Overexpression of Circulating Soluble Nogo-B Improves Diabetic Kidney Disease by Protecting the Vasculature

Diabetic nephropathy (DN), the leading cause of end-stage renal disease in the Western world, is characterized by structural changes in the kidney glomerular filtration barrier (1,2). This leads to enhanced glomerular permeability manifested as albuminuria, representing a common mechanism for renal and extrarenal diabetic vascular complications (3).

A complex network of vascular growth factors regulates the permeability and structure of the glomerular capillary filtration barrier (4). Glomerular levels of vascular endothelial growth factor-A (VEGF-A) and angiopoietin-2 (Angpt2) are upregulated in the early stages of DN, while Angpt1 is downregulated (5–7), a milieu associated with vascular remodeling, endothelial proliferation, and increased capillary permeability (1,4). Blockade of VEGF-A signaling (8) or restoration of Angpt1 levels in podocytes (7) ameliorates albuminuria and glomerular damage in rodent models of early DN. The effects of vascular growth factors on endothelial permeability in DN are partly mediated by nitric oxide (NO) signaling through modulation of endothelial NO phosphorylation (eNOS^Ser1177), which acts in an AKT-dependent manner (9). In diabetes, reduction in NO availability due to eNOS uncoupling (10) has been implicated in the pathophysiology of DN. Podocyte-specific overexpression of Angpt1 activates eNOS (7) in diabetic mice, whereas the beneficial effect of VEGF-A blockade on albuminuria in DN is prevented in eNOS knockout mice (11).

Another pathway involved in vascular remodeling is the neurite outgrowth inhibitor (Nogo) family, which is encoded by one gene with three major isoforms: Nogo-A, -B, and -C (12), mainly expressed in the endoplasmic reticulum (13). Nogo-A and -C are found in the central nervous system and muscle tissue, respectively, while Nogo-B localizes to endothelial and smooth muscle cells within the vasculature (12). In physiology, loss of Nogo-B upregulates eNOS-NO and flow-mediated vasodilation, leading to hypotension (14). Mice lacking both Nogo-A and -B are viable and do not have major apparent vascular defects (15). However, vascular lesions are enhanced in Nogo A/B-deficient mice following injury, which can be prevented by gene delivery of full-length Nogo-B (15,16).

The full-length Nogo-B protein of 49 kDa can be cleaved into a shorter ∼150-amino acid N-terminus fragment (17), which can then be secreted into the circulation as soluble Nogo-B (sNogo-B) (18). This Nogo-B N-terminus (identical in circulating sNogo-B and full-length Nogo-B within the cells) binds to its receptor NgBR, expressed in endothelial cells (ECs) on the cell plasma membrane and in the endoplasmic reticulum, leading to EC proliferation/vascular remodeling (15,19), angiogenesis during development, vascular repair, and cytoskeletal organization (12,20–23). Given the role of this N-terminal fragment of Nogo-B in vascular remodeling, we hypothesized that overexpression of sNogo-B in the circulation could have a protective role in the setting of diabetic kidney disease.

Materials and chemicals were purchased from Sigma-Aldrich (Gillingham, U.K.) and STARLAB (Milton Keynes, U.K.) unless otherwise stated.

To induce diabetes, 8- to 12-week-old (∼20 g in weight) male DBA2J mice were administered with streptozotocin (low-dose multiple-injection protocol) (7,8). Mice were considered diabetic with a fed glycemia >22 mmol/L. Control nondiabetic littermates were injected with vehicle only (citrate buffer). Two weeks later, some diabetic and nondiabetic mice were administered an adeno-associated viral vector expressing 6xHis-Tag/sNogo-B (AAV–sNogo-B) (Supplementary Fig. 1A). The used vector, AAV/DJ, has a specific tropism for the liver and maintains a sustained expression of transgene for 15–17 weeks under the CMV promoter (24). The construct also contains a secretory alkaline phosphatase peptide, which drives the release of the 6xHis-Tag/sNogo-B protein in the circulation. To control for infection, other diabetic and nondiabetic mice were injected with AAV/DJ, driving the expression of GFP under the same promoter (AAV-GFP).

