Interaction of the Type IIa Na/Pi Cotransporter with PDZ Proteins*

The type IIa Na+-dependent inorganic phosphate (Na/Pi) cotransporter is localized in the apical membrane of proximal tubular cells and is regulated by an endocytotic pathway. Because molecular processes such as apical sorting, internalization, or subsequent degradation might be assisted by associated proteins, a yeast two-hybrid screen against the C-terminal, cytosolic tail of type IIa cotransporter was designed. Most of the potential proteins found belonged to proteins with multiple PDZ modules and were either identical/related to PDZK1 or identical to NHERF-1. Yeast trap truncation assays confined the peptide-protein association to the C-terminal amino acid residues TRL of type IIa cotransporter and to single PDZ domains of each identified protein, respectively. The specificity of these interactions were confirmed in yeast by testing other apical localized transmembraneous proteins. Moreover, the type IIa protein was recovered in vitro by glutathioneS-transferase-fused PDZ proteins from isolated renal brush border membranes or from type IIa-expressing oocytes. Further, these PDZ proteins are immunohistochemically detected either in the microvilli or in the subapical compartment of proximal tubular cells. Our results suggest that the type IIa Na/Pi cotransporter interacts with various PDZ proteins that might be responsible for the apical sorting, parathyroid hormone controlled endocytosis or the lysosomal sorting of internalized type IIa cotransporter.

In kidney, reabsorption of filtered inorganic phosphate (P i ) takes place along the proximal tubules and is controlled by a variety of hormones (e.g. parathyroid hormone, PTH) 1 and other factors (e.g. dietary intake of P i ) (1,2). Three structurally unrelated sodium-dependent phosphate (Na/P i ) cotransporter families have been identified (1,3). By immunohistochemistry, it was apparent that members of the type I and the type IIa Na/P i cotransporters are located in the apical membrane of proximal tubular cells (4,5). Targeted inactivation of the type IIa Na/P i cotransporter gene (npt2) provided strong evidence that ϳ70% of Na-dependent P i transport across the brush border membrane is mediated by the type IIa Na/P i cotransporter (6). Furthermore, the type IIa cotransporter represents the major target for the many factors described to regulate proximal tubular P i reabsorption (2). Additionally, reduced proximal P i -reabsorption, as observed in X-linked hypophosphatemia, is due to a decreased abundance of the type IIa Na/P i cotransporter (7).
According to the current mechanistic view, inhibition of proximal tubular Pi-reabsorption, such as by PTH or by a diet of high P i content (acutely given), is achieved by a removal of type IIa cotransporters (2) from the apical membrane. Results obtained from in vivo (rats) and in vitro (OK cells) studies indicated that internalized type IIa Na/P i cotransporters are subjected to degradation in the lysosomes (8,9). Besides Na/P i cotransport, the activity of the brush border Na/H exchanger, NHE-3, is regulated by PTH as well; however, internalization of NHE-3 seems to occur after a delay and not immediately after binding of PTH to its receptor (10). This kinetic difference of PTH action on the type IIa protein and on the NHE-3 exchanger, respectively, points to a regulatory mechanism specific for the type IIa Na/P i cotransporter. Not much is known about the molecular reactions that underlie the internalization of the type IIa cotransporter. Although protein kinases are activated upon the binding of PTH to its receptor (2), the target(s) for activated kinases relevant for the internalization of the type IIa protein has(have) not yet been identified.
The microvillar localization of the type IIa protein and its physiologically controlled abundance in the apical membrane suggest specific interactions of the type IIa protein with other microvillar/subapical proteins. Such interactions may be necessary for the correct apical positioning or may be involved in the signaling pathway that leads to internalization. To identify such candidate proteins, a yeast two-hybrid screen was performed. As a bait, we used the C-terminal 75 amino acid residues (563-637) of the type IIa Na/P i cotransporter for the following reasons: (a) based on the current model of the secondary structure, the C terminus is located at the cytoplasmic surface (11); (b) by truncation studies, evidence was obtained that the C terminus is important for apical expression of the type IIa protein. 2 We found that the C terminus of the type IIa Na/P i cotransporter is a strong interactor to various PDZ proteins. Some of these PDZ proteins were localized in the brush border or in the subapical compartment of proximal tubular cells and thus may be of importance for the polar distribution and/or regulation of the type IIa Na/P i cotransporter.
