Introduction

Proteinuria is the diagnostic hallmark for a cohort of renal disorders, including diabetic and chronic kidney diseases. In addition, proteinuria itself triggers progressive renal dysfunction and is an independent risk factor of end-stage renal disease.¹ Glomerular albuminuria, the most common subtype of proteinuria, results from compromised glomerular filtration barrier (GFB), leading to hyperpermeability and aberrant filtration of albumin.² The GFB consists of 3 layers: glomerular endothelium, glomerular basement membrane, and podocytes.³ Previous studies have focused on the dysfunctions induced by podocyte injury.^4,5 Notably, disruption of the glomerular endothelium causes proteinuria independently.^6–10 Indeed, abnormalities in any of the 3 layers comprising the GFB can lead to increased urinary protein.^11,12

The glomerular endothelium is a continuous cell monolayer lining the inner face of the glomerular basement membrane. Glomerular endothelial cells (GECs), which are morphologically and functionally specialized, are characterized by transcellular pores known as fenestrae.¹³ In addition, GECs are covered by a negatively charged gel coat, the glycocalyx, which protects the endothelium and is a vital component of the architecture for selective transportation.^11,14 These unique features allow rapid transport of water and small molecules and provide a barrier with size and charge selectivity to restrict the efflux of large molecules.³ Unfortunately, GECs are vulnerable to pathological stimuli. Hyperglycemia, oxidative stress, and inflammatory factors contribute to the impairment of the glomerular endothelial barrier, leading to hyperpermeability.^3,15

Although tight and gap junctions exist in the GECs, adherens junctions primarily regulate the permeability of the glomerular endothelium and are often found disrupted under pathological conditions.^2,16 Loss of adherens junctions enables macromolecules to leak from the glomerular endothelium in renal diseases.¹⁷ VE-cadherin (vascular endothelial cadherin) (encoded by the gene CDH5), is the primary transmembrane component of adherens junctions in the glomerular endothelium.¹⁸ It links adjacent cells by forming homophilic contacts via its N-terminal extracellular cadherin repeats.^19,20 VE-cadherin also interacts with components of the glycocalyx to support the structure of the endothelial surface layer.²¹

Several lines of evidence suggest that Notch signaling may directly regulate GFB function. The Notch signaling pathway is an evolutionally conserved cell-cell signaling pathway that directs the differentiation, proliferation, patterning, and fate determination of diverse cell lineages.²² In mammals, the pathway is activated when one of the 4 mammalian Notch receptors interacts with a ligand. This leads to cleavage of the Notch receptor and subsequent release of the NICD (Notch intracellular domain fragment). The NICD translocates into the nucleus and assembles a transcriptionally active complex with a Mastermind family protein and the CSL protein (CBF1, suppressor of hairless, Lag-1) to regulate target gene transcription.²³ Notch1 is expressed in the developing glomerular endothelium and epithelium,²⁴ and coordinated Notch signaling is essential for embryonic nephrogenesis. Aberrant activation or inactivation of Notch signaling induces developmental renal abnormalities.²⁵ In adult renal diseases, proteinuria and the degree of proteinuric nephropathy are positively correlated with Notch activation in podocytes.²⁶ In addition, activation of Notch signaling in endothelium has been observed in patients and animal models of diabetes or chronic kidney disease.^26–29 However, the role of endothelial Notch signaling in glomerular dysfunction has not been well addressed. A model that can specifically activate Notch signaling in the adult glomerular endothelium has, to this point, been lacking.

In this study, we combined transgenes enabling Cre/loxP-mediated DNA recombination and tetracycline-regulated gene expression. This system enabled us to avoid embryonic lethality associated with constitutive activation of Notch1 in the embryonic endothelium and to obtain adult transgenic mice. In these mice, Notch1 signaling was postnatally activated in the glomerular endothelium, resulting in a decrease of VE-cadherin expression and an increase of urinary albumin secretion. In cultured human renal GECs (HRGECs), activation of Notch1 signaling significantly inhibited VE-cadherin expression, reduced the endothelial glycocalyx and increased monolayer permeability. Mechanistically, we found that this was due to altered expression of 2 transcription factors SNAI1 (snail family transcriptional repressor 1) and ERG (Ets related gene), which jointly functioned to suppress VE-cadherin gene transcription in GECs. These results reveal a novel mode by which Notch signaling disrupts the GFB by repressing VE-cadherin expression in the glomerular endothelium, suggesting that endothelial Notch signaling is a potential target for proteinuria treatment in renal diseases.

