Glucagon-Like Peptide 1 and Its Cleavage Products Are Renoprotective in Murine Diabetic Nephropathy

Glucagon-like peptide 1 (GLP-1) is an incretin hormone secreted by intestinal L cells and pancreatic α-cells (1). GLP-1 (7-36 amide or 7-37) improves glucose metabolism by increasing glucose-dependent insulin secretion, inhibiting gastric emptying, and reducing appetite. This has made GLP-1-modifying therapies an attractive target to treat diabetes. GLP-1 (7-36 amide or 7-37) is rapidly inactivated by the ubiquitously present enzyme dipeptidyl peptidase 4 (DPP-4), cleaving off the first two N-terminal amino acids of the peptide. The resulting GLP-1 (9-36 amide or 9-37) metabolite does not activate the known GLP-1 receptor, which has made DPP-4 inhibition a successful strategy to improve glucose metabolism. Additional enzymatic cleavage of GLP-1 by neutral endopeptidase (NEP) results in smaller GLP-1 fragments including GLP-1 (28-36 amide or 28-37) (1).

Recent clinical trials have demonstrated marked differences in cardiovascular outcomes among available incretin-based therapies in patients with diabetes. Importantly, the GLP-1 receptor agonists liraglutide and semaglutide, which mimic the peptide structure of GLP-1, were found to improve cardiovascular and renal outcome and reduce cardiovascular and overall mortality in high-risk patients with diabetes (2,3). In contrast, neither GLP-1 receptor agonists with exendin-based peptide structure (featuring only 53% homology to GLP-1) nor DPP-4 inhibitors, which prevent N-terminal GLP-1 cleavage, were found to reduce cardiovascular events in patients with diabetes (4,5). Importantly, cleavage products of GLP-1 have been found to have GLP-1 receptor-independent metabolic and immune-regulatory actions (1). As these may only be increased by GLP-1-based (liraglutide or semaglutide) but not exendin-based (lixisenatide or exenatide long-acting release [LAR]) GLP-1 receptor agonists or DPP-4 inhibition, these might hold relevance for the differential cardiovascular actions of these drugs. In this study, we directly compared metabolic, renal, and cardiac actions of GLP-1 and its cleavage products GLP-1 (9-37) and GLP-1 (28-37) in the diabetic model of db/db mice on a high-fat diet.

The db/db and control wild-type (wt) mice (db/db: BKS.Cg-Dock7m^+/+Leprdb/J, stock number 000642; controls: C57BLKS/J, stock number 000662; Charles River Laboratories) received 5 × 10¹² particles of AAV-CMV-GLP-1 variants ([7-37 Mut8] or [9-37] or [28-37]; each n = 10) and AAV-CMV-LacZ (n = 13 for db/db and n = 5 for wt) as control. Adeno-associated viruses (AAVs) were generated as previously reported (6). All animals were fed a high-fat Western-type diet (39 kJ% fat, 41 kJ% carbohydrates, and 20 kJ% protein [ssniff EF R/M acc. TD88137AQ8 mod.; ssniff Spezialdiäten GmbH]) for 9 weeks with additional feeding of regular chow diet for the following 5 weeks. Blood pressure was recorded by computerized CODA system (Kent Scientific Corporation) after repeated procedural training of the animals. Glucose tolerance was assessed after overnight fast by i.p. injection of 2 g glucose/kg body weight. Hemodynamics were measured by Millar catheter (Millar) after advancing a 1FR Millar Catheter (SPR 1000) across the right carotid artery through the aortic valve into the left ventricle in mice anesthetized using ketamine (100 mg/kg) and xylazine (10 mg/kg). After hemodynamic stabilization, signals were continuously recorded and analyzed before and after i.p. injection of dobutamine 5 µL/g body weight using iox (emka Technologies). Serum and urinary electrolytes were assessed by the chemistry department of the animal facility of the University Hospital RWTH Aachen using an autoanalyzer. Urinary albumin was measured by ELISA (Bethyl Laboratories) according to the manufacturer’s protocol. Urine collection was performed in metabolic cages for 16 h.

