Podocyte VEGF-A Knockdown Induces Diffuse Glomerulosclerosis in Diabetic and in eNOS Knockout Mice

Vascular endothelial growth factor-a (VEGF-A) and nitric oxide (NO) are essential for glomerular filtration barrier homeostasis, and are dysregulated in diabetic kidney disease (DKD). While NO availability is consistently low in diabetes, both high and low VEGF-A have been reported in patients with DKD. Here we examined the effect of inducible podocyte VEGF-A knockdown (VEGFKD ) in diabetic mice and in endothelial nitric oxide synthase knockout mice (eNOS−/− ). Diabetes was induced with streptozotocin using the Animal Models of Diabetic Complications Consortium (AMDCC) protocol. Induction of podocyte VEGFKD led to diffuse glomerulosclerosis, foot process effacement, and GBM thickening in both diabetic mice with intact eNOS and in non-diabetic eNOS−/−:VEGFKD mice. VEGFKD diabetic mice developed mild proteinuria and maintained normal glomerular filtration rate (GFR), associated with extremely high NO and thiol urinary excretion. In eNOS−/−:VEGFKD (+dox) mice severe diffuse glomerulosclerosis was associated with microaneurisms, arteriolar hyalinosis, massive proteinuria, and renal failure. Collectively, data indicate that combined podocyte VEGF-A and eNOS deficiency result in diffuse glomerulosclerosis in mice; compensatory NO and thiol generation prevents severe proteinuria and GFR loss in VEGFKD diabetic mice with intact eNOS, whereas VEGFKD induction in eNOS−/−:VEGFKD mice causes massive proteinuria and renal failure mimicking DKD in the absence of diabetes. Mechanistically, we identify VEGFKD -induced abnormal S-nitrosylation of specific proteins, including β3-integrin, laminin, and S-nitrosoglutathione reductase (GSNOR), as targetable molecular mechanisms involved in the development of advanced diffuse glomerulosclerosis and renal failure.


INTRODUCTION
Diabetic kidney disease (DKD) is a major complication of both type 1 and type 2 diabetes that leads to renal failure, and the single most frequent cause of end-stage renal disease (ESRD) worldwide (Tuttle et al., 2014). An incomplete understanding of the molecular mechanisms that lead to DKD has precluded the development of effective treatments preventing progression to ESRD (Tufro and Veron, 2012;Reidy et al., 2014). Vascular endothelial growth factor-A (VEGF-A) and nitric oxide (NO) are essential for glomerular filtration barrier homeostasis, and both are disregulated in diabetic nephropathy (Papapetropoulos et al., 1997;Shen et al., 1999;Hohenstein et al., 2006;Tufro and Veron, 2012). Unlike consistently low NO availability in diabetes, both high and low VEGF-A have been observed in patients with DKD (Hohenstein et al., 2006;Baelde et al., 2007;Lindenmeyer et al., 2007). We have shown that podocyte VEGF-A gain-of-function in diabetic mice leads to the development of Kimmelstiel-Wilson-like nodular glomerulosclerosis and massive proteinuria . Similar glomerular phenotype was reported in eNOS deficient type 1 and type 2 diabetic mouse models (Zhao et al., 2006;Nakagawa et al., 2007). Moreover, we showed that VEGF-A gain-of-function in eNOS KO mice also induces nodular glomerulosclerosis, massive proteinuria and renal failure in the absence of diabetes (Veron et al., 2014). These findings demonstrated that NO deficiency and excess VEGF-A have a synergistic deleterious effect that is necessary and sufficient for the development of nodular glomerulosclerosis, the prototypical glomerular phenotype of human advanced DKD (Veron et al., 2014). Endothelial cell or podocyte VEGF-A knockout causes thrombotic microangiopathy in adult mice (Lee et al., 2007;Eremina et al., 2008). Short term VEGF-A knockdown in podocytes induces acute renal failure and proteinuria associated with endotheliosis, mesangiolysis, and microaneurisms , and VEGF-A deletion accelerates DKD in a short term diabetes mouse model (Sivaskandarajah et al., 2012).
Here we examined the effect of podocyte VEGF-A knockdown (VEGF KD ) in diabetic mice and in eNOS −/− :VEGF KD mice. We determined that in the setting of NO deficiency, caused either by diabetic milieu or eNOS knockout, VEGF KD results in diffuse glomerulosclerosis and proteinuria, mimicking human diabetic diffuse glomerulosclerosis of increasing severity. This phenotype is linked to the generation of NO and thiol mediated by changes in S-nitrosoglutathione reductase (GSNOR) and β3-integrin S-nitrosylation that impairs their activity.