To generate LexA DNA binding domain (DNA-BD) fusions, the fragments were inserted in-frame into the vector pBTM116 carrying the TRP1 selection marker (12). All cDNA fragments encoding for N termini were subcloned into the vector pFBL23 (13), where the inserts had reverted orientation relative to LexA. Site-directed mutagenesis (QuickChange, Stratagene) was applied to generate truncations (Ϫ3; Ϫ46; Ϫ56 and Ϫ67) of the C terminus from the type IIa Na/P i cotransporter by introducing stop codons at the respective positions. As a control bait, the subunit of the reverse transcriptase from human immunodeficiency virus (1RTp51) in pBTM116 was used (generously provided by Prof. U. Hü bscher; see Ref. 14).
All fragments were ligated to the C terminus of the GAL4 activation domain in vector pACT2, which delivers the LEU2 nutritional gene for complementation. All constructs were confirmed by sequencing using an upper primer GAL855 (5Ј-TGTTTAATACCACTACAATG-3Ј).
Yeast Transformation and Growth Selection-Bait vectors were transformed into Saccharomyces cerevisiae (strain L40) containing the genotype MATa trp1 leu2 his3 LYS2::lexA-HIS3 URA3::LexA-LacZ (15). Yeast cells were grown in YPDA medium (1% yeast extract, 2% Difco peptone, 2% glucose, 0.003% adenine hemisulfate) or the synthetic minimal Trp dropout medium (SD-Trp) complemented with 0.003% adenine hemisulfate (16). For transformation, a YPDA culture from L40 (20 ml) grown to an A 600 of ϳ1 (18 h) was centrifuged at 1200 ϫ g for 3 min, washed with 40 ml of distilled H 2 O, and resuspended in 1 ml of distilled H 2 O. Aliquots of 50 l were treated with 240 l of polyethylene glycol (50% w/v), 36 l of 1 M LiAc, 25 l of single-stranded DNA (2.0 mg/ml), 50 l of distilled H 2 O, and 1 g of plasmid, according to Gietz and Schiestl (17). Proper expression of the baits was confirmed by Western blotting of total cell lysates (18) with antibodies raised against the C-and N-terminal part of type IIa Na/P i cotransporter (5) or against LexA. In control experiments (data not shown), an autonomous activation of the L40 reporter genes by the baits was verified.
Library Screening-A cDNA library of whole adult mouse kidneys (MATCHMAKER, CLONTECH) was used. A 100-ml culture in SD-Trp medium was set up with four single colonies, grown to A 546 Ͼ 2 (ϳ 36 h) at 30°C (260 rpm) and used to inoculate 1 liter of YPDA to 0.2 A 600 . The culture was propagated to ϳ0.8 A 600 , splitted up in quarters and washed with a total of 400 ml of sterile distilled H 2 O after centrifugation at 4200 ϫ g (4°C) for 15 min. The pellets were resuspended in a total of 160 ml of distilled H 2 O, transferred to 50-ml tubes, and repelleted at 1300 ϫ g for 10 min at 4°C. Each pellet was transfected separately according to Gietz and Schiestl (17), using a reaction mixture with a scaling-up of 50 times from above and with 50 l of cDNA (1 g/l). After an incubation at 30°C followed by a heat shock at 42°C (30 min at 220 rpm), cells were harvested at 1900 ϫ g for 3 min, washed with 80 ml of distilled H 2 O, recombined in 20 ml of distilled H 2 O and plated onto 100 14-cm Petri dishes. After 5 days, 4.4 ϫ 10 6 Trp/Leu/His prototrophies were tested for LacZ expression by a colony-lift assay (19,20), where the permeabilized cells transferred onto Whatman filters were overlaid with 0.2 mg/ml 5-bromo-4-chloro-3-indolyl ␤-D-galactopyranoside (Alexis), 150 mM NaCl, 50 mM Tris-HCl (pH 7.4), and 0.8% agarose. Putative positive colonies were restreaked on selective medium and rescreened by colony-lift assay. Yeast DNA of true positives was isolated (21), and prey plasmids were rescued by transformation into KC8 cells, which carries trpC, leuB, and hisB mutations (22). Protein-protein interactions of purified prey plasmids were reconfirmed in yeast. Insert sizes were checked by BglII digestion and subjected to dideoxy sequencing (Microsynth, Switzerland). Identical sequences were grouped via ClustalW at Pôle Bio-Informatique (Lyonnais) or via Pileup from Genetics Computer Group (Oxford) and overlaps connected (Contig assembly program; Baylor College of Medicine). Searches for protein relationships were performed at the National Center for Biotechnology Information at the National Institutes of Health (Bethesda, MD) using BLAST (23). The modular architecture of proteins was determined by SMART (24).
