Loss of the VEGF₁₆₄ and VEGF₁₈₈ Isoforms Impairs Postnatal Glomerular Angiogenesis and Renal Arteriogenesis in Mice

Abstract

ABSTRACT. Vascular endothelial growth factor (VEGF) is transcribed in the VEGF₁₂₀, VEGF₁₆₄, or VEGF₁₈₈ isoforms, which differ in receptor binding, matrix association, and angiogenic activity. This vascular growth factor has been implicated in the development of the renal vasculature, but the role of the distinct VEGF isoforms remains unknown. In the present report, renal angiogenesis and arteriogenesis were studied in VEGF^120/120 mice, expressing only the short VEGF₁₂₀ isoform. In VEGF^120/120 mice, ingrowth and survival of capillaries in glomeruli, remodeling of peritubular capillaries, vascular coverage by pericytes, and branching of renal arteries were all severely impaired, causing abnormal glomerular filtration and impairing renal function. The arterial branching defect might be related to a reduced expression of renin, a presumed renal arterial branching factor. Glomerulosclerosis and tubular dilation possibly resulted from renal ischemia caused by vascular defects. Thus, VEGF₁₆₄ and VEGF₁₈₈ not only mediate angiogenesis, but they also play an essential role in renal branching arteriogenesis.

The mechanisms that regulate development of capillaries (angiogenesis) and arteries (arteriogenesis) in the kidney remain largely unknown (^1,2). During renal development, mesenchymal cells condense around branches of the ureter into a comma-shaped body, which matures in an S-shaped glomerulus. The latter becomes invaded by endothelial cells (EC), which assemble into a single glomerular capillary loop (cup-shaped glomerulus), which subsequently expands into a complex tuft of branched capillaries (mature glomerulus) (¹). Mesangial cells, which share a common origin with smooth muscle cells and pericytes, also infiltrate the glomerulus. The renal arterial vasculature is a complex, highly patterned network in which a single renal artery ramifies (from the hilus to the cortex) into lobar arteries, arcuate arteries (coursing through the corticomedullary junction), interlobular arteries and, finally, into afferent arterioles (³). Each nephron contains afferent and efferent arterioles, a capillary tuft, and peritubular capillaries. In the mouse, a large fraction of the nephrons and their blood vessels expand after birth (³).

Several angiogenic growth factors and receptors are expressed in the kidney during vascular development. However, few transgenic studies have addressed the functional role of these factors in renal vascular development or disease (for review, see references (² and ⁴). Loss of PDGF-BB or PDGFR-β results in defective glomerular mesangial cell recruitment, but it does not affect endothelial angiogenesis or smooth muscle arteriogenesis (^5–7), whereas deficiency of components of the renin-angiotensin system impairs primarily branching of renal arteries (⁸).

VEGF is a key player in angiogenesis. It is alternatively processed in at least three isoforms in the mouse (VEGF₁₂₀, VEGF₁₆₄ and VEGF₁₈₈), which differ in receptor specificity, mitogenic activity, heparin binding, and tissue-specific expression (⁹). VEGF and its receptors, VEGFR-1, VEGFR-2, and neuropilin-1 (a VEGF₁₆₅-specific receptor), have also been implicated in renal vascular development (^10,11). In embryonic kidney, VEGF is expressed in epithelial cells in S-shaped bodies and in collecting ducts, suggesting a role in attracting capillary sprouts into the developing glomerulus and around tubules (¹⁰). In the adult, persistent VEGF expression in tubular and glomerular epithelial cells in the vicinity of fenestrated capillaries suggests an involvement in the induction and maintenance of fenestrations (^10,12,13). In the embryonic and adult mouse kidney, transcript levels are highest for VEGF₁₆₄ (accounting for approximately 60% of the total VEGF mRNA present), intermediate for VEGF₁₂₀, and lowest for VEGF₁₈₈ (^11,14). VEGF receptors are expressed in glomerular and peritubular EC in the embryo and adult (^10,11,15) and have been implicated in tubulogenesis and mesangial sclerosis (¹⁶). A direct effect of VEGF on renal vascular growth has been demonstrated in cultured embryonic kidney (¹⁷). In addition, conditional VEGF gene inactivation or administration of VEGF antagonists during the early postnatal period impairs renal vascular development (^18,19). Conversely, increased VEGF levels have been detected in vascularized renal tumors or during reparative angiogenesis in inflammatory renal disorders.

Although these findings indicate that VEGF plays a significant role in renal vascular development, the distinct role of the different VEGF isoforms remains unknown. To study the differential role of these distinct isoforms in vascular growth, we studied renal vascular development in VEGF^120/120 mice, expressing only the VEGF₁₂₀ isoform (¹⁴). These mice survive embryogenesis, but suffer ischemic heart failure due to impaired myocardial angiogenesis (¹⁴). Here, we characterized the defects of renal vascular development in VEGF^120/120 mice and discuss the possible roles of VEGF₁₆₄ and VEGF₁₈₈ in renal angiogenesis and arteriogenesis.