For the acute unilateral renal ischemia/reperfusion (I/R) injury model, C57BL/6J mice were anesthetized by ketamine (100 mg/kg) and xylazine (10 mg/kg). Left kidneys were clamped for 30 min. During this period, body temperature was maintained between 34°C and 37.5°C using a temperature-controlled heating system. Analgesia with Temgesic (0.05 mg/kg) was used every 8 h until sacrifice 48 h after surgery. Mice were treated with GLP-1 (7-36) amide (Bachem), GLP-1 (9-36) amide (Bachem), GLP-1 (28-36) amide (Phoenix Pharmaceuticals), exendin-4 (Bachem), and liraglutide (Novo Nordisk), all at a concentration of 100 nmol/L/kg, and the GLP-1 receptor antagonist exendin-9 (Bachem; 1,000 nmol/L/kg) twice daily by i.p. injection starting 12 h before surgery. Blood and PBS-perfused kidneys were collected for further analyses. All experiments were approved by the government of North Rhine-Westphalia, Germany.

Buffy coats were obtained completely anonymized from the Department of Transfusion Medicine, University Hospital RWTH Aachen, which does not require ethical approval in Germany. Human CD4⁺ lymphocytes were isolated from buffy coats as has been previously described (7). Briefly, a Biocoll (Biochrom) gradient centrifugation was used to obtain mononuclear cells, and subsequent negative selection of CD4⁺ T cells was performed by magnetic bead separation (Miltenyi Biotec). All GLP-1 peptides were used at a concentration of 1 nmol/L, exendin-9 was used at a concentration of 10 nmol/L, and CCL5 (RANTES) (Sigma-Aldrich) at a concentration of 100 ng/mL.

After isolation, CD4⁺ lymphocytes were cultured in serum-free media for 24 h. T cell chemotaxis was performed under serum-free conditions in 48-well FluoroBlok cell culture inserts with a pore size of 8 µm and in a density of 2 × 10⁵/mL. CCL5 was used in a concentration of 100 ng/mL, all GLP-1 peptides were used in a concentration of 100 nmol/L, and exendin-9 was used at a concentration of 1,000 nmol/L. Cells were stimulated for 18 h, fixed with 4% paraformaldehyde, stained with DAPI (Vector Laboratories), and counted in five random fields per well.

Serum concentrations of total GLP-1 were analyzed as reported previously (8) using anti-GLP-1 (non-amid. [-Arg-Gly], C-term.; ABS 046-03; Dianova) and anti-Glp-1 (midmolecule specific; HYB 147-12B; Dianova). Serum concentrations of GLP-1 (28-37) were assessed by mass spectrometric analyses using 2.5-dihydroxybonzoic acid as matrix, loaded to the matrix-assisted laser desorption/ionization (MALDI) target (Bruker Daltonik, Bremen, Germany) with subsequent MALDI time-of-flight mass spectrometry (MALDI-TOF MS, MALDI-Lift) fragment MS (MALDI-TOF/TOF MS; Bruker).

Histology and immunohistochemistry was performed using 1-µm paraffin sections stained with periodic acid Schiff (PAS; Sigma-Aldrich) reagent (Merck), Mayers hematoxylin (Sigma-Aldrich) counterstain or sequential antibody incubation, the ABC-Elite reagent (Vector Laboratories), and 3,3′-diaminobenzidine (DAB; Sigma-Aldrich) with nuclei counterstaining using methyl green (Sigma-Aldrich) as reported previously (9,10). Kidney sections were deparaffinized and rehydrated followed by heat-mediated antigen retrieval and blocking of endogenous peroxidase. The sections were incubated for 1 h with the primary antibody (see below) diluted in PBS containing 1% BSA (Sigma-Aldrich). The slides were sequentially incubated with biotinylated secondary antibody (see below), the ABC-Elite reagent (Vector Laboratories), and finally DAB (Sigma-Aldrich) with nickel chloride enhancement used as the chromogen. Nuclei were counterstained with methyl green (Sigma-Aldrich). Tubulointerstitial injury of db/db mice was assessed under consideration of loss of brush border, tubular dilatation, interstitial inflammation, and matrix deposition in PAS sections using the following scoring system: 0 is 0–1%; 1 is 2–25%; 2 is 26–50%; 3 is 51–75%; and 4 is >75% tubulointerstitial injury. Quantification of immune cells was performed by manual counting in 12 view fields per cross section (×40 original magnification). The tubular injury in the I/R model was scored as follows: 0 is normal; 1 is mild tubular dilation, flattened epithelial cells, or loss of brush border; 2 is apoptotic nuclei, loss of tubular epithelial cells, or denudation of basement membrane; and 3 is complete tubular necrosis. Per mouse, 100 cortical tubuli and 100 tubuli in the outer medulla were scored. For computer-assisted morphometric analyses, stained sections were digitalized using a whole-slide scanner (NanoZoomer HT; Hamamatsu Photonics) and processed by NDP.view (Hamamatsu Photonics) and ImageJ (National Institutes of Health). Per slide, 7–13 images in ×40 original magnification were taken, and the percentage of positively stained area (DAB or PAS) of each slide was quantified in a blinded manner. Used antibodies were CD3 (monoclonal antibody) and F4/80 (monoclonal antibody) from AbD Serotec/Bio-Rad, Lipocalin-2 (Polyclonal Antibody) from R&D Systems, and anti-rat IgG from Vector Laboratories.

