Impact of 17β-HSD12, the 3-ketoacyl-CoA reductase of long-chain fatty acid synthesis, on breast cancer cell proliferation and migration
Metabolic reprogramming of tumor cells involves upregulation of fatty acid (FA) synthesis to support high bioenergetic demands and membrane synthesis. This has been shown for cytosolic synthesis of FAs with up to 16 carbon atoms. Synthesis of long-chain fatty acids (LCFAs), including ω-6 and ω-3 polyunsaturated FAs, takes place at the endoplasmic reticulum. Despite increasing evidence for an important role of LCFAs in cancer, the impact of their synthesis in cancer cell growth has scarcely been studied. Here, we demonstrated that silencing of 17β-hydroxysteroid dehydrogenase type 12 (17β-HSD12), essentially catalyzing the 3-ketoacyl-CoA reduction step in LCFA production, modulates proliferation and migration of breast cancer cells in a cell line-dependent manner. Increased proliferation and migration after 17β-HSD12 knockdown were partly mediated by metabolism of arachidonic acid towards COX2 and CYP1B1-derived eicosanoids. Decreased proliferation was rescued by increased glucose concentration and was preceded by reduced ATP production through oxidative phosphorylation and spare respiratory capacity. In addition, 17β-HSD12 silencing was accompanied by alterations in unfolded protein response, including a decrease in CHOP expression and increase in eIF2α activation and the folding chaperone ERp44. Our study highlights the significance of LCFA biosynthesis for tumor cell physiology and unveils unknown aspects of breast cancer cell heterogeneity.
KeywordsBiosynthesis Cancer 17β-Hydroxysteroid dehydrogenase Long-chain fatty acid Unfolded protein response Endoplasmic reticulum
De novo FA synthesis mainly occurs in tissues with high lipogenic activity, including the liver and adipose tissue. Nevertheless, glucose can be readily utilized by tumor cells for FA synthesis, and it was shown that the rate of lipogenesis in mouse hepatoma is comparable to that in non-neoplastic liver tissue . An upregulation of de novo FA synthesis is generally accepted to have a major role in the metabolic rewiring of cancer cells [4, 5, 6]. FASN is abundantly present in human tumors, including breast cancer , and has been the focus of numerous anti-cancer therapeutic attempts . Several other enzymes involved in FA biosynthesis and its regulation are overexpressed or overactivated in malignant tissues .
The role of cytosolic FA synthesis in cancer is a field of intense research; in contrast, de novo synthesis of LCFAs has received less attention. Saturated and monounsaturated LCFAs derive from elongation of palmitic acid, whereas polyunsaturated fatty acids (PUFAs) originate from two essential FAs, the linoleic acid (LA, C18:2) and α-linolenic acid (ALA, C18:3) that generate the ω-6 and ω-3 FA series, respectively  (Fig. 1b). ω-6 PUFAs have a double bond at position 6, when counting as 1 the carbon atom opposite of the carboxylic group, whereas ω-3 PUFAs have a double bond at the corresponding position 3. PUFAs with well-established physiological functions include the ω-6 arachidonic acid (AA) and the ω-3 docosahexaenoic acid (DHA) and eicosapentaenoic acid (EPA). Metabolism of AA gives rise to key pro-inflammatory eicosanoids, including prostaglandins and leukotrienes, which support immune system evasion and promote tumor cell proliferation and angiogenesis [11, 12]. On the other hand, DHA and EPA give rise to metabolites that dampen inflammatory response and limit malignant cell growth . In many tumors, the ratio of ω-6/ω-3 PUFAs has been reported to be elevated compared to adjacent non-cancerous tissue [14, 15, 16]. Several epidemiological studies found a positive correlation between consumption of ω-3 PUFAs and reduced cancer risk [17, 18, 19]. Based on this, a wealth of in vitro, animal and clinical studies have aimed at evaluating the potential benefits of ω-3 PUFA preventive supplementation or utilization as adjuvant antineoplastic therapy . Despite some promising results, so far findings have not led to definitive conclusions. This could be partially attributed to inter-individual differences in the metabolism of diet-supplemented FAs, including the expression and activity of the involved enzymes. Moreover, the possible significance of de novo biosynthesis and metabolism of LCFAs has been scarcely explored.
Interestingly, in two earlier reports, silencing of 17β-HSD12 expression, the enzyme performing the first reduction step in LCFA elongation (Fig. 1a), in one breast cancer and one ovarian cell line led to decreased proliferation, which could be reverted by AA supplementation [21, 22]. These and one additional study found a correlation between high 17β-HSD12 expression and poor prognosis for survival in patients with breast and ovarian tumors . In contrast, knockdown of 17β-HSD12 in mammary epithelial cells was shown to promote cell proliferation, whereas overexpression of the enzyme led to the opposite effect . Recently, an elaborate transcriptomic study of 17 different cancer types in a total of around 8000 patients was performed, with data of the Cancer Genome Atlas and Human protein Atlas Projects. This analysis showed that 17β-HSD12, along with other genes involved in the elongation and desaturation of LCFAs, correlates with either good or poor prognosis, depending on the tumor type . Importantly, 17β-HSD12 was identified as a prognostic gene for favorable outcome in renal cancer and unfavorable in liver cancer patients. Thus, the role of LCFA synthesis in cancer, including that of 17β-HSD12, is insufficiently understood and deserves further research.
In the current study, we addressed the importance of this biosynthetic pathway by analyzing the consequences of 17β-HSD12 silencing in different breast cancer cell lines. This model is particularly interesting, given the heterogeneous molecular profile of breast tumor cells. We analyzed the impact of 17β-HSD12 downregulation on cell proliferation, migration, and energy metabolism. In addition, we explored whether inhibition of LCFA synthesis at the ER could constitute a stress signal for transformed cells, leading to changes in the unfolded protein response (UPR). Our results show that silencing 17β-HSD12 expression has divergent outcomes in cell survival and invasiveness in different cell lines, and leads to a number of alterations in cancer cell physiology, including energy source utilization. This study contributes to the understanding of how the biosynthetic pathway for LCFAs affects homeostasis of malignant cells, which is a necessary step for evaluating the clinical relevance of ω-3 and ω-6 PUFAs for cancer patient treatment.
Materials and methods
Chemicals and reagents
Trifluoromethoxy carbonyl cyanide phenylhydrazone (FCCP), AA, DHA, EPA, thioridazine (Thio), and 2,3′,4,5′-tetramethoxystilbene (TMS) were purchased from Cayman Chemical (Ann Arbor, MI, USA). All other compounds and reagents were purchased from Sigma-Aldrich (Buchs, Switzerland), unless otherwise specified.
