Prostaglandin I₂ Receptor Agonism Preserves β-Cell Function and Attenuates Albuminuria Through Nephrin-Dependent Mechanisms

The therapeutic tools available to the physician caring for an individual with diabetes are more numerous and more varied today than ever before; yet, the long-term complications of diabetes remain a major cause of morbidity and mortality. Diabetic kidney disease, for instance, continues to be the most common cause of end-stage renal disease worldwide, and although blood glucose–lowering strategies may slow the onset of nephropathy, they provide little benefit once renal dysfunction has become established (1). In efforts to alleviate this burden, attention in the past has been drawn to exploration of the coincidental glucose-independent effects of each new class of antihyperglycemic therapy. This has been the case for the thiazolidinediones (2), glucagon-like peptide 1 (GLP-1) receptor agonists (3), dipeptidyl peptidase 4 (DPP-4) inhibitors (4), sodium–glucose cotransporter 2 (SGLT2) inhibitors (5), and even metformin (6). However, more often than not, initial promises have not been realized. An alternative strategy would be to develop (or reposition) an agent with the a priori assumption that, given its known mechanism of action, the tool will simultaneously and independently improve glucose homeostasis and attenuate end-organ injury.

One structurally essential signaling molecule that is shared between pancreatic β-cells and a renal cell type is nephrin, a member of the immunoglobulin superfamily. Nephrin is a transmembrane protein most widely appreciated for its pivotal role in maintenance of the complex architecture of podocytes, the visceral epithelial cells that form the outermost layer of the glomerular filtration barrier and the final obstacles impeding albumin leakage into the urinary filtrate. The presence of nephrin in pancreatic β-cells has been appreciated for more than a decade (7), but only more recently has the protein been identified as being a key regulator of glucose-stimulated insulin secretion (GSIS) (8) and β-cell viability (9). Because nephrin does not possess a natural ligand (10), efforts to therapeutically alter its actions are best addressed through posttranslational modification of the protein. Nephrin phosphorylation mediates β-cell insulin release (10), whereas its dephosphorylation in the kidney has been associated with proteinuria development in patients and in experimental models (11).

In our search for a therapeutic strategy that alters nephrin phosphorylation, we directed our attentions to the role of the prostaglandin I₂, PGI₂ (IP) receptor. PGI₂ is a cyclooxygenase-derived prostanoid that exerts its cell-specific effects through both IP receptor–dependent and IP receptor–independent pathways. The IP receptor signals predominantly through 3′,5′-cyclic adenosine monophosphate (cAMP)–dependent protein kinase A (PKA) activation (12), a pivotal regulator of the podocyte differentiated state (13). In the current study, we hypothesized that IP receptor–mediated PKA activation would augment nephrin phosphorylation, and by taking advantage of a recently developed first-in-class IP receptor agonist MRE-269 (and its prodrug form selexipag) (14–16), we explored these effects in pancreatic β-cells and in podocytes.

The effects of IP receptor agonism were determined in MIN6 cells and in conditionally immortalized human podocytes (gift of Dr. Moin Saleem) (17). Experiments were conducted in the MIN6 B1 clone (Addex Biotechnologies, San Diego, CA), except nephrin overexpression experiments, which were conducted in MIN6 C3 cells (18), previously reported to constitutively express lower levels of nephrin than B1 cells (8). MIN6 C3 cells were provided by Dr. Philippe Halban (University of Geneva, Switzerland) with permission from Dr. Jun-ichi Miyazaki, University of Osaka, who produced the maternal MIN6 cell line (19). Cells were treated with MRE-269 (Cayman Chemical, Ann Arbor, MI) in 0.1% DMSO at a concentration of 10 µmol/L (14,20), with or without preincubation for 30 min with the PKA inhibitor H89 (10 µmol/L) (Sigma-Aldrich, Oakville, Ontario, Canada) (21), or with the IP receptor inhibitor RO1138452 10 µmol/L (Cayman Chemical), 100 nmol/L insulin (Eli Lilly, Toronto, Ontario, Canada), or 10 µmol/L forskolin (Tocris Bioscience, Bristol, U.K.) for 1 h. IP receptor expression was determined in MIN6 B1 cells, human podocytes, and conditionally immortalized mouse podocytes (22).