Kidney and heart tissue was minced and digested by collagenase type IV (Worthington Biochemical Corporation, Lakewood, NJ). Leukocytes were isolated by multiple centrifugation steps (11). Red cell lysis was performed by Pharm Lyse (BD Biosciences, San Jose, CA). The cells were stained with fluorochrome-conjugated antibodies for multicolor flow cytometry analysis with FACS Canto II running FACS Diva 7.0 (BD Biosciences) (db/db experiment) and LSR-Fortessa running FACS Diva 8.0.1 (BD Biosciences) (I/R experiments). Analysis of the data was done with FlowJo (Tree Star).

In the db/db experiment, for renal and cardiac tissue, live cells (Hoechst negative) were gated by the following strategy: T-helper cells were CD45 positive, CD19 negative, and CD4 positive; cytotoxic T cells were CD45 positive, CD19 negative, and CD8 positive; macrophages were CD45 positive, Ly6G negative, and F4/80 positive cells. For blood samples, live cells (Hoechst negative) were gated as follows: T cells were FSC/SSC low, CD19 negative, and then CD4 and CD8 positive cells; and monocytes were Ly6G negative and CD11b and CD115 double positive. Flow cytometric analysis was performed on an FACS Canto II running FACS Diva 7.0 (BD Biosciences) and analyzed with FlowJo (Tree Star). For the I/R experiment, renal leukocytes (7-aminoactinomycin D⁻ [7-AAD], CD45⁺) were gated for T-helper cells (TCRβ⁺ NK1.1⁻, CD4⁺), cytotoxic T cells (TCRβ⁺ NK1.1⁻, CD8⁺), and macrophages (Ly6G⁻; CD11b^high; F4/80^int). Blood leukocytes (7-AAD⁻, CD45⁺) were gated for T-helper cells (TCRβ⁺ NK1.1⁻, CD4⁺), cytotoxic T cells (TCRβ⁺ NK1.1⁻, CD8⁺), and monocytes (Ly6G⁻, CD11b^high, CD115⁺). Flow cytometric analysis was performed on an FACS LSR-Fortessa running FACS Diva 8.0.1 (BD Biosciences) and analyzed with FlowJo (Tree Star). The following FACS antibodies were used: anti-mouse CD8a fluorescein isothiocyanate (FITC), anti-mouse CD115 (c-fms) phycoerythrin (PE), anti-mouse CD11b allophycocyanin (APC), anti-mouse CD4 APC, CD115 (c-fms) monoclonal antibody (AFS98) PE, TCR-β monoclonal antibody (H57–597) PE-cyanine 7, F4/80 monoclonal antibody (BM8), and PE-cyanine 7 (all from eBioscience); APC-Cy7 rat anti-mouse CD45, FITC rat anti-mouse Ly-6G, PE-Cy7 streptavidin, PerCP-Cy5.5 rat anti-mouse CD19, FITC rat anti-mouse Ly-6G, PE mouse anti-mouse NK-1.1, PE rat anti-mouse Ly-6G, Alexa Fluor 647 rat anti-CD11b, PE rat anti-mouse Ly-6G, Alexa Fluor 647 rat anti-CD11b, APC-Cy7 rat anti-mouse CD45, and Hoechst 33342 solution (all from BD Pharmingen); rat anti-mouse F4/80 antigen Alexa Fluor 488 from AbD Serotec; 7-AAD APC/Cy7 anti-mouse NK-1.1 antibody and Brilliant Violet 711 anti-mouse CD8a from BioLegend; 7-AAD from Sigma-Aldrich; and V450 rat anti-mouse CD4 and BV510 rat anti-mouse CD45 from BD Horizon.

