Slit2-Robo Signaling Promotes Glomerular Vascularization and Nephron Development

Significance Statement

Slit2 is a secreted ligand for Robo1 and Robo2 receptors. Slit2 and Robo2 mutations lead to congenital abnormalities of the kidney and ureteric tract, underscoring the requirement of this signaling pathway for kidney development. Previous studies in global mouse knockouts demonstrated that Slit2-Robo2 signaling restricts ureteric epithelium budding. Temporally inducible Slit2 and Robo deletions reveal a novel role of Slit2-Robo signaling in glomerular vascularization in mice. Only the glomerular endothelium, but no other renal endothelial compartment, responded to Slit2 via Robo receptors. Postnatally induced Slit2 gene deletion or a Slit2 ligand trap inhibited glomerular vascularization by reducing endothelial cell proliferation and migration, identifying Slit2 as a driver of glomerular angiogenesis.

Abstract

Background

Kidney function requires continuous blood filtration by glomerular capillaries. Disruption of glomerular vascular development or maintenance contributes to the pathogenesis of kidney diseases, but the signaling events regulating renal endothelium development remain incompletely understood. Here, we discovered a novel role of Slit2-Robo signaling in glomerular vascularization. Slit2 is a secreted polypeptide that binds to transmembrane Robo receptors and regulates axon guidance as well as ureteric bud branching and angiogenesis.

Methods

We performed Slit2-alkaline phosphatase binding to kidney cryosections from mice with or without tamoxifen-inducible Slit2 or Robo1 and -2 deletions, and we characterized the phenotypes using immunohistochemistry, electron microscopy, and functional intravenous dye perfusion analysis.

Results

Only the glomerular endothelium, but no other renal endothelial compartment, responded to Slit2 in the developing kidney vasculature. Induced Slit2 gene deletion or Slit2 ligand trap at birth affected nephrogenesis and inhibited vascularization of developing glomeruli by reducing endothelial proliferation and migration, leading to defective cortical glomerular perfusion and abnormal podocyte differentiation. Global and endothelial-specific Robo deletion showed that both endothelial and epithelial Robo receptors contributed to glomerular vascularization.

Conclusions

Our study provides new insights into the signaling pathways involved in glomerular vascular development and identifies Slit2 as a potential tool to enhance glomerular angiogenesis.

The filtration function of the kidney relies on the presence and proper function of a heterogeneous renal vasculature,^1,2 which receives 20% of cardiac output.³ Ninety percent of the renal blood flow is delivered and filtered via the glomerular capillary loops,³ which consist of endothelial cells and podocytes and are supported by mesangial cells.⁴ Fenestrated endothelium together with the podocyte slit diaphragm and the glomerular basement membrane make up the filtration barrier.⁵ The glomerular filtration barrier allows the retention of large macromolecules and cells in the circulation while filtering through water and small molecules. Despite their importance, how glomerular capillary loops are assembled is still not fully understood. Current models of glomerular vascularization describe that early developing podocytes secrete vascular endothelial growth factor A (VEGFA), a major angiogenic growth factor,⁶ which binds and signals through VEGFR2 on endothelial cells to induce endothelial cell migration into the vascular cleft of the nephron progenitors at the S-shaped body stage.⁷ Other than VEGFA, other factors, such as semaphorin3a⁸ and TGFβ1,⁹ affect the glomerular endothelium, but the cellular events and signaling networks that guide the formation of functional three-dimensional glomerular capillaries remain incompletely characterized.

The delicate structure of the glomerular vasculature also makes it particularly susceptible to diseases. Disruption of the glomerular capillaries is seen in diabetic nephropathy¹⁰ and preeclampsia,¹¹ and it leads to proteinuria. In addition, proteinuria and glomerular endotheliosis are common adverse effects seen in antivascular endothelial growth factor (anti-VEGF) therapy for cancer.¹² Currently, there is no pharmacologic treatment to restore glomerular endothelium function,¹³ which urges more in-depth studies. Moreover, glomerular endothelial cells have been shown to be important for age-related kidney disease.¹⁴ Thus, understanding the signaling events involved in the development of the glomerular vasculature can promote better understanding of this biologic process, providing further insights for bioengineering of kidney tissue in vitro, and help identify novel therapeutic targets to treat glomerular diseases and normalize glomerular function.

Slits are a family of three secreted proteins that bind to Robo1 and -2 receptors.¹⁵ Slit-Robo signaling was first described to regulate axon guidance during development of the central nervous system.¹⁶ In humans, mutations of Slit2 and Robo2 were found in patients with congenital abnormalities of the kidney and ureteric tract disease, underscoring a requirement of this signaling pathway for kidney development.^17–19 Global knockout mouse models showed that Slit2-Robo2 signaling inhibition led to the formation of multiple ureters and perinatal lethality,²⁰ demonstrating that Slit2-Robo2 signaling functions to restrict the budding of the ureteric epithelium to a single site. Further studies showed that Robo2 signaling defects in mice also lead to cystic kidney disease and that Robo2 signaling regulates ciliogenesis, polarization, and differentiation of tubular epithelium via P53 and ciliary proteins.²¹ Slit2-Robo2 signaling also affects glomerular podocyte foot processes by interacting with nephrin.²² Whether Slit-Robo signaling also affects the renal vasculature is still unknown.

Previous work from us and others showed that Slit-Robo signaling promotes angiogenesis in the retina²³ and bone.²⁴ Slit2 signaling via endothelial Robo1 and -2 receptors promotes endothelial cell migration in cooperation with the VEGF signaling pathway.^23,25 Slit2 proangiogenic signaling requires endocytosis via the EndophilinA2 pathway and downstream NCK adaptor proteins that integrate signals from both Slit- Robo and VEGFA-VEGFR2 signaling pathways.^25,26 In this study, we investigated Slit2 function during renal morphogenesis and vasculature development and report that Slit2-Robo1/2 signaling is indispensable for the proper development of nephrons and glomerular vasculature.

Methods

Mice

All mouse experiments were reviewed and approved by the Yale University Institutional Animal Care and Use Committee.

Slit2^fl/fl mice, in which exon 8 is flanked by loxP sites, were generated at the Mouse Clinical Institute–Institut Clinique de la Souris (Illkirch, France; http://www.ics-mci.fr/en/).²³ This allele is predicted to produce a truncated Slit2 protein in which the final amino acid residue is F204, just after the end of the first leucine-rich domain (D1). This truncated protein is, therefore, unable to bind or activate its receptors.²³ Robo1^−/−, Robo2^fl/fl mice were previously described and validated.^23,27 Floxed mice were crossed with RosaCre^ERT2 (Jackson Laboratories)²⁸ and Cdh5 Cre^ERT2 mice.²⁹ Gene deletion in neonatal mice was induced by tamoxifen (TAM) injections (T5648; Sigma-Aldrich; diluted in corn oil, C8267; Sigma-Aldrich). Mice of either sex were used. In each experiment, TAM-injected Cre-negative littermate pups were used as controls. Cre-positive flox-negative mice were also analyzed as additional controls.