B) VEGF KD Diabetic Mice
Diabetes was induced in 6-to 8-week-old male siVEGF mice  (herein called VEGF KD ) by intraperitoneal streptozotocin (STZ) using the low dose AMDCC (Animal Models of Diabetic Complications Consortium) protocol, as previously described Aggarwal et al., 2015). Random blood glucose concentration >300 mg/dl was confirmed a week after the last STZ injection and every 4 weeks along the experiment. Diabetic VEGF KD (DM-VEGF KD ) and non-diabetic (non-DM-VEGF KD ) mice were fed standard (−dox) or doxycycline containing chow (+dox) for 12 weeks to induce VEGF-A knockdown. At the end of the study 24 h urine was collected in metabolic cages; blood and kidney samples were obtained under anesthesia, as we previously described Veron et al., 2012;Veron et al., 2014). All experimental protocols were approved by the Institutional Animal Care and Use Committee at Yale University School of Medicine.

Functional Parameters
Random blood glucose was measured by glucose oxidase biosensor (OneTouch Ultra-2; LifeScan), and BP was measured under anesthesia and analyzed using PowerLab/8SP system (Chart; AD Instruments, Colorado Springs, CO, Unites States) as previously described Veron et al., 2012). Plasma and urine creatinine were measured by HPLC, and glomerular filtration rate (GFR) was assessed by creatinine clearance. Albuminuria was evaluated by Coomassie blue staining and measured by ELISA (Albuwell-M, Exocell), plasma and urine VEGF-A were quantified by ELISA (R&D), NO was measured by colorimetric assay (Cayman), as previously described (Veron et al., 2014), and urine thiols (Cys and GSH) were measured by fluorometric assay (Cayman), following manufacturers' protocol.

Statistical Analyses
Data are expressed as mean ± SEM unless otherwise stated. Statistical significance (p < 0.05) was determined using Prism 8 software by unpaired t test with Welch's correction and one-way Brown-Forsythe ANOVA to compare two or multiple experimental groups, respectively. Mann-Whitney test was used to analyze non-parametric variables.

VEGF-A and NO
To gain insight into the availability of NO and VEGF-A systemically and at the glomerular filtration barrier we measured VEGF-A and NO in plasma and urine. We determined that plasma VEGF-A and urinary excretion are similarly elevated in eNOS −/− :VEGF KD mice, irrespective of podocyte VEGF KD (Figures 6A,B, red bars). In diabetic mice podocyte VEGF KD decreased plasma VEGF-A ( Figure 6A, blue bar), which remained significantly higher (~2-fold) than in nondiabetic mice ( Figure 6A, white/gray bars). Urine VEGF-A excretion was not altered in diabetic mice, irrespective of podocyte VEGF KD ( Figure 6A, blue bars). Conversely, podocyte VEGF KD in non-diabetic mice with intact eNOS significantly decreased VEGF-A excretion ( Figure 6B, white/ gray bars). As expected, NO plasma level was low in eNOS −/− :VEGF KD mice ( Figure 6C, red bars). In mice with intact eNOS, NO plasma level was higher in diabetic (blue bars) than in nondiabetic mice (white bar), but VEGF KD did not significantly decrease plasma NO in any experimental group ( Figure 6C). Surprisingly, NO urinary excretion was similar in non-diabetic mice with deficient or intact eNOS, VEGF KD increased NO excretion two-fold in eNOS −/− :VEGF KD mice, whereas NO excretion increased dramatically (>6 fold) in diabetic mice, irrespective of podocyte VEGF KD ( Figure 6D). No correlation was detected between VEGF-A and NO plasma levels in any experimental group, nor between VEGF-A or NO and albuminuria or creatinine clearance. These findings suggest that urinary NO excretion is not determined only by eNOS or VEGF-A and that the diabetic milieu elicits higher systemic NO and increases NO excretion in the urine, involving additional factors.

DISCUSSION
This study demonstrates that in the setting of bioavailable NO deficiency, caused by diabetic milieu or by eNOS knockout, podocyte VEGF-A knockdown results in diffuse glomerulosclerosis and proteinuria of increasing severity, leading to renal failure in eNOS −/− :VEGF KD mice. We show that podocyte VEGF KD and eNOS −/− induce severe diffuse glomerulosclerosis in the absence of diabetic milieu. Podocyte VEGF KD in diabetic mice prevents diabetes-induced glomerulomegaly but causes diabetic diffuse glomerulosclerosis.
