Vangl2, a planar cell polarity molecule, is implicated in irreversible and reversible kidney glomerular injury

Abstract Planar cell polarity (PCP) pathways control the orientation and alignment of epithelial cells within tissues. Van Gogh‐like 2 (Vangl2) is a key PCP protein that is required for the normal differentiation of kidney glomeruli and tubules. Vangl2 has also been implicated in modifying the course of acquired glomerular disease, and here, we further explored how Vangl2 impacts on glomerular pathobiology in this context. Targeted genetic deletion of Vangl2 in mouse glomerular epithelial podocytes enhanced the severity of not only irreversible accelerated nephrotoxic nephritis but also lipopolysaccharide‐induced reversible glomerular damage. In each proteinuric model, genetic deletion of Vangl2 in podocytes was associated with an increased ratio of active‐MMP9 to inactive MMP9, an enzyme involved in tissue remodelling. In addition, by interrogating microarray data from two cohorts of renal patients, we report increased VANGL2 transcript levels in the glomeruli of individuals with focal segmental glomerulosclerosis, suggesting that the molecule may also be involved in certain human glomerular diseases. These observations support the conclusion that Vangl2 modulates glomerular injury, at least in part by acting as a brake on MMP9, a potentially harmful endogenous enzyme. © 2018 The Authors. The Journal of Pathology published by John Wiley & Sons Ltd on behalf of Pathological Society of Great Britain and Ireland.

PCP is also implicated in acquired kidney disease. Mitotic orientation, a PCP-mediated process, is aberrant in kidney cystogenesis [12]. Glomerular Vangl2 transcripts increased 48 h after the initiation of kidney injury by nephrotoxic nephritis (NTN), a progressive disease model, and NTN is more severe in mice with podocyte-specific Vangl2 deletion [8].
We hypothesised that Vangl2 impacts on glomerular disease by modulating MMP. We tested this by analysing mouse models of irreversible and reversible glomerular injury, both accompanied by leakage of protein into the urine. Irreversible injury was examined in NTN mice, analogous to humans with focal segmental glomerulosclerosis (FSGS). Injection of lipopolysaccharide (LPS) in mice was used to induce reversible glomerular injury, as occurs in humans with minimal change disease (MCD). Our results support the conclusion that Vangl2 modulates glomerular injury, in part by acting as a brake on MMP9.

Transgenic mice
All procedures were approved by the UK Home Office. For specific gene deletion in glomerular podocytes, we used PodCre mice that express Cre recombinase driven by the promoter of podocin, a gene expressed in podocytes from the immature capillary loop stage of glomerular development to maturity [19]. Initially, we examined the specificity of Cre recombination by breeding PodCre + mice with R26R-EYFP mice, which have a loxP-flanked STOP sequence followed by the enhanced yellow fluorescent protein gene (EYFP) inserted into the Gt(ROSA)26Sor locus. Subsequently, to delete Vangl2 in podocytes, we crossed PodCre + mice with Vangl2 flox/flox mice [20], where loxP sites flank exon 4, with PodCre + /Vangl2 /flox/+ mice being mated to generate PodCre + /Vangl2 flox/flox mice and littermate controls, Vangl2 flox/flox without Cre. Primers to detect the PodCre, Vangl2 flox and excised exon 4 of Vangl2 (Δ band) alleles are detailed in the supplementary material, Supplementary materials and methods. Recombination of Vangl2 by Cre generates a premature stop codon that gives rise to a protein lacking the four trans-membrane domains and the C-terminal PDZ-binding domain required for the interaction of Vangl2 with other proteins [21,22]. All transgenic mouse strains were on a C57Bl/6 background for >10 generations.

Murine models of glomerular disease
To induce accelerated NTN [23], a model of irreversible and progressive glomerular damage, male PodCre + /Vangl2 flox/flox and Vangl2 flox/flox mice were pre-immunised by subcutaneous injection of sheep immunoglobulin (0.2 mg) in complete Freund's adjuvant. This was followed by intravenous administration of sheep anti-mouse glomerular basement membrane (GBM) nephrotoxic globulin (200 μl) 5 days later to induce nephritis. Glomerular injury follows, with capillary thrombosis and crescent formation [23].

