Mix (Vazyme, RA103). Real-time PCR analysis was performed with the ChamQ Universal SYBR qPCR Master Mix (Vazyme, Q711) and LightCycler 480 System (Roche). The reverse transcription and real-time PCR conditions were according to the manufacturer’s protocol. The sequences of primers were listed in the Supplementary Table 3.
Protein extraction and western blot analysis
Samples were lysed in fresh RIPA Lysis Buffer (Epizyme, PC102) with protease inhibitor cocktail 100X (Sigma) and phosphatase inhibitor cocktail 100X (Sigma) according to the manufacturer’s protocol. Then the protein concentrations were determined by the BCA protein assay (Beyotime, P0010). The Lysates were boiled for 10 min with the loading buffer (denaturing, reducing,5×, Epizyme, LT101). Western blot analysis used 10% polyacrylamide gel (Epizyme, PG212) and transferred it onto a PVDF transfer membrane (Millipore, IPVH00010). Then the blot membranes were incubated with corresponding antibodies and the bands were visualized by Enhanced chemiluminescence (ECL) (Millipore, WBKLS0500) and ChemiDoc Touching Imaging System (Bio-rad). The blot bands were quantified by Image Lab software (Version 6.0.1, Bio-rad). During quantification, β-actin and Calnexin were used for internal reference. The antibodies and applications were listed in the supplementary Table 1.
Immunofluorescence
Cells were fixed with 4% paraformaldehyde (PFA) and incubated with permeabilizing solution (Sigma, Triton X-100). Then cells were incubated with corresponding antibodies and fluorescent images were captured using a Leica DM2500 microscope. The antibodies and applications were listed in the supplementary Table 1. The confocal imaging analysis was carried out by “Image J” software (National Institutes of Health, NIH).
Co-immunoprecipitation
The co-immunoprecipitation was performed with a classic magnetic protein A/G IP/Co-IP Kit (Epizyme, YJ201) according to the protocol. In brief, cells were lysed and incubated with corresponding antibodies at 4 °C overnight. Then, the lysis was incubated with the magnetic protein A/G at 4 °C overnight. Finally, the lysates were boiled for 10 mins with the loading buffer and for the next experiments. The antibodies and applications were listed in the supplementary Table 1.
Xenografted animal model
The female five-week-aged BALB/c nude mice were obtained from Shanghai Charles River Animal Co., Ltd. Mice, were grouped into six animals randomly and housed in a specific pathogen-free (SPF) environment in the animal center of Renji Hospital. BT-20 cells (1 × 106/100 μl) and MDA-MB-231 cells (1 × 107/100 μl) were injected subcutaneously into the right flank of the mice to establish a breast cancer xenograft model. The tumor was calculated every 5 days after the injection as follows: Tumor volume (V) = tumor longer diameter (a) × tumor shorter diameter2 (b2) /2. When the experiment ended, or the tumor volume reached 1500 mm3, the mice were sacrificed, and the tumors were isolated for further experiments. All experimental procedures were approved by the Institutional Animal Care and Use Committee of Renji Hospital, School of Medicine, Shanghai Jiao Tong University.
Measurement of free cholesterol levels
The free cholesterol levels of TNBC cells were determined using the free cholesterol assay kit (Applygen Technologies, E1016) following the protocol. In brief, cell samples were washed with PBS twice and then lysed and collected in microtubes. The samples and standard cholesterol with different concentrations were mixed and incubated with working solutions provided by the manufacturer for 20 min at room temperature and then measured by a microplate reader (BioTek) at 550 nm.
Measurement of ferrous iron levels (Fe2+)
The level of Fe2+ was detected using the iron assay kit (Sigma-Aldrich, MAK025) according to the operating protocol of the manufacturer. In brief, cell samples were homogenized in Iron Assay buffer and centrifuged at 16,000 × g for 10 min at 4 °C. The test samples and iron standards incubated for 30 min at 25 °C protected from light. Then Iron Probe was added to each well and mixed and incubated for 60 min at 25 °C protected from light. Finally, the ferrous iron levels were measured by a microplate reader (BioTek) at 593 nm.
Measurement of lipid reactive oxygen species (ROS)
The level of lipid ROS was evaluated using the C11-BODIPY591/581 (Thermo Fisher, D-3861). In brief, different groups of cells were incubated with 20 μM C11-BODIPY591/581 for 30 min and protected from the light. Then cells were washed and resuspended in the PBS, detected with a flow cytometer (BD Biosciences, LSRFortessa X-20). The relative ROS level was calculated with the FITC fluorescence intensity.
Measurement of GSH/GSSG ratio
The level of GSH/GSSG ratio was measured by the GSH and GSSG assay kit (Beyotime, S0053) according to the manufacturer’s procedure. In brief, cell samples were lysed, then put into liquid nitrogen and then at 37 °C twice. The samples were centrifuged at 10,000 g for 10 minsat 4 °C. The supernatant samples and standard GSH were mixed with wording buffer and incubated for 5 min at 25 °C. Then NADPH solution was added and the total GSH absorbances were measured by a microplate reader (BioTek) at 412 nm. Special reagents were then used to clear GSH and GSSG absorbances were measured by a microplate reader (BioTek) at 412 nm. The GSH/GSSG ration was calculated according to the protocol.
Label-free relative quantitative proteomics
The label-free relative quantitative proteomics analyses for EMC2-associated proteins were performed at Instrumental Analysis Center at Shanghai Jiaotong University. The proteomics was operated with the Easy nLC1200/Q-Exactive plus mass spectrometer (Thermo Scientific). The raw data were searched against the UniProt-Human database.
Pull-down assay and mass spectrometry (LC-MS) analysis
The LC-MS analysis was performed at Shanghai Luming Biological Technology Co., Ltd. The proteomics was operated with the Ultimate 3000 RSLCnano/Q-Exactive plus mass spectrometer (Thermo Scientific). The raw data were searched against the UniProt-Human database.
Ubiquitination assay
Cells were transfected with combinations of plasmids and siRNA, including HA-ubiquitin plasmids. Before harvest, cells were treated with 10 μg/ml MG132 for 6 h. Then cells were used for the following research to detect the ubiquitination level of target proteins.