All mice were maintained for 12–14 weeks after induction of diabetes before sacrifice and tissues analysis. Prior to sacrifice, blood pressure was assessed with tail cuff methodology, and 24-h urine collection was conducted with mice kept in metabolic cages for creatinine, sNogo-B, and albuminuria determination (7). At sacrifice, full blood and plasma was collected (to assess for HbA_1c, creatinine by high-performance liquid chromatography, and sNogo-B levels). Kidney tissue was harvested for histology, electron microscopy (EM), and lysates of kidney cortex or isolated glomeruli frozen for further analysis (7).

Glomerular Ki67/CD31 and CD31-positive glomerular ECs (GECs) per glomerulus were visualized with a fluorescent microscope (7); an average of 30–40 glomeruli per animal were studied and the means values used in the analysis.

Lectin staining was assessed as indicator of thickness/integrity of the endothelial glycocalyx in frozen kidney sections as described (25). A total of 10–15 capillary loops were analyzed per mouse from five to seven glomeruli, and the mean value for each animal was used in the analysis.

Mesangial volume fraction, glomerular basement membrane (GBM) thickening, podocyte number, and glomerular volume were studied as described (7,8). For detailed methodology, see the Supplementary Material.

sNogo-B in plasma, urine, and cell culture media was assessed by ELISA (BioLegend, San Diego, CA). ELISA was also used for kidney cortex VEGF-A (R&D Systems, Abingdon, U.K.), Angpt1/2 (Biomatik, Wilmington, DE), and VEGFR2 phosphorylation (PathScan; Cell Signaling Technology, Leiden, the Netherlands) levels; results were normalized for micrograms of total kidney cortex protein lysates.

Immunoblotting was performed on cells and tissue lysates (mouse renal cortex and isolated glomeruli) as described (7). The following primary antibodies were used: Nogo-B (N-terminus; R&D Systems); NgBR (Abcam, Cambridge, U.K.); α-tubulin (Santa Cruz Biotechnology, Heidelberg, Germany); β-actin; pan-AKT, GSK3β, phospho-AKT (Ser⁴⁷³), phospho-eNOS (Ser¹¹⁷⁷), and phospho-GSK3β^Ser9 (Cell Signaling Technology); eNOS (Santa Cruz Biotechnology); and total β-catenin (Proteintech, Manchester, U.K.).

Primary lung ECs were isolated from adult C57BL/6J mice and cultured, up to three passages, as described (14).

Human GECs and podocytes were cultured as described (26,27). To examine the effect of high glucose (HG) and VEGF-A in fully differentiated GECs, cells were starved for 12 h in 1% serum followed by incubation in normal glucose (NG; 5 mmol/L glucose plus 20 mmol/L mannitol) or HG (25 mmol/L glucose) for 72 h in the presence or absence of VEGF-A (50 ng/mL).

For immunoprecipitation (IP) experiments, GECs were transfected with 6xHis-Tag/sNogo-B adenovirus (ADV–sNogo-B, a transgene identical to the AAV–sNogo-B construct) at a multiplicity of infection of 100 for 4 h and studied after 24 h; whole protein lysates were obtained and incubated with anti-NgBR (Novus Biologicals, Oxford, U.K.), anti–6xHis-Tag antisera (Thermo Fisher Scientific, Oxford, U.K.), or vehicle IgG. Immunoblotting was conducted on immunoprecipitates obtained with anti–6xHis-Tag and -NgBR antisera.

For the proximity ligation assay (PLA) experiments, cell culture media enriched with 6xHis-Tag/sNogo-B was obtained by transfecting confluent human GECs with ADV–sNogo-B (multiplicity of infection of 100) in full media for 4 h; subsequently, cells were put in 1% serum media and supernatant (containing 6xHis-Tag/sNogo-B protein, ∼6,000 pg/mL) collected after 48 h; collected media was stored at 4°C and used within 6–12 h. Human differentiated GECs were then incubated with 6xHis-Tag/sNogo-B–conditioned media for 1, 5, and 15 min, fixed in 4% paraformaldehyde for 10 min, and then incubated with primary antisera against NgBR (1:100 dilution, rabbit anti–N-terminus NgBR, gift from R.Q.M.) and anti–6xHis-Tag (1:100 dilution) mouse monoclonal antiserum (Thermo Fisher Scientific). The NgBR–sNogo-B interaction was visualized with immunofluorescence following the manufacturer’s instructions (Duolink In Situ Red Starter Kit Mouse/Rabbit).

