Abstract Type 2 diabetes mellitus (T2DM) increases the susceptibility of bone fragility. The underlying mechanisms have, however, remained largely unknown. MicroRNAs (miRNAs) are short single-stranded non-coding RNA molecules with utility as biomarkers due to their easy accessibility and stability in bodily fluids. Here, we aimed to use an unbiased approach to identify miRNAs dysregulated in a polygenic mouse model of T2DM. Genome-wide analysis of miRNAs in serum, BM, and bone from the polygenic TallyHo/JngJ (TH) mice, which recapitulate T2DM in humans, was performed. This analysis was compared to the recommended control Swiss Webster Random/Jackson (SWR/J) and a strain-matched non-diabetic control (TH-ND). When comparing TH mice with TH-ND using an adjusted p-value false discovery rate (FDR) cut-off of 0.2 to identify differentially expressed miRNAs, mmu-miR-466i-5p and mmu-miR-1195 were found to be up-regulated in both serum and in BM. Dysregulated miRNAs were not found in bone tissue. When comparing TH-ND mice with SWR/J using the same FDR cut-off, mmu-miR-351-5p, and mmu-miR-322-3p were upregulated in both BM and serum, while mmu-miR-449a-5p and mmu-miR-6240 were downregulated in BM and serum. Dysregulated miRNAs in BM or cortical bone compared to serum between TH-ND mice and SWR/J were investigated for their cell-type enrichment to identify putative donor cells and their gene target networks. Gene target network analysis revealed genes involved in diabetes-related signaling pathways as well as in diabetic bone disease. Cell-type enrichment analysis identified hsa-miR-449a enriched in immune cells, hsa-miR-592 in hepatocytes and endothelial cells, while hsa-miR-424-3p, hsa-miR-1-3p, and hsa-miR-196b-5p were enriched in mesenchymal stem cells and their derived tissues. In conclusion, our comparative miRNA profiling sheds light on differential expression patterns between SWR/J and both subgroups of TH. No differences were observed between TH and TH-ND, suggesting the genetic background of SWR/J may be responsible for the change of dysregulated miRNA. Keywords: microRNA, type 2 diabetes, TallyHo, biomarker, next-generation sequencing, osteoporosis, circulating microRNA Graphical Abstract Graphical Abstract. [42]Graphical Abstract [43]Open in a new tab Introduction Diabetes is well-defined by elevated blood glucose levels caused by a dysfunction in insulin metabolism. Type 1 diabetes mellitus (T1DM) is characterized by insufficient insulin production resulting from the autoimmune destruction of pancreatic islet beta cells.[44]^1 In contrast, type 2 diabetes mellitus (T2DM) primarily arises from insulin resistance, often influenced by genetic predisposition, obesity, and an unhealthy lifestyle.[45]^2 With the global prevalence of diabetes on the rise, an estimated 463 million people are affected, with T2DM accounting for ~90% of all cases.[46]^3 A range of late-stage complications accompany both forms of diabetes, including macro- and microvascular disease, retinopathy, and neuropathy.[47]^4 Recently, studies have shown patients with T1DM and T2DM have an increased risk of fractures,[48]^5 albeit through different mechanisms. While low BMD is the most common risk factor for patients with T1DM, patients with T2DM have normal or increased BMD.[49]^6 This indicates that the diabetic bone quality is less resistant to fractures. This phenomenon is also known as the diabetic paradox of bone fragility, suggesting that other factors aside from BMD must contribute to the increased fracture risk in T2DM. Substantial evidence has shown T2DM is linked to low bone turnover, leading to a failure to renew microdamage that naturally happens over time, resulting in the accumulation of bone that is structurally weak.[50]^7 In addition, high levels of advanced glycation end-products (AGEs) may contribute to stiff collagen structures and altered bone cell functions in diabetic bone disease. Further, under diabetic conditions, bone formation is significantly impaired.[51]^8 One contributing factor to decreased osteogenesis and increased adipogenesis in diabetic bone disease is the inhibition of Wnt signaling. Additionally, enhanced oxidative stress may hamper osteoblast function.[52]^9^–[53]^15 More recently, microRNAs (miRNAs), short single-stranded non-coding RNAs that regulate gene expression at a transcriptional level,[54]^16^–[55]^19 have been shown to be dysregulated in diabetic bone disease in humans and in rodent models. We previously detected several dysregulated miRNAs in the serum and bone of ZDF rats, a monogenetic model of diabetic bone disease based on a mutation in the leptin receptor gene. Further, we found that these miRNAs are differently affected by bone-specific treatments such as anti-sclerostin antibodies (anti-scl) and parathyroid hormone (PTH), or insulin.