MicroRNA Dysregulation in Liver and Pancreas of CMP-Neu5Ac Hydroxylase Null Mice Disrupts Insulin/PI3K-AKT Signaling

CMP-Neu5Ac hydroxylase (Cmah)-null mice fed with a high-fat diet develop fasting hyperglycemia, glucose intolerance, and pancreatic β-cell dysfunction and ultimately develop characteristics of type 2 diabetes. The precise metabolic role of the Cmah gene remains poorly understood. This study was designed to investigate the molecular mechanisms through which microRNAs (miRNAs) regulate type 2 diabetes. Expression profiles of miRNAs in Cmah-null mouse livers were compared to those of control mouse livers. Liver miFinder miRNA PCR arrays (n = 6) showed that eight miRNA genes were differentially expressed between the two groups. Compared with controls, seven miRNAs were upregulated and one miRNA was downregulated in Cmah-null mice. Specifically, miR-155-5p, miR-425-5p, miR-15a-5p, miR-503-5p, miR-16-5p, miR-29a-3p, and miR-29b-3p were significantly upregulated in the liver and pancreas of Cmah-null mice. These target miRNAs are closely associated with dysregulation of insulin/PI3K-AKT signaling, suggesting that the Cmah-null mice could be a useful model for studying diabetes.


Introduction
Sialic acids are terminal components of glycoconjugate carbohydrate chains and are involved in cell-cell recognition, ligand-receptor, and cell-pathogen interactions [1][2][3][4]. The most common mammalian sialic acids are Nacetylneuraminic acid (Neu5Ac) and N-glycolylneuraminic acid (Neu5Gc). Neu5Gc is generated by hydroxylation of CMP-Neu5Ac to CMP-Neu5Gc, catalyzed by CMP-Neu5Ac hydroxylase (Cmah). An inactivating mutation in the Cmah gene during evolution led to humans losing the common mammalian cell surface molecule N-glycolylneuraminic acid (Neu5Gc) [3,5,6]. The mutated Cmah gene initiated a series of genetic and biochemical changes in the biology of sialic acids that may contribute to several unique aspects of human biology in health and disease [7,8].
The effects of Cmah deactivation have been observed in Cmah-null mice. These mice have accumulation of Neu5Ac, which is also observed in humans. Similarly Cmah-null mice display human-like phenotypes including diminished acoustic sensitivity [9], altered startle response threshold (relative to normal mice) [9], hearing loss in old age [9], skin healing delay in adulthood [9], and a human-like muscular dystrophy phenotype following combined mutation of the Dmd gene [10]. Immunologically, these mice have enhanced B cell proliferation and antibody production, consistent with the observation that Neu5Gc is physiologically downregulated during normal B cell activation [11]. A recent study showed that high-fat diet (HFD)-fed Cmah-null mice exhibit fasting hyperglycemia, glucose intolerance, and pancreaticcell dysfunction that ultimately lead to the development of a type 2 diabetes-like syndrome [12]. Therefore, the HFDfed Cmah-null mouse is considered one of the most suitable animal models of the human metabolic syndrome.
MicroRNAs (miRNAs) are small noncoding RNAs with lengths of approximately 22 nucleotides. miRNAs have essential roles in the regulation of diverse biological pathways and play an important role in insulin resistance in various tissues [13,14]. For example, in obesity and diabetes, it has been demonstrated that miRNAs such as miR-375, miR-29, miR-320, miR-103, mir-107, and miR-126 play a crucial role in the regulation of glucose and lipid metabolism. This occurs through control of pancreatic islet cell function, adipocyte insulin resistance, hepatocyte insulin signaling, and glucose homeostasis [15][16][17][18][19]. miR-33a/b inhibits expression of insulin receptor substrate-2 (IRS-2) in hepatic cells, leading to reduced activation of downstream insulin signaling pathways including AKT and ERK. Antagonism of endogenous miR-33a/b upregulates fatty acid oxidation and the response to insulin in hepatocytes, suggesting this is a potential therapeutic target in the metabolic syndrome [20]. However, the precise molecular mechanisms by which miRNAs are involved in insulin resistance in Cmah-null mice remain obscure. The first aim of the present study was to identify changes in miRNA expression in the liver of Cmah-null mice. The second aim was to analyze miRNA expression profiles and to investigate the involvement of the insulin/PI3K-AKT signaling pathway.