Materials and Methods

VEGF^120/120 Mice

All animal experiments were conducted in accord with the NIH Guide for the Care and the Use of Laboratory Animals. Littermate offspring from VEGF^+/120 breeding pairs were genotyped by PCR as described previously (²⁰). VEGF^+/+ and VEGF^120/120 littermates were used at the indicated age: embryonic day (E) E15.5, E17.5, postnatal day (P) P0.5, P3, or P6. Blood was collected from anesthetized P6 pups via carotid puncture. Plasma levels of electrolytes and creatinine were measured using standard methods.

Immunostaining and Ultrastructural Analyses

Kidneys from anesthetized VEGF^+/+ and VEGF^120/120 littermates were perfusion-fixed with 4% paraformaldehyde and 6-μm paraffin sections were stained (¹⁴). The following antibodies were used: CD34 (Pharmingen, San Diego, CA), smooth muscle α-actin (SMA, Sigma Chemical Co., St Louis, MO), Wilms’ tumor antigen (Santa Cruz Biotechnology, Santa Cruz, CA), Tamm Horsfall glycoprotein (Cortex Biochem, San Leandro, CA), laminin (Sigma), fibronectin (Sigma), collagen IV (gift from Dr. Foidart, Liège, Belgium), and fibrin (Nordic Immunologies, Tilburg, The Netherlands). Immunostaining for VEGF receptor-2 (VEGFR-2/Flk1, Imclone, NY), VEGF receptor-1 (VEGFR-1/Flt1, Imclone, New York, NY), endoglin (Pharmingen), and CD31 (Pharmingen) was performed on cryosections of kidneys frozen in Tissue-Tek immediately after dissection. Renin immunostaining was performed on kidneys fixed in Bouin solution (²¹). Specificity of the stainings was confirmed by replacement of the primary antibody with isotype-matched nonimmune IgG or serum. Morphometric analysis of glomeruli was performed on parasagittal sections after hematoxylin-eosin (H&E) staining, using the Quantimet Q600 imaging system (Leica, Nussloch, Germany). The number of proximal and distal tubuli, loops of Henle, vascularized glomeruli, glomerular capillary loops, and arteries was counted on mid-parasagittal sections stained for PAS, Tamm Horsfall glycoprotein, CD34, laminin, or SMA, respectively. The thickness of the nephrogenic cortex was measured on parasagittal sections perpendicularly to the papillae. Ultrastructural analyses were performed as described (^20,22). Horseradish peroxidase (HRP, Sigma, 10 mg/100 g body wt of 10 mg/ml HRP in 0.15 M NaCl, pH 7.0) was injected intravenously in pups. After 3 min, the kidneys were dissected and postfixed in sodium cacodylate buffer, pH 7.2 containing 2% glutaraldehyde and 3% paraformaldehyde. Small (≈1 mm³) kidney blocks were incubated in DAB/H₂O₂ staining solution and further processed for standard ultrastructural analysis.

In Situ Hybridization

In situ hybridization was performed (²³) using sense and antisense ³⁵S-labeled CTP riboprobes for VEGFR-1 (nct 421 to 1403), VEGFR-2 (nct 3361 to 4440), Tie1 (nct 2853 to 3751), Tie2 (nct 2807 to 3500), and PDGF receptor-β (PDGFR-β, nct 1668 to 3743). Hybridized slides were coated with photographic emulsion (Amersham Life Science, Buckinghamshire, UK) and exposed for 8 wk.

Results

Macroscopic Analyses of VEGF^120/120 Kidneys

Macroscopic examination revealed that the kidneys of VEGF^120/120 mice were smaller than those of VEGF^+/+ (wild-type [WT]) mice at all ages examined. The cross-sectional area of the kidneys, measured on mid-parasagittal sections through the hilus, at P0.5, P3, and P6 was: 4 ± 0.2, 6 ± 0.5, and 12 ± 0.4 mm² in VEGF^+/+ mice versus 2.5 ± 0.2, 3.5 ± 0.3, and 6.6 ± 0.4 mm² in VEGF^120/120 mice (n = 6; P < 0.05). Growth of the kidneys was, however, proportional to that of the total body. The kidney/total body weights (mg/g) at P0.5 and P6 were: 0.52 ± 0.05 and 0.65 ± 0.05 versus 0.48 ± 0.10 and 0.72 ± 0.13, respectively (n = 10; P = NS). VEGF^+/+ kidneys were uniformly reddish pink, whereas VEGF^120/120 kidneys were generally more pale, suggesting impaired kidney perfusion.

Impaired Nephrogenesis in VEGF^120/120 Mice

Glomerulogenesis.

