Linagliptin-Mediated DPP-4 Inhibition Ameliorates Kidney Fibrosis in Streptozotocin-Induced Diabetic Mice by Inhibiting Endothelial-to-Mesenchymal Transition in a Therapeutic Regimen

Diabetic nephropathy is a leading cause of kidney disease, which progresses into end-stage renal disease, requiring kidney replacement therapy (1,2). Glycemic control could be essential for therapies combating diabetic nephropathy, although normalizing the blood glucose levels in such patients with appropriate monitoring is challenging (2). Therefore, to prevent/retard diabetic nephropathy, in addition to achieving proper glycemic control, strategies that are not directly related with blood glucose normalization are required.

Fibrosis in the kidney is the final common pathway of progressive kidney diseases and results in the destruction of the kidney structure and the deterioration of the kidney filtration function (3–8). Kidney fibrosis is caused by prolonged injury associated with the dysregulation of the normal wound healing process and an excess accumulation of extracellular matrix. Kidney fibroblasts play an important role in this fibrotic process, but the origin of the fibroblasts remains unclear and has become the focus of intense debate (2,9). A significant heterogeneity of matrix-producing fibroblasts is thought to exist (2), and diverse origins for fibroblasts have been described, such as residential fibroblasts or pericytes, epithelial-to-mesenchymal transition, and endothelial-to-mesenchymal transition (EndMT) (2). Among these diverse origins of matrix-producing fibroblasts, EndMT seems to be an important origin of myofibroblasts or activated fibroblasts (9).

Dipeptidyl peptidase-4 (DPP-4) inhibitors enhance the activity of endogenous glucagon-like peptide-1 and glucose-dependent insulinotropic polypeptide (10), which have emerged as important prandial stimulators of insulin secretion and have many physiological actions (10,11). Additionally, DPP-4 is distributed throughout the body and cleaves numerous substrates other than incretin hormones (10). Aside from their glucose-lowering action, DPP-4 inhibitors are also associated with potentially organ protective effects due to their diverse, widely distributed, and pleiotropic action (12). Indeed, the kidney is where DPP-4 is expressed at the highest level per organ weight (13). Interestingly, DPP-4 has been associated with cell survival signaling and extracellular matrix remodelings (14–16).

Linagliptin, a new DPP-4 inhibitor, is mainly excreted via the bile; therefore, this drug can theoretically be prescribed for patients with renal dysfunction without adjusting the dosage (17,18). In this study, we tested whether the DPP-4 inhibitor linagliptin could exert its therapeutic benefits in mice with kidney fibrosis associated with type 1 diabetes.

A rat polyclonal anti-mouse CD31 antibody was purchased from emfret ANALYTICS (Eibelstadt, Germany). Mouse monoclonal anti-human CD31 and goat polyclonal anti-mouse DPP-4 antibodies (for tissue labeling) were purchased from R&D Systems (Minneapolis, MN). A rabbit polyclonal anti-αSMA antibody was purchased from GeneTex (Irvine, CA). A rabbit polyclonal anti-SM22α antibody and a monoclonal antibody for VE-cadherin were obtained from Novus Biologicals (Littleton, CO). A polyclonal antibody for FSP1 (S100A4) was purchased from Covance (Princeton, NJ). A goat polyclonal anti–DPP-4 antibody (for Western blotting in human cells) and a rabbit polyclonal anti-GAPDH antibody were obtained from Sigma-Aldrich (St. Louis, MO). Fluorescence-, Alexa Fluor 647–, and Rhodamine-conjugated secondary antibodies were obtained from Jackson ImmunoResearch (West Grove, PA). A horseradish peroxidase–conjugated secondary antibody was purchased from Cell Signaling Technology (Danvers, MA). TGF-β2 was purchased from PeproTech (Rocky Hill, NJ). Apoptosis was detected with an Annexin V assay kit, which was obtained from Clontech Laboratories (Mountain View, CA). DPP-4 activity was monitored by DPP-4 assay kit (BioVision, Milpitas, CA).