Lysates were thawed at 37°C and spun for 3 min as above to remove insoluble debris. Equal amounts of GST fusion proteins (ϳ2 g) were incubated with 25 l of preequilibrated glutathione-agarose beads (Sigma; 50% slurry) in a total volume of 500 l of binding buffer (50 mM Tris-HCl, pH 8, 120 mM NaCl, 0.5% Igepal CA-630, 5 mM dithiothreitol) by rocking at 4°C for 30 min. After absorption, beads were collected by brief centrifugation at 12,000 rpm for 10 s (4°C) and gently washed three times with 500 l of binding buffer containing 0.075% SDS.
Pull-down Experiments-Pull-down experiments were performed either with isolated proximal tubular brush border membranes (BBMV) from mice or with Xenopus laevis oocytes injected with type IIa cRNA. Purified BBMVs (27) were solubilized in binding buffer (see above) for 5 min at 4°C and centrifuged at 16,000 ϫ g for 3 min. Extracts of oocytes injected with IIa cRNA (for details, see Ref. 28) were obtained as described (29). GST fusion protein-loaded beads were incubated for 1 h at 4°C with solubilizates corresponding to 0.05 mg of BBMV protein or 1-2 oocytes, washed three times with binding buffer and used for gel-electrophoresis. Western blotting was performed with a polyclonal anti-IIa cotransporter antibody directed against the N terminus (5), and immunoreactive bands were detected by ECL using a secondary HRPcoupled IgG (Amersham Pharmacia Biotech).
Northern Blotting-Tissue distribution of mRNA expression of identified gene products was studied by Northern blotting using poly(A) ϩ RNAs of adult mice either purchased from CLONTECH or isolated by standard procedures. Full-length inserts were randomly labeled with [ 32 P]dCTP (oligolabeling kit; Amersham Pharmacia Biotech) and used as probes. After hybridization, all blots were washed sequentially with 2ϫ SSC, 1ϫ SSC, and 0.5ϫ SSC (containing 0.1% SDS) at temperatures up to 55°C.
Immunofluorescence and Immunogold Electron microscopy-Immunohistochemical distribution of NaPi-Cap1 and NaPi-Cap2 in mouse kidney was essentially performed as described (30). Polyclonal antisera were raised against synthetic peptides derived from the N termini of mouse NaPi-Cap1 and mouse NaPi-Cap2, respectively. ␤-Actin was visualized by phalloidin-Texas Red. Anti MAST205 antibody was kindly provided by P. Walden (31), and an anti-HSP86 antibody was purchased from Affinity Purified Reagents.

Identification of Type IIa Na/P i Cotransporter-associated
Proteins-An adult mouse kidney cDNA library was screened in yeast against the C-terminal tail (aa 563-637) of the murine type IIa Na/P i cotransporter. Initially, based on selection for growth on synthetic media and on LacZ expression, 138 positive colonies were obtained from 4.4 ϫ 10 6 Leu/Trp/His prototrophs. To confirm the initial positives, plasmids from 104 clones were isolated and used for retransformation of L40 cells harboring the following bait constructs: (a) the C terminus or (b) the N terminus of the type IIa Na/P i cotransporter; and, as controls, (c) a subunit of the reverse transcriptase from human immunodeficiency virus (1RTp51) or (d) the empty vector pBTM116. 93 prey plasmids reappeared blue when cotransformed with the original C-terminal bait, but did not show any interactions with the other baits used; notably no interaction occurred with the N terminus of the IIa Na/P i cotransporter which has been predicted to be located intracellularly as well (11). Inserts (on average around 2 kilobase pairs) from all clones were partially sequenced, grouped according to their alignments, and compared with GenBank entries (Table I).
