Male National Institutes of Health (NIH) mice (age, 4 weeks) were purchased from the Guangdong Medical Laboratory Animal Center (Foshan, China). The animals were maintained in an environmentally controlled breeding room (temperature, 20±2˚C; humidity, 60%±5%, dark/light cycle, 12 h). The mice were fed with standard laboratory chow and water ad libitum. The study was performed in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the Institutional Animal Care and Use Committee of Tsinghua University (Beijing, China). The protocol was approved by the Animal Welfare and Ethics Committee of Tsinghua University. Male NIH mice that were fasted for 24 h (weight, 18±2 g) were used to induce the diabetic model by the intrainperitoneal injection of 100 mg/kg STZ (Sigma-Aldrich, St. Louis, MO, USA). STZ was freshly prepared in 0.1 M pre-cold citric acid buffer (pH 4.5). One week after the injection of STZ, diabetic animal models with fasting blood glucose of >11.1 mmol/l were selected for further trials. Four weeks after STZ induction, the animals were sacrificed by cervical dislocation; pancreatic tissues were removed for further miRNA or RNA extractions. STZ-untreated mice served as the normal controls.

The miRNA chip assay was conducted by the method of Exiqon miRNA array [The 7th generation of miRCURY™ LNA array (v.18.0), KangChen Bio-tech Inc., Shanghai, China]. Three different whole pancreatic tissues were freshly collected from three different diabetic mice. The three diabetic or normal mice had similar blood glucose and body weights. Three equal volumes (∼0.33 ml) of different homogenates from whole pancreatic tissue of each mouse were mixed and subjected to miRNA and mRNA extracts. Total RNA was harvested using TRIzol (Invitrogen Life Technologies, Carlsbad, CA, USA) and the miRNeasy mini kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. Subsequent to having passed RNA quantity measurement using the NanoDrop 1000, the samples were labeled using the miRCURY™ Hy3™/Hy5™ Power labeling kit and hybridized on the miRCURY™ LNA array (v.18.0). Following the washing steps, the slides were scanned using the Axon GenePix 4000B microarray scanner (Molecular Devices Corp., Sunnyvale, CA, USA). Scanned images were imported into the GenePix Pro 6.0 software (Axon Instruments, Molecular Devices Corp.) for grid alignment and data extraction. Replicated miRNAs were averaged and miRNAs with intensities ≥30 in all the samples were chosen for calculating the normalization factor. Expressed data were normalized using the median normalization. Following normalization, differentially expressed miRNAs were identified through fold change filtering.

Following the miRNA microarray analysis, collected total RNAs were also used to conduct the mRNA chip assay. The mRNA chip assay was conducted by the method of NimbleGen Mouse Gene Expression Microarrays (KangChen Bio-tech Inc.). Total RNA from each sample was quantified by the NanoDrop ND-1000 and RNA integrity was assessed by standard denaturing agarose gel electrophoresis. Total RNA of each sample was used for labeling and array hybridization as follows: i) Reverse transcription with the Invitrogen Superscript ds-cDNA synthesis kit; ii) ds-cDNA labeling with the NimbleGen one-color DNA labeling kit; iii) array hybridization using the NimbleGen Hybridization System followed by washing with the NimbleGen wash buffer kit; and iv) array scanning using the Axon GenePix 4000B microarray scanner. Scanned images (TIFF format) were imported into the NimbleScan software (version 2.5) for grid alignment and expression data analysis. Expression data were normalized through quantile normalization and the Robust Multichip Average (RMA) algorithm included in the NimbleScan software. The probe level (*_norm_RMA.pair) and gene level (*_RMA.calls) files were generated following normalization. All the gene level files were imported into the Agilent GeneSpring GX software (version 11.5.1) for further analysis. Genes that were ≥2 out of 2 samples have values greater than or equal to the lower cut-off: 100.0 (‘All Targets Value’) were chosen for data analysis. Differentially expressed genes were identified through fold-change filtering. Pathway analysis and gene ontology (GO) analysis were applied to determine the roles of these differentially expressed genes played in these biological pathways or GO terms. Finally, hierarchical clustering was performed to show distinguishable gene expression profiling among samples.

Reverse expressions of miRNA and mRNA were selected (upregulation of miRNAs by >2 fold and downregulation of mRNAs by <50%; or downregulation of miRNAs by <50% and upregulation of mRNAs by >50%) for miRNA-mRNA target prediction. Three different databases were used for miRNA target predictions: miRanda (www.microrna.org/microrna/getDownloads.do), miRBase (http://www.mirbase.org/) and TargetScan (www.targetscan.org/mmu_60/), and only those targets predicted by at least two databases were selected. Furthermore, based on the bioinformatics prediction, only miRNA target gene pairs with such opposing expression patterns were selected. Figures of miRNA-mRNA interactions were drawn by the cytoscape software (9). Selected mRNA was subjected to further hierarchical clustering analysis, pathway analysis (http://www.genome.jp/kegg/pathway.html) and GO analysis.

According to the miRNA chip report, 64 miRNAs showed >2-fold increase in pancreatic tissues of diabetic mice compared to the normal controls (Table I). However, 72 miRNAs showed ≥50% decrease in pancreatic tissues of diabetic mice compared to the normal controls (Table II). However, functions of the majority of miRNAs remained unclear.

