The dipeptidyl peptidase 4 inhibitor linagliptin ameliorates renal injury and accelerated resolution in a rat model of crescentic nephritis

Dipeptidyl peptidase 4 (DPP‐4) inhibitors are a class of oral glucose‐lowering drugs used in the treatment of type 2 diabetes. In a pilot study using human kidney biopsies, we observed high DPP‐4 expression in early crescent formation. This glomerular lesion occurs in different kidney diseases and is a hallmark in the pathogenesis of renal dysfunction. Therefore, we investigated the potential involvement of DPP‐4 in the pathogenesis of nephritis induced by anti‐glomerular basement membrane (GBM) antibody in rats.

DPP-4 functions as an exopeptidase to cleave or inactivate several physiological substrates and also occurs as a membranous glycoprotein along with a circulating variant in blood (Gorrell, Gysbers, & McCaughan, 2001). By inactivating the incretins, glucagon-like peptide 1 (GLP-1) and glucose dependent insulinotropic peptide (GIP), DPP-4 reduces insulin secretion and thus plays a major role in glycaemic regulation (Neumiller, Wood, & Campbell, 2010). In this respect, different DPP-4 inhibitors are available for the treatment of type 2 diabetes mellitus having different binding and pharmacokinetic characteristics (Schnapp, Klein, Hoevels, Bakker, & Nar, 2016). Interestingly, DPP-4 is found in various tissues and cells with the kidney having high level of enzymatic activity (Wang et al., 2014).  inhibition is currently restricted to the treatment of diabetes mellitus but is potentially beneficial for renal diseases. The CARMELINA clinical trial recently demonstrated reduced progression of albuminuria in diabetes patients treated with linagliptin but no improvement on hard renal endpoints (Hanssen & Jandeleit-Dahm, 2019). However, own unpublished data has shown reduced glomerular DPP-4 activity in renal biopsies from diabetic patients compared to non-diabetic healthy controls, while DPP-4 activity was up-regulated in other renal diseases. The role of renal DPP-4 in the pathogenesis of glomerular diseases and therapeutic effects of DPP-4 inhibition, beyond regulation of the glucose metabolism, is still unclear. DPP-4 cleaves a wide range of other substrates including peptide hormones such as stromal-derived factor 1 alpha (SDF-1α; CXCL12 α) or monocyte chemoattractant protein 1 (MCP-1 or CCL2) (de Meester, Lambeir, Proost, & Scharpe, 2003). Furthermore, DPP-4 can serve as a surface receptor and co-stimulatory protein involved in T-cell activation mediated by interaction with caveolin-1 (Ohnuma et al., 2007;Ohnuma et al., 2015). Therefore, it is conceivable that DPP-4 is directly or indirectly involved in the pathogenesis of crescent formation. While other DPP-4 inhibitors are cleared renally, linagliptin can be given during renal impairment because its clearance primarily occurs non-renally via a hepatobiliary pathway (Fuchs, Binder, & Greischel, 2009;Fuchs, Tillement, Urien, Greischel, & Roth, 2009;Heise et al., 2009;Huttner, Graefe-Mody, Withopf, Ring, & Dugi, 2008). Thus, it can be used in patients with renal impairment and does not require dose adjustment Gallwitz, 2013). Using a rat model for anti-glomerular basement membrane nephritis and different treatment regimens for the DPP-4 inhibitor linagliptin, we followed the hypothesis that DPP-4 is directly or indirectly involved in the pathogenesis of glomerular basement membrane nephritis, especially in crescent formation.
2 | METHODS 2.1 | Human renal tissue specimens and evaluation of glomerular  In our study, 158 formalin-fixed paraffin-embedded specimens of archival kidney biopsies (from the Department of Nephropathology, Friedrich-Alexander University Erlangen-Nürnberg) were used to evaluate glomerular DPP-4 expression in different renal diseases using immunohistochemistry. In our pilot study, we included biopsies from hypertensive and non-hypertensive patients with clinically documented blood pressure. In addition, biopsies from transplanted kidneys without rejection (n = 16) and macroscopically normal portions of kidneys surgically excised due to the presence of a localized neoplasm (control, n = 11) were included as healthy controls. The analysis of archived renal biopsies was approved by the local ethics committee (reference number 4415).