Endoplasmic reticulum enrichment and cell fractionation
The extraction of ER was performed by the ER enrichment kit (Novus Biologicals, NBP2-29482) according to the protocol. In brief, cell samples were homogenized with 1×Isosmotic Homogenization Buffer followed by 100× PIC and centrifuged at 1000 × g for 10 min at 4 °C. The supernatant was centrifuged at 12,000 × g for 15 min at 4 °C. Then the supernatant was centrifuged at 90,000 × g (Beckman Avanti J30I centrifuge with JS24 rotor) for 60 min at 4 °C. The pellet was collected as the total endoplasmic reticulum fraction for the following experiments.
Filipin III staining
Cells were fixed with 4% PFA and washed with PBS. Then cells were incubated with 0.05 mg/ml Filipin III (Sigma, SAE0087) for 1 h. Immunofluorescence images were captured using a Leica DM2500 microscope.
Measurement of aggresome
The aggresome was detected by PROTEOSTAT® Aggresome Detection Kit (Enzo, ENZ-51035) according to the protocol. In brief, cells samples were washed with PBS twice and fixed with 4% PFA for 30 min at room temperature. Then, samples were incubated with a permeabilizing solution (Sigma, Triton X-100) and washed with PBS twice. Then, the Dual Detection Reagent was added and incubated for 30 min at room temperature, protected from light. Finally, the immunofluorescence images were captured using a Leica DM2500 microscope. The calculation was measured by “Image J” software (National Institutes of Health, NIH).
Bioinformatic analysis and filtering strategy
The cancer genome atlas (TCGA) datasets were obtained by the R package “TCGAbiolinks”. Counting data is converted to transcripts per million (TPM) and normalized log2 (TPM + 1) while keeping clinical information intact. The Gene Expression Omnibus (GEO) datasets were obtained from the following website: https://www.ncbi.nlm.nih.gov/gds/?term=GSE. Only cancer tissue was included. The 27 cholesterol biosynthesis genes from the REACTOME pathway were obtained from the following website: https://reactome.org/content/detail/R-HSA-191273. Then we performed consensus analysis using the R package ConsensusClusterPlus (v1.54.0), with number of clusters set to two, repeating 100 times by extracting 80% of the total samples, and setting clusterAlg = “hc” and innerLinkage = ‘ward.D2’.The differential expressed genes (DEGs) analysis was performed with R package Limma (v3.40.2) using the two clusters mentioned before. In the data from TCGA or GEO, we analyzed the adjusted P-values to correct for false positive results. In cholesterol biosynthesis candidate genes filter, 1432 upregulated mRNAs were selected (Log FC > 0.5, P value < 0.001, average expression >5) from TCGA Triple negative breast cancer, 1216 upregulated mRNAs were selected (Log FC > 0.25, P value < 0.1, average expression >7) from GSE106977, 1144 upregulated mRNAs were selected (Log FC > 0.25, P value < 0.001, average expression >5) from GSE76250, and 457 upregulated mRNAs were selected (Log FC > 0.8, P value < 0.05, average expression >5) from GSE38959. The candidate cholesterol biosynthesis-related genes were first selected by the overlapping of the DEGs in the four datasets.
Statistical analysis
Statistical analyses were performed using R-3.6.3 and GraphPad Prism 9.5.1. The experiments were repeated at least three times with similar results. All data were presented as the means ± SDs. Two-sided unpaired Student’s t tests, one way ANOVA tests, chi-square tests, Kaplan–Meier analysis, and log-rank tests were used to evaluate the data according to the figure legends. Differences were listed in the figures as detailed numbers unless P < 0.001. All p values were two-sided unless otherwise specified.
Results
Identification of EMC2 as a positive regulator of cholesterol biosynthesis and negative predictor of overall survival for TNBC
To identify potential clinically related cholesterol biosynthesis genes in TNBC, we initially collected a total of 27 cholesterol biosynthesis genes from the REACTOME pathway (Table 1). Then, public datasets including TCGA and GEO (GSE106977, GSE76250 and GSE38959) were employed to dig the pivotal cholesterol biosynthesis related genes in TNBC. We classified aforementioned 27 genes of each dataset into two groups and analyzed the differential expression of genes (DEGs) between these groups (Fig. S1A). We identified the DEGs that overlapped across four high-throughput analysis datasets (Fig. 1A). Finally, a total of 11 candidate cholesterol biosynthetic genes were determined (Table 2). To further understand the functional roles of these candidate genes in TNBC, we identified the GO and KEGG terms of each candidate genes, which revealed that eight of the candidate genes were involved in processes such as response to unfolded protein, protein folding, endoplasmic reticulum protein-containing complex, protein folding chaperone, and unfolded protein binding (Fig. 1B). The protein-protein interaction (PPI) network was visualized, comprising of 10 genes and 19 nodes (Fig. 1C). Through the integration of functional enrichment and PPI results, a total of 7 candidate genes (DNAJB1, EMC2, HSP90AA1, HSPA5, POLR2K, RPN2, and TAF2) were identified. We then evaluated the correlation between these 7 genes and OS in TNBC by using single-factor Cox regression analysis. Notably, among the candidate genes, only EMC2 harbored an obviously prognostic value in TNBC (Fig. 1D and S1B).