Methods

Data Availability

The data supporting findings of this study are available from the corresponding author upon reasonable request. Please see the Major Resources Table in the Data Supplement.

Transgenic Mice

The ZEG-NICD1 transgenic mice were previously generated in our laboratory³⁰ and are available through the Jackson Laboratory (JAX Stock No. 006850, Bar Harbor, ME). The Tie2-tTA mice express the tTA (tetracycline transactivator; tet repressor fused to a VP16 activator) in endothelial cells under the regulation of a 2.1 kb Tie2 promoter element³¹ (provided by Dr Urban Deutsch, Theodor-Kocher-Institute). The Tet-O-Cre mice carry the Cre recombinase coding sequence downstream of a minimal CMV (cytomegalovirus) promoter coupled to a tetracycline operator element³² (provided by Dr Andras Nagy, Samuel Lunenfeld Research Institute). The 3 lines were backcrossed at least 9 generations onto the C57BL/6J background. Animal experiments complied with ethical standards of the Institute Animal Care Committee of Sunnybrook Health Sciences Center Research Institute. The ZEG-NICD1 mice were bred with Tie2-tTA mice, and double transgenic offspring were then crossed with Tet-O-Cre mice to generate triple transgenic mice. Breeding pairs were maintained on 0.1 mg/mL doxycycline (Sigma-Aldrich, St. Louis, MO) in drinking water with 5% sucrose. Doxycycline was withdrawn on the day the pups were born, and pups were genotyped to identify triple transgenic offspring (ZEG-NICD1/Tie2-tTA/Tet-O-Cre; Table I in the Data Supplement). Sample size was determined by power calculations based on data from previous studies with an α level of 0.05 and 80% power. Only male mice were used in this study to reduce variability and limit the number of animals needed to determine significant differences between experimental groups. Within the animal studies, all the experiments and outcome assessments were blinded through numerical coding of mice and samples.

Urine and Blood Analysis

Urine of the mice was collected for 24 hours using a metabolic cage, and blood was collected by retro-orbital bleeding. Urine albumin and serum albumin were determined by ELISA (R&D Systems, Minneapolis, MN). Urine protein was measured with a BCA protein assay kit (Thermo Fisher Scientific, Waltham, MA). Urine creatinine and serum creatinine was examined using the creatinine detection kits (Arbor Assays, Ann Arbor, MI). Serum blood urea nitrogen was determined by a urea nitrogen detection kit (Invitrogen, Carlsbad, CA).

Laser Capture Microdissection

The healthy human and mouse kidney samples were purchased from US Biomax. The tissue samples were sectioned at 10 µm on aminosilane-coated slides (HistoGene; Arcturus Engineering, Mountain View, CA). Endothelial cells and podocytes were labeled by immunofluorescent staining of CD31 and synaptopodin, respectively. Laser capture microdissection collection was performed by the Arcturus microdissection system (Arcturus Engineering). Targeted cells were collected by transferring them from tissue sections to an adhesive infrared-activated polymer on the CapSure HS laser capture microdissection Cap (Life Technologies, Carlsbad, CA). Total RNA was extracted from the pooled samples with PicoPure RNA Isolation Kit (KIT0204; Life Technologies) according to the manufacturer’s instructions.

Transmission Electron Microscopy

The transmission electron microscopy was performed by the Electron Microscopy Core Facility at Shandong University per the standard protocols.

Dual-Luciferase Reporter Gene Assay

The −1192 to +1 fragment of the VE-cadherin promoter was produced by DNA synthesis and the mutations (−373, −134/−118, −373/−134/−118) was generated by polymerase chain reaction (PCR). The promoter fragments were cloned into pGL3-basic vector (Promega, Madison, WI), and the resulting constructs were transfected into HRGECs using Lipofectamine 3000 (Thermo Fisher Scientific) following the manufacturer’s protocol. A plasmid containing the Renilla luciferase reporter gene under a CMV enhancer/promoter (Promega) was used as an internal control. After 48 hours, the cells were collected by Passive Lysis Buffer and assayed for luciferase activity using a dual-luciferase reporter kit (Promega) on a single tube luminometer (Promega).

Chromatin Immunoprecipitation Assay

HRGECs were fixed with 1% PFA for 10 minutes, washed twice with PBS containing an EDTA-free protease inhibitor mixture (Roche, Basel, Switzerland), and collected by a cell scraper. Fragmentation of genomic DNA was achieved by sonication with a Scientz Sonifier. Immunoprecipitation was performed using a chromatin immunoprecipitation assay kit (17-371; Upstate Bio-technology, Lake Placid, NY) with antibodies for ERG (ab133264, Abcam) and SNAI1 (AF3639, R&D systems) according to the manufacturer’s instructions. Relative IgG (Proteintech) was used as negative control. The target genomic DNA fragment was amplified by semi-quantitative PCR (primer sequences were listed in Table II in the Data Supplement). The PCR products were separated on 1% agarose gel and visualized under UV light.

