Characterization of the interactions of the nephrin intracellular domain Evidence that the scaffolding protein IQGAP1 associates with nephrin Xiao Li Liu 1 , Pekka Kilpela ¨ inen 1 , Ulf Hellman 3 , Yi Sun 1 , Jorma Wartiovaara 4 , Ekaterina Morgunova 1 , Timo Pikkarainen 1 , Kunimasa Yan 5 , Anders P. Jonsson 2 and Karl Tryggvason 1 1 Divisions of Matrix Biology, Department of Medical Biochemistry and Biophysics, Karolinska Institute, Stockholm, Sweden 2 Medical Chemistry, Department of Medical Biochemistry and Biophysics, Karolinska Institute, Stockholm, Sweden 3 Ludwig Institute for Cancer Research, Uppsala, Sweden 4 Electron Microscopy Unit, Institute of Biotechnology, University of Helsinki, Finland 5 Department of Pediatrics, Kyorin University School of Medicine, Mitaka, Tokyo, Japan Keywords Fyn, IQGAP1, phosphoinositide 3-kinase, podocyte, slit diaphragm Correspondence K. Tryggvason, Karolinska Institutet Department of Medical Biochemistry and Biophysics, Division of Matrix Biology, Scheeles Va ¨ g 2 B1, Plan 4 SE-17177, Stockholm, Sweden Fax: +46 8 316165 Tel: +46 8 5248 7720 E-mail: karl.tryggvason@mbb.ki.se (Received 22 July 2004, revised 21 September 2004, accepted 22 September 2004) doi:10.1111/j.1432-1033.2004.04408.x Nephrin is a signalling cell–cell adhesion protein of the Ig superfamily and the first identified component of the slit diaphragm that forms the critical and ultimate part of the glomerular ultrafiltration barrier. The extracellular domains of the nephrin molecules form a network of homophilic and heterophilic interactions building the structural scaffold of the slit dia- phragm between the podocyte foot processes. The intracellular domain of nephrin is connected indirectly to the actin cytoskeleton, is tyrosine phos- phorylated, and mediates signalling from the slit diaphragm into the podo- cytes. CD2AP, podocin, Fyn kinase, and phosphoinositide 3-kinase are reported intracellular interacting partners of nephrin, although the biologi- cal roles of these interactions are unclarified. To characterize the structural properties and protein–protein interactions of the nephrin intracellular domain, we produced a series of recombinant nephrin proteins. These were able to bind all previously identified ligands, although the interaction with CD2AP appeared to be of extremely low stoichiometry. Fyn phosphory- lated nephrin proteins efficiently in vitro. This phosphorylation was required for the binding of phosphoinositide 3-kinase, and significantly enhanced binding of Fyn itself. A protein of 190 kDa was found to associate with the immobilized glutathione S-transferase–nephrin. Peptide mass finger- printing and amino acid sequencing identified this protein as IQGAP1, an effector protein of small GTPases Rac1 and Cdc42 and a putative regula- tor of cell–cell adherens junctions. IQGAP1 is expressed in podocytes at significant levels, and could be found at the immediate vicinity of the slit diaphragm. However, further studies are needed to confirm the biological significance of this interaction and its occurrence in vivo. Abbreviations CD2AP, CD2-associated protein; CHAPS, 3-[(3-cholamidopropyl)dimethyl-ammonio]-1-propanesulfonate; FITC, fluorescein-isothiocyanate; HEK293, human epidermal kidney cells; HT1080, human fibrosarcoma; HT1080 L, mouse fibroblast; pAb 2, polyclonal antibody 2; PI 3, phosphoinositide 3. 228 FEBS Journal 272 (2005) 228–243 ª 2004 FEBS We have seen in recent years the rapid unravelling of the molecular components of the slit diaphragm, a spe- cialized structure connecting the podocyte foot proces- ses. An important finding in this regard was the identification of nephrin as a gene mutated in congen- ital nephrosis NPHS1 [1,2]. NPHS1 patients show a massive proteinuria starting in utero and their podo- cyte foot processes are effaced. Identical symptoms can be brought about in mice by inactivating the nephrin gene by homologous recombination [3]. These mice die from renal failure within 24 h after birth, which under- lines the role of nephrin as a key component of the slit diaphragm. The nephrin protein has an apparent molecular mass of 180 kDa [1]. The extracellular part of nephrin has eight Ig-like modules, one fibronectin type III-like module, and an unknown number of N-linked carbo- hydrate moieties [4,5]. The intracellular domain of 156 amino acids has no homology with other known pro- teins but contains nine tyrosine residues, some of which are phosphorylated [6]. The domain structure and biochemical properties of nephrin brought about the immediate suggestion that nephrin molecules may form dimers through homophilic interactions spanning the slit diaphragm [1]. Very recently, our electron tomography studies showed that the slit diaphragm is a network of intertwined strands containing nephrin [7]. Consistently, biochemical studies have indicated that nephrin has an ability to form homo- and hetero- dimers with NEPH1 [8–10], another component of the slit diaphragm. NEPH1 belongs to the Ig super- family, has five extracellular Ig domains, and its dele- tion leads to heavy proteinuria and early postnatal death [11]. P-cadherin and FAT, a large cadherin pro- tein, have been localized at the slit diaphragm area or close to it [12,13]. Their participation and role in the formation of the slit diaphragm structure is not known. Inactivation of the FAT gene in mice leads to severe renal phenotype with the fusion of foot proces- ses and loss of slit junctions [14], but P-cadherin- knockout mice do not show any abnormalities in the kidney [15,16]. The best-documented intracellular interaction part- ner of nephrin is podocin [17,18]. It is a member of the stomatin protein family and is predicted to form a membrane-associated hairpin-like structure with cyto- solic N- and C-terminal