Nogo-B immunogold staining was conducted in mouse renal cortex tissue as previously described using a sheep polyclonal anti–Nogo-B (N-terminus) antibody (R&D Systems) (8).

Nogo-B immunohistochemistry was conducted in mice and humans (patients with DN or thin basement membrane nephropathy [TBMN]) (Supplementary Table 1) kidney tissue paraffin sections using the streptavidin-biotin complex method with specific sheep polyclonal anti–Nogo-B N-terminus antibody (R&D Systems).

In human tissue, Nogo-B expression was quantified by assessing the percentage glomerular area with positive staining. Ten glomeruli were analyzed for each human biopsy, and the calculated average Nogo-B staining/area of glomeruli was then used for analysis.

To examine the continuous paracrine/autocrine effect of sNogo-B on angiogenesis, human umbilical vein ECs (HUVECs) were transfected with ADV–sNogo-B or identical vector lacking sNogo-B cDNA (control vector) (Supplementary Fig. 1B) for 4 h; cells were then equilibrated (24 h) in complete medium and then plated in 96-well plates (5,000 cells/well) in EGM basal media containing sera (4% v/v) obtained from patients with type 1 diabetes mellitus (T1DM) susceptible to (DN+) or protected toward (DN−) the progression of DN (Supplementary Table 2). Tube formation was performed on Matrigel (BD Biosciences, Berkshire, U.K.) (28) in duplicate and assessed after 24 h. Tube length and number were analyzed in a blinded fashion with the Wimasis WimTube image system (29).

Differences among groups were analyzed by two-tailed Student t test or ANOVA with post hoc least significant difference (LSD) pairwise comparisons test for normally distributed variables. Kruskal-Wallis and Mann-Whitney nonparametric tests were used for not normally distributed variables (albuminuria and sNogo-B). Data are expressed as means ± SD for normally distributed data or as median and interquartile range for not normally distributed data. Analysis was conducted with IBM SPSS-22 software (IBM, New York, NY), and statistical significance was accepted at P ≤ 0.05.

Nondiabetic (ND) and diabetic (D) mice injected with AAV–sNogo-B were characterized by a 12-fold increase in circulating sNogo-B when compared with those injected with control vector (AAV-GFP) (ND-GFP vs. ND–sNogo-B, P = 0.00001; D-GFP vs. D–sNogo-B, P = 0.0001) (Fig. 1A).

Diabetes led to a significant increase in HbA_1c levels (ND-GFP vs. D-GFP, P = 0.00008; ND–sNogo-B vs. D–sNogo-B, P = 0.00004), which was paralleled by a loss of body weight (ND-GFP vs. D-GFP, P = 0.0001; ND–sNogo-B vs. D–sNogo-B, P = 0.00001), increased kidney/body weight ratio (ND-GFP vs. D-GFP, P = 0.006; ND–sNogo-B vs. D–sNogo-B, P = 0.0002), and a modest decrease in systolic blood pressure (ND-GFP vs. D-GFP, P = 0.01; ND–sNogo-B vs. D–sNogo-B, P = 0.003) (Table 1).

Glomerular permeability, measured by albumin excretion rate over 24-h urine collection, was increased in D mice compared with ND mice (ND-GFP vs. D-GFP, P = 0.0001) (Fig. 1B). Overexpression of sNogo-B in the circulation of D mice led to a significant 40–50% reduction in albuminuria (D-GFP vs. D–sNogo-B, P = 0.04) (Fig. 1B). This effect was independent of changes in systemic blood pressure, kidney/body weight, and glycemic control, which were similar in D mice administered either AAV–sNogo-B or AAV-GFP (Table 1). Diabetes-mediated renal hyperfiltration was evidenced by raised creatinine clearance levels (Fig. 1C) (ND-GFP vs. D-GFP, P = 0.02), and AAV–sNogo-B overexpression attenuated this effect (D-GFP vs. D–sNogo-B, P = 0.04) (Fig. 1C).