Accumulation and maturation of glomeruli were impaired in VEGF^120/120 mice. For reasons of standardization and comparison, the number of glomeruli (as well as of all other parameters, see below) were counted on mid-parasagittal sections through the hilus. Compared to VEGF^+/+ mice, VEGF^120/120 mice contained approximately 25% fewer glomeruli (glomeruli/kidney section at P0.5, P3, and P6: 70 ± 2, 144 ± 9, and 136 ± 10 in VEGF^+/+ mice versus 52 ± 2, 107 ± 6, and 104 ± 7 in VEGF^120/120 mice; n = 6; P < 0.05). The nephrogenic cortex was also approximately 25% smaller in mutant mice (at P3: 260 ± 20 μm in VEGF^120/120 mice versus 370 ± 30 μm in VEGF^+/+ mice; n = 5; P < 0.05; Figure 1, A and B). VEGF^120/120 kidneys were smaller; therefore, the glomerular density was comparable in both genotypes (number of glomeruli per mm² at P0.5, P3, and P6: 18 ± 2, 24 ± 3, and 11 ± 3 in VEGF^+/+ mice versus 20 ± 2, 30 ± 4, and 16 ± 3 in VEGF^120/120 mice; n = 6; P = NS). Unlike in VEGF^+/+ mice, a fraction of the glomeruli in VEGF^120/120 mice enlarged and became sclerotic. The number of sclerotic/total glomeruli at P0.5, P3, and P6 was: 0/70 (0%), 1/144 (0.6%), and 0/136 (0%) in VEGF^+/+ mice versus 0/52 (0%), 7/107 (6.5%), and 18/104 (17%) in VEGF^120/120 mice (n = 6; P < 0.05). The majority of these abnormal glomeruli accumulated an unusual amount of collagen type IV, fibrin, fibronectin (not shown), and laminin (Figure 1, C and D), as documented by light microscopy after immunostaining using di-amino-benzidine as color reagent. Ultrastructural analyses confirmed the presence of amorph extracellular material in sclerotic glomeruli (Figure 5, B and D). When laminin was visualized with a more sensitive immunofluorescence staining, excessive matrix deposition was already detectable in fetal mature glomeruli beyond E17.5 (Figure 4, C and D). By P6, sclerotic glomeruli in VEGF^120/120 mice were approximately threefold larger than mature glomeruli in VEGF^+/+ mice (5100 ± 440 μm²versus 1500 ± 75 μm²; n = 5; P < 0.05).

Figure 1.: Abnormal glomerulogenesis and tubulogenesis in VEGF^120/120 mice. (A and B), hematoxylin-eosin (H&E) staining of kidney sections at postnatal day (P) P0.5. The nephrogenic cortex (bar) is thinner in VEGF^120/120 mice (B) than in VEGF^+/+ mice (A) (arrowhead, comma-shape glomerulus; arrow, S-shape glomerulus). (C and D) Unusual accumulation of laminin (LN), immunostained using di-amino-benzidine as color reagent, in sclerotic VEGF^120/120 glomeruli (D) but not in mature VEGF^+/+ glomeruli (C) at P6. (E and F) Fewer and enlarged loops of Henle (arrows) in VEGF^120/120 (F) but not in VEGF^+/+ (E) mice (Tamm Horsfall glycoprotein [THg] staining). Magnification bar: 100 μm in panels A and B; 50 μm in C through F.

Figure 5.: Ultrastructural analysis of abnormal glomerular capillarization. (A and C) VEGF^+/+ mice: the mature glomerulus in panel A contains a peripheral rim of podocytes (P) and several, more centrally located, intact capillaries (C; capillary lumen contains electrodense horseradish peroxidase [HRP]). Panel C shows a capillary loop with its EC and surrounding P at higher magnification. No HRP was used in panel C. (B and D) VEGF^120/120 mice: the sclerotic glomerulus in panel B is completely devoid of capillary loops, which were replaced by an amorphous necrotic debris (*); P are still present. At a higher magnification (panel D), the amorph necrotic debris of a ghost endothelial remnant in a sclerotic glomerulus is displayed (*); the arrow denotes HRP in the residual lumen of the disintegrated capillary loop. (E) Peritubular capillary in a VEGF^+/+ mouse with a small lumen (filled with HRP) and flat aligned EC. Note the tight assembly of the vessel. (F) Dilated peritubular capillary in a VEGF^120/120 kidney exhibiting irregular EC (extremely thin at certain locations: arrows; and cytoplasmic protrusions in the lumen: arrowheads), surrounded by large intercellular spaces. Magnification bar: 10 μm in panels A and B; 5 μm in C and D; and 2 μm in E and F.

Figure 4.: Abnormal glomerular vascularization in VEGF^120/120 glomeruli. (A and B) Immunostaining for VEGFR-2 at P0.5, revealing strongly stained EC in glomerular capillary loops (arrow) and peritubular capillaries (arrowhead) in VEGF^+/+ kidneys (A). In contrast, in VEGF^120/120 glomeruli (B), expression of VEGFR-2 in capillary loops (arrow) and peritubular capillaries (arrowhead) is much weaker. (C and D) Visualization of basement membranes by immunofluorescence staining for LN. In VEGF^+/+ glomeruli at P0.5 (C), numerous capillary loops (identified by their microvascular basement membrane) are orderly arranged and the glomerular basement membrane (GBM) appears regular. In VEGF^120/120 glomeruli (D), there are fewer and disordered capillary loops, the GBM often exhibits a variable shape and thickness, and it appears more irregular and tortuous. Some VEGF^120/120 glomeruli also contain an unusual amount of LN—an initial sign of glomerulosclerosis (*). Magnification bar, 50 μm in all panels.

Tubulogenesis.

