Introduction

Chronic kidney disease (CKD) is a major global health problem associated with significant morbidity and mortality1. Hypertension and diabetes mellitus are known to be the main contributors to the development and progression of CKD in developed countries2. Reducing blood pressure using angiotensin II receptor blockers or angiotensin-converting enzyme is the first-line therapy for delaying CKD progression3. Despite the control of the renin-angiotensin system glycemic control is essential for therapies combating diabetic nephropathy.

Dipeptidyl peptidase (DPP)-4 inhibitors are indicated for the treatment of diabetes, however beneficial effects independent of glucose and renal blood pressure have been shown in animal models of both diabetic4 and non-diabetic CKD5. It is well known that the antidiabetic effects of DPP-4 inhibitors are mediated via increases in levels of the incretin hormones glucagon-like peptide 1 (GLP-1) and glucose-dependent insulinotropic polypeptide (GIP)6. Moreover, recent experimental data demonstrated that linagliptin exerts anti-inflammatory7, anti-oxidant8 and anti-fibrotic9 effects, however the exact underlying renoprotective molecular mechanisms remain unclear.

In diabetic and non-diabetic CKD pathogenesis TGF-β signaling is known to induce the synthesis and accumulation of extracellular matrix (ECM) leading to fibrosis and hypertrophy in various kidney cell types10. TGF-β-induced expression of ECM genes during the development of diabetic nephropathy is mediated by several miRNA species. The functional crosslink between TGF-β and responsive miRNA expression has been demonstrated in renal cells, such as proximal tubular epithelial cells, mesangial cells and podocytes. Increased glomerular miR-21 level are positively associated with the albumin-creatinine ratio (ACR) in patients with diabetic nephropathy and it was postulated that altered miRNA-21 level might serve as an indicator of podocyte damage11. Another prominent pro-fibrotic miRNA is miR-199a which is involved in the regulation of TGF-β signaling12. Increased renal expression of miR-199a was observed in the unilateral ureteral obstruction mouse model and inhibits the expression of caveolin 1 which is a critical process in the activation of fibroblasts by TGF-β13. In contrast, expression of kidney damage protective miRNAs, such as members of the miR-200-, miR-30- and miR-29-families, is decreased in various kidney cell types10. miR-29 family members are down-regulated in CKD such as diabetic nephropathy, focal glomerulosclerosis, membraneous nephropathy and IgA nephropathy14,15. In vitro studies revealed that inhibition of overexpression of and knockdown of miR-29 enhanced TGF-β induced expression of collagens type I and III by renal tubular cells16. Streptozotocin-induced diabetic miR-29 transgenic mice showed improved renal function and better podocyte viability whereas knockdown of miR-29 promoted podocyte apoptosis, proteinuria and subsequent renal dysfunction17. The DPP-4 inhibitor linagliptin restored the level of renal miR-29s, which was associated with the inhibition of TGF-β-induced endothelial to mesenchymal transition (EndMT) in the kidneys of diabetic mice9.

Extracellular vesicles, such as exosomes, are secreted in large quantities from all nephron segments18 into the urine and may provide valuable insights into renal pathophysiology. Urinary exosomal miRNA content is altered in patients with focal segmental glomerulosclerosis19, in type I diabetic patients with incipient diabetic nephropathy20, and in type II diabetic patients with diabetic nephropathy21. Moreover, it was shown that altered urinary exosomal miR-29c level might potentially serve as a predictor of early fibrosis in lupus nephritis22.

In the present study we investigated the effects of linagliptin and compared with those of telmisartan which is one of the most commonly used ARBs on dysregulated miRNAs in the kidney in the rat 5/6 nephrectomy model, one of the most well-established experimental non-diabetic CKD models. We aimed to identify those effects on urinary exosomal miRNA level which might serve as potential novel biomarkers for monitoring both disease progression and treatment effects.

