Laminin β₂ variants associated with isolated nephropathy that impact matrix regulation

The glomerular filtration barrier allows the efficient flow of water and small molecules while preventing plasma proteins from leaking into the urine. The barrier consists of podocytes, endothelial cells, and the glomerular basement membrane (GBM). Nephrotic syndrome, characterized by proteinuria, edema, hypoalbuminemia, and hyperlipidemia, is associated with impaired glomerular filtration. Nephrotic syndrome can be caused by dysfunctions of podocytes, endothelial cells, or GBM, indicating that all 3 components are essential for maintaining the glomerular filtration barrier. Extensive genetic studies of nephrotic syndrome have identified mutations in genes important for establishing and maintaining the barrier (1).

The GBM is a specialized extracellular matrix separating 2 cellular layers, endothelial cells and podocytes. The GBM plays critical roles in filtration and in maintenance of glomerular morphology (2). Basement membrane (BM) is usually composed of 4 major extracellular matrix proteins: laminin, collagen IV, nidogen, and sulfated proteoglycan. The GBM contains an atypical assortment of BM protein isoforms, including collagen α₃α₄α₅(IV) and laminin-521 (LM521; α₅β₂γ₁; ref. 3). Mutations in collagen α₃α₄α₅(IV) cause Alport syndrome, a kidney disease with variable ear and eye defects (4). LM521 is composed of α₅, β₂, and γ₁ chains. Mutations in the laminin β₂ gene (LAMB2) cause Pierson syndrome, which is characterized by congenital nephrotic syndrome with severe ocular and neuromuscular defects (5). Zenker et al. originally identified homozygous or compound heterozygous mutations in the LAMB2 gene, which contains 32 exons, in patients with Pierson syndrome (6). More than 49 mutations have since been reported to be associated with Pierson syndrome (5). Most disease-associated alleles are truncating mutations localized across the entire gene, eliminating laminin β₂ function completely (7). On the other hand, the patients with LAMB2 missense mutations occasionally exhibit higher mean age at onset of renal disease or oligosymptomatic disease variants of Pierson syndrome (5). For example, p.R246Q and p.R246W lead to congenital or infantile nephrotic syndrome with milder or no extrarenal symptoms (5, 6). A patient with Pierson syndrome carrying p.S80R presented with late-onset nephrotic syndrome that was diagnosed when the patient was 6.5 years old (5). Another missense mutation, p.C321R, is associated with congenital nephrotic syndrome, despite less severe extrarenal defects (5). p.C321R affects the formation of disulfide bonds in an EGF-like module and disrupts the structure of the LEa domain, leading to defective secretion of laminin β₂ and podocyte ER stress possibly due to misfolding of the aberrant laminin β₂ (8). Previously, Matejas et al. demonstrated that missense mutations found in Pierson syndrome are apparently clustered in the laminin N-terminal (LN) domain, which is required for the polymerization of laminins (5). Therefore, mutations in the LN domain have been implicated in the perturbation of BM formation.

Laminin is a family of heterotrimeric glycoproteins that mediate cell adhesion via various receptors. Five α, 3 β, and 3 γ chains have been characterized, and 19 isoforms have been identified in various tissues and cell culture media (9). Of the isoforms, 9 contain laminin β₂, and LM521 is the most studied of these. LM521 trimers are recognized by several receptors, including integrin α₃β₁, α₆β₁, and α₆β₄; α-dystroglycan; and Lutheran/basal cell adhesion molecule. The C-terminal 20–amino acid sequence of the laminin β₂ chain modulates the binding affinities of laminins to X2-type integrins, such as α₃β₁ and α₇X2β₁ (10). However, because the laminin globular domains of laminin α chains have been implicated in various cellular interactions, the X2-type integrins are not specific for laminin β₂. The β₂ chain of laminin-421 and -521 are also identified as ligands that bind to N- and P/Q-type voltage-gated calcium channels (VGCCs, ref. 11). Although VGCCs specifically bind to laminin β₂ at neuromuscular junction presynaptic terminals, it is unlikely that these mediate cell adhesion to the β₂ chain. So far, the cell-adhesive activity of laminin β₂ has not been fully examined.

