Abstract

   Aberrant DNA methylation is associated with diabetes, but the precise
   regulatory events that control the levels and activity of DNA
   methyltransferases (DNMTs) is not well understood. Here we show that
   miR-539-5p targets Dnmt3b and regulates its cellular levels. miR-539-5p
   and Dnmt3b show inverse patterns of expression in skeletal muscle of
   diabetic mice. By binding to the 3′ UTR of Dnmt3b, miR-539-5p
   downregulates its levels in C2C12 cells and in human primary skeletal
   muscle cells. miR-539-5p-Dnmt3b interaction regulates Srebf1
   transcription by altering methylation at CpG islands within Srebf1 in
   C2C12 cells. Dnmt3b inhibition alone was sufficient to upregulate
   Srebf1 transcription. In vivo antagonism of miR-539-5p in normal mice
   induced hyperglycemia and hyperinsulinemia and impaired oral glucose
   tolerance. These mice had elevated Dnmt3b and decreased Srebf1 levels
   in skeletal muscle. db/db mice injected with miR-539-5p mimics showed
   improved circulatory glucose and cholesterol levels. Oral glucose
   tolerance improved together with normalization of Dnmt3b and Srebf1
   levels in skeletal muscle. Our results support a critical role of
   miR-539-5p and Dnmt3b in aberrant skeletal muscle metabolism during
   diabetes by regulating Srebf1 transcription; modulating the
   miR-539-5p-Dnmt3b axis might have therapeutic potential for addressing
   altered skeletal muscle physiology during insulin resistance and type 2
   diabetes.

   Keywords: DNMT3b, miR-539-5p, DNA methylation, Srebf1, diabetes,
   skeletal muscle

Graphical abstract

   graphic file with name fx1.jpg
   [29]Open in a new tab
     __________________________________________________________________

   Kesharwani et al. demonstrate that miR-539-5p-Dnmt3b interaction
   regulates Srebf1 in skeletal muscle during diabetes. miR-539-5p
   administration to diabetic mice normalizes the glycemic status and
   improves Srebf1 levels. Data provide crucial leads for addressing
   aberrant skeletal muscle metabolism during diabetes and offer a
   potential tool for therapeutic intervention. Keywords: DNMT3b,
   miR-539-5p, DNA methylation, Srebf1, diabetes, skeletal muscle

Introduction

   Type 2 diabetes is a highly complex metabolic disease marked by insulin
   resistance and/or impaired insulin secretion. A strong genetic
   association is implicated in the manifestation of diabetes,[30]^1 but
   most of these incompletely explain the high discordance among
   monozygotic twins and close correlations with environmental
   factors.[31]^2^,[32]^3 Therefore, non-genetic factors, primarily
   epigenetic modifications, gradually gained acceptance as critical
   contributing causal factors of the onset and progression of diabetes.
   An epigenome association study has reported alterations in DNA
   methylation patterns to be closely linked to whole-body insulin
   sensitivity.[33]^4

   DNA methyltransferases (DNMTs) catalyze the reversible covalent
   transfer of a methyl group to the C-5 position of a cytosine residue in
   the DNA, usually present in stretches of CpG islands.[34]^5 In mammals,
   DNMTs encompass a family of five members: DNMT1, DNMT2, DNMT3a, DNMT3b,
   and DNMT3L. However, only DNMT1, DNMT3a, and DNMT3b demonstrate DNA
   methyl transfer activity. Although DNMT1 is involved in the replication
   process, DNMT3a and DNMT3b are de novo DNMTs and are implicated in
   establishing new DNA methylation patterns. Such patterns undergo
   dynamic and reversible remodeling that determines transcript levels of
   various genes during different cellular processes that consequently
   govern cell fate and function.[35]^6^,[36]^7 Therefore, regulatory
   events that determine the levels and function of DNMTs are important
   for maintenance of normal cellular physiology.

   As with other genes, DNMTs are also regulated at transcriptional,
   translational, and post- translational levels. At the transcriptional
   level, DNMTs are regulated by several transcription factors, such as
   Sp1, Sp3, p53, Rb, Foxo3a, etc. The promoters of DNMT3a and DNMT3b
   contain multiple binding sites for several transcription factors; Sp1
   acts as transcriptional activator, and Sp3 acts as an activator or a
   repressor of Sp1-mediated transcription.[37]^8 In a colon carcinoma
   cell line, HCT116, p53 negatively regulates DNMT1 expression by forming
   a complex with Sp1.[38]^9 The promoters of DNMT1 and DNMT3a also harbor
   binding sites for E2F, and the Rb protein inhibits DNMT expression by
   binding with E2F and consequently promoting global
   hypomethylation.[39]^10 DNMT3b expression is negatively regulated by
   Foxo3a by interacting with specific binding sites on the DNMT3b
   promoter.[40]^11

   Several reports suggest that DNMTs are also regulated by
   post-translational modifications. Cellular levels and activities of
   DNMT1 and DNMT3a have been shown to be influenced by post-translational
   modifications like phosphorylation, methylation, and SUMOylation.
   Although SUMOylation of DNMT3a modifies its ability to interact with
   other proteins ,[41]^12 lysine methylation of DNMT1 affects its
   stability and signals for its proteasomal degradation.[42]^13^,[43]^14
   The cellular activity and genome methylation patterns are determined by
   serine phosphorylation of DNMTs.[44]^14^,[45]^15 It is believed that an
   interplay between the various regulatory post-translational
   modifications of DNMTs determines their stability and activity. The RNA
   binding protein HuR has been shown to post-transcriptionally regulate
   DNMT3b by binding to its 3′ UTR and increasing its levels in colorectal
   cancer cells, affecting DNMT3b-specific target DNA methylation.[46]^16

   In addition, microRNAs (miRNAs) have, in the recent past, been
   identified as major regulators of DNMTs. Particularly members of the
   miR-29 family have been reported to directly target DNMT3a and DNMT3b
   and affect global DNA methylation.[47]^17^,[48]^18 miR-29b targets and
   downregulates Sp1, a major transcriptional regulator of DNMT1, and
   indirectly regulates its levels.[49]^18 miR-148 targets DNMT3b[50]^19
   and DNMT1[51]^20 and promotes DNA hypomethylation of
   methylation-sensitive genes. DNMT1 has been reported to be targeted by
   miR-126 and miR-152 by specific binding at the 3′ UTR, resulting in
   silencing of crucial methylation-sensitive tumor suppressor
   genes.[52]^20^,[53]^21

   Despite these regulatory features of DNMT modulation, not much is known
   about such events in skeletal muscle during diabetes. This is
   significant because it is known that DNMT levels and activity are
   increased in skeletal muscle and subsequently modify the methylome, as
   seen in this tissue during diabetes.[54]22, [55]23, [56]24, [57]25

   In this study, we present data to demonstrate the regulation of DNMT3b
   by miR-539-5p and the resulting consequences of this event for impaired
   skeletal muscle physiology during diabetes.