Cell culture and treatments
All cell lines were purchased from the American Type Culture Collection (ATCC, Manassas, VA, USA) and cultured at 37 °C and 5% CO2. SUM159 cells were cultured in Ham’s F12 nutrient mixture (Thermo Fisher Scientific, Waltham, MA USA), supplemented with 5% fetal bovine serum (FBS) and 5 μg/ml bovine pancreas insulin. MCF7 cells were grown in Dulbecco’s Modified Eagle’s medium (DMEM) containing 4.5 g/l glucose, 10% FBS, and nonessential amino acid mixture. For the growth of MDA-MB-453 and MDA-MB-231 cells, RPMI 1640 medium supplemented with 10% FBS was used. This medium was also used in certain experiments with SUM159 cells. For the experiments with different glucose concentrations, RPMI 1640 without glucose (Thermo Fisher Scientific, #11879020) was supplemented with 4 g/l, 2 g/l, or 0.5 g/l glucose. For the experiments to assess the possible effects of glutamine, RPMI 1640 without glutamine (Bioconcept Ltd., Allschwil, Switzerland, #1-41F01-I) was supplemented or not with 2 mM glutamine. In all culture media, 10 mM HEPES buffer pH 7.4 and a mixture of 100 U/ml penicillin/100 μg/ml streptomycin (BioConcept Ltd.) were added.
For transfection of cells with siRNAs, Lipofectamine RNAiMax (Thermo Fisher Scientific) was used. Reverse transfection was performed, with the seeding of cells and siRNA transfection taking place simultaneously. Per 100,000 cells, 20 pmol siRNA and 1 μl lipofectamine reagent was used. The target sequences recognized by the siRNAs are: mock (scrambled) 5′-UGGUUUACAUGUUUUCUGA-3′, HSD17B12 5′-GAACUAAUAUUGUCGGGAA-3′ (‘siRNA1’-used throughout the study), HSD17B12 5′-GAAAUCGGCAUCUUAGUGA-3′ (‘siRNA2’), HSD17B12 5′-UAAGAUGACACAAUUGGUA-3′ (‘siRNA3’), eIF2α 5′-GUA CAA GAG ACC UGG AUA UTT-3′ and RXRα 5′-AGG ACU GCC UGA UUG ACA ATT-3′ (Microsynth AG, Balgach, Switzerland).
In all cell treatments where compound stock solutions were dissolved in DMSO, the final concentration of the latter did not exceed 0.02%. For FA supplementation, 8 mM FAs were conjugated with 10% FA-free BSA after incubation for 1 h at 37 °C, prior to addition to the cell culture medium.
Proliferation measurements with the real-time cell analyzer (RTCA)
Cell proliferation in real time was monitored using the xCELLigence RTCA DP instrument (ACEA Biosciences, Inc., San Diego, CA, USA). 5000 SUM159, 7500 MCF7, 10,000 MDA-MB-453, or 15,000 MDA-MB-231 cells were reverse transfected with siRNAs or left untreated and seeded on an E-plate View 16. Quadruplets of all samples were prepared and impedance signals were measured every 30 min for 96 h. Data were analyzed using the RTCA 2.0 software.
Assessment of cell number by high-content imaging
For estimation of the cell number at different time points on 96-well plates, we fixed cells with 4% formaldehyde and stained nuclei with 5 μg/ml Hoechst-33342 (Thermo Fischer Scientific). Counting of stained nuclei was performed in high throughput with the ArrayScan® high-content imaging system (Thermo Fischer Scientific), according to a protocol adjusted for each cell line. In every experiment, each sample was represented by six technical replicates (different wells) and 14 fields were randomly chosen in each well for counting.
BrdU cell proliferation assay
To assess cell proliferation, we seeded 5000 SUM159 or 7500 MDA-MB-231 cells in 96-well plates and performed reverse transfection with siRNAs as described above. After 48 h, the conditioned medium was replaced by fresh medium-containing 10 μΜ BrdU for 2.5 h. Cells were subsequently washed twice with PBS, fixed with 4% formaldehyde, washed three times with PBS, and permeabilized with 0.2% Triton for 5 min. After three washes with PBS, DNA was denatured with 2 N HCl for 30 min, and cells were washed three times with PBS and once with PBS glycine (0.75% w/v glycine in PBS). After incubation with PBS glycine for 30 min and blocking with 1% BSA for 30 min, rat monoclonal antibody against BrdU was added (Abcam, Cambridge, UK, Ab6326) in a dilution of 1:360 in 1% BSA. Following three washes with PBS, goat anti-rat IgG Alexa Fluor 555 (Thermo Fischer Scientific) was added in a dilution of 1:360 along with 5 μg/ml Hoechst and incubated for 40 min. Cells were washed four times with PBS and cells positive for BrdU staining counted with the ArrayScan® high-content imaging system (Thermo Fischer Scientific), using the BGRFR_386_23 and BGRFR_659_15 filters for detection of cell nuclei and BrdU-positive cells, respectively.
Migration and adhesion assays
To assess migration potential, cells were transfected with siRNAs and seeded on a 35 mm dish. At 48 h post-transfection, cells were reseeded on a trans-well insert (diameter 6.5 mm, pore size 8 µm; Corning Inc., Lowell, MA, USA). The cell numbers seeded were 7500 for SUM159, 20,000 for MCF7, 30,000 for MDA-MB-453 and 15,000 for MDA-MB-231 cells. The cells were allowed to migrate for 24 h from medium-containing 1% FBS for SUM159 and MDA-MB-231 or 0.5% FBS for MCF7 and MDA-MB-453 cells towards medium-containing 10% FBS. Subsequently, cells were stained with 0.1% crystal violet (w/v in 25% methanol). After removal of the cells at the top chamber, migrated cells were visualized under the 20 × objective of a light microscope (Zeiss Axiovert 100; Carl Zeiss Microscopy GmbH, Feldbach, Switzerland). Images were captured from ten fields using a digital camera (EOS 70D, Canon, Tokyo, Japan) and cells counted with the ImageJ software (National Institute of Health, MD, USA). For the experiments where glutamine was assessed, 48 h after siRNA transfection, 40,000 MDA-MB-231 cells were seeded on a CIM-Plate 16 (ACEA Biosciences Inc.) and migration was measured over a period of 24 h using the xCELLigence RTCA DP instrument.
Cell adhesion assays were performed essentially as elaborated in Ref. .
Quantitative polymerase chain reaction (qPCR)
Oligonucleotide primers used for qPCR
Sense/antisense primers (5′–3′)
Cell lysis and sample preparation for western blot have been described before . The antibodies used for detection of specific proteins and their final concentrations in 5% defatted milk blocking solution were the following.