A Venn diagram shows the potential regulators of cholesterol biosynthesis for TNBC, up-regulated in four datasets (TCGA, GSE106977, GSE76250, GSE38959). 11 genes were overlapped. B Heatmap showing the different functions to identify the candidate gene profiles associated with cholesterol biosynthesis. C PPI network carried by STRING to identify the functional protein association networks. The combined score was computed by combining the probabilities from the different evidence channels and corrected for the probability of randomly observing an interaction according to STRING. D Overall survival analyses stratified by median EMC2 mRNA expression in TNBC in the TCGA dataset and the KM-plotter website. The Kaplan–Meier method was used to estimate overall survival. The log-rank test was used to evaluate the differences between survival curves. E Histograms show the EMC2 mRNA expression profile between breast cancer tissues and normal adjacent tissues in the TCGA and GSE45827 datasets. Wilcoxon rank-sum test and student t test were used as statistical analyses. F Histograms show the EMC2 mRNA expression among different breast cancer subtypes in the TCGA and GSE45827 datasets. Dunn’s test and student t-test were used as statistical analyses. G Histogram shows the EMC2 mRNA expression between breast cancer tissues and normal adjacent tissues in Renji cohort 1. Student t test was used as statistical analysis. H Representative images of different EMC2 protein expression levels in breast cancer using IHC analysis in Renji cohort 2. The scale bar is 500 μm and 100 μm. Staked bars showing the contingency of EMC2 expression between breast cancer tissues and normal adjacent tissues in Renji cohort 2. The Chi-square test was used as statistical analysis. I Disease Free Survival (DFS) and Overall Survival (OS) analyses stratified by median EMC2 protein expression in TNBC in Renji cohort 2. The Kaplan–Meier method was used to estimate survival probability. The log-rank test was used to evaluate the differences between survival curves. J Forest plot showing the associations of different paths-clinical characteristics and EMC2 protein expression in Renji cohort 2. The univariate Cox regression was used as statistical analysis. TNBC triple-negative breast cancer, DEGs differentially expressed genes, TCGA The Cancer Genome Atlas, GEO Gene Expression Omnibus, GO Gene Ontology, KEGG Kyoto Encyclopedia of Genes and Genomes. See also Figs. S1–2.
On the basis of TCGA and GEO datasets, we found that EMC2 expressed higher in breast cancer than in normal tissues (Fig. 1E), and expressed higher in metastatic tissues (Fig. S2A). Among breast cancer subtypes, EMC2 expressed highest in the TNBC than in hormone receptor positive (HR + ) and human epidermal growth factor receptor 2 (HER2 + ) breast cancer (Fig. 1F). EMC2 also showed subtype-specific prognostic value in predicting the overall survival (OS, Fig. S2B) and recurrence-free survival (RFS, Fig. S2C). These results were verified by Renji breast cancer tissues using RT-PCR and immunohistochemical assays (Fig. 1G-I). Moreover, univariate Cox regression analysis indicated that EMC2 was positively related to clinicopathological characteristics in TNBC based on our cohort (Fig. 1J). Totally, EMC2 may potentially be a positive regulator of cholesterol biosynthesis and can predict poor clinical outcome of TNBC.
EMC2 promotes TNBC progression in vitro and in vivo
We initially determined the expression of EMC2 in various TNBC cell lines by analyzing the Cancer Cell Line Encyclopedia (CCLE) and conducting western blot assay (Fig. S3A and S3B). We selected two cell lines (MDA-MB-231 and BT-20 cell) with EMC2 upregulation and one cell line (MDA-MB-468 cell) with EMC2 downregulation as our experimental cells. Then, we transfected EMC2 siRNAs into MDA-MB-231 and BT-20 cell lines to establish knockdown models, while EMC2 overexpression plasmids were transfected into MDA-MB-468 cells to establish an overexpression model. The transfection efficiency was tested at both mRNA level (Fig. S3C) and protein level (Fig. 2A and S3D). Knockdown of EMC2 mitigated the proliferation of MDA-MB-231 and BT-20 cells by using CCK-8 and colony formation assays (Fig. 2B, C and Fig. S3E, F). Conversely, overexpression of EMC2 promoted the cell proliferation in MDA-MB-468 cells (Fig. 2B, C). To further demonstrate the effect of EMC2 on TNBC growth in vivo, we constructed a lentivirus-based stable RNA interference vector utilizing the EMC2-siRNA1 and EMC2-siRNA2 interference sequence in MDA-MB-231 and BT-20 cells. The knockdown efficiency of Lenti-shEMC2-transfected was determined by western blotting (Fig. S4A). Consistently, we validated that stable EMC2 downregulation could prominently slow down TNBC tumor growth in the subcutaneous tumor model (Fig. 2E–G, Fig. S4B–E). Besides, the Ki-67 staining revealed that cell proliferative index was reduced in shEMC2 tumor tissues compared to that in control one in BT-20 models (Fig. 2H and Fig. S4F). All our findings suggested that EMC2 may act oncogenic function in TNBC.
A The efficiency of EMC2 knockdown and overexpression models (MDA-MB-231 and MDA-MB-468) at the protein level by western blot, n = 3. B CCK8 assays were performed in the EMC2 knockdown and overexpression models (MDA-MB-231 and MDA-MB-468), n = 3. Values are means ± SD. Values are means ± SD. pvalues were calculated by one-way ANOVA test and two-tailed unpaired Student t test. C Representative colony formation assay and relative colony formation analyses of EMC2 knockdown and overexpression models (MDA-MB-231 and MDA-MB-468), n = 6. Values are means ± SD. pvalues were calculated by one-way ANOVA test and two-tailed unpaired Student t test. D Representative image of syngeneic tumors (BT-20) at the end of the experiment (shNC, shEMC2-1, and shEMC2-2 groups, n = 6 in each group). The scale bar is 1 cm. E Tumor weight and tumor volume (BT-20) in each group were measured at the end of the experiment was measured in each group. Values are means ± SD. pvalues were calculated by one-way ANOVA test. F The Tumor volume (BT-20) in each group was measured every five days and drawn into a curve. Values are means ± SD. pvalues were calculated using the data at the end of the experiment by one-way ANOVA test. G Representative immunohistochemistry images of the syngeneic mice tumor tissue stained for Ki-67 in BT-20 shNC mice, shEMC2-1 mice, and shEMC2-2 mice. The scale bar is 50 μm. See also Figs. S3, 4.