VEGF^120/120 mice had fewer proximal convoluted tubules (tubules with PAS-positive brush border per optical field at P0.5, P3, P6: 15 ± 1, 17 ± 1, 22 ± 2 in VEGF^+/+ mice versus 11 ± 1, 13 ± 1, 10 ± 1 in VEGF^120/120 mice; n = 10; P < 0.05). Proximal convoluted tubules also enlarged in VEGF^120/120 mice (cross-sectional area at P3, P6: 400 ± 60 μm², 350 ± 30 μm² in VEGF^+/+ mice versus 1800 ± 300 μm², 2100 ± 270 μm² in VEGF^120/120 mice; n = 5; P < 0.05). Tamm-Horsfall glycoprotein immunostaining revealed fewer, but often enlarged and tortuous, loops of Henle in mutant mice (loops per optical field: 23 ± 1 in VEGF^+/+ mice versus 14 ± 1 in mutant mice at P6; n = 6; P < 0.05; Figure 1, E and F). Distal convoluted tubules appeared indistinguishable in size and number in both genotypes. Ultrastructural signs of ischemia (mitochondrial clarification and vacuolization, swelling, plasmalemma blebbing) were detected in the proximal tubules and loops of Henle in VEGF^120/120 mice (Figure 6A).

Figure 6.: Ultrastructural analysis of an abnormal proximal tubule and impaired glomerular filtration. (A) An epithelial cell in a proximal tubule (*) exhibits signs of ischemia, including mitochondrial clarification, disruption of the mitochondrial cristae, cyotplasmic vacuolization, intracellular edema, and swelling. Note also the abnormal peritubular capillary (C) with its irregular EC. (B) After intravenous injection of HRP in VEGF^+/+ mice, minimal amounts of HRP leak through the glomerular filter and are reabsorbed by the brush border of proximal convoluted tubuli (the HRP in the brush border is electrodense). (C) In contrast, in VEGF^120/120 mice, abundant HRP leaks through the glomerular filter and is reabsorbed by the brush border of proximal convoluted tubuli; note also abundant electrodense HRP in intracellular endocytic vesicles (arrowhead). Magnification bar: 2 μm in A; 1 μm in B and C.

Endothelial Cell Defects in Glomerular and Peritubular Vessels

EC in Glomeruli.

To examine vascularization of glomeruli, we used a panel of antibodies specific for endothelial markers (CD34, VEGFR-2/Flk1, VEGFR-1/Flt1, endoglin, and CD31). All antibodies yielded comparable results, but EC stained more strongly for CD34 (Figure 2, A through D) and VEGFR-2 (Figure 4, A and B; see below). In VEGF^+/+ mice, vascularization of glomeruli had already commenced at E15.5 and proceeded during the first week after birth. The number of CD34-positive glomeruli progressively increased during prenatal and postnatal development (Figure 3A). Compared with the weaker staining at E15.5, E17.5 and P0.5 (Figure 2A), CD34 staining became very prominent and abundant at P6, reflecting more mature glomerular vascularization (Figure 2C). Individual capillary loops were visualized using an immunofluorescent staining for laminin, present in the microvascular glomerular basement membrane (Figure 4C and D). Laminin-positive capillary loops always colocalized with CD34- or VEGFR2-immunoreactive EC (not shown). EC initially formed a single capillary loop in cup-shaped glomeruli and progressively expanded into a more complex capillary tuft in mature glomeruli (Figure 4C). On average, the number of capillary loops per glomerulus increased about 1.4-fold from E15.5 until birth (Figure 3B). Ultrastructurally, immature glomeruli contained a few capillaries (not shown), whereas mature glomeruli contained numerous capillaries with EC, apposed to the glomerular basement membrane (GBM) (Figure 5, A and C).

Figure 2.: Abnormal glomerular vascularization in VEGF^120/120 mice. (A and C) Normal glomerular vascularization in VEGF^+/+ mice. Compared with cup-shaped glomeruli at P0.5 (panel A, arrow), CD34-staining in mature glomeruli at P6 (panel C, arrow) is stronger and more abundant. Peritubular capillaries (arrowheads in panels A and C) are disorganized and dilated at P0.5, but they become more regularly spaced by P6. (B and D) Abnormal glomerular vascularization in VEGF^120/120 mice. Compared with wild-type (WT) mice (A and C), the CD34-staining of mutant glomeruli is weaker at P0.5 (arrows in panel B) and P6 (arrow in panel D). CD34-positive endothelial cells (EC) were absent in sclerotic glomeruli at P6 (* in panel D). Peritubular capillaries remain dilated and irregular (arrowheads in panels B and D). Magnification bar, 50 μm in all panels.

Figure 3.: Abnormal glomerular vascularization in VEGF^120/120 mice. Reduced vascularization in VEGF^120/120 glomeruli. (A) The number of CD34-positive vascularized glomeruli per parasagittal section was reduced in VEGF^120/120 compared with WT mice at all ages (mean ± SE; * P < 0.01 versus VEGF^+/+). (B) The number of LN-positive capillary loops per glomerulus was reduced in VEGF^120/120 compared with WT mice (mean ± SE; * P < 0.01 versus VEGF^+/+).

