Epigenetic Histone Methylation Modulates Fibrotic Gene Expression

Abstract

TGF-β1–induced expression of extracellular matrix (ECM) genes plays a major role in the development of chronic renal diseases such as diabetic nephropathy. Although many key transcription factors are known, mechanisms involving the nuclear chromatin that modulate ECM gene expression remain unclear. Here, we examined the role of epigenetic chromatin marks such as histone H3 lysine methylation (H3Kme) in TGF-β1–induced gene expression in rat mesangial cells under normal and high-glucose (HG) conditions. TGF-β1 increased the expression of the ECM-associated genes connective tissue growth factor, collagen-α1[Ι], and plasminogen activator inhibitor-1. Increased levels of chromatin marks associated with active genes (H3K4me1, H3K4me2, and H3K4me3), and decreased levels of repressive marks (H3K9me2 and H3K9me3) at these gene promoters accompanied these changes in expression. TGF-β1 also increased expression of the H3K4 methyltransferase SET7/9 and recruitment to these promoters. SET7/9 gene silencing with siRNAs significantly attenuated TGF-β1–induced ECM gene expression. Furthermore, a TGF-β1 antibody not only blocked HG-induced ECM gene expression but also reversed HG-induced changes in promoter H3Kme levels and SET7/9 occupancy. Taken together, these results show the functional role of epigenetic chromatin histone H3Kme in TGF-β1–mediated ECM gene expression in mesangial cells under normal and HG conditions. Pharmacologic and other therapies that reverse these modifications could have potential renoprotective effects for diabetic nephropathy.

TGF-β1 has been implicated in various human disorders including vascular and renal diseases.^1–3 Diabetic nephropathy (DN) is a chronic renal complication characterized by the thickening of glomerular and tubular basement membranes and progressive accumulation of extracellular matrix (ECM) proteins such as type I and type IV collagens, fibronectin, and laminin in the tubular interstitium and mesangium.^3–5 Induction of profibrotic TGF-β1 by diverse mediators such as high glucose (HG), advanced glycation endproducts (AGEs), and angiotensin II in glomerular mesangial cells (MCs) and other renal cells has been implicated in these events.^2–9 TGF-β1 also increases ECM accumulation through induction of its downstream effector, connective tissue growth factor (CTGF),^10,11 and by decreasing matrix degradation through inhibition of proteases or activation of protease inhibitors such as plasminogen activator inhibitor-1 (PAI-1).¹² A TGF-β1–specific antibody had significant anti-fibrotic effects in animal models of DN, including db/db type 2^13,14 and streptozotocin-induced type 1 diabetic mice¹⁵ and prevented HG-induced increase in matrix protein synthesis in renal cells.^4,16 TGF-β1 can regulate gene expression through Smad transcription factors and E-box–dependent mechanisms.^1,17–20 However, the subtle nuclear chromatin mechanisms involved in TGF-β1–induced expression of key ECM genes in MCs are not clear.

Gene regulation by extracellular stimuli involves not only transcription factors binding to their cognate DNA binding sites but also epigenetic changes in chromatin without alterations in DNA sequence. Post-translational modifications on amino-terminal tails of nucleosomal histones such as histone H3 and H4, including acetylation, methylation, and ubiquitination at key lysines, play key roles in modulating chromatin structure and gene transcription.^21,22 They form a “histone code” that can dictate transcriptional outcomes of gene activation or repression.²³ In general, acetylation of histone H3 lysines (H3KAc) is associated with active gene transcription, whereas methylation (H3Kme) can be associated with either active or inactive gene promoters depending on the position of lysine modified. H3KAc is mediated by histone acetyl transferases and H3Kme by histone methyltransferases (HMTs). HMTs can mono-, di-, or tri-methylate (H3K-me1, -me2, -me3) specific lysine residues, thereby adding another epigenetic regulatory layer.²²

Histone H3K4me is usually associated with gene activation and transcriptional elongation and is mediated by HMTs such as SET1, MLL1–4, and SET7/9.^22,24–26 H3K9me, on the other hand, is generally associated with gene repression and is mediated by HMTs such as SUV39H1, G9a, and SETDB1/ESET.²⁴ Other lysines, including H3K27, H3K36, and H3K79, can also be methylated to various degrees.²⁴ In addition, the discovery of histone lysine demethylases has added another dimension to gene regulation.²⁷ Together, these factors create a fine balance of gene regulation, a disruption of which could result in abnormal gene expression and disease phenotypes.