What is already known
• Linagliptin is a useful therapy for diabetic patients.

What this study adds
• Linagliptin ameliorates crescentic glomerulonephritis by accelerated resolution of crescents and anti-fibrotic effects.
What is the clinical significance • Linagliptin can improve disease progression in non-diabetic renal disease.
2.2 | Glomerular basement membrane nephritis model in the rat and treatment with linagliptin In the passively induced glomerular basement membrane nephritis model, proteinuria was observed from Day 3 onwards and reaching a plateau on Day 5 (Kohda et al., 2004). Anti-glomerular basement membrane antibodies bind to glomerular basement membrane, inducing glomerular injury including severe endocapillary hypercellularity and extracapillary changes, such as capsular adhesion and crescent formation via complement or Fc-receptor-mediated processes. While glomerular CD8-positive T-cells are only barely increased after disease induction, a marked glomerular influx of macrophages can be observed early on Day 1 and further increases during the first 10 days (Kado et al., 2006). To investigate short-term versus long-term effects of linagliptin on the pathogenesis of glomerular basement membrane nephritis, rats were randomized into two different experiments: four groups in a short-term regimen for 14 days ( Figure 1a) and four groups in a long-term regimen for 8 weeks (Figure 1b). The group size in the animal model was calculated in the expectation that administration of the DPP-4 inhibitor would lead to a biologically relevant reduction of crescents by at least 35%. With a determined effect size of 1.26 and a first type error of 5% and a second type error of 20%, a group size of n = 11 was determined using G-Power software (Version 3.1) (Faul, Erdfelder, Lang, & Buchner, 2007). In total, 88 male Wistar Kyoto rats (WKY/NIcoCrl, Charles River, Sulzfeld, Germany) with a body weight of 200-220 g and an age of 6-7 weeks were maintained in a specific pathogen-free facility in a temperature-and lightcontrolled environment and had ad libitum access to chow and water.
The rats are kept on classical aspen wood bedding (sniff-Spezialdiäten GmbH, Soest, Germany) in type IV macrolon cages (Tecniplast Deutschland GmbH, Hohenpeißenberg, Germany) with a maximum population of three. The experimental protocol for the animal studies was approved by the German regional committee for animal care and use, which is equivalent to the US IACUC and authorized by the governmental department ("Regierung von Unterfranken" Permit number: 55.2-2532-2-324) before the animal studies were performed in strict accordance with the German welfare act and adherence to the   Kröller-Schön et al., 2012). In both regimens, rats were either treated preventively (starting with disease induction) or therapeutically (starting on Day 7 after disease induction in short-term regimen and on Week 4 in long-term regimen) (Figure 1a,b). Blood samples and urine were collected for analysis of proteinuria, DPP-4 activity and kidney function. Therefore, rats were housed in metabolic cages before starting the experiment, on Days 6 and 13 or on Weeks 3, 5 and 8 to collect urine for 23 h. On Days 1, 7 and 14 or on Weeks 3, 5 and 8 blood samples taken from tail vein under isofluran anesthesia were collected in tubes containing lithium heparin and after sedimentation for 10 min at 1,300 g plasma supernatants were kept at −20 C F I G U R E 1 Experimental design. Linagliptin effects on pathogenesis of anti-glomerular basement membrane (GBM) nephritis were investigated using an early (a) and a late regimen (b). In both regimens, rats were either treated preventively (starting with disease induction) or therapeutically (starting on Day 7 after disease induction in short-term regimen and on Week 4 in long-term regimen) (a, b). Blood samples and urine were collected for analysis of proteinuria, dipeptidyl peptidase 4 (DPP-4) activity and kidney function. Therefore, rats were housed in metabolic cages before start of the experiment on Days 6 and 13 or on Weeks 3, 5, and 8 to collect urine for 24 h. For investigation of disease progression, renal survival biopsies were taken on Day 7 in the short-term regimen and on Week 4 in the long-term regimen. Finally, rats were killed on Day 14 (a) or 8 weeks after induction of anti-GBM nephritis (b) until analysis of kidney function (urinary creatinine, serum creatinine) and DPP-4 activity. Proteinuria was measured using Bio-Rad Protein Assay (Bio-Rad Laboratories GmbH, Munich, Germany) ( Figure 1). For investigation of disease progression, renal survival biopsies were taken on Day 7 in the short-term regimen and on Week 4 in the longterm regimen. For this purpose, the animals were administered an opioid analgesic buprenorphine (Buprenovet; Bayer, Leverkusen, Germany) in a dose of 0.05 mgÁkg −1 body weight s.c. before surgery.