Electrophoretic Mobility Shift Assay

Nuclear protein extracts were prepared from HRGECs using Pierce NE-PER Nuclear and Cytoplasmic Extraction Reagents (Thermo Fisher Scientific) following the manufacturer’s protocol. Electrophoretic mobility shift assay was performed using a LightShift Chemiluminescent electrophoretic mobility shift assay Kit (Thermo Fisher Scientific). The probes used were as follows: 5′- CAATAACAGGAAACCATCCCAGGGGGAAGTAAACCAG-3′ (probe 1); 5′- GGTGATGACACCTGCCTGTAGCATTCCAA-3′ (probe 2). Equal amounts of nuclear extract were incubated with the biotin-labeled double-stranded probes or control poly (dI:dC) for 20 minutes in binding reaction buffer. Antibodies for ERG (ab133264, Abcam) and SNAI1 (AF3639, R&D systems) were used for supershift assay. The DNA–protein complexes were electrophoresed through a nondenaturing 6% polyacrylamide gel and transferred onto a positively charged nylon membrane (Thermo Fisher Scientific). The membrane was then crosslinked with UV radiation and visualized using chemiluminescence reagents (Millipore).

Statistical Analysis

All values were presented as mean±SEM. Statistical analysis was performed using GraphPad Prism 7 software (GraphPad, San Diego, CA). Data normality was determined by Shapiro-Wilk test. An unpaired, 2-tailed Student t test was applied for 2 groups comparison with normal distribution, or Mann-Whitney test for nonparametric test. Wherever multiple t test was performed, P values were corrected for multiple comparisons using the Holm-Sidak method. For comparisons of multiple groups, 1- or 2-way ANOVA with Bonferroni post hoc comparisons tests was used for parametric test, and Kruskal–Wallis test with Dunn post hoc comparisons tests was applied for nonparametric test. No corrections for multiple testing were made across assays, which represents a weakness of this study. The images with the value closest to the mean value were selected as representative images. A P value <0.05 was considered statistically significant.

Results

Notch1 Is Highly Expressed in Glomerular Endothelium

First, we characterized the expression profile of Notch1 in glomeruli. Immunohistochemical staining of Notch1 in kidney tissues of healthy human, mouse (Figure XIV in the Data Supplement), monkey, and rat (Figure I in the Data Supplement) showed that Notch1 was expressed in GECs of the glomeruli. Double immunostaining of Notch1 and PECAM1 (platelet endothelial cell adhesion molecule 1, also known as CD31), an endothelial-specific marker, further illuminated a high expression level of Notch1 in GECs of human and mouse glomeruli (Figure 1A). GECs and podocytes of human and mouse glomeruli were precisely collected by laser capture microdissection and Notch1 gene expression was examined by quantitative real-time PCR. Figure 1B showed that Notch1 gene expression was significantly higher in GECs than that in podocytes.

• Download figure
• Download PowerPoint

Establishment of an Endothelial-Specific NICD1 Transgenic Adult Mouse Model

Previous studies of endothelial Notch activation in renal function have been limited since global or endothelial cell-specific Notch1 activation induces early embryonic lethality at E9.5 to 10.5.^30,33,34 To circumvent this problem and analyze the phenotype of Notch1 activation in adult endothelium, we previously generated ZEG-NICD1 transgenic mice which utilizes a loxP-flanked β-gal-polyA sequence between a CMV promoter and the coding sequence for Myc-tagged NICD1 with an internal ribosomal entry site (IRES) linked enhanced green fluorescence protein (EGFP).³⁰ The ZEG-NICD1 mice were mated with Tie2-tTA and Tet-O-Cre mice, and the pregnant female mice were maintained on tetracycline analog doxycycline throughout pregnancy. In this way, the Tie2-tTA transgene provides tTA expression in endothelial cells but the tTA is disabled by doxycycline. We obtained live-born triple transgenic ZEG-NICD1/Tie2-tTA/Tet-O-Cre newborns, and these mice continued to survive after removal of doxycycline, which enables tTA to bind to the Tet operator and activate the expression Cre recombinase. Cre excises the β-gal-polyA sequence between the loxP sites, resulting in CMV-driven expression of NICD1 and EGFP (Figure 1C and Figure II in the Data Supplement). To detect the expression of NICD1 in these adult triple transgenic mice, kidney tissues were sectioned and observed for GFP fluorescence. As shown by Figure 1D, GFP-positive cells were observed in glomerular endothelium of ZEG-NICD1/Tie2-tTA/Tet-O-Cre mice. Immunohistochemical staining for the Myc-tag of the NICD1 protein showed positive staining in GECs from ZEG-NICD1/Tie2-tTA/Tet-O-Cre mice (Figure 1E). A Western blot analysis also detected Myc-tagged NICD1 in kidney tissues of ZEG-NICD1/Tie2-tTA/Tet-O-Cre mice (Figure 1F).