To date, it is not known whether promoter histone H3 lysine methylation plays a role in TGF-β1–induced transcription of ECM-associated genes in MCs or whether the effects of HG on such epigenetic events can be mediated through TGF-β1. Here we show that TGF-β1 leads to the enrichment of H3K4me1/2/3 and depletion of H3K9me2/3 marks at ECM-associated gene promoters in rat MCs. A TGF-β1 antibody could reverse HG-induced changes in H3Kme at these fibrotic gene promoters along with reductions in their expression. Furthermore, the H3K4 HMT SET7/9 seemed to play a role in TGF-β1–induced ECM gene expression. These data show novel epigenetic chromatin mechanisms in TGF-β1 actions in MCs related to ECM deposition and DN.

RESULTS

ECM-Associated Genes Are Increased, Whereas, Reciprocally, Repressive H3K9me Levels Are Decreased at Their Promoters in TGF-β1–Treated Rat Mesangial Cells

We first examined whether TGF-β1–induced expression of key ECM-related genes was associated with changes in the repressive epigenetic marks H3K9me2 and H3K9me3 at their promoters. Serum-depleted rat mesangial cells (RMCs) were stimulated with TGF-β1 (10 ng/ml) for various time periods, and gene expression levels were analyzed by RT-QPCR. Collagen-α1(I) chain (Col1a1), CTGF, and PAI-1 mRNA levels were significantly increased by TGF-β1 from 2 to 24 hours compared with control, whereas the housekeeping gene CypA showed no difference under these conditions (Figure 1A). Immunoblotting showed that protein levels of collagen I, CTGF, and PAI-1 (Figure 1B) were also similarly increased by TGF-β1. These results confirmed that TGF-β1 can upregulate ECM-associated genes in RMCs.

Figure 1.

TGF-β1 increases the expression of ECM-associated genes in RMCs. (A) Serum depleted RMCs were stimulated with TGF-β1 (10 ng/ml) for various time periods (0.5 to 24 hours), and mRNA levels of ECM-associated genes (Col1a1, CTGF, and PAI-1) and housekeeping gene cyclophilin A (CypA) were analyzed by RT-QPCR. Gene expression was normalized to internal control β-actin gene, and results are expressed as fold stimulation over control (ctrl) (mean ± SEM; *P < 0.05; **P < 0.01 versus ctrl, n = 3). (B) Western blot analysis of RMC cell lysates from control and TGF-β1–treated RMCs using collagen I, CTGF, PAI-1, and β-actin antibodies. Results shown are representative of two separate experiments.

We next performed chromatin immunoprecipitation (ChIP) assays with H3K9me2- and H3K9me3-specific antibodies. ChIP-enriched DNA samples were analyzed by quantitative PCR (QPCR) using primers spanning Smad binding sites and nearby cis-elements at these promoters (Figure 2A). Levels of both H3K9me2 (Figure 2B) and H3K9me3 (Figure 2C) at the Col1a1, CTGF, and PAI-1 promoters were significantly reduced in RMCs treated with TGF-β1 from 2 to 24 hours compared with control. In contrast, there were no significant differences at the CypA promoter. These results suggest that TGF-β1–induced expression of these genes may be caused, at least in part, by a loss of repressive epigenetic histone modifications at their promoters.

Figure 2.

TGF-β1 decreases H3K9me2/3 levels at ECM-associated gene promoters in RMCs. (A) Map showing locations of Col1a1, CTGF, and PAI-1 promoter primers used for ChIP-QPCRs. TSS, transcription start site; SBE, Smad binding elements. (B and C) Bar graphs showing H3K9me2 (B) and H3K9me3 (C) levels at the indicated gene promoters in control and TGF-β1 (10 ng/ml)-stimulated RMCs. ChIP assays were performed with H3K9me2 and H3K9me3 antibodies as described in Concise Methods. Immunoprecipitated DNA and input DNA were subjected to QPCRs with primers specific for the indicated gene promoters to measure enrichment levels. QPCR data were analyzed using the 2^−ΔΔCt method, and results normalized to input DNA were expressed as fold over respective untreated control (ctrl) cells (mean ± SEM; *P < 0.05; **P < 0.01 versus ctrl, n = 3).

TGF-β1 Enhances H3K4me Levels at the Promoters of ECM-Associated Genes

We next examined whether TGF-β1 could alter promoter levels of H3K4me, an epigenetic “active” mark, using ChIP assays with H3K4me1, H3K4me2, or H3K4me3 antibodies. As shown in Figure 3A, TGF-β1 increased H3K4me1 levels at the Col1a1 and CTGF promoters in RMCs at 2 hours, and this was sustained up to 24 hours. At the PAI-1 promoter, H3K4me1 levels were significantly increased only at 24 hours. TGF-β1 enhanced H3K4me2 levels from 6 to 24 hours at Col1a1 and PAI-1 promoters, with no significant changes at the CTGF promoter (Figure 3B). TGF-β1 significantly increased H3K4me3 levels at the Col1a1, CTGF, and PAI-1 promoters (Figure 3C). These increases in promoter H3K4me1/2/3 levels correlated with the increased expression of these genes by TGF-β1. On the other hand, the CypA promoter showed no significant changes in these marks, confirming specificity. These results suggest that increases in promoter H3K4me may be involved in TGF-β1–induced upregulation of ECM-associated genes in RMCs.