In Vivo Experiments.
All lines were maintained on a congenic C57Bl/6J background. The mice were allowed to eat and drink ad libitum and were fed standard mouse chow (Cargill Agri Purina, Inc., Seongnam-Si, Korea). Twelve-week-old wild type and Cmah-null male mouse in this study were used. All animal experiments were approved and performed under the guidelines of the Konkuk University Animal Care and Experimentation Community (IACUC approval number: KU12045). Cmah <tm1Ykoz> knockout mice were originally made and kindly provided by RIKEN.

MicroRNA
Isolation and cDNA Generation. miRNA was isolated from mouse tissue using the miRNeasy mini kit (Qiagen, Valencia, CA, USA) according to the manufacturer's instruction. Concentration and purity of isolated miRNA were measured using the NanoDrop2000 spectrometer and samples were screened for an A260 : A280 ratio greater than 1.8. Conversion of miRNA to cDNA was carried out using 100 ng of miRNA according to the manufacturer's instructions for the RT 2 miRNA First Strand Kit (SA Biosciences, Frederick, MD).

MicroRNA PCR Array Analysis.
Changes in miRNA expression between samples were measured using a Liver miFinder miRNA PCR Array (SA Biosciences, Frederick, MD). Plates contained 84 of the most abundantly expressed and best-characterized miRNAs. The cDNA from individual samples was used as a template for the PCR array using SYBR green, according to the array instructions in an ABI ViiA 7 system (Applied Biosystems, Foster City, CA, USA). Data were analyzed using SABiosciences RT 2 Profiler PCR Data Analysis software (available at http://pcrdataanalysis.sabiosciences.com/mirna/arrayanalysis.php) and were considered significant at >2.0 fold change. Relative quantitation of each miRNA was determined by normalizing to the small nuclear RNA housekeeping panel (SNORD61, SNORD68, SNORD72, SNORD95, SNORD96A, and RNU6-2) comparing the control wild type (WT) and Cmah-null samples using the 2-ΔΔCt method.

Western Blot Analysis.
Approximately 40 mg of tissue was homogenized using a Tissue-lyser (Qiagen) in ice cold RIPA buffer (with protease inhibitor cocktail) and centrifuged at 13,000 g for 15 min at 4 ∘ C and the pellet discarded. Protein concentration was measured using BCA Protein Assay Kit (Thermo Scientific, Seoul, Korea) and BSA was used to generate a standard curve. Proteins were electrophoresed on 10% SDS-PAGE gels under nonreducing conditions and were then transferred to PVDF membranes. Blots were incubated with primary antibodies for IRS2 (1 : 1000, Cell signaling, MA, USA), PI3K (1 : 1000, Cell signaling, MA, USA), AKT (1 : 1000, Cell signaling, MA, USA), and p-AKT (1 : 1000, Cell signaling, MA, USA), followed by incubation with a secondary goat anti-rabbit IgG antibody conjugated to horseradish peroxidase (HRP) (1 : 5000, Calbiochem, CA, USA). Bands were visualized using an ECL detection system (Amersham Pharmacia Biotech). An antiactin antibody was used to verify equal protein loading. Band intensity was quantified using ImageJ v1.32.

Langerhans islets
Cmah-null mouse pancreas cells (b) Figure 1: Representative pancreas tissues from control and Cmah-null mice. Islet size and number are significantly reduced in Cmah-null mouse compared to that of control mouse. Insert panels demonstrate that -cell numbers in pancreas islets of Cmah-null mouse (b) were significantly lower than those of control (a).