As listed, most of the clones (71 out of 93) coded for PDZ proteins. The majority of them were identical or closely related to the formerly identified rat protein diphor-1 or its human homologue PDZK1 (32)(33)(34) and thereafter were referred as mouse NaPi-Cap1 and mouse NaPi-Cap2, respectively. Other PDZ proteins identified were identical/similar to NHERF-1 (35,36), NHERF-2 (37), and a protein named C2PA composed of a C2 as well as a PDZ domain (accession no. AJ250999). Besides, several clones were identical to the heat shock proteins HSP84 and HSP86 (38) or to MAST205, a microtubulin-associated serine/threonine kinase (31). Others were found only once, such as the mouse protein FHL-2 having four and half LIM domains (39) or as the mouse homologues of rat zetin 1 (accession no. AF245225) and of the human putative transmembrane protein 54TMp (accession no. AF004876). Differences in the representation of these clones could be explained by either the low abundance of corresponding mRNAs or by the presence of diverse truncations of inserts existing in the library.
Specific Binding of Putative Type IIa-associated Proteins in Yeast-To evaluate the specificity of the identified clones for binding the C terminus of the type IIa cotransporter, various baits derived from a number of proximal tubular, apically localized membrane proteins were constructed and tested in yeast for possible interactions with some of the proteins obtained by the screen (Table II). Most of the baits used did not activate the reporter genes in the presence of the different proteins exhibiting binding capacity for the type IIa cotransporter. Exceptional was the C terminus of the type I Na/P i cotransporter (40), which was found to associate with all the PDZ proteins listed as well as weakly with MAST205. Furthermore, the C-terminal tail of the Na/H exchanger, NHE-3, bound NHERF-1, as shown previously (35,41,42), and additionally NaPi-Cap1, but not NaPi-Cap2.
GST Precipitations for in Vitro Corroboration of Interactions-Some of the interactions found by the yeast two-hybrid assay were verified by pull-down experiments using GST fusion constructs and either solubilized mouse kidney BBMV (Fig. 1A) or lysates of X. laevis oocytes expressing the mouse type IIa Na/P i cotransporter (Fig. 1B). From solubilized BBMV, the mature type IIa protein (mass ϳ 85 kDa; see Ref. 5) was precipitated by the following GST fusion constructs (Fig. 1A): GST-NaPi-Cap1, GST fused to the PDZ domain 3 of NaPi-Cap1 (see below), GST-NaPi-Cap2, GST-NHERF-1, GST-NHERF-2, and GST fused to the protein similar to NHERF-2. The type IIa cotransporter was neither pulled down by GST alone nor by GST fused to C2PA, zetin 1, MAST205 (Fig. 1) and to FHL-2, HSP84, or 54TMp (data not shown). Additionally, from lysates of X. laevis oocytes, the type IIa cotransporter was pulled down by GST-MAST205 and to a weak extent by GST-C2PA. Again, GST-zetin 1 was unable to recover the cotransporter (Fig. 1B). As NaPi-Cap1 is localized in the microvilli (see below), GST fused to the C terminus of the type IIa cotransporter was used to precipitate NaPi-Cap1 from solubilized BBMVs. However, no precipitation of NaPi-Cap1 was achieved (data not shown). This could be explained by the fact that NaPi-Cap1 contains four PDZ domains, which may be linked to cytoskeletal or membrane proteins (43,44) and thereby may prevent the pulldown of NaPi-Cap1. Tissue Distribution and Immunolocalization of Mouse NaPi-Cap1 and NaPi-Cap2-In our further studies, we concentrated on the two proteins NaPi-Cap1 and NaPi-Cap2, since fulllength sequences were obtained for both proteins and other proteins, such as the NHERFs, HSPs, and MAST205, have been partially characterized previously (31,35,36,38,45,46). Besides, by immunofluorescence, the heat shock proteins and MAST205 were found in all cells along the nephron and were present in the cytoplasm (data not shown). Therefore, HSPs and MAST205 were assumed not to be implicated in brush border localization and/or regulation of the type IIa Na/P i cotransporter.