Results

DNMT and miRNA expression is altered in skeletal muscle of db/db mice

   Several studies have shown altered patterns of DNA methylation in
   skeletal muscle during diabetes.[58]22, [59]23, [60]24, [61]25 In this
   study, we sought to evaluate whether miRNAs might regulate the
   expression of DNMTs and consequently regulate DNA methylation patterns.
   In a previous study from our laboratory, we had shown that, compared
   with normal db/+ mice, there exists an altered miRNA signature in
   skeletal muscle of diabetic db/db mice.[62]^26 To interrogate whether
   these altered miRNAs might target DNMTs in skeletal muscle, we used
   online tools that predict miRNA-target-interacting sites (TargetScan,
   miRanda, and miRDB). Although Dnmt3b was predicted to be targeted by
   three downregulated miRNAs, miR-539-5p, miR-381-3p, and miR-31-5p,
   Dnmt3a was a potential target of miR-374, miR-455, and miR-883a-3p.
   None of the altered miRNAs were predicted to target Dnmt1. We therefore
   began this study by evaluating the status of Dnmt3a and Dnmt3b in
   skeletal muscle of diabetic mice. Although transcript and protein
   levels of Dnmt3b were significantly elevated in skeletal muscle of
   diabetic mice, those of DNMT3a were unchanged ([63]Figures 1A–1D).
   Concomitantly, compared with normal mice, DNMT activity was
   significantly increased in skeletal muscle of diabetic mice
   ([64]Figure 1E). Although Dnmt1 was not predicted to be targeted by any
   of the altered miRNAs as stated above, we desired to assess the levels
   of DNMT1 in skeletal muscle of normal and diabetic mice. As shown in
   [65]Figure 1F, compared with normal db/+ mice, DNMT1 levels were
   significantly decreased in skeletal muscle of diabetic mice. Because
   DNMT activity is increased ([66]Figure 1E) in skeletal muscles of
   diabetic db/db mice, we think this is the result of increased DNMT3b
   levels. Because DNMT3b levels were increased and also predicted to be
   targeted by the downregulated miRNAs, we probed the status of the
   miRNAs that were predicted to target Dnmt3b and therefore, might be
   responsible for the observed changes in DNMT3b levels ([67]Figures 1B
   and 1D). TargetScan, miRDB, and miRanda revealed Dnmt3b to be targeted
   by the downregulated miRNAs miR-539-5p, miR-381-3p, and miR-31-5p. The
   expression of these three miRNAs was validated in skeletal muscle of
   db/db mice. As in the microarray profiling in a previous study,[68]^26
   compared with normal mice, the levels of all the three miRNAs were
   significantly downregulated in the skeletal muscle during diabetes
   ([69]Figures 1G–1I). Such inverse patterns of expression between Dnmt3b
   and these three miRNAs indicate that Dnmt3b levels might be regulated
   by these miRNAs during diabetes.

Figure 1.

   [70]Figure 1
   [71]Open in a new tab

   Expression of DNMTs and miRNAs is altered in skeletal muscle of db/db
   mice

   (A and B) Total RNA was isolated from the skeletal muscles of normal
   (db/+) and diabetic (db/db) mice, and 1 μg RNA was reverse transcribed
   and subjected to qRT-PCR to assess the transcript levels of Dnmt3a
   (A) and Dnmt3b (B). 18S rRNA was used as the loading control. (C and D)
   Skeletal muscle of normal (db/+) and diabetic (db/db) mice was lysed as
   described under “[72]Materials and methods;” 40-μg lysates were run on
   SDS-PAGE, and the levels of DNMT3a (C) and DNMT3b (D) were evaluated by
   western blot analysis. HSC70 was used as a loading control.
   Densitometric analyses of the expression are shown below. (E) Skeletal
   muscle of normal (db/+) and diabetic (db/db) mice was lysed as
   described under “[73]Materials and methods,” and 20-μg lysates were
   used to measure DNMT activity. (F) DNMT1 protein levels were assessed
   in skeletal muscle of db/+ and db/db mice by western blot analyses,
   where 40 μg protein was resolved on SDS-PAGE and probed with a DNMT1
   antibody. HSC70 was used as the loading control. Densitometric analyses
   of the expression are shown below. (G–I) Total RNA was isolated from
   skeletal muscle of normal (db/+) and diabetic (db/db) mice, and 1 μg
   RNA was reverse transcribed and subjected to qRT-PCR for expression of
   miR-539-5p (G), miR-381-3p (H), and miR-31-5p (I). All experiments were
   performed in at least four animals in each group, and values are
   reported as means ± SEM. ∗p < 0.05, ∗∗p < 0.01.

miR-539-5p targets Dnmt3b by binding to its 3′ UTR

   To validate the above, C2C12 cells were transfected with varied doses
   of miRNA mimics (1–50 nM) corresponding to miR-539-5p, miR-381-3p, and
   miR-31-5p, and the levels of DNMT3b were evaluated by western blot
   analysis. There was a significant dose-dependent decrease in expression
   of DNMT3b in the presence of miR-539-5p ([74]Figure 2A). However, there
   was no significant change in expression of DNMT3b with overexpression
   of miR-381-3p or miR-31-5p in C2C12 cells ([75]Figure 2A). The
   miR-539-5p mimic-mediated (50 nM) decrease in DNMT3b levels was
   significantly prevented in the presence of the miR-539-5p inhibitor
   ([76]Figures 2B and 2C) at the transcript and protein level, suggesting
   a specific effect of miR-539-5p on DNMT3b. Overexpression of miR-539-5p
   mimics at a dose of 50 nM caused an almost 500-fold increase over its
   endogenous levels (∼60,000 copies/cell) (data not shown). To validate
   direct binding of miR-539-5p on Dnmt3b, we generated luciferase
   reporter constructs of the wild-type 3′ UTR (wild type [WT]) of Dnmt3b
   harboring the miR-539-5p binding site ([77]Figure 2D) and also a
   mutated construct (MT) with a swap of 4 nt within the seed region of
   miR-539-5p binding site ([78]Figure 2D). C2C12 cells were
   co-transfected with the WT or the MT 3′ UTR reporter plasmid along with
   the miR-539-5p mimic with or without the miR-539-5p inhibitor. There
   was a significant decrease in the luciferase activity of the WT
   construct in the presence of miR-539-5p, and this was significantly
   prevented in the presence of miR-539-5p inhibitor and with the
   construct harboring a mutation in the miR-539-5p binding site
   ([79]Figure 2E), suggesting that miR-539-5p binds to the 3′ UTR of
   Dnmt3b. This was confirmed using a biotin-labeled miR-539-5p mimic.
   C2C12 cells were transfected with biotin-labeled scramble or
   biotin-labeled miR-539-5p, and after 48 h, cells transfected with
   biotin-labeled miR-539-5p mimics demonstrated binding of the miRNA
   mimic to the Dnmt3b 3′ UTR, as evident by significant enrichment of the
   Dnmt3b transcript in the streptavidin pull-down complex
   ([80]Figure 2F). All of this suggests that miR-539-5p targets Dnmt3b
   and inhibits its expression by binding to its 3′ UTR.

Figure 2.