Gene expression analysis was performed as previously reported (12). Primer sequences for SYBG were: Actb forward, 5′-CTCTAGACTTCGAGCAGGAGATGG-3′ and reverse, 5′-ATGCCACAGGATTCCATACCCAAG-3′; Kim1 forward, 5′-GCAAGGATTCCACTTTCCGTT-3′ and reverse, 5′-GCACCCTGCAGTCATTCAGA-3′; Lcn2 forward, 5′-GGCCTCAAGGACGACAACA-3′ and reverse, 5′-TCACCACCCATTCAGTTGTCA-3′; Tnfa forward, 5′-TCCACTTGGTGGTTTGCTACG-3′ and reverse, 5′-TCCACTTGGTGGTTTGCTACG-3′; Ccl5 forward, 5′-TGCAGAGGACTCTGAGACAGC-3′ and reverse, 5′-GAGTGGTGTCCGAGCCATA-3′; Synpo forward, 5′-CAGCCGCAAATCCATGTTTA-3′ and reverse, 5′-CCGCTGTCTGTACCAGATCCA-3′; and TaqMan BGH polyadenylation signal sequence probe, FAM-5′-TCCCCCGTGCCTTCCTTGACC-3′-TAM, forward, 5′-TCTAGTTGCCAGCCATCTGTTGT-3′, and reverse, 5′-TGGGAGTGGCACCTTCCA-3′.

Data are presented as the mean ± SD with statistical analysis performed by GraphPad Prism 6.0 (GraphPad Software Inc.) as indicated in the figure legends.

To investigate metabolic, renal, and cardiac effects of GLP-1 and its cleavage products, we overexpressed DPP-4-resistant GLP-1 (7-37 Mut8) and two GLP-1 cleavage products, GLP-1 (9-37) and GLP-1 (28-37), in db/db mice on a high-fat diet using an AAV vector system. This led to a stable increase of GLP-1 peptide concentrations in comparison with control groups (Fig. 1A). Overexpression of the DPP-4-resistant GLP-1 receptor-activating GLP-1 (7-37 Mut8) significantly improved glucose metabolism by reducing fasting blood glucose and increasing glucose tolerance, whereas the GLP-1 cleavage products had no effects (Fig. 1B and C). This was associated with a paradoxical increase of body weight of GLP-1 (7-37 Mut8)-treated animals, which only became significant at the final 14-week time point (Fig. 1D). No difference in blood pressure (Fig. 1E and Supplementary Fig. 1A–C) or serum electrolytes (Supplementary Fig. 1D and E) was recorded between groups.

Compared with wt LacZ-transfected mice, diabetic LacZ mice developed prominent renal injury (Fig. 2). None of the GLP-1 constructs reduced albuminuria (Fig. 2A) or altered urinary electrolyte excretion (Supplementary Fig. 1F and G), although glycosuria was significantly reduced by GLP-1 (7-37 Mut8) treatment (Fig. 2B). Further, GLP-1 constructs had no effects on glomerular podocyte damage as detected by podocin or synaptopodin expression (Fig. 2C, E, and F). Still, all GLP-1 peptides significantly improved and even normalized tubular injury (Fig. 2D and E). These findings were confirmed by reduced renal mRNA expression of the tubular injury markers Kim1 and Lcn2 (Fig. 2F), together with lower expression of the proinflammatory cytokines Tnfa and Ccl5 (Fig. 2G). Furthermore, all GLP-1 constructs significantly reduced renal accumulation of F4/80 positive monocytes, macrophages, and dendritic cells and of CD3 positive T cells as shown by immunohistochemistry (Fig. 2E, H, and I). This was confirmed by FACS analyses demonstrating lower renal CD4 and CD8 positive T cells in addition to reduced F4/80 positive cells in all GLP-1 groups (Fig. 3A).