Next-generation sequencing approaches have also identified LAMB2 missense mutations in congenital or childhood-onset nephrotic syndrome without apparent extrarenal abnormalities (12–19). Here we demonstrate that most of these missense variants found in cases with isolated nephropathy affect a region in the LEa-LF-LEb domains in the laminin β₂ short arm. Notably, variants in this region have not been found in patients with classic Pierson syndrome. We examined the effects of these variants on secretion of laminin β₂ in vitro. To explore the biochemical properties that might influence glomerular filtration, recombinant laminin β₂ chains carrying the variants were produced in mammalian cells. We found that laminin β₂ serves as a potentially novel cell-adhesive ligand for integrin α₄β₁. These results unraveled roles of the laminin β₂ short arm in matrix regulation that may underlie the pathogenesis of isolated nephropathy as a mild form of Pierson syndrome.

We previously reported a Japanese patient with isolated nephropathy carrying LAMB2 p.R469Q and p.G699R variants (12). He was born to a healthy parent and appeared normal at birth. He exhibited systemic edema by 21 months and was diagnosed with steroid-resistant nephrotic syndrome. A kidney biopsy at 23 months of age revealed focal segmental glomerulosclerosis (FSGS) and mesangial matrix expansion (Supplemental Figure 1A; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.145908DS1). As seen by electron microscopy, the GBM exhibited a stratified structure (Supplemental Figure 1B). The patient’s renal function declined and progressed to end-stage kidney disease (ESKD) at 26 months of age. After renal transplantation at 7 years of age, he is presently healthy without extrarenal symptoms. Our whole-exome analyses of the patient and his parents showed that c.1406G>A (p.R469Q) and c.2095G>C (p.G699R) in LAMB2 are derived from his father and mother, respectively (Supplemental Figure 2, A and B, and ref. 12). The former, c.1406G>A (p.R469Q), has not been reported in Genome Aggregation Database (gnomAD) (20) or TogoVar (21). Although c.1406G is adjacent to a splice acceptor site, the predictions of bioinformatics tools such as NNSplice and ASSP (22, 23) were comparable to that of WT (Supplemental Figure 2C), suggesting that c.1406G>A does not influence splicing. The allele frequency of c.2095G>C (p.G699R) is 0.01173 in East Asians (gnomAD) and 0.007 in 2 Japanese genome databases (ToMMo 4.7KJPN and GEM-J WGA; available from TogoVar, ref. 21, and Supplemental Table 1).

A Hispanic patient was found to have massive proteinuria by 20 months and diagnosed with steroid-resistant nephrotic syndrome. Sanger sequencing identified a heterozygous c.3232C>T (p.R1078C) variant in LAMB2; its allele frequency is 0.000276 (gnomAD). Additional pathogenic variants were not identified in LAMB2 or in genes associated with early-onset nephrotic syndrome in this patient, but we suspect there is an additional pathogenic variant in LAMB2 that could not be identified. In silico algorithms (SIFT, PolyPhen-2, Align-GVGD) suggested that p.R1078C is likely to be disruptive (24–26). In the ClinVar database (27) there are multiple submissions of this variant in nephrotic syndrome or Pierson syndrome but with scarce clinical or genetic information.

