Structural Basis of Highly Specific Interaction Between Nephrin And

BASIC RESEARCH www.jasn.org

Structural Basis of Highly Speciﬁc Interaction between Nephrin and MAGI1 in Slit Diaphragm Assembly and Signaling

Zhuangfeng Weng,1,2 Yuan Shang,3 Zeyang Ji,3 Fei Ye,3,4 Lin Lin,1 Rongguang Zhang,1,2 and Jinwei Zhu1

Due to the number of contributing authors, the afﬁliations are listed at the end of this article.

ABSTRACT Background The slit diaphragm is a specialized adhesion junction between opposing podocytes, establishing the final filtration barrier that prevents passage of proteins from the capillary lumen into the urinary space. Nephrin, the key structural and signaling adhesion molecule expressed in the slit diaphragm, contains an evolutionally conserved, atypical PDZ-binding motif (PBM) reported to bind to a variety of proteins in the slit diaphragm. Several mutations in NPHS1 (the gene encoding nephrin) that result in nephrin lacking an intact PBM are associated with glomerular diseases. However, the molecular basis of nephrin-PBM–mediated protein complexes is still unclear. Methods Using a combination of biochemic, biophysic, and cell biologic approaches, we systematically investigated the interactions between nephrin-PBM and PDZ domain–containing proteins in the slit diaphragm. Results We found that nephrin-PBM specifically binds to one member of the membrane-associated guanylate kinase family of scaffolding proteins, MAGI1, but not to another, MAGI2. The complex structure of MAGI1-PDZ3/nephrin-PBM reveals that the Gly at the 23 position of nephrin-PBM is the determining feature for MAGI1-PDZ3 recognition, which sharply contrasts with the typical PDZ/PBM binding mode. A single gain-of-function mutation within MAGI2 enabled nephrin-PBM binding. In addition, using our structural analysis, we developed a highly efficient inhibitory peptide capable of specifically blocking the nephrin/ MAGI1 interaction. Conclusions MAGI1 interacts with nephrin-PBM with exquisite specificity. A newly developed, potent inhibitory peptide that blocks this interaction may be useful for future functional investigations in vivo. Our findings also provide possible explanations for the diseases caused by NPHS1 mutations.

J Am Soc Nephrol 29: ccc–ccc, 2018. doi: https://doi.org/10.1681/ASN.2017121275

Slit diaphragm is a highly specialized cell-cell junction formed by the neighboring podocytes and functions Received December 11, 2017. Accepted June 19, 2018. as the final filtration barrier that prevents passage of macromolecules from the capillary lumen into the Published online ahead of print. Publication date available at www.jasn.org. urinary space.1,2 Disruption of the architecture of slit diaphragm can lead to severe nephrotic syn- Present address: Dr. Yuan Shang, Center for Biomedical In- formatics and Biostatistics, University of Arizona, Tucson, drome, with symptoms including massive protein- Arizona. uria, hypoalbuminemia, and edema.3,4 fi Correspondence: Dr. Rongguang Zhang or Dr. Jinwei Zhu, NPHS1,encodingthede ning adhesion molecule Shanghai Institute of Biochemistry and Cell Biology, Chinese inNephrinintheslitdiaphragm,wasidentified as the Academy of Sciences, 333 Haike Road, Shanghai 201203, China. major causative gene for congenital nephrotic syn- E-mail: [email protected] or [email protected] drome of the Finnish type (CNF).5 To date, over 160 Copyright © 2018 by the American Society of Nephrology