Materials and Methods

Experimental design

The aim of this study was to identify the effects of telmisartan and linagliptin on miRNAs differentially expressed in kidneys from 5/6 Nx rats and to assess whether these effects can be accurately quantified in urinary exosomes. As a first, step large-scale miRNA and mRNA expression profiles from the same renal tissue samples were evaluated using the Nanostring and RNA sequencing platforms, respectively. Changes in miRNA expression level were identified by a moderated t-test, evaluating only those miRNAs with at least 1.5-fold expression changes at a P-value <0.05. In addition, P-values were further corrected for multiple testing according to Benjamini-Hochberg. Functional annotations of these miRNAs were searched in several databases including miRWalk (http://www.umm.uni-heidelberg.de/apps/zmf/mirwalk), TargetScan (http://www.targetscan.org), miRBase (http://www.mirbase.org), and PubMed (http://www.ncbi.nlm.nih.gov/pubmed). Inverse correlated miRNA-target mRNA pairs were identified using Ingenuity Pathway Analysis (http://www.qiagenbioinformatics.com). Highly predicted target mRNAs with a Spearman´s correlation coefficient of at least −0.7 were further assessed. Quantitative real-time polymerase chain reaction (PCR) was then used to detect differentially expressed renal miRNAs in urinary exosomes.

Animals

Eight week old Wistar rats were purchased from Charles River Laboratories International, Inc. (Wilmington, MA). This study (animal studies and protocols) was approved by the Committee on the Ethics of Animal Experiments (Landesamt fuer Gesundheit und Soziales), Berlin, Germany and is in accordance with University Guidelines for Use of Laboratory Animals. The animals were assigned into 4 groups: sham operation (sham) + placebo; 5/6 nephrectomy (5/6 Nx) + placebo; 5/6 Nx + telmisartan; 5/6 Nx + linagliptin. The Nx operation was performed as follows: uninephrectomy at Week 1, followed at Week 3 by amputation of the poles of the remaining kidney. Sham operations were performed at the same time points. Linagliptin (83 mg/kg/day in chow) and telmisartan (5 mg/kg/day in drinking water) were administered from Day 4 after the second surgery for 13 weeks. Immediately after this treatment period animals were sacrificed and organs were harvested. Urine albumin-to-creatine ratio (UACR), interstitial fibrosis, plasma DPP-4 activity and urinary DPP-4 activity/creatinine ratios of the experimental groups are summarized in Table S1. Further physiology and biochemistry parameters of animals were described in detail by Tsuprykov et al.23 and isolated total RNA was used for the following experiments.

Nanostring analysis

Total RNA from kidney tissue was extracted using RNeasy Fibrous Tissue Mini Kit (Qiagen, Hilden, Germany). RNA purity and quantity were assessed by NanoDrop 2000 (ThermoScientific, Waltham, MA, USA). Total RNA (100 ng) was used to assess the miRNA expression using the nCounter Rat v1.5 miRNA CodeSet (based on miRBase v17, Nanostring Technologies, Seattle, WA, USA) which contains a library of 420 probes. The purified complexes were quantified on the nCounter Digital Analyzer and analyzed by nSolver software (v1.1; Nanostring Technologies, Seattle, WA). The exact data analysis procedure can be found at: http://www.NanoString.com/media/pdf/MAN_nCounter_Gene_Expression_Data_Analysis_Guidelines.pdf.