Notch1 Activation Elicits Severe Albuminuria in ZEG-NICD1/Tie2-tTA/Tet-O-Cre Mice

The adult ZEG-NICD1 and ZEG-NICD1/Tie2-tTA/Tet-O-Cre mice did not display any obvious abnormalities. No significant difference was found in the white blood cell count, neutrophils count, and blood glucose levels of the 2 groups of mice (Figure IXA through IXC in the Data Supplement), excluding an inflammatory or diabetic phenotype for these mice. To assess the effect of endothelial Notch1 activation on renal filtration function, we examined the concentration of serum creatinine and serum blood urea nitrogen, metabolites that are mainly excreted through glomeruli. As shown in Figure 2A and 2B, the concentrations of serum creatinine and BUN did not exhibit significant alteration between ZEG-NICD1 and ZEG-NICD1/Tie2-tTA/Tet-O-Cre mice. Consistently, glomerular filtration rate was not affected in ZEG-NICD1/Tie2-tTA/Tet-O-Cre mice (Figure XXID in the Data Supplement), indicating that NICD1 transgene expression in the endothelium does not lead to impairment of glomerular filtration. In addition, no significant difference was found in the number of glomeruli, glomerular volume, and urine osmolality of the 2 groups of mice (Figure XXIA through XXIC in the Data Supplement). The ZEG-NICD1/Tie2-tTA/Tet-O-Cre mice exhibited body weight loss (Figure 2C), while measurements of systolic blood pressure showed no significant difference between 2 groups of mice, excluding the impairment of hypertension on renal function (Figure 2D). We observed significantly decreased albumin concentration in serum (Figure 2E) and correspondingly increased daily protein excretion in urine of the ZEG-NICD1/Tie2-tTA/Tet-O-Cre mice (Figure 2F). Consistently, the ZEG-NICD1/Tie2-tTA/Tet-O-Cre mice exhibited featured albuminuria, indicated by ≈4.12-fold increase of albumin in urine (Figure 2G). This was further supported by the dramatically increased urinary albumin/creatinine ratio (Figure 2H). These results showed that aberrant Notch1 activation in endothelium causes increased albumin excretion, suggesting that Notch signaling regulates the barrier function of glomerular endothelium.

• Download figure
• Download PowerPoint

VE-Cadherin Expression Is Down-Regulated by Activation of Notch1 Signaling in GECs In Vivo and In Vitro

VE-cadherin is the primary component of endothelial adherens junction and indispensable for glomerular endothelial filtration barrier.³⁵ To examine VE-cadherin expression, the kidneys of the transgenic mice were isolated, homogenized, and subjected to Western blotting or quantitative real-time PCR examination. The results showed that the mRNA and protein levels of VE-cadherin were significantly decreased in the ZEG-NICD1/Tie2-tTA/Tet-O-Cre mice (Figure 3A and 3B). Fluorescent staining of VE-cadherin also exhibited a weaker fluorescent intensity in the glomeruli of the ZEG-NICD1/Tie2-tTA/Tet-O-Cre mice (Figure 3C). To further address these findings in vitro, we infected HRGECs with lentiviral particles comprising a C-terminal flag-tagged NICD1 coding sequence. Quantitative real-time PCR and Western blotting showed elevated expression of the conventional Notch downstream target HES1 (Hes family bHLH transcription factor 1) (Figure 3D and 3E), demonstrating that Notch signaling was adequately activated. Successful infection was also validated by detection of Flag-tagged NICD1 (Figure 3E). In addition, the mRNA and protein levels of VE-cadherin were decreased in HRGECs infected with NICD1 lentiviral particles (Figure 3D and 3E). Immunofluorescent staining showed a decreased VE-cadherin expression in the membrane of HRGECs with NICD1 infection (Figure 3F). The permeability of the HRGEC monolayer was examined by FITC (fluorescein isothiocyanate)-Dextran tracer of 70 kDa, a comparable average molecular weight of albumin. The results exhibited that NICD1 infection significantly enhanced the passage of FITC-Dextran tracer through the HRGEC monolayer (Figure 3G). This was further supported by trans-endothelial electrical resistance measurement, which indicated a comparatively lower value in NICD1-infected HRGECs (Figure 3H and 3I).