Figure 3.

TGF-β1 upregulates H3K4me1/2/3 levels at ECM-associated gene promoters. H3K4me1 (A), H3K4me2 (B), and H3K4me3 (C) levels at indicated promoters in RMCs treated without (control, ctrl) or with TGF-β1 (10 ng/ml) for various time periods. ChIP assays were performed as described in Figure 2 with specific antibodies, and results normalized to input DNA were expressed as fold enrichment over respective untreated ctrl (mean ± SEM; *P < 0.05; **P < 0.01 versus ctrl, n = 3).

TGF-β1–Specific Antibody Reverses HG-Induced Inhibition of Repressive H3K9me in RMCs

Serum-depleted RMCs were pretreated with TGF-β1–specific antibody (25 μg/ml) or control mouse IgG (25 μg/ml) and then treated with either normal glucose (5 mM) plus 25 mM mannitol (NG), or HG (30 mM) for 48 hours. Gene expression was evaluated by RT-QPCR, whereas promoter enrichments of H3K9me2 or H3K9me3 were assessed by ChIP-QPCRs. As shown in Figure 4A, HG significantly increased Col1a1, CTGF, and PAI-1 mRNA levels compared with NG. These changes were significantly inhibited by TGF-β1 antibody but not control IgG. More interestingly, ChIP assays showed that HG significantly decreased H3K9me2 (Figure 4B) and H3K9me3 levels (Figure 4C) at the Col1a1, CTGF, and PAI-1 promoters compared with NG, and this inhibitory effect of HG was significantly reversed by TGF-β1 antibody but not IgG control. There were no significant differences at the CypA promoter. These results suggest that HG-induced ECM gene expression in RMCs is associated with decreased repressive marks H3K9me2 and H3K9me3, which can be reversed, at least in part, by the TGF-β1 antibody.

Figure 4.

TGF-β1–specific antibody reverses HG-induced expression of ECM-associated genes and HG-induced changes in H3K9me2/3 at their promoters in RMCs. (A) mRNA levels of ECM-associated genes in RMCs. Serum-depleted RMCs were pretreated with TGF-β1–specific antibody (25 μg/ml) or mouse IgG (25 μg/ml) for 1 hour and then treated with NG (5 mM glucose + 25 mM mannitol) or HG (30 mM glucose) for 48 hours. Gene expression was analyzed by RT-QPCR, and results are expressed as fold stimulation over NG cells (mean ± SEM; **P < 0.01 versus NG; ^##P < 0.01 versus HG, n = 3). (B) H3K9me2 and (C) H3K9me3 enrichment levels at ECM and CypA gene promoters in RMCs treated with NG or HG pretreated with IgG or TGF-β1 antibodies. ChIP assays were performed as described in Figure 2, and results are expressed as fold enrichment relative to NG (mean ± SEM; *P < 0.05 versus NG; ^#P < 0.05 versus HG, n = 3).

TGF-β1–Specific Antibody Reverses HG-Induced Increases in Promoter H3K4me in RMCs

We next tested whether HG can increase active H3K4me marks at the promoters of ECM-associated genes and whether this can be reversed by the TGF-β1 antibody. Serum-depleted RMCs were pretreated with TGF-β1 antibody or IgG for 1 hour and then treated with NG or HG for 48 hours, followed by ChIP assays with specific antibodies. Results showed that HG significantly increased H3K4me1 (Figure 5A), H3K4me2 (Figure 5B), and H3K4me3 (Figure 5C) levels at the Col1a1, CTGF, and PAI-1 promoters compared with NG. TGF-β1 antibody, but not IgG, significantly attenuated these HG-induced increases (except for H3K4me3 at the Col1a1 promoter; Figure 5, A–C). In contrast, the CypA promoter showed no significant differences among all of the groups. These findings, coupled with the effects seen on H3K9me2/3, show the mediatory role of TGF-β1 in HG-induced epigenetic events at promoters of ECM genes and their subsequent expression and that blockade of these events may be a key mechanism for the anti-fibrotic and renoprotective effects of the TGF-β1 antibody.

Figure 5.

