MicroRNAs have emerged as important regulators of gene expression in obesity and diabetes (Krutzfeldt and Stoffel 2006
; Tang et al. 2008
). However, to date there has not been a comprehensive survey of miRNA expression in multiple tissues from an animal model of obesity-induced type 2 diabetes. In this study we provide a quantitative measurement of the expression of ~220 miRNAs in islets, adipose tissue, and liver from diabetes resistant (B6) and diabetes susceptible (BTBR) mice.
To the best of our knowledge, this is the first study to survey miRNA expression in isolated pancreatic islets in the context of obesity and diabetes. In addition to islets, we report on the relative expression of miRNAs in adipose tissue and liver, identifying miRNAs that are commonly expressed across these 3 tissues, as well as those that show dramatic tissue-specific expression patterns. Further, we report on many miRNAs that are differentially regulated by obesity and/or genetic differences that exist between B6 and BTBR mice. Finally, by profiling hepatic miRNA expression in an F2-intercross, we demonstrate that miRNA abundance, like mRNA abundance, is a heritable trait, giving rise to miR-eQTLs, or genomic loci that govern their expression. The information provided by this study will aid in understanding the relationship between miRNA regulation and type 2 diabetes.
In islets, the expression of two miRNAs (132 and 212) each increased ~14 and ~4-fold in response to obesity in B6 and BTBR mice, respectively. This obesity-dependent increase in miRNA expression observed in B6 mice was the largest of any miRNA profiled in islets, adipose tissue, or liver. MiRNAs 132 and 212 are genomically located on Chr 11, ~200 base pairs apart, suggesting they may be generated from a common pri-miRNA molecule (Vo et al. 2005
), explaining why obesity regulates their expression equally. Interestingly, these two miRNAs are either not expressed or expressed at a low level in liver and adipose tissue, suggesting that obesity-dependent regulation is unique to islets. Previous work has shown that they are highly expressed in neurons and may regulate neuronal differentiation (Vo et al. 2005
; Wayman et al. 2008
). MiR-132 has also been shown to modulate the circadian clock in the superchiasmatic nucleus via the MAPK/CREB-dependent pathway (Cheng et al. 2007
). Interestingly, a circadian clock has been identified in islets that regulates the expression of several key islet genes, including Glut2
, and Stx1a
(Allaman-Pillet et al. 2004
). Further investigation would be needed to determine whether the obesity-dependent regulation of miR-132 regulates the islet circadian clock. Finally, miRNAs 132 and 212 were shown to be highly up-regulated in ovarian granulosa cells in response to the hormone hCG (Fiedler et al. 2008
). It would be interesting to determine if changes in circulating levels of hCG that is known to occur during pregnancy (Braunstein et al. 1976
), induces the expression of these miRNAs in pancreatic islets.
There are two consensus cAMP-response element binding (CREB) sites located proximal to the genomic position of miRNAs 132 and 212 (Vo et al. 2005
). Consistent with these CREB sites playing a role in their regulation, we have found that incubating islets with IBMX or forskolin, compounds that are known to elevate cellular cAMP levels (Barnett et al. 1994
), results in a ~2-fold increase in the expression of miRNAs 132 and 212 (unpublished observations). Further, the incretin hormone Glp-1, also known to elevate cellular cAMP levels (Thorens and Waeber 1993
), induces the expression of these miRNAs in mouse islets and Ins-1 cells (Shang et al., manuscript in preparation). Glp-1 stimulates glucose-stimulated insulin secretion, stabilizes insulin mRNA levels, promotes β-cell replication and suppresses apoptosis in β-cell lines and isolated islets (Drucker 2003
), all of which has resulted in Glp1 offering a new target for the treatment of diabetes. Interestingly, the receptor for Glp1 (Glp1R
) contains a coding SNP between B6 and BTBR, converting a highly-conserved histidine at position 47 to arginine. While we do not know if this amino acid substitution alters the activity of Glp1R
, our data suggests that the difference in the magnitude of the obesity-dependent induction of miRNAs 132 and 212 reflects differential Glp-1 responsiveness between B6 and BTBR mice, which may contribute to the increased susceptibility to diabetes of the BTBR mice.
Fatty acids have been shown to affect miRNA expression in the mouse insulinoma cell line, Min6 (Lovis et al. 2008
). The expression of miR-34a was increased >2-fold when Min6 cells were treated with 1 mM palmitate. Interestingly, we found that the expression of miR-34a was elevated ~3 and ~5-fold as a function of obesity in B6 and BTBR mice, respectively. The greater obesity-dependent induction of miR-34a expression in BTBR mice may in part reflect increased plasma lipids, which we have previously reported for BTBR-ob/ob
mice (Clee et al. 2005
). Lovis and colleagues (Lovis et al. 2008
) also report that miR-34a over-expression in Min6 cells lead to decreased glucose-stimulated insulin secretion targeting Vamp2
and increased apoptosis by targeting Bcl2
Poy et al. were the first to show that miR-375 is abundantly expressed in pancreatic islets (Poy et al. 2004
). Here, we show that miR-375 was the second most abundantly expressed miRNA in islets and further, was expressed ~600-fold higher in islets than in adipose tissue and was not expressed in liver. Recently, whole-body miR-375 knockout (375KO
) mice have been generated (Poy et al. 2009
). The mice have elevated plasma glucose levels due to a reduction in β-cell mass and an increase in the number of α-cells per islet. Further, when challenged with severe obesity (Lepob/ob
mice do not show enhanced β-cell proliferation, which is necessary to compensate for the peripheral insulin resistance resulting from the obesity. Our data reveals that miR-375 expression in islets is ~50% lower in BTBR-ob/ob
mice than in B6-ob/ob
mice. We recently reported that obesity does not stimulate β-cell proliferation in BTBR mice, whereas B6 mice demonstrate a ~3-fold increase in response to obesity (Keller et al. 2008
). Perhaps the loss of miR-375 expression in the islets of BTBR-ob/ob
mice is involved in their inability to increase β-cell replication in response to obesity.