The effects of selexipag (14,15) on pancreatic β-cell function were assessed in nondiabetic and streptozotocin (STZ)-induced diabetic male C57BL/6 mice (The Jackson Laboratory, Bar Harbor, ME) aged 8 weeks (n = 8–10 per group). Diabetes was induced by administering a daily intraperitoneal injection of STZ (55 mg/kg) in 0.1 mol/L citrate buffer (pH 4.5) (or citrate buffer control) after a 4-h fast for 5 consecutive days. Mice were randomized to receive the IP receptor prodrug selexipag (3 mg/kg in 0.9% saline containing 12% DMSO by twice daily gavage; Cayman Chemical) (14) or vehicle beginning with the first intraperitoneal injection of STZ. After 2 weeks, fasting blood glucose levels (OneTouch UltraMini; LifeScan Canada Ltd., Burnaby, British Columbia, Canada) were determined, and GSIS was determined by administering an intraperitoneal glucose load (3 g/kg) after a 15-h fast, with blood collected from the saphenous vein 30 min after the injection (23). Plasma insulin was determined by ultrasensitive ELISA (Alpco, Salem, NH).

Diabetes was induced in male C57BL/6 mice, aged 8 weeks (n = 9), by daily intraperitoneal injections of STZ (55 mg/kg) for 5 days, as already described. Mice were treated with selexipag (3 mg/kg) by twice-daily gavage beginning with the first intraperitoneal injection of STZ and were individually housed in metabolic cages for 24 h after 2 and 3 weeks of treatment before determination of urine albumin excretion by AssayMax Mouse Albumin ELISA (Assaypro, St. Charles, MO).

The effects of selexipag on renal function and podocyte integrity were assessed in STZ-induced diabetic endothelial nitric oxide synthase (eNOS) knockout (eNOS^−/−) mice. Male C57BL/6 and eNOS^−/− mice, 8 weeks of age (The Jackson Laboratory) (n = 6–10 per group) were randomized to receive citrate buffer or STZ, as already described, and an additional group of eNOS^−/− mice administered STZ received selexipag (3 mg/kg) by twice-daily gavage beginning with the first intraperitoneal injection of STZ (n = 9). Animals were monitored for 3 weeks. Only mice with a blood glucose >15 mmol/L were considered diabetic. At the end of the study period, systolic blood pressure (SBP) was measured using a CODA noninvasive blood pressure system (Kent Scientific, Torrington, CT) (24). Urine albumin excretion was measured by ELISA after metabolic caging for 24 h. None of the mice in studies one, two, or three received supplemental insulin.

All experimental procedures adhered to the guidelines of the Canadian Council on Animal Care and were approved by St. Michael’s Hospital Animal Care Committee.

Immunoblotting was performed with primary antibodies in the following concentrations: IP receptor 1:1,000 (Cayman Chemical), GAPDH 1:5,000 (Abcam, Cambridge, MA), Y1176 and Y1193 phosphorylated nephrin 1:10,000 (Abcam), total nephrin 1:1,000 (R&D Systems, Minneapolis, MN) or 1:1,000 (Origene, Rockville, MD), phosphorylated PKA-IIα regulatory subunit (Ser 96) 1:1,000 (phospho-PKA-IIα reg; Santa Cruz Biotechnology, Dallas, TX), total PKA-IIα reg 1:1,000 (Santa Cruz Biotechnology), PKAα catalytic subunit 1:1,000 (PKAα cat; Santa Cruz Biotechnology), and cleaved caspase 3 1:1,000 (Cell Signaling Technology, Danvers, MA). Densitometry was performed using ImageJ 1.46r software (National Institutes of Health, Bethesda, MD). Additional controls included immunoblotting mouse podocytes for IP receptor after overnight transfection with 75 pmol/L IP receptor small interfering RNA (siRNA) (Thermo Scientific, Rockford, IL) and preincubating the IP receptor primary antibody with a 10-fold excess of the immunizing peptide (Bio Basic Inc., Markham, Ontario, Canada) for 2 h at 37°C before immunoblotting.