TGF-β1–specific antibody reverses HG-induced H3K4me at ECM gene promoters in RMCs. (A–C) Bar graphs showing the H3K4me1 (A), H3K4me2 (B), and H3K4me3 (C) levels at ECM and CypA gene promoters in RMCs pretreated with TGF-β1–specific antibody (25 μg/ml) or mouse IgG (25 μg/ml) for 1 hour, followed by treatment with NG or HG for 48 hours. ChIP assays were performed as described in Figure 2, and results are expressed as fold enrichment relative to NG (mean ± SEM; *P < 0.05 versus NG; ^#P < 0.05 versus HG, n = 3).

SET7/9 Expression Is Increased by TGF-β1 in RMCs

We next examined the role of SET7/9, a H3K4 mono-methyltransferase, in ECM gene expression. We first observed that SET7/9 mRNA levels were increased by TGF-β1 (10 ng/ml) in a time-dependent fashion in RMCs (Figure 6A). SET7/9 protein levels were also increased (Figure 6, B and C). Pretreatment with actinomycin D (2 μg/ml) abolished TGF-β1 induced SET7/9 mRNA (Figure 6D), showing that TGF-β1 may regulate SET7/9 at the level of transcription.

Figure 6.

TGF-β1 induces the expression of SET7/9 mRNA and protein in RMCs. (A) SET7/9 mRNA expression in RMCs after TGF-β1 treatment for various time periods was analyzed by RT-QPCR, normalized to internal control β-actin gene, and expressed as fold stimulation over control (ctrl) (mean ± SEM; *P < 0.05; **P < 0.01 versus ctrl, n = 3). (B) Cell lysates from control (ctrl) RMCs and cells treated with TGF-β1 for indicated time periods were analyzed by immunoblotting with SET7/9 and β-actin antibodies. (C) SET7/9 protein levels were quantified by scanning densitometry, and results are expressed as fold over ctrl (mean ± SEM; *P < 0.05; **P < 0.01 versus ctrl, n = 3). (D) SET7/9 mRNA expression in RMCs pretreated with or without actinomycin D followed by stimulation with TGF-β1. Gene expression was analyzed by RT-QPCR and normalized to internal control β-actin gene, and results were expressed as fold stimulation over no actinomycin D and no TGF-β1 treatment (mean ± SEM; *P < 0.05; **P < 0.01 versus no actinomycin D and no TGF-β1; ^$P < 0.01 versus no actinomycin D and TGF-β1, n = 3).

TGF-β1 Increases SET7/9 Recruitment to ECM-Associated Gene Promoters

Next, we examined whether TGF-β1 alters SET7/9 occupancy using ChIP assays with SET7/9 antibodies. Results showed that SET7/9 recruitment was significantly increased at the Col1a1, CTGF, and PAI-1 promoters from 2 to 24 hours after TGF-β1 stimulation compared with control, with no change at the CypA promoter (Figure 7A). The SET7/9 recruitment pattern was quite similar to the increased H3K4me1 (Figure 3A), except at the PAI-1 promoter, suggesting a key role for SET7/9 in TGF-β1–mediated increase in H3K4me1 in the induction of ECM genes.

Figure 7.

TGF-β1 enhances SET7/9 recruitment at ECM-associated gene promoters, and the TGF-β1–specific antibody reverses SET7/9 occupancy under HG conditions in RMCs. (A) SET7/9 recruitment at the indicated gene promoters in RMCs stimulated with TGF-β1 (10 ng/ml) for various time periods. (B) Inhibition of HG-induced SET7/9 enrichment at ECM gene promoters by TGF-β1 Ab. Serum-depleted RMCs were pretreated with TGF-β1–specific antibody (Ab) (25 μg/ml) or mouse IgG (25 μg/ml) for 1 hour and then treated with normal glucose (5 mM) plus mannitol (25 mM) (NG) or high glucose (30 mM) (HG) for 48 hours. SET7/9 recruitment was determined by ChIP assays with SET7/9 Ab as described in Figure 2. Results were normalized to input, and SET7/9 occupancy was expressed as fold enrichment over respective control samples (mean ± SEM; *P < 0.05; **P < 0.01 versus ctrl, n = 3) (A) or over NG samples (mean ± SEM; *P < 0.05 versus NG; ^#P < 0.05 versus HG, n = 3) (B).

TGF-β1–Specific Antibody Can Block HG-Induced SET7/9 Recruitment

We next observed that HG can also enhance SET7/9 recruitment to the Col1a1, CTGF, and PAI-1 promoters compared with NG (Figure 7B). Furthermore, the TGF-β1 antibody (but not IgG control) could significantly block these HG-induced increases in SET7/9 recruitment (Figure 7B). SET7/9 occupancy at the CypA promoter showed no differences under these conditions. These results suggest a mediatory role for TGF-β1 in HG-induced SET7/9 recruitment and increased H3K4me1 at ECM gene promoters.