Histological analysis
Kidneys were fixed in 4% paraformaldehyde, dehydrated, wax-embedded and sectioned at 5 μm. Periodic acid Schiff (PAS) staining was used to detect basement membranes and sclerosis. Glomerular morphology in 12-week-old male PodCre + /Vangl2 flox/flox and Vangl2 flox/flox mice was examined by two blinded assessors and designated as normal (little PAS-positive material and normal capillary loops) or abnormal (PAS in >50% of the tuft). At least 30 glomeruli from four separate mice in each genotype were evaluated. Results for each category were expressed as a percentage of the total glomeruli assessed. In NTN mice, thrombosis (PAS-positive areas of occluded capillary loops) was scored using a scale of 0-4 depending on the number of quadrants affected within the glomerular tuft (each tuft divided into four quadrants for scoring purposes) [23]. Fifty glomeruli were assessed per sample by a blinded assessor, and an average score was obtained for each kidney.
Renal function assessment, immunofluorescence staining, Western blotting, electron microscopy, podocyte culture and RT-qPCR Details are provided in supplementary material, Supplementary materials and methods.

Studies of human kidney tissue
We interrogated microarray data obtained from microdissected glomeruli from two independent cohorts of renal patients. Cohort I included patients with FSGS (n = 10), MCD (n = 5) and living donor (LD) healthy controls (n = 18) from the European Renal cDNA Bank [25] (see supplementary material, Table S1) where RNA had been hybridised to Affymetrix HG-U133 Plus 2.0 microarrays (Santa Clara, CA, USA) [26]. Cohort II included microarray data from the public domain [GEO database: www.ncbi.nlm.nih.gov/geo; project GSE108109; Affymetrix Human Gene 2.1 ST arrays (Santa Clara, CA, USA)]. This project includes mRNA expression data from human renal biopsies with FSGS (n = 16), MCD (n = 5) and controls (LDs) (n = 6). A single probe-based analysis tool, ChipInspector (Genomatix Software GmbH, Munich, Germany), was used for transcript annotation, total intensity normalisation, significance analysis of microarrays and transcript identification based on significantly changed probes [27]. The statistic algorithm in ChipInspector is a T-test that creates artificial background data by randomly permuting the array results. Each probe has a score on the basis of its fold-change relative to the standard deviation of repeated measurements for this probe. Probes with scores higher than a certain threshold are deemed significant. This threshold is the Delta value. The permutations of the dataset are then used to estimate the percentage of probes identified by chance at the identical Delta. Thus, a relation of significant probes to falsely discovered probes can be given for each Delta threshold. This relation is the false discovery rate (FDR), a stringency indicator. Analysis was carried out using all default settings as recommended by the software provider, with an FDR of 0% and a median false positive of 0% [27].

Statistics
Datasets [mean ± standard error of mean (SEM)] were analysed using GraphPad Prism (GraphPad Software, La Jolla, CA, USA). Differences between two groups were analysed using an unpaired t-test. When comparing more than two groups, differences were analysed using one-way analysis of variance (ANOVA) with Bonferroni's multiple comparison post hoc tests. Data affected by two variables were analysed using two-way ANOVA with Bonferroni's multiple comparison post hoc tests unless otherwise stated. Statistical significance was set at p ≤ 0.05.