As illustrated in Fig. 2, NaPi-Cap1 mRNA (around 2.5 kilobase pairs) was detected in liver, kidney, small intestine (two transcripts) and testis (one transcript), but was not detected in other organs, such as heart, brain, and lung. NaPi-Cap2 mRNA was expressed only in kidney and intestine and was absent in all other organs looked for. The renal localization of NaPi-Cap1 and NaPi-Cap2 was assessed by immunohistochemistry using polyclonal sera raised against synthetic peptides derived from the corresponding N termini. Specificity of these antisera was tested on immunoblots using the GST fusion constructs (results not shown). In mouse kidney sections, both NaPi-Cap1 and NaPi-Cap2 were detected exclusively in proximal tubules. Immunostaining was absent in other nephron segments or in glomeruli and was not observed in blood vessels or in interstitial cells (Fig. 3A and data not shown). Immunostaining was uniform along the entire proximal tubules (S1, S2, and S3 segments) and was indistinguishable between the cortical and juxtamedullary nephrons. Proximal tubular location of NaPi-Cap1 was also confirmed by in situ hybridization (results not shown).
In proximal tubular cells, both NaPi-Cap1 and NaPi-Cap2 were immunodetected at the apical side. Whereas NaPi-Cap1 TABLE II Binding specificity of identified proteins toward other apical brush border proteins of proximal tubular cells Putative type IIa cotransporter-associated proteins and anticipated cytosolic C or N termini of various apically localized proximal tubular proteins were sequentially transformed in yeast. Colony-lifts were used to determine the extent of ␤-galactosidase expression: intensity of color is represented by (ϩϩϩϩ) for very strong, (ϩϩϩ) for strong, (ϩϩ) for weak, (ϩ) for very weak and (Ϫ) for none. In addition, the following C-and N-terminal baits were used: sodium/glucose cotransporter SGLT-1, peptide transporter Pept-2, calcium receptor CaR, glycoprotein-associated amino acid transporter b o,ϩ AT, and sodium/sulfate cotransporter NaSi-1 (for further descriptions of these proteins, see "Experimental Procedures"). All these baits as well as the controls (the DNA-BD alone, the viral protein p51, and the basolaterally expressed sulfate/oxalate/bicarbonate anion exchanger, Sat-1) did not activate the reporter gene in the presence of the prey constructs listed.

Bait
NaPi-Cap1 NaPi-Cap2 a CT signifies the putative cytosolic C terminus of the protein. b NT signifies the putative cytosolic N terminus of the protein.

FIG. 1. Type IIa cotransporter is recovered in vitro by various proteins identified by the two-hybrid approach.
Purified GST alone or GST fusion proteins (see "Experimental Procedures") immobilized on glutathione-agarose were incubated in the presence of lysates from murine kidneys BBMVs (A) or from X. laevis oocytes expressing the mouse IIa Na/P i cotransporter (B). After intense washing, the final samples were denatured in the absence of a reducing agent, and transferred material was immunoblotted using a polyclonal antibody raised against the type IIa Na/P i cotransporter. The band of ϳ85 kDa represents the mature Na/P i cotransporter (see also Ref. 5). As controls for the stability of the type IIa protein, aliquots of the lysates were analyzed as well.

FIG. 2. Expression pattern of mouse NaPi-Cap1 and NaPi-Cap2
mRNA. Poly(A) ϩ RNA from multiple mice tissues was probed with randomly labeled inserts of NaPi-Cap1 (A) and NaPi-Cap2 (B). The multiple tissue blot was additionally probed for ␤-actin (C). was strictly associated with the microvilli, NaPi-Cap2 was predominantly located in the subapical compartment, but was not detected in the microvilli. However, faint immunostaining for NaPi-Cap2 was also found throughout the cytoplasm (Fig.  3A). By immunogold electron microscopy, evidence was obtained that NaPi-Cap2 was associated with vesicular structures within the subapical compartment. Again, microvillar localization was observed only for NaPi-Cap1 (Fig. 3B).