domains. Podocin has been localized to the slit diaphragm area [19], and it is mutated in a form of autosomal recessive familial focal segmental glomerulosclerosis (SRN1) [20]. Podocin is required for recruitment of nephrin into lipid rafts [21], and podocin-knockout mice die from renal failure within a few days of birth [22]. In addition to nephrin, it interacts with the NEPH proteins and CD2-associ- ated adaptor protein CD2AP [18,23]. Furthermore, nephrin itself has been reported to coimmunoprecipi- tate with CD2AP [24] that connects several membrane proteins to the actin cytoskeleton [25]. The importance of CD2AP for glomerular ultrafiltration is emphasized by the fact that CD2AP-knockout mice develop neph- rotic syndrome 1–2 weeks after birth that leads to renal failure and death at 6–7 weeks of age [24]. More- over, heterozygous CD2AP+ ⁄ –, mice are haploinsuffi- cient and display glomerular changes at 9 month’s of age with a histological pattern similar to that in human focal segmental glomerulosclerosis [26]. CD2AP has been localized close to the attachment site of the slit diaphragm [27], and its interaction with nephrin has been examined also in vitro, although the mapping of interacting domain for nephrin binding resulted in controversial reports [27,28]. Studies employing gradient centrifugation and immunofluores- cence techniques have demonstrated that nephrin is indeed either directly or indirectly associated with the actin cytoskeleton [29–31]. Altered morphological characteristics of podocytes such as foot process effacement are typical to many experimental and human glomerulopathies [32,33]. All these changes are reversible and recovery from these anomalies can only be achieved by the reformation of foot processes and slit diaphragms. Nephrin expression levels and subcellular localization are affected in sev- eral renal diseases and proteinuric diseases such as dia- betes [34]. This implies that there has to be signalling from the slit diaphragm into the podocytes, and vice versa. Nephrin is very probaly a major player in this signalling. Transfection of nephrin into HEK293 cells activates protein kinase p38 and c-jun aminoterminal kinase (JNK), thereby activating transcription factor AP-1 [17]. Nephrin has been found to be dislocated to the apical pole of the narrowed filtration slits and tyro- sine phosphorylated, when the slit diaphragm is dis- rupted by in vivo injection of antibodies recognizing a podocyte-specific 9-O-acetylated GD3 ganglioside [6]. Fyn kinase has been shown to bind and phosphorylate nephrin [35], and we have observed that tyrosine phosphorylation of nephrin is induced robustly, when nephrin is clustered on the cell surface, by using anti-nephrin Igs [36]. In addition, a central signalling molecule phosphoinositide 3 (PI 3)-kinase has been reported to associate with nephrin [37]. These observations demonstrate how a complicated network of interactions and signalling is needed to organize and maintain the slit diaphragm. They also indicate how important the elucidation of the protein– protein interactions formed by nephrin are for the X. L. Liu et al. Intracellular domain of nephrin FEBS Journal 272 (2005) 228–243 ª 2004 FEBS 229 understanding of the function and regulation of the glomerular ultrafiltration barrier. Here, as a step towards this goal, we have studied the functions of the nephrin intracellular domain. Utilizing recombinant nephrin proteins, we analysed the requirements for the interactions with the known intracellular partners. Fur- thermore, we identified IQGAP1, an effector protein of small GTPases Rac1 and Cdc42, as a potential new interacting partner of nephrin. Nephrin and IQGAP1 were found to colocalize both in cultured cells and in kidney sections, but the demonstration of the biologi- cal significance requires further studies. Results The intracellular domain of mouse nephrin is com- posed of 156 amino acids, of which 116 are conserved in human and rat nephrin [38]. However, it is not possible to show any homology to other known proteins in databases. Nevertheless, there are several putative docking and phosphorylation sites in neph- rin (Table 1). Prediction of secondary structures was carried out with several programs that were able to recognize only two a-helical segments (http:// www.expasy.ch, http://xray.bmc.uu.se/sbnet/prosal.html). The first one, RRRLRRLAEE(1100–1110), follows immediately the transmembrane domain, and was detected by all programs employed, while the second, EEDRIRNEY(1120–1128), was suggested by most programs. Using these in silico analyses as a background, we proceeded to analyse the properties of the recombinant intracellular domain. For these studies, we produced and purified the intracellular domain without any tags, as well as a series of glutathione S -transferase (GST)- fusion proteins containing various parts of the intracel- lular domain. The GST-fusion proteins migrated as expected from their calculated molecular masses. However, in particular, the GST–C terminus and GST–nephrincyt were found to be partially degraded, indicating that the C-terminal half of the intracellular domain is sensitive to bacterial proteases (Fig. 1A). Western blot analysis with the antibody pAb 2 gener- ated against the intracellular domain of human neph- rin confirmed that all polypeptides detected in Fig. 1A originate from nephrin (results not shown). We first tested whether these fusion proteins were able to pull- down proteins found in previous studies to associate with nephrin. As demonstrated in Fig. 1B, both podo- cin and Fyn kinase readily associated with GST–C ter- minus and GST–nephrincyt. CD2AP did not interact with the fusion proteins in the initial experiments, and we therefore investigated its association with nephrin in