EMC2 positively regulates the intracellular cholesterol biosynthesis pathway
To explore the underlying molecular mechanisms of EMC2 in TNBC cells, we performed protein mass spectrum (MS) of both BT-20 EMC2 knockdown cells and control cells (Source data). Gene set enrichment analysis (GSEA) analysis uncovered that proteins altered by EMC2 downregulation were tightly associated with cholesterol biosynthesis (Fig. 3A). GO and KEGG of our MS data also exhibited that differential expressed proteins were mostly enriched in cholesterol biosynthesis (Fig. 3B). GSEA analysis on the basis of TCGA TNBC data displayed that EMC2 upregulation was mostly correlated with genes involved in cholesterol biosynthesis pathway (Fig. 3C). In BT-20 and MDA-MB-231 cells, cholesterol levels were significantly diminished by EMC2 knockdown (Figs. 3D, E and S5). On the contrary, cholesterol contents were enhanced by EMC2 overexpression in MDA-MB-468 cells (Fig. 3F). We then verified whether the EMC2 knockdown-induced phenotype was due to cholesterol depletion. Supplement of cholesterol partially restored the cell viability in EMC2 downregulation TNBC cells (Fig. 3G). Conversely, cholesterol deprivation by using low lipid charcoal-stripped serum (CFS) or the cholesterol depletion agent methyl-β-cyclodextrin (MCD) attenuates proliferation mediated by EMC2 overexpression (Fig. 3H), supporting the crucial role of cholesterol homeostasis in EMC2 function. Altogether, these results suggested that EMC2 may fulfill its tumorigenic function via regulating cholesterol synthetic metabolism in TNBC.
A Gene set enrichment analysis (GSEA) revealed pathway alterations of BT-20 EMC2 knockdown cells and control cells, 3 replications were examined in each group. The top 5 pathways (P < 0.05 and FDR q < 0.25) ranked by absolute normalized enrichment scores are shown. NES score and nominal Pvalue were given by GSEA software. B The KEGG and GO analysis was used to enrich the functional differences of BT-20 EMC2 knockdown cells and control cells, 3 replications were examined in each group. The pathways (P.adjust <0.05 and FDR q < 0.25) were ranked by normalized Z-score and log10(P.adjust). GO analyses include biological process (BP), cellular component (CC), and molecular function (MF). C Enrichment plot of the cholesterol biosynthesis pathway from GSEA analysis between EMC2 high and low expression in TNBC in the TCGA dataset. NES score and nominal Pvalue were given by GSEA software. D The cellular free cholesterol concentration after EMC2 knockdown in MDA-MB-231 and BT-20 cells, n = 3. Values are means ± SD. pvalues were calculated by a one-way ANOVA test. E Filipin III staining showing the cellular free cholesterol content in EMC2 knockdown models (MDA-MB-231 and BT-20). The boxed areas are shown in the lower panels. The scale bar is 50 μm and 20 μm, n = 3 fields/group. F The cellular free cholesterol concentration after EMC2 overexpression in MDA-MB-468 cells, n = 3. Values are means ± SD. pvalues were calculated by a two-tailed unpaired Student t test. G The CCK-8 assay after the knockdown of EMC2 while adding exogenous cholesterol (10 μM) in the medium in the MDA-MB-231 and BT-20 cells, n = 3. The pvalue was compared between 96 h of the siEMC2-1 and siEMC2-1 + CHOL groups. Values are presented as mean ± SD, tested by a one-way ANOVA test. H The CCK-8 assay after the overexpression of EMC2 was mixed with media containing CFS or complete serum or removing the cholesterol by MCD (5 mM for 6 h each day) in the MDA-MB-468 cells, n = 3. The p value was compared between 96 h of the oeEMC2 and oeEMC2+CFS/MCD groups. Values are presented as mean ± SD, tested by a one-way ANOVA test. GSEA Gene Set Enrichment Analysis, CHOL cholesterol, CFS charcoal-stripped serum, MCD methyl-β-cyclodextrin. See also Fig. S5.
EMC2-mediated TNBC growth is partially affected by FDFT1 expression
To dig the exact molecule by which EMC2 affected to orchestrate cholesterol biosynthesis, we analyzed protein expression of genes related to cholesterol biosynthesis using our MS data. Heatmap of differential proteins in siNC and siEMC2 BT-20 cells revealed that FDFT1 ranked as the most significant decreased one in EMC2 knockdown cells (Fig. 4A). FDFT1, named farnesyl-diphosphate farnesyltransferase 1, encodes the protein responsible for the initial step in cholesterol biosynthesis. This enzyme facilitates the dimerization of two molecules of farnesyl diphosphate through a two-step reaction, ultimately leading to the formation of squalene, which plays a crucial role in the process of cholesterol biosynthesis [24]. Actually, FDFT1 knockdown could indeed decrease intracellular cholesterol levels in TNBC cells (Fig. S6A–C). We subsequently demonstrated that EMC2 interference could downregulate FDFT1 protein level (Fig. 4B) and EMC2 overexpression could increase FDFT1 protein level (Fig. 4C), without affecting the mRNA level of FDFT1 in TNBC cells (Fig. 4D). Immunohistochemical staining of serial sections from the TNBC tissues disclosed that FDFT1 expression was positively related to EMC2 expression (Fig. 4E and S6D). Next, we validated the involvement of FDFT1 in EMC2 knockdown-induced proliferation suppression by overexpressing FDFT1 in EMC2-silenced TNBC cells. Our consequences exhibited that the overexpression of FDFT1 could partially ameliorated intracellular free cholesterol contents and TNBC cell growth inhibition caused by EMC2 downregulation (Fig. 4F–I, S6E).
A Heatmap of genes related to the cholesterol biosynthesis pathway between shNC and shEMC2 transfection BT-20 breast cancer cells via proteomics. The cholesterol biosynthesis genes were ranked by abundance fold change. 3 replications were examined in each group. B Western blot showing the expression level of FDFT1 after EMC2 knockdown in MDA-MB-231 and BT-20 cells, n = 3. C Western blot showing the expression level of FDFT1 after EMC2 overexpression in MDA-MB-468, n = 3. D RT-qPCR showing the mRNA expression level of FDFT1 in EMC2 knockdown and overexpression models (MDA-MB-231, BT-20, and MDA-MB-468), n = 3. Values are means ± SD. pvalues were calculated by one-way ANOVA test and two-tailed unpaired Student t test. E Representative images of FDFT1 IHC staining according to different EMC2 protein expression levels in breast cancer. The scale bar is 500 μm and 100 μm. F Filipin III staining showing the cellular free cholesterol content in EMC2 knockdown models (MDA-MB-231and BT-20) with or without FDFT1 overexpression. The scale bar is 50 μm, n = 3 fields/group. G The cellular free cholesterol concentration in EMC2 knockdown models (MDA-MB-231and BT-20) with or without FDFT1 overexpression, n = 3. Values are means ± SD. Values are means ± SD. pvalues were calculated by a one-way ANOVA test. H The CCK-8 assay in EMC2 knockdown models (MDA-MB-231and BT-20) with or without FDFT1 overexpression, n = 3. Values are presented as mean ± SD, tested by a one-way ANOVA test. I The clone formation assays in EMC2 knockdown models (MDA-MB-231and BT-20) with or without FDFT1 overexpression, n = 3. Values are presented as mean ± SD, tested by a one-way ANOVA test. See also Fig. S6.