Including these cases, 14 cases carrying LAMB2 variants that presented with nephropathy without ocular or neurologic abnormalities have been reported in the literature (Table 1). These patients presented with abnormal urinalysis to varying degrees, ranging from mild hematuria to nephrotic range proteinuria. Among the patients, 4 of their cases progressed to ESKD in the infantile period (cases 2, 3) or in early childhood (cases 4, 5). Other cases maintained normal kidney function in childhood and even at school age. This feature is quite different from the clinical picture of typical Pierson syndrome, in which patients often progress to ESRD in the early infantile period. With respect to genetic involvement, 1 patient (case 14) carries splice site variants in LAMB2, and the other patients carry at least 1 missense variant. These missense variants are distributed in exons 2–29 (Figure 1A). This is in clear contrast to missense mutations found in typical patients with Pierson syndrome, which cluster in exons 2–9 and 25–28. Missense variants in a region spanning from exon 10 to 24, which encode the LEa-LF-LEb domains in the laminin β₂ short arm, were found in 9 patients (64.3%) with isolated nephropathy (cases 4–13), despite the mutations in that region not being reported in previously reported classical Pierson syndrome patients with ocular or neurological manifestations (Supplemental Table 2). Furthermore, none of these 9 patients progressed to ESRD in the infantile period. Therefore, we hypothesized that variants in this region may be associated with the mild phenotype caused by LAMB2 abnormalities.

Figure 1

LAMB2 variants. (A) The missense variants found in isolated nephropathy (upper arrowheads, Table 1) and Pierson syndrome (lower arrowheads, Supplemental Table 2). Vertical bars and horizontal lines represent exons and introns of the human LAMB2 gene, respectively. Numerals indicate exon numbers. p.R469Q, p.G699R, and p.R1078C are indicated by red arrows. Boxes indicate the domain structure of laminin β₂. (B) Amino acid sequence alignments of laminin β₂ derived from multiple species.

Table 1

The missense variants in isolated nephropathy

Recently, variants in LAMB2 have been found in patients with nephropathy but without apparent extrarenal manifestations. These cases also present with less severe renal phenotypes. Proteinuria is often found in childhood, and some cases maintain renal function in adolescence, which is a clear contrast to the clinical course of typical Pierson syndrome. Electron microscopic analysis of kidneys of these patients demonstrated a thin or stratified GBM forming a basket weave pattern (Supplemental Figure 1 and ref. 13). Notably, laminin β₂ in the GBM from a proteinuric patient carrying compound heterozygous truncating (p.Y76TfsTer) and missense (p.G699R) variants was detected (13), indicating that p.G699R did not cause a secretion defect. We noted that the distribution of missense variants causing Pierson syndrome and those of isolated nephropathy were different: variants in a region spanning from exon 10 to exon 24, which correspond to the LEa-LF-LEb domains in the laminin β₂ short arm, were exclusively found in patients with isolated nephropathy. Although it is known that LN domains mediate the self-polymerization of laminins, the functions of the LEa-LF-LEb domains have been unclear. Here we found that the p.R469Q, p.G699R, and p.R1078C variants affecting the LEa-LF-LEb domains found in isolated nephropathy cases did not abrogate laminin secretion but altered heparin and laminin-111 binding. These data suggest that the LEa-LF-LEb domains regulate matrix interactions, and conformational changes in this segment lead to altered GBM structure. Because these variants do not cause extrarenal defects, the altered LEa-LF-LEb domains must maintain those laminin β₂ functions required in ocular and neurological tissues. The present results suggest a novel mechanism underlying isolated nephropathy caused by LAMB2 defects, which is clinically and genetically distinct from Pierson syndrome.

The glomerular filtration barrier is permeable to water and small molecules and selectively restricts macromolecules such as albumin (34). Because proteinuria precedes podocyte abnormalities in Lamb2-knockout mice, the GBM plays a critical role in selective filtration (29). The GBM had been believed to exhibit not only size selectivity but also charge selectivity to produce urine (31). However, because the reduction of negative charge using knockout mice does not affect the filtration barrier (35, 36), the hypothesis of charge selectivity has been challenged (37). Therefore, the mechanism whereby laminin β₂ influences the size selectivity of the GBM is an important concept. In this study, WT laminin β₂ bound heparin under not only low ionic but also acidic conditions, indicating that heparin binding is pH dependent. The pH dependency of laminin β₂ binding to heparin is likely due to the basic pI of the β₂ LN domain. The basic or neutral pI of the β₂ LN domain is conserved in the mouse and human (Supplemental Figure 4) whereas that of the β₁ LN domain is acidic. This may be why laminin β₁ imperfectly compensates for the functions of β₂, despite their structural similarity (38). Our solid phase binding assays also showed that the laminin β₂ variants, but not WT β₂, bound to heparin under neutral conditions. The recombinant proteins carrying p.R469Q, p.G699R, and p.R1078C bound similarly to heparin. The binding of mutant amino acid residues to heparin was minimal under physiologic conditions.