Bioinformatics Analysis for Putative MicroRNA Targets
and Pathway. miRNA target prediction and pathway analysis was performed using DIANA-mirPath. This allowed identification of molecular pathways potentially altered by the expression of single or multiple miRNAs (http://diana.imis.athena-innovation.gr/DianaTools/index .php?r=mirpath/index) [21]. The software performs enrichment analysis of multiple miRNA target genes comparing each set of miRNA to all available pathways provided by the Kyoto Encyclopedia of Genes and Genomes (KEGG). Priority scores (enrichment value) are assigned based upon the predicted strength of the miRNA interactions with components of the target pathway.

Immunohistochemistry (IHC).
For IHC, pancreatic tissues fixed in neutral buffer with 10% formalin were used. The tissues were embedded in paraffin and were sectioned at 3-5 m. Sections used for the quantification of histopathology were stained with Hematoxylin QS to provide backgrounds. All photomicrographs were acquired using a Zeiss Axiophot microscope (Carl Zeiss, Oberkochen, Germany) equipped with an Olympus DP70 high-resolution digital microscope camera (Olympus, Center Valley, PA). The Sniper and Hematoxylin QS were all purchased from Vector Laboratories (Burlingame, CA).

Altered miRNA Expression in the Liver and Pancreas of
Cmah-Null Mice. To minimize differences in age, genetic background, sex, and time of sacrifice, which could affect metabolism, all animals with same age, sex, and genetic background were sacrificed at Zeitgeber time 9 (3 hrs before lights off). When Cmah homozygous mice with same age and sex from Cmah +/− × Cmah +/− crosses were examined and weighted, homozygous Cmah-null mice were indistinguishable from their wild type or heterozygous littermates in all respects. However, in both size and numbers of pancreatic islets in Cmah-null mice were significantly decreased ( Figure 1). Also numbers of -cells in pancreatic islets were significantly lower than those of control (insert panels of Figure 1). Consistent with a previous report [9,11], however, germline inactivation of the Cmah gene has only a minor effect on development such as hearing loss in old age, skin healing delay in adulthood, and abnormal B-cell numbers. Taken together, these observations suggest that Cmah-null mice are closely associated with dysfunction of pancreatic islets or B-cells.

Identification of Putative Target Pathways Using In Silico Analysis.
To identify the biological pathways overrepresented among the predicted targets of differentially expressed miRNAs in Cmah-null derived livers and pancreas, we examined pathway enrichment analysis by using DIANA-mirPath. The result showed putative pathways to be involved in a broad range of biological processes associated with the target genes (Table 1). Further, an investigation was carried out in statistically significant signaling pathways regulated by selected miRNAs. Among them, we found two major signal pathways such as insulin signaling (miR-155-5p, miR-425-5p, and miR-15a-5p) and PI3K-AKT signaling (miR-503-5p, miR-16-5p, miR-29a-3p, and miR-29b-3p) pathways ( Table 2). The percentage of genes involved in the insulin/PI3K-AKT signaling pathway was higher than for other pathways. The biological processes in which these genes (Pik3r1, Prkar2a, Sos1, Pik3r3, Fasn, Exoc7, Mapk3, Flot2, Raf1, and mTOR) are involved are shown in Table 2.