After anaesthesia induced with isoflurane, the left flank was shaved and disinfected with KODAN tincture forte (Schülke & Mayr GmbH, Norderstedt, Germany), the flank was opened with a 1.5 cm incision and the kidney was carefully lifted out and fixed with two sterile swabs. Then the upper pole of the kidney was resected with a scalpel and the cut surface was closed with a collagen sponge (Resorba Medical GmbH, Nürnberg, Germany). Finally, the kidney was returned to its original position, the abdominal incision was closed with an absorbable suture and a continuous seam as well as the skin was closed with single button seams. Postoperatively, activity, fluid intake, weight, coat texture and behaviour of the animals were assessed but showed no signs of post-operative pain. However, analgesia given over at least 3 days post-operatively. Finally, the rats were killed under isoflurane anaesthesia on Day 14 by bleeding ( Figure 1a supplemented with 0.2% procaine (Steigerwald Arzneimittel GmbH, Darmstadt, Germany) followed by 0.9% NaCl and the vena cava was opened at the same time.

| Isolation of glomeruli for quantitative RNA analysis
Glomeruli were isolated from rat kidneys using the graded sieving method as described previously (Wittmann et al., 2008). Total glomerular RNA was isolated using RNeasy Mini columns (Qiagen, Hilden, Germany). Reverse transcription reactions and Real-time PCR were performed using Power SYBR Green on a 7500 Fast Real time PCR system (both Applied Biosystems, Weiterstadt, Germany) according to the manufacturer's instructions. Real-time PCR data were analysed using the SDS v1.3 software (Applied Biosystems) and relative expression of target gene mRNA levels was calculated using the comparative delta Ct method (Dimmler et al., 2003).
Normalization was conducted against the endogenous 18S rRNA levels applied to the resulting relative fold changes. The list of used primers is provided in Table S1.

| Immunohistochemical staining
For all immune staining, kidneys were fixed in formalin (R. Langenbrinck, Emmendingen. Germany) or zinc fixative, embedded in paraffin and cut into sections of 2 μm. After deparaffinization, endogenous peroxidase was blocked using 3% H 2 O 2 (Merck KGaA, Darmstadt, Germany) for immunohistochemistry. Antigen retrieval was done using target retrieval solution (Dako Deutschland GmbH, Hamburg, Germany) and boiling in a pressure cooker for 2.5 min (this step could be omitted when using zinc fixative was used). After blocking for 10 min with 1% BSA (Merck KGaA) in 50 mM of Tris (Roth GmbH, Karlsruhe, Germany) pH 7.4, sections were incubated overnight at 4 C for immunohistochemical staining for using the antibodies (Table S2) diluted in 1% BSA (Merck KGaA) in 50 mM of Tris pH 7.4. Primary antibodies were detected using secondary antibodies as shown in Table S3; ABC-kit and ImmPACT-DAB were used as substrate (all from Vector Laboratories, Burlingame, CA, USA). The experimental detail provided for immunohistochemistry conforms with BJP guidelines. All immuno-related procedures involved comply with the editorial on immunoblotting and immunohistochemistry .