VEGF^120/120 mice exhibited severe glomerular vascularization defects. CD34-staining revealed that fewer glomeruli were vascularized at all ages, and, by P6 approximately 40% fewer glomeruli contained CD34-positive EC (Figure 3A). Notably, the vascularization deficit (40% fewer vascularized glomeruli) was greater than the glomerulogenesis deficit (25% fewer glomeruli) in VEGF^120/120 mice. When glomeruli became vascularized, they contained fewer and more weakly stained CD34-positive EC (Figure 2, B and D) and fewer laminin-positive capillary loops (Figure 3B; Figure 4D) than their WT littermates. Similar findings were obtained with the other endothelial markers (not shown). Ultrastructural analyses revealed that EC detached from their underlying basement membrane and exhibited signs of cellular degeneration and death (blebbing, vacuolization, nuclear desintegration). In sclerotic glomeruli, the entire capillary bed often disappeared and was replaced by amorphous granular necrotic debris (Figure 5, B and D). These data suggest that loss of VEGF₁₆₄ and VEGF₁₈₈ severely impaired development, maintenance, and integrity of glomerular capillaries.

Peritubular Capillaries.

CD34 staining revealed that peritubular capillaries in kidneys from VEGF^+/+ and VEGF^120/120 neonates at P0.5 were irregular, dilated, and randomly scattered amid the tubules (Figure 2, A and B). By P6, peritubular capillaries in VEGF^+/+ mice had a small smooth lumen and were assembled in an organized pattern between tubules at regular distances (Figure 2C). In contrast, peritubular capillaries in VEGF^120/120 kidneys remained dilated, were tortuous and irregular, and lacked symmetric patterning (Figure 2D). Ultrastructurally, the peritubular capillaries in WT mice contained intact, regular, and tightly interacting EC (Figure 5E), whereas EC in peritubular capillaries from mutant mice contained luminal cytoplasmic protrusions and blebs or were thin and elongated (Figure 5F). The larger intercellular spaces around the peritubular capillaries in VEGF^120/120 kidneys might represent a sign of intercellular edema resulting from impaired inter-endothelial and/or pericyte-endothelial cell contacts or abnormal cell-matrix interactions (Figure 5F). VEGF^120/120 kidneys likely suffered ischemia, because the distance between neighboring peritubular capillaries at P6 was enlarged in transgenic as compared with WT kidneys (39 ± 2 μm in VEGF^+/+ kidneys versus 66 ± 3 μm in VEGF^120/120 kidneys; n = 5; P < 0.05).

Endothelial Differentiation.

EC differentiation was further analyzed by immunolocalization and in situ hybridization. VEGFR-2 immunoreactivity was undetectable in immature comma-stage glomeruli, but it was present in cells surrounding these immature glomeruli, i.e., in putative EC that were about to sprout into the glomerulus (not shown). VEGFR-2 was also expressed in vascularized glomeruli (Figure 4A) and in cells, lying intermingled between tubules, i.e., at the location where peritubular capillaries develop. Comparable findings were obtained by in situ hybridization of VEGFR-2 transcripts (not shown). Similar expression patterns were also observed after immunostaining for VEGFR-1, endoglin, and CD31 as well as after in situ hybridization for VEGFR-1 and the endothelial receptors Tie1 and Tie2 (not shown). In general, endothelial markers appeared to be expressed in similar cell types in both genotypes. VEGF^120/120 glomeruli were, however, more weakly stained, because they contained fewer capillary loops (Figure 4B). Staining became even undetectable in sclerotic glomeruli when endothelial defects were severe. VEGF₁₆₅ has also been implicated in the development of endothelial fenestrations (²⁴). Nonetheless, ultrastructural analyses revealed endothelial fenestrations in capillaries of VEGF^120/120 mice, like in VEGF^+/+ mice (not shown).

Pericyte/Mesangial Cell and Smooth Muscle Cell Defects in VEGF^120/120 Mice

Pericytes.

PDFGR-β is a marker of pericytes around peritubular capillaries and of pericyte-like mesangial cells in glomeruli (^5,25). In VEGF^+/+ mice, PDGFR-β was initially detected in cells around avascular glomeruli, presumably in mesangial precursors (Figure 7A). When the glomeruli became vascularized by capillary loops, PDGFR-β was expressed by cells (presumably mesangial cells) within the mature glomerulus. PDGFR-β was also detectable in VEGF^120/120 glomeruli, although its expression was generally weaker (Figure 7B). Mesangial cells can be recognized by electron microscopy as voluminous cells intermingled in a dense mesangial substance between capillary loops in mature vascularized glomeruli. Putative mesangial cells were observed in vascularized glomeruli in both genotypes. However, when VEGF^120/120 glomeruli became sclerotic and lost their capillary network, mesangial cells also gradually disappeared. PDGFR-β was also detected in pericytes around peritubular capillaries (not shown). Fewer pericytes were detected around peritubular capillaries in transgenic kidneys, although their precise number could not be reliably quantified.

Figure 7.: Smooth muscle cell and pericyte/mesangial cell defects in VEGF^120/120 mice. (A and B) In situ hybridization of PDGFR-β in VEGF^+/+ (A) and VEGF^120/120 (B) kidneys, revealing reduced expression of PDGFR-β in VEGF^120/120 glomeruli at P0.5. Expression of PDGFR-β was detected within glomeruli (panel A, arrowhead) and in the immediate vicinity of immature glomeruli (panel A, arrow). (C and D) smooth muscle cell a-actin staining, revealing an arcuate artery (AA), interlobular arteries (IA), and afferent/efferent arterioles (AEA). IA and AEA are more numerous in VEGF^+/+ (C) than in VEGF^120/120 (D) mice. (E and F) Renin expression (arrows) at arterial branch points is reduced in VEGF^120/120 (F) compared with VEGF^+/+ (E) mice. Magnification bar: 100 μm in panels C, D, E, and F; 50 μm in A and B.