Mechanistically, we show that compensatory local NO and thiols generation prevent severe proteinuria and GFR loss in diabetic mice with intact eNOS, and we identify abnormal S-nitrosylation of specific proteins, including GSNOR, laminin, and β3-integrin, as novel molecular pathways potentially involved in advanced diffuse glomerulosclerosis.
High circulating VEGF-A in diabetic mice stimulates NOS leading to NO production, protecting the integrity of the  (Du et al., 2001;Nakagawa, 2008;Tufro and Veron, 2012). VEGF-A is a survival factor for all glomerular cell types and stimulates endothelial and mesangial cell proliferation (Tsurumi et al., 1997;Feliers et al., 2005;Guan et al., 2006;Lee et al., 2007), and thereby mediates glomerular hypertrophy and angiogenesis in DKD (Farquhar et al., 1959;Stout et al., 1993;Flyvbjerg et al., 2002;Veron et al., 2010;Veron et al., 2011;Tufro and Veron, 2012). Here we show that in diabetic mice podocyte VEGF KD abrogates VEGF-A-mediated glomerular hypertrophy, leading to diffuse glomerulosclerosis with modest albuminuria and normal creatinine clearance. In contrast, eNOS −/− :VEGF KD mice have a decreased ability to increase NO when podocyte VEGF KD is induced, despite similarly elevated circulating VEGF-A, thereby becoming more susceptible than diabetic mice to deleterious effects of local VEGF KD , resulting in mesangiolysis, extensive podocyte foot process effacement, GBM thickening, and a notably severe diffuse glomerulosclerosis phenotype reminiscent of advanced diabetic diffuse glomerulosclerosis (Farquhar et al., 1959;Tsurumi et al., 1997;Nakagawa, 2008). Moreover, VEGF KD and eNOS deficiency have a synergistic effect exacerbating proteinuria (>15 fold either individual genotype) and leading to renal failure, consistent with the more severe morphologic phenotype.
Previous studies demonstrated that glomerular hypertrophy and hyperfiltration occurring in diabetic mice are VEGF-A dependent (Flyvbjerg et al., 2002;Veron et al., 2010;Veron et al., 2011;Tufro and Veron, 2012), and showed that short term podocyte VEGF knockdown results in decreased glomerular size in non-diabetic mice . Here we extend this observation documenting that long term podocyte VEGF-A knockdown leads to significant decrease in glomerular size in non-diabetic mice and abrogates the glomerulomegaly typically observed in diabetic mice. Diabetic mice with podocyte VEGF KD developed diffuse glomerulosclerosis associated with inflammatory infiltrates and no evidence of endothelial injury or thrombotic microangiopathy (TMA). This phenotype is partially similar to that described in diabetic VEGF-A knockout mice (Sivaskandarajah et al., 2012), suggesting a dose effect of VEGF-A loss-of-function. Most mouse models of DKD show glomerular hypertrophy, mesangial, and extracellular matrix expansion (reviewed in (Brosius et al., 2009;Alpers and Hudkins, 2011). In contrast, few mouse models show advanced diabetic nodular glomerulosclerosis (Zhao et al., 2006;Nakagawa et al., 2007;Hudkins et al., 2010;Alpers and Hudkins, 2011;Veron et al., 2011;Takahashi and Harris, 2014;Aggarwal et al., 2015) or diabetic diffuse glomerulosclerosis (Alpers and Hudkins, 2011;Sivaskandarajah et al., 2012). To our knowledge, the mechanisms leading to such distinct glomerular lesions remain undefined. eNOS KO mice are susceptible to developing renal failure in the setting of diabetes (Zhao et al., 2006;Nakagawa et al., 2007;Hudkins et al., 2010;Kakoki et al., 2010;Alpers and Hudkins, 2011;Yuen et al., 2012;Takahashi and Harris, 2014), reduced renal mass (Nakayama et al., 2009), and VEGF-A gain-offunction (Veron et al., 2014). We have previously shown that podocyte VEGF-A gain-of-function in eNOS KO mice causes massive proteinuria and renal failure (Veron et al., 2014), not unlike those described here in eNOS −/− :VEGF KD + dox mice, illustrating that a relatively narrow range "normal" VEGF-A expression and signaling at the glomerular filtration barrier are required to maintain GFR and selective permeability, as has been previously observed in other genetic and experimental models (Eremina et al., 2008;Sivaskandarajah et al., 2012;Yuen et al., 2012). Despite the similar functional consequences of podocyte VEGF-A gain-of-function and knockdown