J Am Soc Nephrol 29: ccc–ccc, 2018 ISSN : 1046-6673/2909-ccc 1 BASIC RESEARCH www.jasn.org mutations in NPHS1 have been discovered in the patients with Significance Statement CNF.6 NPHS1 knockout mice showed defective assembly of the slit diaphragm and impaired barrier function of podocytes.7 Although nephrin plays crucial roles in formation of the podocyte slit Nephrin consists of extracellular eight Ig-like domains, a fibro- diaphragmand in dynamic regulationof signaling pathways intheslit nectin type III-like domain, and an unstructured intracellular diaphragm, the mechanisms at molecular level are poorly un- derstood. In addition, several mutations in NPHS1 (the gene en- domain (Figure 1A). Nephrins form the unique zipper-like struc- coding nephrin) that affect nephrin’s atypical PDZ-binding motif ture via the homotypic interactions between the Ig domains from (PBM) have been associated with glomerular diseases. Using a va- the adjacent podocytes.8 The intracellular domain of Nephrin riety of approaches, the authors demonstrate that nephrin’sPBM (Nephrin-CT) contains an evolutionarily conserved atypical specifically interacts with the PDZ3 domain of MAGI1, a member of fi PDZ domain-binding motif (PBM; “2LPFELRGHLV”)atthe the membrane-associated guanylate kinase family. The ndings provide a molecular basis of nephrin/MAGI1 interaction in slit di- C-terminus, not consistent with the canonical type I, type II, or aphragm assembly and signaling, as well as possible insights at the type III PBMs (Figure 1A).9 Mutations of NPHS1 lacking partial molecular level for diseases caused by alterations in NPHS1. intracellular domain including the C-terminal PBM lead to congenital nephrotic syndrome.6 Specifically, a truncating mutation of Nephrin (p.L1240fs12863) that lacks the very C-terminal Va- insights into Nephrin-mediated protein complex in slit line of PBM and instead bears an additional 45 amino acids was diaphragm assembly and signaling. The potent inhibitory peptide identified in patients with steroid-resistant nephrotic syndrome could serve as a useful manipulating tool to dissect differential (SRNS)(Figure1A,SupplementalFigure1).10 These genetic data rolesofMAGIisoformsinfutureinvestigations. implied that Nephrin, in addition to playing key roles in the assembly of the slit diaphragm, also serves as an intracellular signaling hub to orchestrate the dynamic regulation of the slit METHODS diaphragm. There is much evidence indicating that Nephrin forms Protein Expression and Purification multicomponent complexes with a variety of PDZ domain– The coding sequences of MAGI1-PDZ3 (residues 820–920) and containing proteins in the slit diaphragm, including membrane- Nephrin-PBM (residues 1247–1256) were amplified from associated guanylate kinase family proteins (e.g.,MAGI1/2,ZO-1, mouse brain complementary DNA. For the isothermal titration and CASK),11,12 the Par6/Par3 complex,13 scribble,14 etc. Nephrin- calorimetry (ITC) assay, wild-type or various mutants of PBM–mediated protein complexes play essential roles in mainte- MAGI1-PDZ3 and Nephrin-PBM were cloned into a modified nance of glomerular filtration barrier integrity.11 However, little version of pET15b vector and pET32a vector, respectively. His6- 2+ is known about whether the atypical PBM of Nephrin directly tagged proteins were purified with Ni -NTA agarose affinity interacts with these PDZ-containing proteins. If it does, how is it chromatography, followed by size-exclusion chromatography specifically recognized by its target(s)? We believe that a better (SEC). For the glutathione-S-transferase (GST) pull-down assay, understanding of the molecular basis underlying the Nephrin- MAGI1-PDZ3 or Nephrin-PBM was fused to the C-terminus of PBM–mediated complex formation can provide valuable insights GSTusing the pGEX-4T-1 vector and purified by GSH-Sepharose into the physiologic roles of Nephrin in the assembly and dynamic affinity chromatography, followed by another round of SEC. All of regulation of the slit diaphragm, as well as the pathogenesis of the constructs were expressed in Escherichia coli BL21 (DE3) host nephrotic syndrome caused by alterations of NPHS1. cells at 16°C for 16–18 hours. In this study, we systematically studied the interactions between For the reconstitution of the MAGI1-PDZ3/Nephrin-PBM fi fi Nephrin-PBM and PDZ domain–containing proteins in complex, His6-MAGI1-PDZ3 was rstly puri ed as described slit diaphragm and found that only MAGI1 bound directly to above, and the His6-tag was then removed by incubation with Nephrin-PBM. We further discovered that MAGI1-PDZ3 binds human rhinovirus 3C protease overnight, followed by another to Nephrin-PBM with high specificity and affinity. The precise round of SEC. The commercial synthetic Nephrin-PBM peptide molecular basis underlying this interaction was elucidated by was mixed with MAGI1-PDZ3 in a molar excess of 3:1 (approx- solving the complex structure. Unexpectedly, the atypical PBM imately 15 mg/ml total complex protein in PBS buffer; pH 7.4). of Nephrin binds to MAGI1-PDZ3 with a mode distinct from the All of the peptides used in this study were commercial syn- canonical PDZ/PBM interactions. In addition to the canonical thesized by ChinaPeptide Co., Ltd. binding interface at the aB/bB groove, the N-terminus of Neph- rin-PBM adopts a short helix structure to engage the hydrophobic GST Pull-Down Assay surface formed by bB/bC of PDZ3. Interestingly, the highly con- GST-tagged MAGI1-PDZ3 was incubated with thioredoxin- served residue Gly(23) of Nephrin-PBM plays a critical role in the tagged Nephrin-PBM in the assay buffer (50 mM Tris [pH 8.0], formation of the MAGI1/Nephrin complex, which is different 100mMNaCl,1mMDTT,and1mMEDTA)for1hourat4°C.For from the typical PDZ/PBM binding modes. We further inhibitory peptide competition experiments, GST-tagged Neph- developed a potent inhibitory peptide that can specifically block rin-PBMwasincubatedwithHis-taggedMAI1-PDZ3inthepres- the MAGI1/Nephrin interaction in vitro and in vivo. Our bioche- ence of the PBM_H-2T peptide or the PBM_G-3A peptide in the mic, biophysic, and cell biologic studies could provide valuable assay buffer. After incubation, the mixture was centrifuged at

2 Journal of the American Society of Nephrology J Am Soc Nephrol 29: ccc–ccc,2018 www.jasn.org BASIC RESEARCH

individually for 30 minutes. After washing three times, the bound proteins were eluted by boiling with 20 ml23 SDS- PAGE loading dye and detected by Western blotting with anti-Flag antibody (Sigma).

ITC Assay The ITC assay was performed on a MicroCal ITC-200 (Malvern Panalytical, UK) at 25°C. All of the protein samples were in the buffer containing 50 mM Tris (pH 8.0), 100 mM NaCl, 1 mM EDTA, and 1 mM DTT. Wild- type and various mutants of Nephrin-PBM (approximately 0.6 mM) and MAGI1-PDZ3 (approximately 0.04 mM) were placed in a syringe and the sample cell, respectively. The Nephrin-PBM fragments were injected into the MAGI1-PDZ3 proteins with 2 mlof Nephrin-PBM fragments at a time, at 2-minute intervals. The titration data were analyzed and ﬁtted by Origin 7.0 from MicroCal, using the one-site binding model.