Smooth Muscle Cells.

Immunostaining for smooth muscle alpha-actin (SMA) revealed that the number of interlobular arteries doubled from birth until P6 in VEGF^+/+ mice, whereas the number of afferent/efferent arterioles increased fivefold (Table 1). In VEGF^120/120 mice, there were consistently fewer SMA-stained arteries at all postnatal ages (Figure 7, C and D). Lobar arteries developed normally, but further branching into arcuate, interlobular, and afferent/efferent arterioles progressively failed in VEGF^120/120 mice. The afferent/efferent arterioles of the glomeruli were most significantly affected (Table 1; Figure 7D). Immunoreactivity for renin, implicated in the branching of renal arteries (^21,26), was detected in the smallest identifiable renal arterioles in VEGF^+/+ mice (Figure 7E), consistent with its expression pattern at arterial branch points (²¹). The number of renin-expressing branch points per kidney section increased from 20 ± 1 at P0.5 to 45 ± 2 at P3 in WT mice. In contrast, there were fewer renin-positive branch points in transgenic kidney sections at P0.5 (14 ± 1; n = 5; P < 0.05) and at P3 (15 ± 1; n = 5; P < 0.05; Figure 7F).

Table 1: Impaired arterial branching and arteriogenesis in VEGF^120/120 mice^a

Impaired Renal Function in VEGF^120/120 Mice

The glomerular filter consists of capillary EC, a GBM (shared by EC and podocytes), and epithelial podocytes (²⁷). In contrast to the severe vascular defects, only minimal epithelial defects were observed in VEGF^120/120 mice. Immunostaining for Wilms tumor antigen staining revealed that the visceral epithelial podocytes were represented in normal numbers (podocytes per glomerular section at P0.5, P6: 17 ± 1, 15 ± 1 in VEGF^+/+ mice versus 16 ± 1, 17 ± 1 in VEGF^120/120 mice; n = 6; P = NS). Ultrastructurally, parietal epithelial cells in the capsule of Bowman and podocytes appeared normal in most glomeruli in VEGF^120/120 mice and persisted, even when glomeruli became sclerotic. Immunofluorescence staining for laminin revealed that the GBM was often irregular, tortuous, and of variable thickness in mature glomeruli in VEGF^120/120 mice beyond E17.5 (Figure 4, C and D).

The functional consequences of the vascular defects on renal function were evaluated. Because of the fragility of the VEGF^120/120 neonates, the GFR could not be determined by measuring creatinine clearance. Glomerular filtration in P6 VEGF^120/120 mice was impaired, as revealed by the elevated levels of plasma creatinine (0.23 ± 0.01 mg/dl in WT mice versus 0.75 ± 0.03 mg/dl in VEGF^120/120 mice; n = 5; P < 0.05) and urea (100 ± 12 mg/dl in WT mice versus 290 ± 45 mg/dl in VEGF^120/120 mice; n = 5; P < 0.05). Sodium chloride levels were normal in transgenic mice. To estimate proteinuria, neutral HRP was injected intravenously. Only minimal amounts of HRP normally pass through the glomerular filter, and the little amount of HRP in the urine is reabsorbed by the proximal tubuli (²⁸). As expected, HRP slightly leaked through the glomerular filter and was detected in the brush borders of proximal tubules in VEGF^+/+ mice (Figure 6B). In contrast, HRP leaked through the glomerular filter in the urine in VEGF^120/120 mice and was detected on the brush border, in microvilli, and in intracellular endocytic vesicles in proximal tubules (Figure 6C). Leakage of HRP in VEGF^120/120 mice was already detectable at P0.5, but it increased by P6.

Discussion

This study demonstrates that VEGF₁₆₄ and VEGF₁₈₈ are essential for vascular development in the kidney and that VEGF₁₂₀, by itself, is insufficient to promote the development and/or support the maintenance of the renal vasculature. Remarkably, the longer isoforms of VEGF, a growth factor believed to be highly specific for EC, were not only essential for angiogenesis (Figure 8A) but also for arteriogenesis (Figure 8B). In addition, correct branching of the renal arterial tree depends on the presence of all three VEGF isoforms (Figure 8B).

Figure 8.: Glomerular and arterial defects in VEGF^120/120 mice. (A) In WT mice, avascular epithelial cell aggregates (comma stage) are progressively colonized by EC (S stage; EC in red) and perivascular mesangial cells (cup stage; mesangial cells in green). The immature capillary network further develops into a highly branched complex capillary tuft in the mature glomerulus. In VEGF^120/120 mice, avascular glomeruli arise but their subsequent vascularization is impaired; fewer capillary loops develop in nascent glomeruli, the glomerular capillary tuft remains immature, and other capillaries disintegrate (indicated by the shaded pink color). Glomerulosclerosis develops in parallel with the disappearance of the capillary network. (B) The renal arterial network in WT mice consists of lobular arteries (LA), arcuate arteries (AA), and interlobular arteries (IA), which finally branch into smaller afferent/efferent arterioles (AEA). In VEGF^120/120 mice, postnatal arterial branching is impaired and only the LA (developing before or at birth) are normally represented. In contrast, the AA, but especially the IA and AEA (developing primarily postnatally) are significantly underrepresented, and a fraction of glomeruli becomes sclerotic.