| Immunofluorescence multiple staining for co-localization studies
The kidney biopsies were pretreated for immunofluorescence multiple staining in the same way as for immunohistochemistry, omitting the blocking of endogenous peroxidase. Combinations of primary antibodies (Pax8/DPP-4/synaptopodin; Pax8/podocalyxin; synaptopodin/ SDF-1α; PCNA/Pax8/ED1) were incubated simultaneously overnight at 4 C (Table S2). Immunofluorescence multiple staining was performed using fluorescence-labelled secondary antibodies as listed in Table S3 and nuclei were stained with DAPI (Merck KGaA). Sections were covered with Mowiol mounting medium (Calbiochem, La Jolla, USA) and examined by a confocal laser scanning microscope (LSM Zeiss 710 and Zen software, Zeiss GmbH, Jena, Germany) and Image-J software (Abramoff, Magalhaes, & Ram, 2004). In negative controls, the primary antibody was omitted and replaced with blocking solution.

| Quantitative evaluation of renal histopathologic changes and immunohistology
In at least 30 glomeruli (at 400× magnification) and 15 fields of vision (at 200× magnification) per cross section from each kidney with the glomerular cross sections with crescents in %, the degree of glomerulosclerosis (glomerularsclerosis index) and tubular injury (tubular injury index) were evaluated using semi-quantitative scores as described previously (Schlote et al., 2013). Fifty randomly selected glomeruli per kidney section were examined for each rat at 40× magnification and the mean cell numbers per glomerular cross section that stained positive for Pax8, Sox9, Ki67, WT1, SDF-1α, ED1 and CD3 were evaluated. Glomerular staining for the myofibroblast marker α-smooth muscle actin (SMA), fibronectin, sirius red and DPP-4 in glomeruli of human biopsies were analysed using a semi-quantitative score ranging from 0 to 4 (0, no staining; 1, marginal staining; 2, obvious staining in ≤25%; 3, >25%; 4, >50% of the glomerular section area). CD163-positive cells and CD3-positive cells were counted in 25 randomly selected cortical fields at 20× magnification and demonstrated as cell number per mm 2 .

| Detection of DPP-4 activity in tissue and plasma
Levels of DPP-4 activity in plasma were detected using an assay with Gly-Pro 4-methoxy-β-naphtylamide (Sigma Aldrich) as substrate as described previously in detail (Wang et al., 2014). In situ renal DPP-4 activity was detected using air-dried cryosections from rat kidneys as described previously (Luippold, Mark, Klein, Amann, & Daniel, 2018).
Samples were subsequently washed with assay buffer, underwent nuclear staining using hemalaun and mounted with Aquatex (Merck, Darmstadt, Germany). Kidney sections from non-treated rats pretreated with linagliptin 1 μM served as negative controls.
2.8 | ELISA for detection of MCP-1 and SDF-1α plasma levels MCP-1 plasma levels were quantified using a sandwich ELISA kit according to the protocol provided by the manufacturer (R&D, MJE00, Minneapolis, USA). SDF-1α/CXCL12 plasma levels were quantified by Natural and Medical Science Institute at University of Tübingen (Reutlingen, Germany) using a sandwich ELISA.

| Data and analysis
All data were analysed in a blinded fashion and presented as scatter plots showing each single data point representing the number of independent values and means ± SEM using bars and whiskers. Group sizes in the animal experiment were n = 11 and in human biopsy study at least n = 10. Only for the complex co-localization studies in  (Figure 4a,c), initial preventive treatment did not significantly lower proteinuria in the short-term experiment (Figure 4a). In the long-term preventive linagliptin treatment group, proteinuria was lowered by more than 25% on Weeks 3-5 but did not reach the level of significance at endpoint on Week 8 (Figure 4b). Long-term therapeutic linagliptin treatment was less efficient but also showed a tendency to lower proteinuria and serum urea levels (Figure 4b,d). Serum creatinine was increased by less than 20% in this model (Figure 4e,f) with a tendency to reduce levels in the preventive long-term DPP-4 inhibitor group (Figure 4f). Rats injected with the anti-glomerular basement membrane antibody showed cellular crescents in more than 50% of glomerular cross sections (Figure 5a-c). In the short-term experiment, glomerulosclerosis and tubular injury were significantly lower on Day 7 (Figure 5d,f) after preventive DPP-4 inhibition with no effect on crescent formation (Figure 5b). In contrast, long-term linagliptin treatment significantly reduced crescent formation in both preventive and therapeutic treatment (Figure 5c), indicating that most likely crescent formation and resolution were affected. However, glomerulosclerosis and tubular injury were significantly ameliorated by 20-25% only after preventive DPP-4 inhibitor treatment (Figure 5e,g).