Quantitative Real-Time PCR

Total RNA was isolated from whole kidney using the RNeasy Kit (74134; Qiagen). One microgram of RNA was used as a template for reverse transcription using the iScript cDNA Synthesis kit (170–8891; Bio-Rad). qPCR was performed in duplicate using SYBR Green Supermix (1798880; Bio-Rad) and specific primers (Qiagen). GAPDH and actin were used as internal references to normalize between samples. Then, the relative gene expression levels compared with the average of control groups were calculated.

Immunostaining

For immunostaining on sections, kidneys were collected at postnatal day 7 (P7) or 6 weeks and fixed in 4% PFA at 4°C overnight. The kidneys were submerged in 30% sucrose (573113; Sigma-Aldrich) at 4°C and snap frozen; then, 6-μm sections were cut with a cryostat (CM1520; Leica). The sections were blocked in PBS containing 0.05% Tween-20 and 5% donkey serum for 1 hour and incubated overnight at 4°C with primary antibodies diluted in the same blocking buffer, followed by incubation with secondary antibodies and 100 μg/ml DAPI diluted in blocking buffer. The primary antibodies used were anti–Cytokeratin 8 (anti-CK8; TROMA-I; Developmental Studies Hybridoma Bank), Lotus Tetragonolobus Lectin (LTL)-FITC (FL-1321; Vector Laboratories), anti-endomucin (anit-Emcn; HM1108; Hycult Biotech), anti-Vegfr2 (AF644; R&D Systems), anti–Tamm Horsfall Protein (anti-THP; Millipore AB733), anti-podocalyxin (AF1556; R&D Systems), anti–Erg-1/2/3 (anti-ERG; 92513; Abcam), anti-Robo2 (64158; Abcam), anticleaved caspase3 (9226S; Cell Signaling), anti-Ki67 (PA5–19462; Invitrogen), anti-EHD3 (NBP2–31896; Novus Biology), anti-ESM1 (AF1999; R&D Systems), anti-CK8 (TROMA-I; DSHB), anti–Wilms tumor 1 (WT1) (sc-7385; Santa Cruz), anti-NG2 (AB5320; Millipore), and anti-Pdgfrβ (14–1402–82; Invitrogen). For soluble Flt1-Fc staining, kidney sections were blocked as described above and then incubated in 1 μg/ml recombinant mouse soluble Flt-1 FC chimera (471-F1–100; R&D Systems) overnight at 4 °C. Sections were washed with PBS and incubated with anti-human IgG secondary antibodies at room temperature for 1 hour. Alexa Fluor–coupled secondary antibodies were all from Invitrogen. Images were acquired with an SP8 confocal microscope (Leica) or with an LSM800 confocal microscope (Zeiss).

In Vivo Slit2 Binding Using Alkaline Phosphatase (AP) Staining

To obtain alkaline-phosphatase–conjugated full-length Slit2 protein (Slit2-FL-AP) protein, the Slit2 full-length sequence was inserted into a pAPtag5 plasmid (catalog no. L301; GenHunter Corporation). Slit-Nter-AP was obtained by inserting the N-terminal sequence (nucleotides 378–3640) into pAPtag5 plasmid. Top 10 competent Escherichia coli (C404010; Thermofisher) was used for amplification of plasmids. Slit2-AP protein was obtained by transient transfection of the plasmids into HEK cells using Lipofectamine 2000 (11668; Thermofisher). Supernatant containing Slit2-AP protein was harvested after 36 hours and stored at −80°C. Renal cryosections were blocked and then incubated overnight with Slit2-AP protein at 4°C or 2 hours at room temperature. Slides were washed with TBS. Endogenous AP activity was inactivated by 2 hours of incubation in TBS at 65°C, and Slit2 binding was revealed by adding NBT/BCIP Substrate (34042; Thermofisher).

Pharmacologic Slit2 Inhibition

We used Robo1-Fc (1749-RB; R&D Systems) or control IgG1-Fc (110-HG-100; R&D Systems). Mice were injected with 5 μg/g Robo1-Fc or control IgG1-Fc at P0, P1, P2, P3, and P5. Samples were collected at P7. Dosage and validation of successful trapping of endogenous Slit2 are described in another study by our group.³⁰

Western Blot

Kidneys were rinsed with PBS and homogenized in RIPA buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, and 0.1% SDS) containing protease and phosphatase inhibitor cocktails (11697498001 and 04906837001; Roche). Protein concentration was determined by a BCA protein assay kit (23225; Pierce). Equal amounts of protein per sample were loaded and separated by 4%–15% SDS-PAGE gel (5671084; Bio-Rad). The proteins were transferred to a nitrocellulose membrane and blocked in 5% BSA in TBST for 1 hour; then, they incubated in primary antibodies overnight at 4°C. The membranes were washed three times with TBST for 10 minutes, and HRP-conjugated secondary antibodies (1:5000; Vector Lab) were added for 1 hour at room temperature. The membranes were washed with TBST, treated with Supersignal Chemiluminescent substrate (Thermo Fisher Scientific, Rockford, IL), and visualized on a ChemiDoc MP imaging system (Bio-Rad Laboratories, Mississauga, ON, Canada). Band intensity was quantified using ImageJ software. Antibodies used for WB were anti-Vegfa (ab1316; R&D Systems), anti-Vegfr2 (2479S; Cell Signaling), anti-Slit2 (PA5–31133; Invitrogen), and antiactin (A1978; Sigma-Aldrich).

Perfusion Assay

Fifteen microliters of 0.3 μg/μl Isolectin-B4 Alexa Fluor 488 Conjugate (I21411; Invitrogen) were administered via retro-orbital injection of P7 mice. Dye was left to circulate for 15 minutes, and the kidney samples were collected and processed as described above. Thick sections of 60 μm were cut and imaged.

Transmission Electron Microscopy

Mice were perfused with prewarmed perfusion fixative (4% PFA and 0.5% glutaraldehyde in PBS). Kidney samples were cut into half and placed in fixation buffer (2% PFA and 2.5% GA in cacodylate buffer) for 2 hours at room temperature. Samples were washed with 0.1 M cacodylate and postfixed in 1% osmium tetroxide. Then, samples were dehydrated in dimethoxypropane, and embedded into epoxyresin, followed by ultrathin sectioning. Ultrathin sections were stained with 0.3% potassium ferrocyanide in 2.0% osmium tetroxide and 4.0% uranyl acetate and then imaged on an FEI Tecnai G2 Spirit BioTWIN electron microscope equipped with an SIS Morada 11-megapixel CCD camera. For quantification of podocyte morphology at superficial cortical glomeruli of P7 kidneys, low-magnification EM pictures of entire glomeruli were taken. A rectangle surrounding the podocyte cell body was drawn, and side length was measured. The longer side was considered as width, and the shorter side length was considered as height. The foot process width was measured on six high-magnification TEM images per mouse.