To investigate whether this immune-regulatory effect of GLP-1 and its cleavage products was restricted to the kidney or also present systemically, we performed FACS analysis of blood and heart tissue. Consistent with the literature, we found reduced circulating T cells in db/db mice in comparison with wt controls (13) (Fig. 3A). Interestingly, all GLP-1 constructs restored circulating T cell numbers to levels of wt mice. Still, all GLP-1 constructs significantly reduced myocardial tissue T cell accumulation, resembling the effect observed in the kidney (Fig. 3A). This, however, was not associated with a difference in cardiac function as assessed by Millar catheter with and without dobutamine stress (Fig. 3B–D). Treatment with each of the GLP-1 constructs was associated with reduced mortality of db/db mice on a high-fat diet (mortality of GLP-1 [7-37 Mut8]: 20%; GLP-1 [9-37]: 30%; GLP-1 [28-37]: 30% in comparison with control [LacZ]: 46%), which became significant in a combined analysis for all GLP-1 groups in comparison with control (LacZ) db/db (P < 0.05) (Fig. 3E).

To further investigate the mechanisms of renoprotection, we treated a human tubular cortical cell line (HK-2) and a murine renal tubular cell line in addition to primary human T cells under in vitro conditions with GLP-1 and its cleavage products. Although GLP-1 peptides did not affect interleukin-6 secretion of HK-2 cells or Lcn2 expression of murine renal tubular cells in response to different stressors (data not shown), all GLP-1 peptides inhibited CCL5 (RANTES)-dependent T cell migration (Fig. 4A and B). This was not attenuated by cotreatment with the GLP-1 receptor antagonist exendin-9 (Fig. 4C and D).

To further characterize the immune-modulatory actions of GLP-1 and its cleavage products under in vivo conditions, we used the unilateral renal I/R injury mouse model. Pretreatment with GLP-1 or its cleavage products significantly reduced renal infiltration of proinflammatory Ly6C-high monocyte-derived macrophages (MoMFs) (Fig. 4E) while not affecting Ly6C-low MoMFs (Supplementary Fig. 2A) 48 h postreperfusion. Further, all GLP-1 constructs significantly reduced renal infiltration of CD4 and CD8 positive T cells (Fig. 4E). This was not associated with a differential expression of the renal tubular injury marker Lcn2 or histological assessment of tubular injury (Supplementary Fig. 2B–D). In addition, GLP-1 (28-36) by trend reduced circulating monocyte count (Fig. 4F). This immunomodulatory pattern was not attenuated by cotreatment with the GLP-1 receptor antagonist Ex-9, which by itself reduced renal infiltration of Ly6C-high MoMFs and CD4 positive T cells (Fig. 4G) while increasing circulating blood CD4 and CD8 positive T cells (Fig. 4H). To investigate whether synthetic GLP-1 receptor agonists exert similar immune modulatory actions and whether differences exist between GLP-1-based and exendin-based GLP-1 receptor agonists, we treated mice with liraglutide or exendin-4 in the same model. Only liraglutide significantly decreased renal CD4 positive T cell infiltration (Fig. 4I). Still, neither liraglutide nor exendin-4 affected CD8 positive T cell nor macrophage infiltration (Fig. 4I) while also not attenuating circulating leukocyte profile (Fig. 4J).

In this study, we found GLP-1 (7-37 Mut8) and the GLP-1 cleavage products GLP-1 (9-37) and GLP-1 (28-37) to reduce tubulointerstitial renal damage and have immune-modulatory effects in the chronic diabetes model of db/db mice. A similar immune-modulatory effect by all GLP-1 cleavage products was found in the acute model of renal I/R injury. Further, all GLP-1 constructs inhibited in vitro T cell migration, suggesting direct immune-modulatory actions.