| Linagliptin modulated inflammatory response in glomerular basement membrane nephritis
Inflammatory cells are important pathophysiologic mediators in crescent formation. Therefore, we analysed how linagliptin treatment influenced renal inflammation in experimental glomerular basement membrane nephritis. Linagliptin time dependently F I G U R E 2 Glomerular expression of dipeptidyl peptidase 4 (DPP-4) in healthy and kidneys with glomerulonephritis (GN). Glomerular DPP-4 expression was detected using immunohistochemistry as shown in healthy controls (a, b), lupus nephritis (c), and pauci immune GN (d) using immunohistochemistry (brown staining) and quantified using semiquantitative scoring in kidney sections from healthy and patients with different renal diseases: control (Ctrl), transplanted kidneys without rejection (NTx NR), acute humoral rejection (antibody-mediated rejection; ABMR), IgA nephropathy ( (Figure 7e). F I G U R E 3 Dipeptidyl peptidase 4 (DPP-4) activity in biopsies from healthy and antiglomerular basement membrane (GBM) nephritic rats. In situ DPP-4 activity was detected in cryosections of healthy rats (a), on Day 7 (b) and Day 14 (c) after induction of anti-GBM nephritis using Gly-Pro 4-methoxy-β-naphtylamide (a Gly-Pro-pNA) s a substrate. On Day 14, DPP-4 activity was inhibited in both the preventive (d) and therapeutic (e) linagliptin groups. Plasma DPP-4 activity was measured before and after disease induction using Gly-Pro-pNA as a substrate (f). Each group consisted of 11 rats. *P < 0.05

| DPP-4 inhibition reduced podocyte stress and renal fibrosis
Induction of glomerular basement membrane nephritis resulted in a marked loss of podocytes, which was hardly ameliorated by F I G U R E 4 Linagliptin treatment ameliorated renal function in experimental anti-glomerular basement membrane (GBM) nephritis. Changes in renal function were monitored in shortterm (a, c, e) and late (b, d, f) anti-GBM nephritis in the rat by measuring proteinuria (a, b), serum urea (c, d) and serum creatinine (e, f ). Unless otherwise stated in the graphs, each group consisted of 11 rats. *P < 0.05 linagliptin preventive treatment in the short-term experiment, as assessed by detection of WT1-positive cells (Figure 9a Figure 10f). F I G U R E 5 Linagliptin treatment ameliorated histopathologic changes in experimental anti-glomerular basement membrane (GBM) nephritis. Kidney damage was evaluated using periodic acid-Schiff (PAS) stained kidney sections. Representative glomeruli for healthy and crescentic anti-GBM nephritic rats including treatment groups were shown (a). Bar = 50 μm. The percentage of glomerular cross sections with cellular crescents in short-term (b) and late anti-GBM nephritis (c) was evaluated. Glomerulosclerosis was analysed using an index (glomerularsclerosis index; GSI) (d, e) and the tubular injury was investigated by a tubular injury score (TSI) (f, g) in both short-term (b, d, f) and late regimen (c, e, g). Unless otherwise stated in the graphs, each group consisted of 11 rats. *P < 0.05

| Linagliptin transiently increased DPP-4 substrate SDF-1α (CXCL12 α) with minor changes in downstream pathways
As a potential mediator of linagliptin effects, we investigated the DPP-4 substrate in our nephritis model. Total serum SDF-1α was significantly increased in both linagliptin-treated groups (Figure 11a), while mRNA expression of SDF-1α was similarly increased in all nephritic groups in the short-term experiment and not altered by linagliptin treatment (Figure 11b), indicating reduced degradation of this chemokine. SDF-1α was expressed by podocytes and was absent in glomerular lesions, as shown by double immunofluorescence staining with synaptopodin (Figure 11g,h). However, not uncommon for a secreted protein, we could not detect significant differences in glomerular SDF-1α expression, evaluated by immunohistochemistry, between treated and non-treated anti-glomerular basement membrane nephritic groups (data not shown). Furthermore, we investigated glomerular SDF-1α signalling via ERK and STAT pathways. Both pathways were markedly increased in anti-glomerular basement membrane nephritic rats on Day 14 as well as Week 8 after disease induction, as shown by P-ERK1/2-and P-STAT3-positive cells (Figure 11c,d). In short-term glomerular basement membrane nephritis on Week 2 preventive linagliptin treatment increased ERK1/2 and STAT3 phosphorylation of glomerular cells in number compared to untreated nephritic control but failed to reach the level of significance due to high variability (Figure 11c,d). In contrast, in the long-term experiment, P-ERK1/2 -positive glomerular cells were significantly reduced in therapeutically linagliptin-treated rats, showing the same F I G U R E 6 Macrophage infiltration in experimental anti-glomerular basement membrane (GBM) nephritis was modulated by linagliptin treatment. Immunohistopathologic examination revealed the number of ED1-positive cells per glomerular cross section (a, b) and the number of CD163-positive cells per mm 2 (C, D) in short-term (a, c) and late trial (b, d). Serum MCP-1 at endpoint of each regimen was measured by an ELISA (e, f). Representative staining for glomerular ED1 staining was shown for all groups of the short-term experiment (G). Unless otherwise stated in the graphs, each group consisted of 11 rats. *P < 0.05 tendency for the group treated with the preventive regimen ( Figure 11c). On Week 8, the activation of STAT3 was not influenced by linagliptin treatment (Figure 11d). Since we could show in a previous study that transcription factor Sox9 was highly up-regulated in parietal epithelial cells and Sox9-positive cells could be also detected on the glomerular tuft in anti-glomerular basement membrane nephritic rats, we expected that Sox9 is an important regulator in the pathophysiology of crescent formation. Therefore, we investigated

| DISCUSSION
DPP-4 inhibitor are widely used in patients with diabetes mellitus to improve glycaemic control by inhibition of incretin breakdown (Mulvihill, 2018). Due to its anti-inflammatory and anti-fibrotic effects, DPP-4 inhibitors are recently suggested for treatment of chronic kidney disease (Kanasaki, 2018). There are several preclinical studies using animal models for acute injury (Chen et al., 2017;Glorie et al., 2012;Reichetzeder et al., 2017) and chronic kidney disease (Hasan et al., 2019;Tsuprykov et al., 2016) confirming the reno-protective effects of DPP-4 inhibition. However, the role of DPP-4 in pathogenesis and progression of glomerulonephritis was not investigated in detail (Higashijima, Tanaka, Yamaguchi, Tanaka, & Nangaku, 2015). Since we had observed markedly increased DPP-4 expression in cellular crescents found in human pauci F I G U R E 7 T-cell infiltration in experimental anti-glomerular basement membrane (GBM) nephritis was transiently modulated by linagliptin treatment. CD3-positive cells per glomerular cross section (a, b) and per mm 2 (d, e) were determined in immunohistochemical staining in short-(a, d) and long-term (b, e) experiment. Representative images of CD3 immunohistochemical staining for all groups on Day 14 were shown. Bar = 100 μm (c). Unless otherwise stated in the graphs, each group consisted of 11 rats. *P < 0.05  . Glomerular proliferation was assessed by Ki67 staining in short-term (e) and long-term regimens (f). Co-localization studies were done by confocal immunofluorescence microscopy using the proliferation cell nuclear antigen (PCNA) as a proliferation marker, Pax8 and ED1; and representative stainings were shown for each group on Day 14 (g, bar = 50 μm). Proliferating Pax8-positive (h; n = 5 per group) and ED1positive macrophages (i; n = 5 per group) were determined on glomerular cross sections. Unless otherwise stated in the graphs or legend, each group consisted of 11 rats. (a) *P < 0.05 Pax8-positive parietal epithelial cells on the glomerular tuft were reduced in the preventive and therapeutic treatment with linagliptin 14 days after disease induction and in our long-term study. Since linagliptin also reduced Pax8-positive cells on glomerular tuft when therapy started on Week 5 after disease induction, we suggest that DPP-4 inhibition rather accelerates resolution of crescentic lesions than preventing their formation. Cellular crescents can be resolved but upon gradual encasement of parietal epithelial cells with extracellular matrix a fibrous crescent develops, a lesion that is considered as irreversible (Anguiano, Kain, & Anders, 2020).
