SGLT2 Inhibition by Intraperitoneal Dapagliflozin Mitigates Peritoneal Fibrosis and Ultrafiltration Failure in a Mouse Model of Chronic Peritoneal Exposure to High-Glucose Dialysate

Peritoneal dialysis (PD) is limited by glucose-mediated peritoneal membrane (PM) fibrosis, angiogenesis, and ultrafiltration failure. Influencing PM integrity by pharmacologically targeting sodium-dependent glucose transporter (SGLT)-mediated glucose uptake has not been studied. In this study, wildtype C57Bl/6N mice were treated with high-glucose dialysate via an intraperitoneal catheter, with or without addition of selective SGLT2 inhibitor dapagliflozin. PM structural changes, ultrafiltration capacity, and peritoneal equilibration testing (PET) status for glucose, urea, and creatinine were analyzed. Expression of SGLT and facilitative glucose transporters (GLUT) was analyzed by real-time PCR, immunofluorescence, and immunohistochemistry. Peritoneal effluents were analyzed for cellular and cytokine composition. We found that peritoneal SGLT2 was expressed in mesothelial cells and in skeletal muscle. Dapagliflozin significantly reduced effluent transforming growth factor (TGF-β) concentrations, peritoneal thickening, and fibrosis, as well as microvessel density, resulting in improved ultrafiltration, despite the fact that it did not affect development of high-glucose transporter status. In vitro, dapagliflozin reduced monocyte chemoattractant protein-1 release under high-glucose conditions in human and murine peritoneal mesothelial cells. Proinflammatory cytokine release in macrophages was reduced only when cultured in high-glucose conditions with an additional inflammatory stimulus. In summary, dapagliflozin improved structural and functional peritoneal health in the context of high-glucose PD.


INTRODUCTION 48
Peritoneal dialysis (PD) as a renal replacement therapy for individuals with end stage renal 49 disease relies on the peritoneum and its properties as a dialyzer membrane. Glucose-based PD 50 fluid (PDF) generates an osmotic gradient that promotes water and solute clearance across the 51 peritoneal membrane. However, glucose-containing PDF is non-physiological and as a result, 52 in most PD patients structural and functional changes occur over time, resulting in decreased 53 dialysis efficiency and ultimately technique failure.
[1] While our understanding of the 54 molecular mechanisms of such PD-related structural and functional aberrations of the 55 peritoneum has grown considerably over the last decades, successful translation of 56 7/23 automated microvessel imaging NanoZoomer 2.0-HT Scan System (Hamamatsu Photonics) 148 was used at 20x magnification (resolution: 0.46 μ m/pixel). The slide scanner automatically 149 detects the region of interest (ROI) containing the tissue and automatically determines a valid 150 focal plane for scanning. As PDF penetration level reaches 400 µm, an area reaching 400 µm 151 below mesothelial cell layer was annotated as ROI and microvessel density was quantified by 152 microvessel algorithm v1 (Aperio Image Scope, Leica). 153

RNA extraction and real-time quantitative PCR 155
Total RNA was extracted from harvested anterior peritoneal walls not affected by the 156 peritoneal catheter using RNeasy mini kit (Qiagen) and reverse-transcribed using Promega 157 kits. Real-time quantitative PCR analysis was performed on a LightCycler480 (Roche) real-158 time PCR system using SybrGreen as well as TaqMan technologies; β -tubulin and Rn18S 159 mRNA were used as reference genes. Quantification was conducted using the delta-delta Ct HPMC were derived from omentum samples of 3 human controls as described 167 previously [4] and grown to 80% confluence. In short, HPMC were isolated with 168 trypsin/EDTA digestion method from omentum tissue obtained from patients with normal 169 kidney function undergoing elective abdominal surgery. Informed consent was obtained for 170 the use of omentum tissue and the study was approved by the institutional ethics committee 171 (Hannover Medical School #17/6715). The cells were grown in RPMI1640 medium 172 supplemented with 10% fetal bovine serum (FBS), 100 U/ml penicillin and 100 mg/ml 173 streptomycin 174 Immortalized MPMC were generated in our lab and cultivated as described 175 previously [4]. Briefly, the cells were grown to 80% confluence in RPMI1640  in the single mesothelial cell layer, but also in submesothelial skeletal muscle (Figure 1a, 213 upper row). Antibody specificity against SGLT1 and SGLT2, respectively, was confirmed in 214 kidney tissue from the same animals (Figure 1a We then evaluated the peritoneal transcriptional expression of SGLT2, SGLT1 and 231 several GLUTs known to be expressed in the peritoneum. We found a strong upregulation of 232 SGLT2 expression in mice receiving high glucose PDF, whereas SGLT1 expression was 233 unaltered (Figure 2b). Most notably, pharmacological inhibition of SGLT2 with 234 dapagliflozin completely abrogated PD-induced upregulation of SGLT2. Glucose transporters 235 demonstrated differential regulation, GLUT1 and 3 being upregulated and GLUT4 down-236 regulated, respectively, in response to chronic exposure to PDF. This regulation was 237 unaffected by dapagliflozin (Figure 2c).  (Figures 4a-b). Dapagliflozin-treated animals demonstrated reduced microvessel 276 density (p=0.06). Of note, while PDF-treated animals showed a significant increase in 277 vascular endothelial growth factor A (VEGF-A) levels in peritoneal effluents, dapagliflozin-278 treated animals had similar levels of VEGF-A, suggesting that dapagliflozin-mediated 279 reduction of angiogenesis was independent of VEGF-A (Figure 4c). 280 281