Dietary and genetically-imposed obesity has previously been shown to regulate miRNA expression in mouse adipose tissue (Xie et al. 2009
). Microarray chips were used to profile adipose tissue collected from either chow or high-fat diet-fed B6 mice, as well as adipose tissue from B6-lean and B6-ob/ob
mice. The expression of several miRNAs changed in response to either obesity challenge, including miRNAs 107, 103, 30c, 30a-5p, 222, and 221. Furthermore, these same miRNAs were differentially expressed in differentiated versus non-differentiated 3T3-L1 pre-adipocytes. Our data supports an obesity-dependent change in the expression of these miRNAs (, adipose tissue panel). Moreover, we did not see a strain difference between B6 and BTBR for the influence of obesity, suggesting that these miRNAs are not involved in the pathogenesis of diabetes, but rather play a critical role in adipogenesis, in agreement with Xie et al. (Xie et al. 2009
). Sun et al. reported that the let-7 miRNA family was highly expressed in adipocytes and further, was up-regulated during 3T3-L1 adipogenesis (Sun et al. 2009
). However, while our data supports the finding that the let-7 miRNA family is highly expressed in adipose tissue (let-7 a, b, c, d were all > 8,000 copies/cell), we did not see a change in their expression as a function of obesity in either B6 or BTBR mice (Supplementary Table 1
Of the 3 tissues profiled in our study, the liver contained the greatest number of miRNAs whose expression patterns showed a strain difference in their response to obesity. The expression of ~43 miRNAs showed a small decrease (>60%) in response to obesity in B6 mice, but not in BTBR mice, suggesting that strain plays an important role in the regulation by obesity of hepatic miRNA expression. In order to reveal the specific gene loci mediating this regulation, we profiled miRNAs in the liver of a sub-set of F2-ob/ob mice generated as part of a larger ongoing study (Keller et al., manuscript in preparation).
We found that ~10% of the miRNAs profiled in the liver of the F2 mice showed heritability. Twenty-one miRNAs have a miR-eQTL across the genome (LOD > 5.3, p
-value <0.05). The most significant linkage was observed for miR-214, which mapped to a locus on Chr 17 with a LOD score of ~16. Previous work has shown that the transcription factor Twist1
regulates the expression of miR-214 (Lee et al. 2009
). MiRNA-214 is genomically located on Chr 1, therefore its linkage to Chr 17 is trans
, indicating the presence of a regulatory element. Given that Twist1
is genomically located on Chr 12, our data suggests a new regulatory element is involved in miR-214 regulation. Interestingly, of the 21 miRNAs to show significant linkage, all but 3 mapped to loci in trans
, suggesting that the expression of many miRNAs are under the control of specific regulatory elements.
The 3 miRNAs mapping in cis
were miRNAs 34a, 31 and 295. MiR-34a had the second most significant miR-eQTL, mapping to Chr 4 with a LOD score of ~13. The tumor suppressor gene, p53
has been shown to directly regulate miR-34a expression (Chang et al. 2007
). However, the p53 binding site in the promoter for miR-34a does not include any SNPs between B6 and BTBR, suggesting once again, that our data has revealed novel regulatory pathways that control hepatic miRNA expression. Finally, six out of the 21 miRNAs illustrated in , had trans
-linkage to a region on Chr 12. Using a lower LOD threshold of 3.0, 209 miRNAs show linkage across the genome, 26% of which have a miR-eQTL on Chr 12. This suggests that Chr 12 may contain a general regulatory factor involved in the processing of up to 25% of the miRNAs expressed in liver.
To the best of our knowledge, our study is the first to report the genetic architecture of hepatic miRNA expression. Not only does this reveal specific genetic loci governing the control of miRNAs, we believe that it may allow us to identify which mRNAs are targets for specific miRNAs. For each miRNA that showed significant linkage, we were able to identify hundreds of mRNA transcripts that showed linkage to the same genomic region as the miRNA. This will enable us to utilize the linkage data from the two profiling studies to identify putative mRNA targets for a given miRNA. Future work will utilize the linkage profiles for miRNAs and co-mapping mRNAs to construct causal transcriptional networks to identify direct versus indirect mRNAs for a particular miRNA.
In conclusion, our study provides a quantitative measure ~220 miRNAs in tissues (islets, adipose and liver) that play a key role in glucose homeostasis in the context of obesity-induced type 2 diabetes. We identified many miRNAs that undergo a significant change in expression in response to obesity and/or genetic differences between diabetes resistant (B6) and diabetes susceptible (BTBR) mice. We also report for the first time that hepatic miRNA expression is a heritable trait, which can be used to identify putative mRNA targets. It is estimated that the expression of up to 30% of all genes may be effected by miRNAs (Brennecke et al. 2005
; Lewis et al. 2005
). Additionally, a miRNA can potentially influence the expression of hundreds to thousands of mRNA transcripts. Elucidating the regulatory mechanisms that mediate obesity and genetic-dependent regulation of miRNA expression will aid in understanding the role(s) miRNAs play in the context of obesity-induced type 2 diabetes.