SET7/9 Knockdown Can Attenuate TGF-β1–Induced Expression of ECM-Associated Genes

We further studied the functional role of SET7/9 in TGF-β1–induced gene expression. RMCs were first transfected with various concentrations of siRNA oligonucleotides targeting SET7/9 (siSET7/9) or control siRNAs (siNeg). After 48 hours, total RNA from transfected cells was analyzed by RT-QPCR. As shown in Figure 8A, SET7/9 mRNA levels were significantly reduced in RMCs transfected with siSET7/9, with maximum inhibition at 300 ng and no further decrease at 450 ng. Next, RMCs were transfected with 300 ng of siSET7/9 or siNeg, and 72 hours later, lysates were analyzed by immunoblotting with SET7/9 antibody. As shown in Figure 8B, siSET7/9 also significantly reduced SET7/9 protein levels compared with siNeg.

Figure 8.

SET7/9 is involved in TGF-β1–induced regulation of ECM-associated genes in RMCs. (A) RMCs were transfected with various concentrations of SET7/9 siRNA (siSET7/9) or control (siNeg) oligonucleotidies. Forty-eight hours after transfection, SET7/9 mRNA levels were analyzed by RT-QPCR. Results were normalized to internal control β-actin gene, and SET7/9 levels were expressed as fold over siNeg 150 ng (mean ± SEM; **P < 0.01 versus siNeg 150 ng). (B) SET7/9 protein levels in RMCs transfected with 300 ng of siSET7/9 or siNeg oligonucleotidies. Total cell lysates were prepared 72 hours after transfection and immunoblotted (IB) with SET7/9 or β-actin antibodies. Bar graph below represents the quantification of SET7/9 protein levels determined by densitometry and expressed as fold over siNeg (mean ± SEM; **P < 0.01 versus siNeg, n = 3). (C) Global levels of H3K4me1, H3K4me2, and H3K4me3 in RMCs transfected with siNeg or siSET7/9 were determined by immunoblotting with indicated antibodies (mean ± SEM; **P < 0.01 versus siNeg, n = 3). (D) RMCs were transfected with 150 ng of siSET7/9 or siNeg oligonucleotidies, serum depleted for 24 hours, and stimulated with or without TGF-β1 (10 ng/ml) for 6 hours. ECM-associated genes (Col1a1, CTGF, and PAI-1) and CypA mRNA expression were analyzed by RT-QPCR and normalized to internal control β-actin gene. Results were expressed as fold over siNeg (mean ± SEM; **P < 0.01 versus ctrl siNeg; ^†P < 0.05 versus siNeg + TGF-β1, n = 3).

We next examined whether SET7/9 siRNA can affect H3K4me in RMCs because previous studies showed that SET7/9 can regulate H3K4me1 in endothelial cells and monocytes^28–30 and induce H3K4me2 in pancreatic islet cells,³¹ whereas others reported that SET7/9 may not mediate H3K4me1³² or that it can activate transcription by also methylating nonhistone protein substrates.^33–36 We transfected RMCs with siSET7/9 and examined global H3K4me levels by immunoblotting. Total H3K4me1 levels were clearly decreased in SET7/9 knockdown RMCs compared with siNeg, whereas H3K4me2 and H3K4me3 levels were not affected (Figure 8C). Thus, SET7/9 most likely mediates H3K4me1, but not H3K4me2 or H3K4me3, in RMCs. Furthermore, TGF-β1–induced Col1a1, CTGF, and PAI-1 mRNA levels were significantly attenuated by siSET7/9 compared with siNeg (Figure 8D). In contrast, CypA mRNA levels were not affected (Figure 8D). These results further support a key role for SET7/9 in modulating TGF-β1 responses in RMCs.

DISCUSSION

In this report, we first confirmed that TGF-β1 can upregulate ECM-associated genes Col1a1, CTGF, and PAI-1 in RMCs. We then examined changes in key epigenetic chromatin marks, including histone H3K9me2/3 and H3K4me1/2/3 levels, at these gene promoters. Our results showed that these marks and the H3K4 HMT and SET7/9 are involved in TGF-β1– and HG-induced up-regulation of ECM-associated genes in RMCs. Furthermore, a TGF-β1 antibody could reverse HG-induced changes in the levels of these promoter marks, and this correlated with the antibody-induced inhibition of their expression in RMCs.