Dapagliflozin modulates intraperitoneal inflammatory response 282
After demonstrating ameliorating effects of SGLT2 inhibition on development of peritoneal 283 fibrosis, angiogenesis and UF failure in a high glucose milieu we were interested to evaluate 284 its effects on intraperitoneal inflammation. We therefore analyzed the composition of 285 intraperitoneal cell influx in effluents obtained after a 120 min dwell of PDF across all 286 groups. Consistent with previous findings from our group,[4] chronic PDF exposure led to a 287 significant increase in peritoneal cell count, predominantly leukocytes. Significantly different 288 changes were noted for T cells, B cells, polymorphonuclear neutrophils (PMN) and 289 macrophages (Figure 5a). While dapagliflozin had no effect on T and B cell composition, we 290 noted a significantly reduced amount of PMN and an increase in macrophages beyond the 291 PDF-mediated level. Concurrently, intraperitoneal cytokine levels measured in effluents by 292 ELISA demonstrated increases of pro-inflammatory markers IL-6, TNF-α and MCP-1 after 293 PDF exposure (Figure 5b). Interferon-and anti-inflammatory interleukin-10 also increased 294 in effluents of PDF-exposed mice compared with saline-treated controls, but there were no 295 significant differences in PDF-exposed animals treated with or without dapagliflozin. Again, 296 similarly with pro-fibrotic changes, there was a non-significant trend towards high glucose-297 independent increase of pro-inflammatory mediators MCP-1 and TNF-α in animals receiving 298 saline and dapagliflozin (Figure 5b). 299 300 Dapagliflozin abrogates pro-inflammatory signaling in murine and human peritoneal 301 mesothelial cells and exerts glucose-independent anti-inflammatory effects on murine 302 peritoneal macrophages 303 As SGLT2 inhibition significantly ameliorated in vivo fibrotic and functional changes and had 304 equivocal effects on inflammatory response, we wanted to further analyze the effects of 305 dapagliflozin on mesothelial cells and macrophages in vitro. In murine omentum-derived 306 mesothelial cells, only SGLT2, but not SGLT1 transcription was upregulated in response to 307 dapagliflozin in a high glucose environment (Figure 5a). 308