   [81]Figure 2
   [82]Open in a new tab

   miR-539-5p inhibits Dnmt3b levels by binding to its 3′ UTR

   (A) Differentiated C2C12 cells were transfected with the scramble (Scr)
   or mimics of miR-539-5p, miR-381-3p, or miR-31-5p (1–50 nM). Upon
   termination of incubation at 48 h, cells were lysed, and 40 μg protein
   was subjected to western blot analysis using a DNMT3b antibody. HSC70
   was used as a loading control. Densitometric analyses of the blots are
   shown below. (B) Differentiated C2C12 cells were transfected with the
   Scr or the miR-539-5p mimic with or without its inhibitor. After 48 h,
   transcript levels of Dnmt3b were quantified by qRT-PCR. 18S rRNA was
   taken as the loading control. (C) Cells incubated as in (B) were lysed,
   and lysates (40 μg) were assessed for DNMT3b protein levels by western
   blot analysis using a DNMT3b antibody. HSC70 was used as a loading
   control. Densitometric analysis of the blots is shown below. (D)
   Depiction of the miR-539-5p binding site in the 3′ UTR of Dnmt3b and
   the mutations (red) incorporated in the miRNA binding site of the
   Dnmt3b 3′ UTR (MT) as described under “[83]Materials and methods.” (E)
   C2C12 cells were transfected with wild-type (WT) or mutated (MT) 3′ UTR
   luciferase constructs of Dnmt3b together with the Scr or the miR-539-5p
   mimic with or without its inhibitor. After 48 h of incubation, cells
   were harvested, and luciferase activity was measured as described under
   “[84]Materials and methods.”. Renilla luciferase activity was
   normalized to firefly luciferase activity. (F) C2C12 cells were
   transfected with biotin-labeled Scr or biotin-labeled miR-539-5p mimic
   (50 nM), and after 48 h, cells were harvested, lysed, and pulled down
   using streptavidin-linked Dynabeads. Enrichment of Dnmt3b mRNA in
   biotin-labeled Scr- or miR-539-5p mimic-transfected cells was
   quantified by real-time PCR using Dnmt3b-specific primers. (G) Primary
   human skeletal muscle cells were transfected with the Scr or the
   miR-539-5p mimic with or without its inhibitor. Upon termination of
   incubation (48 h), DNMT3b expression was assessed by western blot
   analysis using a DNMT3b antibody. HSC70 was used as a loading control.
   Densitometric analysis of the same is given alongside the blot. All
   experiments were performed in at least 3 sets for each group. (H and I)
   1 μg RNA from skeletal muscle of chow diet- and HFD-fed mice (n = 5)
   was reverse transcribed and subjected to qRT-PCR to evaluate the
   expression of miR-539-5p (H) and Dnmt3b (I). Sno234 and 18S rRNA,
   respectively, were used as normalization controls. Values are means ±
   SEM. ∗p < 0.05, ∗∗p < 0.01.

   Such a miR-539-5p and Dnmt3b interaction was validated in primary human
   skeletal muscle cells, where miR-539-5p significantly inhibited
   expression of DNMT3b, and this inhibition was prevented in the presence
   of a miR-539-5p inhibitor ([85]Figure 2G). The inverse pattern of
   expression between miR-539-5p and Dnmt3b in db/db mice ([86]Figure 1)
   was validated in a high-fat diet (HFD) model of obesity and diabetes.
   Compared with chow-diet fed mice, although expression of miR-539-5p was
   significantly downregulated ([87]Figure 2H), Dnmt3b transcript levels
   were significantly upregulated ([88]Figure 2I) in skeletal muscle of
   HFD-fed mice. This suggests that miR-539-5p levels are downregulated in
   skeletal muscle during obesity and diabetes and that, by targeting
   Dnmt3b, this miRNA might be responsible for the deregulated skeletal
   muscle physiology in these states.

miR-539-5p targets Dnmt3b and regulates Srebf1 levels in C2C12 cells

   Because DNMT activity and DNMT3b levels are upregulated in skeletal
   muscle during diabetes, and this would likely hypermethylate and
   promote gene silencing, to assess the cellular relevance, we looked for
   consequent effects on a set of downregulated genes in skeletal muscle
   of db/db mice taken from a recent study from our laboratory.[89]^27
   Pathway enrichment analysis (using DAVID Pathway Analysis 6.8) of the
   downregulated genes revealed pathways of AMPK and insulin signaling to
   be significantly overrepresented.[90]^27 Both of these pathways are
   critical in skeletal muscle metabolism, and three common genes, Prkab2,
   Srebf1, and Irs1 ([91]Figure 3A), from these pathways are significantly
   downregulated in skeletal muscles of db/db mice, as revealed by RNA
   sequencing.[92]^27 The expression of these genes in skeletal muscles of
   db/+ and db/db mice was validated by qRT-PCR, and, as in the RNA
   sequencing data,[93]^27 these genes were significantly downregulated in
   db/db mice skeletal muscle ([94]Figure 3B). We hypothesized that this
   downregulation might be mediated by increased DNMT3b levels, and to
   investigate this, we examined the presence of CpG islands within these
   genes. All three genes, Prkab2, Srebf1, and Irs1, harbored varied
   lengths of CpG islands across their gene lengths ([95]Figure 3C),
   indicating a possibility of them being regulated by DNA methylation.

Figure 3.

   [96]Figure 3
   [97]Open in a new tab

   Srebf1, Irs1, and Prkab2 harbor CpG islands, and their transcript
   levels are downregulated in skeletal muscle of db/db mice

   (A) AMPK and insulin signaling pathways are enriched among the
   downregulated genes in db/db mouse skeletal muscle, with Prkab2,
   Srebf1, and Irs1 commonly mapped to both pathways. (B) Skeletal muscle
   RNA (1 μg) from db/+ and db/db mice was reverse transcribed and
   subjected to qRT-PCR for transcript expression of Prkab2, Srebf1, and
   Irs1 using gene-specific primers. 18S rRNA was used as the loading
   control. Data are from at least four animals in each group, and values
   are means ± SEM. ∗∗p < 0.01, ∗∗∗p < 0.001. (C) Presence of CpG islands
   (enclosed within red boxes) across the Prkab2, Srebf1, and Irs1 genes
   as identified using the UCSC Genome Browser.