various conditions. We were able to detect a weak, occasional interaction in the lysis buffer containing 1% (v ⁄ v) Chaps as a detergent, but not in 1% (v ⁄ v) digito- nin, 1% (v⁄ v) NP-40, or 0.5% (v ⁄ v) Triton X-100. Recently, we and others have reported that nephrin is tyrosine phosphorylated, probably by Fyn kinase in vivo [35,36]. Here we show direct phosphorylation by Fyn kinase in vitro. Fyn was able to phosphory- late the untagged full-length intracellular domain, as well as all GST–nephrin fusion proteins indicating that there are phosphorylation sites on both halves of the intracellular domain (Fig. 2A). Furthermore, several degradation fragments of GST–C terminus Table 1. Amino acid sequence analysis of nephrin. Sequences presented are of mouse nephrin. Only sites conserved in human, mouse and rat are mentioned. Putative site Sequence in nephrin Consensus sequence and ⁄ or reference Docking and phosphorylation site for Src kinase LYDEV(1207–1211) YE ⁄ D ⁄ TE ⁄ N ⁄ DI ⁄ V ⁄ M ⁄ L[42] Docking site for Nck adaptor proteins LYDEV(1207–1211) YDEP ⁄ D ⁄ V [42] Proline-rich sequence for SH3 domain binding PQLPP(1160–1164) PXXP [63] a-Helical region RRRLRRLAEE(1100–1109) http://www.expasy.ch EEDRIRNEY(1120–1128) http://xray.bmc.uu.se/sbnet/prosal.html Repeated sequences GHLYDEVE(1188–1195) http://www.ebi.ac.uk/Radar/ GPLYDEVQ(1205–1212) Protein kinase C phosphorylation site SEK(1112–1114) S ⁄ TXR ⁄ K SMR(1155–1157) http://c.expasy.org/tools/scanprosite/ Casein kinase II phosphorylation site STAE(1145–1148) S ⁄ TXD ⁄ E SMRD(1155–1158) http://c.expasy.org/tools/scanprosite/ TLEE(1166–1169) Most probable tyrosine phosphorylation sites Y1153, Y1208, Y1225, Y1232 http://www.cbs.dtu.dk/services/NetPhos/ Most probable serine or threonine phosphorylation sites S1119, S1142, S1145, S1155, S1160, S1171 http://www.cbs.dtu.dk/services/NetPhos/ Intracellular domain of nephrin X. L. Liu et al. 230 FEBS Journal 272 (2005) 228–243 ª 2004 FEBS were phosphorylated, whereas only the full-length GST-1173 was phosphorylated at a detectable level. Prolonged incubation or increased amount of Fyn kinase did not significantly increase phosphorylation suggesting that the in vitro phosphorylation was efficient and close to quantitative (Fig. 2B). Next, we wanted to investigate whether in vitro phosphorylation provides new properties for the nephrin intracellular domain. First, we examined whether the phosphoryl- ated GST–nephrin fusion proteins were able to bind phosphatidylinositol 3-kinase that has been reported recently to associate with nephrin [37]. We found that PI 3-kinase bound only to the phosphorylated GST– nephrincyt and GST-1173 implying that one of the SH2 domains in the p85 subunits of PI 3-kinase is involved in the interaction (Fig. 2C). These findings are also well in line with the prediction that the segment YYSM(1153–1156) constitutes a docking site for the SH2 domains of PI 3-kinase. Having demonstrated the functionality of the phos- phorylated GST–nephrin fusion proteins, we proceeded to investigate how the phosphorylation effects on the interaction of nephrin with Fyn and podocin. Interest- ingly, we observed that the tyrosine phosphorylation enhanced association of Fyn with nephrin, but did not have any significant effect on the interaction with podocin (Fig. 3). Identification of IQGAP1 as a putative interacting partner for nephrin The above described findings encouraged us to utilize GST–nephrincyt in a search for novel interactors of nephrin. Indeed, incubation of metabolically labelled podocyte, HEK293, HT1080, and L-cell lysates with the fusion protein conjugated to the glutathione-Seph- arose beads resulted in the strong and reproducible binding of a 190 kDa protein, as demonstrated by ana- lysing the eluates by SDS ⁄ PAGE and autoradiography (Fig. 4A). To identify the 190 kDa protein, it was purified in quantities sufficient for peptide mapping by MALDI- TOF-MS (Fig. 4B). Pooled fractions from three separ- ate purifications were concentrated, and loaded on a one-dimensional SDS ⁄ PAGE gel, which was then stained by silver. The bands containing the 190 kDa protein were excised, and the gel pieces treated as described in Experimental procedures. After overnight in-gel digestion with trypsin, the resulting peptide mixture was analysed by MALDITOF-MS. Fifty-seven peptide masses were determined, and used to search the database. Up to 30 masses corresponded with com- puted masses of tryptic peptides of mouse IQGAP1 (Table 2), a 189 kDa effector protein of small GTPases Rac1 and Cdc42. These peptides covered 19% of the mouse IQGAP1 sequence (GenBank accession number AF240630). These results allowed an unequivocal iden- tification of the 190 kDa protein as IQGAP1. In addi- tion, we sequenced three of the mapped peptides and two other peptides, which were not detected in the ori- ginal MALDITOF-MS analysis. All obtained peptide sequences could be positioned in the mouse IQGAP1 sequence, confirming the identification. Fig. 1. The C-terminal half of the nephrin intracellular domain is pro- tease-sensitive and interacts with Fyn and podocin. (A) GST fusion proteins containing various parts of the nephrin intracellular domain were produced and purified by standard methods in the presence of protease inhibitors. The purified proteins were analysed by SDS ⁄ PAGE and Coomassie Brilliant Blue-staining. (B) Cell lysates prepared from parental HEK293 cells, or from those stably expres- sing recombinant human podocin, were incubated with immobilized GST fusion proteins, after which equal amounts of bound proteins were separated on two SDS ⁄ PAGE gels. Gels were subjected either to Western blot analysis with anti-Fyn or anti-podocin Igs, or to Coomassie Brilliant Blue-staining to control the loading of the GST proteins to the