Crystallography Thebestcrystalsofthe MAGI1-PDZ3/Nephrin- PBMcomplex were obtained by the sitting-drop vapor diffusion method at 16°C. The crystals weregrownin0.5Msodiumchloride, 0.01 M magnesium chloride hexahydrate, and 0.01 M hexadecyltrimethylammonium bromide. The crystals were cryoprotected with the mother liquor supplemented with Figure 1. Nephrin-PBM specifically interacts with PDZ3 domain of MAGI1. (A) Domain 25% glycerol and flash-frozen in liquid ni- organization of Nephrin. Sequence of Nephrin-PBM (-LPFELRGHLV) is shown. Ne- trogen. The diffraction data were collected at phrotic syndrome-causing mutations located at the cytoplasmic region of Nephrin are 100 K with a wavelength of 0.97791 Å and listed. (B and C) GST pull-down analyses of the interactions between the Nephrin-PBM processed with the HKL3000 package.15 The and various PDZ-containing proteins expressed in the slit diaphragm. The domain structure of the complex was solved by organizations of PDZ-containing proteins are shown (B). (D) GST pull-down analyses of the molecular replacement method using the interactions between various MAGI1 fragments and Nephrin-PBM. (E) Analytical the structure of MAGI1-PDZ3 (protein fi gel ltration-based analysis showing that MAGI1-PDZ3 forms a 1:1 stoichiometric data bank code: 2Q9V) using the software complex with Nephrin-PBM. (F) Quantitative measurement of the binding affinity suits of PHASER.16 The initial model between MAGI1-PDZ3 and Nephrin-PBM using an ITC-based assay. was rebuilt automatically and refined by PHENIX.17 Further model building and ad- 13,0003 rpm for 5 minutes and the supernatant was then loaded justment were completed using COOT.18 The final refinement into 20 ml GSH-Sepharose 4B slurry beads pre-equilibrated with statistics of the complex structure are listed in Supplemental the assay buffer, to incubate for 15 minutes at 4°C. After washing Table 1. Structural diagrams were prepared by Pymol. three times, the complex was boiled with 20 ml23 SDS-PAGE loading dye and detected by Coomassie brilliant blue staining. Coimmunoprecipitation Assay Flag-tagged PDZ-containing proteins andvarious MAGI1 frag- Flag-tagged full-length MAGI1 was cotransfected with GFP-tagged ments were overexpressed in HEK293T cells. Cells were lysed in wild-type or mutants of Nephrin full-length constructs into ice-cold cell lysis buffer (50 mM Hepes [pH 7.4], 150 mM NaCl, HEK293T cells. At 24 hours after transfection, the cells were lysed 10% glycerol, 1 mM EGTA, 1% Triton and protease inhibitor in ice-cold cell lysis buffer for 1 hour at 4°C, followed by centrifu- cocktail)for1hourat4°C.Aftercentrifugationat13,0003 rpm gation at 13,0003rpm for 10 minutes. The supernatants were each for 10 minutes at 4°C, the supernatants were incubated with GST- incubated with anti–Flag-conjugated agarose beads (Sigma) for 1 Nephrin-PBM on preloaded GSH-Sepharose 4B slurry beads hour at 4°C. For inhibitory peptide competition experiments, the

J Am Soc Nephrol 29: ccc–ccc, 2018 Three-Dimensional Structure of MAGI1/Nephrin Complex 3 BASIC RESEARCH www.jasn.org supernatants were incubated with the PBM_H-2T peptide or the Previous studies have demonstrated that PDZ2–3tandems PBM_G-3A peptide before incubation of anti–Flag-conjugated aga- of MAGI1 binds to Nephrin-CT.20,21 We further demonstrated rose beads. After washing three times, the samples were resuspended that the PDZ3 domain is sufficient for MAGI1 to bind to with SDS-PAGE dye and analyzed by Western blotting using Nephrin-PBM through a truncation-based approach (Figure anti-Flag and anti-GFP (Santa Cruz Biotechnology) antibodies. 1D). MAGI1-PDZ3 and Nephrin-PBM formed a stable 1:1 stoichiometric complex in solution (Figure 1E). ITC-based Cellular Localization Assay assay showed that Nephrin-PBM binds to MAGI1-PDZ3 HEK293T cells were transiently transfected with 0.5 mgofeach with a Kd of approximately 0.54 mM (Figure 1F), which is plasmid per well using a lipofectamine 2000 reagent according to much stronger than those of canonical PDZ/PBM interactions the manufacturer’s instructions (Invitrogen, Thermo Fisher Sci- (Kd approximately 10 mM). entific) in 12-well plates. Cells were cultured for 12 hours in DMEM containing 10% FBS in 5% CO2, before fixation. The Structural Characterization of Nephrin-PBM/MAGI1- cell images were acquired on a Leica TCS SP8 confocal micro- PDZ3 Complex scope (Hyvolution) with a 633 oil immersion lens. Fluorescence We further determined the crystal structure of the Nephrin- intensities of each image were analyzed using ImageJ software. PBM/MAGI1-PDZ3 complex at 1.8-Å resolution (Supplemental To quantify membrane colocalization of Nephrin and MAGI1, Table 1). In the complex structure, MAGI1-PDZ3 adopts a ca- we used the Pearson correlation coefficient (R). Data were nonical PDZ domain folding consisting of six b-strands (bA–F) analyzed using a t test. Twenty cells were randomly selected as and two helices (aAandaB) (Figure 2A); unlike the canonical quantification for each group and values (means6SD) were PBM, which folds into a b-strand to insert into the concave calculated from three independent experiments. groove formed by aB/bB of PDZ (Figure 2B), Nephrin-PBM adopts a unique conformation to contact PDZ3 with the C-terminus Accession Code of PBM, binding to the aB/bB groove and N-terminus of PBM, The atomic coordinate of MAGI1-PDZ3/Nephrin-PBM com- which forms a short a-helix coupling with the bB/bCsurfaceof plex has been deposited to the Protein Data Bank under the PDZ3 (Figure 2, A and C). accession code 5ZYS. Analysis of the PDZ3/PBM binding interface revealed several unique features (Figure 2D): (1) the N-terminus of PBM forms another hydrophobic core (interaction between RESULTS Phe(27)PBM and Leu848PDZ3/Tyr858PDZ3) in addition to the canonical PDZ core (interaction between Val(0)PBM and Specific Interaction between Nephrin-PBM and Phe843PDZ3/Phe845PDZ3/Met900PDZ3); (2) His(22)PBM is not MAGI1-PDZ3 involved in the binding, although the residue at 22position Various scaffold proteins have been found to link Nephrin with usually determines the binding type of a typical PBM9;and(3) downstream cytoskeleton components in slit diaphragm. Toun- Arg846PDZ3 forms hydrogen bonds with the backbones of Gly cover the detailed composition of Nephrin-PBM–assembled (23)PBM and Arg(24)PBM. Importantly, the residues involved in protein complexes, we first systematically studied the inter- the binding interface are conserved in MAGI1 (Supplemental actions between Nephrin-PBM and the PDZ-containing pro- Figure 2) and Nephrin (Figure 2E) from different species, sug- teins expressed in slit diaphragm. Flag-tagged PDZ domain(s) gesting that the interaction is highly conserved during evolution. of ten PDZ-containing proteins expressed in slit diaphragm To probe the role of the hydrophobic interaction between were tested individually for their bindings to GST-tagged the N-terminus of PBM and the bB/bC surface of PDZ3, mu- Nephrin-PBM in vitro (Figure 1, A and B). Surprisingly, tations of residues in this interface (i.e.,L848DPDZ3 and F(27) none of these PDZ domain(s) bound to Nephrin-PBM except DPBM) were introduced. These mutants led to significant forMAGI1(Figure1C).Mostnotably,MAGI2,theclosest weakening of both the PDZ3/PBM interaction (Figure 3A) homolog to MAGI1, did not bind to Nephrin-PBM, indicat- and the full-length Nephrin/MAGI1 interaction (Figure 3B). ing that the MAGI1/Nephrin-PBM interaction is highly spe- Substitution of Arg846PDZ3 with Ala, Ser, or even Lys led to cific. MAGI1 is a unique member of the membrane-associated obvious decreased PDZ3/PBM interaction (Figure 3A). guanylate kinase family of proteins, which is enriched in the Arg846 is totally conserved in the PDZ3 domain from differ- glomerular podocytes.19 MAGI1 was reported to bind to ent species of MAGI1 (Supplemental Figure 2), but not con- Nephrin and the interaction was required for Rap1 activation served among different PDZ domains of the MAGI1 protein in podocyte to maintain long-term slit diaphragm struc- (Figure 2F), which may explain why other PDZ domains of ture.20,21 Membrane localization of Nephrin was diminished MAGI1 did not bind to Nephrin-PBM. when MAGI1 was depleted in podocytes.20 The combination It is noted that, in addition to forming a hydrogen bond with of heterozygous deletion of NPHS1 and homozygous deletion Arg846PDZ3, the backbone of Gly(23)PBM (adopting unique of MAGI1 resulted in spontaneous glomerulosclerosis,20 dihedral angles of w [82.24°] and c [2177.40°], which are not indicating a genetic complementation of the two genes permitted by other residues) allows the PBM peptide to (NPHS1 and MAGI1). adopt a sharp turn to facilitate the upstream of PBM to interact