mRNA library preparation, sequencing and data analysis

RNA sequencing libraries were prepared using Illumina’s TruSeq RNA Sample Prep Kit - v2 (Illumina Inc.) according to the manufacturer’s instructions. The library concentrations were then quantified with the Quant-iT PicoGreen dsDNA Assay Kit (Quant-iT) using CLARIOstar (BMG LABTECH) and the library quality was determined by checking cDNA fragment size using a DNA1000 Kit on the Agilent Bioanalyzer 2100 (Agilent Tech Inc.). Libraries were then normalized to 2 nM and subjected to cluster generation on a cBot system followed by single-read sequencing of 52 bp on an Illumina HiSeq. 2000 instrument (Illumina Inc.). Quality check (QC) of the obtained reads was performed using FASTQC v0.10.1. Based on RPKMs/FPKMs as obtained in the subsequent analysis (1.6.2), principal component analysis (PCA) and hierarchical clustering analysis were further used to identify outliers. Outliers were removed from subsequent analysis. Reads were then mapped to the human reference genome hg19 (GRCh37 Ensembl v. 70, primary assembly) using STAR v2.3.0e. Mapped reads were QCed using RNA-SeQC v1.1.7. Gene expression intensities are represented as either uRPKM/uFPKM or mRPKM/mFPKM values and these were calculated based on Ensembl v70 gene annotations using cufflinks v2.2.1 for Linux_x86_64. RPKM/FPKM values account for the different lengths of genes as well as different number of reads measured for the respective sample. Only those reads that could unambiguously be assigned to the respective gene were considered in the analysis to determine uRPKM/uFPKM values. For mRPKM/mFPKM values multiple mapping reads were also considered. SAM to BAM conversion was done using picard-tools-1.77. Differential expression analysis was conducted based on read counts per gene. Read counts per gene (Ensembl v70 gene annotations) were either obtained using htseq-count from HTSeq v0.5.3p9 and samtools v0.1.18 or based on cufflinks v2.2.1 results. Fold changes and their respective significance were computed based on the read counts obtained for each gene using R and Bioconductor packages edgeR, DESeq2 or voom in conjunction with limma.

Integrated analysis of miRNA and mRNA expression

Spearman correlation coefficients (rs) between mRNA and miRNA expression values for each sample were calculated. Correlation coefficients ranged from −1 to 1 and rs < −0.7 were considered for further evaluation.

Urinary exosomal miRNA expression

The animals were placed in metabolic cages to obtain 24-hour urine samples. Urine samples from three animals were pooled due to the limited amount of urine, resulting in 4 samples in total per experimental group. Urinary exosomal miRNAs were isolated using the exoRNeasy Serum/Plasma Maxi Kit (QIAGEN, Hilden, Germany) and the expression of the miRNAs was screened using Taqman Fast Advanced MasterMix (Applied Biosystems) and the rodent Taqman miRNA Array, Card A V.2.0 (Applied Biosystems). The gene expression analysis was run on a SDS7900HT real-time PCR system (Applied Biosystems by ThermoFisher Scientific). TaqMan MicroRNA Reverse Transcription Kit (Applied Biosystems) and Megaplex RT Primers, Rodent Pool A v2.1 (Applied Biosystems) were used to reverse transcribe miRNAs. The RT products were pre-amplified using Megaplex RT Pre-Amp Primers, Rodent Pool A v2.1 (Applied Biosystems). U6snRNA was used for normalization. PCR reactions were performed with the TaqMan gene expression master mix (Life Technologies) according to the manufacturer´s protocol on a SDS7900HT real-time PCR System. Raw Ct values were calculated using the SDS software v2.4. Threshold cycle (Ct) values > 35 were excluded from any calculations. Fold change of expression was calculated using the comparative Ct method (2−ΔΔct)24.

Statistical analysis

Differential expression was calculated applying the linear models approach incorporated in the limma software package25,26 using Tibco Spotfire version 6.5.2 (TIBCO Software, Palo Alto, USA). miRNAs >1.5-fold and mRNAs >1.5-fold differentially expressed between the compared groups with a P-value adjusted for multiple testing <0.05 were considered to be significantly differentially expressed. Adjustment of P-values for multiple testing was performed according to Benjamini and Hochberg27. Heatmaps, principal component analyses and volcano plots were generated via Tibco Spotfire (TIBCO Software). Hierarchical clustering was calculated using unweighted pair group method with arithmetic mean as the default clustering method and Euclidean distance measure.

Statistical analysis for the PCR data was performed using ANOVA and Tukey’s multiple comparisons test. For group-wise comparisons, expression was considered significantly changed if the respective fold change was >1.5 and P-value <0.05.