13/23
Pharmacological inhibition of SGLT2 decreased both glucose consumption and 309 uptake in HPMC. [15] We therefore asked whether dapagliflozin decreases intracellular 310 glucose content in murine mesothelial cells and macrophages cultured under high glucose 311 conditions. Expectedly, glucose concentration in lysates of both MPMC and macrophages 312 significantly increased in high glucose conditions (Figure 6b). Dapagliflozin   Glucose has been implicated as a major mechanism of peritoneal membrane pathophysiology 344 in PD. [16,17] The chronic peritoneal glucose exposure induces significant systemic sequelae. 345 We have previously shown that daily dialytic glucose exposure is associated with vascular 346 complement and TGF-ß activation and closely correlated with the degree of vasculopathy. In the present study, we demonstrate that both SGLT1 and SGLT2 are expressed in the 364 peritoneal membrane in mice and humans. In mice, chronic exposure of the peritoneal 365 membrane to a high glucose milieu in PDF regulated the expression of glucose transporters 366 such as GLUT1, GLUT3, GLUT4 and SGLT2. We show for the first time that SGLT2 367 inhibition via intraperitoneal application of dapagliflozin ameliorates structural and functional 368 changes in PDF-induced peritoneal fibrosis. Dapagliflozin/PDF treatment reduced peritoneal 369 thickening and fibrosis and improved ultrafiltration compared to animals treated with PDF 370 alone. These changes are in keeping with evidence from other organs where SGLT2 inhibition 371 has been associated with antifibrotic effects, most prominently in the kidney but also in the 372 heart and the liver. The effluent cytokine analyses suggest a VEGF-independent mechanism of reduced 390 peritoneal vascularization by SGLT2 inhibition and argue in favor of pathways such as 391 modulation of angiopoietin 1/2,[31] but determination of tissue cytokine abundance may be 392 more sensitive and valid. 393 As mentioned above, we observed some glucose-independent detrimental side effects 394 of dapagliflozin. The dose of 1 mg/kg of dapagliflozin used in our experiments has been 395 shown as safe and well tolerated in mice if given systemically up to 12 weeks. [32,33] In order 396 to achieve effective concentrations of dapagliflozin in the peritoneal cavity, we used an 397 intraperitoneal way of administration. It should be noted that despite local application of 398 dapagliflozin, systemic action was observed, as reflected by presence of glucosuria in 24 h 399 urine collections of mice treated with the SGLT2 inhibitor. This suggests uptake by peritoneal 400 blood capillaries or by lymphatics, which is not surprising given the low molecular weight of 401 dapagliflozin.
However, we cannot fully exclude intraperitoneal accumulation of 402 dapagliflozin leading to increased local concentration, which might possibly result in local 403 toxic effects. It would be interesting to test whether systemic application of dapagliflozin will 404 still have a protective effect at the PM without possible detrimental glucose-independent side 405 effects observed by local application. In the setting of a high glucose environment, however, 406 we saw significant benefits of additional dapagliflozin application. 407 Despite the marked reduction of peritoneal fibrotic changes and microvessel density 408 with SGLT2 inhibition in our PDF exposure model, dapagliflozin action on glucose-mediated 409 peritoneal inflammatory response was more complex. While cell influx to the peritoneal 410 16/23 cavity was unchanged with regard to T and B cells, we noted a significant reduction of PMN 411 and increase in peritoneal macrophages. At the same time, dapagliflozin administered in the 412 absence of a high glucose milieu showed a trend towards PM thickening (p=0.10), whereas 413 increases in proinflammatory cytokines such as IL-6, TNF-α and MCP-1, and anti-414 inflammatory cytokines such as IL-10 and IFN-were not statistically significant. Given 415 these equivocal in vivo results, we tried to pinpoint the influence of a high glucose milieu and 416 SGLT2 inhibition on inflammatory responses of mesothelial cells and macrophages in vitro. milieus. To this end, we stimulated macrophages with LPS after treating them with either 430 normal or high glucose conditions in presence or absence of dapagliflozin. We observed that 431 when macrophages were not challenged by LPS, dapagliflozin had no effect on 432 proinflammatory marker release, whereas MCP-1 and TNF-α were significantly reduced by 433 SGLT2 inhibition in the presence of an inflammatory stimulus such as LPS. This effect was 434 seen in cells cultured under normal glucose conditions (Figure 6d), suggesting that SGLT2 435 inhibition can shift macrophages to M2 polarization independently from glucose. This 436 observation is in line with a recently published novel mechanism of SGLT2 inhibitors-437 mediated M2 polarization through a glucose-independent reactive oxygen and nitrogen 438 species-dependent STAT3-mediated pathway.
[28] In line with this, when macrophages were 439 cultured under high glucose condition and therefore already shifted to M2 prior to LPS 440 stimulation,[13] anti-inflammatory effects of dapagliflozin were still observed albeit less 441 pronounced and reached statistical significance for MCP-1 only (Figure 7). 442 In summary, our data demonstrate the presence of SGLT2 in the murine and human 443 peritoneum, its regulation by glucose in mice and beneficial effects of its inhibition by 444  Antibody specificity is demonstrated in kidney positive controls, which show specific and distinct staining patterns of the proximal tubule brush-border membrane for SGLT1 (asterisk) and SGLT2 (arrowheads), respectively. Staining of the mesothelial cell layer for SGLT1 and SGLT2, respectively, is denoted by arrows. Blue staining denotes DAPI, scale bar=100 µm.