4 Journal of the American Society of Nephrology J Am Soc Nephrol 29: ccc–ccc,2018 www.jasn.org BASIC RESEARCH

with bB/bC of PDZ3. Importantly, substitution of Gly(23) with Ala completely disrupted the PDZ3/PBM interaction, sub- stantiating the critical structural role of Gly (23) in the binding process (Figure 3A). Consistently, the G(23)A mutant also blocked the binding of the full-length Nephrin to MAGI1 (Figure 3B). Analysis of PBMs of other reported targets of MAGI1-PDZ3, such as endothelial cell-selective adhesion molecule (ESAM)22 and coxsackievirus and adenovirus receptor,23 showed that the residues at 23 position are both Gly (Supplemental Figure 3A). ESAM-PBM binds to MAGI1-PDZ3 with a Kd of approximately 2.1 mM, whereas G(23) AESAM showed undetectable binding to MAGI1-PDZ3 (Supplemental Figure 3B). These data collectively suggest that Gly at the 23 position of PBMs may be a determining feature in their specific binding to MAGI1-PDZ3. Wenext wanted to investigate the roles of the above key elements for the MAGI1/ Nephrin interaction in living cells. Because of the relatively low expression level of the full-length Nephrin in the heterologous cells, wehypothesizethatfusionoftheNephrin-CT to a membrane-associated protein could per- fectly mimic the native Nephrin protein at plasma membrane. Pleckstrin homology (PH) domain, one of the largest domain fam- iliesformembraneassociation,isoftenusedas Figure 2. Complex structure of MAGI1-PDZ3/Nephrin-PBM reveals a unique binding lipid marker or membrane recruitment tag. mode of PDZ/PBM interaction. (A) Ribbon diagram of the MAGI1-PDZ3/Nephrin-PBM Myosin X (MyoX) contains three PH domains complex structure. MAGI1-PDZ3 and Nephrin-PBM are colored in pink and green, re- that form a tandem and cooperatively bind to spectively. (B) A cartoon showing the canonical type-I PDZ binding mode. The carboxyl the cellular membrane, with extremely group of type-I PBM (purple) forms extensive hydrogen bonds with the “2GLGF” motif high affinity.24,25 Therefore, we made the of PDZ. The residue at 22 position (Ser or Thr) of type-I PBM determines the binding MyoX-PH123-Nephrin-CT (MyoX-PH- specificity by forming the hydrogen bond with His from aB of PDZ. (C) A cartoon showing Nephrin-CT) chimera protein. As expected, the unexpected MAGI1-PDZ3/Nephrin-PBM (green) binding mode. N-terminus of MyoX-PH-Nephrin-CT effectively recruits b b Nephrin-PBM contacts with the hydrophobic surface of B/ C of PDZ3. Note that His893 MAGI1 to the cell membrane (Figure 3, D a c from B of PDZ3 has no contact with Nephrin-PBM. represents hydrophobic residue. and G), whereas MyoX-PH alone could not (D) Stereo view of detailed interface of MAGI1-PDZ3 (pink)/Nephrin-PBM (green) com- (Figure 3, C and G). Fully consistent with the plex. Dotted lines represent hydrogen bonds. A water molecule is shown in brown. (E) in vitro binding data, neither the R846A Sequence alignment of Nephrin-PBM from different spices. In this drawing, totally con- 2 served and conserved residues are colored in red and green, respectively. Residues mutant of MAGI1 nor the G( 3)A mutant involved in binding to Nephrin-PDZ3 are shown with a blue dot at the bottom. of Nephrin could colocalize with wild-type (F) Structure-based sequence alignment of different PDZ domains of MAGI1. Conserved Nephrin or MAGI1 at plasma membrane, residues between different PDZ domains are highlighted with pink boxes. The totally respectively (Figure 3, E–G). conserved and conserved residues in MAGI1-PDZ3 form different species are colored in red and green, respectively (related to Supplemental Figure 2). Residues involved in A Potent Inhibitory Peptide Capable binding to Nephrin-PBM and Ser/Thr at 22 position of canonical PBM are shown with red of Blocking Nephrin/MAGI1 and blue dots at the bottom, respectively. Interaction MAGI1-PDZ3 is a type I PDZ domain, with a His (H893) at aB capable of forming a