   For validation, C2C12 cells were transfected with the scramble or
   miR-539-5p mimic with or without its inhibitor for 48 h. There was no
   change in the transcript levels of Prkab2 or Irs1 in the presence of
   miR-539-5p. Interestingly, miR-539-5p caused a significant increase in
   the transcript levels of Srebf1, which was prevented in the presence of
   its inhibitor ([98]Figure 4A), suggesting that, because Srebf1 harbors
   CpG islands, presumably by inhibiting DNMT3b levels, miR-539-5p
   decreases methylation of Srebf1 and increases its levels. To validate
   this, C2C12 cells were transfected with the scramble or the miR-539-5p
   mimic, and the methylation status of Srebf1 was evaluated by methylated
   DNA immunoprecipitation (MeDIP). Using the UCSC genome browser, two
   CpG-rich regions were captured within Srebf1: site 1 spanning across
   the promoter and exon 1 and site 2 within intron 1 ([99]Figure 4B).
   Compared with scramble-transfected cells, there was a significant
   decrease in methylation at site 1, but there was no significant change
   at site 2 ([100]Figure 4C), suggesting a role of this miRNA in
   regulating Srebf1 methylation by targeting Dnmt3b and, consequently,
   regulating its levels. This methylation status was also validated
   in vivo in normal (db/+) and diabetic (db/db) mouse skeletal muscle,
   and, as shown in [101]Figure 4D, there was significant hypermethylation
   of Srebf1 in db/db mouse skeletal muscle compared with normal db/+
   mice. CpG site 2 also demonstrated significantly increased
   methylation in db/db mouse skeletal muscle; this is possibly mediated
   by factors other than miR-539-5p in vivo. To investigate whether Dnmt3b
   inhibition alone might affect Srebf1 levels, C2C12 cells were
   transfected with Dnmt3b small interfering RNA (siRNA), and this caused
   significant inhibition of DNMT3b levels ([102]Figures 4E and 4F)
   associated with notable increases in Srebf1 transcript levels at 100 nM
   (Figure 4G). This was confirmed using nanaomycin A, a selective DNMT3b
   inhibitor. As shown in [103]Figures 5A and 5B, compared with control
   cells, although nanaomycin A (NA) significantly inhibited DNMT3b levels
   at 10 μM, there was no significant change in the transcript or protein
   levels of DNMT1 and DNMT3a at any of the doses used. This change in
   DNMT3b levels was not evident at 1 or 5 μM. In Hep3B cells, similar
   patterns of NA-induced decreases in DNMT3b protein levels at higher
   doses have been shown by Lai et al.[104]^28 Interestingly, however,
   there was a modest but significant decrease in DNMT activity in the
   presence of 5 μM NA, although this inhibitory effect was more
   pronounced at 10 μM concentration ([105]Figure 5C). Such NA-induced
   inhibition of DNMT activity at 5 μM without a change at the transcript
   or protein level has been shown by Kuck et al.[106]^29 This was
   consequently accompanied by an increase in the transcript levels of
   Srebf1 at 10 μM but not at 5 μM ([107]Figure 5D), even though there was
   a modest decrease in DNMT activity at 5 μM. This may possibly be due to
   the fact that the modest decrease in DNMT activity at 5 μM NA is not
   sufficient to exert effects on Srebf1 transcription. These results
   suggest that Dnmt3b is a crucial regulator of Srebf1 and that, by
   targeting Dnmt3b, miR-539-5p regulates cellular Srebf1 levels. To
   confirm that the miR-539-5p-Dnmt3b axis is important in determining
   cellular Srebf1 levels, endogenous miR-539-5p levels in C2C12 cells
   were inhibited using the miR-539-5p inhibitor, and levels of its
   target, Dnmt3b and Srebf1 were evaluated. As shown in
   [108]Figures 5E–G, inhibition of endogenous miR-539-5p led to a
   significant increase in DNMT3b levels, and this was accompanied by a
   decrease in the transcript levels of Srebf1. To validate whether such
   decreases in Srebf1 are due to miR-539-5p inhibitor-mediated increases
   in Dnmt3b, we inhibited Dnmt3b levels in the presence of the miR-539-5p
   inhibitor. Dnmt3b levels were inhibited using specific siRNAs, and, as
   shown in [109]Figures 5H and 5I, inhibiting Dnmt3b levels
   significantly rescued miR-539-5p inhibitor-mediated decrease in
   Srebf1 transcript levels. To explore whether Dnmt3b overexpression can
   rescue miR-539-5p effects on Srebf1 transcription, C2C12 cells were
   transfected with the empty vector or the Dnmt3b overexpression vector
   (0.5–1 μg). Transfection with the Dnmt3b overexpression vector led to
   an approximately 300- and 700-fold dose-dependent increase in
   endogenous Dnmt3b levels at 0.5- and 1-μg doses, respectively.
   Subsequently, transfection of C2C12 cells with the miR-539-5p mimic
   together with the Dnmt3b overexpression vector (1 μg) significantly
   rescued the miR-539-5p-mediated increase in Srebf1 transcript levels
   ([110]Figure 5J). This indicates that the effects of miR-539-5p on
   Srebf1 are mediated through DNMT3b.

Figure 4.

   [111]Figure 4
   [112]Open in a new tab

   miR-539-5p regulates Srebf1 transcript levels by inhibiting Dnmt3b
   expression and altering CpG methylation on the Srebf1 gene in C2C12
   cells

   (A) Differentiated C2C12 cells were transfected with the Scr or the
   miR-539-5p mimic with or without its inhibitor. After 48-h incubation,
   cells were harvested, and expression of transcript levels of Prkab2,
   Srebf1, and Irs1 (as described under “[113]Materials and methods”) was
   evaluated using qRT-PCR. 18S rRNA was used as the loading control. (B)
   Schematic of CpG islands across the Srebf1 gene in mice. Each CpG site
   was randomly split into three regions, and primers were designed to
   amplify these regions within each CpG island using primers sets (P1,
   P2, and P3) as shown. (C) Differentiated C2C12 cells were transfected
   as in (A) with Scr or the miRNA mimic, and upon termination of
   incubation, genomic DNA was isolated, sheared, and immunoprecipitated
   (2 μg) with a 5-methylcytosine antibody (2 μg) or non-immune rabbit IgG
   (2 μg) and incubated overnight at 4°C. Methylation-enriched CpG regions
   within the Srebf1 gene were quantified by real-time PCR using CpG
   region-specific primers. (D) Genomic DNA isolated from skeletal muscle
   of normal db/+ and diabetic db/db mice was sheared, and the methylation
   status within Srebf1 was quantified by real-time PCR as described in
   (C). Differentiated C2C12 cells were transfected with the Scr or Dnmt3b
   siRNA (50–100 nM), and after 48 h, cells were harvested, and DNMT3b
   expression was quantified at the transcript (E) and protein (F) levels.
   18S rRNA and HSC70 were used as normalization controls, respectively.
   (G) Differentiated C2C12 cells transfected with the Scr or Dnmt3b siRNA
   (50–100 nM) were harvested, and 1 μg RNA was reverse transcribed to
   evaluate the transcript levels of Srebf1 by qRT-PCR. 18S rRNA was used
   as the normalization control. Values are means±SEM of at least three
   independent experiments. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. ns,
   non-significant.

Figure 5.

   [114]Figure 5
   [115]Open in a new tab

   Dnmt3b inhibition rescues the miR-539-5p inhibitor-induced decrease in
   Srebf1 levels in C2C12 cells

   (A and B) C2C12 cells were incubated in the presence of a DNMT3b
   inhibitor (nanaomycin A [NA]) at doses of 1, 5, and 10 μM. Control
   cells were incubated in the presence of DMSO. After 48 h, DNMT1,
   DNMT3a, and DNMT3b levels were evaluated by qRT-PCR (A) and western
   blot (B). 18S rRNA and HSC70, respectively, were used as the loading
   controls. D, DMSO. (C) Cells incubated as in (A) were assayed for
   activity of DNMT as described under “[116]Materials and methods.” Data
   were normalized to the total protein content. (D) Total RNA was
   isolated from cells incubated as in (A), and the transcript levels of
   Srebf1 were quantified using specific primers. 18S rRNA was used as the
   normalization control. (E–G) C2C12 cells were transfected with the Scr
   or miR-539-5p inhibitor (25–50 nM), and after 48 h, the status of
   DNMT3b (E and F) and Srebf1 (G) were evaluated by qRT-PCR and western
   blot analysis. HSC70 and 18S rRNA were taken as the normalizing
   controls for western blot and qRT-PCR experiments, respectively. (H and
   I) C2C12 cells were transfected with the Scr or miRNA-539-5p alone
   (50 nM) or along with Dnmt3b siRNA (100 nM), and after 48 h, Dnmt3b
   (H) and Srebf1 (I) transcript levels were evaluated by qRT-PCR. 18S
   rRNA was taken as the endogenous control. (J) C2C12 cells were
   transfected with the empty vector or the Dnmt3b overexpression vector
   (1 μg) together with the Scr or miR-539-5p (50 nM); after 48 h, total
   RNA was isolated, and the transcript levels of Srebf1 were evaluated by
   qRT-PCR. 18S rRNA was used as the endogenous control. All experiments
   were performed at least three times for each group, and values are
   reported as means ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

In vivo antagonism of miR-539-5p induces hyperglycemia and impairs oral
glucose tolerance in mice

   To determine the physiological significance of the miR-539-5p-DNMT3b
   interaction, mice were injected intravenously (i.v.) with the scramble
   or the miR-539-5p antagomir at a dose of 5 mg/kg. A schematic of the
   injection schedule and experimental protocol is shown in
   [117]Figure 6A. Blood glucose was measured randomly on the next day
   after every injection, and there was significant increase in blood
   glucose levels in antagomir-injected mice compared with
   scramble-injected mice ([118]Figure 6B). After 6 and 12 h of fasting,
   there was a significant increase in fasting glucose levels in
   miR-539-5p antagomir-administered mice ([119]Figure 6C) together with
   increased circulatory insulin levels ([120]Figure 6D). An oral glucose
   tolerance test (OGTT) demonstrated significant impairment in glucose
   tolerance in antagomir-injected mice compared with scramble-injected
   mice ([121]Figure 6E). Compared with scramble-injected mice, serum
   triglycerides and cholesterol levels were also significantly elevated
   in miR-539-5p antagomir-injected mice ([122]Figures 6F and 6G). All of
   these results demonstrate that inhibition of miR-539-5p impairs glucose
   metabolism and alters serum lipid profiles in mice.