Genetic downregulation of Vangl2 in podocytes worsens experimental nephritis
Thus, deletion of podocyte Vangl2 led to modest aberrations of glomerular morphology, but this did not lead to increased albuminuria or kidney excretory failure. Therefore, we proceeded to investigate possible roles for Vangl2 in experimentally induced glomerular disease. We first used a model of irreversible and progressive glomerular damage, accelerated NTN. Here, mice are pre-immunised with sheep immunoglobulin, and 5 days later, nephritis is induced by nephrotoxic globulin (see supplementary material, Figure S4A). Previous work has shown that glomerular Vangl2 transcripts increased 48 h after the initiation of NTN [8].
We also examined whether genetic deletion of podocyte Vangl2 affected immune cell infiltration because this modulates the initiation and progression of NTN [28]. We assessed numbers of F4/80 + positive macrophages [29] in glomerular tufts ( Figure 2F) and in areas surrounding glomeruli ( Figure 2G). In glomerular tufts before injury, very few F4/80 + positive macrophages were detected in either genotype, and there was no significant change following NTN injury. After induction of nephritis, F4/80 + cells around glomeruli increased in Vangl2 flox/flox (p < 0.01) and Pod-Cre + /Vangl2 flox/flox kidneys, although in the latter case, this did not reach statistical significance (p = 0.06), but there was no difference between genotypes.
We examined the expression of collagen IV, an MMP9 substrate [30] and a key GBM component [31], using immunofluorescent staining of kidney sections, at baseline and during NTN, with an antibody reactive to all collagen IV chains. Quantification was performed by assigning a score of 0 to glomeruli with staining in <50% of the tuft area and a score of 1 to glomeruli with staining in >50% of the tuft ( Figure 4I, J). There was no significant difference between Vangl2 flox/flox and PodCre + /Vangl2 flox/flox before induction of NTN. During NTN, the collagen IV score was reduced in both Vangl2 flox/flox and PodCre + /Vangl2 flox/ flox mice versus healthy controls (p < 0.01), but no difference was observed between the two genotypes ( Figure 4M). We also quantified ZO-1, a tight junction protein [32] degraded by MMP9 in cultured podocytes [33], by immunofluorescent staining using the same scoring system ( Figure 4K, L). In nephropathic mice, ZO-1 immunostaining in PodCre + /Vangl2 flox/flox kidneys was reduced (p < 0.02) to approximately half the level measured in Vangl2 flox/flox organs ( Figure 4N).

Podocyte Vangl2 deletion enhances LPS-induced glomerular injury and modulates MMP9
Next, we determined whether Vangl2 plays a role in another glomerular disease model, LPS-induced reversible glomerular injury. Here, podocytes are injured through the activation of the toll-like receptor 4, leading to FP effacement within 24-48 h, followed by resolution after 72 h [24]. One day after LPS administration, the urinary albumin/creatinine ratio was, on average, three-fold greater (p < 0.05) in PodCre + /Vangl2 flox/flox versus Vangl2 flox/flox mice ( Figure 5A). Albuminuria continued to increase in both groups until 48 h, with a non-significant (p = 0.68) tendency for higher values in PodCre + /Vangl2 flox/flox  Single-probe analysis for selected PCP transcripts in microdissected glomeruli from two independent cohorts of renal patients with focal and segmental glomerulosclerosis (FSGS); MCD and living kidney donors (LD) used as controls. Values are expressed as fold-change compared to LD. Significantly upregulated genes are shown in red and downregulated genes in blue. Transcripts with a fold-change above 1.5 or below 0.667 are displayed in bold.

Levels of PCR transcripts in human glomerular disease
To begin to examine the human relevance of this work, we assessed levels of transcripts encoded by PCP genes (VANGL1, VANGL2, CELSR1, CELSR2, DISHEVELED 1-3, FRIZZLED3, PRICKLE1 and PRICKLE2) in glomeruli from biopsies of individuals with either FSGS or MCD from two different cohorts of patients (Table 1). In cohort I, significant increases versus healthy controls were observed for all PCP transcripts examined in FSGS, including VANGL2, which was upregulated >1.5-fold. Similar findings were observed in cohort II, with significant increases found in 6 of the 10 genes evaluated, one of which was VANGL2. In contrast, in samples from MCD patients in cohort I, only VANGL1 and PRICKLE1 were significantly increased versus healthy kidneys, whereas CELSR1 levels were downregulated by 0.8-fold. There were no significant changes in any of the PCP genes examined in the MCD patients in cohort II.