Cmah-Null Mice Have Defects in the Insulin/PI3K-AKT Signaling Pathway.
To determine whether the loss of Cmah was responsible for differential expression of miRNAs, RT-qPCR and western blot analysis were performed using the liver and pancreas of Cmah-null and control mice. Cmahnull livers had significant downregulation of IRS2, AKT, and Foxo1 mRNA expression, whilst Foxo1 mRNA expression was significantly upregulated in the pancreas (Figure 3(a)).
PI3Ks are composed of a catalytic subunit (p110) and a regulatory subunit. Various isoforms of the catalytic subunit (p110 , p110 , p110 , and p110 ) have been isolated, and the regulatory subunits that associate with p110 , p110 , and p110 are p85 and p85 . Among them, we used the p110 isoform of PI3K, because it plays a fundamental role in insulin signaling and control of hepatic glucose and lipid metabolism [22]. In addition, Akt is activated by phospholipid binding and activation loop phosphorylation at Thr308 by PDK1 and by phosphorylation within the carboxy terminus at Ser473. Therefore, we used the p-AKT (Phospho-Akt [Thr308]) antibody in this study. RT-qPCR and western blotting analysis confirmed that mRNA and protein levels of IRS2, PI3K, and p-AKT were reduced in livers from Cmah-null mice (Figures 3(a) and  3(b)). This suggests that dysregulation of miRNA may affect both mRNA stability and protein translation. To confirm this possibility, we analyzed mRNA expression of genes involved in insulin/PI3K-AKT signaling by RT-qPCR. Expression of three randomly selected genes was significantly lower in Cmah-null mice than in controls (Figure 4(a)). As shown    Relative mRNA expression (fold changes) Relative mRNA expression (fold changes) * * * * * * * * * * * * * * * in Figure 4(b), we proposed an interaction of miRNAs with key components of the insulin/PI3K-AKT signaling pathway. This putative insulin/PI3K-AKT signaling pathway suggested that functional loss of Cmah gene during human evolution is closely associated with Type II diabetes due to altered miRNAs and target genes interaction.

Expression Validation of Target Genes by Using Three Different miRNAs.
To determine whether the miRNAs dysregulate expression of their predicted target genes, we transfected 3 different types of miRNA mimics (hsa-miR-15a-5p mimic, hsa-miR-29a-3p mimic, and hsa-miR-29b-3p mimic) and 1 positive (hsa-miR-1 mimic) or negative control siRNA (AllStar negative control siRNA) into HepG2 cells. At 48 hrs after transfection, each cell was harvested and was provided for RT-qPCR analysis. In this study, the hsa-miR-1 mimic and AllStar siRNA were used as positive control and negative control for miRNA mimic experiments, respectively. A positive control, has-miR-1 mimic, markedly decreased the expression of HDAC4 in HepG2, compared to negative control group, whereas a negative control do not altered expression of HDAC4 gene (Figure 5(a)). We next determined the effect of miR-15a on mRNA expression of Fasn, Col1a2, Col4a2, and Ccnd2 genes. Transfection of HepG2 cells with miR15a mimic significantly decreased the mRNA level of Col1a2, Col4a2, and Ccnd2, whereas Fasn mRNA expression was not changed ( Figure 5(b)). Furthermore, HepG2 cells transfected with miR-29a and miR-29b mimic significantly downregulated its target mRNA expression such as Col3a1 and Col1a1 genes, compared to those of mock group (Figures 5(c) and 5(d)). Taken together, our observation strongly demonstrated that miRNAs dysregulation in Cmah-null mouse-derived liver could impair insulin/PI3K-AKT expression signaling during transition from glucose tolerance to intolerance.