Yeast Trap Truncation Assays to Specify the Interaction of Type IIa Cotransporter with Mouse NaPi-Cap1, NaPi-Cap2, and NHERF-1-Both NaPi-Cap1 and NaPi-Cap2 encompass four PDZ domains. To analyze which of the four PDZ domains of NaPi-Cap1 or NaPi-Cap2, respectively, was responsible for the interaction with the C terminus of the type IIa Na/P i cotransporter, each of the four PDZ domains was tested separately by trap two-hybrid assays. As depicted in Fig. 4A, solely PDZ domain 3 of NaPi-Cap1 and NaPi-Cap2 increased significantly the activity of ␤-galactosidase, in accordance with GST-NaPi-Cap1/PDZ3 by which the type IIa cotransporter was pulled down (see Fig. 1). The PDZ domain 3 of both NaPi-Caps was also found to be mostly important for the interaction with the C terminus of the type I Na/P i cotransporter (data not shown). Furthermore, our results indicated that the type IIa cotransporter via its C terminus interacts predominantly with the first PDZ domain of NHERF-1 (Fig. 4B).
The last three C-terminal amino acid residues of the type IIa cotransporter sequence are TRL and thus may represent a PDZ-binding cassette (41). To refine if this putative PDZ binding determinant was responsible for the interaction with identified PDZ proteins, the last three amino acid residues (TRL) of the type IIa cotransporter were deleted and the respective bait was assayed in yeast in the presence of either NaPi-Cap1, NaPi-Cap2 or additionally NHERF-1 and C2PA. As illustrated in Fig. 5A, the removal of the three C-terminal amino acid residues, TRL, completely abolished the interaction of the Cterminal tail of the IIa Na/P i cotransporter with NaPi-Cap1, NaPi-Cap2, NHERF-1, and C2PA. Similarly, the last three FIG. 3. Localization of NaPi-Cap1 and NaPi-Cap2 in mice proximal tubules. A, cryosections were stained with polyclonal antibodies raised against NaPi-Cap1 and NaPi-Cap2 (upper row) or with phalloidin red to localize ␤-actin (middle row). Merged pictures are shown in color. Colocalization of NaPi-Cap1 with microvillar ␤-actin is indicated in yellow. Bar size, 50 m. B, immunogold electron microscopy. As indicated by arrowheads, NaPi-Cap1 was associated with microvilli whereas NaPi-Cap2 was detected as attached to vesicular structures in the subapical compartment. Eventually, NaPi-Cap1 was also present in the intermicrovillar clefts (open arrowheads).

FIG. 4. Single PDZ domain interactions of mouse proteins
NaPi-Cap1, NaPi-Cap2, and NHERF-1 with the C terminus of the mouse type IIa Na/P i cotransporter in yeast. Yeast cells containing the C terminus as a bait were transformed with the indicated prey constructs, and interactions were analyzed by liquid ␤-galactosidase assays with permeabilized liquid cultures of total yeast lysates. All prey constructs were also tested for possible LacZ activation in yeast harboring either the DNA-BD from the empty vector (pBTM116) or the control bait protein p51, as shown in B but not in A. All these controls were negative, indicating specific interactions of the C terminus with the PDZ preys listed. amino acid residues of the type I Na/P i cotransporter are represented by TRL and their deletion also blunted the interaction with the PDZ proteins mentioned above (data not shown). On the other hand, the C terminus, as a ligand for the heat shock proteins HSP84 or HSP86, did not depend on the last three amino acid residues because complex formation was disrupted not before 46 amino acid residues were removed (Fig. 5B). DISCUSSION Proximal tubular apical location of the type IIa Na/P i cotransporter and its mode of regulation (endocytosis, lysosomal sorting in the subapical region) (2) suggest a complex interaction pattern of the type IIa Na/P i cotransporter with microvillar proteins as well as with proteins localized in the subapical compartment. In microvilli, such interacting proteins may play a role in scaffolding and also may anchor components involved in the signaling pathways. In the subapical compartment, type IIa cotransporter-associated proteins can be envisaged to play not only a role in apical sorting of newly synthesized cotransporters, but also to be important for the initial steps leading to the routing of internalized cotransporters to the lysosomes. A priori, every domain of the type IIa cotransporter oriented toward the cytoplasm could represent a possible site for such interactions. Predictions of the secondary structure and studies on the topology of the type IIa cotransporter suggested 8 -10 transmembrane segments and 5 intracellular regions including the N and the C termini (11). In this study, we have used the whole C terminus of the type IIa cotransporter as a bait to screen a cDNA library of adult mouse kidneys by the yeast two-hybrid strategy.