17β-HSD12 0.05 μg/ml (SIGMA-Aldrich HPA016427), α-tubulin 0.062 μg/ml (GeneTex, Irvine, CA, USA, #GTX628802), actin 0.67 μg/ml (Santa Cruz Biotechnology, sc-1616), PPIA 0.05 μg/ml (Abcam, Ab41684), PERK 0.1 μg/ml (Cell signaling Technology, Danvers, MA, USA, #3192), eIF2α 0.1 μg/ml (Cell Signaling Technology, #9722S), peIF2α 0.3 μg/ml (Cell Signaling Technology, #119A11), ATF4 0.1 μg/ml (Cell Signaling Technology, #11815S), ATF6 0.25 μg/ml (Cell Signaling Technology, #65880S), CHOP 1.2 μg/ml (Cell Signaling Technology, #2895S), sXBP1 0.1 μg/ml (Cell Signaling Technology, #12782S), GRP78 0.62 μg/ml (BD Bioscience, San Jose, CA, USA, #610978), ERp44 1:1000 dilution , PDI 0.1 μg/ml (Abcam, Cambridge, UK, Ab2792), ERp72 1:1000 dilution (Stressgen), GRP94 1:1000 dilution , FADS1 0.5 μg/ml (Santa Cruz Biotechnology, sc-134337), FADS2 0.5 μg/ml (Abcam, ab72189), G6PD 1 μg/ml (Bethyl Laboratories, TX, USA, A300-404A), PGD 1:1000 (Abcam, ab129199) and RXRα 0.1 μg/ml (Santa Cruz Biotechnology, sc-515929), Akt 0.1 μg/ml (Abcam, ab126811), pAkt Ser473 0.1 μg/ml (Cell Signaling Technology, #9271), Caspase-9 (Cas-9)/cleaved Cas-9 0.1 μg/ml (Cell Signaling Technology, #9508), Caspase-3 (Cas-3) 0.1 μg/ml (Cell Signaling Technology, #9665), cleaved Cas-3 0.1 μg/ml (Cell Signaling Technology, #9664), Bcl-2 0.1 μg/ml (Cell Signaling Technology, #2870), Bax 0.1 μg/ml (Cell Signaling Technology, #5023), and ELOVL5 1 μg/ml (Origene, #TA315700). Samples were not boiled for detection of ELOVL5, due to the protein aggregation occurring otherwise. The signal density of protein bands was measured using the ROI manager tool of ImageJ. For relative protein-level quantification, the density of each band was compared to that of the loading control (actin, α-tubulin or PPIA) in every lane. Different loading controls were used depending on the size of the proteins of interest and the species origin of the primary antibody.
Immunofluorescence staining and microscopy
Indirect immunofluorescence staining has been described earlier . Visualization of the fluorescence signal and image acquisition was performed utilizing the 20 × objective of a DMI4000 B Microscope (Leica Microsystems, Wetzlar, Germany) and processed with the ImageJ software.
Seahorse XF Cell Mito Stress Test
Measurements of oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were performed in real time with the Seahorse XF96 analyzer (Agilent Technologies, CA, USA). The different parameters of mitochondrial respiration were evaluated using the Seahorse XF Cell Mito Stress Test, according to the manufacturer’s instructions. Cells were transfected 48 h prior to the assays with siRNAs and seeded in six replicates per sample on a Seahorse XF96 V3 PS Cell Culture Microplate (Agilent Technologies). In this assay, baseline cellular OCR was measured, representing both mitochondrial and non-mitochondrial respiration (Suppl. Figure 7a). The proportion of mitochondrial OCR linked with ATP production is measured after addition of the ATPase inhibitor oligomycin at a concentration of 1 μΜ. Subsequently, 0.5 μM of the uncoupling agent FCCP is added, which collapses the electron gradient across the mitochondrial membrane and leads to a compensatory maximal oxygen consumption from the mitochondria, thereby enabling measurement of maximal respiration. The spare respiratory capacity is calculated by deduction of the basal mitochondrial respiration from this value. Finally, a mixture of 0.5 μM of antimycin and 0.5 μM rotenone is added, inhibiting function of complex III and I, and thus, completely shutting down mitochondrial oxygen consumption. This allows for calculation of non-mitochondrial OCR. Deduction of this value from that obtained after oligomycin treatment yields the portion of OCR resulting from proton leak. Basal respiration is the initial OCR value excluding non-mitochondrial respiration. Eto or Thio was added at a concentration of 40 μΜ 15 min prior to the commencement of the assay. Following the measurements, values were normalized to the protein content of each sample using the Pierce™ BCA Protein Assay Kit (Thermo Fisher Scientific, MA, USA). Data were analyzed employing the Wave 2.6.0 software.
Quantification of AA
To quantify AA in cells, the competitive AA ELISA assay was used (OKEH02583, Aviva Systems Biology, CA, USA), in line with the manufacturer’s guidelines. AA was measured in transfected cells 48 h post-siRNA delivery. Cells were subjected to five freeze/thaw cycles to break up cell membranes, and samples were centrifuged for 20 min at 1200g, 4 °C. Subsequently, the supernatant was collected and assayed on the ELISA plate. The amount of AA was calculated based on the standard curve of serially diluted standards provided with the kit. To normalize to the protein content of each sample, the Pierce™ BCA Protein Assay was employed.
Quantitative and statistical analysis
All experiments in this study were performed at least three times independently and values shown in graphs are the mean of these experiments (with the exception of xCELLigence experiments where a representative measurement is shown in Fig. 1c). Error bars represent standard deviations (SD). Statistical significance of the difference between conditions was calculated with a two-tailed Student’s t test. The levels of significance are as follows: *p < 0.05, **p < 0.01, and ***p < 0.001.
Effect of 17β-HSD12 downregulation on breast cancer cell proliferation and migration
To assess the impact of 17β-HSD12 on proliferation and migratory potential of tumor cells, we used four breast cancer cell lines, each representing one of the three main molecular subtypes of breast cancer; the MCF7 that is positive for the estrogen receptor-α (ER) and progesterone receptor (PR) but negative for the human epidermal growth factor receptor 2 (HER2), the MDA-MB-453 that is negative for ER and PR but positive for HER2, and the SUM159 and MDA-MB-231 that are triple-negative (ER−, PR−, HER2−). First, we evaluated the efficiency of siRNA knockdown over time (24 h, 48 h, and 72 h) in all cell lines by western blot, using a specific antibody against 17β-HSD12. In all cases, we found a time-dependent decrease in protein levels after siRNA transfection compared to mock- and non-transfected cells (Suppl. Figure 1). Next, we tested the effect of 17β-HSD12 downregulation on cell proliferation with live measurements up to 96 h after siRNA transfection using the xCELLigence RTCA (Fig. 1c, left panel). In parallel, we performed endpoint measurements of cell numbers at 24 h, 48 h, and 72 h, by Hoechst-33342 staining of nuclei and high-content imaging analysis (Fig. 1c, right panel). The results from both approaches were in accordance. Downregulation of 17β-HSD12 led to decreased proliferation of SUM159 cells and increased proliferation of MCF7, MDA-MB-453 and MDA-MB-231 cells. The time after knockdown when differences in proliferation were statistically significant varied, which can partly be attributed to the different growth rates among cell lines. To validate the specificity of the siRNA against 17β-HSD12 used in the above experiments (‘siRNA1’), we employed two additional siRNAs (‘siRNA2’ and ‘siRNA3’) in experiments with Hoechst-33342 staining and high-content imaging for the SUM159 and MDA-MB-231 cells and obtained similar results (Suppl. Figure 2).
Subsequently, we explored whether the reduction in cell number following 17β-HSD12 knockdown in SUM159 cells could be due to increased apoptosis. We found unchanged ratios of cleaved Cas-9/Cas-9, as well as cleaved Cas-3/Cas-3 at 48 h post-transfection with 17β-HSD12 siRNA (Suppl. Figure 3a). In addition, the ratio of the apoptosis regulators Bax/Bcl-2, which is used as an indicator of apoptotic levels, was not altered (Suppl. Figure 3a). Moreover, we examined whether ferroptosis, a non-apoptotic form of cell death involving iron-dependent accumulation of lipid reactive oxygen species, could contribute to the reduced cell number after 17β-HSD12 downregulation in SUM159 cells. For this purpose, we treated cells with 1 μΜ of the ferroptosis inhibitor ferrostatin and knocked-down 17β-HSD12. Our data showed a significant reduction in cell number upon ferrostatin treatment, as observed in the absence of the inhibitor (Suppl. Figure 3b). Taken together, the above results do not support an involvement of apoptosis or ferroptosis as mechanisms implicated in the 17β-HSD12-knockdown phenotype in SUM159 cells.