The solid phase binding assay using truncated recombinant proteins showed that heparin binding sites are localized in the N-terminus of β₂. The variants may lead to a conformational change in the β₂ short arm, causing heparin binding sites to become exposed. The heparin binding property of laminin β₂ allows us to hypothesize a model in which the short arm of β₂ variants binds to heparan sulfate on agrin in the GBM (Figure 8A). This binding seems to disturb laminin polymerization under physiologic conditions, resulting in a large-pore GBM that can be permeable to macromolecules such as albumin. Similar to the β₂ variants, the WT β₂ short arm binds to heparan sulfate under acidic conditions. To test this model, future investigations on the effect of pH variation in mice would be required.

Figure 8

Schematic of laminin β₂ activity in the GBM. (A) The disorder of permselectivity on GBM with β₂ variants. The polymerization of laminin-521 containing WT β₂ chain forms the GBM that restricts permeability of macromolecules such as albumin (ALB) (left). The N-terminus of β₂ variants binds to heparan sulfates on agrin, when the LEa-LF-LEb variants are present (right). This binding disturbs laminin polymerization, resulting in larger GBM pores with increased permeability to albumin. (B) Dysfunction of GBM with laminin β₂ variants. The heparin-binding activity of laminin β₂ variants leads to a large-pore GBM under neutral conditions, resulting in proteinuria. Increased laminin binding also results in the accumulation of β₂ variants in the GBM. The increased cell adhesive activity would affect glomerular cells.

Once laminins are secreted into the extracellular space, they polymerize to form a meshwork via interactions among LN domains of the α, β, and γ short arms (33). The pairings of α-α, α-β, α-γ, and β-γ are possible in the presence of Ca²⁺ (39). Although the LN domain of β₂ binds to the N-termini of α₁, ₂, ₅; β₁, ₂; and γ₁, ₃, our solid phase binding assays showed that WT β₂ short arm weakly bound to laminin-111. On the other hand, the mutant β₂ short arms readily bound to laminin-111 in a calcium-independent manner. These in vitro results suggest a model in which the variants expose a cryptic laminin binding site in the β₂ short arm under physiologic conditions. Structural or biochemical analyses using samples from patients or knock-in mice carrying these variants are required to confirm the model.