Quantifications

Glomerular area was measured on three glomeruli per image and five 20× images of the superficial cortical or juxtamedullary region. THP⁺ tubular length was measured on three to four 10× images of sagittal sections through the kidney center, and the continuous length of 15 individual tubules per image was measured. For THP and CK8 tubular area, 20 individual tubules on four 20× images were measured to calculate the average tubular area. LTL⁺ tubular area was normalized to the imaged area using the ImageJ analyze particle function. Total ERG⁺ endothelial cell number was counted with the ImageJ analyze particle function on five 10× images per sample and normalized to the area. To quantify the number of endothelial cells, podocytes, and mesangial cell coverage, kidney sections were stained with ERG, WT1, NG2, respectively, in addition to Pdx staining. Five 20× images of superficial cortical and juxtamedullary regions were taken. The number of WT1⁺ podocytes or ERG⁺ endothelial cells per glomerulus was quantified on 15 glomeruli per mouse. The average NG2 intensity of 20 glomeruli was measured. To quantify vascularized glomeruli, five 10× images were taken, and the number of glomeruli with Emcn⁺ capillaries was counted and normalized to the total glomerular number. To measure the glomerular number, six 4× images were taken, and total glomeruli number of individual images was counted and normalized to the area. When measuring separately the superficial cortical versus the juxtamedullary, the outermost 200-μm area was quantified, and the number of superficial cortical glomeruli was normalized to this area. To quantify the EC invasion to the vascular cleft, the length of endothelial cells within vascular cleft was measured and normalized to the total length of vascular cleft in ten comma- and S-shaped bodies on five 20× images per sample.

Blood and Urine Analyses

Urine creatine levels were measured by the QuantiChrom Creatinine Assay Kit (BioAssay Systems). Urine albumin level was measured by ELISA (catalog no. E90–134; Bethyl Laboratories, Inc.). BUN level was measured by the colorimetric method with the Stanbio Procedure 0580 Stanbio Excel Chemistry Analyzer.

Statistical Analyses

All graphs of quantitative data show mean value ± SEM. In scatterplot graphs, each dot represents average data from one mouse or as otherwise indicated in figures. Significance of all data was analyzed with Prism 7 using the Mann–Whitney U test method or ANOVA test where there are three groups. A P<0.05 was considered significant.

Results

Neonatal Slit2 Deletion Affects Renal Tubules and Glomeruli

We took advantage of a conditional Slit2 allele²³ that was intercrossed with a Rosa-Cre^ERT2 line²⁸ to generate temporally induced global Slit2 deletions by TAM injection (Slit2^iKO). Embryonic deletion of Slit2 starting at E13.5 resulted in formation of cystic kidneys with severe tubular dilations at E18.5 (Supplemental Figure 1, A and B), similar to the phenotype of Robo2 mutants described previously.²¹ Interestingly, immunostaining with antibodies recognizing the endothelial markers Emcn³¹ and the ERG1–3 transcription factors³² showed that Slit2 deletion also affected developing glomeruli and reduced glomerular endothelial cell numbers in E18.5 embryonic kidneys (Supplemental Figure 1, C and D).

To better understand the effects of Slit2 in nephrogenesis, we deleted Slit2 neonatally by injecting TAM at P0–P2, when nephrogenesis in the superficial cortex of the kidney is still ongoing,^33,34 and we analyzed kidneys upon completion of nephrogenesis at P7–P9³³(Figure 1A). TAM-injected Cre-negative littermates and Cre-positive, flox-negative mice were used as controls. qPCR analysis of renal Slit2 expression confirmed efficient gene deletion in Slit2^iKO mutants (Figure 1B), which had slightly but significantly reduced body and kidney weight at P7 (Figure 1, C–E). H&E staining confirmed a reduced kidney size in Slit2^iKO when compared with controls and showed that the renal cortex region of Slit2^iKO appeared disorganized, whereas the kidney pelvis and larger arteries and veins therein developed normally (Figure 1C).

Figure 1.

Neonatal Slit2 deletion affects renal tubules and superficial cortical glomeruli. (A) Time line for deletion of the Slit2 fl/fl allele by IP TAM injections to neonates. Samples were collected at P7 or P9. (B) Validation of Slit2 deletion. Each dot represents individual mice. (C) Hematoxylin & Eosin (H&E) staining of P9 kidneys of the indicated genotypes. The ureter (U), medulla (M), and cortex region (C) were present in both control and Slit2^iKO. Scale bar: 0.5 mm. (D) Body weight and (E) kidney weight of P7 control and Slit2^iKO mice. Each dot represents individual mice. (F) Higher-magnification images from boxed areas in (C) show disorganization of MR tubules and poor development of glomeruli in the superficial cortical region of Slit2^iKO kidneys (arrows). A, artery; V, vein. Scale bar: 100 µm. (G) Quantification of superficial cortical and juxtamedullary glomerular area of P7 kidney sections. Each dot represents an individual mouse. (H) P7 kidney cryosections immune labeled with THP antibody recognizing TAL and nuclear DAPI (blue). Scale bar: 50 µm. (I) Quantification of THP⁺ tubular length and area. Each dot represents an individual mouse. (J) Slit2-FL-AP binds to MR tubules and to glomerular capillary loops (arrows) in Slit2^fl/fl and Slit2^iKO mice. Note the poorly developed vascular loop in Slit2^iKO. Scale bar: 50 µm. (K) Quantification of CK8⁺ tubular area and total LTL⁺ tubular area. Each dot represents an individual mouse. *, P<0.05; **, P<0.01; ***, P<0.001; Mann-Whitney U test.

Higher-magnification views of the H&E-stained renal cortex showed that Slit2^iKO mice presented underdeveloped superficial cortical glomeruli when compared with Cre-negative littermate controls and that the medullary rays (MRs), which are composed of collecting ducts (CDs) and thick ascending limbs (TALs), were abnormally shaped or missing in the Slit2^iKO mice (Figure 1F).