In our study, we analysed SDF-1α as a potential player of linagliptin-mediated reduction of Pax8-positive parietal epithelial cells on glomerular tuft. The chemokine SDF-1α is degraded by DPP-4 and expressed by podocytes (Miglio, Vitarelli, Klein & Benetti, 2017;Romoli et al., 2018) and thus a potential mediator of linagliptin effects.
We could confirm the increase of circulating SDF-1α in nephritic rats after linagliptin in short-term treatment. SDF-1α can interact with two different receptors CXCR4 and CXCR7 that exert numerous functions (Miglio et al., 2017) and are both expressed in parietal epithelial cells and podocytes (Romoli et al., 2018). SDF-1α was described as an intrinsic podocyte progenitor feedback mechanism: podocytes produce SDF-1α to keep podocyte progenitor cells quiescent. Loss of podocytes was associated with reduced SDF-1α secretion and hereby activation of podocyte progenitors from Bowman's capsule (Romoli et al., 2018;Sayyed et al., 2009). In vitro experiments demonstrated that SDF-1α strongly induced notch2 in podocytes without induction F I G U R E 9 Dipeptidyl peptidase 4 (DPP-4) inhibition ameliorated podocyte stress. Using immunohistochemistry, the number of WT1-positive podocytes on the glomerular tuft was evaluated in the short-term (a) and long-term regimens (b). Representative WT1-positive cells on glomerular tuft in all groups were shown on Day 14. Bar = 50 μm (c). Glomerular expression of nephrin was analysed by real-time PCR on Weeks 2 and 8 (d).
Immunostaining for desmin was used to evaluate podocyte stress (e). Representative pictures of desmin staining were shown for all groups on Day 14. Bar = 50 μm (f). Unless otherwise stated in the graphs, each group consisted of 11 rats. *P < 0.05 of notch2 signalling (Romoli et al., 2018). In our study, glomerular notch2 expression was significantly increased but not changed by linagliptin treatment ( Figure S3E). However, it might be that locally Macrophages are another important cell type in crescent formation that proliferate and undergo apoptosis during pathogenesis of crescentic glomerulonephritis (Lan et al., 1997). On Day 14, we could observe increased glomerular macrophage infiltration and proliferation in linagliptin-treated groups, which might be responsible for the lacking improvement of the disease at this time point.
Accelerated macrophage acquisition in the short-term experiment could be influenced by the chemokine MCP-1, which is also a F I G U R E 1 0 Dipeptidyl peptidase 4 (DPP-4) inhibition reduced renal fibrosis in experimental anti-glomerular basement membrane (GBM) nephritis. Glomerular (a) and cortical fibrosis (b) were evaluated using a sirius red score for immunohistological staining in all rat groups on Week 4 and Week 8. Quantitative reverse-transcriptase polymerase chain reaction analysis showed the relative mRNA fibronectin (c) and SMA expression (d) for each group on Week 8. Glomerular SMA occurrence was assessed in both short-term (e) and late regimen (f) for each animal. Representative alpha-SMA staining was shown for all groups on Day 14 (G). Bar = 50 μm. Unless otherwise stated in the graphs, each group consisted of 11 rats. *P < 0.05 substrate of DPP-4. Using ELISA, we could not confirm increased MCP-1 serum levels after linagliptin treatment, but the ELISA did not differentiate between active and cleaved MCP-1. Therefore, differences in MCP-1 activity by linagliptin treatment cannot be excluded.  (Wagner, Klemann, Stephan & von Horsten, 2016). While glomerular crescents are the hallmark of severe glomerular damage and important for disease progression, podocytes are essential for renal filtration (Pavenstadt, Kriz, & Kretzler, 2003). The number of podocytes, as assessed by WT1-positive cells, was not different in linagliptin-treated groups, which might explain why proteinuria was not significantly ameliorated. However, there was a tendency F I G U R E 1 1 Linagliptin transiently increased the dipeptidyl peptidase 4 (DPP-4) substrate SDF-1α with minor changes in downstream pathways. Total serum SDF-1α was significantly increased in linagliptin-treated rats compared to healthy and non-treated