Our results indicate that laminin β₂ serves as a potentially novel cell-adhesive ligand for integrin α₄β₁. Integrin α₄β₁ plays a major role in the regulation of immune cell recruitment to inflamed endothelia and other sites of inflammation. Previous studies have reported that integrin α₄β₁ is a receptor for vascular cell adhesion molecule-1 (VCAM-1) and fibronectin (40). In fibronectin, integrin α₄β₁ recognizes a motif containing the sequence EILDVPST within the alternatively spliced connecting segment-1 (41), with the LDV sequence being most critical (42). Homology search of the rat laminin β₂ amino acid sequence revealed that 2 LDVs are localized in the LN and LCC domains, respectively. The N-terminal LDV sequence is likely exposed to integrin α₄β₁ because the position is close to the signal peptide cleavage site. However, this LDV is conserved in rodents but not humans. We also searched for the LDV sequence in human laminin β₂ and found that it is localized in the LF domain. This LDV is conserved in mammals except for rodents. Pulido et al. reported the crystal structure of the laminin β₂ LF domain (43). Because the position of the LDV is located at the molecular surface, integrin α₄β₁ would be able to bind to the sequence. Our recent study showed that the β₂ short arms are polymerized toward the center of the GBM, suggesting that it is difficult for the binding site to be accessible to integrins (32). The excessive deposition of β₂ variants in the GBM may expose adhesive sites for integrin α₄β₁–positive cells (Figure 8B). In this regard, expression of integrin α₄β₁ was detected in the endothelial cells and mesangial cells of glomeruli (Supplemental Figure 5). Therefore, the exposure of the integrin binding site might strengthen cell adhesion, leading to glomerular dysfunction. Integrin α₄β₁ is a major cell surface receptor on immune cells such as lymphocytes. Moreover, Jurkat and Raji cells derived from human lymphocytes adhere to laminin β₂ via integrin α₄β₁ (data not shown). The nephritogenic immune responses are driven by CD4⁺ T cells (44). CD4⁺ T cells spontaneously adhere within uninflamed glomerular capillaries (45), and the accumulation of immune cells such as CD4⁺ T cells, neutrophils, and macrophages is often observed in inflamed glomeruli (44). In addition to VCAM-1 on glomerular endothelial cells, laminin β₂ may also serve as an adhesive ligand for inflammatory cells in nephritis.

Increased laminin deposition in the GBM has been reported in a Col4a3^–/– mouse model of Alport syndrome (46, 47). LAMB1 and LAMA2 are normally found in the mesangial matrix, but ectopic deposition is detected in Alport GBM (46). Furthermore, increased laminin α₅ mRNA was detected in Alport glomeruli (47). The increased deposition of laminin was also reported in patients with Alport syndrome (48). These increases may be due to increased production or altered turnover of extracellular matrix molecules triggered by instability caused by the collagen abnormality. We hypothesize that the mechanism of GBM laminin anomalies observed in Alport syndrome is different from that predicted here due to variations in the LEa-LF-LEb domains of LAMB2.

The variants in the LEa-LF-LEb domains found in isolated nephropathy are not common, but the allele frequencies of some variants are around 1% in specific ethnic populations. For example, the frequency of c.2095G>C (p.G699R; rs28364667) is 0.012 in the East Asian population and 0.007 in Japanese databases. Variant c.3071C>T (p.P1024L; rs368506627) is observed at 0.0098 in the South Asian population (gnomAD, ref. 20). Many Mendelian disorders, such as Pierson syndrome, are caused by highly penetrant rare variants. Recent empirical evidence has shown that variants at low frequency or rare variants are associated with complex diseases. The recessive nature of these variants and the residual function of the mutant proteins make it difficult to determine their pathogenicity. Mouse models with altered LEa-LF-LEb domains to mimic the defect or exome- or whole genome–based variant analysis in large cohorts, including several target populations, will be required to further uncover the contribution of disease-related variants in this region to the pathogenesis of proteinuric kidney disease.

Antibodies and regents. Monoclonal antibodies against human integrin α₂ (P1E6), α₃ (P1B5), α₄ (P4G9), α₅ (P1D6), α₆ (P5G10), and α_V (P3G8) (Developmental Studies Hybridoma Bank) were purified on Protein G Sepharose (Cytiva) from conditioned media of hybridoma cells purchased from Developmental Studies Hybridoma Bank. Anti–integrin α₁ (FB12), β₁ (6S6), and β₃ (25E11) antibodies were purchased from Merck. Anti–β-actin antibody conjugated with peroxidase was from Medical & Biological Laboratories Co. Ltd. Alexa Fluor 647–conjugated rat monoclonal antibody against mouse integrin α₄ (R1-2) and isotype control rat IgG2bκ (RTK4530) were purchased from BioLegend. Anti-podocin polyclonal antibody (catalog ab50339) was from Abcam. Human recombinant laminin-111 (LM111), laminin-511 (LM511), and laminin-521 (LM521) were purchased from BioLamina. Mouse EHS laminin was from BD Biosciences and Takako Sasaki (Oita University School of Medicine, Oita, Japan). Biotinylated heparin with an average mass of 12.5 kDa was from Celsus Laboratories, Inc.