Importantly, even the shortest nonapeptide GLP-1 (28-37), which gets liberated by NEP-dependent cleavage from GLP-1 (7-37 Mut8) and GLP-1 (9-37), proved sufficient to provide comparable renoprotective and immune-regulatory capacities to the full-length DPP-4-resistant GLP-1 (7-37 Mut8) peptide. This occurred in a glucose-independent manner, with glucose metabolism only being improved by GLP-1 (7-37 Mut8) but none of the GLP-1 cleavage products in the db/db model. GLP-1 (7-37 Mut8)-dependent improvement of glucose metabolism might explain the paradoxical increase of body weight as an indicator of improved health status of db/db mice burdened by severe diabetes. This interpretation is supported by a similar paradoxical increase of body weight in response to chronic sodium–glucose cotransporter 2 (SGLT2) inhibition in the same model (14).

GLP-1 (9-36 amide) and GLP-1 (28-36 amide) have been found to exhibit GLP-1 receptor-independent properties, which include inhibition of weight gain, improvement of glucose metabolism, and reduction of oxidative stress (15–17). Still, GLP-1 cleavage products did not affect body weight or glucose metabolism in our study. This discrepancy might relate to the distinct mouse models used, which will require further investigation.

Importantly, the GLP-1 receptor agonists liraglutide and semaglutide, which are based on the peptide structure of GLP-1 and incorporate the modified GLP-1 (28-37) C-terminus (with arginine to lysine substitution at amino acid 34), were recently found to reduce cardiovascular and renal events in patients with diabetes and high cardiovascular risk (2,3). In contrast, neither DPP-4 inhibition, which prevents GLP-1 degradation, nor the GLP-1 receptor agonists lixisenatide and exenatide LAR, which are both based on the peptide structure of exendin with low amino acid homology to GLP-1, were found to improve cardiovascular outcome in clinical studies (4,5). It seems possible that beneficial effects on end-organ injury depend on an increased availability of GLP-1 cleavage products provided by liraglutide and semaglutide, but not exendin-based GLP-1 receptor agonists like lixisenatide or exenatide LAR. Similar to GLP-1, liraglutide is cleaved by DPP-4 in the Ala8-Glu9 position of the N-terminus and degraded by NEP to several metabolites (18). However, liraglutide and exendin-4 did only partially mimic the immune-modulatory actions of GLP-1 cleavage products after renal I/R injury in our study. This differential response might be due to slow metabolization and long half-life (11–15 h) of liraglutide in comparison with the rapidly metabolized native GLP-1 peptides (half-life: 1 to 2 min) (19). The prolonged half-life of liraglutide is reached by replacement of Lys34 by Arg and derivatization of the GLP-1 protein backbone in the Lys26 position with a glutamate spacer bound to a 16C fatty acid, allowing reversible plasma albumin binding (20). Tracer studies in humans suggest <11% of radiolabeled liraglutide to be available as metabolites within the first 24 h after injection (18). First application of liraglutide 12 h before renal I/R injury, as performed in our study, might therefore have been too short to allow sufficient metabolization and creation of cleavage products. Still, only very little liraglutide is excreted from the organism, suggesting full in vivo metabolization (18). This might lead to relevant concentrations of cleavage products under chronic treatment conditions, which will require further investigation. In addition, the specific biological activity of the modified GLP-1 cleavage product as being present in the peptide structure of liraglutide should be investigated.

Renoprotective effects of GLP-1 and its analogs are consistently reported in preclinical and clinical studies (1,21), although less potent than observed for renin-angiotensin-aldosterone system blockade or SGLT2 inhibition (22). None of the GLP-1 constructs attenuated albuminuria or glomerular injury in our model. Still, all GLP-1 peptides reduced renal tubulointerstitial damage and immune-cell infiltration. This was not mediated by differences in blood pressure as assessed by noninvasive tail cuff method, which might, however, underestimate central blood pressure and has limitations of restrain-induced stress response. As alternative mechanism GLP-1 cleavage products have been reported to attenuate macrophage recruitment to atherosclerotic lesions of apolipoprotein E-deficient mice (8) and to reduce T cell activation under in vivo and in vitro conditions (1,7,15). Given the chronic inflammatory nature of type 2 diabetes, these anti-inflammatory/immunomodulatory properties might contribute to a glucose-unrelated prognostic benefit of GLP-1. These actions are most likely independent of the known GLP-1 receptor, which has earlier been found to be dispensable for cardioprotective effects of GLP-1 and its cleavage products (23). Consistently, application of GLP-1 receptor agonists did not or only partially recapitulated the immunomodulatory phenotype of GLP-1 cleavage products after renal I/R injury in our study, whereas antagonization of the GLP-1 receptor by exendin-9 did not reverse immunomodulatory actions of GLP-1 (28-36) in the same model. Alternative mechanisms might include the existence of an additional receptor or point to receptor-independent actions of the peptides, with GLP-1 (28-36 amide) being found to directly enter cells and improve cellular metabolism by targeting mitochondria (16).