Cells were preincubated with isobutylmethylxanthine (100 µmol/L) (Sigma-Aldrich) for 30 min and treated with MRE-269 (10 µmol/L) or DMSO for 5 min before measurement of cAMP with a commercial kit (R&D Systems).

For determination of gene expression in mice kidneys, tissue was collected from control and STZ-induced diabetic C57BL/6 and eNOS^−/− mice after 3 weeks (n = 5–6 per group). Measurement of gene expression was performed using SYBR Green on an ABI Prism 7900HT Fast PCR System (Applied Biosystems, Foster City, CA). Primers for mouse and human IP receptor were from Origene. Primer sequences for mouse insulin, human RPL32, and mouse RPL13a are given in Supplementary Table 1.

MIN6 cells were washed in PBS and incubated for 45 min at 37°C in sterile filtered Krebs-Ringer buffer (KRB) containing 1 mmol/L glucose, washed in KRB, and then incubated for 60 min in KRB containing varying amounts of glucose (1, 2.8, or 11.1 mmol/L) in the presence or absence of MRE-269 (10 µmol/L). Insulin levels in the culture supernatant were measured by ELISA. PKA inhibition was achieved by incubating cells with 10 µmol/L H89. Nephrin knockdown experiments were performed by transfecting MIN6 cells overnight with 75 pmol/L nephrin siRNA using Lipofectamine 2000 (Life Technologies, Grand Island, NY).

Human wild-type (WT) nephrin fused with monomeric EGFP inserted into the pEZ-M98 vector was obtained from GeneCopoeia (Rockville, MD). End-to-end primers having a mutation in the respective sites were designed using NEBaseChanger. The Q5 Site-Directed Mutagenesis Kit (New England Biolabs Inc., Ipswich, MA) was used to insert the mutations in human nephrin with the two tyrosine residues at positions 1176 and 1193 replaced with phenylalanine separately (Y1176F, Y1193F) or together (Y1176F/Y1193F). The entire nephrin insert was sequenced to confirm the substitution and to exclude unintended mutations (The Centre for Applied Genomics, The Hospital for Sick Children, Toronto, Ontario, Canada). Primer sequences are reported in Supplementary Table 1. The three constructs were separately transfected into MIN6 C3 cells using Lipofectamine 2000.

MIN6 C3 cells transfected with WT-nephrin or mutated Y1176F, Y1193F, or Y1176/Y1193F nephrin were seeded into 96-well plates (1 × 10⁴ cells) in the presence or absence of 300 µmol/L H₂O₂ (Sigma-Aldrich) and 10 µmol/L MRE-269. After 16 h, media were replaced with media containing 5 mg/mL MTT for 3 h before measurement of absorbance at 590 nm.

Immunohistochemistry was performed with antibodies in the following concentrations: insulin 1:800 (Cell Signaling Technology), nephrin 1:200 (R&D Systems), CD2-associated protein (CD2AP) 1:400 (Thermo Scientific), and Wilms tumor 1 (WT1) 1:900 (Santa Cruz Biotechnology). Sections were scanned with the Aperio ScanScope system (Aperio Technologies Inc., Vista, CA). β-Cell mass was calculated as the product of the proportional pancreatic area staining for insulin and the wet pancreas weight (25). The proportional glomerular area positively immunostaining for nephrin or CD2AP was determined in 30 random glomerular profiles in each kidney section using ImageScope 11.1 software (Leica Microsystems, Concord, Ontario, Canada). Nuclei positively immunostaining for WT1 were counted in 30 glomerular profiles from each kidney section (26).

Cortical tissue (1 mm³) was dissected from male C57BL/6 mouse kidneys (n = 3), fixed in paraformaldehyde, and embedded in LR White resin. Ultrathin sections on nickel grids were immersed in 80 mmol/L ammonium chloride in PBS, then in PBS containing 0.2 mol/L glycine and 0.5% BSA, followed by 10% normal goat serum. The grids were incubated overnight at 4°C with the primary antibody (rabbit polyclonal to IP receptor, 300 μg/mL; Cayman Chemical) diluted 1:50 in PBS and 0.5% BSA. The grids were washed in PBS and 0.5% BSA and incubated with 10 nm gold-conjugated goat anti-rabbit IgG diluted 1:20 in PBS and 0.5% BSA at room temperature. A grid without primary antibody was used as a negative control. The grids were washed and stained with 2% aqueous uranyl acetate and examined using a Philips CM100 transmission electron microscope (Electron Microscopy Research Services, Newcastle University, Newcastle upon Tyne, U.K.).