Increasing evidence shows that H3K9me2 and H3K9me3 marks are recognized by heterochromatin protein 1 and generally correlate with gene silencing and transcriptional repression.^24,37,38 Our previous study showed that, in vascular smooth muscle cells (VSMCs) derived from diabetic db/db mice, H3K9me3 levels were decreased at key inflammatory genes promoters and inversely correlated with the increased expression of these genes under basal and TNF-α–treated conditions.³⁹ Furthermore, human VSMCs and endothelial cells cultured under HG conditions also exhibited decreased levels of H3K9me3,^30,39 suggesting that a loss of the repressive H3K9me3 mark can increase the expression of pathologic genes under diabetic conditions. Our current results showed for the first time that TGF-β1 decreased H3K9me2 and H3K9me3 levels on Col1a1, CTGF, and PAI-1 promoters, and this inversely correlated with increased expression of these genes, thereby further supporting the notion that a relief of transcriptional repression caused by decreases in repressive chromatin histone modifications may contribute to increased expression of fibrotic genes by TGF-β1.

We also showed that up-regulation of H3K4me1/2/3 marks usually associated with active chromatin occurs in parallel with the downregulation of H3K9me2/3 by TGF-β1, suggesting that this can further contribute to the increased gene expression. Recent studies showed that methylated histone H3K4 correlates with transcriptionally competent chromatin and is associated with active genes.^22–26 This supports our observation that TGF-β1 increased the expression of Col1a1, CTGF, and PAI-1 in RMCs, and this positively correlated with increased H3K4me1, H3K4me2, and H3K4me3 at their promoters.

Evidence showed that HG treatment caused dynamic changes in H3K4me2 and H3K9me2 in human monocytes⁴⁰ and H3K4me1 in endothelial cells,²⁹ whereas H3K9me3 levels were decreased in VSMCs from diabetic db/db mice relative to control db/+ and in HG-treated VSMCs³⁹ and endothelial cells.³⁰ This is in line with our results showing increases in H3K4me1–3 and decreases in H3K9me2–3 at the Col1a1, CTGF, and PAI-1 promoters under HG conditions in RMCs, which correlated with HG-induced up-regulation of these genes. Furthermore, the TGF-β1 antibody could reverse HG-induced epigenetic alterations.

This study also showed an increase, not only in the recruitment of H4K4 HMT SET7/9 at the Col1a1, CTGF, and PAI-1 promoters, but also in the expression of SET7/9 in RMCs stimulated by TGF-β1, suggesting that this HMT is involved in TGF-β1–induced up-regulation of ECM genes in RMCs. This was supported by our observations that SET7/9 gene silencing partially, but significantly, blocked TGF-β1–induced ECM-associated genes. Knockdown of SET7/9 with siRNAs could decrease global H3K4me1, but not H3K4me2 or H3K4me3, levels, suggesting that SET7/9-mediated H3K4me1 plays a key role, at least in part, in ECM-associated gene expression and that SET7/9 may be a potential therapeutic target for fibrotic disorders such as DN. The observed increases in H3K4me1 might synergize with other complementary events occurring at these promoters, such as increases in H3K4me2–3 and decreases in H3K9me2/3, to enhance ECM gene expression in response to TGF-β1 (Figure 9). Furthermore, the HMTs and histone demethylases (HDMs), as well as histone acetyl transferases (HATs) and histone deacetylases (HDACs) regulating these other chromatin marks, may also play cooperative roles. Additional studies are needed to assess these factors, including the role of HMTs that mediate H3K4me2/3 and H3K9me2/3.

Figure 9.

Schematic representation of histone H3 lysine methylation in TGF-β1–mediated fibrotic gene expression in mesangial cells. Diabetic conditions and TGF-β1–induced expression of ECM-associated genes Col1a1, CTGF, and PAI-1 in MC result in fibrosis in the pathogenesis of DN through increases in promoter levels of active chromatin marks (H3K4me1, H3K4me2, and H3K4me3) and promoter occupancy or expression of K4-HMTs such as SET7/9, and in parallel, through decreases in promoter levels of repressive chromatin marks H3K9me2 and H3K9me3. Under diabetic (HG) conditions, a TGF-β1 antibody can reverse these chromatin modifications to exert renoprotective effects.

It was suggested that SET7/9-mediated H3K4 methylation functions in transcriptional activation by competing with HDACs to enhance H3-K9 acetylation and prevent H3-K9 methylation.^41,42 SET7/9 can also methylate nonhistone proteins including p53,^33,43 DNMT1,⁴⁴ TAF10,³⁴ and p65.^35,36 Furthermore, SET7/9 could regulate a subset of TNF-α–induced NF-κB–dependent inflammatory genes in monocytes²⁸ and HG-induced expression of NF-κB p65 and inflammatory genes in endothelial cells.²⁹ These data show the diverse physiologic roles of SET7/9 in gene transactivation. It is possible that, in this study, SET7/9 may also act through methylating other nonhistone proteins and possibly even Smads. Further studies are needed to evaluate these aspects.