Knockdown of EMC2 mediates FDFT1 degradation through ERAD pathway
We continue to explore how EMC2 modulates FDFT1 abundance. Since we have proved that EMC2 only affects FDFT1 at protein level, the effect of EMC2 knockdown on the stability of endogenous FDFT1 protein in the presence of the protein synthesis inhibitor cycloheximide (CHX) were examined. The results indicated that FDFT1 protein exhibited obviously more rapid degradation in siEMC2 cells than that in siNC TNBC cells (Fig. 5A). We also employed a protein proteasome inhibitor (Mg132) to probe whether EMC2 exerted a role in FDFT1 protein degradation, showing that while the initial FDFT1 protein levels may differ between control and EMC2-knockdown cells, the degradation curves in both groups become flat and stable after Mg132 treatment (Fig. 5B). Thus, our preliminary findings suggested that EMC2 may regulate protein level of FDFT1 at post-translational level during the initial phase of protein synthesis through the proteasome degradation pathway.
A Effect of EMC2 knockdown on FDFT1 protein stability in the MDA-MB-231 cells and the BT-20 cells treated with protein synthesis inhibitor Cycloheximide (CHX) 0.1 mg/ml for the indicated time. The graph on the right shows the amount of FDFT1 protein remaining after CHX treatment as a percentage of the starting FDFT1 protein level. B Effect of EMC2 knockdown on FDFT1 protein stability in the MDA-MB-231 cells and the BT-20 cells treated with ubiquitination inhibitor Mg132 10 μg/ml for the indicated time. The graph on the right shows the accumulation of FDFT1 protein using the time point 0 h in siNC cells as reference. C GO analysis was used to enrich the functional differences of FDFT1-interacting proteins using BT-20 EMC2 knockdown cells and control cells, 3 replications were examined in each group. The pathways (P.adjust <0.05 and FDR q < 0.25) were ranked by gene ratio and P.adjust. GO analyses include biological process (BP), cellular component (CC), and molecular function (MF). D The Venn diagram shows the E3 ligase binding specifically with the BT-20 shEMC2.1279 proteins interacting with FDFT1 in the siNC group and 1275 proteins interacting with FDFT1 in the siEMC2 group. The E3 ligases list was shown in Table S3. E The interaction of FDFT1 and HRD1 was detected by co-immunoprecipitation in MDA-MB-231 and BT-20 cells after EMC2 knockdown, n = 3. F MDA-MB-231 and BT-20 cells were transfected with HA-Ubiquitin, siNC or siEMC2 as indicated. Ploy ubiquitination of FDFT1 was then examined by immunoprecipitation with beads and analyzed. Cells were treated with Mg132 (10 μg/ml) for 6 h before harvested, n = 3. G The shEMC2 MDA-MB-231 and BT-20 cells were treated with the ERAD pathway inhibitor Eer I (10 μM) for the indicated time for the western blot detection of FDFT1 and EMC2 expression, n = 3. H The presence of EMC2 and FDFT1 in different fractions in the shEMC2 MDA-MB-231 and BT-20 cells was analyzed by western blot, using antibodies against FDFT1 and EMC2, the ER protein calnexin and cytosolic β-actin. Cells were transfected and treated with Eer I (10 μM) as indicated, n = 3. I The immunofluorescence of FDFT1 (pseudo-color: green) and endoplasmic reticulum biomarker Calnexin (pseudo-color: red) after knockdown of EMC2 in the MDA-MB-231 cells and the BT-20 cells for 48 h, n = 3. Hoechst: nuclear counterstaining (pseudo-color: blue). The scale bar is 20 μm. See also Fig. S7.
We next unveil the potential approach by which EMC2 downregulation may induce the degradation of FDFT1. we performed Co-immunoprecipitation (Co-IP) MS to examine the alteration in FDFT1-interacting proteins between siNC and siEMC2 BT20 cells. GO enrichment analysis revealed that FDFT1-interacting proteins were significantly enriched in proteasome degradation-related functions (Fig. 5C). Then, we identified the E3 ubiquitin-protein ligases that bind to FDFT1 in the siEMC2 cells rather than in siNC cells. Intriguingly, we found HMG-CoA reductase degradation 1 (HRD1) (Fig. 5D), a component of ERAD that plays a pivotal role in the ubiquitin-dependent degradation of misfolded endoplasmic reticulum proteins, was among FDFT1 interacting proteins in siEMC2 cells [25]. Co-IP assay confirmed the physical interaction between HRD1 and FDFT1 in EMC2 knockdown cells (Fig. 5E), while there is no direct interaction between HRD1 and FDFT1 in control cells (Fig. S7A). Knockdown of EMC2 prominently elevated the level of FDFT1 ubiquitination by HA-ubiquitination (HA-Ub) pull-down assay (Fig. 5F). To further demonstrate that EMC2 could prevent FDFT1 from ERAD, Eeyarestain I (Eer I), a specific inhibitor of ERAD, was utilized to treat EMC2-knockdown TNBC cells. We observed that EMC2 induced-FDFT1 downregulation was notably restored upon Eer I treatment (Fig. 5G). And the cell fractionation of ER also validated that Eer I effectively recovered FDFT1 levels downregulated by silenced EMC2 in the ER (Fig. 5H). Furthermore, the localization of FDFT1 in ER was reduced in EMC2 downregulated cells by confocal fluorescence microscope (Fig. 5I, S7B). To some extent, these observations suggested that EMC2 could guarantee FDFT1 protein quality control against the ERAD pathway.