For PKA knockdown experiments, cultured human podocytes were transfected with 50 nmol/L PKAα cat siRNA (Cell Signaling Technology) for 16 h before treatment with MRE-269 (10 µmol/L) or vehicle for 1 h. To determine the effect of MRE-269 on cell death, podocytes were serum starved for 48 h in the presence or absence of MRE-269 (10 µmol/L) or vehicle. Podocyte cell death was evaluated by flow cytometry using a MACSQuant Analyzer (Miltenyi Biotec, Cambridge, MA) after labeling with phycoerythrin-annexin V and 7-aminoactinomycin D (7-AAD) and by immunoblotting for cleaved caspase 3.

Podocytes were treated with vehicle or 10 µmol/L MRE-269 for 1 h, fixed in 4% paraformaldehyde, and permeabilized with 0.1% Triton X-100 before staining with 25 µg/mL Alexa Fluor 488 phalloidin (Thermo Scientific) (24). Images were obtained using an Olympus BX60 fluorescence microscope, and staining intensity was determined in 50 cells for each treatment using Adobe Photoshop CS6 (27).

Transcytosis of albumin was detected using total internal reflection fluorescence microscopy, as recently described (28). Briefly, rhodamine-albumin (Sigma-Aldrich) was allowed to bind to the apical membrane at 4°C for 10 min, after which excess albumin was washed off. Cells were heated at 37°C, allowing bound albumin to be internalized, and then cells were imaged by total internal reflection fluorescence microscopy to visualize albumin-containing vesicles fusing at the basolateral membrane. The signal profiles of every vesicle were tracked in a masked and automated fashion using a MATLAB algorithm to determine whether an exocytosis event had occurred. Before imaging, cells were pretreated with MRE-269 (10 µmol/L) or vehicle for 1 h, with or without overnight transfection with 75 pmol/L nephrin siRNA.

A minimum of 30 glomeruli were examined in periodic acid-Schiff–stained kidney sections from each mouse. The degree of mesangial matrix accumulation was subjectively graded on a scale of 0 to 4, as previously described (26).

Podocyte density and number were determined by transmission electron microscopy. Images of the entire glomerular profile were taken through three randomly selected glomeruli from five to six mice per group, and podocyte density was estimated using the method of Weibel and Gomez (29). Podocyte density was multiplied by mean glomerular volume (determined on periodic acid-Schiff–stained kidney sections [30]) to give an estimate of podocyte number per glomerulus (31).

Data are expressed as mean ± SEM. Statistical significance was determined by one-way ANOVA with a Fisher least significant difference post hoc comparison for comparison of multiple groups and the Student t test for comparison between two groups (or Mann-Whitney test for nonparametric data). Skew-distributed data were log-transformed before statistical comparison. Statistical analyses were performed using GraphPad Prism 6 for Mac OS X (GraphPad Software Inc., San Diego, CA).