• Download figure
• Download PowerPoint

DLL4 (delta-like ligand 4), a Notch ligand predominantly expressed in the endothelium, is essential for activation of Notch signaling.³⁶ To examine the role of Notch activation in HRGECs under physiological conditions, we cultured HRGECs on DLL4 coated dishes. Quantitative real-time PCR and Western blot showed that both the mRNA and protein levels of Hes1, were significantly increased (Figure 4A and 4B), indicating the activation of Notch signaling. Consistent with the results in NICD1-infected HRGECs, DLL4 coating dramatically down-regulated the mRNA and protein expression level of VE-cadherin (Figure 4A and 4B). Immunofluorescent staining showed that the accumulation of VE-cadherin on cell borders was significantly decreased in DLL4-treated HRGECs (Figure 4C). Functional studies demonstrated that the passage of 70 kDa FITC-Dextran tracer was markedly increased (Figure 4D), and trans-endothelial electrical resistance was decreased in DLL4-treated HRGECs (Figure 4E and 4F). However, treatment with dibenzazepine, a γ-secretase inhibitor which prevents the cleavage of Notch receptor, did not significantly alter VE-cadherin expression or HRGEC permeability (Figure III in the Data Supplement). Collectively, both the in vivo and in vitro studies showed that Notch activation decreases VE-cadherin expression and induces hyperpermeability in GECs.

• Download figure
• Download PowerPoint

Notch1 Activation Impairs Glomerular Endothelial Glycocalyx

To examine whether the transcellular filtration function of GECs was impaired by constitutive activation of Notch signaling in ZEG-NICD1/Tie2-tTA/Tet-O-Cre mice, transmission electronic microscopic imaging was performed to visualize the glomerular endothelial glycocalyx. As shown in Figure 5A, glomerular endothelial glycocalyx of kidney tissues of ZEG-NICD11/Tie2-tTA/Tet-O-Cre mice was distinguished from that of ZEG-NICD1 mice with lower thickness and decreased attachment area on the endothelial surface. Nevertheless, the thickness of the glomerular basement membrane and the morphology of foot processes barely exhibited any abnormalities in ZEGNICD1/Tie2-tTA/Tet-O-Cre mice compared with those in ZEG-NICD1 mice. To confirm this result in vitro, we seeded HRGECs into DLL4-coated plates. Fluorescent staining showed that DLL4 treatment remarkably reduced the fluorescent intensity of wheat germ agglutinin, a lectin that selectively binds to N-acetylglucosamine in the glycocalyx, and syndecan-4, a key component of the glycocalyx, indicating that the glomerular endothelial glycocalyx was impaired by DLL4 stimulation (Figure 5B and Figure IV in the Data Supplement). Next, we investigated whether VE-cadherin is involved in the maintenance of glomerular endothelial glycocalyx. Our results showed that knockdown of VE-cadherin resulted in decreased fluorescent intensity of wheat germ agglutinin and syndecan-4, consistent with the findings in DLL-4 treated HRGECs (Figure 5C and Figure IV in the Data Supplement). These results suggest that VE-cadherin is required for stability of the glycocalyx of GECs.

• Download figure
• Download PowerPoint

VE-Cadherin Mediates Notch1 Activation-Induced Hyperpermeability in GECs

To evaluate the role of VE-cadherin in Notch activation-induced permeability, we knocked down (Figure 6A) or overexpressed (Figure 6F) VE-cadherin in NICD1-infected or DLL4-treated HRGECs. NICD1 infection increased the passage of FITC-dextran (Figure 6B and 6D) and decreased trans-endothelial electrical resistance value (Figure 6C and 6E), and knockdown of VE-cadherin dramatically enhanced these effects (Figure 6B through 6E). Besides, Notch activation failed to induce additional permeability in HRGECs with VE-cadherin knockdown (Figure 6B through 6E). Overexpression of VE-cadherin resulted in markedly reduced FITC-dextran passage (Figure 6G and 6I) and increased the trans-endothelial electrical resistance value (Figure 6H and 6J) of the HRGEC monolayer. Upon overexpression of VE-cadherin, NICD1 infection or DLL4 treatment did not increase permeability of the HRGEC monolayer (Figure 6G through 6J). These results suggest that VE-cadherin is the downstream target of Notch signaling in the regulation of glomerular endothelial permeability.