Eight-week-old male CD-1 mice (Sankyo Laboratory Service, Tokyo, Japan) were used in all of the diabetic experiments. The mice were injected intraperitoneally with streptozotocin (STZ; 200 mg/kg). The induction of diabetes was confirmed as a blood glucose level >16 mmol/L 2 weeks after STZ injection. By 20 weeks after the induction of diabetes, the diabetic mice were divided into two groups (linagliptin [5 mg/kg body weight (BW)/day in drinking water] and untreated). Dose of linagliptin was decided based on the dose-dependent effects of this drug between 3 and 30 mg/kg BW/day (19). Linagliptin was diluted directly in drinking water. All mice were killed 24 weeks after the induction of diabetes. Linagliptin was provided by Boehringer Ingelheim (Ingelheim, Germany) with a material transfer agreement. Blood pressure was monitored by the tail cuff method with BP-98A (Softron Co., Beijing, China).

Frozen sections (5 μm) were used for the detection of in vivo EndMT. Cells undergoing EndMT were identified by double-positive labeling with CD31-αSMA or CD31-FSP1. The immunolabeled sections were analyzed by fluorescence microscopy (Biozero; Keyence, Osaka, Japan). For each mouse, ×300 magnification pictures were obtained from six different areas, and quantification was performed.

The glomerular surface area was calculated in 10 glomeruli per mouse using ImageJ software. To evaluate the mesangial matrix area (%), we used a point counting method. We analyzed 10 Periodic acid-Schiff (PAS)-stained glomeruli from each mouse on a digital microscope screen grid containing 540 (27 × 20) points in Adobe Photoshop Element 6.0. The number of grid points in the mesangial area (both matrix and cells) was divided by the total number of points in the glomerulus to obtain the percentage of relative mesangial matrix area in a given glomerulus. Masson trichrome stain (MTS) were imaged and analyzed by ImageJ software, and fibrotic areas were quantified. In each mouse, six pictures (original magnification ×100) were evaluated.

Deparaffinized (2 min in xylene, four times; 1 min in 100% ethanol, twice; 30 s in 95% ethanol; 45 s in 70% ethanol; and 1 min in distilled water) mouse kidney sections were used for DPP-4 labeling. Immunohistochemistry was performed with a Vectastain ABC Kit (Vector Laboratories, Burlingame, CA). The DPP-4 primary antibody was diluted 1:100. In negative controls, the primary antibody was omitted and replaced with blocking solution.

Human dermal microvascular endothelial cells (HMVECs; Lonza, Basel, Switzerland) cultured in endothelial basal medium (EBM) supplemented with endothelial growth medium (EGM)-2 kit medium were used in this experiment. When the HMVECs on the adhesion reagent (Kurabo Bio-Medical, Osaka, Japan) reached 70% confluence, 5 ng/mL recombinant human TGF-β2 for 48 h was placed in the experimental medium (HuMedia-MVG in serum-free RPMI at a 1:3 ratio) with or without linagliptin (100 nmol/L) preincubation for 2 h. In the control well, vehicle (DMSO) was added (3 × 10⁻⁵ dilution of DMSO in final concentration). The protein lysate was harvested for Western blot analysis.

The passaged HMVECs were placed in six-well plates and cultured with EBM supplemented with EGM-2 until reaching 70–80% confluence, and then the cells treated with TGF-β2 (5 ng/mL) in the presence or absence of linagliptin (100 nmol/L) were incubated with a medium containing HuMedia MVG and RPMI-1640 (1:3). At the same time, the control group was incubated in the same medium without TGF-β2 or linagliptin. In the control well, vehicle (DMSO) was added (3 × 10⁻⁵ dilution of DMSO in final concentration). Using a pipette tip at an angle of ∼30°, each well received a straight scratch, simulating a wound. After 24 and 48 h had passed, the number of cells that migrated into the wounded area was counted under a light microscope. Six different areas were evaluated in each group, and the experiment was repeated twice with similar results.