J Am Soc Nephrol 29: ccc–ccc, 2018 Three-Dimensional Structure of MAGI1/Nephrin Complex 5 BASIC RESEARCH www.jasn.org

Figure 3. Mutations of specific residues in the interface disrupt Nephrin/MAGI1 interaction. (A) A summary of the binding affinities between PDZ3 (or its mutants) and PBM (or its mutants) measured by ITC assays. (B) A single mutation in MAGI1 and Nephrin significantly impaired the interactions between full-length MAGI1 and Nephrin in the co-immunoprecipitation assay. (C–F) The cellular colocalization analyses of Nephrin, MAGI1, and their mutations in transfected HEK293T cells. Because the expression level of the full- length Nephrin is extremely low, a chimera protein with the Nephrin-CT fused to the C-terminal end of GFP-Myosin X-PH123 (GFP- MyoX-PH) was effectively localized at the membrane. Fluorescence intensities of the area marked by the white lines (left panels) are shown in the right panels. Green and red curves represent the fluorescence intensities of GFP and RFP, respectively. Colocalizations of Nephrin and MAGI1 are indicated by white arrows. Scale bars, 5 mm. (G) Quantification of colocalization coefficient (Pearson) (R) between wild-type and mutants of Nephrin and MAGI1. Values are means6SD from three independent experiments (n=20 for each group), analyzed with t test; ****P,0.001.

