Lysine-specific demethylase 2 suppresses lipid influx and metabolism in hepatic cells

Abstract

Cells link environmental fluctuations, such as nutrition, to metabolic remodeling. Epigenetic factors are thought to be involved in such cellular processes, but the molecular basis remains unclear. Here we report that the lysine-specific demethylase 2 (LSD2) suppresses the flux and metabolism of lipids to maintain the energy balance in hepatic cells. Using transcriptome and chromatin immunoprecipitation-sequencing analyses, we revealed that LSD2 represses the genes involved in lipid influx and metabolism through demethylation of histone H3K4. Selective recruitment of LSD2 at lipid metabolism gene loci was mediated in part by a stress-responsive transcription factor, c-Jun. Intriguingly, LSD2 depletion increased the intracellular levels of many lipid metabolites, which was accompanied by an increased susceptibility to toxic cell damage in response to fatty acid exposure. Our data demonstrate that LSD2 maintains metabolic plasticity under fluctuating environment in hepatocytes by mediating the cross talk between the epigenome and metabolism.

FIG 1

Increased expression of lipid transport and metabolism genes in LSD2-depleted hepatocytes. (A) LSD2 mRNA expression in mouse tissues. qRT-PCR values were normalized to the expression levels of the Rplp0 gene. Values are the means ± SDs for four samples. *, P < 0.05 versus liver tissue. BAT, brown adipose tissue. (B) Efficiency of LSD2 knockdown (KD). LSD2 expression after the introduction of three different siRNAs was detected at the mRNA (qRT-PCR) and protein (by Western blotting) levels in HepG2 cells. Cells were harvested 4 days after introduction of siRNAs. *, P < 0.05 versus control (Cont) siRNA. (C) Venn diagram of the genes changed 1.5-fold or more by three different siRNAs against LSD2. A total of 1,362 probe sets were detected with a >1.5-fold change between the control and LSD2-KD cells. Of these, 906 probe sets were upregulated, while 456 were downregulated. Control siRNAs were used as controls. (D) Significant upregulation of lipid metabolism genes by LSD2-KD as revealed by gene set enrichment analysis. Nominal (NOM) P values and false discovery rates (FDRs) are indicated. A full list of the identified gene sets is shown in Fig. S2 in the supplemental material.

FIG 2

Increased expression of lipid metabolism genes by LSD2-KD in hepatocyte cell lines. (A to C) Expression changes of lipid metabolism genes after LSD2-KD were confirmed in HepG2 (A), Hep3B (B), and Huh-7 (C) cells. Genes identified by microarray analysis as LSD2 regulated were subjected to validation by qRT-PCR. Values were normalized to the expression level of the RPLP0 gene and are shown as the fold difference against control siRNA-introduced samples. Values are the means ± SDs for triplicate samples. *, P < 0.05 versus control siRNA.

FIG 3

Genome-wide distribution of LSD2-bound sites in HepG2 cells. (A) Validation of anti-LSD2 antibody. Control HepG2 cells and HepG2 cells into which LSD2 siRNA had been introduced, in combination with use of pcDNA3-Flag or pcDNA3-Flag-LSD2, were cultured for 48 h before collection. (B) LSD2 is enriched near the TSS but excluded from the TSS itself. LSD2 peaks in HepG2 cells were detected using the MACS algorithm with ChIP-seq peaks of input sample as a control. (C) Venn diagram of the overlap between genes adjacent to the robust LSD2 ChIP-seq (green circle) peaks and genes with >1.5-fold expression change in response to LSD2-KD, as assessed by microarray analysis (purple circle). A total of 1,362 probe sets, shown in Fig. 1C, were mapped to 919 genes after removal of redundant or nonannotated probes. For the detection of robust LSD2 peaks using the MACS algorithm, criteria of >25 reads and >5-fold enrichment were applied. (D) Distribution of up- or downregulated genes among the LSD2 target genes identified in panel C (fully listed in Table S2 in the supplemental material).

FIG 4

LSD2 is enriched on lipid metabolism genes and associated with the H3K4 monomethylation level. (A) Correlation of LSD2 peaks with H3K4me1 enrichment. LSD2 peaks exhibit a marked accumulation near H3K4me1 peaks, while input peaks do not. (B) LSD2 peaks do not overlap H3K27me3 peaks. (C) The accumulation of H3K4me1 by LSD2-KD occurs predominantly near the LSD2 binding sites. H3K4me1 peaks are significantly enhanced in LSD2-KD over the control (orange line), and those enhanced in the control over LSD2-KD (blue line) are shown. (D) Venn diagram of the overlap between neighboring genes with increased H3K4me1 enrichment in response to LSD2-KD and genes with >1.5-fold expression change in response to LSD2-KD as assessed by microarray analysis. (E) Distribution of the up- or downregulated genes among LSD2-dependent genes identified in panel D (fully listed in Table S3 in the supplemental material).