Discussion
Targeted genetic downregulation of Vangl2 in podocytes enhanced the severity of both accelerated NTN and LPS-induced glomerular damage. In each proteinuric model, genetic deletion of Vangl2 in podocytes was associated with an increased ratio of active-MMP9 to inactive-MMP9. These observations support the conclusion that Vangl2 modulates glomerular injury in mice, at least in part by acting as a brake on MMP9, a potentially harmful endogenous enzyme. In addition, by interrogating data from two cohorts of renal patients, we report increased VANGL2 transcript levels in glomeruli of individuals with FSGS, providing evidence that the molecule may also be involved in certain human glomerular diseases. Previously, we [10] and others [9] showed that Loop-tail (Lp) mice with homozygous point mutations in Vangl2 had malformed kidneys containing fewer ureteric tree collecting duct branches and fewer mature glomeruli. However, the Vangl2 Lp/Lp mouse is not an ideal model to define the specific glomerular roles of Vangl2. First, the mutation would affect Vangl2 in both nephron and collecting duct lineages. Accordingly, because nephrons including glomerular and collecting duct development are interdependent, the glomerular phenotype could be a secondary effect. Second, homozygous Lp mutants die neonatally, precluding their use in testing roles for Vangl2 in glomerular function and disease in adulthood.
To circumvent this, we used a conditional Vangl2 flox/flox mouse [19] and deleted Vangl2 specifically in glomerular podocytes. In this model, we found that there were no alterations in other PCP components in the kidney at the transcriptional level but cannot rule out possible differences in their localisation. Indeed, the Lp mutation affects the localisation of certain PCP components such as Pk2 [34], Frizzled 3 [21] and Vangl1 [35]. Based on our previous observations on the Loop-tail mouse [10] and other evidence supporting a role for Vangl2 in podocyte morphology [36], we initially hypothesised that a lack of podocyte Vangl2 might result in impaired glomerular morphology and function. We found that the kidneys of 12-week-old PodCre + /Vangl2 flox/flox did contain a slight but statistically significant increased proportion of morphologically abnormal glomeruli. This is likely explained by the fact that podocin promoter-driven Cre expression, and thus Vangl2 recombination, would start in immature glomeruli, in the capillary loop stage. Kidney function, however, appeared preserved in adults as assessed by plasma creatinine and urinary albumin levels. Our results concur with Rocque and colleagues [8], who showed that podocyte-specific deletion of Vangl2 using the same PodCre line in our study led to smaller glomeruli at 2 weeks of age, but this also did not lead to any changes in albuminuria. Furthermore, genetic deletion of podocyte Scribble, encoding another PCP core protein, did not lead to any changes in glomerular morphology or function [37]. Collectively, these results suggest that the knockdown of an individual PCP component does not have a major effect on glomerular biology of otherwise healthy mice. On the other hand, our observations on Vangl2 and Celsr1 compound heterozygous mice showed a more severe foetal glomerular defect than either mouse alone [11]. Future studies on mice lacking multiple PCP components could provide more insights into the potential role of this pathway in glomerular morphogenesis.
A key finding in our study is that glomerular injury, induced by nephrotoxic serum or LPS, is aggravated in mice with genetic downregulation of podocyte Vangl2 compared with controls. Although there was no difference in albumin excretion between PodCre + /Vangl2 flox/flox and Vangl2 flox/flox mice before glomerular injury, we cannot rule out that the increased proportion of morphologically abnormal glomeruli seen in PodCre + /Vangl2 flox/flox mice makes these animals more susceptible to injury. In future, inducing NTN in mice in which Vangl2 is deleted in adulthood by an inducible PodCre allele [38] should help unravel whether the above modest glomerular maturation defect is playing a confounding role in worsening the severity of nephritis in mice with podocyte-specific Vangl2 depletion.