beads. X. L. Liu et al. Intracellular domain of nephrin FEBS Journal 272 (2005) 228–243 ª 2004 FEBS 231 Characterization of the IQGAP1–nephrin interaction IQGAP1 was purified from L-cells rather than podo- cytes as L-cells proliferate more rapidly, and are there- fore more convenient for a large-scale application. In order to demonstrate that the 190 kDa protein bound to GST–nephrincyt from the podocyte cell lysate was IQGAP1, we carried out pull-down assays from these lysates. The eluates were analysed by Western blotting with anti-IQGAP1 Igs, and, as expected, a protein of 190 kDa recognized by the antibodies was found to bind to GST–nephrincyt (Fig. 5B). We also used var- ious deletion constructs to map the areas of the neph- rin intracellular domain responsible for IQGAP1 binding. All constructs containing the C-terminal half of the intracellular domain (amino acids 1167–1256) bound IQGAP1, whereas we could not detect any interaction between IQGAP1 and N-terminal half on the nephrin intracellular domain, or between IQGAP1 and GST alone. The last 11 residues of the intracellu- lar domain are fully conserved between human, mouse and rat nephrin. However, a fusion protein lacking this part did bind IQGAP1, indicating that the region was not needed for IQGAP1 binding. We were not able to convincingly demonstrate the nephrin – IQGAP1 inter- action in NPH5 cells by the immunoprecipitation method, as IQGAP1 tended to coimmunoprecipitate also with the Finn-minor mutant form of nephrin lack- ing nearly the whole intracellular domain. Tyrosine phosphorylation of nephrin did not appear to effect significantly on the in vitro interaction of GST–neph- rins and IQGAP1 (Fig. 5B). Nephrin and IQGAP1 colocalize in the HEK293 cell line expressing recombinant nephrin and IQGAP1 can be found at the slit diaphragm As the binding experiments described above were all in vitro studies, and the immunoprecipitation experi- ments did not give a definite demonstration of the interaction in intact cells, we performed immunolocali- zation studies to see whether the subcellular distribu- tion of nephrin and IQGAP1 supports the possibility that IQGAP1 is an intracellular binding partner of nephrin. First, formaldehyde-fixed NPH5 cells were double-labelled for nephrin and IQGAP1. IQGAP1 is known to localize at the cortical cytoskeleton and cell–cell adhesion sites [39,40]. The monoclonal anti- IQGAP1 Ig was found to stain the cortical cytoskele- ton in the NPH5 cells, and also diffusely the plasma membrane (Fig. 6B). The strongest nephrin staining was detected at the plasma membrane and the cortical cytoskeleton, and due to overexpression, nephrin could also be found in the cytoplasm (Fig. 6A). The expres- sion patterns of IQGAP1 and nephrin were overlap- ping (Fig. 6C). Our pull-down and Western blotting experiments indicated that IQGAP1 is expressed in podocytes at significant levels (Fig. 4A; X. L. Liu & P. Kilpela ¨ inen, unpublished observations). To examine the distribution of IQGAP1 in podocytes in detail, we carried out Fig. 2. The intracellular domain of nephrin is efficiently phosphorylated by Fyn kinase in vitro. Phosphorylation of a tyrosine residue within the N-terminal half of the domain generates a binding site for phosphoinositide 3-kinase. (A) After in vitro phosphorylation of the different nephrin proteins with a commercial Fyn kinase preparate,  0.5 lg of each protein was analysed by SDS ⁄ PAGE and Western blotting with anti-phosphotyrosine Igs. (B) The untagged full-length intracellular domain of nephrin was in vitro phosphorylated with the Fyn kinase prepa- rate for varying incubation times or by using different amounts of the kinase. The results demonstrate that already incubation with 1· amount of the kinase preparate for 90 min (conditions used in the panel A experiment) results in the efficient phosphorylation of the fusion protein. (C) Equal amounts of immobilized, nonphosphorylated and phosphorylated GST fusion proteins were incubated with HEK293 lysates, after which bound proteins were analysed by SDS ⁄ PAGE and Western blotting with a monoclonal anti-PI 3-kinase Ig. An identical gel was stained with Coomassie Brilliant Blue. A third gel with same samples was analysed by Western blotting with anti-phosphotyrosine Igs. Intracellular domain of nephrin X. L. Liu et al. 232 FEBS Journal 272 (2005) 228–243 ª 2004 FEBS immunoelectron microscopic studies of human kidney sections (Fig. 6D–F). In immunogold labelling, IQ- GAP1 was distributed in podocyte foot-processes exclusively intracellularly and the cytoplasmic aspect of the slit diaphragm or its vicinity was often notably labelled, whereas the label was rarely found near to other sites of the podocyte surface (Fig. 6D,E). In the cytoplasm the label was quite often found in densities (Fig. 6F) that could correspond to the endoplasmic reti- culum better resolved in well-fixed nonimmunosamples. Immunogold staining with the anti-nephrin Ig pAb 2 gave fairly abundant labelling that was located like the label for IQGAP1 but was concentrated to the slit diaphragm region and not found at other membrane sites (results not shown). Discussion The work that resulted in the identification of the neph- rin gene as a gene mutated in congenital nephrotic syn- drome of Finnish type (NPHS1) [2,41], demonstrated also the crucial role of the nephrin intracellular domain for the structure of the slit diaphragm. A number of mutations or deletions in the intracellular domain lead to severe NPHS1. Analysis of the amino acid sequence reveals that this domain is significantly conserved between different species [38], but it does not show any homology to other known proteins. It contains a few very typical consensus sequences found in the most proteins, such as two putative phosphorylation