Loss of VEGF₁₆₄ and VEGF₁₈₈ caused several endothelial defects. Compared with WT mice, 25% fewer glomeruli developed, but up to 40% fewer glomeruli were vascularized by CD34-positive capillaries in VEGF^120/120 mice, indicating that ingrowth of capillaries in nascent glomeruli was impaired. When glomerular vascularization occurred, fewer individual capillary loops per glomerulus developed, they failed to expand into a complex tuft, and they ultimately disintegrated. How can these vascular defects be explained? First, the longer VEGF isoforms may provide essential signals for growth, survival, and infiltration of EC in glomeruli. Compared with VEGF₁₆₄, the VEGF₁₂₀ isoform binds with a lower affinity to VEGFR-1 (²⁹), is less potent in stimulating endothelial growth (³⁰), and might therefore stimulate endothelial function less efficiently than VEGF₁₆₄. Second, VEGF₁₆₄ and VEGF₁₈₈, in binding the extracellular matrix, may provide matrix-associated guidance cues that facilitate endothelial migration in the glomerulus. VEGF is produced by visceral epithelial podocytes and mesangial cells (^12,13). These cells may lay down a VEGF-gradient or “trail” along which EC migrate to the correct location in the glomerulus to form their complex vascular tuft. Such cues would likely be absent in VEGF^120/120 mice, in which diffusion of the soluble VEGF₁₂₀ isoform would misguide endothelial migration. Third, matrix-associated VEGF isoforms may also play a role in keeping the capillaries tightly apposed to mesangial cells and podocytes within the glomerulus. Fourth, the VEGF₁₆₄ (or even perhaps the VEGF₁₈₈) isoform may interact with neuropilin-1, which is expressed in glomerular capillaries (¹¹). Neuropilin-1, as a co-receptor of VEGFR-2 (^31,32), could provide specific or more potent VEGF signals, essential for optimal glomerular capillarization. Fifth, we cannot exclude that the renal phenotype was influenced by the relative overexpression of a single isoform. Indeed, the VEGF^120/120 mice expressed the VEGF₁₂₀ isoform at a level comparable to the total expression of the three isoforms in WT mice (¹⁴).

These findings extend previous reports that endothelial fenestration, a typical feature of differentiated glomerular EC, still occurred in VEGF^120/120 mice. This is not all that surprising, because VEGF₁₂₀ alone induces endothelial fenestrations in vitro (²⁴), and fenestrations still developed after suppression of VEGF in vivo (¹⁹). Thus, endothelial fenestration can be modulated by, but is not critically dependent on, VEGF.

Loss of VEGF₁₆₄ and VEGF₁₈₈ also impaired arteriogenesis by pericytes and smooth muscle cells and impaired the accumulation of pericyte-like mesangial cells in glomeruli. Reduced recruitment of pericytes might also have contributed to the immature remodeling of peritubular capillaries. The long VEGF isoforms could act directly on mesangial cells (³³) or on pericytes (³⁴), because VEGF stimulates migration and proliferation of these cells in vitro (³⁵). Alternatively, it is possible that the impairment of pericyte recruitment is attributable to the endothelial defects in VEGF^120/120 mice, because EC release pericyte-recruitment signals such as PDGF-BB (³⁶), and expression of the latter is reduced in VEGF^120/120 mice (¹⁴).

Little is known about the role of arteriolar growth in glomerular development. Nonetheless, both processes are closely linked, because glomeruli can only mature into functional filtration units when they become vascularized. The finding that the number of arterioles is more remarkably reduced than the number of mature glomeruli in VEGF^120/120 kidneys may suggest a primary defect in arterial branching with secondary impairment of glomerular development. The renal arterial branching defects in VEGF^120/120 kidneys are reminiscent of those in mice lacking components of the renin-angiotensinogen system (⁸). Renin, a key enzyme in the production of angiotensin II, is expressed at the front of migrating smooth muscle cells during branching of renal arteries (^21,26). Notably, expression of renin was significantly reduced in VEGF^120/120 kidneys. VEGF is known to stimulate expression of angiotensin-converting enzyme (³⁷) and production of angiotensin II (³⁸). In turn, VEGF is also upregulated by renin and angiotensin II (^39,40). Thus, a positive angiogenic feedback pathway between both factors may influence renal arteriogenesis.

VEGF^120/120 mice had significantly fewer glomeruli and a thinner nephrogenic cortex. However, glomerular density i.e., glomeruli per mm²) was not reduced in VEGF^120/120 mice because VEGF^120/120 kidneys were smaller. Thus, loss of the large VEGF isoforms did not affect glomerulogenesis per se, but rather impaired vascular maturation of glomeruli. It may not be surprising that condensation of mesenchymal cells into avascular comma-shaped bodies and S-shaped glomeruli is not prevented in VEGF^120/120 mice, because these initial steps of glomerulogenesis occur in the absence of glomerular vascularization and avascular glomeruli receive oxygen via diffusion from nearby vessels. Nonetheless, the generally impaired vascularization potential of VEGF^120/120 mice may explain why these mice and their organs failed to grow to their normal size. Vascularization of the kidney occurs in an outward direction (i.e., from medulla to cortex). Defective vascularization will therefore cause a greater degree of ischemia in the more cortical regions and will consequently impair growth of the nephrogenic cortex, where new glomeruli arise from mesenchymal cell condensations. Growth and organ retardation also occur in other transgenic mice with defective postnatal vascularization (^41,42).