Discussion
This study observed that expression of the miRNAs miR-155-5p, miR-425-5p, miR-15a-5p, miR-503-5p, miR-16-5p, miR-29a-3p, and miR-29b-3p in the liver of Cmah-null mice may downregulate components of the insulin/PI3K-AKT signaling pathway in concert with other genes. miRNAs play a key role in regulating many physiological and pathological processes in humans, including diabetes and the related metabolic syndrome. It has been well demonstrated that miRNAs such as miR-375, miR-29, miR-320, miR-103, miR-107, and miR-126, play a crucial role in regulating glucose and lipid metabolism through control of pancreatic islet cell function, adipocyte insulin resistance, hepatocyte insulin signaling, and glucose homeostasis [15][16][17][18][19]. In this study, differential miRNA expression was found in Cmahnull mice liver using Liver miFinder miRNA PCR array. The majority of miRNAs differentially affected included genes encoding components of the insulin/PI3K-AKT signaling pathway. As the liver consumes more oxygen at rest than other organs and is thought to play a role in the systemic control of glucose and lipid utilization [23], gene expression and metabolic measurements in the liver were studied. These metabolic pathways are partly regulated at the transcriptional level, perturbation of which leads to pathological changes. In livers of Cmah-null mice, a rapid increase in miRNA expression was observed. Consistent with these results, insulin/PI3K-AKT signaling-related gene expression was significantly decreased. These results suggest that the key molecules altered in the pathway in Cmah-null mice are associated with a specific group of miRNAs.
Heterogeneous genetic background and lifestyle make it difficult to study the mechanism(s) of metabolic syndrome in humans. Presently, mouse models are a popular tool used to study human metabolic disorder as they provide evidence that can be extrapolated to validate the human condition. Because the phenotype of HFD-fed Cmah-null mice is similar to that of patients with type 2 diabetes, the Cmah-null mouse is one of the most suitable animal models of metabolic syndrome. It is not known, however, how impaired glucose tolerance and -cell dysfunction arise in HFD-fed Cmah-null mice. To fill the gap of the phenotype of mice in relation to the previous observations [12], differential miRNA expression was examined in livers from control and Cmah-null mice using a pathway-focused glucose metabolism PCR array (data not shown). In Kyoto Encyclopedia of Genes and Genomes (KEGG), the molecular pathway analysis identified miRNAs in the upregulated or downregulated genes. Each case was linked to at least two different cellular pathways even if crosslinking was observed at low levels. Network-1 shows insulinrelated signaling pathway (Table 2 and Figure 4(b) and includes 17 genes: Rheb, Socs1, Pik3r1, Eif4e2, Prkar2a, Prkx, Crkl, Sos1, Bad, Acaca, Pik3r3, Fasn, Exoc7, Mapk3, Flot2, Raf1, and Mtor. Network-2 functions in PI3K-AKT signaling pathway (Table 2 and Figure 4(b) and includes 19 genes :  Pik3r1, Prkar2a, Sos1, Pik3r3, Fasn, Exoc7, Mapk3, Flot2,  Raf1, Mtor, Col3a1, Col1a1, Col1a2, Col5a3, Bcl2l11, COl4a2, Cdkn1b, Ccnd2, and Bcl2. These data suggest that simultaneous impairment of insulin processes required for normal liver function could contribute to the development of metabolic disorders such as hyperglycemia and diabetes in Cmah-null mice.
Pirola et al. [24] suggested that liver cells mimic the effects of hyperinsulinemia, leading to downregulation of both PI3K/AKT signaling pathways and glucose uptake via a decrease in IRS-1/2 docking molecules. Because PI3K is the key molecule controlling the decrease in IRS-1/2 as downregulation of IRS-1/2 is effectively prevented by PI3K inhibition [24]. The same mechanism could be involved in liver cells of Cmah-null mouse. Definitely, for IRS-1/2, a PI3Kdependent pathway, with PI3K itself acting as an IRS-1 serine/threonine kinase, appears to be involved, although IRS-2 degradation is controlled by a PI3K-mTOR-dependent mechanism. Moreover, tissue-specific mechanisms may control IRS degradation, as IRS-1 degradation in adipocytes appears to be mTOR-dependent [24,25]. The degradation of IRS proteins is promoted by serine/threonine phosphorylation and provides a molecular link to insulin resistance [26,27].
In this study, mRNA and protein levels of IRS2 in the livers of Cmah-null mice were significantly downregulated. As a result, IRS1 and IRS2 downstream, AKT1 and mTOR mRNA expression is also downregulated. As shown in Figure 4(b) miR-155-5p miR-15a-5p, and miR-425-5p in the case of insulin signaling and miR-29b-3p, miR-29a-3p, miR-16-5p, and miR-503-5p in the case of PI3K-AKT1-mTOR signaling were significantly dysregulated. This suggests that an insulindependent pathway in Cmah-null mice causes the increased glucose production. Previous studies showed that mice bearing a human-specific Cmah genetic mutation developed an obesity-related metabolism and diabetes [12]. It was shown that sialylation with Neu5Gc plays an important role in insulin production by cells and that its loss contributes to -cell dysfunction in mice. Taken together, these studies highlight the potential of the Cmah-null mouse to enhance scientific understanding of human metabolic syndrome.