sites for protein kinase C and three for casein kinase II. However, a much more conspicuous and significant feature are the six conserved tyrosines found in human, mouse and rat [38], some of which (Tyr1191, Tyr1208 and Tyr1232, numbering according to mouse sequence) match with the consensus sequence of Src family kin- ase and Nck adaptor protein SH2 domain docking site [42]. These sites correspond also to Src family kinase phosphorylation sites. It is worth pointing out that the amino acid sequences around these tyrosines are indeed very similar (LYDEV, LYDEV, IYDQV), and these could be phosphorylated by the same kinase, for exam- ple by Fyn. Phosphorylation at multiple tyrosines pro- vides several docking sites, which could, for example, result in the assembly of large protein complexes involved in the organization and maintenance of the slit diaphragm. As the three putative docking sites are similar, one possibility is that such complexes could contain oligomers of same ligand that adjust or lock nephrin at the right location. Finally, as a curiosity, Tyr1128 is flanked by the most conserved sequence that can be found around the tyrosines in nephrin. However, this site is an unorthodox substrate for non- receptor tyrosine kinases as it has a negatively charged residue at position )1 before Tyr1128 (EYEES) [43]. On the contrary, several receptor tyrosine kinases have this kind of substrate sequences. As the nephrin intracellular domain did not show any homology to other known proteins, molecular modelling was difficult, and we could only predict the conformation for two short sequences that potentially form a -helixes. These were both located close to the N-terminus of the intracellular domain. When expres- sing and purifying the bacterial recombinant proteins, we noticed that the intracellular domain appears to consist of two clearly separate parts, the N-terminal Fig. 3. Tyrosine phosphorylation of nephrin enhances association with Fyn, but does not have an effect on the interaction with podo- cin. Cell lysates prepared from parental HEK293 cells, or from those stably expressing recombinant human podocin, were incuba- ted with equal amounts of nonphosphorylated and phosphorylated GST proteins immobilized to the glutathione-Sepharose beads. After extensive washings, bound proteins were analysed by SDS ⁄ PAGE and Western blotting with anti-Fyn or anti-podocin Igs. The anti-Fyn blot was reprobed with anti-nephrin Igs, and the podocin pull-down samples were subjected to SDS ⁄ PAGE and Coomassie Brilliant Blue-staining to show equal loading of the phosphorylated and nonphosphorylated fusion proteins to the beads. The eluates were also analysed by SDS ⁄ PAGE and Western blotting with the anti-phosphotyrosine Igs. X. L. Liu et al. Intracellular domain of nephrin FEBS Journal 272 (2005) 228–243 ª 2004 FEBS 233 half (amino acids 1102–1173 in mouse nephrin) and the C-terminal half (1167–1256). The N-terminal part was resistant to proteolysis during production, purifi- cation and storage implying that it may form a tightly packed domain. On the contrary, the C-terminal half was very sensitive to proteolysis occurring already dur- ing production in Escherichia coli BL21. It contains most of the putative consensus sites for tyrosine phos- phorylation and protein–protein interactions, and as we show in this study, all known interacting partners of the intracellular domain except PI 3-kinase, bind to this region. The C-terminal half may form a less tightly folded domain that has a flexible structure. This might be needed to create several interaction sites, or to render the area more accessible to the interacting proteins. Nephrin has been found to be associated and likely phosphorylated by Src family kinase Fyn in vivo [35,36]. In this study, we showed that Fyn kinase directly phosphorylates nephrin in vitro (Fig. 2A,B). Phosphorylation was efficient and apparently close to quantitative. Both N- and C-terminal halves were phosphorylated. Phosphorylation at the N-terminal part never reached the same intensity as that at the C-terminal half implying that the anti-phosphotyrosine Ig may detect phosphotyrosines in the C-terminus more efficiently, or that the C-terminal half may be more accessible to kinases, or simply that the C-ter- minal half contains more phosphorylation sites than the N-terminal half. The last possibility is the most likely one, as the three highly potent tyrosine phos- phorylation sites for Src family kinases, Tyr1191 Tyr1208 and Tyr1232, are all within this region. Fur- thermore, judging from the phosphorylation of degra- dation fragments, the C-terminal half contains at least two phosphorylation sites. Progress with the investigations of the nephrin intra- cellular domain have been relatively rapid, and have to date resulted in the identification of several putative interaction partners for nephrin [17,18,24,35–37]. The biological significance and function of some interactions is obvious, whereas those of others are more difficult to predict. Podocin is able to interact with nephrin, NEPH1, and CD2AP, all components of the slit dia- phragm [17,18,23]. It may function as a scaffolding pro- tein, and it augments nephrin signalling by facilitating recruitment of nephrin to the slit diaphragm area [17,21]. An adapter protein CD2AP has been reported to interact both with podocin and nephrin, and puta- tively connects the slit diaphragm to the actin cytoskele- ton [18,24,25]. Src family kinase Fyn phosphorylates nephrin, and associates with it [35,36]. This phosphory- lation may be critical for the integrity of the glomerular filtration barrier, as the Fyn kinase-knockout mice display a renal phenotype having structurally distorted or coarsened podocyte foot processes, and in some, Fig. 4. The intracellular