Construction of expression vectors. cDNA clones encoding full-length mouse laminin β₁ (provided by Albert E. Chung, University of Pittsburgh, Pittsburgh, Pennsylvania) and rat laminin β₂ (generated in-house) were used to construct expression vectors. The fragments encoding laminin β₁ LN-LEa domains (B1N) and short arm (B1SA) were amplified using primer sets described in Supplemental Table 3 and subcloned into a human IgG₁ Fc expression vector. The EcoRI-XbaI fragment encoding the laminin β₂ LN-LEa domains (B2N) was subcloned into a human IgG₁ Fc expression vector. To construct the β₂ short arm expression vector, the fragment encoding laminin β₂ LF-LEb domains was amplified using primer sets described in Supplemental Table 3 and subcloned into the XbaI site of the B2N_Fc expression vector. The R469Q, G699R, and R1078C human laminin β₂ variants were separately engineered into analogous sites of the rat cDNA by site directed mutant PCR using primer sets described in Supplemental Table 3. Laminin β₂ carrying the C321R variant was prepared in our previous study (28).

Preparation of serum-free conditioned media and cell lysates. HEK293 cells were obtained from the American Type Culture Collection. The cells were maintained in DMEM containing 10% fetal calf serum. To express WT and mutant proteins, the expression vectors were transiently transfected into the cells in 24-well plates using Lipofectamine 2000 (Thermo Fisher Scientific). The next day, the growth media were replaced with 500 μL of serum-free medium. After 3 days’ culture, the conditioned media were harvested and clarified by centrifugation twice at 100g for 5 minutes and 9730g for 10 minutes. Then 20 μL of conditioned media were applied for immunoblotting. The transfectants were also lysed with 500 μL of 10 mM Tris-HCl (pH 7.5), 150 mM NaCl, 10 mM EDTA, 1% NP-40, and protease inhibitor cocktail (Merck). The cell lysates were clarified by centrifugation once at 9730g for 10 minutes. Then 20 μL of cell lysate was applied for immunoblotting.

Immunoblotting. Cell lysates and conditioned media were separated on 5%–20% gradient gels (ATTO) under reducing conditions and transferred to a PVDF membrane (Merck). Conditioned media and lysates were immunoblotted with anti-human IgG₁ Fc antibody (Jackson ImmunoResearch catalog 109-066-098). The bound antibodies were visualized with the ECL Western blotting detection system (Cytiva). The images of bands were captured with WSE-6100 Lumino Graph I (ATTO).

Expression and purification of recombinant proteins. HEK293 cells transfected with the expression vectors were selected using 100 μg/mL Zeocin (Thermo Fisher Scientific). All further cell culture was carried out in the presence of the antibiotic. The selected cells were grown to confluence in culture dishes with DMEM containing 10% FBS. The confluent cells were incubated in serum-free DMEM for 4 days. The conditioned media were harvested and clarified by centrifugation twice at 100g for 5 minutes and 9730g for 10 minutes. The recombinant proteins were purified from the conditioned culture media by Protein A Sepharose (Cytiva). The eluted fractions were pooled and dialyzed against PBS(-). The purified proteins were separated by SDS-PAGE using 5%–20% gels under reducing conditions. The separated proteins were stained with Coomassie Brilliant Blue (nacalai tesque).