We further investigated the changes accompanying the altered migration in SUM159 and MDA-MB-231 cells after 17β-HSD12 knockdown. Specifically, we analyzed the mRNA expression levels of genes involved in cell migration and adhesion during cancer metastasis. In SUM159 cells, an increase in matrix metalloproteinase 2 (MMP2) was observed after 17β-HSD12 silencing (Fig. 2b). Although no expression of MMP2 could be detected in MDA-MB-231 cells, 17β-HSD12 knockdown led to decreased expression of MMP1 and of the intermediate filament protein vimentin (VIM). One would expect that these changes lead to decreased adhesion in SUM159 cells due to increased MMP2 levels and increased adhesion in MDA-MB-231 cells due to decreased MMP1 and VIM levels. However, in cell adhesion assays using fibronectin-coated dishes, no difference in cell adhesion was observed after 17β-HSD12 downregulation in neither cell line (Fig. 2b).
Investigation of the differential impact of 17β-HSD12 on proliferation and migration of SUM159 and MDA-MB-231 cells
After excluding the culture medium as a determining factor, we compared the expression in the two cell lines of the different ELOVL enzymes (ELOVL2, ELOVL5, and ELOVL7) and the fatty acid desaturases (FADS1 and FADS2), since these enzymes preferentially catalyze distinct steps during PUFA biosynthesis (Fig. 1b). We observed that the mRNA and protein expression of ELOVL5 was significantly lower in the SUM159 cells (Fig. 3b, c). Although FADS2 mRNA levels were higher in MDA-MB-231 compared to SUM159 cells, protein expression was comparable in both cell lines (Fig. 3b, c). The mRNA of ELOVL7 that is also able to elongate certain LCFAs in a tissue-restricted manner is present at equally low levels in both cell lines. Interestingly, we could not detect any mRNA or protein expression of ELOVL2 in neither cell line. Downregulation of 17β-HSD12 did not lead to changes in expression of the analyzed enzymes, with the exception of decreased FADS1 mRNA and protein levels in both cell lines (Fig. 3b, c), which may further accentuate the 17β-HSD12-knockdown phenotype.
Rescue of the effect of 17β-HSD12 silencing on cancer cell proliferation and migration
One of the most extensively investigated AA metabolites is prostaglandin E2 (PGE2), which is produced through the enzymatic function of cyclooxygenase 2 (Cox2). It is possible that in MDA-MB-231 cells, 17β-HSD12 antagonizes this pathway of AA metabolism by further elongating it towards longer FAs that do not support growth and migration. Interestingly, we found that the PGE2 cell surface receptor 2 (PTGER2) is expressed in MDA-MB-231 but not SUM159 cells and its levels are significantly upregulated after 17β-HSD12 downregulation (Fig. 4c). We could not detect mRNA expression of other PGE2 receptors (PTGER1, PTGER3, and PTGER4) in MDA-MB-231 or SUM159 cells. Subsequently, we examined whether the Cox2 pathway of AA metabolism could be responsible for the increased proliferation upon 17β-HSD12 silencing in MDA-MB-231 cells by treating mock- or 17β-HSD12 siRNA-transfected cells with the selective inhibitor Celecoxib (CEL). CEL treatment reverted the increased proliferation due to 17β-HSD12 downregulation in MDA-MB-231 cells (Fig. 4c; Suppl. Figure 6c). In contrast, CEL treatment had no effect on the decreased proliferation of SUM159 cells after 17β-HSD12 knockdown, supporting that this pathway does not mediate the effect of 17β-HSD12 downregulation in this cell line (Fig. 4c; Suppl. Figure 6c). Furthermore, CEL was able to inhibit the increased migration triggered by 17β-HSD12 knockdown in the MDA-MB-231 cells (Fig. 4d).
We then explored whether other pathways of AA metabolism are involved in the increased proliferation after 17β-HSD12 silencing in MDA-MB-231 cells by analyzing mRNA expression of the responsible enzymes. First, we tested the lipoxygenase enzymes (LOX5, LOX12, and LOX15) through which leukotrienes are produced. We did not find mRNA expression of any LOX enzyme in MDA-MB-231 or SUM159 cells, with the exception of LOX5 that was expressed at low levels in both. We then tested for expression of CYP450 enzymes, which convert AA to a plethora of biologically active metabolites, including epoxyeicosatetraenoic acids (EETs) and hydroxyeicosatetraenoic acids (HETEs). Among the CYP450 enzymes analyzed (CYP1A2, CYP2J2, CYP4V2, CYP1B1, and CYP4F22), we detected measurable mRNA levels of CYP4V2 and CYP1B1. CYP4V2 was expressed at comparable levels in the two cell lines, but CYP1B1 mRNA was substantially expressed only in MDA-MB-231 cells, suggesting a possible role of this enzyme in the observed effects (Fig. 4e). To test whether CYP1B1 could mediate the increase in proliferation and migration induced after 17β-HSD12 downregulation in MDA-MB-231 cells, we used the CYP1B1 inhibitor TMS. TMS treatment abolished the 17β-HSD12 downregulation-dependent increase in proliferation and migration of MDA-MB-231 cells (Fig. 4f). AA incorporated into membrane phospholipids has been shown to promote proliferation through the Akt survival pathway. Unexpectedly, knockdown of 17β-HSD12 in MDA-MB-231 cells decreased the ratio of activated Akt phosphorylated on Ser473 to total Akt levels (Suppl. Figure 6d). Intriguingly, measurement of AA levels with a competitive ELISA assay showed that the levels of cellular AA are lower in SUM159 compared to MDA-MB-231 cells (Fig. 4g). Furthermore, 17β-HSD12 knockdown did not alter AA levels in SUM159, while it decreased its levels in MDA-MB-231 cells. This could be due to upregulation of AA processing by COX and CYP enzymes to downstream metabolites, which supports the results of the previously described findings (Fig. 4c, d, f).
Impact of 17β-HSD12 silencing on energy metabolism
To further explore possible sources of energy and/or building blocks permitting increased proliferation and migratory potential of MDA-MB-231 cells after 17β-HSD12 silencing, we investigated whether glutamine metabolism, a major pathway in supporting cancer cell growth, was involved . Our results showed that glutamine depletion for 48 h did not alter the 17β-HSD12 knockdown-dependent increase in proliferation and migration (Suppl. Figure 8a).