Results

Effects of telmisartan and linagliptin on differentially expressed miRNAs in kidneys from 5/6 Nx rats

Principal component analysis revealed that the miRNA expression profiles were relatively similar within each experimental group whereas the respective groups clustered separately (Fig. 1). Compared with sham-operated animals, in 5/6 Nx-placebo-treated rats, the renal expression of the 13 miRNAs miR-142-3p, miR-195, miR-21, miR-150, miR-199a-3p, miR-199a-5p, miR-214, miR-223, miR-290, miR-322, miR-497, miR-532-5p and miR-542-5p was significantly up-regulated (Fig. 2a), whereas the renal expression of the 7 miRNAs miR-144, miR-190, miR-203, miR-29b, miR-29c, miR-32 and miR-451was significantly down-regulated (Fig. 2b). Telmisartan significantly ameliorated the expression of miR-150, miR-199a-5p, miR-199a-3p, miR-223 and miR-214, whereas linagliptin significantly ameliorated the expression of miR-150, miR-199a-3p and miR-322 compared to the 5/6 Nx placebo group (Fig. 2a). Significant restoring effects on the expression of miR-29b, miR-29c and miR-32 were observed for both telmisartan- and linagliptin-treated 5/6 Nx rats compared to placebo-treated 5/6 Nx rats. The degree of fold changes of distinct up- and down-regulated renal miRNA expression profiles correlates with the degree of renal interstitial fibrosis (Supplementary data, Figs. S1 and S2). The functions of the deregulated miRNAs based on published data are summarized in Table 1. Most of the deregulated miRNAs in the kidney from 5/6 Nx + placebo rats were involved in fibrotic processes, such as the 8 miRNAs miR-21, miR-29b, miR-29c, miR-150, miR-199a-3p, miR-199a-5p, miR-203 and miR-214. The 6 miRNAs miR-142-3p, miR-144, miR-223, miR-290, miR-322 and miR-451 are involved in cell development and/or differentiation while the remaining 6 miRNAs miR-190, miR-195, miR-32, miR-497, miR-532-5p and miR-542-5p are involved in cell cycle processes mostly via acting as tumor suppressors (Table 1).

Figure 1
figure 1

Principal component analysis of miRNA expression profiling results. Displayed are the first three major components from the principal component analysis. (green = sham+placebo; red = 5/6 Nx + placebo; blue = 5/6 Nx + telmisartan; yellow = 5/6 Nx + linagliptin).

Figure 2
figure 2

Up- (a) and down-regulated (b) renal miRNA expression. MiRNA levels of sham-operated (green), placebo-treated 5/6 Nx (red), telmisartan-treated (blue) and linagliptin-treated 5/6 Nx (yellow) rats revealed by Nanostring analysis. Absolute values are displayed for each animal by normalized reporter cell counts (RCC).

Table 1 Effects of telmisartan and linagliptin on differentially expressed miRNAs in kidney from 5/6 Nx rats.

Integrated miRNA-mRNA analysis

To identify target mRNAs of the 20 differentially expressed miRNAs in kidneys from placebo-treated 5/6 Nx rats, RNA was sequenced from the corresponding kidney tissues. The principal component analysis revealed that most of the expression profiles of the sham + placebo rats clustered separately to the 5/6 Nx groups whereas the treated 5/6 Nx groups are spread across the 5/6 Nx groups (Supplementary data, Fig. S3). Deregulated mRNA expression is depicted in the heat map in Fig. 3 and all deregulated genes are summarized in Supplementary Table 2. In total, the expression of 1404 genes is deregulated (1.5-fold; P < 0.05) in placebo-treated 5/6 Nx rats compared to the sham-operated control rats: the expression of 1015 genes is up-regulated whereas the expression of 389 genes is down-regulated. The effects of telmisartan and linagliptin on these deregulated genes are shown in Fig. 3 and summarized in Supplementary Table 2.

Figure 3
figure 3

Renal mRNA expression. Heatmap of renal gene expression of sham-operated (green box), placebo-treated 5/6 Nx (red box), telmisartan-treated (blue box) and linagliptin-treated 5/6 Nx (yellow box). Hierarchical clustering of all significantly deregulated mRNAs is depicted according to their log2-transformed expression levels (red = high; dark grey = low).