EMC2 interacts with HSP90 to maintain FDFT1 protein quality
It was demonstrated that unfolded or misfolded protein can be purged through ERAD pathway [26]. A variety of studies demonstrated that HSP90AA1 plays a critical role in the folding and localization of its client proteins in the ER, and suppression of HSP90AA1 can lead to degradation of its client protein through the ubiquitin-protease pathway [27]. It was reported that EMC2 could bind to the molecular chaperone heat shock protein HSP90, promoting the correct insertion of the protein into the ER for proper folding [22]. HSP90AA1 is an isoform of HSP90 [28]. Thus, we hypothesized that EMC2 may interact with HSP90 to ensure FDFT1 correct folding. Immunofluorescent staining testified the co-localization of EMC2 and HSP90 (Fig. 6A). Endogenously expressed EMC2 interacted with HSP90 was demonstrated in the Co-IP assay by using HSP90 antibody and EMC2 antibody (Fig. 6B). These results verified an interaction between EMC2 and HSP90. Subsequently, we observed that downregulation of HSP90 could decrease the protein level of FDFT1 by western blot assay and do not influence the EMC2 expression (Fig. 6C). Co-IP assay demonstrated that FDFT1 could interact with HSP90 but not with EMC2 directly (Fig. 6D), whereas this interaction was attenuated by EMC2 knockdown (Fig. 6E). To further demonstrate that EMC2 could affect FDFT1 folding, the localization of aggresomes with protein aggregates was detected. EMC2 and HSP90 knockdown could both raise the amounts of aggresomes (Fig. 6F, G). These findings more or less indirectly indicated that the EMC2 could cooperate with HSP90 to sustain the quality control of FDFT1 in the ER.
A Immunofluorescence staining of EMC2 (pseudo-color: green) and HSP90 (pseudo-color: red) in MDA-MB-231 cells and the BT-20 cells, n = 3. Hoechst: nuclear counterstaining (pseudo-color: blue). The scale bar is 50 μm and 20 μm. B The interaction of EMC2 and HSP90 was detected by co-immunoprecipitation in the MDA-MB-231 and BT-20 cells, n = 3. C Western blot showing the expression level of EMC2 and FDFT1 after HSP90 knockdown in MDA-MB-231 and BT-20 cells, n = 3. D The interaction of FDFT1 and HSP90, FDFT1 and EMC2 were detected by co-immunoprecipitation in the MDA-MB-231 and BT-20 cells, n = 3. E The interaction of FDFT1 and HSP90 after EMC2 knockdown for 48 h was detected by co-immunoprecipitation in the MDA-MB-231 and BT-20 cells, n = 3. F Representative images and relative quantifications of aggresome (pseudo-color: red) and nuclear (pseudo-color: blue) in TNBC cells (MDA-MB-231 and BT-20) transfected with siNC and siEMC2 for 48 h, n = 3.The scale bar is 20 μm. p values were calculated by a one-way ANOVA test. G Representative images and relative quantifications of aggresome (pseudo-color: red) and nuclear (pseudo-color: blue) in TNBC cells (MDA-MB-231 and BT-20) transfected with siNC and siEMC2 for 48 h, n = 3. The scale bar is 20 μm. p values were calculated by a one-way ANOVA test.
Inhibition of EMC2 induces ferroptosis partially through decreased cholesterol biosynthesis
To further search out whether EMC2 enhanced TNBC cells viability through elevating cholesterol biosynthesis leading to ferroptosis resistance, we performed GSEA analysis based on the TCGA TNBC dataset and discovered that genes in EMC2 high expression group are significantly enriched in ferroptosis (Fig. 7A). Firstly, we performed a series of experiments to verify that EMC2 could contribute to ferroptosis suppression. EMC2 knockdown caused aberrant smaller mitochondria with increased membrane density (Fig. 7B), increased intracellular concentrations of iron (Fig. 7C and S8A) and lipid reactive oxygen species (ROS) (Fig. 7D and S8B), two surrogate markers for ferroptosis, and the diminished GSH/GSSG ratio in TNBC cells (Fig. 7E and S8C). These findings were further validated in the overexpression model using MDA-MB-468 cells (Fig. S8D–F). We also observed EMC2 knockdown downregulated the ferroptosis defense enzyme GPX4 expression (Fig. 7F, G), which could be induced by the decreased GSH/GSSG ratio. We also treated EMC2-knockdown cells with a ferroptosis inhibitor ferrostatin-1 (Ferr-1) and observed that the inhibitor partially rescued the proliferation defect caused by EMC2 downregulation (Fig. S8G, H). Finally, we determined the positive correlation between EMC2 and GPX4 by using our clinical breast tissues in Renji cohort 3 (Fig. 7H). Besides, we performed several experiments in ER+ (MCF7) and HER2+ (SKBR3) breast cancer cell lines and found t that knockdown of EMC2 caused FDFT1 downregulation, cholesterol content decline and ferroptosis increase. The tendency is as the same in TNBC cells, but more specific in TNBC cell lines (Fig. S9).
A Enrichment plot of the ferroptosis from GSEA analysis between EMC2 high and low expression in TNBC in the TCGA dataset. NES score and nominal Pvalue were given by GSEA software. B Transmission electron microscopy of MDA-MB-231 and BT-20 transfected with siNC and siEMC2 for 48 h. Black arrowheads, mitochondria. The scale bars are 2 μm and 500 nm. A minimum of 5 cells per treatment were examined. C The relative ferrous iron concentration levels in the MDA-MB-231 cells after EMC2 knockdown for 48 h, n = 3. Values are means ± SD. pvalues were calculated by a one-way ANOVA test. D The relative lipid ROS levels in the MDA-MB-231 cells after EMC2 knockdown for 48 h, n = 3. The relative levels of lipid ROS were quantified using flow cytometry of C11-BODIPY lipid peroxidation. Values are means ± SD. pvalues were calculated by a one-way ANOVA test. E The ratio of reductive GSH to oxidative GSH in the MDA-MB-231 cells after EMC2 knockdown for 48 h, n = 3. Values are means ± SD. pvalues were calculated by a one-way ANOVA test. F Western blot showing the expression level of GPX4 after EMC2 knockdown in MDA-MB-231 and BT-20 cells, n = 3. G The immunofluorescence of GPX4 (pseudo-color: red) after knockdown of EMC2 in the MDA-MB-231 cells and the BT-20 cells, n = 3. The scale bar is 50 μm. H Representative images of GPX4 IHC staining according to different EMC2 protein expression levels in breast cancer. The scale bar is 500 μm and 100 μm. See also Fig. S8.