How might Vangl2 downregulation lead to enhanced kidney injury? Possible mechanisms include cytoskeletal rearrangements affecting cell morphology [39] or changes in inflammation [14]. However, in this study, we focused on the effect of Vangl2 downregulation on MMPs, which modulate tissue remodelling. First, we examined which MMPs were altered in cultured podocytes following Vangl2 downregulation and found increased transcript levels of Mmp9. Furthermore, in both injury models, Vangl2 mutants had increased ratios of active-MMP9 to inactive MMP9. MMP9 has previously been shown to be produced by podocytes [16,33] and altered in a number of glomerular diseases, including lupus nephritis with active, fibrocellular crescents [40], DN [33,41]; viral-associated glomerulonephritis [42], membranous [43] and hypertensive [44] nephropathy. MMP9 is also induced by activation of the toll-like receptor 4 [45,46], which mediates the actions of LPS. The exact mechanism of how PCP proteins regulate MMPs is not fully understood. One possibility is through the regulation of vesicular trafficking [13]. Alternatively, Vangl2 can regulate cell surface integrin αvβ3 expression and adhesion to fibronectin, laminin and vitronectin [47].
We subsequently examined some of the mechanisms through which increased MMP activity might aggravate glomerular disease in NTN. MMPs were originally characterised by their ability to break down ECM [48]; therefore, we examined collagen IV, a key component of the glomerular ECM [31]. However, we did not find any difference in collagen IV expression between Pod-Cre + /Vangl2 flox/flox and Vangl2 flox/flox animals with NTN. Recent studies using proteomic approaches have shown that the glomerular ECM is composed of over 140 structural and regulatory components [49], and future experiments could examine the detailed ECM proteome of nephropathic PodCre + /Vangl2 flox/flox and Vangl2 flox/flox animals. LPS injury in mice also results in remodelling of the GBM 24 h later. A glomerular microarray study found elevated levels of transcripts encoding collagen IV α1 and α2 chains alongside laminin α5β2γ1 [50], both of which normally predominate in immature glomeruli [51]; we postulate that MMP9 may play a role in this process.
In vitro, podocyte exposure to exogenous MMP9 was shown to degrade ZO-1, a tight junction protein [32]. We also examined the distribution of ZO-1 in vivo following NTN and found a significant reduction in Pod-Cre + /Vangl2 flox/flox mice. Ultrastructure assessment of mice kidneys with NTN has shown that tight junction formation is an early abnormality in NTN, preceding FP effacement and podocyte bridge formation [52]. The authors postulated that podocyte-to-podocyte tight junction function may be a compensatory mechanism to maintain glomerular filtration barrier integrity. Therefore, the loss of ZO-1 in PodCre + /Vangl2 flox/flox mice with NTN may lead to filtration barrier disruption and account for the enhanced albuminuria seen in these mice. MMP9 has also been shown to upregulate podocyte integrin-linked kinase (ILK) secretion [33], a kinase known to induce podocyte de-differentiation and detachment in disease conditions [53]; whether it is upregulated in the setting of dysfunctional PCP remains to be elucidated. Further studies inhibiting MMPs in PCP-deficient mice would help to delineate their role in this pathway. Chemical inhibition of MMP activity has already been shown to be beneficial in some models of glomerular damage [54,55], and based on our observations, we would predict a similar role in dysfunctional PCP-associated glomerular damage.
To begin to examine the relevance of our mouse studies to human disease, we examined PCP transcripts in microdissected glomeruli from FSGS and MCD patients. In data from two independent cohorts of FSGS patients, significant increases versus LDs were observed for the majority of PCP transcripts examined. Importantly, in both FSGS cohorts, VANGL2 was upregulated >1.5-fold, suggesting this molecule may have an important biological role in FSGS. It should be noted that there were some discordant results between the two cohorts analysed (e.g. in the number of PCP transcripts found to be significantly altered), and follow-up studies should confirm the microarray data by RT-qPCR and assess VANGL2 at the protein level in human glomeruli. In accord with the human data, a significant upregulation of Dvl2, Fz3, Pk1 and Vangl2 glomerular transcripts was detected in NTN mice 48 h after the induction of disease [8]. Interestingly, changes in glomerular ECM deposition are a feature of FSGS and NTN [56,57], whereas in MCD or LPS glomerular disease, where there is no sclerosis or excess ECM deposition, the majority of PCP genes examined were unaltered. Collectively, the finding of VANGL2 upregulation in FSGS, coupled with the observation that glomerular disease is worsened in mice deficient for Vangl2 in podocytes, suggests that increased PCP gene expression in glomerular disease is likely to be a protective compensatory response.