• Download figure
• Download PowerPoint

Notch1 Activation Decreases VE-Cadherin Expression Through Upregulation of SNAI1 and Downregulation of ERG Transcription Factor

SNAI1 acts as one of the transcription repressors of the VE-cadherin gene during embryonic development and in pathological circumstances.³⁷ In HRGECs, we found that NICD1 infection significantly increased the mRNA and protein expression of SNAI1 (Figure 7A and 7B). Knockdown of SNAI1 by RNA interference remarkably increased mRNA and protein levels of VE-cadherin in the HRGECs (Figure 7C and 7D). The chromatin immunoprecipitation assay using HRGECs showed that SNAI1 bound to the −373 E-box element of the VE-cadherin promoter, which has been identified in human dermal microvascular endothelial cells,³⁸ and NICD1 infection further significantly enhanced this binding effect (Figure 7E and 7F), suggesting that SNAI1 mediates NICD1-repressed VE-cadherin expression in GECs. Nevertheless, knockdown of SNAI1 did not completely rescue Notch-induced VE-cadherin reduction (Figure 7G and 7H and Figure V in the Data Supplement), implying that additional regulators are involved.

• Download figure
• Download PowerPoint

ERG is an endothelial-specific transcription factor that regulates endothelial differentiation and vascular network formation.^39,40 In human umbilical vein endothelial cells, ERG activates VE-cadherin expression through the −134/−118 ETS (E26 transformation-specific) motif.^41,42 This sequence is evolutionarily conserved in mammals (Figure VI in the Data Supplement). We found that mRNA and protein levels of ERG was dramatically reduced in HRGECs with NICD1 infection (Figure 7I and 7J), indicating that ERG is a downstream target of Notch signaling. ERG was overexpressed by transfection of an expressing plasmid and overexpression of ERG significantly increased VE-cadherin mRNA and protein levels (Figure 7K and 7L). The chromatin immunoprecipitation assay demonstrated that ERG bound to the −134/−118 ETS binding elements, and the binding affinity was reduced by NICD1 infection in HRGECs (Figure 7M and 7N). Consistent with the result of SNAI1 knockdown, ERG overexpression did not completely abolish NICD1-induced decrease of VE-cadherin (Figure 7O and 7P and Figure VII in the Data Supplement), suggesting that both SNAI1 and ERG might act as downstream executors of Notch signaling in the downregulation of VE-cadherin expression in GECs.

SNAI1 Coordinates With ERG to Regulate the VE-Cadherin Promoter Activity

To dissect the molecular mechanisms underlying Notch-repressed VE-cadherin transcription, we scrutinized the binding of SNAI1 and ERG to the promoter of the VE-cadherin gene. On the electrophoretic mobility shift assay gel, the probe with the −373 E-box element or the −134/−118 ETS elements was shifted by a nuclear extract of HRGECs and supershifted by the nuclear extract together with SNAI1 or ERG antibodies, respectively (Figure 8A). In addition, we generated luciferase reporter gene constructs of the VE-cadherin core promoter fragment (−1192 to +1, WT) and the correspondence fragments with mutations of −373 E-box and −134/−118 ETS binding elements (Figure 8B). The dual-luciferase reporter gene assay showed that mutation of the −373 E-box did not significantly change the VE-cadherin promoter activity (Figure 8C, second column), whereas mutation of the −134/−118 ETS elements decreased the VE-cadherin promoter activity (Figure 8C, third column), indicating the differential effects of the −373 E-box and −134/−118 ETS elements on VE-cadherin transcription. Combination mutation of −373 E-box and −134/−118 ETS elements also decreased VE-cadherin promoter activity, which was still higher that of the −134/−118 mutant (Figure 8C, fourth column). In addition, NICD1 infection reduced the transcriptional activity of the wildtype VE-cadherin promoter (Figure 8C, first set) as well as the promoter with individual mutation of −373 E-box or −134/−118 ETS element (Figure 8C, second and third set). Double mutation of these elements completely abolished the transcriptional repression of NICD1 infection on the VE-cadherin promoter (Figure 8D, fourth set). SNAI1 siRNA interference up-regulated promoter activities of WT and −134/−118 mutant, but not the −373 or −373/−134/−118 mutant (Figure 8E). Analogically, ERG overexpression increased promoter activities of WT and −373 mutant, but not the −134/−118 or −373/−134/−118 mutant (Figure 8F). Taken together, Notch1 signaling works with SNAI1 to enhance VE-cadherin repression while simultaneously impairing ERG-mediated VE-cadherin transcription.