Next, we examined the effects of cholesterol on ferroptosis. The addition of cholesterol resulted in the part restoration of GPX4 protein levels (Fig. 8A) and GSH/GSSG ratio (Fig. 8B), while more reduction of iron level (Fig. 8C) and lipid ROS (Fig. 8D) induced by EMC2 downregulation. When treating EMC2 overexpressed TNBC cells with MCD, GPX4 expression and GSH/GSSG ratio were reduced, while Fe2+ and lipid ROS were increased (Fig. 8E, F). Meanwhile, CFS cultured EMC2 overexpressed TNBC cells led to these ferroptosis-related indicators at the same trend compared with MCD treated TNBC cells (Fig. S10A–D). Since FDFT1 is a key regulator of cholesterol biosynthesis, we also examined that FDFT1 knockdown or overexpression could facilitate or impair the ferroptosis sensitivity (Fig. S11). Re-overexpression of FDFT1 in EMC2-knockdown cells could offset the decrease of the above ferroptosis-related indicators induced by EMC2 downregulation to some extent (Fig. 8G–I). Totally, all these consequences, at the first time, partly manifested that EMC2 may mediate the ferroptosis defense by increasing FDFT1-regulated cholesterol biosynthesis.
A Western blot showing the expression the expression level of GPX4 after the knockdown of EMC2 while adding exogenous cholesterol in the MDA-MB-231 and BT-20 cells. Cells were transfected and treated with cholesterol (10 μM) as indicated, n = 3. B The ratio of reductive GSH to oxidative GSH changes after the knockdown of EMC2 while adding exogenous cholesterol in the MDA-MB-231 and BT-20 cells. Cells were transfected and treated with cholesterol (10 μM) as indicated, n = 3. Values are presented as mean ± SD, tested by a one-way ANOVA test. C The relative ferrous iron concentration level changes after the knockdown of EMC2 while adding exogenous cholesterol in the MDA-MB-231 and BT-20 cells. Cells were transfected and treated with cholesterol (10 μM) as indicated, n = 3. Values are presented as mean ± SD, tested by a one-way ANOVA test. D The relative lipid ROS level changes after the knockdown of EMC2 while adding exogenous cholesterol in the MDA-MB-231 and BT-20 cells. The relative levels of lipid ROS were quantified using flow cytometry of C11-BODIPY lipid peroxidation. Cells were transfected and treated with cholesterol (10 μM) as indicated, n = 3. Values are presented as mean ± SD, tested by a one-way ANOVA test. E The left graph of the western blot showing the expression level of GPX4 after the overexpression of EMC2 while treated with MCD in the MDA-MB-468 cells. Cells were transfected and treated with MCD (5 mM for 6 h) as indicated, n = 3. The right bar chart shows the ratio of reductive GSH to oxidative GSH changes after the overexpression of EMC2 while treated with MCD in the MDA-MB-468 cells. Cells were transfected and treated with MCD (5 mM for 6 h) as indicated, n = 3. Values are presented as mean ± SD, tested by a one-way ANOVA test. F The left bar chart shows the relative lipid ROS level changes after the overexpression of EMC2 while treated with MCD in the MDA-MB-468 cells. Cells were transfected and treated with MCD (5 mM for 6 h) as indicated, n = 3. Values are presented as mean ± SD, tested by two-tailed two-way ANOVA tests. The right bar chart shows the relative ferrous iron concentration level changes after the overexpression of EMC2 while treated with MCD in the MDA-MB-468 cells. Cells were transfected and treated with MCD (5 mM for 6 h) as indicated, n = 3. Values are presented as mean ± SD, tested by a one-way ANOVA test. G The ratio of reductive GSH to oxidative GSH changes after the knockdown of EMC2 while transfected with FDFT1 overexpression plasmid for 48 h in the MDA-MB-231 and BT-20 cells, n = 3. Values are presented as mean ± SD, tested by a one-way ANOVA test. H The relative lipid ROS level changes after the knockdown of EMC2 while transfected with FDFT1 overexpression plasmid for 48 h in the MDA-MB-231 and BT-20 cells, n = 3. Values are presented as mean ± SD, tested by a one-way ANOVA test. I The relative ferrous iron concentration level changes after the knockdown of EMC2 while transfected with FDFT1 overexpression plasmid for 48 h in the MDA-MB-231 and BT-20 cells, n = 3. Values are presented as mean ± SD, tested by a one-way ANOVA test. See also Figs. S9–11.
Discussion
Reprogramming of cholesterol metabolism has been regarded as the specific hallmark during TNBC development [29]. However, what are the roles of cholesterol and how cholesterol biosynthesis is modulated in TNBC are still not well understood. In this study, we first identified that EMC2 was a positive regulator of cholesterol synthesis and could promote TNBC progression by diminishing ferroptosis. Moreover, we first clarified the biological process of cholesterol biosynthesis from the perspective of regulating FDFT1, a key molecule involved in squalene synthesis. Mechanistically, we demonstrated that EMC2 could prevent FDFT1 degradation in an ERAD-dependent pathway by directly interacting with HSP90. Besides, another remarkable finding was the demonstration that increased ferroptosis resistance related to GPX4 was tightly linked to EMC2-regulated cholesterol biosynthesis in TNBC cells. In brief, our innovative findings implied that EMC2 could orchestrate TNBC progression by enhancing the intracellular cholesterol concentration to mediate ferroptosis defense. EMC2 could maintain FDFT1 stability and correct location on the ER membrane through protecting it from ERAD (Fig. S12). Based on these findings, targeting EMC2 could be utilized as a promising strategy to regress tumor growth in the future.