Most of the proteins identified (71 out of 93) represented PDZ proteins (see Table I). Two similar proteins with four modular PDZ repeats were identified: NaPi-Cap1, being identical to the previously described proteins diphor-1 and PDZK1 (3,32,33), and NaPi-Cap2, being 28% identical to NaPi-Cap1. Other PDZ proteins found belonged either to the NHERF family (NHERF-1, NHERF-2, and a protein similar to NHERF-2) (35)(36)(37) or were related to the C2PA protein, which harbors a PDZ domain and a C2 domain in addition. Further, two large groups of proteins comprised the heat shock proteins HSP84 and HSP86 (38,45) and the microtubulin-associated serine/threonine kinase MAST205 (31). Interaction of above proteins with the C terminus of the type IIa cotransporter was specific (with one exception) since both the N terminus of the type IIa cotransporter and the C or N termini of various other proximal tubular, apically localized membrane proteins failed to stimulate LacZ expression in yeast. Interestingly, all identified PDZ proteins bound also to the C terminus of the type I Na/P i cotransporter, which has been described to be localized in the apical membrane as well (4). However, if such an interaction with the type I cotransporter reflects any physiological significance, it remains to be shown if the C terminus of the type I Na/P i cotransporter is faced toward the cytoplasm and not to the extracellular space.
Additional evidence for an interaction of the type IIa cotransporter with identified proteins was obtained from pull-down experiments, demonstrating that at least all PDZ proteins (NaPi-Cap1/2, C2PA, and NHERF isoforms) were (in vitro) binding partners for the mature type IIa Na/P i cotransporter as present in isolated proximal tubular brush border membranes or/and in oocytes of X. laevis injected with type IIa cRNA.
A first mandatory step to establish the physiological relevance of positive interactions in yeast was to determine the location of the putative candidate proteins in renal tissue, i.e. such proteins were required to be expressed in microvilli and/or subapical compartment of proximal tubular cells. The proteins NaPi-Cap1 and NaPi-Cap2 showed strict and NHERF-1 mostly proximal tubular location. NaPi-Cap1 (see also Ref. 44) and NHERF-1 (47) were found exclusively in microvilli, whereas NaPi-Cap2 was distributed predominantly in the subapical compartment of proximal cells. Interestingly, as revealed by immunogold electron microscopy, NaPi-Cap2 appeared to be associated with vesicular structures in the subapical compartment. At present, the nature of such vesicles is not known and vesicular proteins eventually linked to NaPi-Cap2 remain to be determined.