Integrating miRNAs and their corresponding target mRNAs (both 1.5-fold deregulated at P < 0.05), resulted in 113 renal miRNA/target mRNA pairs with a Spearman’s correlation coefficient rs < −0.7. The predicted targets displayed in Table 2 are based on either experimental observations or a high level of prediction. Inverse correlations were observed for the four up-regulated miRNAs miR-532-5p, miR-199a-5p, miR-199a-3p and miR-142-3p and the two down-regulated miRNAs miR-203 and miR-29. The effects of telmisartan and/or linagliptin on these identified miRNA/target mRNA pairs are characterized by the loss of Spearman´s correlation coefficient of rs

Table 2 Differential regulated miRNAs in 5/6 Nx + placebo/+telmisartan/+linagliptin compared to Sham + placebo rats and their corresponding target mRNAs.

 < −0.7.

To further characterize the effects of telmisartan and linagliptin on the effects of mRNA-miRNA interactions in 5/6-Nx rats pathway analysis was performed using MSigDB28. Hypergeometric testing using the hallmark gene set from MSigDB revealed that the down-regulated miRNAs corresponds to up-regulated target mRNAs that are involved in fibrotic pathways, such as epithelial mesenchymal transition and inflammatory processes, such as TNFα signaling, IL-6/JAK/STAT3 signaling or IL-2/STAT5 signaling (Fig. 4). Down-regulated mRNAs that correspond to up-regulated miRNAs are involved in oxidative phosphorylation and adipogenesis (Fig. 4).

Figure 4
figure 4

Gene set enrichment analysis of miRNA-mRNA pairs according to 38 hallmark pathways. Effects on distinct pathways are represented by colors indicating enrichment scores. miR-29 represents miR-29a/b/c.

Urinary exosomal miRNA expression profiling

To assess which of the aforementioned treatment affected miRNA expression profiles are reflected in urinary exosomes qRT-PCR was employed using TaqMan Array Cards, which contain all deregulated miRNAs described above. Only miRNAs with the highest expression levels in the kidney - miR-21, miR-29b, miR-29c and miR-203 - could be detected in urinary exosomes. The levels of miR-21 and miR-203 were not affected across all groups in urinary exosomes. Urinary exosomal levels of miR-29b and miR-29c were decreased in placebo-treated 5/6 Nx rats compared to the sham-operated group. Telmisartan significantly restored the levels of both urinary exosomal miRNAs miR-29b and miR-29c whereas linagliptin restored the level of only miR-29c compared to the 5/6 Nx placebo group (Fig. 5a). Correlation analysis revealed that down-regulated urinary exosomal miR-29c expression was significantly negatively correlated with increasing UACR (r2 = −0.81; P = 0.0003, Fig. 5b).

Figure 5
figure 5

Urinary exosomal miRNA expression. Urinary exosomal levels of miR-21, miR-203, miR-29b and miR-29c in sham-operated (green), placebo-treated 5/6 Nx (red), telmisartan-treated (blue) and linagliptin-treated 5/6 Nx (yellow). (a) Relative urinary exosomal miRNA expression analyzed by qRT-PCR was normalized to the mean expression of sham-operated control rats. Significant differences 5/6 Nx + placebo are indicated by *(P < 0.05). (b) Correlation between fold change of urinary exosomal miR-29c expression and UACR. Data were compared by Spearman’s correlation coefficient.

Discussion

This study shows that the DPP-4 inhibitor linagliptin and the ARB telmisartan induced effects on renal miRNA and their corresponding target mRNA expression in non-diabetic 5/6 Nx rats. Integrating both renal miRNA and mRNA expression profiles enriched pathways, which are deregulated in 5/6 Nx rats, such as fibrotic pathways including epithelial mesenchymal transition and inflammatory processes, such as TNFα signaling, IL-6/JAK/STAT3 signaling or IL-2/STAT5 signaling. In particular, telmisartan and linagliptin exert their strongest effects on fibrotic processes.