Loss of VEGF₁₆₄ and VEGF₁₈₈ triggered glomerulosclerosis and caused dilation of proximal tubules and loops of Henle. Although these pathologic changes are present in several renal disorders (⁴³), tubular dilation and glomerulosclerosis also develop in response to renal ischemia (⁴⁴). The vascular defects in VEGF^120/120 mice likely impaired kidney oxygenation, and ultrastructural signs of ischemia in mutant mice were indeed observed in the proximal tubules and loops of Henle—known to be most susceptible to ischemic stress (⁴⁴). Not all these pathologic changes may, however, be attributable to ischemia alone and could also be due to a deprivation of trophic or survival signals or perhaps even of other unknown activities, provided by the long VEGF isoforms. VEGF is indeed known to have such trophic effect on endothelial as well as on renal epithelial cells in vitro (^16,17). Another possible cause of the tubular dilation may be the reduced renin levels in the VEGF^120/120 mice, because mice with abnormal renin expression also suffered tubular dilation (⁴⁵).

The renal phenotype in VEGF^120/120 mice may also be at least partially due to abnormal signaling of VEGF directly on renal epithelial cells. VEGF receptors have been identified on renal tubular epithelial cells (¹⁶), whereas VEGF stimulates proliferation and survival of renal epithelial cells (^16,17) and tubulogenesis in embryonic kidneys in vitro (¹⁷). Thus, absence of the long VEGF isoforms may deprive epithelial cells from critical signals. In contrast to the abnormal tubules, podocytes appeared generally normal in VEGF^120/120 kidneys. We cannot, however, exclude that the leakiness of the glomerular filter in VEGF^120/120 mice might not have been at least partially caused by impaired podocyte function, because these cells are essential for GBM formation and, when stressed, can contribute to glomerulosclerosis (⁴⁶). Podocytes lack signaling VEGF receptor tyrosine kinases, but a recent study (⁴⁷) reported that podocytes express neuropilin-1. The physiologic relevance of its expression remains, however, untested because this receptor lacks cytoplasmic signaling domains and is itself not known to transmit intracellular signals. Podocytes might, however, respond to signals produced by EC, and the endothelial dysfunction might thus have affected podocyte-related functions in VEGF^120/120 mice.

Few transgenic studies have reported severe renal vascular defects. The endothelial and smooth muscle cell defects in the VEGF^120/120 mice significantly differ from those found in mice lacking PDGF-BB or its receptor PDGFR-β (^5–7). In the latter, the defect appears to more specifically affect mesangial cell recruitment, with no abnormalities in peritubular capillary remodeling or renal arterial branching. This periendothelial-restricted phenotype in mice with impaired PDGF signaling is consistent with the presumed role of PDGF-BB in periendothelial cell recruitment after EC assembly. These results provide additional support to the hypothesis that VEGF acts at an earlier stage during vascular development than PDGF-BB but also unveils the importance of continued VEGF signaling for maturation and branching of muscularized vessels.

This study extends our initial analyses in VEGF^120/120 mice and reveals that these defects are not restricted to the heart (¹⁴). More recent analyses revealed additional vascular defects in the retina (resulting in partial outgrowth of retinal vessels and persistence of hyaloid vessels; Stalmans et al., unpublished findings) and in bone (resulting in impaired bone vascularization and endochondral bone formation; Maes et al., unpublished findings). We cannot exclude that the renal phenotype might be influenced by the impaired hemodynamic performance or by the vascular defects in other organs in VEGF^120/120 mice. However, vascular abnormalities were already observed in kidneys and bones before birth, when no circulatory defects were detected, suggesting that the renal vascular defects were not only secondary to hemodynamic insufficiency and arguing for a critical role of the various VEGF isoforms in vascular development in these organs. The precise relative contribution of each of these possible mechanisms remains to be determined.

Collectively, the different isoforms are not only important for capillary angiogenesis, but also for arteriogenesis and the maturation of capillaries. The severe angiogenic defects in VEGF^120/120 mice illustrate the functional complementarity of the VEGF isoforms in angiogenesis and implicate that, VEGF₁₆₄, which is the most abundantly expressed isoform in developing and adult mouse kidney, may be necessary for an optimal angiogenic response during renal development.

The authors thank G. Theilmeier for helpful dicussions, Inge Cartois, Ivo Cornelissen, Maria De Mol, Kristel Deroover, Sandra Jansen, Marleen Lox, Ann Mandervield, Peggy Van Wesemael, and Sabrine Wyns (The Center for Transgene Technology and Gene Therapy, Leuven Belgium) for technical assistance, and A. Vandenhoeck for assistance with artwork. This work was supported in part by the European Community (Biomed BMH4-CT98–3380), Actie Levenslijn (#7.0019.98), and Fund for Scientific Research, Flanders, Belgium. (G012500). VM is a recipient of a Marie Curie postdoctoral fellowship from the European Community.