Another fundamental cellular energy source is glucose. Through glycolysis, glucose can be converted to pyruvate and then lactate, resulting in extrusion of protons into the extracellular medium. Pyruvate produced through glycolysis can also enter the TCA cycle in mitochondria, which produces NADH and FADH2 that are utilized in OXPHOS to generate more ATP. Alternatively, glucose can enter the pentose phosphate pathway (PPP) that produces NADPH as well as important building blocks, such as ribonucleotides. The first rate-limiting enzyme in this metabolic pathway is glucose 6-phosphate dehydrogenase (G6PD), which converts glucose 6-phosphate to 6-phosphogluconate, thereby generating NADPH. To gain more insight into the role of glucose after 17β-HSD12 knockdown, we first measured in real-time ECAR as a measure of glycolysis. Our results showed no significant difference in the levels of ECAR after 17β-HSD12 silencing in MDA-MB-231 or SUM159 cells (Fig. 6b). Inhibition of β-oxidation with Eto led to a compensatory increase in glycolysis in both cell lines, whereas Thio had no effect. However, Thio caused significant decrease in glycolysis after 17β-HSD12 knockdown only in SUM159 cells. To conclude whether glucose could play another role in growth of these cell lines following 17β-HSD12 silencing, we cultured SUM159 and MDA-MB-231 cells in RPMI 1640 medium-containing normal (2 g/l), higher (4 g/l), or lower (0.5 g/l) glucose concentration (Fig. 6c). Surprisingly, increasing the amount of glucose in MDA-MB-231 cells dampened the 17β-HSD12-knockdown phenotype, and decreasing glucose concentration led to a further accentuation of the phenotype. This was opposite to what we observed in SUM159 cells, where increased glucose rescued the reduction in proliferation following 17β-HSD12 silencing (Fig. 6c). No change in the 17β-HSD12-knockdown phenotype was observed at lower glucose concentrations. Interestingly, G6PD expression was found to be significantly increased in SUM159 after knockdown of 17β-HSD12, while it was unaltered in MDA-MB-231 cells (Fig. 6d). No difference in the expression of the second NADPH-producing enzyme of the PPP, phosphogluconate dehydrogenase (PGD) was observed in neither cell line after 17β-HSD12 knockdown (Suppl. Figure 8b). Taken together, the above results point towards a significant difference in glucose utilization between the two cell lines.
Alteration in ER stress response and folding pathways after 17β-HSD12 knockdown
Long-chain fatty acids elongation takes place at the ER, the compartment where the cellular machineries for protein folding and stress response pathways reside. Nutrient depletion or defects in biosynthetic pathways induce a multitude of stress signals, although this has not been investigated for LCFAs. To address this, we first assessed the effects of 17β-HSD12 silencing on the UPR pathway. The UPR is activated through diverse stressors, including overload of protein folding demand in the ER, oxidative stress, and shortage in energy sources. It is well recognized that cancer cells exploit the UPR to sustain their high proliferation rates, resulting in constitutive overactivation of several UPR components . The UPR consists of three main branches ; the protein kinase R-like ER kinase/eukaryotic initiation factor 2α (PERK/eIF2α), the inositol-requiring enzyme 1α/split X-box-binding protein 1 (IRE1α-Xbp1), and the activating transcription factor 6 (ATF6).
Since the UPR pathway is closely related to protein folding, we also examined a potential influence of 17β-HSD12 downregulation on several protein folding components. Among the proteins involved in folding are the heat-shock responsive proteins, including Grp78 and Grp94, which were not affected by silencing of 17β-HSD12 (Suppl. Figure 10a). One of the most important families of folding-control proteins is the protein disulfide isomerases (PDI). Although the expression of two members of this family, namely PDI and ERp72, was not changed following 17β-HSD12 downregulation (Suppl. Figure 10b), there was a significant increase in ERp44 expression at the mRNA and protein levels in both SUM159 and MDA-MB-231 cells (Fig. 8d; Suppl. Figure 10c). Given that eIF2α was overactivated after 17β-HSD12 knockdown, we examined whether it could be involved in stabilizing the ERp44 mRNA. In contrast to this hypothesis, knockdown of eIF2α led to a small rise in the levels of ERp44 (Suppl. Figure 10d).
The rapid growth of cancer cells requires a high amount of energy and macromolecules, including carbohydrates, proteins, lipids, and nucleic acids. These demands are largely met by metabolic rewiring, which entails de novo synthesis of cellular building blocks. The role of FAs in malignant transformation is highly relevant, with a multitude of alterations in FA synthesis and break down through β-oxidation reported in several types of cancer . Besides, certain FAs act as signaling molecules and are involved in metabolic adaptation. This work addressed the role of LCFA biosynthesis for cancer cells. Of the four enzymatic entities participating in LCFA synthesis, 17β-HSD12 has garnered attention regarding its role in cancer through studies yielding conflicting results. While some studies reported an association of 17β-HSD12 with more aggressive tumor progression and poor outcome [21, 22, 23], others found a suppressive role . Using different breast cancer cells, we showed that silencing of 17β-HSD12 either increases or decreases cancer cell proliferation and migration depending on the cellular context. Importantly, two triple-negative breast cancer cell lines, MDA-MB-231 and SUM159, exhibit opposite phenotypes with increased and decreased proliferation and migration, respectively, after 17β-HSD12 knockdown, emphasizing the heterogeneity of breast cancer cells (results summarized in Fig. 8e). In a previous report, 17β-HSD12 downregulation caused a dramatic decrease in proliferation after culturing the cells in serum-free medium during knockdown, a condition that severely hinders cell growth . Our experiments were performed under normal culture conditions, highlighting that even under sufficient energy supply, impaired LCFA synthesis affects cancer cell proliferation and invasiveness. The effect of 17β-HSD12 knockdown on cell migration was followed by changes in the levels of MMPs that were not accompanied by altered cell adhesion. This suggests that they likely represent offset mechanisms to initial alterations in the molecular network controlling cell substrate attachment after 17β-HSD12 silencing.
To understand the differential impact of 17β-HSD12 downregulation, we compared MDA-MB-231 and SUM159 cells. Differences in the culture medium could not explain the observed divergent phenotypes of the two cell lines following 17β-HSD12 silencing; therefore, we explored the expression of enzymes participating in LCFA synthesis, focusing on ω-6 and ω-3 PUFAs. Since HACDs 1–4 show tissue but not substrate specificities, and only one TER and one KAR enzyme have been identified so far [38, 39], we tested whether distinct expression of ELOVLs with different substrate specificities directs differential FA synthesis in the two cell lines. Expression of ELOVL2, which is indispensable for the elongation of PUFAs with 22 carbon atoms [40, 41] (Fig. 1b), was not detected in the two cell lines, suggesting that this process does not take place in these cells and resembling the situation in proliferating T cells . In addition, we found that ELOVL5 mRNA and protein expression was lower in SUM159 compared to MDA-MB-231 cells. ELOVL5 elongates 18 carbon PUFAs as well as PUFAs with 20 carbon atoms such as AA and EPA, and several studies indicated a preference of this enzyme for ω-3 FAs [40, 42, 43, 44, 45]. This could mean that at the low ELOVL5 expression in SUM159 cells, the activity is directed towards EPA over AA elongation. The FADS enzymes also display substrate specificities in PUFA elongation (Fig. 1b). Although we found much lower FADS2 mRNA levels in SUM159 than in MDA-MB-231 cells, western blot revealed similar protein expression in both cell lines. This discrepancy between mRNA and protein levels could be due to a higher rate of FADS2 mRNA translation in SUM159 compared with MDA-MB-231 cells, resulting in comparable protein levels. Nevertheless, because protein expression is not significantly different between the two cells, FADS2 unlikely is responsible for the observed differences in their proliferation after 17β-HSD12 knockdown. Future experiments should include a complete proteomics profiling in the two cancer cell lines to unveil key differences in other enzymes participating in synthesis of saturated and monounsaturated LCFAs.