domain of nephrin interacts with a 190-kDa protein. (A) 35 S- labeled extracts from podocytes, HT1080 cells, L-cells, and HEK293 cells were incu- bated with immobilized GST or the GST fusion protein containing the full-length nephrin intracellular domain. Bound proteins were resolved by SDS ⁄ PAGE, and analysed by autoradiography. (B) The 190-kDa protein was purified from the L-cell extract using GST–nephrin affinity chromatography. An aliquot corresponding to a 1 ⁄ 50 portion of the purified protein is analysed here by SDS ⁄ PAGE and silver staining. Three similar large-scale preparations were performed to obtain enough protein for MALDITOF-MS. Intracellular domain of nephrin X. L. Liu et al. 234 FEBS Journal 272 (2005) 228–243 ª 2004 FEBS podocytes foot processes are completely effaced [35,44]. PI 3-kinase has been found to be able to bind to nephrin via the p85 subunit, and in a cell culture model this binding is related to activation of antiapoptotic PI 3-kinase ⁄ AKT pathway [37]. However, further char- acterization of nephrin–ligand interactions is required, and especially their regulation is nearly completely unknown. In this study, we could detect binding of podocin, Fyn, and PI 3-kinase to GST–nephrin. We tested binding of CD2AP to GST–nephrin using four different detergent conditions and two cell lines in pull- down assays. Two different antibodies were employed to visualize CD2AP in Western blot analysis. Nonethe- less, we were able to detect CD2AP-binding only occa- sionally, and only in a lysis buffer containing 1% Chaps. Neither did in vitro phosphorylation of GST– nephrin induce CD2AP-binding. It is likely that neph- rin–CD2AP interaction is of very low stoichiometry also in vivo. It has not been detected in all previous studies [17], and there are two contradictory reports on which domains of CD2AP are involved in the nephrin-binding [27,28]. Consequently, it is possible that the biologically more critical interaction occurs between podocin and CD2AP, and it may be mainly the lack of this inter- action that contributes to the phenotype found in the CD2AP-knockout mice [24]. Surprisingly, more Fyn kinase bound to the phos- phorylated GST–nephrin than to the unphosphory- lated fusion protein. This may imply that nephrin is phosphorylated processively by Fyn. In this process, Fyn at first would phosphorylate a site in nephrin that becomes a high affinity binding site for the SH2 domain. Interaction between this site and the SH2 domain of Fyn facilitates phosphorylation of subse- quent tyrosines in nephrin – alternatively Fyn could phosphorylate other substrates that are forming a pro- tein complex with the intracellular domain of nephrin. The processive phosphorylation by Src family kinases has been demonstrated with several multiphosphory- lated substrates [45,46]. Interestingly, another important signalling molecule, PI 3-kinase bound only to the N-terminal half of the intracellular domain. The classical motif for the binding of PI 3-kinase SH2 domains is YXXM [42]. The match- ing sequence YYSM is conserved in mouse and rat (Tyr1153 in mouse), but is replaced in human nephrin by the sequence YYRSL. This segment is the most probable candidate for the PI 3-kinase binding site also as the phosphorylated tyrosine in GST-1173 is obviously located very close to the C-terminal end of the fusion protein. The PI 3-kinase binding site is apparently needed mainly for signalling purposes, and this same tyrosine may not be involved in the forma- tion of an intracellular protein complex around nephrin or in the regulation of contacts with the cytoskeleton. Tyrosine phosphorylation in the C-terminal half of nephrin has more likely this kind of functions. Having confirmed the functionality of GST–nephri- ncyt, we employed GST–nephrincyt affinity chromato- Table 2. Peptide masses obtained by MALDITOF-MS analysis after in-gel tryptic digestion of the 190-kDa protein identify IQGAP1 as an interacting partner for nephrin. Measured masses are obtained by MALDITOF-MS analysis after in-gel tryptic digestion of the 190- kDa protein. Data base searches with this set of peptide masses assigned a total of 30 peptides to mouse IQGAP1 corresponding to a sequence coverage of 19%. The peptide mass error was less than 100 p.p.m. with all except one peptide. The identification was confirmed by sequencing three peptides shown in bold and two other peptides not detected in the original MALDITOF-MS analysis. Their sequences, TLQALQIPAAK and LFQTALQEEIK, correspond, respectively, to residues 557–567 and 1025–1035 of the mouse IQ- GAP1 sequence. Measured mass Computed mass Sequence Residues in IQGAP1 Start To 725.453 725.386 LQYFR 829 833 735.453 735.373 IPYGMR 1156 1161 746.487 746.444 IQAFIR 842 847 769.434 769.401 MHQARK 818 823 800.439 800.429 ATGLHFR 105 111 800.439 800.425 DIRNQR 1488 1493 826.521 826.527 LIVDVIR 1391 1397 851.431 851.432 MVVSFNR 1054 1060 908.486 908.475 LGNFFSPK 81 88 934.464 934.512 EEYLLLR 1018 1024 934.464 934.454 QDKMTNAK 267 274 1040.576 1040.634 ALQSLALGLR 327 336 1117.543 1117.595 LIFQMPQNK 989 997 1117.543 1117.566 LDNSIRNMR 1131 1139 1123.602 1123.638 IIGNLLYYR 1186 1194 1140.508 1140.563 YGIQMPAFSK 192 201 1148.563 1148.585 TCLDNLASKGK 1533 1543 1258.676 1258.706 MREEVITLIR 892 901 1274.681 1274.701 MREEVITLIR 892 901 1318.665 1318.713 LFQTALQEEIK 1025 1035 1408.769 1408.851 RLAAVAAINAAIQK 388 401 1414.729 1414.781 LGLAPQIQDLYGK 162 174 1482.670 1482.666 ATFYGEQVDYYK 1517 1528 1531.698 1531.719 IFYPETTDIYDR 131 142 1564.613 1564.646 FDVPGDENAEMDAR 1369 1382 1566.757 1566.843 TLINAEDPPMIVVR 857 870 1715.841 1715.908 GVLLEIEDLQANQFK 1572 1586 1723.791 1723.837 EEIQSSISGVTAAYNR 723 738 1752.870 1752.911 IELEKYGIQMPAFSK 187 201 1752.870 1752.961 VDQIQEIVTGNPTVIK 1038 1053 1889.909 1889.948 FALGISAINEAVDSGDVGR 