In our first series of experiments, we set out to determine whether IP receptor agonism could promote insulin release by pancreatic β-cells. To do this, we exposed MIN6 cells, a pancreatic β-cell line, to the IP receptor agonist MRE-269. We first confirmed expression of the IP receptor in MIN6 cells by immunoblotting (Fig. 1A) and real-time PCR (cycle threshold: IP receptor, 23.9 ± 1.1; RPL13a, 21.3 ± 1.1), observing an ∼1,000-fold increase in intracellular cAMP with MRE-269 treatment (Fig. 1B). This increase in cAMP preceded an increase in insulin mRNA levels (Fig. 1C) and an earlier augmentation in GSIS (Fig. 1D), whereas pretreatment with H89 (an inhibitor of the cAMP-dependent kinase, PKA) prevented MRE-269–augmented GSIS (Fig. 1E). Because the posttranslational modification of nephrin has previously been linked to GSIS, we speculated that the cAMP/PKA-dependent effects of MRE-269 in MIN6 cells could be mediated through nephrin phosphorylation. Immunoblotting confirmed the presence of nephrin in MIN6 cells and an increase in tyrosine 1176/1193 phosphorylation of the protein with MRE-269, whereas pretreatment with H89 prevented nephrin phosphorylation (Fig. 1F) and nephrin knockdown with sequence-specific siRNA (Supplementary Fig. 1) prevented MRE-269–potentiated GSIS (Fig. 1G). To determine whether nephrin phosphorylation and GSIS are causatively related, we overexpressed nephrin in the MIN6 C3 clone (which constitutively expresses lower levels of nephrin [8]) using mutant constructs in which the phosphorylated tyrosine residues 1176 or 1193, or both, had been replaced (Fig. 1H). With this approach, we observed that mutation of Tyr-1176/1193 abrogated the augmentation in GSIS with MRE-269 (Fig. 1I). Finally, because nephrin signaling has recently been linked to β-cell survival, we also determined the effect of MRE-269 on MIN6 cell viability. MRE-269 prevented the decrease in MIN6 viability induced by H₂O₂, and as was observed with GSIS, transfection of MIN6 C3 cells with Tyr-1176/1193 phosphorylation-resistant nephrin negated this effect (Fig. 1J).

Having found that IP receptor agonism promotes insulin release and preserves viability in MIN6 cells, we next questioned whether the same approach could preserve β-cell function in vivo. To investigate this, we administered the IP receptor agonist prodrug selexipag to STZ-C57BL/6 mice. Daily intraperitoneal injections of STZ (55 mg/kg) were administered to mice for 5 days, and selexipag treatment was begun coinciding with the first intraperitoneal injection. Blood glucose levels 2 weeks later were marginally but significantly lower with selexipag treatment than with vehicle (Fig. 2A). This attenuation of hyperglycemia was accompanied by 1) an augmentation in plasma insulin levels 30 min after administration of an intraperitoneal glucose load that equated to an ∼50% increase in residual insulin release compared with STZ-C57BL/6 mice treated with vehicle (Fig. 2B) and 2) a preservation in β-cell mass with selexipag (Fig. 2C–G). Next, to determine whether IP receptor agonism could attenuate albuminuria independently of its effects on β-cell function, we treated a further group of STZ-C57BL/6 mice with selexipag and compared serial blood glucose measurements and urine albumin excretion after 2 and 3 weeks of treatment. Although blood glucose levels increased between 2 and 3 weeks of STZ-induced diabetes in selexipag-treated mice, urine albumin excretion declined (Supplementary Fig. 2).

Observing a glucose-independent effect of IP receptor agonism in attenuating albuminuria in STZ-C57BL/6 mice, we sought to determine whether the IP receptor/cAMP/PKA/nephrin–dependent signaling pathway we had elucidated in MIN6 cells was also active in glomerular podocytes. Immunoblotting and real-time PCR confirmed the presence of IP receptor protein (Fig. 3A and Supplementary Fig. 3) and mRNA (cycle threshold: human podocytes IP receptor, 29.1 ± 0.3; RPL32, 21.6 ± 0.3; mouse podocytes IP receptor, 25.4 ± 0.3; RPL13a, 19.6 ± 0.2) in immortalized podocytes of both human and mouse origin, whereas immunogold electron microscopy confirmed the presence of IP receptor protein in mouse podocytes in vivo (Fig. 3B–F).

Similar to our observations in MIN6 cells, treatment of cultured human podocytes with MRE-269: 1) increased intracellular cAMP levels (Fig. 4A), 2) activated PKA (Fig. 4B) in an IP receptor–dependent manner (Fig. 4C), and 3) phosphorylated nephrin (Fig. 4D). Inhibition of PKA with H89 (Fig. 4E) or knockdown of PKA with siRNA (Fig. 4F) prevented nephrin phosphorylation. This increase in nephrin phosphorylation induced by MRE-269 was accompanied by an increase in podocyte F-actin content (Fig. 4G) and a reduction of ∼50% in the transcytosis of labeled albumin across podocyte monolayers (Fig. 4H), the latter effect being abrogated by nephrin knockdown (Fig. 4H). Podocyte cell death, in contrast, was unaffected by MRE-269 (Supplementary Fig. 4). Finally, because podocytes express the insulin receptor (32), nephrin was recently shown to interact with the B isoform of the receptor (33), and we had observed that IP receptor agonism stimulates β-cell insulin release, we questioned whether insulin itself could promote podocyte nephrin phosphorylation. Immunoblotting cultured human podocytes revealed that nephrin phosphorylation was unaffected by exposure to exogenous insulin, whereas the adenylyl cyclase activator forskolin caused a marked increase (Supplementary Fig. 5).