Solid phase binding assays. Heparin binding assays were carried out with recombinant proteins coated onto high-protein-binding capacity 96-well microtiter plates (IWAKI). The plates were blocked with PBS(-) containing 1% BSA for 1 hour at room temperature. A total of 10 μg/mL of biotinylated heparin in 10 mM Tris-HCl (pH 7.5) containing 0, 150, and 500 mM NaCl or 10 mM BisTris-HCl (pH 6.1 and 7.1) containing 150 mM NaCl was added into the wells. After incubation for 1 hour at room temperature, the bound heparin was detected by addition of streptavidin-conjugated horseradish peroxidase, followed by addition of 1 mg/mL o-phenylenediamine and 0.012% H₂O₂. The absorbance was measured at 450 nm with Multiskan GO microplate spectrophotometer (Thermo Fisher Scientific). For laminin-111 binding assay, 96-well microtiter plates were coated with 10 μg/mL of EHS laminin-111. After blocking with PBS(-) containing 1% BSA, 10 μg/mL of Fc-tagged recombinant proteins in the binding buffer (10 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM CaCl₂, 1 mM MgCl₂ containing 0.05% Tween 20) were incubated at room temperature for 1 hour. The bound recombinant proteins were detected with a biotinylated anti-human IgG Fc antibody (Jackson ImmunoResearch) and quantified as described above.

Cell attachment assays. For adhesion assays, 96-well microtiter plates (IWAKI) were coated with recombinant proteins and blocked with PBS(-) containing 1% BSA. The dissociated HEK293 cells were suspended in serum-free DMEM and plated at 2 × 10⁴ cells/50 μL/well. After incubation at 37°C for 0.5 or 1 hour, the attached cells were stained with Diff-Quik (SYSMEX). The stained cells were counted using image cytometer BZ-X800 (Keyence). To identify the receptors for laminin β₂, 10 μg/mL of monoclonal antibodies against different integrins were preincubated individually with cells in a volume of 50 μL of serum-free medium (2 × 10⁴ cells/well) at room temperature for 10 minutes. The preincubated cells were transferred to plates coated with laminin β₂ LN-LEa domains fused with Fc-tag (B2N_Fc) and then incubated further at 37°C for 1 hour. After incubation, the attached cells were quantified as described above.

Staining of renal biopsy specimens. Kidney biopsy specimens were fixed in 10% neutral buffered formalin and embedded in paraffin. The tissues were sectioned and stained with periodic acid–methenamine silver stain–Masson’s trichrome and periodic acid–Schiff.

Electron microscopy. After fixation in glutaraldehyde fixative (1.6% v/v paraformaldehyde, 2.5% v/v glutaraldehyde, 0.2 M sodium cacodylate buffer pH 7.4, 5 mM CaCl₂, 10 mM MgCl₂), tissues were embedded in epon resin and sectioned. Imaging was performed using a HITACHI H-7100.

Immunohistochemistry. C57BL/6J mice (72 weeks old, male) were purchased from Charles River Laboratories Japan. Mice were sacrificed under deep anesthesia with Thiopental. Mouse kidneys were frozen in Tissue-Tek optimum cutting temperature compound (Sakura Finetek) and were cut at 7 μm in a cryostat. After blocking with 10% normal goat serum, the sections were incubated with rat monoclonal antibody against mouse integrin α₄ and rabbit polyclonal antibody against podocin. The integrin α₄ antibody conjugated with Alexa Fluor 647 was used. Rabbit IgG was detected with secondary antibody conjugated with Alexa Fluor 594 (Thermo Fisher Scientific catalog A11037). After several washes, sections were mounted in 90% glycerol containing 0.1× PBS and 1 mg/mL p-phenylenediamine. Images were captured using a BZ-X800 (Keyence).

Statistics. Comparisons between groups were analyzed using 1-way ANOVA with Tukey’s multiple-comparison test. Statistical analyses were performed in R Studio (Version 1.3.1093). Significance, set at P < 0.05, is depicted as follows in figures: *P < 0.001, **P < 0.05.

Study approval. Informed consent was obtained from all the members of a family according to the guidelines of the Ethics Committee of Yamagata University (protocol number 2018-190) and the University of New Mexico School of Medicine. Clinical data of patients, including laboratory data, imaging data, and kidney biopsy, were obtained for diagnostic purposes and with informed consent. Animal studies were approved by the Animal Research Committee of Tokyo University of Pharmacy and Life Sciences (P20-33).