Urinary sNogo-B was measured in a subset of animals. Diabetes was paralleled by an increase in 24-h urine sNogo-B in both mice administered AAV-GFP or AAV–sNogo-B (Fig. 1D) (ND-GFP vs. D-GFP and ND–sNogo-B vs. D–sNogo-B, P ≤ 0.002); no differences were observed in urine sNogo-B between AAV-GFP or AAV–sNogo-B mice within the ND and D group.

The results above implicate sNogo-B as having a protective role in glomerular capillary permeability in diabetic kidneys. We therefore explored whether elevated circulating levels of sNogo-B might stabilize the glomerular vasculature and confer a healthier GEC and podocyte phenotype in diabetes.

Firstly, we examined GEC proliferation to assess the effect of sNogo-B on abnormal vascular remodeling that occurs in early diabetic glomerulopathy (1,7) (Fig. 2A and B). Diabetes increased the number of CD31⁺/Ki67⁺ cells per glomerulus by sevenfold, which was attenuated in mice with sNogo-B overexpression in the circulation (ND-GFP vs. D-GFP, P = 0.0001; D-GFP vs. D–sNogo-B, P = 0.005) (Fig. 2B). The total number of CD31⁺ cells/glomeruli was not different between animals with increased sNogo-B in the circulation in either ND or D mice (mean ± SD CD31⁺ cells/glomeruli: ND-GFP, 7.6 ± 1.37; ND–sNogo-B, 7.7 ± 1.8; D-GFP, 9.7 ± 2.4; D–sNogo-B, 9.5 ± 3.0; n = 7 animals/group, P = NS).

We performed lectin staining as an indicator of thickness/integrity of the glomerular endothelial glycocalyx, which is known to contribute to vascular permeability (30). The estimated lectin thickness was significantly reduced in D mice but restored to baseline levels in mice administered AAV–sNogo-B (ND-GFP vs. D-GFP, P = 0.009; D-GFP vs. D–sNogo-B, P = 0.0004) (Fig. 2C).

Using EM (Supplementary Fig. 2), we found that diabetes was also paralleled by a significant decrease in the number of podocytes per glomerulus (Fig. 2D) (ND-GFP vs. D-GFP, P = 0.0002), increased the mesangial volume fraction (Fig. 2E) (ND-GFP vs. D-GFP, P = 0.02), but had no effect on GBM width (Fig. 2F) or glomerular volume (Table 1). The reduction in podocyte number observed in D mice was ameliorated by elevated circulating sNogo-B levels (D-GFP vs. D–sNogo-B, P = 0.01) (Fig. 2D). Overexpression of sNogo-B in the circulation did not alter diabetes-mesangial expansion (Fig. 2E), GBM width (Fig. 2F), or glomerular volume in diabetic mice (Table 1).

To identify the molecular mechanisms by which sNogo-B might alter the endothelium in diabetes, we examined levels of the vascular growth factors VEGF-A, Angpt1, and Angpt2 in kidney cortex lysates. VEGF-A protein levels were significantly increased in D mice (Fig. 3A) (ND-GFP vs. D-GFP, P = 0.009) and were reduced when sNogo-B was overexpressed in the circulation (D-GFP vs. D–sNogo-B, P = 0.03) (Fig. 3A).

In accord with these findings, phosphorylation levels of the main VEGF-A receptor, VEGFR2, in the kidney cortex were significantly elevated in D mice, an effect that was dampened in mice with elevated circulating sNogo-B levels (ND-GFP vs. D-GFP, P = 0.003; D-GFP vs. D–sNogo-B, P = 0.01) (Fig. 3B). Angpt1/Angpt2 ratio was lower in D mice and was not altered by sNogo-B overexpression in the circulation (ND-GFP vs. D-GFP, P = 0.03; ND–sNogo-B vs. D–sNogo-B, P = 0.001) (Fig. 3C).