Having observed that MRE-269 promotes nephrin phosphorylation and attenuates albumin transcytosis in cultured podocytes, in our final series of experiments we set out to verify the glucose-independent attenuation of albuminuria in vivo. For these experiments, we studied the effects of selexipag in diabetic eNOS^−/− mice, one of the mouse models considered to more closely approximate human disease (34,35). We selected this particular model for two additional reasons. Firstly, we earlier reported that STZ-eNOS^−/− mice develop acute, heavy albuminuria and a podocyte-predominant pathology (24). Secondly, eNOS^−/− mice are more susceptible to chemically induced diabetes than WT C57BL/6 mice, despite being on the same strain background (36). We speculated that this susceptibility would enable us to separate the renal effects of selexipag from its actions at the pancreas.

We first determined that renal IP receptor expression did not differ between STZ-eNOS^−/− mice and control groups (IP receptor mRNA:RPL13a mRNA [arbitrary units]: C57BL/6, 1.1 ± 0.2; STZ-C57BL/6, 1.1 ± 0.1; eNOS^−/−, 0.9 ± 0.2; STZ-eNOS^−/−, 0.9 ± 0.2). Separate control and STZ-induced diabetic C57BL/6 and eNOS^−/− mice were then monitored for 3 weeks with an additional group of STZ-eNOS^−/− mice treated with selexipag (Fig. 5). Blood glucose levels were marginally, albeit nonsignificantly, lower in selexipag-treated STZ-eNOS^−/− mice than in untreated STZ-eNOS^−/− mice (Fig. 5D), and there was no difference in SBP (Fig. 5E) or mesangial matrix accumulation (Fig. 5F–K). Despite these slight or absent effects, albuminuria was reduced by >50% with selexipag treatment of STZ-eNOS^−/− mice (Fig. 5L).

To determine whether the antialbuminuric effect of selexipag in STZ-eNOS^−/− mice was accompanied by a change in podocyte structure or podocyte function we 1) surveyed the glomerular architecture by transmission electron microscopy; 2) immunostained kidney sections for nephrin, CD2AP, and WT1; and 3) immunoblotted kidney homogenates for nephrin tyrosine 1176/1193 phosphorylation. Podocyte number and density were reduced in STZ-eNOS^−/− mice and were unaffected by selexipag treatment (Fig. 6A and B), podocyte foot process effacement was also unaffected by selexipag (Supplementary Fig. 6), and total nephrin, CD2AP, and WT1 were unchanged across the groups (Fig. 6C–H and Supplementary Fig. 7). In contrast, nephrin phosphorylation was increased in selexipag-treated STZ-eNOS^−/− mice compared with untreated STZ-eNOS^−/− mice (Fig. 6I).

PGI₂-dependent pathways play important roles in glucose homeostasis and end-organ function under normal circumstances and in diabetes. In the current study, we explored these actions in pancreatic β-cells and in glomerular podocytes. In each case, pharmacological agonism of the PGI₂, IP receptor caused a PKA-dependent increase in phosphorylation of the transmembrane protein nephrin, which augmented GSIS and preserved viability in pancreatic β-cells and attenuated albumin passage in podocytes. These observations demonstrate the feasibility of pharmacologically modulating a pathway that is common to both the pancreatic β-cell and a cell type that is vulnerable to the consequences of pancreatic β-cell dysfunction.