According to one of our hypotheses, the production of different LCFAs in MDA-MB-231 and SUM159 cells could account for the different phenotypes after 17β-HSD12 knockdown. Indeed, supplementation with EPA following 17β-HSD12 silencing rescued the decreased proliferation and migration of SUM159 cells. It is unclear whether this is a direct effect or caused through formation of downstream metabolites. AA or DHA supplementation failed to revert the 17β-HSD12-knockdown phenotype, supporting an EPA-specific effect. This finding was unexpected, since EPA was associated with suppression of breast cancer cell growth . However, EPA treatment was shown to increase proliferation of RAW264.7 macrophages and of a B-lymphocyte cell line, which in the latter was attributed to the absence of EPA-to-DHA conversion [47, 48]. Due to swift conversion of EPA to DHA in certain cell types, the effects of the two FAs cannot always be distinguished. In MDA-MB-231 cells, AA exacerbated the 17β-HSD12 siRNA-induced rise in cell proliferation, whereas EPA and DHA had no impact. We reasoned that this might be due to AA conversion towards eicosanoids promoting cancer cell growth and invasiveness. Indeed, blocking the activities of COX2 and CYP1B1 that are involved in this metabolism reverted the effect of 17β-HSD12 knockdown on MDA-MB-231 cells, indicating that in this cell line, elongation of AA competes with its processing through eicosanoid synthesis pathways. Of note, CYP1B1 is not expressed in SUM159 cells, and inhibiting COX2 did not alter the proliferation phenotype after 17β-HSD12 silencing. Importantly, 17β-HSD12 silencing led to a change in AA levels exclusively in MDA-MB-231 cells. Downstream metabolism of AA through COX and CYP enzymes may, consequently, promote cancer cell growth and migratory potential. Taken together, our results suggest that the metabolism of AA and EPA is implicated in the observed phenotypes after 17β-HSD12 silencing in MDA-MB-231 and SUM159 cells, respectively. In addition, differential AA metabolism emerges as a crucial divergent point between the two cell lines, which offers an explanation for the differential 17β-HSD12-knockdown phenotype. After AA release from membrane phospholipids by phospholipase A2, cell proliferation was found to be induced via activation of the serine–threonine kinase Akt [49, 50, 51]. Surprisingly, our results showed reduced phosphorylation of Akt at Ser473 at 48 h after 17β-HSD12 downregulation. This finding may be explained by a negative feedback loop after a transient increase in Akt phosphorylation or alternative survival pathways in this cell system.
Since our analysis covered only a small fraction of the possible FAs affected by 17β-HSD12 downregulation, future lipidomics analysis should provide a comprehensive picture of the changes following defects in LCFA synthesis in different cancer cell lines. Reliable methods for detailed and time-effective analysis of lipidomes in cells and various organisms have lately been developed . Our preliminary experiments in SUM159 cells indicate a shift from longer to shorter chain LCFAs upon 17β-HSD12 knockdown (data not shown).
Since FAs are hardly soluble in aqueous solutions, for our experiments, FAs were conjugated with BSA, as previously described [53, 54]. Before entering cells, FAs dissociate from BSA and are intracellularly transported in a free form either through diffusion or protein-mediated transport . In addition, free FAs can bind to G protein-coupled receptors at the cell surface, thereby exerting signaling effects that can affect tumor growth and chemoresistance [56, 57]. The most important receptors mediating LCFA action are GPR40 and GPR120, whereby GPR40 was shown to potentiate some of the effects of free FAs in MDA-MB-231 cells [58, 59]. Thus, it would be interesting to explore whether the phenotypes observed after 17β-HSD12 knockdown and treatment of cells with FAs are influenced by these receptors.
By altering the metabolic phenotype of cancer cells, impaired LCFA synthesis could mediate effects on cell proliferation upon 17β-HSD12 silencing. Our results revealed that 17β-HSD12 knockdown decreased ATP production through OXPHOS in SUM159 cells, whereas MDA-MB-231 cells appeared resistant. It should be noted that these experiments were performed 48 h after 17β-HSD12 siRNA transfection, a time point when SUM159 cell proliferation was not yet significantly reduced, implying that the decreased OXPHOS precedes impairments of proliferation and suggesting a causative effect. Inhibition of peroxisomal β-oxidation with Thio markedly decreased maximal respiration and spare respiratory capacity, which may have crucial consequences when cells grow under conditions requiring mobilization of their energy reserves. Strikingly, blockage of peroxisomal β-oxidation led to an increased proton leak. Proton leak is considered a protective mechanism against reactive oxygen species (ROS) generation and, in turn, ROS are known to promote proton leak . The observed rise in proton leak could be a consequence of increased ROS after Thio treatment or a reaction to increased ROS formation. Blocking peroxisomal β-oxidation could augment the amount of free LCFAs, which have been shown to carry protons across the mitochondrial membrane, by shuffling between the protonated and non-protonated states [61, 62]. Inhibiting mitochondrial β-oxidation severely blocks OXPHOS in both cell lines, indicating that breakdown of FAs is a major contributor to mitochondrial ATP production. This is mainly mediated by acetyl-CoA production, the ultimate product of β-oxidation, which can readily enter the TCA cycle that subsequently fuels the electron transport chain . Since we found β-oxidation to be a prevailing factor in sustaining ATP production through OXPHOS, we examined whether it mediates the increased proliferation observed after 17β-HSD12 downregulation in MDA-MB-231 cells. In line with our hypothesis, inhibition of mitochondrial or peroxisomal β-oxidation attenuated the 17β-HSD12-knockdown effect on the proliferation of these cells. Although peroxisomal β-oxidation does not directly yield ATP, further breakdown of FAs in mitochondria links the two β-oxidation pathways.
A key metabolite supporting cancer cell growth is glucose. Based on live measurements of ECAR, we concluded that 17β-HSD12 downregulation does not affect glycolysis in neither cell line. However, the expression of G6PD, the enzyme catalyzing the first and rate-limiting enzyme in the PPP of glucose metabolism, was upregulated upon silencing of 17β-HSD12 only in SUM159 cells. Through the enzymatic conversion of glucose 6-phosphate, G6PD generates NADPH, an essential co-factor for the synthesis of short-chain FAs by FAS in the cytosol. The importance of the PPP for tumor cells has recently started to be recognized . Silencing of G6PD MCF7 breast cancer cells reduced lipid synthesis and proliferation . Importantly, in the present study, increased glucose concentration in the culture medium fully restored proliferation after 17β-HSD12 knockdown in SUM159 cells. In contrast, increased glucose concentration suppressed the increased proliferation upon 17β-HSD12 downregulation in MDA-MB-231 cells, while decreased glucose concentration further enhanced proliferation. These results imply that the cellular pathways in which glucose participates are competitive to the ones that MDA-MB-231 cells utilize to boost proliferation after 17β-HSD12 knockdown. This may be explained by competition between glycolysis and the β-oxidation pathway that sustains high proliferation after 17β-HSD12 silencing. Although both glycolysis and β-oxidation produce acetyl-CoA that fuels the TCA cycle, β-oxidation is more efficient by providing a higher number of carbon atoms per molecule . β-oxidation of FAs can thus generate more ATP molecules than glycolysis. In accordance, we demonstrated that the majority of OXPHOS in both cancer cell lines is fueled by β-oxidation.