EMC is a conservative protein family that comprise of six subunits in yeast, while ten subunits in human [22, 30]. It participates in a variety of biological processes, including ER protein quality control, organelle communication, ER stress as well as lipid homeostasis [23]. However, the effects of EMC in various cancers are still less investigated. To date, there are limited articles reporting the role and function of EMC subunits in various cancers. EMC6 was reported to play tumor-suppressive roles in glioblastoma cells and gastric cancer cells [31, 32]. Our study firstly unveiled that EMC2 could promote TNBC progression via in vitro and in vivo experiments. Besides, we also identified that EMC2 could be an independent predictor for poor prognosis in TNBC. One research based on bioinformatic analysis indicated that high expression of EMC2 is closely related to the poor clinical outcome of bladder urothelial carcinoma, breast cancer and uveal melanoma, reduced immune cells infiltration in breast cancer, and co-expressed with a variety of oncogenes and drug resistant genes [23]. Totally, all above researches indirectly supported our finding that EMC2 could be an oncogenic gene in TNBC. Furthermore, our study clarified a unique molecular mechanism for the oncogenic function of EMC2 which defenses TNBC ferroptosis by promoting cholesterol synthesis.
Previous studies reported that FDFT1 expression is positively associated with tumor cell proliferation, metastatic ability and histologic grade [15, 33]. Consistent with these publications, our study firstly elucidated that FDFT1 could promote TNBC growth and FDFT1 fulfills its oncogenic role through enhancing intracellular cholesterol synthesis. Generally, squalene biosynthesis is one of the critical processes of the de novo synthesis of cholesterol [4]. FDFT1 encoding squalene synthase (SQS) is mainly responsible for catalyzing two farnesyl diphosphate molecules to synthesize squalene [2, 34, 35]. This finding provides direct evidence that FDFT1 plays independent role in stimulating cancer progression by regulating cholesterol synthesis. It has been reported that SQS inhibitor Zaragozic Acid A could block cholesterol synthesis to induce prostate cancer cell LNCaP death [33]. JLA-9, one plantaricins from lactic acid bacteria, harbors the analogous binding patterns as SQS inhibitors, which could kill lung carcinoma cell line (A549) [36]. To date, no study reported that blocking SQS could restrain breast cancer cells growth and downregulating FDFT1 expression could delay TNBC cells proliferation. We found out that knocking down FDFT1 could inhibit the proliferation of TNBC cells. Based on existing reports and our findings, targeting FDFT1 is expected to become a novel strategy for the treatment of TNBC.
Our study innovatively uncovered that EMC2 may be a critical regulator of FDFT1 so as to orchestrate cholesterol biosynthesis in TNBC. So far, few research is dedicated to investigating the regulatory network of FDFT1, especially in breast cancer. We first explained the exact regulatory mode of FDFT1 from the perspective of posttranscriptional level. EMC2 guaranteed the FDFT1 protein quality through protecting it from ERAD. As one of the two key protein quality control mechanisms in tumor cells, ERAD is primarily responsible for enhancing proteasome activity to degrade misfolded proteins in the ER to adapt to rapid growth and metabolic requirements [19, 37]. FDFT1 is mainly located in the endoplasmic reticulum (ER), which is a cellular compartment for protein folding [38, 39]. Protein-folding ability could make adjustments according to the demand of cell growth or enhanced intracellular metabolism, otherwise excessive misfolded proteins would accumulate in the ER, causing degradation from the ERAD pathway [37, 40]. Interestingly, our study first observed that EMC2 bound with HSP90 to facilitate FDFT1 correct folding in the ER, thus decreasing unfolded or misfolded FDFT1 through ERAD. Ordinarily, HSP90 serves as a chaperone molecule principally for pivotal protein folding and maturation in the ER [41]. Besides, HSP90 overexpresses in many cancer tissues and its absence could lead to cell lethality [42]. Ryuichiro et.al reported that HSP90 could bind with SCAP–SREBP2 complex to stabilize them localizing in the ER, then preventing premature proteasome-dependent degradation of this complex, improving SREBP2 activity, and resulting in lipids biosynthesis augment [43]. However, this investigation did not disclose the details how HSP90 stabilizes SREBPs. Our study revealed that EMC2 interacts with HSP90 to improve the folding ability of FDFT1 and enhance cholesterol synthesis. We also demonstrated that EMC2 mediated TNBC growth partially dependent on FDFT1. The direct causal association between elevated EMC2 and increased FDFT1 protein level provides significant evidence for exploring the potential therapeutic target to treat TNBC in which both EMC2 and FDFT1 were highly expressed.
Intriguingly, we illustrated that EMC2 could lead to ferroptosis resistance in TNBC partly through increasing cholesterol synthesis. In fact, the researches barely focused on the role and function of EMC2 in breast cancer. It was only reported that ECM2 is a ferroptosis-related gene in breast cancer through bioinformatic analyses based on TCGA database [23]. However, whether EMC2 positively or reversely modulates ferroptosis is still not well elucidated. Our study firstly verified that EMC2 may potentially be a negative regulator of ferroptosis through affecting cholesterol biosynthesis pathway in TNBC cells. Previous study unveiled that FDFT1 was a negative regulator of ferroptosis in white people with hepatocyte cancer (HCC) through bioinformatic analyses [44]. Our work also revealed that FDFT1 could lead to ferroptosis defense by increasing intracellular cholesterol synthesis in TNBC. Furthermore, we found that EMC2 could be a positive regulator of FDFT1, so it is plausible that EMC2 may act as a positive regulatory factor in TNBC. However, in liver cancer cells, EMC2 was negatively related to GPX4 expression and could facilitate ferroptosis [44], the underlying mechanism how cholesterol contributes to GPX4 expression to have impact on ferroptosis inhibition requires further exploration.
Totally, our study identified EMC2 as a tumor-promoting factor, which plays a critical role to facilitate cholesterol biosynthesis to resist ferroptosis in TNBC. In short, our findings may provide abundant theoretical basis for a deeper understanding of the regulatory mode of cholesterol synthesis in malignant tumor and accelerate the development of promising strategies for cancer therapy by targeting EMC2 and FDFT1.