PGI₂ exerts its effects on cellular function in an autocrine/paracrine manner by binding to its cell surface IP receptor and in an intracrine manner by binding to the nuclear receptor, peroxisome proliferator–activated receptor β/δ (PPARβ/δ). The IP receptor may signal through several different second messengers in renal cells (e.g., cAMP, intracellular calcium, diacylglycerol/protein kinase C) (37). However, its most commonly described signaling pathway involves a conformational change in the G-protein–coupled IP receptor, which activates G_s and causes a buildup of intracellular cAMP and the downstream activation of PKA (38). This pathway is well recognized for its role in renal (patho)physiology (39). However, PGI₂ also acts at the level of the pancreatic islet. In the INS1E pancreatic β-cell line, overexpression of PGI₂ synthase enhanced insulin secretion (40), and in transplanted islets, addition of the PGI₂ analog, beraprost sodium, to collagenase solution has been reported to improve viability (41).

Nephrin is an essential component of the glomerular slit diaphragm, which is a modified adherens junction that forms a zipper-like structure joining adjacent interdigitating foot processes of podocytes (42). It helps to maintain the complex architecture of podocyte foot processes preventing the free passage of macromolecules into the urinary space. In addition to its structural role conferred through organization of the actin cytoskeleton, nephrin also acts as a signaling molecule. Of the 10 tyrosine residues present in the cytoplasmic domain of human nephrin, 7 are conserved in mice and rats (43). Phosphorylation of three of these (Tyr-1176, Tyr-1193, and Tyr-1217) by the Src kinase Fyn results in the recruitment of the Nck family of SH2/SH3 adaptor proteins modulating podocyte actin dynamics (43). Analogously, Tyr-1176/1193 phosphorylation of nephrin in MIN6 cells was recently found to enhance GSIS (10). However, by contrast, the additional phosphorylation of Tyr-1217 promoted nephrin degradation (10). Nephrin is found both on the plasma membrane and on insulin vesicles in β-cells (8), and it regulates insulin exocytosis through interaction with the actin cytoskeleton and phosphorylation-dependent trafficking (10). In the current study, we found that IP receptor agonism phosphorylated nephrin at Tyr-1176/1193 in podocytes and in MIN6 cells. Nephrin phosphorylation in MIN6 cells was accompanied by enhanced GSIS and preserved viability, whereas mutation of the phosphorylated Tyr-1176/1193 residues abrogated these effects. Although preservation of islet β-cell mass may explain some of the reduction in blood glucose levels with selexipag treatment in STZ-C57BL/6 mice, a separate secretagogue-like effect is supported by the findings in MIN6 cells. However, unlike secretagogue therapies currently available in the clinic, IP receptor agonism alone is not sufficient to induce insulin release, emphasizing that its effects occur downstream of glucose-sensing, plausibly by priming β-cells through nephrin phosphorylation (8).

Although nephrin is present and functional in the pancreatic β-cell, its most apparent physiological role is in the preservation of glomerular filtration barrier integrity, most readily evidenced by the heavy proteinuria that occurs in individuals who possess a mutation in the nephrin-encoding gene, NPHS1 (44). Unlike the well-established function of nephrin in podocytes, the presence and actions of the IP receptor in podocytes has been uncertain. One earlier report demonstrated IP receptor mRNA in human podocytes in vivo by in situ hybridization (45). However, to our knowledge, its expression in rodent podocytes has not previously been demonstrated. Here, we detected IP receptor mRNA by real-time PCR and protein by immunoblotting in conditionally immortalized podocytes of mouse and human origin. We observed in vivo expression of the IP receptor in mouse podocytes by immunogold electron microscopy, with gold labeling at the plasma membrane and within cytoplasmic vesicles consistent with G-protein–coupled receptor internalization. Moreover, functional activity of the protein was evidenced by a >500-fold increase in cAMP upon incubation of human podocytes with the IP receptor agonist MRE-269, which subsequently resulted in IP receptor–dependent PKA activation, nephrin phosphorylation, and the prevention of albumin transcytosis.