6 Journal of the American Society of Nephrology J Am Soc Nephrol 29: ccc–ccc,2018 www.jasn.org BASIC RESEARCH hydrogen bond with a Ser or Thr at 22 position of a canonical rescued MAGI2/Nephrin-PBM interaction (Figure 5D). The fact type I PBM (Figure 2, B and C). However, the residue at the 22 that changing only one residue of MAGI2-PDZ3 to the corre- position of Nephrin-PBM is a His that does not bind to sponding residue in MAGI1-PDZ3 can convert MAGI2 into a His893PDZ3 (Figure 2C). Thus, we reason that substitution of Nephrin-PBM binding protein further illustrates the exquisite His(22)PBM with a Thr would further enhance the binding be- specificity of the binding between MAGI1 and Nephrin. tween PDZ3 and PBM. Our modeled structure also suggested that the sidechain of T(22)PBM_H-2T would form a hydrogen bond with the sidechain of His893PDZ3 (Figure 4A). The DISCUSSION PBM_H-2T peptide binds to MAGI1-PDZ3 with a Kd of approximately 27 nM, which is about 20-fold higher than that of wild- In this study, we systematically investigated the interaction type PBM/PDZ3 interaction (Figures 1F and 4B). between the evolutionarily conserved atypical PBM of Nephrin We next tested whether the PBM_H-2T peptide can block the and various PDZ-containing proteins in the slit diaphragm and Nephrin/MAGI1 interaction in vitro. As expected, GST-tagged demonstrated that only MAGI1 could bind to Nephrin-PBM. Nephrin-PBM could robustly pull down MAGI1-PDZ3 (Figure We further elucidated the molecular basis of the exquisite 4C, lane 1). Addition of the PBM_H-2T peptide in the reaction binding specificity of MAGI1-PDZ3/Nephrin-PBM complex mixture significantly reduced the interaction between Nephrin- via combination of the biochemic, biophysic, and cell biologic PBM and MAGI1-PDZ3 (Figure 4C, lanes 2 and 3), whereas the approaches. In the podocytes, MAGI1 connects Nephrin with PBM_G-3A mutant peptide had no effect on the binding (Figure cytoskeletons via binding to various proteins such as dendrin, 4C, lane 4). We further evaluated whether the PBM_H-2T peptide actinin, CD2AP, etc., which is essential for stable assembly of can block the full-length Nephrin/MAGI1 interaction. Similarly, slit diaphragm (Figure 5E). It is worth noting that the specific the PBM_H-2T peptide significantly reduced the interaction be- Nephrin/MAGI1 interaction is not only required for slit tween full-length Nephrin and MAGI1, whereas the PBM_G-3A diaphragm assembly, but also critical for Nephrin’sroleas mutant peptide had no effect on the binding in a coimmunopre- an intracellular signaling hub in slit diaphragm homeostasis. cipitation assay (Figure 4D), further confirming the specificity A recent study suggested that Nephrin/MAGI1 interaction is of the PBM_H-2T peptide in blocking the Nephrin/MAGI1 crucial for proper activity of Rap1 in the podocytes.20 Rap1 interaction in vitro. activation is tightly balanced to sustain proper podocyte func- We next investigated whether the PBM_H-2T peptide can tion. How does Nephrin/MAGI1 complex promote the acti- effectively modulate the membrane colocalization of Nephrin vation of Rap1? We recent discovered that MAGI1 can bind and MAGI1 in vivo. Satisfyingly, the membrane localization of directly to RapGEF2, a guanine nucleotide exchange factor of MAGI1 in cells expressing the PBM_H-2T peptide was signifi- Rap1 that is required for Rap1 activation (Z.F. Weng, J.W. Zhu, cantly reduced compared with those expressing the PBM_G-3A unpublished data). Thus, it is most likely that the Nephrin- peptide (Figure 4, E and G), indicating that expression of the MAGI1-RapGEF2-Rap1 axis plays critical role in maintaining PBM_H-2T peptide, most likely through disrupting the interac- the long-term slit diaphragm structure (Figure 5E). tion between Nephrin-atypical PBM and MAGI1, can block Notably, although global MAGI1 knockout mice demon- membrane colocalization of Nephrin and MAGI1 in living cells. strated normal glomerular histology and function into adult- hood, NPHS1 but not NEPH1 heterozygosity is capable of A Gain-of-Function Mutant of MAGI2 Rescues providing podocytes with a “second hit” when combined Nephrin-PBM Binding with loss of MAGI1, leading to spontaneous FSGS.20 These An unexpected finding in our study was that MAGI2 does not data suggested a specific genetic interaction between NPHS1 bind to Nephrin-PBM, although the amino acid sequences of and MAGI1. Such a multihit scenario is typical for glomerular PDZ3 domains from MAGI1 and MAGI2 are highly conserved disease initiation and progress, where two or more insults may (Figure 5A). We further found that MAGI3 also does not bind to be necessary for glomerular disease pathogenesis.20 For exam- Nephrin-PBM (Figure 5C). Detailed analysis of the sequence ple, bigenic heterozygosity of Cd2ap and either Synaptopodin alignment, together with careful analysis of complex structure, or Fyn resulted in spontaneous proteinuria and in FSGS-like 26 revealed that only two residues (Tyr858MAGI1 and His861MAGI1) glomerular damage ; combinations of heterozygous NPHS2 are different between MAGI1-PDZ3 and MAGI2-PDZ3 (Figure mutations and a heterozygous NPHS1 mutation lead to 27 5A). In the structure, His861MAGI1 forms a hydrogen bond with FSGS. Interestingly, MAGI2, the closest homolog to Glu(26)PBM; Tyr858MAGI1, together with L848MAGI1, forms hy- MAGI1, did not bind to Nephrin-PBM in our biochemic drophobic interactions with Phe(27)PBM (Figure 5B). We first study, although the MAGI2 knockout mice presented with mutated Ala805 (the corresponding residue in MAGI1 is progressive proteinuria, diffused podocyte foot process efface- 28 His861) in MAGI2 to His (A805HMAGI2). Surprisingly, the ment, and died of kidney failure. Mutations of MAGI2 have A805HMAGI2 mutant partial rescued the binding between recently been identified to cause congenital nephritic syn- MAGI2 and Nephrin-PBM (Figure 5D). Further substitution drome in human.29 MAGI2 is tightly associated with Nephrin of Leu802 (the corresponding residue in MAGI1 is Tyr858) in the regulation of podocyte cytoskeleton and slit diaphragm 11 into Tyr (i.e., A805H/L802YMAGI2 double mutant) completely dynamics. Loss of MAGI2 function led to a significant

J Am Soc Nephrol 29: ccc–ccc, 2018 Three-Dimensional Structure of MAGI1/Nephrin Complex 7 BASIC RESEARCH www.jasn.org

Figure 4. A potent inhibitory peptide is capable of disrupting Nephrin/MAGI1 interaction. (A) Combined stick and ribbon repre- sentation showing a structural homology model (on the basis of the Nephrin-PBM/MAGI1-PDZ3 structure in this study) for PBM_H-2T/ MAGI1-PDZ3 complex using the SWISS MODEL tool. The sequences of Nephrin-PBM and PBM_H-2T are shown above. (B) ITC-based measurements of binding affinity between the PBM_H-2T peptide and MAGI1-PDZ3. (C) The PBM_H-2T peptide effectively blocked Nephrin-PBM/MAGI1-PDZ3 interaction in a GST pull-down assay. (D) The PBM_H-2T peptide effectively blocked full-length Nephrin/ MAGI1 interaction in a coimmunoprecipitation assay. (E and F) The membrane colocalization analyses of Nephrin and MAGI1 in living cells when expressing the PBM_H-2T peptide or the PBM_G-3A mutant peptide. Fluorescence intensities of the area marked by the white lines in e1, e2, and e3 are shown in f1, f2, and f3, respectively. The expression levels of the RFP-PBM_H2T and RFP-PBM_G-3A peptides are also included. (G) Quantification of colocalization coefficient (Pearson) (R) between Nephrin and MAGI1 revealing that the PBM_H-2T peptide significantly blocks the Nephrin/MAGI1 in living cells. Values are means6SD from three independent experiments (n=20 for each group), analyzed with t test; ****P,0.001. decrease of Nephrin and dendrin at slit diaphragm.28 Although of Nephrin than PBM, or it may associate with Nephrin via our data indicate that Nephrin-PBM does not bind to MAGI2, it other adaptor protein(s). Further work is required to dissect could not be ruled out that MAGI2 interacts with other regions the differential roles of MAGI1 and MAGI2 in the podocytes.