It has recently been recognized that the UPR pathway is of paramount importance for tumor cell growth, due to its role in maintaining excessive bioenergetic needs and growth under hypoxic and nutrient-limited conditions. Importantly, the UPR is constantly alert to changes in lipid metabolism. For example, membrane lipid composition directly triggers the UPR, irrespective of proteostasis, and perturbations in ER membrane lipid composition activate several UPR components, including PERK and IRE1 [67, 68]. Interestingly, we found that 17β-HSD12 downregulation induced the phosphorylation/activation of eIF2α. Downregulation of eIF2α drastically decreased 17β-HSD12 levels, implying that 17β-HSD12 expression is controlled by eIF2α, and mimicked some of the effects observed after 17β-HSD12 knockdown, such as the decreased proliferation of SUM159 cells. The increased phosphorylation of eIF2α after 17β-HSD12 knockdown could thus represent a compensatory mechanism. Upon activation, eIF2α induces a global reduction in protein synthesis, but specifically enhances translation of certain mRNAs through the control of translation from upstream open reading frames (sORF) . The only mRNA that has been unequivocally shown to be subjected to such regulation is that of ATF4. Since ATF4 was reduced when eIF2α was activated in the SUM159 cells, and we did not observe any general attenuation of protein synthesis in neither cell line after 17β-HSD12 knockdown, we conclude that activation of eIF2a is a response to the decreased 17β-HSD12 mRNA levels. In this context, it would be interesting to investigate whether 17β-HSD12 is an mRNA stabilized by eIF2α and whether other lipid metabolizing enzymes could be subject to this type of regulation.
Another UPR component modified by 17β-HSD12 knockdown in SUM159 and MDA-MB-231 cells was CHOP, exhibiting a decrease at both mRNA and protein levels. CHOP is a transcription factor that can be activated by both the eIF2a/ATF4 and ATF6 branches of the UPR, and has been associated with ER stress-induced apoptosis . However, CHOP is abundantly expressed in MDA-MB-231 and SUM159 cells at steady state, pointing towards an alternative function of this factor in these cells. In addition, CHOP expression was reduced after 17β-HSD12 knockdown in both cell lines, independently of the proliferation rate. Recent evidence suggests that CHOP possesses a dual function; it regulates apoptosis after extensive stressful stimuli, such as amino acid deprivation, but also induces the expression of autophagy-related genes at initial stages of starvation [71, 72]. Therefore, CHOP expression in tumor cells at basal levels may control autophagy, a cellular process with a crucial role for survival of many tumors . Among others, autophagy encompasses lipid degradation to free FAs (lipophagy) that fuels β-oxidation . It is thus possible that CHOP sustains growth of tumor cells through maintaining signals for optimal autophagy levels, including lipid recycling. In addition, studies in CHOP−/− mice have shown that it directly regulates the expression of master regulators of lipid metabolism, such as sterol regulatory element-binding proteins (SREBPs) and the peroxisome proliferator-activated receptor alpha (PPARα) . This favors a role of CHOP in sustaining cancer cell growth through the control of metabolic genes. The mechanism of how CHOP expression was reduced after 17β-HSD12 downregulation remains to be discovered. In liver, FXR/RXR heterodimers regulate CHOP expression; therefore, we tested the potential involvement of RXRα, which is abundantly expressed in MDA-MB-231 and SUM159 cells. 17β-HSD12 knockdown diminished RXRα mRNA and protein levels, and RXRα downregulation led to decreased CHOP levels, as also observed after 17β-HSD12 downregulation. Although a direct link cannot be drawn from these experiments, our results suggest RXRα as a candidate regulator of CHOP expression.
Examination of folding-control proteins revealed activation of ERp44 as a common denominator of 17β-HSD12 silencing in MDA-MB-231 and SUM159 cells. This increase of ERp44 did not seem to be controlled by eIF2α; silencing eIF2α led to a small increase in ERp44 expression, which could be due to the decreased 17β-HSD12 levels upon eIF2α silencing. These data further support that eIF2α is upstream of 17β-HSD12 and controls its expression. ERp44 is an interesting member of the PDI family, since it shunts between the ER, ER–Golgi intermediate compartment (ERGIC), and cis-Golgi [76, 77]. It is dormant in the ER but active in more distal compartments where it binds to its client proteins, releasing them only when their maturation is complete. Among the proteins falling into ERp44 quality control are multimeric proteins, such as immunoglobin M (IgM) and adiponectin [78, 79, 80]. Increased ERp44 expression may be a consequence of high demand in folding of specific proteins or triggered by other stimuli, such as redox status imbalance provoked by 17β-HSD12 silencing. ERp44 knockdown induced apoptosis in HeLa cells, while ERp44 expression inhibited migration in the human non-small lung cancer cell line A549 through a pathway involving inositol 1,4,5-trisphosphate receptor type 2 (IP3R2) [81, 82]. Given that the ERp44 increase after 17β-HSD12 downregulation was followed by decreased and increased proliferation/migration in SUM159 and MDA-MB-231 cells, respectively, it is likely that ERp44 upregulation is an early event following impaired LCFA synthesis. Its specific involvement in tumor cell homeostasis warrants further investigation.
High 17β-HSD12 expression has been associated with poor prognosis in breast and ovarian cancer patients [21, 22, 23]. At the same time, 17β-HSD12 was found to be a prognostic marker gene for favorable and adverse outcome in patients with kidney and liver cancer, respectively . Our study explains these apparently contradicting results, since we showed that the phenotype after 17β-HSD12 silencing could diverge, depending on the cellular context. The complex biosynthetic pathway of LCFA synthesis, the pleiotropic effects of the produced FAs, as well as the differential responses of different types of cells to impaired biochemical processes, could all underlie the divergent phenotype upon downregulation of 17β-HSD12 expression. The notion that genes can be categorized as cancer promoting or suppressing is slowly shifting, after the realization that genes can either promote or suppress the growth of cells depending on the cancer type, underscoring the heterogeneous nature of tumor cells. To our knowledge, this is the first study reporting the cellular changes following impaired LCFA synthesis due to 17β-HSD12 silencing and sets a basis for further understanding of this aspect of lipid metabolism for malignant transformation of cells.
This work was supported by the Swiss National Science Foundation to A.O. Grant 31003A-179400. We thank Dr. Michael Witting, Helmholtz-Zentrum Munich, for sharing his expertise and knowledge in lipid analysis methods.
MT and AO planned the study. MT designed and performed experiments, analyzed data, and wrote the manuscript. PS, AD, HN, and NM performed experiments and analyzed data. AO supervised the work and wrote the manuscript.
Compliance with ethical standards
Conflict of interest
The authors declare that they have no conflict of interest.