Quantifications of glomeruli on H&E-stained sections confirmed that superficial cortical glomeruli were smaller in Slit2^iKO, whereas the juxtamedullary glomeruli that had developed before gene deletion was induced³³ were of similar size when compared with control littermates (Figure 1G, Supplemental Figure 2A). The total number of glomeruli was similar between control and Slit2^iKO kidneys (Supplemental Figure 2B). To quantify the tubular defects observed in Slit2^iKO mice, we performed immunostainings with specific markers for different renal tubules, including THP labeling of TAL, CK8 labeling of CD, and LTL staining of proximal convoluted tubules (PCTs). Slit2 deletion reduced the length of THP⁺ TALs and increased their surface area (Figure 1, H and I), but it did not significantly change the surface area of CK8⁺ CDs or of LTL⁺ PCTs (Figure 1K, Supplemental Figure 2, B and C). To determine if Slit2 protein signaled directly to the renal tubules and glomeruli that were affected by its loss of function, we visualized Slit2 binding to kidney cryosections with a Slit2-FL-AP (Figure 1J). Slit2-FL-AP bound to vascular loops in glomeruli but not to any other renal vascular beds (Figure 1J). Slit2-FL-AP also bound to the abluminal side of the tubular epithelium of the MRs but not to PCTs (Figure 1J). Slit2 binding MRs and cortical glomeruli were abnormally shaped in Slit2^iKO mice (Figure 1J). Taken together, we found that Slit2 binds glomeruli and MRs and that neonatal Slit2 deletion affected glomerular vascularization and TAL tubulogenesis.

To determine if Slit2 affected renal development by binding and signaling through Robo1 and -2 receptors, we incubated P7 WT C57/Bl6 kidney sections with Slit2-FL-AP or with an AP-conjugated N-terminal fragment of Slit (Nter-Slit-AP), which contains the Robo receptor binding sequences.³⁵ Both Slit2-AP proteins stained renal tubules and glomeruli with a similar pattern (Supplemental Figure 3A), suggesting that Slit2 bound to Robo1 and -2 receptors in the kidney. To test the specificity of Slit2 binding to Robos, we generated TAM-inducible global Robo receptor knockouts using Robo1^−/−; Robo2^fl/fl mice intercrossed with the Rosa-Cre^ERT2 driver (Robo1,2^iKO).²³ TAM was administered at P0, P1, and P2, and mice were analyzed at P7 (Supplemental Figure 3B). Slit2-FL-AP staining of kidney sections showed dramatic reduction of Slit2 binding in the Robo1,2^iKO mice (Supplemental Figure 3C), confirming the specificity of Slit2-AP staining to renal Robo receptors, as well as the successful Robo1 and -2 receptor knockout. Robo1^−/−, Robo2^fl/fl mice showed weaker Slit2-FL-AP staining when compared with C57/Bl6 mice (Supplemental Figure 3, A and C), indicating that both Robo receptors contributed to Slit2 binding in the kidney. In support of this idea, Robo1 expression was visualized by β-galactosidase staining of an LacZ cassette inserted into the Robo1 locus.³⁶ Robo1 β-galactosidase was highly expressed by renal tubules but also detected at lower levels within glomeruli (Supplemental Figure 3D).

Slit2 Promotes Glomerular Vascularization

We next compared glomerular vascular development in Slit2^fl/fl controls and Slit2^iKO mice by staining P7 kidney sections following neonatal gene deletion (Figure 2A). In control mice, Slit2-FL-AP bound to the basolateral side of glomerular vascular loops, although we could not distinguish whether the binding was to the endothelial cell membrane or to podocyte foot processes facing the glomerular basement membrane (Figure 2A). Virtually all glomeruli were positive for Slit2-FL-AP binding, regardless of their position in the superficial cortical or juxtamedullary region (Supplemental Figure 4A), whereas Slit2-AP did not bind to renal peritubular capillaries, vasa recta, or larger arteries and veins (Figure 1J, Supplemental Figures 3 and 4A). In Slit2^iKO mice, superficial cortical glomerular loops still bound Slit2-FL-AP and Nter-Slit-AP but were poorly developed, with fewer loops (Figure 2A, Supplemental Figure 4, A and B), whereas juxtamedullary glomeruli had well-developed vascular loops similar to controls (Supplemental Figure 4A). As glomeruli in the superficial cortical region were mainly formed postnatally (after the onset of gene deletion), whereas juxtamedullary ones were formed during embryonic development (before the onset of gene deletion),^33,37 this result was consistent with Slit2 promoting glomerular vascularization and revealed a previously unknown role of Slit2 in promoting glomerular angiogenesis.

Figure 2.

Slit2 promotes glomerular vascularization. (A) Slit2-FL-AP staining of P7 superficial cortex after TAM treatment at P0–P2. Note the Slit2-binding glomerular capillary loops and poorly developed loop in Slit2^iKO. Scale bar: 10 µm. (B) P7 kidneys with the indicated genotypes were cryosectioned and immunolabeled with antibodies recognizing the indicated proteins. Note the glomeruli without endothelial cells in Slit2^iKO (arrows). JM, juxtamedullary region; SC, superficial cortical region. Scale bar: 100 µm. (C) Quantification of the total ERG⁺ endothelial cell (EC) number normalized to image field and the percentage of vascularized glomeruli in control and Slit2^iKO. (D and F) High-magnification views of glomeruli in (D) the superficial cortical and (F) the juxtamedullary region. Yellow arrowheads point to individual ERG⁺ endothelial nuclei in glomeruli. Scale bar: 10 µm. (E and G) Quantification of ERG⁺ endothelial cell number in individual glomeruli as well as percentage of vascularized glomeruli at (E) the superficial cortex and (G) the juxtamedullary region. Each dot represents an individual mouse. **, P<0.01; ***, P<0.001; Mann-Whitney U test.

To further characterize the angiogenic defects, we performed immunostaining of kidney cryosections with antibodies recognizing endothelial cells (Emcn and ERG) and podocalyxin, which stains the cell body of podocytes and the luminal membrane of blood vessels. Except for cortical glomeruli, the overall renal vasculature patterning was similar between Slit2^iKO and littermate control kidneys (Figure 2B, Supplemental Figure 4C). Counting of ERG⁺ endothelial nuclei on cross-sections through the entire kidney revealed no significant change of total endothelial cell number in Slit2^iKO mice, supporting that Slit2 deletion did not affect the overall renal vascular density (Figure 2C). Counting the number of glomeruli containing capillary loops and normalizing it to the number of all podocalyxin-labeled glomeruli revealed a 20% decrease of vascularized glomeruli in Slit2^iKO mutants when compared with Cre-negative littermate controls or age-matched Cre-positive, flox-negative mice (Figure 2C, Supplemental Figure 4D). When quantifying separately the percentage of vascularized glomeruli in superficial cortex and juxtamedullary region, we observed a significant reduction of vascularized glomeruli in the superficial cortex but not in the juxtamedullary region (Figure 2, D–G). Further quantification of the number of ERG⁺ endothelial nuclei in individual glomeruli revealed a significant decrease of endothelial cell number per glomerulus in Slit2^iKO superficial cortical glomeruli, and some glomeruli were completely devoid of any endothelial cells (Figure 2, D and E). By contrast, the number of endothelial cells in juxtamedullary glomeruli was similar between mutants and controls (Figure 2, F and G).