[56]^20 A meta-analysis study reported 40 miRNAs to be significantly dysregulated in T2DM patients. This includes miR-375, miR-142-3p, and miR-144, along with 2 tissue miRNAs, miR-199a-3p and miR-233. Furthermore, another study investigated stress-related miRNA biomarkers in relation to T2DM and discovered that miR-148b, miR-223, miR-130a, miR-19a, and miR-26b are potential biomarkers for T2DM.[57]^21^,[58]^22 To understand the role of miRNA in diabetic bone disease with a polygenetic perspective and assess if we would find similar miRNAs regulated as in monogenetic T2DM models, we aimed to analyze serum and bone tissue samples from TallyHo/JngJ mice. The TallyHo/JngJ strain is a polygenetic murine model that recapitulates T2DM with similar features as seen in humans. Previous studies have reported that they develop a bone phenotype characterized by reduced BMD, altered bone microarchitecture, and increased bone turnover when compared to SWR/J mice as their recommended control.[59]^23^–[60]^25 In this study, we employed an untargeted profiling technique, next generation sequencing (NGS), to quantify miRNA levels in serum and ulna samples from TallyHo/JngJ. With the generated NGS, data we performed hierarchical clustering analysis, univariate and multivariate statistical analysis, as well as gene target network analysis and cell-type enrichment analysis to identify the most promising miRNA biomarker candidates for diabetic bone disease. Materials and methods Animals Animal procedures were approved by the institutional Animal Care Committee of the Technische Universität Dresden and the Landesdirektion Sachsen (TVV 2017/20). Twelve-week-old male mice TALLHYHO/Jng and SWR/J were purchased from Jackson Laboratory and house under institutional guidelines. Animals were maintained in groups up to 5 animals in a light–dark cycle of 12/12 h at room temperature in filter-top cages and had ad libitum access to their respective drinking water and standard chow diet. Body weight measures, serum analysis, and assessment of bone mass and bone microarchitecture were performed at the TUD in all mice as described by Emini et al.[61]^26 To investigate changes in circulating and bone miRNA expression as a consequence of the TH-ND phenotype, serum, BM, and cortical bone samples were harvested at Week 12 at the Bone Lab located in the TUD (Technical University of Dresden, Germany) and used for RNA extraction. RNA extraction from cortical bone was performed at TUD, while RNA extraction from serum and BM and NGS were performed at TAmiRNA GmbH for a genome-wide screen of miRNA levels in a subset of 45 serum samples as shown in [62]Table 1. BM and cortical bone tissue samples were collected from the same rats used for serum collection. Table 1. Summary of the animals used for the study. Genotype Serum Bone (cortical) Bone marrow Total TallyHo/JngJ 5 5 6 16 Strain-matched non-diabetic control 5 5 5 15 Swiss Webster Random/Jackson 5 4 5 14 Total 15 14 16 45 [63]Open in a new tab Serum, bone marrow, and cortical bones were collected from 11 male TallyHo/JngJ mice, 6 TH (diabetic) and 5 TH-ND (non-diabetic), and 5 male SWR/J mice at the age of 12-wk-old. Micro-CT, serum, and tissue analyses were performed as indicated in the paper submitted paper by Emini et al.[64]^26 RNA extraction from serum and bone marrow Total RNA was extracted from 100 μL serum and flushed bone marrow using the miRNeasy Mini Kit (Qiagen, Germany) as described by Kocijan et al.[65]^27 100 μL of each sample were mixed with 1000 μL of Qiazol and 1 μL of a mix of 3 synthetic spike-in controls (Exiqon, Demark). After a 10-min incubation at room temperature, 200 μL of chloroform were added to the lysates, followed by centrifugation at 12 000 × g for 15 min at 4 °C. 650 μL of the upper aqueous phase were transferred to a miR-Neasy mini column where RNA was precipitated with 750 μL ethanol, followed by automated washing with RPE and RWT buffer in a QiaCube liquid handling robot. Finally, total RNA was eluted in 30 μL nuclease free water and stored at −80 °C. The RNA yield was analyzed for all samples. Due to low RNA concentrations in biofluids, which render most RNA quantifications inaccurate, serum RNA extraction efficiency was assessed through RT-qPCR analysis of spike-ins added prior to the RNA extraction step. Spike-in controls showed acceptable variation by RT-qPCR