Figure 6.

   [123]Figure 6
   [124]Open in a new tab

   In vivo antagonism of miR-539-5p induces hyperglycemia and
   hyperinsulinemia in mice

   (A) Schematic of the in vivo experimental pipeline. C57BL/6 mice were
   injected i.v. with Scr or miR-539-5p antagomir at a dose of 5 mg/kg.
   Three injections were given, one every other day. An OGTT was done on
   day 6, and on day 8, mice were euthanized. (B) Blood glucose levels
   were measured randomly the following day after every injection. (C and
   D) Fasting blood glucose levels at 6 and 12 h of fasting (C) and
   circulatory insulin levels (D) were measured in Scr- and miR-539-5p
   antagomir-injected mice. (E) OGTT was performed on Scr- and miR-539-5p
   antagomir-injected mice after 12 of fasting. (F and G) Serum
   triglyceride (F) and cholesterol (G) levels were measured in sera of
   Scr- and miR-539-5p antagomir-injected mice as described under
   “[125]Materials and methods.” Data are from at least four animals in
   each group, and values are means ± SEM. ∗p < 0.05 compared with
   Scr-injected mice.

miR-539-5p inhibition in vivo alters DNMT3b and Srebf1 expression in skeletal
muscle of mice

   Because, as shown above, administration of the miR-539-5p antagomir
   induces hyperglycemia, hyperinsulinemia, and hyperlipidemia in mice, we
   sought to examine DNMT3b and Srebf1 levels in skeletal muscle of these
   mice. miR-539-5p antagomir administration to mice significantly
   increased DNMT3b levels in skeletal muscle ([126]Figures 7A and 7B) and
   decreased the transcript levels of Srebf1 ([127]Figure 7C), indicating
   that, by increasing DNMT3b levels, miR-539-5p antagonism decreases
   Srebf1 transcription in skeletal muscle tissue in vivo. However,
   compared with scramble-injected mice, neither Dnmt3b nor Srebf1
   transcript levels were altered in the livers of miR-539-5p
   antagomir-injected mice ([128]Figures 7D and 7E), indicating that the
   effects of miR-539-5p on Dnmt3b and Srebf1 are possibly tissue
   specific. This was evident because transfection of hepatic Hepa 1-6
   cells with the miR-539-5p (50 nM) mimic did not cause any significant
   change in the transcript levels of Dnmt3b or Srebf1 ([129]Figure 7F).

Figure 7.

   [130]Figure 7
   [131]Open in a new tab

   In vivo inhibition of miR-539-5p induces DNMT3b expression and inhibits
   Srebf1 expression in skeletal muscle

   (A) Total RNA (1 μg) isolated from skeletal muscle of Scr- and
   miR-539-5p antagomir-injected mice was reverse transcribed and assessed
   for the transcript levels of Dnmt3b by qRT-PCR. 18S rRNA was used as
   endogenous control. (B) Skeletal muscle tissue from Scr- and miR-539-5p
   antagomir-injected mice was homogenized in lysis buffer, and 40 μg
   protein was run for SDS-PAGE and subjected to western blot analysis for
   detection of DNMT3b using specific antibodies. HSC70 was used as a
   loading control. (C) 1 μg of RNA from skeletal muscle tissue of Scr-
   and miR-539-5p antagomir-injected mice was reverse transcribed, and
   transcript levels of Srebf1 were evaluated by qRT-PCR. 18S rRNA was
   used as a normalizing control. (D and E) Total RNA (1 μg) from the
   livers of Scr- and miR-539-5p antagomir-injected mice was reverse
   transcribed, and transcript levels of Dnmt3b (D) and Srebf1 (E) were
   evaluated by qRT-PCR. 18S rRNA was used as a normalization control. (F)
   Mouse hepatic Hepa 1-6 cells were transfected with the Scr or
   miR-539-5p mimic (50 nM). After 48 h, cells were harvested to evaluate
   the expression of Dnmt3b and Srebf1 transcripts by qRT-PCR, normalized
   to 18S rRNA. Data are from at least 4 animals in each group, and data
   in (F) are from experiments in three independent sets; values are
   means ± SEM. ∗p < 0.05 compared with Scr-injected mice.

In vivo administration of miR-539-5p normalizes hyperglycemia, improves oral
glucose tolerance, and restores miR-539-6p, Dnmt3b, and Srebf1 levels in
skeletal muscle of db/db mice

   A schematic of the in vivo protocol is shown in [132]Figure 8A. Random
   and fasting glucose levels were measured, and after the final
   injection, an OGTT was done, and skeletal muscle tissues were
   collected. As shown in [133]Figure 8B, compared with db/db mice
   injected with the scramble, miR-539-5p-administered mice showed a
   significant improvement in random blood glucose levels that were
   measured the next day after each injection. Fasting glucose levels and
   oral glucose tolerance also significantly improved in
   miR-539-5p-administered db/db mice compared with scramble-administered
   db/db mice ([134]Figures 8C and 8D). miR-539-5p significantly improved
   serum triglyceride and cholesterol levels in db/db mice, although
   cholesteryl esters remained unchanged ([135]Figures 8E and 8F).
   Compared with scramble-injected db/db mice, in skeletal muscle of
   miR-539-5p-administered db/db mice, there was an increase in the levels
   of miR-539-5p and a concomitant decrease in the levels of Dnmt3b
   ([136]Figures 8G and H). Srebf1 levels also significantly improved upon
   miR-539-5p administration to db/db mice ([137]Figure 8I).

Figure 8.

   [138]Figure 8
   [139]Open in a new tab

   miR-539-5p administration improves hyperglycemia and normalizes
   miR-539-5p, Dnmt3b, and Srebf1 levels in skeletal muscle of diabetic
   db/db mice

   (A) Schematic of the experimental design for db/+ and db/db animals.
   db/db mice were injected with the Scr to miR-539-5p mimics (i.v.
   through the tail vein) at a dose of 5 mg/kg body weight. db/+ mice were
   injected with Scr. Three injections were given, one every other day. An
   OGTT was done on day 6, and on day 8, mice were euthanized. (B and C)
   Random blood glucose levels on the following day after every injection
   (B) and fasting blood glucose levels at 6 and 12 h of fasting (C) were
   measured in db/+ and db/db mice. (D) OGTT was performed after an oral
   glucose dose (2 g/kg body weight), and blood glucose levels were
   measured at time intervals of 30, 60, 90, and 120 min after the glucose
   feed. (E and F) Serum triglyceride (E) and cholesterol (F) levels were
   measured in sera of db/+ and db/db mice. (G–I) Total RNA was isolated
   from skeletal muscle tissue of db/+ mice injected with the Scr or db/db
   mice injected with the Scr or miR-539-5p mimics, and the transcript
   levels of miR-539-5p (G), Dnmt3b (H), and Srebf1 (I) were assessed by
   qRT-PCR. Sno234 or 18S rRNA was taken as the normalization control.
   Values are means ± SEM of four animals in each group. ∗p < 0.05,
   ∗∗p < 0.01, ∗∗∗p < 0.001 compared with db/+ mice injected with Scr;
   #p < 0.05, ##p < 0.01 compared with db/db mice injected with Scr.