Slit2 Deletion Reduced Cortical Glomeruli in Young Adult Mice

We next determined if the renal effects of neonatal Slit2 deletion persisted in young adult mice at 6 weeks of age (Figure 3A). qPCR analysis of renal Slit2 expression confirmed efficient gene deletion in 6-week-old Slit2^iKO mutants (Figure 3B) and revealed that kidney weight and body weight of Slit2^iKO were similar to littermate controls in young adult mice (Figure 3, C and D). H&E staining showed a similar organization of renal cortex in controls and Slit2^iKO; however, the number of glomeruli at the superficial cortex was reduced in Slit2^iKO (Figure 3E). Quantification of glomerular numbers on H&E-stained kidney sections revealed a significant reduction at the superficial cortex, whereas the glomeruli number in the juxtamedullary region was similar to that in littermate controls (Figure 3F). The glomeruli that were present in the superficial cortex contained similar numbers of ERG⁺ endothelial cells when compared with control littermates (Figure 3G). Likewise, quantifying the TAL area by THP staining revealed no differences between control and mutants (Figure 3H). Slit2 protein still bound to vascular loops of glomeruli and MRs of young adult mice (Figure 3I). To test if Slit2^iKO impaired renal function, we collected urine and blood samples of 6-week-old control and Slit2^iKO mice and measured the urine albumin-creatinine ratio (ACR) and BUN. We found that the ACR was slightly elevated in the urine of Slit2^iKo mice when compared with controls, but neither ACR nor BUN were significantly different between groups (Figure 3, J and K).

Figure 3.

Slit2 deletion reduces glomerular number in young adult mice. (A) Time line for deletion of the Slit2 fl/fl allele by TAM injections to neonates. Samples were collected at 6 weeks. (B) Validation of Slit2 deletion efficiency. Each dot represents an individual mouse. (C) Body weight and (D) kidney weight of control and Slit2^iKO mice at 6 weeks. Each dot represents an individual mouse. (E) Hematoxylin & Eosin staining of 6-week kidneys of the indicated genotypes. Yellow arrows point to individual glomeruli at the superficial cortex. A, artery; V, vein. Scale bar: 100 µm. (F) Quantification of glomeruli at superficial cortical and juxtamedullary regions. Each dot represents an individual mouse. (G) Quantification of endothelial cell (EC) number expressing the ERG protein per superficial cortical glomerulus. Each dot represents an individual mouse. (H) Quantification of THP⁺ tubular area. Each dot represents an individual mouse. (I) Slit2-FL-AP binds to glomerular capillary loops of the 6-week-old mouse kidneys. Scale bar: 50 µm. (J and K) Urine ACR and BUN of 6-week-old control and Slit2^iKO mice. Each dot represents an individual mouse. *, P<0.05 **, P<0.01; Mann-Whitney U test.

Slit2 Inhibition with a Ligand Trap Reduces Glomerular Vascularization

To confirm that neonatal cortical glomeruli required Slit2 for their vascularization, we used a pharmacologic approach. Neonatal wild-type C57/Bl6 mice were treated with a commercial rat Robo1-Fc protein (5 μg/g, IP) or control IgG-Fc from P0 to P5 and euthanized at P7 (Supplemental Figure 5A). Quantification of the number of ERG-positive endothelial nuclei in glomeruli showed that Robo1-FC–treated mice developed fewer cortical glomerular vessels, whereas juxtamedullary vessels were unaffected (Supplemental Figure 5, B and C), hence mimicking the effects seen in Slit2^iKO mice. Moreover, TALs in the Robo1-Fc–treated mice were dilated as seen in Slit2^iKO (Supplemental Figure 5D).

Slit2 Regulates Glomerular Vascularization in a Robo-Dependent Manner

We next asked if Robo receptors mediated Slit2-regulated glomerular vascular development. TAM was administered to Robo1,2^iKO at P0, P1, and P2, and mice were analyzed at P7 (Figure 4A). TAM-injected Robo1^−/− littermates²³ and age-matched wild-type mice were used as controls. Along with reduced Slit2-AP staining of kidney sections in the Robo1,2^iKO mice (Supplemental Figure 3C), quantification of the number of ERG-positive endothelial nuclei showed that Robo1,2^iKO had fewer endothelial cells in the superficial cortical but not in the juxtamedullary glomeruli compared with Robo1^−/− littermates or with wild-type control mice (Figure 4, B and C). Robo1,2^iKO mice also reproduced the dilated TALs seen in Slit2^iKO mice (Figure 4D), supporting that Robo1 and -2 transduce Slit2 signals in kidney tubules and glomeruli.

Figure 4.

Slit2 signaling through Robo1 and -2 receptors promotes glomerular vascularization. (A) Time line for Rosa-Cre^ERT2–mediated global deletion of the Robo2 fl/fl allele on a Robo1-null background by TAM injections to neonates. Samples were collected at P6 and P7. (B and C) Immunostaining of kidney cryosections with indicated antibodies. Note (B) the reduced endothelial cell number in superficial cortical but not (C) juxtamedullary glomeruli of Robo1,2^iKO. Right panels show quantification of endothelial cell (EC) number expressing ERG protein in individual glomeruli at the superficial cortex versus the juxtamedullary region. Each dot represents an individual mouse. Scale bar: 10 µm. (D) Representative images of THP⁺ TAL, counterstained with nuclear DAPI (blue). Scale bar: 50μm. The right panel shows quantification of the TAL tubular area. Each dot represents an individual mouse. n=5 Robo1^+/+, n=5 Robo1^−/− Robo2^fl/fl, and n=5 Robo1^−/− Robo2^fl/fl Rosa-Cre^ERT2 mice. *, P<0.05; one-way ANOVA test.