Evidence showed that TGF-β1 signaling through acetylation of Smad3 itself can up-regulate gene transcription,⁴⁵ and acetylation of histones in the TGF-βRII promoter regions can up-regulate its transcription.⁴⁶ A few studies in diabetic kidneys have shown the involvement of HDACs in TGF-β1–mediated ECM production and kidney fibrosis.^47,48 These studies suggest a role for HDACs in the pathogenesis of renal fibrosis and TGF-β1 actions by possibly silencing key protective genes. In unpublished observations, we showed that TGF-β1 can also increase levels of the active chromatin histone acetylation mark H3K9Ac at the PAI-1 promoter, and this correlated with increased PAI-1 expression in MCs, suggesting that a balance between active and repressive marks at the K9 position may control TGF-β1–induced gene expression. Further studies are needed to determine such functional interplay between these two antagonistic histone modifications. It is also well known that tubulointerstitial disease contributes substantially to DN. In the future, it would be interesting to explore analogous approaches that examine epigenetic mechanisms in matrix gene expression in tubular epithelial and/or interstitial cells exposed to TGF-β1. Evidence showed that HDAC inhibitors attenuate fibrotic changes in tubulointerstitial injury.⁴⁹ Epigenetic mechanisms may affect ECM expression by also regulating the expression of metalloproteases and other proteases, because histone acetylation has been shown to regulate metalloproteases in some cells.⁵⁰

Taken together, our studies showed that TGF-β1 and HG can promote significant changes in promoter histone H3K4 and H3K9 methylation in MCs that correlate with parallel increases in the expression of genes related to ECM accumulation and the pathogenesis of DN. Because histone lysine methylation has been associated with metabolic memory in other cells,^29,30,39 it is possible that these epigenetic changes may also contribute to sustained diabetic renal complications that persist despite glycemic control.

CONCISE METHODS

Materials

Recombinant human TGF-β1 (240-B) and pan-specific TGF-β antibody (MAB1835) were from R&D systems (Minneapolis, MN); normal mouse IgG(12–371), normal rabbit IgG (PP64B), and protein A agarose/salmon sperm DNA(16–157) were from Upstate, (Billerica, MA). The following antibodies were used in the Western blotting analyses and ChIP assays: anti-CTGF (ab6992), anti-collagen I (ab6308), anti-histone H3 dimethyl K9 (ab1220), anti-H3 trimethyl K9 (ab8898), anti-H3 monomethyl K4 (ab8895), anti-H3 dimethyl K4 (ab32356), and anti-H3 trimethyl K4 (ab8580) from Abcam (Cambridge, MA); purified mouse anti-PAI-1 (612024) from BD Biosciences (San Jose, CA); anti-β-actin (A5441) from Sigma (St. Louis, MO); and anti-SET7/9(07–314) from Upstate Biotechnology. SET7/9 ON-TARGETplus siRNA (J-059399[09–12]) was from Thermo Scientific, and Silencer Negative Control #1 siRNA (AM4611) was from Ambion. Nucleofection kits (VPI-1004) were from Lonza (Allendale, NJ). Actinomycin D (114666) was from Calbiochem (La Jolla, CA). Reverse transcriptase kits and SYBR Green PCR Master Mix kits were from Applied Biosystems (Foster City, CA), and RNA-STAT60 reagent was from Tel-Test (Friendswood, TX). Primers for β-actin internal standards were from Ambion (Austin, TX). All other primers were designed using bioinformatics software and synthesized by IDT (Coralville, IA). Sequences of primers used in this study are listed in Table 1.

Table 1.

Primer sequences

Primer	Forward Primer	Reverse Primer	Annealing Temperature
cDNA primers
rCol1α1	TGGTGCTCCTGGTATTGCTG	cggaCGTTTTCCTTCTTCTCcG	58
rCTGF	cagctCCGAGAAGGGTCAAGCtG	AACAGGCGCTCCACTCTGTG	61
rPAI-1	cggaCTTCTTCAAGCTCTTCcG	TGAAATAGAGGGCGTTCACCAG	58
rSET7/9	ACAGAAGAAGGGAAGCCACA	CGGACTCATAAGGGTCTGGA	58
rCypA	TATCTGCACTGCCAAGACTGAGTG	CTTCTTGCTGGTCTTGCCATTCC	58
rβ-actin	CTGCCCTGGCTCCTAGCAC	cggacGCAGCTCAGTAACAGTCcG	62
ChIP primers
rCol1α1pro	GGCTGGAGAAAGGTGGGTCT	CCCAGGTATGCAGGGTAGGA	58
rCTGFpro	ATCAGGAAGGGTGCGAAGAG	TCCACATTCCTCCGTCTGAA	58
rCypApro	TATCTGCACTGCCAAGACTGAGTG	CTTCTTGCTGGTCTTGCCATTCC	58
rPAI-1pro	gacaatATGTGCCCTGTGATTGtC	AGGCTGCTCTACTGGTCCTTGC	60