Discussion

   In this study, we present data to show that miR-539-5p targets DNMT3b
   and consequently regulates Srebf1 transcript levels in skeletal muscle
   during diabetes.

   Aberrant DNA methylation is linked to several diseases, including
   cancers, heart diseases, inflammation, skin diseases, etc.[140]30,
   [141]31, [142]32, [143]33 DNA methylation mediated by DNMTs is a
   covalent modification of DNA and almost exclusively occurs at cytosine
   residues in CpG islands that are frequently located within the promoter
   regions and close to transcription start sites. Such modifications
   determine the spatial and temporal patterns of transcript levels within
   a cell and signify the importance of DNMT regulation in cellular
   physiology. The activity and cellular levels of DNMTs are regulated at
   the level of activation, stabilization, and recruitment at specific
   sites and heterochromatin regions on the DNA.[144]^34 This may be
   mediated by transcription factors, piRNA complexes, other noncoding
   RNAs, and post-translational modifications.[145]^35^,[146]^36

   In the recent past, miRNAs have been identified as significant
   regulators of DNMTs. Members of the miR-29 family target DNMT3a and
   DNMT3b and affect global DNA methylation patterns.[147]^17^,[148]^18
   Other miRNAs, like miR-126, miR-152, and miR-148, have also been shown
   to directly target and alter levels of DNMTs. Embryonic stem cells from
   Dicer-deficient mice have significantly decreased levels of DNMT1,
   DNMT3a, and DNMT3b.[149]^37^,[150]^38 Our current study demonstrates
   regulation of DNMT3b by miR-539-5p in skeletal muscle during diabetes.
   DNMT3b and miR-539-5p display inverse patterns of expression in
   skeletal muscle of db/db and HFD diabetic mice. By binding to the 3′
   UTR of Dnmt3b, miR-539-5p significantly decreases the transcript and
   protein levels of DNMT3b. These data suggest a critical role of miRNAs
   in mediating post-transcriptional regulation of DNMTs at the onset and
   during progression of diverse diseased states.

   Abnormal DNMT levels trigger aberrant methylation patterns of genes
   that consequently regulate their expression levels and cell function.
   Interestingly, DNMT3b, in its dual capacity, controls diverse cellular
   events. It provides accessory functions by supporting the catalytic
   activity of other DNMTs, HDACs, etc.; mediates methylation; represses
   gene expression; and regulates essential pathways. On the other hand,
   its intrinsic catalytic function alone at least in part determines its
   methyltransferase activity, which is evident by hypomethylation of
   target gene loci in Dnmt3b^−/− and Dnmt3b^CI/CI mice (expressing
   catalytically inactive Dnmt3b) compared with WT mice.[151]^39
   Generally, hypermethylation of DNA regions close to transcription start
   sites is associated with suppression of gene expression and vice
   versa.[152]^5 It is believed that the transcriptional decrease because
   DNA hypermethylation is possibly due to decreased binding of
   transcription factors to their respective binding sites. Our study
   shows that miR-539-5p, by inhibiting DNMT3b levels, increases the
   transcript levels of Srebf1. Srebf1 harbors potential CpG islands at
   two regions, and the one spanning the transcription start site is
   hypomethylated in the presence of miR-539-5p, and this elevates Srebf1
   transcription. Although our data support the fact that such
   hypomethylation is due to a miR-539-5p-targeted decrease in DNMT3b
   levels and, consequently, of its activity, a contribution of its
   independent methyltransferase catalytic activity alone cannot be ruled
   out and needs further investigation, especially because, despite being
   dispensable, the catalytic activity of DNMT3b is critically associated
   with genomic elements that control tumorigenesis in mice by regulating
   processes involved in cellular transformation, including that of
   phosphatidylinositol 3-kinase (PI3K)-Akt, mitogen-activated kinase
   (MAPK), etc.[153]^40

   Srebf1 encodes a helix-loop-helix leucine zipper transcription factor,
   SREBP1, that binds to the sterol regulatory element primarily found in
   genes involved in cholesterol biosynthesis and lipid homeostasis. The
   encoded protein is located at the endoplasmic reticulum and, upon
   cleavage, translocates to the nucleus to initiate target gene
   transcription. The Srebf1 gene is expressed in various
   tissues,[154]^41^,[155]^42 and abnormal levels are linked to several
   diseases, like diabetes, obesity, NAFLD, Parkinson’s disease, and
   hepatocellular carcinoma.[156]43, [157]44, [158]45, [159]46

   With respect to insulin resistance and type 2 diabetes, genetic
   variability within the Srebf1 gene is suggested to play a significant
   role. Genome-wide studies in the European population depict a strong
   link of type 2 diabetes with the Srebf1 locus,[160]^47 and
   single-nucleotide polymorphisms within this gene have been associated
   with obesity, insulin resistance, and abnormal lipid
   levels.[161]^46^,[162]^48^,[163]^49 Irs2 gene transcription is
   repressed by increased hepatic levels of this protein, which impairs
   insulin sensitivity and promotes hepatic triglyceride accumulation and
   fatty liver.[164]^50 Hepatic levels of the Srebf1-encoded protein
   SREBP1c are regulated by nutrient stimuli such as glucose and
   polyunsaturated fatty acids.[165]^51 Insulin upregulates SREBP1c levels
   in skeletal muscle, which increases the levels of glycolytic and
   lipogenic genes.[166]^52 A recent study from our laboratory
   demonstrated decreased transcript levels of Srebf1 in skeletal muscle
   of diabetic mice.[167]^27 Srebf1 transcript levels have been shown to
   be downregulated in skeletal muscle of diabetic subjects,[168]^53 and,
   interestingly, these diabetic subjects had poor metabolic control and a
   negative correlation between Srebf1 transcript levels and circulatory
   HbA1c levels.

   Our results in the current study show that, by targeting Dnmt3b,
   miR-539-5p increases Srebf1 transcript levels in skeletal muscle cells.
   In vivo antagonism of miR-539-5p increases DNMT3b levels and
   downregulates the transcript levels of Srebf1. This was accompanied by
   hyperglycemia, hyperinsulinemia, hypercholesterolemia, and
   hypertriglyceridemia. Hence, decreased Srebf1 levels as seen in
   skeletal muscle during diabetes[169]^27 are due to the increased levels
   and activity of DNMT3b and the consequent hypermethylation and
   silencing of Srebf1, all triggered by decreased miR-539-5p levels. Our
   data show that, within skeletal muscles of db/db mice, there is
   significant hypermethylation at these regions compared with normal db/+
   mice. Similar hypermethylation patterns on the Srebf1 promoter have
   been reported in the liver and adipose tissue of diabetic
   subjects.[170]^54 However, in our study, compared with
   scramble-injected mice, transcript levels of Srebf1 were not altered in
   the livers of miR-539-5p antagomir-injected mice. This might be due to
   the fact that miRNAs are known to exert tissue-specific effects, and
   miR-539-5p probably does not specifically affect the Dnmt-3b-Srebf1
   axis in the liver. Our data also demonstrate that in vivo
   administration of miR-539-5p to diabetic db/db mice normalizes
   hyperglycemia and improves oral glucose tolerance together with
   normalization of Dnmt3b and Srebf1 levels in the skeletal muscle.