Given that upregulation of sNogo-B in the circulation was paralleled by altered VEGF-A signaling, for which the effects on endothelial permeability in DN are regulated by eNOS activation (9), we examined phosphorylation of eNOS. Diabetes led to a significant increase in the ratio of phosphorylated eNOSs^Ser1177/total eNOS (Fig. 3D and E) (ND-GFP vs. D-GFP, P = 0.017), which was prevented by sNogo-B overexpression (D-GFP vs. D–sNogo-B, P = 0.05) (Fig. 3D and E). As protein kinase B (AKT) activation is involved in EC proliferation and eNOS phosphorylation (9), we investigated the phosphorylation at the AKT^Ser473 site. The ratio of AKT^Ser473 phosphorylation/total AKT was upregulated in D mice, an effect that was significantly attenuated in mice injected with AAV–sNogo-B (ND-GFP vs. D-GFP, P = 0.0001; D-GFP vs. D–sNogo-B, P = 0.04) (Fig. 4A). The level of total AKT protein was downregulated by diabetes (ND-GFP vs. D-GFP, P = 0.002) (31), but elevated circulating levels of sNogo-B had no effect on total AKT protein expression in kidney cortex lysate.

We then investigated activation of GSK3β, a known substrate of AKT (32) that has been postulated as a therapeutic target for diabetic glomerulopathy (33). The ratio of phosphorylated GSK3β^Ser9/total GSK3β was upregulated in kidney cortex lysates of D mice when compared with ND mice (Fig. 4B) (ND-GFP vs. D-GFP, P = 0.006), consistent with previous reports (33). Notably, sNogo-B overexpression in the circulation was paralleled by a significant downregulation of GSK3β^Ser9 phosphorylation/total GSK3β in D mice (D-GFP vs. D–sNogo-B, P = 0.04) (Fig. 4B). Total GSK3β protein expression was not altered in any of the experimental groups studied. As a known substrate of GSK3β, an important player in DN (34), we studied total β-catenin (35) protein levels in kidney cortex lysate. In line with GSK3β activation status, β-catenin was significantly upregulated in D mice when compared with ND mice (ND-GFP vs. D-GFP, P = 0.007) (Fig. 4C). sNogo-B overexpression in the circulation led to a significant downregulation of β-catenin in kidney cortex of D mice (Fig. 4C) (D-GFP vs. D–sNogo-B, P = 0.01), whereas β-catenin protein levels were unaffected in ND animals.

Next, we examined the physical interaction between sNogo-B and NgBR by IP and PLA experiments in human GECs (26). Firstly, we showed that GECs expressed NgBR in vitro. We then could detect an interaction between 6xHis-Tag/sNogo-B and NgBR with IP experiments (Supplementary Fig. 3A).

The sNogo-B/NgBR interaction was further confirmed with PLA experiments in which the sNogo-B/NgBR interaction was observed by identification of positive red/orange dots signals in GECs incubated with “conditioned media” containing sNogo-B protein as early as 1 min (Supplementary Fig. 3B).

To further explore the renoprotective mechanisms of sNogo-B overexpression in the circulation, we investigated the expression of full-length Nogo-B in mouse and human kidney tissue in diabetes. With immunogold staining, combined with EM, we found that full-length Nogo-B protein was localized in both GECs and podocytes in ND male adult mice (Fig. 5A). Immunohistochemical staining confirmed the expression of full-length Nogo-B in the glomerular tuft and, as described, in cortical collecting duct (36) (Fig. 5B). Full-length Nogo-B protein, assessed by immunoblotting, in glomeruli enriched kidney cortex and isolated glomeruli lysates was two- to threefold downregulated in D mice when compared with ND controls (Fig. 5C) (ND-GFP vs. D-GFP, P ≤ 0.045); sNogo-B overexpression in the circulation restored the diabetes-induced loss of full-length Nogo-B (D-GFP vs. D–sNogo-B, P ≤ 0.015) (Fig. 5C). NgBR was expressed in isolated glomeruli but not modulated by diabetes or elevated sNogo-B circulating levels (Fig. 5D).

We also investigated the expression of full-length Nogo-B in human kidney biopsies obtained from two distinct glomerular diseases: DN and TBMN. DN represents a progressive proteinuric disease, while TBMN represents a benign nonproteinuric, nonprogressive disorder, featuring a uniformly thinned GBM (37) without any other significant glomerular pathology (Supplementary Table 1). Nogo-B expression was lower in glomeruli of kidney biopsies of patients with DN when compared with patients with TBMN (TBMN vs. DN, P = 0.0001) (Fig. 5E).