Recently it was demonstrated that linagliptin was comparable to telmisartan in preventing CKD progression in non-diabetic, non-glucose dependent 5/6 Nx rats23. Tsuprykov et al.23 demonstrated that linagliptin’s action was pronounced with respect to reduction of renal interstitial fibrosis and glomerular hypertrophy, whereas telmisartan’s beneficial action was mostly due to its blood pressure-lowering effect. In this study, among all 14,605 sequenced mRNAs and 420 detected miRNAs in the kidney both mRNA and miRNA expression levels were not significantly different between telmisartan and linagliptin treated 5/6 Nx rats. Nevertheless, the extent of the beneficial effects exerted by each drug was different. The expression of one gene, namely Renin, was approximately 16-fold (adj- P-value = 0.4) stronger induced in the kidneys of telmisartan treated 5/6 Nx rats compared to linagliptin-treated rats. There was a trend that the beneficial effects characterized by the effects on the deregulated expression levels in 5/6 Nx are more pronounced after telmisartan treatment compared to linagliptin treatment (Table S2). Moreover, a combination of telmisartan and linagliptin might have additive effects as demonstrated by Alter et al.5. However, this was not investigated in this study and should be considered in future studies. The pharmacological effects of telmisartan and linagliptin are attributed to the interference with multiple pathways. An overlap on gene expression level has been previously described23 and confirmed here in addition with miRNA expression level. It is worthwhile to mention that the extent of deregulation is different for individual pathways. Nevertheless, proteomic analyses revealed that post-translational modifications might further differentiate the mechanisms of both drugs23. In particular, linagliptin induced a significant upregulation of the HNRNPA1 fragment, an important regulator of the atrial natriuretic peptide-dependent guanylate cyclase pathway, which might contribute to the anti-fibrotic effects. Further investigations are needed to integrate all omics approaches (miRNomics, transcriptomics, proteomics) to further dissect the mode of action of both drugs. In contrast to animal models of diabetic CKD9,29 and kidney cell lines exposed to high glucose concentrations30 the TGFβ/SMAD2/3 pathway which is a hallmark of renal fibrosis was not directly modulated by linagliptin in the non-diabetic 5/6 Nx model. Nevertheless, indirect modulation of renal fibrosis can occur via pro- and anti-fibrotic miRNAs and our study shows that the degree of deregulation of the renal miRNA expression correlates with increasing interstitial fibrosis.

The renal expression of the up-regulated pro-fibrotic miRNAs miR-150, miR-199a-3p and miR-199a-5p is significantly attenuated after telmisartan or linagliptin treatment. miR-199a-5p is up-regulated during the fibrogenic response to tissue injury13 and is a general marker for fibrogenesis in various animal models such as bleomycin-induced lung fibrosis, CCl4-induced liver fibrosis and unilateral ureteral obstruction model of kidney fibrosis13. Our study revealed that most of the protein-encoding genes potentially regulated by miR-199a-5p are involved in cytoskeletal organization and cell proliferation processes such as ARHGEF5, VEGFA or SDPR. SDPR encodes Cavin 2, which is involved in maintaining endothelial morphology and function31 and inhibition of Cavin 2 expression might cause a loss of cellular integrity and accelerate endothelial mesenchymal transition.

Using different CKD models it was shown that members of the miR-29 family (miR-29s) protect against renal fibrosis by inhibiting endothelial mesenchymal transition and preventing the deposition of ECM32,33. The miR-29 family members share a common seed region sequence and are predicted to target the same genes. However, differential regulation of members of the miR-29 family might be a result of the subcellular distribution, suggesting their functional consequence may not be identical. Our study revealed that the expression of miR-29a is not altered between experimental groups. In previous studies it appeared that the three miR-29s may be regulated by distinct mechanisms. In the renal medulla of the Dahl salt-sensitive (SS) rat and a consomic rat strain derived from it, miR-29a is more abundantly expressed compared to miR-29b and miR-29c and the three miR-29 species respond differently to 3 days of a high-salt diet, which might be a consequence of distinct post-transcriptional processing16. In addition, our study does not distinguish between miR-29s-3p and -5p-arms. Nevertheless, it is expected that both arms are affected during fibrotic processes in the kidney8,34. Predicted target genes for the miR-29 family members largely overlap35 which includes genes encoding proteins that are involved in the synthesis of the ECM, including collagen isoforms, laminins, fibrilins, elastins, matrix metalloproteinases, and integrins36,37. It is well known that TGFβ/Smad signalling negatively regulates the expression of miR-29s and that DPP-4 inhibitors are able to normalize these changes9,15. The present study shows that linagliptin/telmisartan-induced restoration of renal miR-29s expression was accompanied with a decrease in direct target mRNA expression of ECM genes such as COL1A1, COL1A2, COL3A1, COL4A1, COL11A1, FBN1 and MMP2. A similar relation was observed between linagliptin and telmisartan induced changes in miR-29b/c level and collagen type III protein expression (not significant) (Fig. S4). Moreover, normalized miR-29s levels are associated with decreased expression of TGFB2 and BMP1.