The bottom side of the migration chamber (Cell Culture Insert; BD Falcon, San Jose, CA) was coated with Matrigel (BD), and 1,000 HMVECs were passaged in the upper migration chamber. Twenty-four hours after passage, the medium was changed to the experimental medium (1:3 HuMedia-MVG:RPMI1640) in both the upper and the bottom wells. Subsequently, cells were exposed to TGF-β2 in the presence or absence of linagliptin (100 nmol/L), while the control group was incubated with the same medium, lacking TGF-β2 and linagliptin. In the control well, vehicle (DMSO) was added (3 × 10⁻⁵ dilution of DMSO in final concentration). After 48 h, the cells were washed with PBS, followed by fixation with formaldehyde (3.7% in PBS) at room temperature for 2 min. After washing twice with PBS, the cells were permeabilized with 100% methanol for 20 min at room temperature. Then, the cells were washed twice with PBS and stained with hematoxylin-eosin. After scraping off the nonmigratory cells (upper well) with a cotton swab, the number of migrated cells was counted under a light microscope. Six different areas were evaluated in each group, and the experiment was repeated twice with similar results.

The protein lysates were denatured by boiling in SDS sample buffer at 100°C for 5 min and then centrifuged (17,000g for 10 min at 4°C); subsequently, the supernatant was separated on SDS-polyacrylamide gels. Separated protein lysates were blotted onto polyvinylidene fluoride (PVDF) membranes (Pall Corporation, Pensacola, FL) by the semidry method. After blocking with TBS-T (Tris-buffered saline containing 0.1% Tween 20) containing 5% nonfat dry milk or 5% BSA, the membranes were incubated with the primary antibodies of the target molecules (1:1,000 for all primary antibodies) in TBS-T containing 5% BSA at 4°C overnight. The membranes were washed three times and incubated with 1:2,000 diluted horseradish peroxidase–conjugated secondary antibodies (Cell Signaling Technology) for 1 h at room temperature. The immunoreactive bands were visualized with an enhanced chemiluminescence (ECL) detection system (Pierce Biotechnology, Rockford, IL) by ImageQuant LAS 4000 (GE Healthcare Life Sciences, Uppsala, Sweden).

The total RNA was isolated using a commercially available kit (RNeasy Mini Kit; QIAGEN, Hilden, Germany). The concentration of RNA was quantified by photometry at 260/280 nm, and the quality of the RNA was determined by the ratio of the 18S/28S ribosomal band intensities in an ethidium bromide–containing 1% agarose gel after electrophoresis. The sense cDNA was prepared using an Ambion WT Expression Kit (Ambion, Austin, TX), and target hybridizations were performed using a Mouse Gene 1.0 ST Array (Affymetrix, Santa Clara, CA), according to the manufacturer’s instructions. Hybridization was performed for 17 h at 45°C in a GeneChip Hybridization Oven 640 (Affymetrix). After washing and staining in a GeneChip Fluidics Station 450, hybridized cDNAs were detected using the GeneChip Scanner 3000. The digitalized image data were processed using the GeneChip Operating Software version 1.4. Because replicate assays were not performed, the signal intensities of the selected genes that were upregulated or downregulated by at least twofold compared with a control group were extracted using GeneSpring GX software package version 12.5 (Agilent Technologies, Santa Clara, CA). After hierarchical clustering, the results were illustrated as a heat map. Ingenuity Pathway Analysis (Ingenuity Systems, Inc., Redwood City, CA) was used to select the specific function-related genes.