Endothelial Robo Receptors Contribute to Glomerular Vascularization

Robo2 immunostaining revealed high expression of Robo2 at the capillary loops of glomeruli (Figure 5A). Robo2 labeled the macula densa cells of the juxtaglomerular apparatus (Figure 5A) and podocytes (Supplemental Figure 6A), as reported previously.²² In addition, we observed partial colocalization of Robo2 with Emcn (Figure 5A) but not with Pdgfrβ⁺ mesangial cells (Supplemental Figure 6A). The partial colocalization of Robo2 with endothelial cells suggested that endothelial Robo signaling could contribute to glomerular vascularization. To test this hypothesis genetically, we generated TAM-inducible endothelial-specific Robo receptor knockouts using Robo1^−/−; Robo2^fl/fl mice intercrossed with the endothelial-specific Cdh5-Cre^ERT2 driver (Robo1,2^iECKO)²⁹ (Figure 5B). TAM was administered at P0, P1, and P2, and mice were analyzed at P7. Endothelial-specific recombination of CDH5Cre in kidney glomeruli was verified using recombination of the mTmG locus,³⁸ and we observed GFP expression in Emcn⁺ glomerular endothelial cells but not in other cell types (Supplemental Figure 6B). Slit2-AP staining of kidney sections from these mice showed that Slit2 robustly bound to cortical tubules of Robo1,2^iECKO mice, indicating that endothelial deletion did not affect tubular Slit2 binding (Figure 5C). Slit2-AP still bound to cortical and juxtamedullary glomerular vascular loops (Figure 5C), showing that endothelial-specific Robo deletion was not sufficient to abolish Slit2 binding. Immunostaining showed that endothelial deletion of Robo2 abolished the colocalization of Robo2 with Emcn but not with Pdx (Figure 5D). The superficial cortical glomeruli in Robo1,2^iECKO mice had underdeveloped or missing capillary loops (Figure 5E). Quantification of the number of ERG-positive endothelial nuclei confirmed that Robo1,2^iECKO mice developed fewer endothelial cells in superficial cortical but not in juxtamedullary glomeruli when compared with Robo1^−/− Robo2^fl/fl littermates or with nonlittermate control mice (Figure 5E). Hence, endothelial Robo signaling contributed to glomerular vascular development in response to Slit2. By contrast to inducible Robo1 and -2 global or endothelial deletion, Robo4^−/− mice developed no vascular defects in the kidney (Supplemental Figure 6C), which is in line with lack of Slit2 binding to Robo4 and lack of a detectable vascular phenotype in Robo4^−/− mice.^39–42

Figure 5.

Endothelial Robo1 and -2 receptors contribute to glomerular vascularization. (A) Costaining of Robo2 and endothelial Emcn reveals overlapping (white) expression pattern. G, glomerulus; MD, macula densa. Scale bars: 10 µm; 2 µm in zoom. (B) Time line for Cdh5-Cre^ERT2–mediated endothelial-specific deletion of the Robo2 fl/fl allele on a Robo1-null background by TAM injections to neonates. Samples were collected at P6 and P7. (C) Slit2-AP binding to Robo1,2^iECKO superficial cortical or juxtamedullary kidney sections. Note that Slit2-AP can still bind to the capillary loops when endothelial Robo receptors are deleted. However, Robo1,2^iECKO mice show glomerular capillary loop defects in the superficial cortical region (arrows). Scale bar: 20 µm. (D) Costaining of Robo1,2^iECKO kidney with antibodies against the indicated proteins. Scale bars: 10 µm; 2 µm in zoom. (E) Immunostaining shows reduced endothelial cell number in Robo1,2^iECKO superficial cortical glomeruli. The right panels show quantification of endothelial cells per glomerulus. Each dot represents the average number of endothelial cells from 15 glomeruli of individual mice. n=5 Robo1^+/+, n=5 Robo1^−/− Robo2^fl/fl, and n=4 Robo1^−/− Robo2^fl/fl Cdh5-Cre^ERT2 mice. Scale bar: 10 µm. *, P<0.05; one-way ANOVA test.

Slit2 Promotes Glomerular Endothelial Proliferation and Migration

To determine the mechanism underlying reduction of glomerular vascularization in Slit2^iKO mice, we quantified proliferation and apoptosis of endothelial cells. At P7, endothelial proliferation and cell division were reduced in Slit2^iKO when compared with controls, whereas cell apoptosis was similar between genotypes (Figure 6A). NG2⁺ mesangial cells, which are recruited by endothelial cells⁵ and normally function to support the vascular loops, were reduced in the superficial cortical glomeruli of Slit2^iKO mice (Figure 6B, Supplemental Figure 7A). The number of WT1⁺ podocytes per glomerulus was similar between controls and Slit2 mutants (Figure 6C, Supplemental Figure 7B). The underdeveloped endothelium in Slit2^iKO cortical glomeruli expressed the glomerular endothelial-specific marker Ehd3⁴³ in addition to other general endothelial markers, indicating that the molecular identity of glomerular endothelium was not changed after Slit2 knockout (Figure 6D). As glomerular vascularization depends on Vegf signaling,⁷ we tested Vegfa and Vegfr2 expression in kidney protein extracts by western blotting and observed increased Vegfa expression but reduced Vegfr2 expression in Slit2^iKO when compared with littermate controls (Figure 6, E–G). To evaluate Vegfa expression histologically, we incubated kidney sections with a commercial Vegfr1 (Flt1)-extracellular domain-Ig fusion protein, which binds Vegfa.⁴⁴ Flt1-Fc binding was increased in the cortical region of Slit2^iKo mice when compared with control littermates, in particular in poorly vascularized glomeruli (Supplemental Figure 7, C and D), which is consistent with hypoxia regulation of Vegfa expression in those areas. As Vegf signaling enhances endothelial migration into the developing glomeruli and the S-shaped body,⁷ we evaluated if this process was affected in Slit2^iKO mice. To do so, mice were treated with TAM at P0–P2 and analyzed at P3, when comma-shaped and S-shaped bodies could still be observed in the kidney. Vegfr2 immunostaining was reduced in Slit2^iKO angiogenic endothelial cells invading the vascular cleft, and the leading tip cells appeared blunted (Figure 6H). Further immune labeling with an antibody recognizing the endothelium tip cell marker ESM1⁴⁵ confirmed blunted tip cell appearance (Figure 6I). We quantified migration into the vascular cleft by measuring the length of the vascular cleft and the length occupied by endothelial cells in the cleft, and we observed a significant decrease in Slit2^iKO (Figure 6J). Altogether, these data indicate that Slit2 promoted glomerular vascularization by stimulating both endothelial proliferation and migration.

Figure 6.

Slit2 promotes glomerular endothelial proliferation and migration. (A) Quantification of endothelial cell (EC) proliferation and apoptosis in cryosectioned and immunostained P7 kidneys using antibodies recognizing pH3, Ki67, and cleaved Capase3. Each dot represents the average number from six images of individual mice. n=5 Slit2^fl/fl mice, and n=6 Slit2^iKO mice. (B) Quantification of mesangial NG2 antibody staining intensity in the superficial cortical glomeruli. Each dot represents an individual mouse. (C) Quantification of podocyte number per superficial cortical glomerulus. Each dot represents an individual mouse. (D) Immunostaining of glomerular endothelial-specific marker Ehd3 and nuclear DAPI on cryosectioned P7 kidney. Scale bar: 10 µm. (E) Western blot of whole-kidney lysates of P7 Slit2^iKO and control mice, showing protein expression of Vegfr2, Vegfa, Slit2, and actin. (F and G) Quantification of Vegfa and Vegfr2 protein level. Each dot represents an individual mouse. (H) Representative image of a P3 Pax2⁺ S-shaped body and endothelial cells labeled with Vegfr2 (yellow arrows). Scale bar: 10 µm. (I) P3 Pax2⁺ comma-shaped body and endothelial cells labeled with Esm1 and Emcn (yellow arrows). Scale bar: 10 µm. (J) Quantification of endothelial cell coverage length per vascular cleft. Each dot represents an individual mouse. *, P<0.05; **, P<0.01; Mann-Whitney U test.