8 Journal of the American Society of Nephrology J Am Soc Nephrol 29: ccc–ccc,2018 www.jasn.org BASIC RESEARCH

nephrotic syndrome. The mutant protein lacks the very C-terminal Valine of PBM and instead bears an additional 45 amino acids at the C-terminus (Supplemental Figure 1, A and B). On the basis of our structure, the mutant would diminish the Nephrin/MAGI1 interaction, thus impairing Nephrin/MAGI1-mediated function in the podocytes. Indeed, the impaired PBM of the p.L1240fs12863 mutant did not bind to MAGI1-PDZ3 (Supplemental Figure 1C). In addition, several truncation mutations (e.g., p.R11093, p.Q11483, and p.R11603; Figure 1A) have been identiﬁed in patients with the congenital nephrotic syndrome.6 These mutations all lack the PBM and would deﬁnitely disrupt the Nephrin/ MAGI1 interaction, leading to improper podocyte function. Therefore, our work may provide possible explanations for the glomerular disease-associated mutants in NPHS1.

ACKNOWLEDGMENTS

We thank the Shanghai Synchrotron Radiation Facility Beamline 18U1 for x-ray beam time, the National Center for Protein Science Shanghai (Protein Expression and Purification system, and Molecular Imaging System) for their instrument support and technical assistance, Professor Qing Lu (Shanghai Jiao Tong University) for technical support during imaging processes, and Professor Mingjie Zhang (Hong Kong University of Science and Technology) for the critical reading of the Figure 5. A single mutation converts MAGI2 into a Nephrin-PBM binder. (A) Structure- manuscript. based sequence alignment of PDZ3 domains of MAGI1–3. The totally conserved and conserved residues are colored in red and green, respectively. The residues involved in J.Z.andR.Z.designedtheexperiments.Z.W.and the binding but not conserved among MAGI1–3 are highlighted by a black triangle at the Z.J. performed the experiments. Y.S., L.L., F.Y., bottom. (B) The combined surface and ribbon representations of the PDZ3/PBM com- and Z.J. contributed to x-ray data collection and plex. The hydrophobic residues, positively charged residues, and negatively charged structure determination. Z.W., R.Z., and J.Z. ana- residues of PDZ3 are colored in yellow, blue, and red, respectively. (C) GST pull-down lyzedthedata.J.Z.wrotethemanuscript.Allauthors assay showing that Nephrin-PBM did not bind to MAGI2 and MAGI3. (D) Substitution of approved the final version of the manuscript. Ala805 of MAGI2 with His would partially rescue the binding to Nephrin-PBM. Further This work was supported by grants from the replacement of Leu802 to Tyr would enhance the MAGI2/Nephrin-PBM interaction. (E) National Natural Science Foundation of China The proposed schematic model illustrating the molecular basis of Nephrin/MAGI1 in- (U1532121, 31470733, and 31770779) to J.Z., the teraction in slit diaphragm assembly and signaling. StrategicPriorityResearchProgramoftheChinese Academy of Sciences (XDB08030104) and the We believe that the potent inhibitory peptide capable of block- Chief Scientist Program of Shanghai Institutes for Biological Sciences, ing the Nephrin/MAGI1 interaction would offer a valuable Chinese Academy of Sciences to R.Z. manipulating tool in these studies. Intriguingly, a protein-truncating mutation (p.L1240fs12863) DISCLOSURES of Nephrin was identified in patients with steroid-resistant None.

J Am Soc Nephrol 29: ccc–ccc, 2018 Three-Dimensional Structure of MAGI1/Nephrin Complex 9 BASIC RESEARCH www.jasn.org