Fully differentiated human GECs (26) incubated in NG or HG condition, with VEGF-A (50 ng/mL or vehicle) or a combination of HG and VEGF-A for 72 h, showed a 30–40% downregulation of full-length Nogo-B expression (NG vs. HG, NG vs. VEGF-A, and NG vs. HG plus VEGF-A, P ≤ 0.04) (Fig. 6A). HG and/or VEGF-A–mediated full-length Nogo-B downregulation was paralleled by an upregulation of sNogo-B secretion in the supernatant (NG vs. HG, NG vs. VEGF-A, and NG vs. HG + VEGF-A, P ≤ 0.03) (Fig. 6B).

Finally, we explored the paracrine/autocrine effects of secreted sNogo-B overexpression on the ability of HUVECs, known to express NgBR (19), to differentiate in a tubelike structure (tube formation assay) (38) when incubated with the serum of patients with T1DM with (DN+) or without (DN−) DN (Supplementary Table 2).

sNogo-B overexpression led to a fivefold increase in sNogo-B released from cells into the supernatant when compared with ADV-control–transfected cells (ADV-control vs. ADV–sNogo-B within DN− and DN+, P = 0.0001) (Fig. 7D).

HUVECs incubated with serum of patients with T1DM/DN+ had decreased numbers and lengths of tubes when compared with cells incubated with serum of patients with T1DM/DN− (ADV-control, DN− vs. DN+, tube length, P = 0.01; tube number, P = 0.04) (Fig. 7A–C). Increased sNogo-B in the supernatant was paralleled by a correction of the impaired tube formation observed in cells incubated with DN+ serum (DN+, ADV-control vs. ADV–sNogo-B, tube length, P = 0.02; tube number, P = 0.001) (Fig. 7A–C).

In this study, we demonstrated, in DBA2/J diabetic mice, that systemic overexpression of the N-terminal fragment sNogo-B, which binds to NgBR, improved diabetic glomerulopathy as evidenced by a reduction in albuminuria. This finding was associated with reduced GECs proliferation, restoration of glycocalyx thickness, maintenance of podocyte number, and dampening of VEGF-A signaling and eNOS, AKT, and GSK3β phosphorylation as assessed in glomeruli-enriched kidney cortex lysates. Finally, sNogo-B prevented the impairment of tube formation, which occurred when ECs were exposed to sera from patients with T1DM with DN.

The reduction in diabetes-mediated albuminuria seen in mice with sNogo-B overexpression in the circulation was possibly partly driven by a parallel fall in diabetes-mediated glomerular hyperfiltration; this was not accompanied by an increase in glomerular volume, which is in accord with other reports examining the early (up to 16 weeks of diabetes) phase of DN in DBA2J mice (39). Other potential confounders for the reduced albuminuria could be that sNogo-B overexpression in the circulation lowers blood pressure or be related to changes in glycemic control (2). However, we did not find any change in systemic blood pressure or glycemic control between diabetic mice with overexpression of sNogo-B in the circulation or control mice ruling out these possibilities. As previously described, the blood pressure observed in diabetic DBA2J mice was slightly lower than in nondiabetic ones (39).

Urinary sNogo-B was elevated in diabetes, likely due to the increased permeability of the filtration barrier. However, the excretion of sNogo-B was not increased in either nondiabetic or diabetic mice administered AAV-sNogoB, a finding most likely to be attributed to the large molecular weight of the 6xHis-Tag/sNogo-B (∼80 kDa) (Supplementary Fig. 1C).

As albuminuria results from defects in the glomerular filtration barrier, we initially focused on the glomerular endothelium because the biological effects of sNogo-B and its receptor NgBR are primarily on the vasculature (14,19). Elevated circulating levels of sNogo-B prevented diabetes-mediated GEC proliferation as seen in the early stages of DN (1,7,40). sNogo-B overexpression in the circulation was also associated with amelioration of the diabetes-induced loss of lectin, an important component of the endothelial glycocalyx, known to contribute to vascular permeability (25,41). These changes were paralleled by a reduction in VEGF-A signaling. VEGF-A is a potent stimulator of EC proliferation (42), and dampening VEGF-A signaling ameliorates albuminuria and glomerular damage in DN (4). VEGF-A has also been implicated in the regulation of the thickness of the endothelial glycocalyx in diabetic mice with the antiangiogenic VEGF-A_165b isoform able to restore the diabetes-induced loss of the endothelial glycocalyx and albuminuria (41). There have been a few reports linking the sNogo-B/NgBR and VEGF-A/VEGFR2 systems, suggesting a concerted vasculoprotective role (20,43).