Cell Culture

All animal studies were performed according to protocols approved by the Institutional Animal Care and Use Committee (IACUC). Primary cultures of RMCs were obtained by explant culture of renal glomeruli isolated from Sprague-Dawley rats and cultured in RPMI 1640 medium as described.⁵¹ RMCs were serum depleted in medium containing 0.2% BSA before stimulation. Cells between 6 and 12 passages were used.

Transient Transfections

RMCs were plated in 100-mm culture dishes and transfected the next day (70% confluent) with SET7/9 ON-TARGETplus siRNA (siSET7/9) or Silencer Negative Control #1 siRNA (siNeg) using Nucleofection reagent as described.⁵² This yielded transfection efficiencies of 50 to 60%. About 6 hours after transfection, cells were washed, and fresh medium containing 0.5% FBS was added. The next day, cells were placed in serum-free RPMI 1640 medium containing 0.2% BSA for 24 hours, treated with or without TGF-β1 (10 ng/ml), and processed for RNA or protein extraction or ChIP assays at the indicated time periods.

RNA Isolation and Real-Time QPCR

Total RNA was isolated from RMCs using RNA-STAT60 reagent according to the manufacturer's instructions. Total RNA (2 μg) was used to synthesize cDNA using Moloney murine leukemia virus (MuLV) reverse transcriptase and random hexamers in a final volume of 20 μl as described by the manufacturer. QPCRs using SYBR green reagent with gene-specific primers (listed in Table 1) and β-actin gene primers (internal control) were performed in triplicate in a final volume of 20 μl in an ABI 7300 real-time PCR thermal cycler (Applied Biosystems). Dissociation curves were run to detect nonspecific amplification, and we confirmed that single products were amplified in each reaction. Relative gene expression levels were calculated after normalization with internal control β-actin gene using the 2^−ΔΔCt method, where Ct is the threshold value.³⁹ Results were expressed as fold over control.

Western Blotting

RMCs were lysed in 1.5× SDS sample buffer, fractionated on 4 to 15% SDS-PAGE gels (Bio-Rad, Hercules, CA), and immunoblotted with antibodies to PAI-1 (1:10,000), CTGF (1:5000), collagen I (1:2000), and SET7/9 (1:2000) as reported earlier.¹¹ The blots were stripped and reprobed with an antibody to β-actin (1:200,000). Immunoblots were developed using a chemiluminescence method and scanned with a GS-800 densitometer to determine the intensity of protein bands with Quantity One software (Bio-Rad).

ChIP Assays

ChIPs were performed, and ChIP-enriched DNA was analyzed by real-time QPCR as described earlier.³⁹ Briefly, cells were fixed with 1% formaldehyde at 37°C for 10 minutes, washed with cold PBS containing protease inhibitors, and lysed in Tris, pH 8.1, containing 1% SDS, 1 mM PMSF, and complete protease inhibitor cocktail. Cell lysates were sonicated to fragment chromatin to 500-bp size, diluted in ChIP dilution buffer, and immunoprecipitated overnight at 4°C with indicated specific antibodies, with IgG control, or without antibody (no antibody control). Next day, immune complexes were collected on protein-A agarose beads, and the beads were washed to remove nonspecific binding. DNA was eluted from the beads, cross-links were reversed, and DNA was extracted. ChIP-enriched DNA samples and input DNA samples were analyzed by QPCR with SYBR reagent in a real-time PCR machine (ABI 7300; Applied Biosystems) using primers specific for Col1a1, CTGF, or PAI-1 promoters spanning Smad binding elements or the control cyclophilin A promoter. All reactions were performed in triplicate in a final volume of 20 μl. Dissociation curves were run to detect nonspecific amplification, and we confirmed that single products were amplified in each reaction. QPCR data were analyzed using the 2^−ΔΔCt method as described earlier³⁹ and normalized with input samples. Results were expressed as fold over control. In all of the experiments, we verified that ChIP samples obtained with specific antibodies exhibited significant enrichment relative to IgG or no antibody controls.

Statistical Analysis

Data are expressed as mean ± SEM of multiple experiments. Paired t tests were used to compare two groups or ANOVA with Dunnet post tests for multiple groups, using PRISM software (GraphPad, San Diego, CA). Statistical significance was determined at the 0.05 level.

DISCLOSURES

None.