REFERENCES 17. Adams PD, Afonine PV, Bunkóczi G, Chen VB, Davis IW, Echols N, et al.: PHENIX: A comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr D Biol Crystallogr 66: 213–221, 2010 1. Greka A, Mundel P: Cell biology and pathology of podocytes. Annu Rev 18. Emsley P, Cowtan K: Coot: Model-building tools for molecular graph- Physiol 74: 299–323, 2012 ics. Acta Crystallogr D Biol Crystallogr 60: 2126–2132, 2004 2. GrahammerF,SchellC,HuberTB:Thepodocyteslitdiaphragm–from a thin grey line to a complex signalling hub. Nat Rev Nephrol 9: 587–598, 2013 19. Patrie KM, Drescher AJ, Goyal M, Wiggins RC, Margolis B: The mem- 3. Perico L, Conti S, Benigni A, Remuzzi G: Podocyte-actin dynamics in brane-associated guanylate kinase protein MAGI-1 binds megalin and is – health and disease. Nat Rev Nephrol 12: 692–710, 2016 present in glomerular podocytes. J Am Soc Nephrol 12: 667 677, 2001 4. Assady S, Wanner N, Skorecki KL, Huber TB: New insights into podo- 20. Ni J, Bao S, Johnson RI, Zhu B, Li J, Vadaparampil J, et al.: MAGI-1 cyte biology in glomerular health and disease. J Am Soc Nephrol 28: interacts with nephrin to maintain slit diaphragm structure through – 1707–1715, 2017 enhanced Rap1 activation in podocytes. J Biol Chem 291: 24406 5. Kestilä M, Lenkkeri U, Männikkö M, Lamerdin J, McCready P, Putaala H, 24417, 2016 et al.: Positionally cloned gene for a novel glomerular protein–nephrin–is 21. Hirabayashi S, Mori H, Kansaku A, Kurihara H, Sakai T, Shimizu F, et al.: mutated in congenital nephrotic syndrome. Mol Cell 1: 575–582, 1998 MAGI-1 is a component of the glomerular slit diaphragm that is tightly – 6. Schoeb DS, Chernin G, Heeringa SF, Matejas V, Held S, Vega-Warner V, et al.; associated with nephrin. Lab Invest 85: 1528 1543, 2005 Gesselschaft für Paediatrische Nephrologie (GPN) Study Group: Nineteen 22. Wegmann F, Ebnet K, Du Pasquier L, Vestweber D, Butz S: Endothelial novel NPHS1 mutations in a worldwide cohort of patients with congenital adhesion molecule ESAM binds directly to the multidomain adaptor – nephrotic syndrome (CNS). Nephrol Dial Transplant 25: 2970–2976, 2010 MAGI-1 and recruits it to cell contacts. Exp Cell Res 300: 121 133, 2004 7. Putaala H, Soininen R, Kilpeläinen P, Wartiovaara J, Tryggvason K: The 23. Excoffon KJ, Hruska-Hageman A, Klotz M, Traver GL, Zabner J: A role for murine nephrin gene is specifically expressed in kidney, brain and the PDZ-binding domain of the coxsackie B virus and adenovirus receptor pancreas: Inactivation of the gene leads to massive proteinuria and (CAR) in cell adhesion and growth. JCellSci117: 4401–4409, 2004 neonatal death. Hum Mol Genet 10: 1–8, 2001 24. Lu Q, Yu J, Yan J, Wei Z, Zhang M: Structural basis of the myosin X 8. Conti S, Perico L, Grahammer F, Huber TB: The long journey through PH1(N)-PH2-PH1(C) tandem as a specific and acute cellular PI(3,4,5)P(3) renal filtration: New pieces in the puzzle of slit diaphragm architecture. sensor. Mol Biol Cell 22: 4268–4278, 2011 Curr Opin Nephrol Hypertens 26: 148–153, 2017 25. Gassama-Diagne A, Yu W, ter Beest M, Martin-Belmonte F, Kierbel A, 9. Zhang M, Wang W: Organization of signaling complexes by PDZ-do- Engel J, et al.: Phosphatidylinositol-3,4,5-trisphosphate regulates the main scaffold proteins. Acc Chem Res 36: 530–538, 2003 formation of the basolateral plasma membrane in epithelial cells. Nat 10. Philippe A, Nevo F, Esquivel EL, Reklaityte D, Gribouval O, Tête MJ, Cell Biol 8: 963–970, 2006 et al.: Nephrin mutations can cause childhood-onset steroid-resistant 26. Huber TB, Kwoh C, Wu H, Asanuma K, Gödel M, Hartleben B, et al.: nephrotic syndrome. JAmSocNephrol19: 1871–1878, 2008 Bigenic mouse models of focal segmental glomerulosclerosis involving 11. Lehtonen S, Ryan JJ, Kudlicka K, Iino N, Zhou H, Farquhar MG: Cell pairwise interaction of CD2AP, Fyn, and synaptopodin. JClinInvest junction-associated proteins IQGAP1, MAGI-2, CASK, spectrins, and 116: 1337–1345, 2006 alpha-actinin are components of the nephrin multiprotein complex. 27. Löwik M, Levtchenko E, Westra D, Groenen P, Steenbergen E, Proc Natl Acad Sci U S A 102: 9814–9819, 2005 Weening J, et al.: Bigenic heterozygosity and the development of 12. Fukasawa H, Bornheimer S, Kudlicka K, Farquhar MG: Slit diaphragms steroid-resistant focal segmental glomerulosclerosis. Nephrol Dial contain tight junction proteins. JAmSocNephrol20: 1491–1503, 2009 Transplant 23: 3146–3151, 2008 13. Hartleben B, Schweizer H, Lübben P, Bartram MP, Möller CC, Herr R, 28. Ihara K, Asanuma K, Fukuda T, Ohwada S, Yoshida M, Nishimori K: et al.: Neph-Nephrin proteins bind the Par3-Par6-atypical protein ki- MAGI-2 is critical for the formation and maintenance of the glomerular nase C (aPKC) complex to regulate podocyte cell polarity. JBiolChem filtration barrier in mouse kidney. Am J Pathol 184: 2699–2708, 2014 283: 23033–23038, 2008 29. Bierzynska A, Soderquest K, Dean P, Colby E, Rollason R, Jones C, et al.; 14. Hartleben B, Widmeier E, Wanner N, Schmidts M, Kim ST, Schneider L, UK study of Nephrotic Syndrome: MAGI2 mutations cause congenital et al.: Role of the polarity protein Scribble for podocyte differentiation nephrotic syndrome. JAmSocNephrol28: 1614–1621, 2017 and maintenance. PLoS One 7: e36705, 2012 15. Otwinowski Z, Minor W: [20] Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol 276: 307–326, 1997 16. McCoy AJ, Grosse-Kunstleve RW, Adams PD, Winn MD, Storoni LC, Read This article contains supplemental material online at http://jasn.asnjournals. RJ: Phaser crystallographic software. J Appl Cryst 40: 658–674, 2007 org/lookup/suppl/doi:10.1681/ASN.2017121275/-/DCSupplemental.

AFFILIATIONS

1State Key Laboratory of Molecular Biology, Shanghai Science Research Center, Chinese Academy of Sciences Center for Excellence in Molecular Cell Science, Shanghai Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences, University of Chinese Academy of Sciences, Shanghai, China; 2School of Life Science and Technology, ShanghaiTech University, Shanghai, China; and 3Division of Life Science, State Key Laboratory of Molecular Neuroscience, and 4Center of Systems Biology and Human Health, Hong Kong University of Science and Technology, Kowloon, Hong Kong, China

10 Journal of the American Society of Nephrology J Am Soc Nephrol 29: ccc–ccc,2018