for all serum RNA samples, proving an acceptable quality for their use in further experiments. Microcapillary electrophoresis (Bioanalyzer, Agilent) was performed to determine the RNA concentration and the RNA integrity (RIN) in all tissue RNA samples. RNA quality was also high and remained stable among bone marrow RNA samples, as RIN values were in a range of 8.7-9.4 for all samples. RNA extraction from cortical bone Cortical bones were harvested from all mice, bone marrow was flushed, and the bone was homogenized with a mortar at the Bone Lab in the TUD (Dresden), and samples were immediately snap frozen in liquid nitrogen and stored at −80 °C. Total RNA from bone tissue was extracted using TRIzol reagent (Invitrogen, Darmstadt, Germany), respectively, following the manufacturer’s protocol and quantified using a Nanodrop spectrophotometer (Peqlab, Erlangen, Germany). Similar to the bone marrow RNA samples, RNA concentration and RIN were determined for all cortical bone tissue RNA samples to confirm their quality. The quality of the cortical bone tissue RNA samples was lower than the one of bone marrow RNA samples, but similar to bone marrow samples, the RIN values remained stable among cortical bone RNA samples, as RIN values were in a range of 2.0-2.7 for all samples. All cortical bone tissue RNA samples were stored at −80 °C. Library preparation for small RNA-Seq Based on previous studies performed by Khamina et al.,[66]^28 library preparation was performed using the RealSeq-Biofluids Plasma/Serum miRNA Library kit for Illumina sequencing (RealSeq Biosciences, 600-00048; protocol 20181220_RealSeq-BF_CL) according to the manufacturer’s protocol. Briefly, 8.5 μL of extracted RNA were used as input, using the same RNA concentration (11.8 ng/μL) for all cortical bone tissue and bone marrow RNA samples. Adapter-ligated libraries were circularized, reverse transcribed and amplified. Library PCR was performed with Illumina primers included in the kit and using 22 cycles for serum samples, 18 cycles for bone marrow samples, and 29 cycles for cortical bone tissue samples. In total, 15 miRNA libraries were prepared from serum samples, 14 miRNA libraries were prepared from cortical bone samples, and 16 from bone marrow samples. All 45 libraries were analyzed for library fragment distribution using the Agilent DNA1000 kit (Agilent Technologies, 5067-1504) with Agilent DNA1000 reagents (Agilent Technologies, 5067-1505). The generated libraries were pooled in an equimolar proportion, and the obtained pool was size-selected with the BluePippin system using a 3% agarose cassette with a target range of 100-250 kb (Sage Science, BDQ3010) to remove DNA fragments outside of the target range. The pooled and purified libraries were analyzed for fragment distribution on an Agilent High Sensitivity DNA kit (Agilent Technologies, 5067-4626) with Agilent High Sensitivity DNA reagents (Agilent Technologies, 5067-4627). The library pool was then sequenced on an Illumina NextSeq550 (single-read, 75 bp) according to the manufacturer’s protocol at the Vienna BioCenter Core Facilities (VBCF), Vienna, Austria. RT quantitative polymerase chain reaction analysis Starting from total bone marrow and serum RNA samples, cDNA was synthesized using the miRCURY LNA RT kit (Qiagen, Cat No. 339340). In total, 2 μL of total RNA were used for serum samples and 40 ng from bone marrow per 10 μL RT reaction. To monitor RT efficiency and presence of impurities with inhibitory activity, a synthetic RNA spike-in (cel-miR-39-3p) was added to the RT reaction. PCR amplification was performed in a 96-well plate format using miRCURY SYBR Green qPCR (Qiagen, Cat No. 339347) and miRCURY LNA miRNA primers miR-592, miR-449a, miR-351-5p, and miR-322-3p (Qiagen Cat No. 339306). qPCR was performed in a Roche LC480 II instrument (Roche, Germany). All steps were performed according to the manufacturer’s instructions. To calculate the cycle of quantification values (Cq-values), the second derivative method was used. Spike-in control cel-miR-39-3p values were used for monitoring data quality and showed acceptable variation. Cq-values of endogenous serum miRNAs were normalized to the RNA spike-in controls by subtracting the individual miRNA Cq-value from the RNA spike-in Cq, thus obtaining delta-Cq (dCq) values that were used for the statistical analysis. Similarly, Cq-values of endogenous bone marrow miRNAs were normalized to the mean of 2 established reference RNAs—5S and SNORD65—by subtracting the individual miRNA Cq-value from the mean Cq of the references, thus obtaining delta-Cq (dCq) values