Total RNA was isolated using the miRNeasy Kit (QIAGEN) according to the manufacturer's instructions. Quality-confirmed total RNA samples were assayed and qualified in duplicate using the microRNA microarray. The input for the Agilent microRNA labeling system was 100 ng total RNA. Dephosphorylated and denatured total RNA was labeled with cyanine 3-pCp and subsequently hybridized to the Agilent mouse microRNA microarray release version 15 using the microRNA Complete Labeling and Hyb Kit (Agilent). After hybridization for 20 h, the slides were washed with Gene Expression Wash Buffer Kit (Agilent) and measured using an Agilent Scanner G2565BA. Agilent Feature Extraction Software version 9.5.1 and GeneSpring GX software version 12.5 (Agilent) were used for data processing, analysis, and monitoring.

Frozen kidney tissues (one each from a control, diabetic, and linagliptin-treated diabetic mouse; samples were kept at −70°C) were first placed on the RNA later-ICE (Life Technologies) for 16 h at −20°C before the subsequent homogenization process to avoid RNA degradation while extracting high-quality microRNA. MicroRNA was extracted using miRNeasy Mini Kit (QIAGEN) according to the manufacturer’s instructions for homogenized samples. The complementary DNA was generated by a miScript II RT Kit (QIAGEN) using the hiSpec buffer method. MicroRNA expression was quantified using miScript SYBR Green PCR Kit (QIAGEN) using 3 ng of complementary DNA. The primers to quantify Mm_miR-29a, Mm_miR-29b, and Mm_miR-29c were the miScript primer assays predesigned by QIAGEN. The mature microRNA sequences were 5′-UAGCACCAUCUGAAAUCGGUUA for Mm_miR-29a, 5′-UAGCACCAUUUGAAAUCAGUGUU for Mm_miR-29b, and 5′-UAGCACCAUUUGAAAUCGGUUA for Mm_miR-29c. All experiments were performed in triplicate, and Hs_RNU6-2_1 (QIAGEN) was used as an internal control.

For the transfection studies, HMVECs, which were maintained in EBM supplemented with EGM-2, were passaged in six-well plates with nonproliferative medium (HuMedia-MVG and RPMI at a ratio of 1:3). The HMVECs were transfected with 100 nmol/L of antagomiR for microRNA 29a, microRNA 29c, microRNA 29a+c (FASMAC, Kanagawa, Japan), microRNA 29b inhibitor (QIAGEN), or mimetics for microRNA 29s (29a-3p, UAGCACCAUCUGAAAUCGGUUA; 29b-3p, UAGCACCAUUUGAAAUCAGUGUU; 29c-3p, UAGCACCAUUUGAAAUCGGUUA) using Lipofectamine 2000 transfection reagent (Invitrogen, Carlsbad, CA) according to the manufacturer’s instructions.

The cells were incubated for 6 h with Lipofectamine and the anti-microRNA complex in antibiotic-free medium, after which the medium was replaced with fresh medium before the cells were incubated for another 48 h. Upon the termination of the incubation, the cells were scraped using RIPA buffer (with the addition of phenylmethylsulfonyl fluoride, sodium vanadate, and protease inhibitor) to assess DPP-4 protein expression (Sigma-Aldrich) using the Western blot technique.

For the luciferase assay to analyze the activity of 3′ untranslated region (UTR) in human DPP-4, we cloned the fragment of human DPP-4 3′ UTR sequence by PCR using the primer set (forward, ATAGAGCTCAATAGCTAGCAGCACAGCACACCAAC; reverse, ATATCTAGA GTGTCCATATGCCAGTGCGGTTTAGG) and BAC clone human RP11 178A14 as template. Both purified PCR fragment and pmirGLO Dual-Luciferase microRNA Target Expression Vector (Promega) were enzyme digested (SacI and XbaI), purified, and ligated (Thermo Fisher Scientific). The sequence of DPP-4 3′ UTR was confirmed, and amplified vector DNA (300 ng/well in 12-well plate) was transfected into HMVECs. In the presence of control microRNA, mimetic, antagomiR, or inhibitor of 29s (75 nmol/L in final concentration), transcriptional activity was evaluated by Dual-Luciferase Reporter Assay System (Promega) in triplicate samples.