To investigate the effects of IP receptor agonism on albuminuria development in vivo, we studied the actions of selexipag in STZ-eNOS^−/− mice, taking advantage of their increased susceptibility to chemically induced diabetes to help exclude the confounding effects of increased circulating insulin levels. eNOS has been shown to protect against STZ-induced diabetes, possibly through increased islet sensitivity or altered drug uptake through changes in vascular permeability and pancreatic blood flow, although the precise mechanisms have not been defined (36). In our studies, we found that treatment of STZ-eNOS^−/− mice with selexipag augmented renal nephrin phosphorylation and attenuated albuminuria without affecting other indices of podocyte injury or blood pressure, the latter effect being consistent with observations from clinical trials of selexipag in people with pulmonary arterial hypertension (46). Whether augmentation of nephrin phosphorylation in the setting of podocyte loss can preserve renal function in patients has yet to be determined. It is noteworthy, however, that podocytes are not the sole (nor the predominant) site of IP receptor expression in the kidney. The receptor is also present in the vasculature, mesangial cells, macula densa, and tubule epithelial cells. It plays an important role in the regulation of renal hemodynamics, tubule transport, and renin secretion (47), whereas PGI₂ analog therapy may reduce renal fibrosis and inflammation (48). Selexipag treatment improved endothelial dysfunction in a rat model of pulmonary hypertension (14), and endothelial dysfunction is a common occurrence in people with diabetic kidney disease (49). These alternative actions represent possible additional advantages of IP receptor–directed therapies for diabetic kidney disease, and they cannot be excluded as contributing to the anti-albuminuric effect of selexipag observed in the current study. Nevertheless, in vitro at least, a role for nephrin modulation in preventing albumin passage is supported by the attenuation of albumin transcytosis when MRE-269 was applied to podocyte monolayers and an abrogation of this effect with nephrin knockdown.

Over the decades, PGI₂ analogs have been developed and evaluated for their effects in a wide range of conditions. However, the progression of analog therapies to the clinic has been obstructed by off-target actions and unacceptable adverse effects caused by rapid vasodilatory fluctuations (14). The nonprostanoid IP receptor agonist, MRE-269 and its prodrug form, selexipag, which were developed to circumvent these limitations, have an extended elimination half-life (over 4 h in rats) (14) and a binding affinity for the human IP receptor that is 130-fold greater than that for other human prostanoid receptors (20). In the phase 3 GRIPHON (Prostacyclin [PGI₂] Receptor Agonist in Pulmonary Arterial Hypertension) study, selexipag treatment reduced the occurrence of the primary end point of death from any cause or a complication related to pulmonary arterial hypertension (hazard ratio 0.60; 99% CI 0.46–0.78; P < 0.001) (16). Adverse events typical of prostacyclin therapies (e.g., headache, diarrhea, and nausea) occurred more frequently with selexipag and were generally mild to moderate, with only a minority leading to discontinuation of therapy (16). Whether the advancement in small-molecule therapeutics will affect the development of new agents for nonrespiratory diseases remains to be seen.

In summary, the current study describes a pharmacological strategy that induces the posttranslational modification of nephrin. The IP receptor/cAMP/PKA/nephrin pathway modulates insulin secretion from pancreatic β-cells and permselectivity in glomerular podocytes. These observations highlight the feasibility of adopting a therapeutic strategy that simultaneously targets both glucose homeostasis and end-organ function in diabetes.

Funding. These studies were supported by a research grant from the Natural Sciences and Engineering Research Council of Canada (RGPIN-2015-05802) to W.L.L. and by a Diabetes Innovation Award from Novo Nordisk and a Canadian Institutes of Health Research Operating Grant (MOP-133631) to A.A. S.N.B. is supported by a Keenan Family Foundation KRESCENT Post-doctoral Fellowship, by a Heart and Stroke/Richard Lewar Centre of Excellence Fellowship Award, and by a Banting & Best Diabetes Centre Hugh Sellers Postdoctoral Fellowship. S.M. is supported by a Canadian Diabetes Association Post-doctoral Fellowship. A.S.B. is supported by a Queen Elizabeth II/Dr. Arnie Aberman Graduate Scholarship in Science and Technology and a Yow Kam-Yuen Graduate Scholarship in Diabetes Research from the Banting & Best Diabetes Centre.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. S.N.B., S.M., W.L.L., and A.A. designed the experiments. S.N.B., S.M., B.B.B., K.E.W., S.L.A., A.S.B., Y.L., K.T., and P.M.A. conducted the experiments. S.N.B., S.M., K.E.W., A.S.B., P.M.A., W.L.L., and A.A. analyzed and interpreted the data. S.N.B. and A.A. wrote the paper. A.A. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented at the 75th Scientific Sessions of the American Diabetes Association, Boston, MA, 5–9 June 2015.