FIG 5

LSD2 epigenetically regulates lipid metabolism genes. (A) Epigenetic landscape of the ACOX1 gene locus. ChIP-seq results of LSD2 binding in untreated HepG2 cells and H3K4me1 enrichment in control and LSD2-KD HepG2 cells are shown. H3K4me1 peaks induced in response to LSD2-KD were detected by the MACS algorithm and are shown above the H3K4me1 histogram of LSD2-KD sample. ENCODE data for H3K4me2, H3K27ac, p300, FAIRE, and c-Jun in HepG2 cells are also shown. c-Jun peaks that were detected using MACS algorithm are shown above the histogram. The red box indicates the putative cis-regulatory region that harbors LSD2 peak, while the green box indicates the TSS. (B) ChIP-qPCR analyses of chromatin modifications of the ACOX1 gene locus. The TSS and regions a to c (as indicated in panel A) were selected for qPCR analyses. LSD2 enrichment was tested in untreated HepG2 cells, whereas H3K4me1, H3K4me2, and H3K27ac levels were examined in control and LSD2-KD cells. LSD2-KD1 was used for this experiment. All values are means ± SDs from triplicate samples. *, P < 0.05 versus control siRNA.

FIG 6

LSD2 functionally cooperate with c-Jun. (A) Correlation of LSD2 peaks with c-Jun enrichment. LSD2 peaks exhibited a marked accumulation near c-Jun peaks, while input peaks did not. (B) Convergence of c-Jun and LSD2 peaks. Note that c-Jun was enriched near LSD2 peaks, while c-Myc was not. (C) c-Jun-dependent enrichment of LSD2 at TPA-responsive element (TRE). LSD2 occupancy was tested under c-Jun-KD in HepG2 cells. siGENOME nontargeting siRNA SMARTpool 1 (Cont-S; Dharmacon) was used as a control. (D) Upregulated expression of lipid metabolism genes under LSD2-KD was cancelled by the loss of c-Jun. siRNAs targeting LSD2 and c-Jun genes were simultaneously introduced into HepG2 cells. Different siRNAs were prepared as controls against LSD2-KD (Cont) and c-Jun-KD (Cont-S). All values are means ± SDs from triplicate samples. *, P < 0.05 versus control siRNA.

FIG 7

Accumulation of fatty acids and toxic lipid metabolites in LSD2-KD cells. (A) Heat map showing relative abundances of 275 metabolites detected by CE-TOFMS and LC-TOFMS in HepG2 cells under LSD2-KD. Both samples and metabolites were subjected to hierarchical clustering. Clusters 1 and 3 were enriched with metabolites that were increased by LSD2-KD2, while cluster 2 was enriched with the decreased metabolites. The metabolites included in these clusters are listed in Fig. S7 in the supplemental material. Data from triplicate samples of control and LSD2-KD2 were used. (B and C) Metabolomic data of representative metabolites showing increased levels in LSD2-KD. Fatty acids (B) and lipid intermediates (C) are shown. The numbers shown with the fatty acids indicate the number of carbons and the unsaturated bonds. All values are means ± SDs of triplicate results of LC-TOFMS analyses. *, P < 0.05 versus control according to Student's t test. (D) Increase of fatty acid uptake in response to LSD2-KD. (Top) Fluorescence microscopic images of BODIPY-dodecanoic acid (DA) incorporated into HepG2 cells. Cells were exposed to BODIPY-DA for 30 min before analysis. The scale bars represent 50 μm. (Bottom) Cells were treated with BODIPY-DA for 10 min, followed by flow cytometric analysis. The histogram shows the representative results for triplicate samples. Values are means ± SDs from triplicate samples. (E) LSD2-KD evokes lipotoxicity under fatty acid exposure. HepG2 cells after the introduction of siRNA were treated with bovine serum albumin (BSA) or with 50 μM oleic acid (OA) for 24 or 48 h, and then the cells were counted. The final molar ratio of OA to BSA was 6:1. All values are means ± SDs from three samples. (F) OA at 200 μM alone inhibits cell growth. All values are means ± SDs from four samples. *, P < 0.05 versus BSA according to Student's t test.

FIG 8

LSD2 is reduced under MCD-induced lipotoxic liver damage. (A) Hematoxylin and eosin staining of liver sections from 10-week-old male C57BL/6J mice fed a ND, HFD, or MCD diet for 4 weeks. The scale bar represents 50 μM. (B) Serum ALT level in mice. Values are means ± SDs from six samples. (C) LSD2 mRNA expression in the livers of HFD- or MCD-fed mice. qRT-PCR values were normalized to the expression levels of the Rplp0 gene and are shown as the fold difference from ND-fed mice. Values are the means ± SDs from six samples. *, P < 0.05 versus ND or HFD. (D) LSD2 protein level in the livers of HFD- or MCD diet-fed mice. The Western blot bands were quantitated using ImageQuant TL software (GE Healthcare). Signal intensities of LSD2 bands were normalized to those of histone H3. Values are means ± SDs from four samples. (E) Correlation plot of hepatic LSD2 mRNA expression versus serum ALT level. Pearson product-moment correlation coefficient and P values are indicated.

FIG 9

Proposed model for homeostatic control of lipid metabolism by LSD2. In the normal state, LSD2 represses fatty acid transport and metabolism genes by erasing H3K4me1 near the TRE occupied by c-Jun, contributing to maintenance of the intracellular lipid level. When LSD2 is lost, the proper expression of the lipid metabolism genes becomes compromised, leading to excessive influx and metabolism of lipids that result in the impaired cell growth. acyl-CoA, acyl coenzyme A.