   This study identifies miR-539-5p as a novel regulator of Srebf1
   transcription and provides key information for modulating its levels
   for therapeutic intervention. We provide evidence showing that, in
   skeletal muscle during diabetes, miR-539-5p levels are downregulated,
   accompanied by increased Dnmt3b and decreased Srebf1 levels. As in the
   liver and adipose tissue, Srebf1-encoded SREBP1c in skeletal muscle
   mediates insulin effects on specific genes, and its regulation of
   glycolytic and lipogenic genes indicates its relevance in maintaining
   muscle insulin sensitivity.[171]^52 These events could be important in
   aberrant skeletal muscle metabolism in pathophysiological states such
   as obesity and diabetes.

Materials and methods

Animal experiments

   Ten- to twelve-week-old male normal mice (C57BLKs lep^db/+; body
   weight, 23.06 ± 0.47 g; blood glucose, 104.5 ± 2.19 mg/dL) and diabetic
   mice (C57BLKs lep^db/db; body weight, 42.04 ± 2.19 g; blood glucose,
   390.25 ± 57.85 mg/dL) (n = 4) were obtained from the animal house
   facility of the CSIR-Central Drug Research Institute (CDRI) (Lucknow,
   India) and maintained with ad libitum food and water at the animal
   house facility of the CSIR-Institute of Genomics and Integrative
   Biology (IGIB) (Delhi, India). Seven-week-old C57BL/6 male mice
   obtained from the animal house facility of the CSIR-IGIB were fed a
   chow diet (10% calories from fat) or HFD (60% calories from fat)
   (Research Diets, NJ, USA) for 6 months; there was an approximately 40%
   increase in body weight, and glucose levels increased by around 45% in
   HFD-fed animals compared with chow diet-fed mice. Mice were euthanized,
   and skeletal muscle was isolated and stored in RNAlater (Ambion, TX,
   USA) or transferred to −80°C for further experiments. Another set of
   10- to 12-week-old male C57BL/6 mice procured from the animal house
   facility of the CSIR-IGIB were randomly assigned to two groups (n = 5
   per group) and maintained at a 12:12 h light:dark cycle with ad libitum
   food and water. Animals were injected i.v. (three injections, one every
   other day) with the scramble or miR-539-5p antagomir (Dharmacon, CO,
   USA) at a dose of 5 mg/kg body weight using in vivo-jetPEI
   (Polyplus-Transfection, France) according to the manufacturer’s
   instructions. Blood glucose of mice was measured randomly the next day
   after every injection. The day after the third injection, animals were
   fasted overnight for an OGTT. On the eighth day after the first
   injection, mice were euthanized, and blood samples and skeletal muscle
   and liver tissue were collected for further experiments. For in vivo
   injections in normal db/+ mice (C57BLKs; body weight, 24.66 ± 0.76 g;
   blood glucose, 121.25 ± 11.03 mg/dL) and diabetic db/db mice (C57BLKs;
   body weight, 46.94 ± 1.52 g; blood glucose, 225.0 ± 19.32 mg/dL) (n =
   5), db/db mice were injected with the scramble or miR-539-5p mimics at
   a dose of 5 mg/kg body weight i.v. via the tail vein using
   in vivo-jetPEI (three doses, one every other day). db/+ animals were
   injected with a scramble oligonucleotide. One day after the last
   injection, animals were fasted overnight, and an OGTT was performed as
   described below. Animals were then euthanized, and blood and skeletal
   muscle tissue were collected and stored for metabolic profiling and
   qRT-PCR as described. In-vivo-injected oligonucleotides had 2′O-methyl
   modifications with phosphorothioate linkages and a cholesterol moiety
   linked at the 3′ end by a hydroxyprolinol linkage (Sigma, MO, USA). All
   animal experiments were approved by the Institute Animal Ethical
   Committee and regulated by the Committee for the Purpose of Control and
   Supervision of Experiments on Animals (CPCSEA) (New Delhi, India).

DNMT activity assay

   DNMT activity in skeletal muscle of normal and diabetic mice or
   cell lysates was measured using the colorimetric DNMT activity
   quantification kit (Abcam, UK) according to the manufacturer’s
   instructions. Briefly, 20 μg of lysate (lysed in RIPA lysis buffer
   containing protease and phosphatase inhibitor) was incubated with
   Adomet buffer (containing S-adenosyl methionine) in a universal DNMT
   substrate-coated microplate, and the methylated DNA (substrate) was
   quantified using a capture antibody (anti-5-methycytosine antibody) and
   a detection antibody. Absorbance was measured in a microplate reader
   (Tecan, Switzerland) at 450 nm, and DNMT activity was expressed as
   optical density (OD) per hour per microgram of protein.

Cell culture and treatment

   C2C12, mouse myoblast, and mouse Hepa1-6 cells were obtained from the
   National Centre for Cell Science (NCCS) (Pune, India) and maintained in
   high-glucose Dulbecco’s modified Eagle’s medium (DMEM) supplemented
   with 10% (v/v) fetal bovine serum (Gibco, Cambridge, UK) and
   antibiotic/antimycotic solution (Gibco) at 37°C in 5% CO[2]. At a
   confluence of 70%–80%, the medium of C2C12 cells was replaced with
   high-glucose DMEM supplemented with 2% (v/v) horse serum (Gibco), to
   promote differentiation of myoblasts into myotubes, which were visible
   within 4 days of incubation. Differentiated cells were transfected with
   the scramble or mimics of miR-539-5p, miR381-3p, or miR-31-5p (1–50 nM)
   (Dharmacon). Wherever mentioned, along with the miR-539-5p mimic, the
   scramble or the miR-539-5p inhibitor (Dharmacon) was co-transfected at
   a concentration of 50 nM using Lipofectamine RNAimax (Invitrogen, MA,
   USA) and OptiMEM (Gibco). After 48 h of incubation, cells were
   harvested for qRT-PCR and western blotting to evaluate target gene
   expression. For the miR-539-5p inhibitor studies, C2C12 cells
   transfected with the scramble or miR-539-5p inhibitor (25–50nM) were
   harvested after 48 h to evaluate the levels of DNMT3b and Srebf1. For
   validation in primary cells, human primary myoblast cells (PromoCell,
   Germany) were cultured in 12-well plates (CellBIND, Corning, NY, USA)
   in skeletal muscle growth medium (PromoCell) according to the
   manufacturer’s instructions. Upon attaining 80% confluence, cells were
   transfected as above and harvested for further use. For Dnmt3b siRNA
   treatment, differentiated C2C12 cells were transfected with the
   scramble or Dnmt3b siRNA (50–100 nM) (Sigma) using Lipofectamine
   RNAimax (Invitrogen) and OptiMEM (Gibco); upon termination of
   incubation at 48 h, cells were harvested for qRT-PCR and western
   blotting as described below. To confirm the effects of DNMT3b
   inhibition, differentiated C2C12 cells were incubated in the presence
   of DMSO or the DNMT3b inhibitor NA (Calbiochem, MA, USA) at a dose of
   1, 5, and 10 μM as described by Lai et al.[172]^28 After 48 h, cells
   were harvested, lysed, and processed as described below to evaluate
   Dnmt1, Dnmt3a, Dnmt3b, and Srebf1 levels. Whether DNMT3b mediates the
   effects of miR-539-5p on Srebf1 was studied by co-transfecting C2C12
   cells with the miR-539-5p inhibitor (50 nM) with the scramble or Dnmt3b
   siRNA (100 nM), and after 48 h, Dnmt3b and Srebf1 transcript levels
   were evaluated by qRT-PCR as described below. To explore the effect of
   overexpression of miR-539-5p in the liver, mouse hepatic Hepa 1-6 cells
   were transfected with the scramble or miR-539-5p mimic (50 nM). After
   48 h of incubation, cells were harvested to evaluate the expression of
   miR-539-5p, Dnmt3b, and Srebf1.