• Download figure
• Download PowerPoint

Discussion

Previous studies revealed a pivotal role for the renal glomerular endothelium in the pathogenesis of proteinuria independent of podocytes.^2,9,10 The integrity of glomerular endothelium is essential for the filtration selectivity of solutes of different molecular weight or charge.^3,14 Loss of VE-cadherin is a hallmark for disruption of the integrity of endothelial cell barrier.^23,43 In this study, we found that endothelial-specific activation of Notch1 signaling in adult mice leads to an increase in albumin excretion. Both in vivo and in vitro studies showed that Notch1 activation induces VE-cadherin downregulation, which consequently increases pericellular and transcellular permeability. Simultaneously, Notch1 activation lead to enhanced expression of the transcriptional repressor SNAI1, alongside decreased expression of the transcriptional activator ERG. SNAI1 and ERG work in concert to mediate Notch1 activation-induced VE-cadherin downregulation through the −373 E-box and −134/−118 ETS binding elements on the VE-cadherin promoter, respectively. Our study revealed a novel role of endothelial Notch signaling in the pathogenesis of proteinuria.

Notch signaling is critical for the formation and maturation of renal glomeruli during embryonic development.²⁵ In the adult kidney, Notch signaling is activated in series of renal glomerular diseases that develop proteinuria.^26,44 The level of proteinuria and the degree of proteinuric nephropathies are positively correlated with the expression level of Notch1 ligands or the activated NICD.²⁶ Previous studies on the role of Notch signaling in the development of proteinuria mainly focused on podocytes, the outermost layer of the GFB.^45,46 Overexpression of NICD1 in podocytes in vivo causes severe proteinuria, as well as podocyte apoptosis, foot process effacement and nephrin endocytosis.^5,47,48 Inhibition of Notch signaling by the γ-secretase inhibitor or depletion of CSL significantly ameliorates NICD1-induced proteinuria,⁴⁸ and genetic deletion of Notch1 receptor protects podocytes from diabetic nephropathy.⁴⁹ Endothelial cell injury and dysfunction commonly occurs in the retina, heart, and kidney of patients with diabetes.⁵⁰ These disorders are reported to be closely correlated with activated Notch signaling.^51,52 Endothelial Notch activation was also observed in patients and animal models of chronic kidney disease.^27,28 However, the functional roles of glomerular endothelial Notch signaling have not been fully investigated. Animal models for activation of Notch signaling in adult vasculature have been limited due to the early lethality caused by Notch activation in embryonic vascular cells.³³ In this study, we established a transgenic mouse model in which Notch signaling is postnatally activated in endothelial cells, and severe proteinuria was observed without any signs of inflammation, diabetes, or hypertension. Although Notch activation in liver sinusoidal endothelial cells may affect hepatocyte homeostasis⁵³ and possibly contribute to decreased plasma albumin, it does not result in proteinuria. Our findings addressed the significance of glomerular endothelium in proteinuria and confirmed the pathological roles of endothelial Notch signaling in this renal disorder.

The permeability of glomerular endothelium is highly selective to solute molecular size. It is hyperpermeable to water and small molecules but obstructive to plasma proteins.² This function is achieved through intercellular junctions, fenestrae and the loosely attached endothelial surface layer, which affects glomerular paracellular and transcellular permeability,⁵⁴ respectively. Dysfunctions of both transcellular and paracellular filtration have been observed in glomerular proteinuria.⁵⁵ VE-cadherin is the central component of the paracellular filtration barrier that associates with β-catenin, γ-catenin, and p120 to assemble adherens junctions.⁵⁶ We demonstrated here that aberrant activation of Notch signaling induces VE-cadherin downregulation in GECs, which directly disrupts the assembly of adherens junctions and increases the paracellular permeability of albumin. The glomerular endothelial surface layer, composing negatively charged glycocalyx and glycocalyx-absorbed plasma components on the luminal surface of endothelium, is the major structure of transcellular filtration barrier for albumin excretion through fenestrae with size and charge selectivity.³ Our results demonstrate that aberrant Notch1 activation reduces the glomerular endothelial glycocalyx. This could be recapitulated by gene silencing of VE-cadherin, suggesting a mechanistic link between the Notch pathway, VE-cadherin function and loss of the glomerular endothelial glycocalyx, a common event in diabetic nephropathy and other glomerular diseases. Several studies have provided evidence that VE-cadherin interacts with components of the endothelial glycocalyx, such as fibrin,^57,58 which help maintain the stability of the endothelial surface layer. As a competitive binding partner for β-catenin, decreased VE-cadherin level releases β-catenin to activate Wnt signaling,^59,60 which down-regulates several components of glycocalyx, including syndecans,⁶¹ glypican-3,^62,63 and CD44 (hyaluronan).⁶⁴ Collectively, aberrant activation of Notch signaling impaired both the paracellular and the transcellular filtration barrier of the glomerular endothelium by inhibition of VE-cadherin expression.