Perfusion Defects in Slit2^iKO Superficial Cortical Glomeruli

To determine if defective vascularization of cortical glomeruli in Slit2^iKO mice led to functional glomerular perfusion defects, we treated pups with TAM between P0 and P2 and injected them at P7 intravenously with fluorescently labeled IsolectinB4 (IB4). IB4 bound to the luminal endothelial membrane and allowed us to visualize patent vessels in the kidney. Pups were euthanized 15 minutes after IB4 injection, and 60-μm-thick sections of the kidney were analyzed by confocal microscopy. Figure 7A shows a representative section through a Cre-negative kidney, where perfused vascular loops are seen in cortical as well as juxtamedullary glomeruli. Slit2^iKO kidneys displayed vascularized glomeruli in the juxtamedullary region, similar to those seen in control mice, whereas the superficial cortical region displayed far fewer normally perfused glomeruli. Apart from these glomeruli, the remainder of the kidney appeared to be perfused normally, and overall morphology of kidney vessels was similar between Cre-negative littermates and Slit2^iKO (Figure 7A) (data not shown). Higher magnification of IB4-injected kidney glomeruli counterstained with podocalyxin showed perfused vessels embedded within podocalyxin-positive podocytes in the control mice (Figure 7B). By contrast, Slit2^iKO cortical glomeruli either displayed small rudimentary vascular loops or were devoid of any perfusion, with vessels remaining at the base of the glomeruli without invading them (Figure 7B, Supplemental Figure 8). Quantification of perfused glomerular area normalized to glomeruli size showed a significant reduction of superficial cortical glomerular perfusion (Figure 7B), whereas the perfusion of juxtamedullary glomeruli was not affected (Figure 7C).

Figure 7.

Slit2^iKO leads to perfusion defects of the superficial cortical glomeruli. (A) Confocal image of perfused glomeruli visualized by intravenous injection of IB4 into mice 15 minutes before sample collection. Note that there are fewer properly perfused glomeruli in the superficial cortical region (arrows) in Slit2^iKO compared with control, whereas perfusion of juxtamedullary glomeruli is not changed. Scale bar: 100 µm. (B and C) Immunostaining with podocalyxin shows glomeruli (arrows). Note the absent or reduced perfusion in the (B) superficial cortical but not (C) juxtamedullary glomeruli of Slit2^iKO. Right panels show quantification of the perfused glomerular region, normalized to glomerular size. Each dot represents the average number from six glomeruli on an individual image field. n=3 Slit2^fl/fl and n=3 Slit2^iKO. Scale bar: 10 µm. **, P<0.01; Mann-Whitney U test.

Podocyte Defects in Slit2^iKO Glomeruli

Electron microscopy of P7 mice after neonatal Slit2 deletion showed that podocytes around glomerular vessels were forming specialized foot processes in control mice (Figure 8A), whereas Slit2^iKO superficial cortical glomeruli lacked endothelial cells, and their podocytes maintained a cuboidal shape characteristic of immature columnar epithelium⁵ (Figure 8A). We quantified the change of podocyte morphology by measuring the width-height ratio of podocyte cell bodies, which was decreased in Slit2^iKO mice superficial cortical glomeruli (Figure 8A), supporting podocyte differentiation defects in the absence of vasculature. In the juxtamedullary region, podocytes still formed foot processes and a filtration barrier with endothelial cells in Slit2^iKO mice, but mild foot process effacement was observed in these glomeruli (Figure 8B), as reported previously in Robo2 mutants.²² In 6-week-old Slit2^iKO and control littermates that had been treated with TAM as neonates, podocytes formed filtration barriers with endothelial cell in both superficial cortical and juxtamedullary glomeruli. However, widened foot processes were observed in both the superficial cortex and juxtamedullary region of Slit2^iKO mice (Figure 8, C and D).

Figure 8.

P7 and 6-week-old Slit2^iKO exhibit ultrastructural defects of glomeruli. (A) Representative EM images from P7 mice with the indicated genotypes treated with TAM at P0–P2, showing podocytes (PCs), glomerular basement membrane (GBM), and capillary lumen (CL) in superficial cortical glomeruli of Slit2^fl/fl, whereas capillary lumens are missing (arrows) in Slit2^iKO. The right panel shows quantification of podocyte cell shape. Each dot represents an individual podocyte. n=5 Slit2^fl/fl, and n=5 Slit2^iKO. Scale bar: 1 µm. (B) Filtration structure of P7 juxtamedullary glomerulus. The right panel shows the quantification of foot process width. Each dot represents an individual foot process. n=5 Slit2^fl/fl and n=5 Slit2^iKO. Scale bar: 0.5 µm. (C and D) Representative transmission electron microscopy images from 6-week-old mice with the indicated genotypes treated with TAM at P0–P2, illustrating filtration structure of (C) superficial cortical and (D) juxtamedullary glomerulus. Right panels show the quantification of foot process width. Each dot represents an individual foot process. n=3 Slit2^fl/fl and n=3 Slit2^iKO. RBC, red blood cell; BS, Bowman’s space. Scale bar: 0.5 µm. *, P<0.05; ****, P<0.0001; Mann-Whitney U test.

Discussion

Our results reveal a novel role of Slit2-Robo signaling during vascularization of glomeruli and nephrogenesis in the postnatal kidney. Using temporally inducible global Slit2 deletion during midgestational embryonic development, we observed cystic kidney formation similar to that previously described in global Robo2 mutants, suggesting that Slit2 interacts with Robo2 to regulate tubular polarization and ciliogenesis via Baiap2 and p53.²¹ As a novel finding, we also observed glomerular angiogenesis defects in the embryonic Slit2^iKO kidneys, supporting proangiogenic effects of Slit2 signaling in glomerular angiogenesis. To better evaluate angiogenesis without the extensive and confounding tubular dilations, we extended the analysis of Slit2-Robo function to the postnatal kidney using a temporal window in the first postnatal week when nephrogenesis in the cortical region of the kidney is still ongoing, whereas it has already been completed in the medullary region.^33,37 Using this time window, we uncovered that interruption of Slit2-Robo signaling still resulted in abnormalities of nephron structure, but these were milder when compared with the embryonic deletion. We observed that MRs but not PCTs bound to Slit2 protein and that TALs of MRs in the Slit2^iKO were malformed, whereas neither CDs nor PCTs were measurably affected by postnatal Slit2 deletion. Two potential mechanisms could account for the TAL tubular defects. As the TALs in Slit2^iKO were shorted but enlarged, Slit2 signaling could regulate their elongation by guiding the cell division plane. Robo2 guides myocardial cell alignment and polarity,⁴⁶ supporting the idea that depletion of Slit2-Robo may lead to loss of polarity and inability to divide along the long axis of the tubular epithelium. Another possibility is that the final differentiation and reorganization of tubular epithelial junctions are induced by shear stress.⁴⁷ Therefore, the reduced perfusion of cortical glomeruli could lead to insufficient shear stress in the tubules and tubular differentiation defects. Nevertheless, kidneys recovered from the neonatal Slit2 deletion, and tubular morphogenesis was not measurably altered in 6-week-old Slit2^iKO kidneys.