in eNOS −/− mice, their morphologic phenotypes are strikingly different and parallel two histologic variants of DKD described in humans: nodular or diffuse glomerulosclerosis, respectively (Farquhar et al., 1959;Stout et al., 1993). These mouse models provide the opportunity to examine the molecular pathogenic mechanisms leading to nodular or diffuse glomerulosclerosis, which are poorly understood in humans. The Kimmelstiel-Wilson-like nodular glomerulosclerosis reported in eNOS −/− mice with excess glomerular VEGF-A is associated with decreased laminin S-nitrosylation (Veron et al., 2014). Here we demonstrate that the severe diffuse glomerulosclerosis observed in eNOS −/− :VEGF KD (+dox) mice is associated with increased S-nitrosylation of glomerular proteins. As opposed to loss of laminin S-nitrosylation in the setting of excess VEGF-A (Veron et al., 2014), podocyte VEGF-A knockdown increased laminin S-nitrosylation in eNOS −/− :VEGF KD (+dox) mice associates with severe diffuse glomerulosclerosis, suggesting that laminin nitrosylation might prevent the development of glomerular nodules, probably by regulating the secretion or polymerization of 521-laminin heterotrimers (Cheng et al., 1997).
Reversible S-nitrosylation of specific Cys residues, like Tyr phosphorylation, regulates protein-protein interactions and modulates protein function Hess and Stamler, 2012). We have recently shown that diabetic milieu dysregulates S-nitrosylation of other relevant podocyte proteins: myosin9A, RhoA and actin, activating RhoA and disrupting podocyte function in a partially reversible manner (Li et al., 2021). Thus, we examined additional S-nitrosylated proteins expressed in the kidney. GSNOR is a ubiquitous denitrosylase whose function is regulated by S-nitrosylation (Liu et al., 2001;Guerra et al., 2016;Stomberski et al., 2019). De-nitrosylation reduces GSNOR enzymatic activity in mouse cells and tissues (Brown-Steinke et al., 2010) and leads to GSNO accumulation, representing a major source of NO independent of NOS (Liu et al., 2001;Stomberski et al., 2019), although in vitro purified GSNOR or plant extracts decrease reductase activity upon exposure to NO donors (Guerra et al., 2016). GSNOR decreased activity was recently reported in type 2 diabetes patients and was shown to contribute to hepatic insulin resistance in an obesity mouse model (Qian et al., 2018). We determined that SNO-GSNOR was significantly decreased in eNOS −/− :VEGF KD (+dox) and diabetic VEGF KD (+ dox) mice. Consistent with GSNOR de-nitrosylation, we detected several fold increase in urine NO, GSH-, and Cys-thiols excretion in eNOS −/− :VEGF KD (+dox) and DM-VEGF KD (+dox) diabetic mice. The precise cellular origin of urine NO and thiols (ultrafiltrate, glomerular, or tubular cells) remains to be determined. We posit that GSNOR de-nitrosylation underlies the compensatory mechanism providing an alternative NO source in diabetic and eNOS −/− :VEGF KD (+dox) mice (Figure 8). This compensatory mechanism may support normal renal function and relatively low albuminuria in DM-VEGF KD (+dox) mice, but does not prevent the development of diffuse glomerulosclerosis. The SNO-GSNOR mediated alternate source of NO supports renal function in eNOS −/− :VEGF KD (− dox) mice, but it fails to do so when podocyte VEGF KD is induced (+ dox), leading to massive proteinuria and renal failure, as well as severe diffuse glomerulosclerosis, suggesting incomplete compensation or an FIGURE 8 | Proposed model of podocyte VEGF KD driven diffuse glomerulosclerosis in DM-VEGF KD and eNOS:VEGF KD mice. (A) Strong compensatory NO and thiol generation prevents GFR loss, attenuates proteinuria and diffuse glomerulosclerosis in diabetic VEGF KD mice, while limitation of this compensatory mechanism in eNOS:VEGF KD mice worsens the renal phenotype, leading to renal failure. (B) Reduced GSNOR S-nitrosylation increases GSNO and promotes increased S-nitrosylation of proteins, altering their signaling pathways: (C) decreased nephrin and VEGFR2 signaling and high SNO-β3-integrin inhibit β3-integrin activity leading to podocyte and endothelial cell injury; high SNO-laminin and low VEGFR2 signaling may contribute to the severe diffuse glomerulosclerosis described herein in eNOS:VEGF KD + dox mice.