The effects on the endothelium may also be related to the changes seen in Wnt/β-catenin signaling. Activation of Wnt/β-catenin signaling (with cellular β-catenin accumulation) results in endothelial dysfunction, inflammation, and fibrosis and has been implicated in DN (34). Inactive GSK3β is also known to promote EC proliferation (44), with a phosphorylated (inactive) GSK3β favoring accumulation of cellular β-catenin (45), a promoter of angiogenesis and enhancer of VEGF-A/VEGFR2 signaling (46). In our experiments, we showed that sNogo-B reduced GSK3β phosphorylation/β-catenin, which may contribute to preventing the angiogenesis response to diabetes-mediated vascular damage and dampen VEGF-A/VEGFR-2 signaling.

Increased eNOS phosphorylation and NO production have been implicated in hyperfiltration in the early phases of diabetic glomerulopathy (47,48), and the role of eNOS in oxidative stress (when in an “uncoupled state” [49]) is implicated in DN and GEC/podocyte damage (50). The reduction in diabetes-mediated eNOS phosphorylation, observed in kidney cortex lysates of diabetic mice with elevated sNogo-B circulating levels, would result in reduced eNOS-mediated NO production and oxidative stress, events that could contribute to a reduction in glomerular hyperfiltration (51) and in the prevention of diabetes-mediated loss of GECs glycocalyx and podocyte detachment. sNogo-B overexpression in the circulation associates with inhibition of diabetes-mediated modulation of VEGF-A/VEGFR2 and AKT/GSK3β/β-catenin system, signaling pathways that are implicated in podocyte injury/detachment and albuminuria (52–54). Finally, it might be possible that sNogo-B may have a direct effect on podocytes, as evidenced by our in vitro data showing expression of NgBR in human podocyte cells (Supplementary Fig. 4).

Full-length Nogo-B (assessed with a specific N-terminus antiserum) was found to be expressed in isolated glomeruli and glomeruli-enriched kidney cortex lysates and downregulated in diabetic mice. Full-length Nogo-B expression levels were also lower in glomeruli of proteinuric DN (a progressive chronic kidney disease) biopsy kidney samples when compared with TBMN, a benign glomerular disease without proteinuria, suggesting that the presence of proteinuria/albuminuria associates with reduced glomerular full-length Nogo-B levels. Whether low glomerular Nogo-B causes albuminuria is yet to be determined, but in experimental models of diabetic retinopathy, lack of full-length Nogo-B attenuated HG-induced cell migration and tube formation (EC differentiation) (55). In contrast to our findings, studies performed in humans’ and rodents’ renal biopsies (Nephroseq database, University of Michigan) showed no change in full-length Nogo-B mRNA expression in diabetic kidney tissue when compared with control tissue.

We also demonstrate that HG and/or VEGF-A determine, in GECs in culture, an increase in sNogo-B in the supernatant and a parallel reduction in cellular full-length Nogo-B (probed with specific N-terminus antiserum). These results support the cleavage of the full-length Nogo-B N-terminus as previously described (17), and future work will have to address the mechanisms behind this phenomenon.

We then explored the putative vasculoprotective role of sNogo-B in an angiogenesis experimental model in HUVECs cultured with sera of patients with T1DM susceptible to (DN+) or protected toward (DN−) the progression of DN (56). We demonstrate that circulating factors (sera from T1DM/DN+), other than glucose, negatively affect EC tube formation, and sNogo-B overexpression in the supernatant rescued this defect favoring EC differentiation toward a stable tube structure, an effect likely related to the patients’ DN status rather than being driven by renal impairment, given that the average of renal function of the two groups of patients studied was 80–100 mL/min/1.73 m². Importantly, artificial upregulation of the N-terminus of Nogo-B (sNogo-B) in the supernatant is able, via a paracrine autocrine mechanism, to ameliorate the ability of ECs to migrate and differentiate into tubule-like structures (38), which translates into a more stable vasculature.