623 641 1953.882 1953.967 LPYDVTPEQALSHEEVK 1112 1128 2020.976 2020.998 NVIFEIGPTEEVGDFEVK 1587 1604 2054.108 2054.079 LEGVLAEVAQHYQDTLIR 568 585 X. L. Liu et al. Intracellular domain of nephrin FEBS Journal 272 (2005) 228–243 ª 2004 FEBS 235 graphy and peptide mass fingerprinting to search for new intracellular interaction partners for nephrin. This work resulted in the identification of IQGAP1, an effector protein of small GTPases Rac1 and Cdc42 and a putative regulator of cell–cell adheren junctions [39,40]. According to the GST pull-down assays, the interaction took place strictly and specifically between the C-terminal half of the nephrin intracellular domain and IQGAP1. The binding was not lost even after rel- atively harsh washing (four times in the lysis buffer containing 0.5 m NaCl), and IQGAP1 could be detec- ted in the pull-down samples also by silver staining of SDS ⁄ PAGE gels (Fig. 5C). Compared to the inter- actions with the previously identified ligands, that between nephrin and IQGAP1 appeared to be clearly the strongest and most specific one. However, despite the fact that these two proteins are according to our immunoelectron microscopic and immunocytochemical findings physically located very close to each other, immunoprecipitation could not demonstrate direct association between nephrin and IQGAP1 in HEK293 cells. This could imply that the nephrin–IQGAP1 interaction is transient, taking place only during certain situations, such as during the formation of the slit structure, and possibly even before nephrin is associated with other intracellular interactors. Although it is not clear whether the nephrin– IQGAP1 interaction is a biologically relevant one, study of the IQGAP1 literature strongly supports the notion that this could be the case. Rac1 and Cdc42 belong to the Rho subfamily of small GTPases that has been implicated in the regulation of a wide range of biological processes, including cell motility and adhesion, cytokinesis, cell morphology and polariza- tion, and cell growth [47]. They affect these processes mostly by controlling the reorganization of the actin cytoskeleton, but they also regulate signal transduction pathways affecting gene transcription in the nucleus. Rac proteins stimulate the formation of lamellipodia and membrane ruffles and are involved also in actin polymerization, cell–cell adhesion and cell motility, whereas Cdc42-activation triggers the formation of filopodia and affects cell–cell adhesion. IQGAP1 accu- mulates at the polarized leading edge and areas of Fig. 5. Characterization of the nephrin–IQGAP1 interaction. (A) Mouse podocyte lysates were incubated for 4 h with various immobilized GST proteins, after which bound proteins were analysed by SDS ⁄ PAGE and Western blotting with monoclonal anti-IQGAP1 Ig. The results show that the binding site for IQGAP1 resides within the C-terminal half of the nephrin intracellular domain. (B) L-cell lysates were incubated with equal amounts of nonphosphorylated and phosphorylated GST proteins immobilized to glutathione-Sepharose beads. The bound proteins were analysed, as in panel A, with monoclonal anti-IQGAP1 Ig. An identical SDS ⁄ PAGE gel was run to show equal loading of the phosphorylated and nonphosphorylated fusion proteins to the beads. The eluates were also analysed by SDS ⁄ PAGE and Western blotting with the anti-phosphotyrosine Igs. The results demonstrate that IQGAP1-binding to nephrin is not affected by the nephrin tyrosine phos- phorylation. (C) GST–nephrin immobilized to glutathione-Sepharose was incubated with the L-cell lysate, after which the beads were washed four times with the lysis buffer containing indicated concentrations of NaCl. Bound proteins were fractionated by SDS ⁄ PAGE and visualized by silver staining. A significant fraction of bound IQGAP1 remains associated even after washing with 0.5 M NaCl. Intracellular domain of nephrin X. L. Liu et al. 236 FEBS Journal 272 (2005) 228–243 ª 2004 FEBS membrane ruffling, as well as at the cell–cell adheren junctions [39,40,48]. It can bind to the activated GTP-bound Rac1 and Cdc42 maintaining them in an activated state, but not to the inactive GDP-bound forms [39,48]. Calmodulin is another signalling mole- cule IQGAP1 binds to [48], and it has also a well-documented actin-binding activity [49,50]. Thus IQGAP1 is a scaffolding protein connecting Ca 2+ ⁄ calmodulin and Rac1 ⁄ Cdc42-mediated signalling and cytoskeleton [51]. Very interesting is the recent observation that IQGAP1 interacts with CLIP-170, a member of a protein family that specifically accumulates at the plus ends of growing microtubules [52]. The CLIP- 170–IQGAP1 complex appears to function as a linker between the plus ends of microtubules and cortical actin meshwork. It may recruit the microtubules at special cortical sites leading to cell polarization. Acti- vated Rac1 or Cdc42, both known as key regulators of cell polarization [53], is present as a third member in the CLIP-170–IQGAP1 complex. Furthermore, IQGAP1 was reported recently to interact with S100B [54] that colocalizes with IQGAP1 at the polarized leading edge and areas of membrane ruf- fling in glioma cells. The strongest S100B immuno- reactivity was found in cells that are characterized by long processes. In these cells, a colocalization of S100B and IQGAP1 was evident at plasma mem- brane and in the growing processes. However, maybe the most notable issue is that IQGAP1 is involved in the formation and maintenance of cell–cell adheren junctions. It has been suggested to associate both with E-cadherin–catenin and nectin–afadin systems, and to regulate E-cadherin-mediated cell–cell adhe- sion [40,55]. IQGAP1 may, together with cell adhe- sion proteins, organize F-actin to form specific structures and morphology at the