Frontiers in Pharmacology | www.frontiersin.org February 2022 | Volume 12 | Article 788886 additional VEGF KD related pathway, including iNOS activation, which we have not evaluated. GSNOR function is influenced by subcellular localization and modulated by VEGF and NOS signaling (Stomberski et al., 2019). The novel finding that VEGF KD increases β3-integrin S-nitrosylation in eNOS −/− glomeruli might be linked to diffuse glomerulosclerosis. Laminin-521, the mature GBM laminin, binds αvβ3-integrin through interaction between α5-laminin and β3-integrin, transducing FGF and VEGF signals (Genersch et al., 2003). S-nitrosylation of β3integrin causes conformational changes that lead to decreased integrin signaling (Walsh et al., 2007). β3integrin S-nitrosylation in endothelial cells induces loss of integrin activity (Walsh et al., 2007;Robinson et al., 2009). We previously showed that VEGF KD decreases αvβ3-integrin activity in non-diabetic kidneys and cultured podocytes . Here we find that VEGF KD increases glomerular β3-integrin S-nitrosylation in eNOS −/− :VEGF KD (+dox) mice, likely decreasing β3-integrin signaling. Decreased β3-integrin inside-out activation disrupts nephrin-VEGFR2-β3 integrin signaling in podocytes Veron et al., 2012), as well as VEGFR2-β3 integrin signaling in endothelial cells (Robinson et al., 2009), leading to podocyte and endothelial injury, and eventually to diffuse glomerulosclerosis, as observed in eNOS −/− :VEGF KD (+ dox) mice. ( Figure 8C). Whether increased S-nitrosylation impairs binding of β3integrin and laminin-521 remains to be determined. Both decreased (Yoo et al., 2015) and increased (Maile et al., 2014) β3-integrin activity have been implicated as a mechanism of diabetic kidney disease, suggesting a context dependent role. Blockade of αvβ3-integrin activity by a monoclonal antibody improved early markers of diabetic nephropathy in pigs (Maile et al., 2014) probably by interfering with excessive VEGF-A signaling (Robinson et al., 2009;Bertuccio et al., 2011). Thus, we propose that in the setting of VEGF KD and NO deficiency, low β3-integrin activity associated with increased S-nitrosylation of β3-integrin and laminin impair growth and survival signals, resulting in severe glomerular filtration barrier disruption, leading to massive proteinuria and renal failure ( Figures 8B,C).
Collectively, these findings suggest that S-nitrosylation contributes to the tight regulation of glomerular homeostasis by modulating several important signaling pathways in DKD models. Our findings support a model whereby laminin S-nitrosylation is instrumental to prevent glomerular nodule development, while GSNOR denitrosylation and increased β3-integrin S-nitrosylation lead to diffuse glomerulosclerosis in the setting of low podocyte VEGF-A.
Further studies are needed to address several limitations of this study: evaluate diabetic eNOS −/− :VEGF KD mice, perform a broad molecular phenotyping, confirm in cultured glomerular cell types the S-nitrosylation abnormalities identified in eNOS −/− :VEGF KD + dox kidneys and assess SNO-protein dysregulation in diabetic mice. Such additional studies will provide insight into how S-nitrosylation modulates several signaling pathways that are critical for glomerular homeostasis in DKD.
In summary, VEGF KD in eNOS −/− :VEGF KD mice causes renal failure, massive proteinuria, and severe diffuse glomerulosclerosis in the absence of diabetes. VEGF KD in diabetic mice with intact eNOS prevents diabetes-induced glomerulomegaly, causes diabetic diffuse glomerulosclerosis, and compensatory NO generation attenuates proteinuria and prevents GFR loss. Together, these models are reminiscent of human DKD phenotypes associated with low VEGF-A expression (Baelde et al., 2007;Lindenmeyer et al., 2007). Mechanistically, VEGF KD in eNOS −/− :VEGF KD mice induces increased glomerular β3-integrin S-nitrosylation, likely disrupting nephrin-VEGFR2-β3-integrin signaling (Genersch et al., 2003;Robinson et al., 2009;Bertuccio et al., 2011;Veron et al., 2012). Our observations highlight a potentially targetable novel regulatory pathway that protects the glomerular filtration barrier up to a point in mouse models that mimic human DKD.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusion of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The animal study was reviewed and approved by Institutional Animal Care and Use Committee at Yale University School of Medicine.