Further pharmacological judgment of the renoprotective potential of GLP-1 cleavage products will require head-to-head comparison with renin-angiotensin-aldosterone system-blocking agents and/or SGLT2 inhibitors.

GLP-1 and its cleavage products did not improve cardiac function in the diabetic model of db/db mice in our study, despite a reduction of myocardial immune cell recruitment. This is in contrast to the cardioprotective effects of GLP-1 reported in other experimental models but is consistent with recent clinical studies, which found no benefit of GLP-1 receptor agonists liraglutide or albiglutide in patients with heart failure (24,25).

In conclusion, we found GLP-1 and its cleavage products GLP-1 (9-37) and GLP-1 (28-37) to have glucose-independent renoprotective effects. As GLP-1 cleavage products are increased by some but not all GLP-1 receptor agonists, this might be relevant for the differential cardiovascular benefit within this drug class.

Acknowledgments. The authors thank Katharina Steiner from the Department of Internal Medicine I, University Hospital RWTH Aachen; Simon Otten, Marie Cherelle Timm, and Christina Gianussis from the Institute of Pathology, University Hospital RWTH Aachen; and Hiltrud Königs from the Electron Microscopy Facility, University Hospital RWTH Aachen, for technical assistance.

Funding. This work was supported by the START program from the Faculty of Medicine at the RWTH Aachen University to M.L. and financial research grants from the German Research Foundation SFB/Transregio 219 (SFB-TRR) “Mechanisms of Cardiovascular Complications in Chronic Kidney Disease” to M.L. and P.B.; SFB/TRR 57 “Mechanisms of Organ Fibrosis”, BO 3755/3-1, and BO 3755/6-1 to P.B.; the German Ministry of Education and Research (BMBF Consortium STOP-FSGS number 01 GM1518A to P.B.); the Corona Foundation to N.M. and M.L.; and the Interdisciplinary Centre for Clinical Research within the Faculty of Medicine at the RWTH Aachen University (K7-3 to P.B.).

Duality of Interest. N.M. has received support for clinical trial leadership from Boehringer Ingelheim; served as a consultant to Amgen, Bayer, Boehringer Ingelheim, Sanofi, Merck Sharp & Dohme, Bristol-Myers Squibb, AstraZeneca, and Novo Nordisk; has received grant support from Boehringer Ingelheim and Merck Sharp & Dohme; and has served as a speaker for Amgen, Bayer, Boehringer Ingelheim, Sanofi, Merck Sharp & Dohme, Bristol-Myers Squibb, AstraZeneca, Eli Lilly and Company, and Novo Nordisk but declines all personal compensation from pharmaceutical or device companies. M.L. serves as an advisor to Merck Sharp & Dohme, Boehringer Ingelheim, Novo Nordisk, and Amgen and received speaker honoraria from Merck Sharp & Dohme, Boehringer Ingelheim, Novo Nordisk, Amgen, Sanofi, Bayer, Eli Lilly and Company, and AstraZeneca. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. J.M. was responsible for experimental design and performance and wrote the first draft of the manuscript. B.M.K. performed I/R experiments and histological analyses and edited the first draft of the manuscript. J.O. and F.T. were responsible for FACS analysis, critical discussion of the results, and editing of the manuscript. R.S. helped with the Millar catheter and FACS performance and analysis. V.J. and J.J. performed peptide MS analysis. C.L. provided critical discussion of the results and review and editing of the manuscript. N.M. provided critical discussion of the results and editing of the manuscript. P.B. was responsible for histological analyses, critical discussion of the results, and manuscript editing. M.L. was responsible for experimental design and wrote the final manuscript. M.L. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.