RNA isolation and qRT-PCR

   RNA from cells and tissues was isolated using TRIzol according to the
   manufacturer’s protocol and quantified using a plate reader (Infinite
   200 Pro, Tecan, Switzerland) with 260/280 ratios between 1.9 and 2.0.
   1 μg of RNA was used for cDNA synthesis using RevertAid reverse
   transcriptase (Thermo Fisher Scientific, USA) and random hexamers
   (Thermo Fisher Scientific, MA, USA). Quantitative PCR was performed to
   evaluate gene and miRNA expression in a StepOne Plus real-time PCR
   machine (Applied Biosystems, MA, USA) using SYBR Green Master Mix
   (Applied Biosystems, USA) and primers as in [173]Table S1. 18S rRNA and
   Sno234 were used as normalization controls for genes and miRNA
   expression, respectively. Relative gene expression was calculated by
   the ΔCt method.[174]^55

Western blotting

   Protein samples were isolated from cells and tissues using RIPA lysis
   buffer (Sigma) containing phosphatase (Calbiochem, Germany) and
   protease inhibitors (Calbiochem, Germany). 40 μg of protein was run on
   SDS-PAGE and subjected to western blot analysis using DNMT1 (Cell
   Signaling Technology, MA, USA; 1:2,000 dilution), DNMT3a (Cell
   Signaling Technology, MA, USA; 1:2,000 dilution), and DNMT3b (Abcam,
   UK; 1:2,000 dilution) antibodies. Immunoreactive bands were detected
   using horseradish peroxidase (HRP)-linked secondary antibodies (Genei,
   India), followed by detection with femtoLUCENT HRP-Plus reagent
   (G-Bioscience, MO, USA). HSC70 was used as the loading control for
   normalization. Densiometric analyses for western blots were performed
   using AlphaEase FC imaging analysis software (Alpha Innotech, CA, USA),
   and the intensity of each band was normalized to background intensity.

Cloning, mutagenesis, and luciferase assay

   Luciferase reporter constructs containing the 3′ UTR of Dnmt3b
   harboring the binding site for miR-539-5p were generated in a psiCheck2
   vector (Promega, WI, USA) using primers as in [175]Table S1.
   Substitution mutation was done in the binding site of miR-539-5p using
   the XL site-directed mutagenesis kit (Agilent Technologies, Canada) and
   mutation-specific primers ([176]Table S1). Differentiated C2C12 cells
   were co-transfected with the WT (100 ng) or the MT 3′ UTR reporter
   plasmid (100 ng) along with the scramble or the miR-539-5p mimic
   (50 nM) with or without the miR-539-5p inhibitor (50 nM). A dual
   luciferase assay (Promega, USA) was performed after 48 h of incubation,
   and luminescence was measured in a multimode reader (Infinite M200 Pro,
   Tecan, Switzerland). Renilla luciferase values were normalized to those
   of firefly luciferase. As described by Phatak and Donahue,[177]^56 to
   validate the binding of miR-539-5p with the 3′ UTR of Dnmt3b mRNA,
   C2C12 cells were transfected with biotin-labeled scramble or
   biotin-labeled miR-539-5p mimic (50 nM) (QIAGEN, USA). After 48 h of
   incubation, cells were harvested, lysed, and incubated overnight with
   Dynabead-linked Streptavidin C1 (Thermo Fisher Scientific, USA) at 4°C.
   Dynabeads were pulled down on a magnetic stand, and the bound RNA was
   isolated by using TRIzol. Enrichment of Dnmt3b mRNA with biotin-labeled
   miR-539-5p mimic was quantified by real-time PCR using Dnmt3b-specific
   primers and compared with that of scramble-transfected cells.

   The complete coding sequence of mouse Dnmt3b was cloned into a
   pcDNA3.1(+) vector (Thermo Fisher Scientific, USA) using specific
   primers ([178]Table S1). Differentiated C2C12 cells were transfected
   with the empty vector or the Dnmt3b overexpression vector (0.5–1 μg)
   alone or with the miR-539-5p mimic (50 nM) using Lipofectamine 2000
   (Invitrogen, USA) according to the manufacturer’s protocol. After 48 h,
   cells were lysed, total RNA was isolated, and the transcript levels of
   Srebf1 were assessed by qRT-PCR as described above.

Bioinformatics analysis

   miRNA target prediction was performed using three online public
   databases: TargetScan ([179]http://www.targetscan.org/vert_72/), mirDB
   ([180]http://mirdb.org/mirdb/index.html), and miRanda
   ([181]http://microrna.org/). The presence of CpG islands within the
   genes was analyzed using the UCSC Genome Browser
   ([182]https://genome.ucsc.edu/).

MeDIP

   Differentiated C2C12 cells were transfected with scramble (50 nM) or
   miR-539-5p mimic (50 nM) for 48 h. Upon termination of incubation,
   genomic DNA was isolated from these cells and from skeletal muscle of
   db/+ and db/db animals using phenol/chloroform/isoamyl alcohol.
   Isolated genomic DNA was randomly sheared into 200- to 700-bp fragments
   with six sets of sonication cycles (15-s pulse and 30-s incubation
   between each pulse) using an ultrasonic homogenizer (Takashi, Japan).
   Sonicated genomic DNA was quantified in a plate reader (Infinite 200
   Pro, Tecan, Switzerland); 2 μg of sonicated genomic DNA was used for
   immunoprecipitation with 5-methylcytosine antibody (2 μg) (Epigentek,
   USA) or non-immune rabbit immunoglobulin G (IgG) (Epigentek, USA) and
   incubated overnight at 4°C. Methylation-enriched CpG regions within the
   Srebf1 gene were quantified by real-time PCR using CpG region-specific
   primers (primers against each CpG island split into regions of
   150–180 bp; [183]Table S1).

OGTT

   Scramble- and miR-539-5p antagomir-injected mice or db/+ and db/db mice
   were fasted for 12 h for an OGTT. Fasting blood glucose was measured
   prior to an oral glucose dose (2 g/kg body weight) for the OGTT. Blood
   glucose levels were measured at time intervals of 30, 60, 90, and
   120 min after the oral glucose feed.

Blood and tissue collection

   Blood glucose was measured using the OneTouch Select glucometer (Life
   Scan Europe, Switzerland). For insulin, cholesterol, and triglyceride
   analysis, blood samples were collected from mice during euthanization,
   centrifuged at 1,300 × g for 10 min at 4°C to collect serum, and stored
   at −20°C for further use. Insulin levels were measured using the
   insulin ELISA kit (G-Biosciences, USA) according to the manufacturer’s
   protocol. Serum triglycerides, cholesterol, and cholesteryl ester
   levels were measured using the respective kits following the
   manufacturer’s instructions (BioVision, CA, USA). Skeletal muscle was
   isolated from all groups of mice and stored in RNAlater (Ambion, USA)
   at −80°C or directly transferred to −80°C for protein analysis.

Statistical analysis

   All experiments were performed at least three times, and data are
   represented as means ± SEM. Student’s t test and one-way ANOVA were
   performed for statistical significance; p < 0.05 was considered
   statistically significant.

Data availability

   The datasets generated and/or analyzed during the current study are
   available from the corresponding author upon reasonable request.

Acknowledgments