The data are expressed as the mean ± SEM. The nonparametric Mann-Whitney U test was used to determine significance, which was defined as P < 0.05, if not specifically mentioned. GraphPad Prism software (version 5.0f) was used for the statistical analysis.

We used a fibrotic diabetic kidney disease model, STZ-induced diabetes in male CD-1 mice. The immunohistochemistry analysis revealed robust DPP-4 immunolabeling in the glomerular basement membrane, tubules, and peritubular vascular cells in the kidney of STZ-induced diabetic CD-1 mice when compared with control CD-1 mice (Fig. 1A and B). Such peritubular vascular cells were likely endothelial cells upon evaluation by immunofluorescence microscopy (Fig. 1C–E). Western blot analysis using whole kidney revealed that in the kidneys of diabetic mice, the DPP-4 protein levels were upregulated compared with control kidneys (Fig. 1F and G). As Sugimoto et al. (20) elegantly showed, STZ-induced diabetic CD-1 mice exhibit kidney fibrosis, with both glomerulosclerosis and tubulointerstitial fibrosis occurring ∼16 weeks after the onset of diabetes (Supplementary Fig. 1), and at 24 weeks after the initiation of diabetes, diabetic mice exhibited severe fibrosis when compared with control mice (Fig. 2A, B, D, and E). Linagliptin-treated diabetic mice exhibited restored normal kidney structures (Fig. 2C and F). Morphometric analysis of the kidneys revealed that diabetic mice displayed significantly enlarged glomeruli, mesangial expansion, and relative areas of Masson trichrome–positive interstitial fibrosis (Fig. 2G–I), whereas linagliptin restored normal kidney histology and normal architecture (Fig. 2G–I). Urine albumin levels were elevated in STZ mice, and linagliptin treatment suppressed the trend of the urine albumin levels (Fig. 2J) as observed in human clinical trial (21). These antifibrotic effects of linagliptin in diabetic mice are associated with a suppressed trend of TGF-β1 and DPP-4, and significant suppression of TGF-β2 protein levels in the kidney (Fig. 2K–N). Linagliptin inhibited DPP-4 mRNA (Fig. 2O) and diabetes-enhanced DPP-4 activity in kidney and plasma of diabetic mice (Fig. 2P and Q).

When compared with control mice, diabetic CD-1 mice exhibited lower blood pressure, lighter BW, higher blood glucose, and increased kidney and liver weight (Fig. 3A–E). Linagliptin treatment in STZ-induced diabetic CD-1 mice caused no alteration in blood pressure, body weight, blood glucose level, or organ weight (for kidney, liver, and heart) when compared with untreated diabetic mice (Fig. 3A–E). The heart weight was lighter in all diabetic mice but showed insignificant changes in all groups analyzed (Fig. 3F).

Kidney fibroblasts play essential roles in kidney fibrosis and originate from diverse sources (2,9,22). In our analysis, we analyzed EndMT, a recently described important source of kidney fibroblasts (2,9,23). When EndMT was analyzed by quantifying cells that coexpressed endothelial marker CD31 and a mesenchymal marker, either α-smooth muscle actin (αSMA) or FSP1, diabetic CD-1 mice exhibited a significantly increased number of cells in the process of the EndMT when compared with control kidneys (Fig. 4A, B, D–F, and H). Linagliptin-treated mice exhibited significantly fewer cells undergoing EndMT in the kidney when compared with untreated diabetic mice (Fig. 4C, D, G, and H).