cell–cell adhesion sites in epithelial cells. It is possible that the neph- rin–IQGAP1 interaction plays a similar role in the slit diaphragm, which, in fact, has been suggested to be a modified adherens junctions, as a-, b- and c- catenins as well as P-cadherin have been found in Fig. 6. Nephrin and IQGAP1 colocalize in cultured cells. IQGAP1 is found at the slit diaphragm in human kidneys. HEK293 cells stably expressing human nephrin were double-stained for nephrin (A) and IQGAP1 (B). Panel C shows the merged picture, indicating that nephrin and IQGAP1 colocalize at the sites of cell–cell contacts. Some of the cells had already lost nephrin expression. (D–F) IQGAP1-immunogold label in podocyte foot-processes. FP, foot processes; SD, slit diaphragm; GBM, glomerular basement membrane; arrowheads, gold label. In D, shorter and longer stretches of slit diaphragm are seen in oblique tangential view in filtration slit between foot processes of human kidney podocytes. Note intracellular gold-label for IQGAP1 (arrowheads) mostly along SD. In higher magnification (E) gold label for IQGAP1 is seen close to cell surface both at the SD area and elsewhere. In F, IQGAP1-label deeper in the podocyte cytoplasm is associated with cytoplas- mic density at the level of cross-cut SD. X. L. Liu et al. Intracellular domain of nephrin FEBS Journal 272 (2005) 228–243 ª 2004 FEBS 237 [...]... mice [57] that do not have any obvious defects in kidney functions with protamine sulfate or anti -nephrin Igs The slit diaphragm is a highly specialized structure, the formation of which requires strictly targeted protein transport and localization These processes and the functional activity of nephrin are regulated via a plethora of protein protein interactions The function and regulation of nephrin. .. in the formation of the slit diaphragm and foot processes, as well as in the recovery after disruption of the filtration barrier [32,33] One possibility is that the nephrin IQGAP1 interaction plays a role mainly during the formation of the slit diaphragm and the specific actin structures, or in such situations as recovery processes after injury Therefore, it would be interesting to challenge the IQGAP1- knockout... recognition site provided the translation initiation codon ATG, thus adding one extra residue, a methionine residue, to the N-terminus of the recombinant protein To generate the constructs for GST nephrin fusion proteins, the cDNA encoding the fulllength cytoplasmic domain of mouse nephrin (amino acids 1102–1256) with BamHI and EcoRI sites at the ends was produced by PCR using the mouse nephrin cDNA [38] as... understood before the protein complex associated with its intracellular domain is described and characterized There are a number of questions waiting to be addressed The most important question is how this complex responds to different physiological or pathological states For example, do changes occur in the molecular composition of the complex? Or how are the regulation of the nephrin function and the rearrangement.. .Intracellular domain of nephrin podocytes close to the slit diaphragm [12] While revising this manuscript, nephrin was reported to form a multiprotein complex with cadherins and p120 in kidney glomeruli [56] which further supports a role for IQGAP1 as a biological ligand of nephrin It is apparent that the slit-diaphragm is connected to the actin cytoskeleton that has a well-documented... constructs The expression construct for the untagged full-length human nephrin intracellular domain (amino acids 1088– 1241) was generated by PCR using the human nephrin cDNA [2] as a template The upstream and downstream primers contained cutting sites for NdeI and XhoI, respectively The amplified fragment was ligated to the corresponding sites in the pET24a(+) vector (Novagen Inc., Madison, WI, USA) The NdeI... by sequencing with BigDye Terminator Cycle Sequencing Ready Reaction kit (PE Biosystems, Foster City, CA, USA) The expression construct encoding the full-length human podocin in the pcDNA3.1 vector has been described elsewhere [59] Experimental procedures Production and purification of recombinant bacterial proteins Antibodies The GST fusion proteins and the untagged nephrin intracellular domain were... described above They were precleared by 1-h end-to-end incubation with the glutathione-SepharoseTM 4B beads (Amersham Pharmacia Biotech AB) before use in GST pull-down assays Intracellular domain of nephrin one Sepharose beads were extensively washed with the lysis buffer containing 1% (v ⁄ v) Triton X-100 and 1.0 m NaCl Pull-down assay GST nephrin fusion proteins or GST alone bound to the glutathione-Sepharose... regulation of the nephrin function and the rearrangement of the actin cytoskeleton connected? If IQGAP1 is a biologically significant ligand for nephrin, it is possible that the cell adhesive properties of nephrin and ⁄ or its connections to the cytoskeleton are regulated in concert with cytoskeletal rearrangements X L Liu et al USA) Mouse monoclonal anti -IQGAP1 was from Zymed Laboratories Inc (San Francisco,... were mixed with 500–1000 lL of cell extracts, and incubated for 4 h with end-to-end mixing at 4 °C After washing with the lysis buffer, bound proteins were eluted by boiling in the SDS-sample buffer The samples were separated by SDS ⁄ PAGE under reducing conditions, and subjected to Western blot analysis When analyzing bound proteins from cells labelled with l-[35S]methionine ⁄ cysteine, the gel was . Characterization of the interactions of the nephrin intracellular domain Evidence that the scaffolding protein IQGAP1 associates with nephrin Xiao. blotting with monoclonal anti -IQGAP1 Ig. The results show that the binding site for IQGAP1 resides within the C-terminal half of the nephrin intracellular domain.