We also found that Slit2-Robo signaling is a potent regulator of glomerular vascularization. Slit2 protein bound to renal glomerular vasculature but not to any other vascular bed in the kidney, suggesting a glomerular-specific angiogenic function of this protein. In line with this idea, postnatal deletion of Slit2 or Slit2 ligand trapping reduced vascularization of developing glomeruli by inhibiting endothelial proliferation and invasion in the angiogenic cleft of S-shaped bodies. Interestingly, postnatal Slit2 inhibition only affected vascularization of cortical glomeruli but not the juxtamedullary glomeruli or other renal vasculature, whereas embryonic deletion of Slit2 led to an overall reduction of glomerular vascularization. These results are consistent with a model whereby Slit2 promotes glomerular angiogenesis but is largely dispensable for the already established glomerular vasculature, thereby affecting actively growing embryonic or postnatal superficial glomeruli, whereas juxtamedullary glomeruli that have developed before birth no longer depend on postnatal Slit2 for their maintenance. Consistent with this view, extending the time window of Slit2 deletion to 6 weeks reduced superficial cortical glomerular numbers. The approximately 25% reduction of glomeruli matches the reduction of vascularized glomeruli at P7, suggesting that avascular glomeruli may undergo degeneration. Interestingly, endothelial cells of remaining glomeruli as well as the TALs were similar between control and mutant at 6 weeks of age. We did not observe glomeruli with moderate vascular defects, and the kidney weight and body weight were normalized in 6-week-old Slit2^iKO, which suggests there may be compensatory growth of glomerular ECs and TALs, as seen after nephrectomy and congenital renal agenesis.^48,49 The reduction of glomeruli number in 6-week-old Slit2^iKO caused a mild but nonsignificant increase of ACR. We speculate that glomerular defects reduce functional nephrons in Slit2^iKO kidneys, which are compensated for by the large functional reserve capacity of the kidney.⁵⁰ Nonetheless, the reduction of functional nephrons is known to increase susceptibility for kidney diseases,⁵¹ and further research with acute injury challenge might reveal functional renal defects in Slit2^iKO mice.

These data identify Slit2 as a novel and specific factor that enhances glomerular vascularization. Interestingly, other angiogenic growth factors, such as Angiopoietin-Tie2 signaling, selectively affect vasa recta of the kidney,⁵² suggesting that combinations of different growth factors are used in different compartments of the kidney vasculature to induce vascular heterogeneity^2,53–55 and ultimately, functional complexity in the different renal vascular beds.

Previous studies from our laboratory showed that combined deletion of both Robo1 and Robo2 is required to abolish Slit2 signaling in endothelial cells during retinal angiogenesis,²³ and we detected both Robo1 and Robo2 expression in glomeruli. Thus, we studied the glomerular phenotype by deleting both Robo1 and Robo2 receptors in mice for this study. Slit2 binding to glomeruli was reduced in Robo1^−/− mice when compared with wild-type mice, and it was abolished in mice with deletions of both Robo1 and -2 receptors. Glomerular vascular defects were similar in Slit2 and Robo1 and -2 global knockout mice, strongly suggesting that Slit2 signals via both Robo receptors during this process. Immunostaining of Robo2 with endothelial and podocyte markers showed that Robo2 was expressed by both podocytes and endothelial cells at their basal membrane. Endothelial-specific Robo2 deletion on a Robo1-null background abolished the colocalization of Robo2 with Emcn but did not affect Robo2 expression by podocytes. Robo1,2^iECKO showed a milder phenotype compared with Slit2 and global Robo1 and -2 knockout, which could be due to incomplete Robo2 deletion or to a contribution from nonendothelial Robo2 receptors in podocytes in guiding glomerular vascularization. We favor the latter explanation because Robo2 expression remained in podocytes of Robo1,2^iECKO mice, and we observed foot process effacement in the juxtamedullary glomeruli of P7 and in the glomeruli of 6-week-old Slit2^iKO kidneys. A previous study reported foot process effacement in Robo2 global or podocyte-specific deletion mutants,²² suggesting that the foot process defect is due to loss of Slit2 signaling through Robo2 in podocytes.

Mechanistically, we found Slit2 deletion leads to decreased endothelial proliferation and reduced Vegfr2 expression on angiogenic tip cells invading the vascular cleft of S-shaped bodies, suggesting that defective Vegfr2 signaling in the absence of Slit2 may cause the glomerular vascular defects. Previous studies using human umbilical vein endothelial cells showed that Slit2-Robo1 and -2 and Vegfa-Vegfr2 cooperate to drive endothelial polarization and migration via NCK-mediated activation of Rac and CDC42, respectively.²⁶ Activation of small GTPases occurs following clathrin-independent endocytosis via the EndophilinA2 pathway.²⁵ However, preliminary analysis of endothelial-specific NCK 1 and 2 mutants showed no effects on glomerular vascular development (data not shown), indicating that a different downstream signaling pathway is involved in the renal vasculature.

In summary, we identify Slit2 as a novel and glomerular-specific angiogenic factor that might contribute to the nephron loss seen in patients with congenital abnormalities of the kidney and ureteric tract and Slit2 mutation⁵⁶ and could be useful to treat conditions associated with glomerular endothelial dysfunction.

Disclosures

All authors have nothing to disclose.

Funding

This project has received funding from National Institutes of Health grants 1R01HLI125811, 1R01EY025979-01, and P30 EY026878 and Fondation Leducq Transatlantic Network of Excellence grant Arterial Flow As Attractor for Endothelial Cell Migration (ATTRACT). The George M. O'Brien Kidney Center at Yale receives funding from National Institutes of Health grant P30 DK079310.

Supplemental Material

This article contains the following supplemental material online at http://jasn.asnjournals.org/lookup/suppl/doi:10.1681/ASN.2020111640/-/DCSupplemental.

Supplemental Figure 1. Renal defects following midgestation Slit2 deletion.

Supplemental Figure 2. Characterization of morphologic changes in Slit2^iKO at P7.

Supplemental Figure 3. Validation of the specificity of Slit2-FL-AP staining.

Supplemental Figure 4. Characterization of the glomerular vascular effect of Slit2.

Supplemental Figure 5. Pharmacologic inhibition of Slit2 with Robo1-Fc.

Supplemental Figure 6. Contribution of endothelial Robos to the glomerular phenotype.

Supplemental Figure 7. Phenotypic change of superficial cortical glomeruli at P7.

Supplemental Figure 8. IB4-488–perfused glomeruli.