In cultured endothelial cells, TGF-β2 induced the suppression of CD31 with concomitant upregulation of SM22α in HMVECs, suggesting that TGF-β2 induces EndMT (Fig. 5A–C); TGF-β2–induced EndMT was inhibited by linagliptin pretreatment (Fig. 5A–C). Wound healing cell invasion assays revealed that TGF-β2 also induced the migration of fibroblast-like EndMT cells (Fig. 5D, E, J, G, and H), while linagliptin inhibited the invasion of those cells (Fig. 5F, I, and J). The Boyden chamber cell migration assay also revealed that linagliptin inhibited the endothelial cells’ transmigration through Matrigel (Fig. 5K–N). When analyzing the molecular mechanisms of TGF-β2–induced EndMT effects, we found that linagliptin inhibited TGF-β2–induced Smad3 phosphorylation in endothelial cells (Fig. 5O). Linagliptin inhibited TGF-β2–induced protein expression, mRNA expression, and activity of DPP-4 in endothelial cells (Fig. 5P–R). Linagliptin and generic DPP-4 inhibitor KR62436 inhibited TGF-β2–induced endothelial cell apoptosis (Supplementary Fig. 2).

A heat map indicated the differences among controls, diabetic mice, and diabetic mice treated with linagliptin (Supplementary Fig. 3A). The diabetes-associated changes in the expression of some genes were restored by linagliptin treatment (Supplementary Fig. 3A). To focus on fibrosis, we selected fibrosis-associated genes based in the Ingenuity Pathways Analysis database and plotted them in a scattered format (Supplementary Fig. 3B). Compared with controls, 13 genes were upregulated and 2 genes were downregulated in the diabetic mouse (Supplementary Fig. 3C). However, the expression of these genes tended to be reversed into normal levels after linagliptin treatment.

Finally, to identify the underlying mechanisms of how DPP-4 is increased in the diabetic kidney, we analyzed the microRNA profiles of the animals and found that the microRNA 29 family tended to be suppressed in diabetic mouse kidneys when compared with control kidneys (although no significant difference by Welch test yet) (Fig. 6A). According to the prediction of microRNA targets by TargetScan (http://www.targetscan.org/vert_60/), DPP-4 may be regulated by the microRNA 29 family. Quantitative analysis revealed that microRNAs 29a, -b, and -c were suppressed in the diabetic kidney when compared with control kidneys, and linagliptin restored such diabetes-suppressed microRNA 29s (Fig. 6B–D). Similarly TGF-β2–suppressed microRNA 29s were restored by linagliptin in in vitro analysis (Supplementary Fig. 4). When an antagomiR for microRNA 29a was transfected into endothelial cells, DPP-4 protein levels were indeed increased (Fig. 6E and F), whereas microRNA 29c antagomiR transfection did not change the DPP-4 level (Fig. 6E and F). The cotransfection of antagomiRs for microRNAs 29a and -c resulted in persistently higher DPP-4 protein levels in the endothelial cells compared with controls (Fig. 6E and F). In addition, the inhibitor for microRNA 29b induced DPP-4 protein expression in HMVECs (Fig. 6G and H), suggesting that microRNAs, specifically the microRNA 29 family, regulate DPP-4 in the diabetic kidney. TGF-β2–enhanced DPP-4 protein expression was significantly suppressed by microRNA 29b mimetics, and some trend of suppression was found in microRNA 29a and -c mimetic–transfected cells (Fig. 6I and J). TGF-β2–induced EndMT and endothelial cell migration was largely inhibited by microRNA 29 mimetic transfection (Fig. 6I and K–T). In contrast, antagomiRs and inhibitor for microRNA 29s induced EndMT phenotype and migration (Supplementary Fig. 5). TGF-β2–stimulated luciferase activity of pmirGLO Dual-Luciferase microRNA Target Expression Vector containing 3′ UTR fragment of DPP-4, where a microRNA 29 binding site was involved, was significantly suppressed by microRNA 29 mimetics (Fig. 6U). In contrast, antagomiR or inhibitor of microRNA 29s significantly induced DPP-4 3′ UTR luciferase activity (Fig. 6V).