Although the type IIa cotransporter was precipitated by both NHERF-2 and its likely isoform, a conclusion about this kind of association would be premature since the renal localization of NHERF-1 isoforms have not yet been verified by morphological techniques. With respect to C2PA, preliminary immunohistochemical studies indicated that this protein is barely localized in the proximal tubules (data not shown). Furthermore, immunofluorescence data disclosed that the heat shock proteins HSP84/86 and the microtubulin-associated serine/threonine kinase (MAST205) were localized along the whole nephron and were neither specifically localized in the brush border nor in the subapical region (data not shown). Therefore, these proteins are probably not implicated in either apical sorting or endocytosis of the type IIa cotransporter. It can, however, not be excluded if HSP84/86 or MAST205 plays a role in protein synthesis or lysosomal routing of the type IIa cotransporter; the latter has been shown previously to be dependent on an intact microtubular network (48). Further studies on other proteins which have emerged from the two hybrid approach were impossible, due to the lack of suitable antibodies. Based FIG. 5. Implication of the C-terminal amino acid motif (TRL) of the type IIa Na/P i cotransporter for interaction with identified proteins. Yeast cells, transformed with the C terminus or with truncated C termini (CT-3, CT-46, or CT-56), were retransformed with various PDZ proteins (A) or the heat shock proteins HSP84/86 (B), as depicted. LacZ activation (␤-galactosidase) was determined in total lysates of yeast. None of the prey constructs activated autonomously the reporter genes when transformed in yeast containing either the DNA-BD alone (empty vector) or the viral protein p51 as controls (results not shown). on the criteria discussed above (positive yeast assay, pull-down and immunolocalization), NaPi-Cap1/2 and NHERF-1 were supposed to associate with the type IIa cotransporter. But it remains to be shown if all these interactions also occur under in vivo conditions. PDZ domains are stretches of 80 -90 amino acids and have been assigned to an increasing number of proteins having physical organizer functions in the formation of protein complexes of various sizes. Such protein complexes may allow a proper arrangement of components involved in signal transduction pathways and/or may allow a correct recruitment of membrane-inserted proteins, as has been shown e.g. for synaptosomal membrane transporters and channels (41,42,49). In addition, evidence for PDZ interactions in the localization of proteins to the apical membrane has been accumulated (50,51). The proteins NaPi-Cap1, NaPi-Cap2, and NHERF-1, on which our studies concentrated, contain two or four PDZ domains. However, their interactions with the C terminus of the type IIa cotransporter were confined to only one of these domains, i.e. PDZ3 of NaPi-Cap1/2 and PDZ1 of NHERF-1. As shown by others, NaPi-Cap1 also interacts with a small transmembrane, cancer-associated protein, MAP17 (43), and with the multidrug resistance-associated protein, MRP2 (44). The latter two proteins are supposed to interact with PDZ domains 1 and/or 4 of NaPi-Cap1. Presumably, other PDZ domains of NaPi-Cap1 and NaPi-Cap2 may be occupied by cytoskeletal proteins or by members of the ERM family of actin-binding proteins (52). As for NHERF-1, it was initially identified as a protein which binds to the Na/H exchanger NHE-3 (35,36), but interaction of NHERF-1 with other proteins, such as the ␤ 2adrenergic receptor (53), the cystic fibrosis transmembrane conductance regulator (50,54), and the B1 subunit of H-ATPase (47), has been reported as well.
Often, PDZ folds recognize a C-terminal amino acid motif of the D(S/T)XV type; however, many variations thereof appeared to anchor PDZ domains as well (41,42,49). The last three C-terminal amino acid residues of the type IIa cotransporter, TRL, resemble a PDZ binding motif. Indeed, deletion of the amino acid residues TRL is sufficient to completely abrogate the interaction with the PDZ proteins NaPi-Cap1, Napi-Cap2, and NHERF-1.
In summary, a yeast two-hybrid screen with the C terminus of the type IIa Na/P i cotransporter as a bait revealed several proteins that are likely to be associated with the IIa cotransporter. Physiological relevance of the identified proteins is met by their proximal tubular as well as apical localization. The proteins NaPi-Cap1, NaPi-Cap2, and NHERF-1 are of particular interest and may mediate apical positioning and/or endocytosis of the type IIa Na/P i cotransporter. These multidomain PDZ proteins are presumably cosequestered by type IIa cotransporters at the membrane and are therefore assumed to build up a complex network consisting of the type IIa Na/P i cotransporter, other membrane proteins, cytoskeletal proteins, or components of signaling pathways (33,43). It seems of interest that NHERF-1 recruits ezrin, which in turn functions as a protein kinase A anchoring site for conferring cAMPmediated regulation of NHE-3 (41,46,55). Thus, NHERF-1 seems to serve as an organizer for signaling complexes necessary for the regulation of a variety of proteins, e.g. permitting phosphorylation of NHE-3 (35). It must be determined if NHERF-1 exerts a similar function in the regulation of the type II cotransporter.
The observation that the C terminus of the type IIa cotransporter recognizes PDZ motifs of several proteins (NaPi-Cap1/2 and NHERF-1) suggests that these proteins may differently affect apical sorting and/or endocytosis. Thus, in the case of the type IIa Na/P i cotransporter, such interactions may not be of static nature but rather may be dynamically regulated by the activation status of specific signaling pathways.