- 1.Wakil SJ (1989) Fatty acid synthase, a proficient multifunctional enzyme. Biochemistry 28(11):4523–4530Google Scholar
- 3.Medes G, Thomas A, Weinhouse S (1953) Metabolism of neoplastic tissue. IV. A study of lipid synthesis in neoplastic tissue slices in vitro. Cancer Res 13(1):27–29Google Scholar
- 7.Kuhajda FP, Jenner K, Wood FD, Hennigar RA, Jacobs LB, Dick JD, Pasternack GR (1994) Fatty acid synthesis: a potential selective target for antineoplastic therapy. Proc Natl Acad Sci USA 91(14):6379–6383Google Scholar
- 15.Abel S, De Kock M, van Schalkwyk DJ, Swanevelder S, Kew MC, Gelderblom WC (2009) Altered lipid profile, oxidative status and hepatitis B virus interactions in human hepatocellular carcinoma. Prostaglandins Leukot Essent Fatty Acids 81(5–6):391–399. https://doi.org/10.1016/j.plefa.2009.08.003 Google Scholar
- 17.Wakai K, Tamakoshi K, Date C, Fukui M, Suzuki S, Lin Y, Niwa Y, Nishio K, Yatsuya H, Kondo T, Tokudome S, Yamamoto A, Toyoshima H, Tamakoshi A, JACC Study Group (2005) Dietary intakes of fat and fatty acids and risk of breast cancer: a prospective study in Japan. Cancer Sci 96(9):590–599. https://doi.org/10.1111/j.1349-7006.2005.00084.x Google Scholar
- 21.Nagasaki S, Suzuki T, Miki Y, Akahira J, Kitada K, Ishida T, Handa H, Ohuchi N, Sasano H (2009) 17Beta-hydroxysteroid dehydrogenase type 12 in human breast carcinoma: a prognostic factor via potential regulation of fatty acid synthesis. Cancer Res 69(4):1392–1399. https://doi.org/10.1158/0008-5472.CAN-08-0821 Google Scholar
- 22.Szajnik M, Szczepanski MJ, Elishaev E, Visus C, Lenzner D, Zabel M, Glura M, DeLeo AB, Whiteside TL (2012) 17beta Hydroxysteroid dehydrogenase type 12 (HSD17B12) is a marker of poor prognosis in ovarian carcinoma. Gynecol Oncol 127(3):587–594. https://doi.org/10.1016/j.ygyno.2012.08.010 Google Scholar
- 25.Uhlen M, Zhang C, Lee S, Sjostedt E, Fagerberg L, Bidkhori G, Benfeitas R, Arif M, Liu Z, Edfors F, Sanli K, von Feilitzen K, Oksvold P, Lundberg E, Hober S, Nilsson P, Mattsson J, Schwenk JM, Brunnstrom H, Glimelius B, Sjoblom T, Edqvist PH, Djureinovic D, Micke P, Lindskog C, Mardinoglu A, Ponten F (2017) A pathology atlas of the human cancer transcriptome. Science. https://doi.org/10.1126/science.aan2507 Google Scholar
- 26.Tsachaki M, Mladenovic N, Stambergova H, Birk J, Odermatt A (2018) Hexose-6-phosphate dehydrogenase controls cancer cell proliferation and migration through pleiotropic effects on the unfolded-protein response, calcium homeostasis, and redox balance. FASEB J 32(5):2690–2705. https://doi.org/10.1096/fj.201700870RR Google Scholar
- 27.Engeli RT, Rhouma BB, Sager CP, Tsachaki M, Birk J, Fakhfakh F, Keskes L, Belguith N, Odermatt A (2016) Biochemical analyses and molecular modeling explain the functional loss of 17beta-hydroxysteroid dehydrogenase 3 mutant G133R in three Tunisian patients with 46, XY Disorders of Sex Development. J Steroid Biochem Mol Biol 155(Pt A):147–154. https://doi.org/10.1016/j.jsbmb.2015.10.023 Google Scholar
- 28.Baker-LePain JC, Sarzotti M, Fields TA, Li CY, Nicchitta CV (2002) GRP94 (gp96) and GRP94 N-terminal geldanamycin binding domain elicit tissue nonrestricted tumor suppression. J Exp Med 196(11):1447–1459Google Scholar
- 29.Flanagan L, Van Weelden K, Ammerman C, Ethier SP, Welsh J (1999) SUM-159PT cells: a novel estrogen independent human breast cancer model system. Breast Cancer Res Treat 58(3):193–204Google Scholar
- 31.Van den Branden C, Roels F (1985) Thioridazine: a selective inhibitor of peroxisomal beta-oxidation in vivo. FEBS Lett 187(2):331–333Google Scholar
- 48.Verlengia R, Gorjao R, Kanunfre CC, Bordin S, de Lima TM, Martins EF, Newsholme P, Curi R (2004) Effects of EPA and DHA on proliferation, cytokine production, and gene expression in Raji cells. Lipids 39(9):857–864Google Scholar
- 50.Hii CS, Moghadammi N, Dunbar A, Ferrante A (2001) Activation of the phosphatidylinositol 3-kinase-Akt/protein kinase B signaling pathway in arachidonic acid-stimulated human myeloid and endothelial cells: involvement of the ErbB receptor family. J Biol Chem 276(29):27246–27255. https://doi.org/10.1074/jbc.M103250200 Google Scholar
- 57.Wang X, He S, Gu Y, Wang Q, Chu X, Jin M, Xu L, Wu Q, Zhou Q, Wang B, Zhang Y, Wang H, Zheng L (2019) Fatty acid receptor GPR120 promotes breast cancer chemoresistance by upregulating ABC transporters expression and fatty acid synthesis. EBioMedicine. https://doi.org/10.1016/j.ebiom.2018.12.037 Google Scholar
- 60.Brookes PS (2005) Mitochondrial H(+) leak and ROS generation: an odd couple. Free Radic Biol Med 38(1):12–23. https://doi.org/10.1016/j.freeradbiomed.2004.10.016 Google Scholar
- 61.Gutknecht J (1988) Proton conductance caused by long-chain fatty acids in phospholipid bilayer membranes. J Membr Biol 106(1):83–93Google Scholar
- 62.Schonfeld P, Schild L, Kunz W (1989) Long-chain fatty acids act as protonophoric uncouplers of oxidative phosphorylation in rat liver mitochondria. Biochim Biophys Acta 977(3):266–272Google Scholar
- 66.Lodish HF (1999) Molecular cell biology. W.H. Freeman, New York; Macmillan, BasingstokeGoogle Scholar
- 67.Hou NS, Gutschmidt A, Choi DY, Pather K, Shi X, Watts JL, Hoppe T, Taubert S (2014) Activation of the endoplasmic reticulum unfolded protein response by lipid disequilibrium without disturbed proteostasis in vivo. Proc Natl Acad Sci USA 111(22):E2271–E2280. https://doi.org/10.1073/pnas.1318262111 Google Scholar
- 71.B’Chir W, Chaveroux C, Carraro V, Averous J, Maurin AC, Jousse C, Muranishi Y, Parry L, Fafournoux P, Bruhat A (2014) Dual role for CHOP in the crosstalk between autophagy and apoptosis to determine cell fate in response to amino acid deprivation. Cell Signal 26(7):1385–1391. https://doi.org/10.1016/j.cellsig.2014.03.009 Google Scholar
- 76.Anelli T, Sannino S, Sitia R (2015) Proteostasis and “redoxtasis” in the secretory pathway: tales of tails from ERp44 and immunoglobulins. Free Radic Biol Med 83:323–330. https://doi.org/10.1016/j.freeradbiomed.2015.02.020 Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.