Both the inhibition of kidney fibrosis and the restoration of normal kidney structure are fundamental processes to research for developing therapies to combat progressive chronic kidney disease, including diabetic nephropathy. Although kidney fibroblasts have been implicated in the pathogenesis of kidney fibrosis, it would be challenging to inoculate only kidney fibroblasts as therapeutic targets. In our analysis, DPP-4 inhibition by linagliptin in the diabetic kidney seems to be a powerful therapy that inhibits the fibroblast-activating process.

We found that DPP-4 protein expression was increased in the whole kidney lysate, endothelial cells, tubules, and glomeruli of STZ-induced diabetic kidneys from CD-1 mice. Both the protein expression and activity levels of DPP-4 are higher in kidneys of high fat–fed, STZ-injected mice (24). Liu et al. (25) reported that the administration of the DPP-4 inhibitor vildagliptin for 24 weeks prevented kidney damage in STZ-induced diabetic male Sprague-Dawley rats. In our analysis, we used fibrotic STZ-induced diabetic CD-1 mice and tested whether intervention with linagliptin starting 20 weeks after the induction of diabetes could rescue fibrotic kidneys. We found that introducing this intervention to such fibrotic kidneys significantly ameliorated kidney fibrosis without altering physiological parameters, suggesting that increased DPP-4 in fibrotic diabetic kidneys is a therapeutically valuable target. Our microarray analysis clearly demonstrated that diabetic kidneys exhibited profibrotic signaling and that linagliptin restored a normal profile in the diabetic kidney.

We have not yet realized an entire series of DPP-4–mediated, kidney-damaging signals in our models, but we believe that signaling associated with endothelial cell survival and EndMT is involved. Linagliptin inhibited both the matrix- and fibroblast-generating pathways associated with EndMT and fibroblast-like EndMT cell invasion. Such enhanced DPP-4 activity is associated with matrix metalloproteinases, which are responsible for tissue remodeling (26). Furthermore, Takahashi et al. (27) recently showed that DPP-4 inhibition by vildagliptin ameliorates the heart failure and collagen deposition associated with inhibiting the TGF-β–Smad signaling pathway. We found that linagliptin inhibited TGF-β2–induced Smad3 phosphorylation.

Our analysis reveals that the microRNA 29 family is one of the potential molecular regulators of DPP-4 in the kidney and endothelial cells as well. The microRNA 29 family protects organs from fibrotic damage (28–31); therefore, DPP-4 inhibition and the subsequent inhibition of the fibroblast activation pathway may be involved in the antifibrotic effects of the microRNA 29 family. DPP-4 has been implicated in fibrogenic pathologies (32–36). Regarding this, DPP-4 inhibition by linagliptin increased microRNA 29s both in vivo and in vitro. MicroRNA 29 suppressions in either diabetic kidneys or TGF-β2–stimulated endothelial cells are associated with EndMT and induction of DPP-4. The inhibition of each microRNA 29 can induce the EndMT feature in HMVECs and migration; the role of each microRNA 29 in DPP-4 level regulation is somewhat complicated. Since microRNA 29a inhibition results in the strong induction of DPP-4 3′ UTR–luciferase activity, in basal conditions microRNA 29a likely emerges as the main regulator of DPP-4 3′ UTR; TGF-β2–induced DPP-4 protein level and cell migration were most efficiently inhibited by microRNA 29b, suggesting that each microRNA 29 could play distinct roles and cooperate in the homeostasis of cells in a context-dependent manner. Also, our data indicate that TGF-β2/DPP-4/miR29s display cross-talk mechanisms in the onset of kidney fibrosis, and linagliptin-mediated inhibition of DPP-4 would diminish profibrotic signaling cross-talk in the kidney. Among human fibrotic diseases, patients with hepatitis C viral infections, the condition associated with microRNA 29 suppression, also exhibit high DPP-4 activity; such DPP-4 activity could be a potential target for combating such liver diseases (37). Further study is required to determine whether DPP-4 activity is greater in type 1 or type 2 diabetic kidney diseases, or in other fibrotic chronic and acute diseases, including kidney disease, and their association with microRNA 29s levels in humans.