ones on Day 14 (a). Relative mRNA SDF-1α expression was evaluated by quantitative reverse-transcriptase polymerase chain reaction analysis on Week 8 (b). P-ERK1/2-positive (c) and P-STAT3-positive cells per gcs (d) in immunohistological staining were elevated in all anti-glomerular basement membrane (GBM) nephritic groups on Day 14 and Week 8. Sox9-positive cells on the glomerular tuft per gcs were evaluated in the short-term (e) and long-term (f) experiments using immunohistochemistry. Immunofluorescence triple-staining of DAPI with SDF-1α and synaptopodin in non-treated anti-GBM nephritic rats (g) showed in a closer look (h) immunofluorescence staining of SDF-1α and synaptopodin in non-treated glomeruli on Day 14. Bar = 20 μm. Unless otherwise stated in the graphs, each group consisted of 11 rats. *P < 0.05 to more WT1 positive cells, higher expression of nephrin and significantly less podocyte stress in preventive and therapeutic linagliptintreated groups in the long-term experiment.
We also observed anti-fibrotic effects by linagliptin therapy, as assessed by sirius red stain for collagens, fibronectin and the myofibroblast marker α-SMA, which might be responsible for the beneficial effects on kidney function and regeneration. Anti-fibrotic action of DPP-4 inhibition therapy was shown in many different preclinical studies using models for diabetic nephropathy (Kanasaki et al., 2014;Shi, Koya, & Kanasaki, 2016) or chronic kidney disease (Tsuprykov et al., 2016). Linagliptin reduced TGF-β signalling in several models by different molecular pathways (Gupta & Sen, 2019). Furthermore, DPP-4 was shown to interact with cation-independent mannose 6-phosphate receptor (CIM6PR), thus being involved in TGF-β activation (Gangadharan Komala, Gross, Zaky, Pollock, & Panchapakesan, 2015). This interaction can be inhibited by linagliptin treatment (Gangadharan Komala et al., 2015). It remains unclear if this mechanism leads to the observed anti-fibrotic effects in our glomerulonephritis model and must be evaluated in later studies. However, the anti-fibrotic action of linagliptin might be prevent the development of irreversible fibrous crescent formation and hereby allows its resolution. This theory could explain that in the long-term experiment, therapeutic administration of linagliptin is not able to dissolve already irreversibly remodelled crescent lesions, so that only a partial improvement could be achieved.
Clearly, our study is limited by focusing on selected DPP-4 substrates and we cannot exclude that additional DPP-4 substrates or interaction partners are involved. One key substrate of DPP-4 is, for example, GLP-1, but linagliptin effects on crescent formation mediated by the incretin GLP-1 are unlikely since in humans neither podocytes nor parietal epithelial cells express the GLP-1 receptor on their surface (Pyke et al., 2014).
In conclusion, our study showed reno-protective effects of DPP-4 inhibition. In the short-term treatment, DPP-4 inhibition increased glomerular macrophage proliferation and numbers while simultaneously Pax8-positive parietal epithelial cell proliferation was reduced leading to lower Pax8-positive cells on glomerular tuft.
These early DPP-4 inhibitor effects might contribute to reduced crescent formation and improved crescent resolution, fibrosis and podocyte stress observed in the long-term experiment when rats were treated preventively and to some extent also therapeutically with linagliptin in a model of crescentic glomerulonephritis. The DPP-4 substrate SDF-1 is might be an important mediator of these effects, but we assume a complex effect of DPP-4 inhibition with currently unknown additional players. The observed effects, however, are not yet sufficient to suggest DPP-4 inhibitor treatment as a single therapy and linagliptin effects on crescent formation fit only partly to current concepts of crescentic lesion pathogenesis, indicating complex interactions that need further investigation. Nevertheless, DPP-4 inhibitor treatment by linagliptin not only is a secure therapy option for diabetic patients with kidney injury but also can improve progression of renal disease.