According to the gene chip report, 507 mRNAs were upregulated by >2-fold while 570 mRNAs were downregulated by ≥50% (available upon request). Among the upregulated mRNAs, certain genes from viral myocarditis, tight junction, pentose and glucuronate interconversions, autoimmune thyroid disease and cocaine addiction pathways appeared to be enhanced in the pancreatic tissues of diabetic mice (Table III). Among the downregulated mRNAs, certain genes from pancreatic secretion, endocrine and other factor-regulated calcium reabsorption, Alzheimer's disease, endometrial cancer, mammalian target of rapamycin (mTOR) signaling pathway, T1DM, neurotrophin signaling pathway, aminoacyl-tRNA biosynthesis, maturity onset diabetes of the young, oxidative phosphorylation and T2DM pathways were attenuated (Table IV). Gene expression analysis identified that pancreatic islet cells showed severe impairments of functions in STZ-induced diabetic mice. These impaired functions included abnormal expressions of insulin synthesis, glucose or energy metabolism, cell skeleton and apoptosis genes. However, the factors that caused these wide changes of mRNAs remain to be discovered.

Whether there were certain correlations between these miRNA and mRNA remained unclear, as miRNA have the potential to inhibit the expression of mRNAs. In the present study, the target genes of miRNAs were predicted. Notably, certain upregulated miRNAs were accompanied by certain downregulated targeted mRNAs, while some downregulated miRNAs were accompanied by some upregulated targeted mRNAs (Fig. 1). Through software prediction, it is easy to observe that 1–18 miRNAs can co-target one common gene and one miRNA can target 1–23 different genes. Therefore, miRNA may form a complicated network to regulate the gene expression. There are 246 binding points between upregulated miRNA and downregulated mRNAs. Simultaneously, there are 583 binding points between downregulated miRNA and upregulated mRNAs.

Furthermore, the signaling pathway analysis regarding those changed targeted mRNAs and reversely changed miRNAs was conducted. The upregulated miRNAs were associated with the changes of the following signaling pathways: Transcriptional misregulation in cancer, dopaminergic synapse, insulin secretion, morphine addiction, alcoholism, endometrial cancer, Hippo signaling pathway, acute myeloid, leukemia, pancreatic secretion, dorso-ventral axis formation, mTOR signaling pathway, viral carcinogenesis, salivary secretion, forkhead box O (FoxO) signaling pathway, hepatitis C, phosphatidylinositol 3-kinase (PI3K)-protein kinase B (Akt) signaling pathway and Gap junction pathways (Table V). Among these signaling pathways, insulin secretion, pancreatic secretion, mTOR signaling pathway, FoxO signaling pathway and PI3K-Akt signaling pathway appeared to be closely associated with the development of diabetes. Other signaling pathways (transcriptional misregulation in cancer, dopaminergic synapse, morphine addiction, alcoholism, endometrial cancer, Hippo signaling pathway, acute myeloid leukemia, dorso-ventral axis formation, hepatitis C and Gap junction) require further validation on whether these pathways were directly or indirectly associated with the development of diabetes.

The downregulated miRNAs were associated with the changes of the following signaling pathways: Axon guidance, Ras-proximate-1 (Rap1) signaling pathway, pentose and glucuronate interconversions, estrogen signaling pathway, melanogenesis, retrograde endocannabinoid signaling, nucleotide excision repair, Huntington's disease, and amino sugar and nucleotide sugar metabolism (Table VI). Among these signaling pathways, pentose and glucuronate interconversions, amino sugar and nucleotide sugar metabolism appeared to be enhanced in the pancreatic tissues of diabetic mice, as KL, UGDH and phosphomannomutase 1 expression was upregulated. The Axon guidance, Rap1 signaling pathway, estrogen signaling pathways and nucleotide excision repair pathway were associated with the regulation of actin cytoskeleton, cell adhesion, cell cycle and cell apoptosis.

For the insulin or pancreatic secretion pathway, the miRNAs and mRNA pairs: mmu-miR-705/ATP1A3, mmu-miR-450a-2-3p/CCKAR, mmu-miR-30b-3p/GNAS, mmu-miR-139-5p/KCNMA1 and mmu-miR-669m-3p/KCNMA1 may be involved in the potential cross-talks between miRNAs and mRNAs (Table VII).

For the FoxO signaling pathway, the miRNAs and mRNA pairs: mmu-miR-323-3p/mmu-miR-369-3p/SOS1, mmu-miR-302a-3p/STK11, mmu-miR-302b-3p/mmu-miR-491-5p/GADD45A and mmu-miR-133a-3p/mmu-miR-133b-3p/mmu-miR-1956/GABARAPL1 may be involved in the potential cross-talks between miRNAs and mRNAs (Table VIII).

For the PI3K-Akt signalingpathway, the miRNAs and mRNA pairs: mmu-miR-139-5p/GNB1, mmu-miR-881-3p/SPP1, mmu-miR-337-3p/YWHAB, mmu-miR-323-3p/mmu-miR-369-3p/SOS1, mmu-miR-1956/PPP2R1B and mmu-miR-1956/mmu-miR-668-3p/mmu-miR-668-3p/PPP2R2A may be involved in the potential cross-talks (Table IX).

For the mTOR signaling pathway, the miRNAs and mRNA pairs: mmu-miR-302a-3p/STK11, mmu-miR-1953/STRADA and mmu-miR-302a-3p/RRAGD may be involved in the potential cross-talks (Table X).

As shown in the above results, there were certain promising cross-talks that mediated the development of diabetes but require further validation (Table XI).