In our analysis, linagliptin inhibited kidney fibrosis and restored normal kidney structure. Although convincing, there are several limitations. First, the concentration we used was very high (5 mg/kg BW) when compared with that used in the clinic (∼100 μg/kg BW). Second, according to our data, we believe that DPP-4 inhibition by any DPP-4 inhibitor can ameliorate kidney fibrosis. However, it is not yet clear whether other DPP-4 inhibitors can efficiently suppress the profibrotic program in the kidney due to differences in the mechanisms/metabolism of each drug. Third, we focused on EndMT as the origin of activated fibroblasts. However, such origins of fibroblasts in kidney fibrosis are still controversial and the focus of intense debate (2,9), and it could be possible that diverse origins of the kidney fibroblast–activating pathway (38) were affected by linagliptin. Finally, the analysis for distinct roles of each microRNA 29 in the regulation of DPP-4 and kidney fibrosis would require further investigation.

Diabetic kidney disease represents a serious health problem worldwide and can develop into end-stage renal disease, which requires kidney replacement therapy. Progressive kidney fibrosis determines residual kidney function, and the restoration of the normal architecture to the fibrotic kidney would constitute a fundamental therapy. We reported here that the antifibrotic effects of linagliptin are beneficial for diabetic kidney disease, regardless of the blood glucose levels, via the suppression of the activated fibroblast-generating EndMT pathway at least in part. Linagliptin may be safe for use among kidney disease patients, given its ability to be exclusively eliminated via the bile; this specificity suggests that this DPP-4 inhibitor has potential utility for therapeutic use in combating kidney fibrosis in diabetes.

Funding and Duality of Interest. This work was partially supported by grants from the Japan Society for the Promotion of Science to K.K. (23790381), M.Ka. (24790329), T.N. (24659264), M.Ki. (24591218), and D.K. (25282028 and 25670414), research grants from the Japan Research Foundation for Clinical Pharmacology to K.K. (2011), and the Takeda visionally research grant to K.K. (2013). This work was partially supported by a Grant for Promoted Research awarded to K.K. (S2011-1 and S2012-5) and a Grant for Collaborative Research awarded to D.K. (C2011-4 and C2012-1) from Kanazawa Medical University. K.K. was also supported by several foundational grants, including grants from the Daiichi Sankyo Foundation of Life Science, the Ono Medical Research Foundation, the Novartis Foundation (Japan) for the Promotion of Science, the Takeda Science Foundation, and the Banyu Foundation. S.P.S. is supported by the Japanese Government Ministry of Education, Culture, Sports, Science and Technology Fellowship Program. S.S. and J.H. are supported by foreign scholar grants from Kanazawa Medical University.

The authors thank Boehringer Ingelheim for providing the linagliptin with a material transfer agreement. K.K. and D.K. received lecture fees from Boehringer Ingelheim and Eli Lilly and Company. Both Boehringer Ingelheim and Eli Lilly and Company donated to Kanazawa Medical University and were not directly associated with this project. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. K.K. proposed the original idea and design of the experiments, supervised experiments, provided intellectual input, and wrote the manuscript. S.S. performed the quantification of the histological analysis and most of the in vitro analyses and DPP-4 activity measurements. M.Ka. took care of the mice, performed the histological analysis, provided intellectual advice, and edited the manuscript. J.H. performed some of the animal experiments and some of the in vitro analysis. T.N. performed some of the in vitro analysis regarding EndMT. Y.N. and Y.I. performed the microRNA and mRNA array analysis. M.Ki. participated in the discussions. S.P.S. performed in vivo and in vitro quantitative PCR for mRNA and microRNA, Western blot, transfection, cloning of DPP-4 3′ UTR vector, and quantification. D.K. provided intellectual input. S.S. and S.P.S. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the American Society of Nephrology Renal Week, Atlanta, GA, 9 November 2013.