A recent study by Polacheck et al⁶⁵ uncovered a noncanonical Notch signaling pathway that is involved in regulating endothelial adherens junction integrity. Mechanistically, shear stress is able to induce Notch1 cleavage and releases the Notch1 TMD (transmembrane domain), which then catalyzes the formation of a VE-cadherin-associated structural complex to drive adherens junction assembly through formation of cortical actin bundles in human dermal microvascular endothelial cells.⁶⁵ In addition, Mack et al⁶⁶ have reported that shear stress may adjust Notch signaling to regulate intracellular calcium signaling to ensure the homeostasis of cell-cell junctions of aortic endothelium. Both studies demonstrated that inhibition of Notch signaling increases endothelial cell permeability in response to shear stress. However, Polacheck et al also showed Notch inhibition in human dermal microvascular endothelial cells does not affect VE-cadherin expression, which is consistent with several studies^67–69 and our findings with GECs. Therefore, loss of Notch signaling may impair endothelial barrier integrity through modulation of Notch TMD-mediated mechanosensory complexes or intracellular calcium, but not by regulation of the basal expression of level of VE-cadherin. Our results show that VE-cadherin is repressed with Notch activation in GECs, leading to hyperpermeability of the glomerular endothelium. Each mechanism described may function to increase permeability in specific contexts, suggesting an appropriate level of Notch signaling is needed for endothelial barrier function. In addition, endothelial cells manifest remarkable heterogeneity in structure, morphology, metabolism, and function, depending on time, space, and physiological or pathological conditions.⁷⁰ As described above, the glomerular endothelium comprises a subset of highly specified endothelial cells featured by transcellular fenestrae and glycocalyx.⁷¹ Underlying the phenotypic specificity of GECs is the discrete molecular basis in gene and protein regulation.^3,13,72 VE-cadherin in particular appears to be regulated distinctly in GECs. Following IL (interleukin)-1β treatment, the expression of VE-cadherin was increased in the HRGECs but decreased in the human umbilical vein endothelial cells.³⁵ Aberrant activation of Notch signaling induces hyperpermeability in retinal and cerebral disorders,^25,67,73,74 and inhibition of Notch signaling is demonstrated to be beneficial. Thus, Notch signaling may regulate endothelial barrier function in a vascular bed-specific manner.

SNAI1 is involved in the regulation of VE-cadherin expression during embryonic heart development and tumor angioinvasion.^37,38 In addition, SNAI1 is a direct downstream target of Notch signaling.⁷⁵ Nevertheless, the regulatory mechanism of Notch-SNAI1-VE-cadherin has not been examined in GECs. In this study, we validated that SNAI1 acts an intermediate effector between glomerular endothelial Notch signaling and VE-cadherin expression. ERG belongs to the ETS transcription factor family and regulates the expression of a number of endothelial cell-specific genes to maintain vascular homeostasis.⁴⁰ Deletion of Erg gene in endothelial cells causes embryonic lethality with severe vascular defects and abnormal cardiac morphogenesis.³⁹ ERG is also required for the maintenance of adherens junction integrity.⁴² Here, we demonstrated that ERG is a novel downstream target of Notch signaling. Our study demonstrated that both the transcription activator and inhibitor mediate Notch-induced VE-cadherin downregulation.

In summary, we characterized a novel role of Notch signaling in the disruption of the glomerular endothelial barrier by downregulation of VE-cadherin through the transcription factors SNAI1 and ERG. Our findings might help understand the molecular mechanisms underlying glomerular proteinuria and provide novel therapeutic targets for this disorder.

Acknowledgments

We thank Dr Urban Deutsch and Dr Andras Nagy for providing the transgenic mice and scientific discussion.

Source of Funding

This study was supported by the National Nature Science Foundation of China (91939110, 81570255, and 81873473), Academic Promotion Program of Shandong First Medical University (2019QL014) and Shandong Taishan Scholarship (J. Liu).

Supplemental Materials

Expanded Materials & Methods

Online Figures I–XXII

Online Tables I–V

References76–88

Footnotes