In the non-diabetic 5/6 Nx model, CKD progression is more pronounced in 5/6 Nx rats compared to 5/6 Nx mice. This is characterized by a marked increase in renal interstitial fibrosis and markers for glomerular filtration rate, such as cystatin C, in 5/6 Nx rats compared to 5/6 Nx mice. Furthermore, renal gene expression representing markers of renal fibrosis and inflammation, such as COL1A1, COL1A3, TIMP1 and TGF-β, are significantly increased in 5/6 Nx rats compared to 5/6 Nx mice23,38. Overall, it seems that mice tolerate the 5/6 Nx surgery better than rats. 5/6 Nx surgery results in only moderate effects on both mRNA and miRNA expression in mice, which might contribute to the different impact in treatment effects of linagliptin between mice and rats. Moreover, the kidney fibrosis phenotype in mice is largely dependent upon the strain specificity39. In addition, Srivastava et al.34 observed discrepancies regarding miR-29s expression level between diabetic mice (STZ treated CD-1 mice) and less fibrotic diabetic 129 Sv mice.

Recent studies revealed that miR-29s levels in urinary exosomes are significantly decreased in CKD patients14 and in patients with lupus nephritis22. Our study demonstrates that linagliptin/telmisartan induced restoration of miR-29c levels can be demonstrated in urinary exosomes highlighting the importance of urinary exosomes as potential biomarkers for disease progression and monitoring of treatment effects. Moreover, miR-29c levels in urinary exosomes correlate with renal function (albuminuria). One prospective, randomized, controlled study including urine sampling has been completed recently (the MARLINA-T2D trial; Efficacy, Safety and Modification of Albuminuria in Type 2 Diabetes Subjects with Renal Disease with LINAgliptin) to investigate potential short-term albuminuria-lowering effects of linagliptin40. In the MARLINA study, individuals at early stages of diabetic kidney disease were included and it was demonstrated that linagliptin significantly improved glycemic control but did not significantly lower albuminuria; in addition, there was no significant change in eGFR40. Furthermore, detection of clinically relevant renal effects of linagliptin may require longer treatment. The long-term effects of linagliptin renal outcomes were assessed in the CARMELINA (NCT01897532) study. Urine samples obtained from this study could be used to further substantiate the current findings. The CARMELINA study demonstrated positive effects of linagliptin on progression of albuminuria category - change from non-albuminuria to micro/macroalbuminuria or change from microalbuminuria to macroalbuminuria (P = 0.003)41,42. Our latest study indicated that the renoprotective effects of linagliptin cannot solely be attributed to the GLP-1/GLP-1R pathway, highlighting the importance of other signaling pathways influenced by DPP-4 inhibition such as collagen I homeostasis, HNRNPA1, YB-1, thymosin β4 and TGF-β136.

Our study showed that the linagliptin and telmisartan-induced restorative effects on miR-29c expression were reflected in urinary exosomes, suggesting that miRNA profiling of urinary exosomes might be used as a useful biomarker for addressing effects during CKD progression and treatment with drugs, since urinary miRNAs are easily to access. This might be of particular impact for human studies, because in human studies kidney tissue to analyze treatment effects are usually not available.