Collectively, our studies provide the first evidence that a primary increase of sNogo-B in the circulation protects the glomerular vasculature in diabetes. More studies will need to further dissect sNogo-B mechanisms of action and explore its potential benefit in diabetic animals with established kidney disease. sNogo-B could represent a novel targetable pathway for the treatment of diabetic chronic vascular complications.

Acknowledgments. The authors thank Dr. W. Sessa (Yale University, New Haven, CT) for the sNogo-B construct, Dr. S. Satchell (Bristol University, Bristol, U.K.) for the human GECs, and Dr. M. Saleem (Bristol University, Bristol, U.K.) for the human podocyte cell lines. The authors also thank the patients who participated in this research.

Funding. This work was supported by British Heart Foundation Project Grant PG/16/41/32138 to G.E.M., D.A.L., and L.G., Heart Research UK (PhD studentship to J.P. and grant RG2619/12/15 to D.A.L. and L.G.), the Diabetes Research Wellness Foundation (startup grant to L.G.), and startup support from the Henry Lester Trust Limited to J.P. M.F. was supported by an European Renal Association–European Dialysis and Transplant Association nonclinical fellowship. X.B. was supported by National Key R&D Program of China (2018YFC1314000), National Natural Science Foundation of China (grant 81873616), and “Group-type” Special Support Project for Education Talents in Universities (G619080438, 4SG19002G, 4SG19044G). X.B. and F.F.H. were supported by the National Natural Science Foundation of China (grant 81100496) and Guangdong Natural Science Foundation (grant 2016A030313581). C.A.R. was supported by a visiting fellowship from the Italian Society of Nephrology and a Kidney Research UK fellowship (TF_001_20171120). Support was also received from the National Institute for Health Research, a Biomedical Research Centre award to Guy’s and St Thomas’ NHS Foundation Trust in partnership with King’s College London. D.A.L. is supported by the Medical Research Council (grant MR/P018629/1), Diabetes UK (grants 13/0004763 and 15/0005283), Kidney Research UK (grant RP36/2015), and the National Institute for Health Research Biomedical Research Centre at Great Ormond Street Hospital for Children NHS Foundation Trust and University College London.

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

Author Contributions. I.H.-D., G.E.M., D.A.L., and L.G. led data discussion, interpretation, and manuscript writing with input from all other coauthors. I.H.-D., D.A.L., and L.G. contributed to experimental design and data collection together with J.P., C.A.R., X.B., J.K., K.E.W., M.F., G.E.F., F.A., A.E.H., F.F.H., G.E.M., and R.Q.M. D.A.L. and L.G. conceived the experiment. All authors approved the final manuscript. L.G. 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.

Data and Resource Availability. The data sets generated and/or analyzed during the current study are available from the corresponding author upon reasonable request. The viral vectors generated and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Prior Presentation. Parts of this study were presented in abstract form at the 53rd Annual Meeting of the European Association for the Study of Diabetes, Lisbon, Portugal, 11–15 September 2017; 54th Annual Meeting of the European Association for the Study of Diabetes, Berlin, Germany, 1–5 October 2018; American Society of Nephrology’s Kidney Week 2017, New Orleans, LA, 31 October–5 November 2017; American Society of Nephrology’s Kidney Week 2018, San Diego, CA, 23–28 October 2018; 54th ERA-EDTA Congress, Madrid, Spain, 3–6 June 2017; 28th Annual General Meeting of the European Diabetic Nephropathy Study Group, Copenhagen, Dennmark, 22–23 May 2015; 29th Annual General Meeting of the European Diabetic Nephropathy Study Group, Pisa, Italy, 20–21 May 2016; 30th Annual General Meeting of the European Diabetic Nephropathy Study Group, Helsinki, Finland, 19–21 May 2017; and 31st Annual General Meeting of the European Diabetic Nephropathy Study Group, Groningen, the Netherlands, 18–19 May 2018.