In expression systems and in yeast, Na/H exchanger regulatory factor (NHERF)-1 and NHERF-2 have been demonstrated to interact with the renal brush border membrane proteins NHE3 and Npt2. In renal tissue of mice, however, NHERF-1 is required for cAMP regulation of NHE3 and for the apical targeting of Npt2 despite the presence of NHERF-2, suggesting another order of specificity. The present studies examine the subcellular location of NHERF-1 and NHERF-2 and their interactions with target proteins including NHE3, Npt2, and ezrin. The wild-type mouse proximal tubule expresses both NHERF-1 and NHERF-2 in a distinct pattern. NHERF-1 is strongly expressed in microvilli in association with NHE3, Npt2, and ezrin. Although NHERF-2 can be detected weakly in the microvilli, it is expressed predominantly at the base of the microvilli in the vesicle-rich domain. NHERF-2 appears to associate directly with ezrin and NHE3 but not Npt2. NHERF-1 is involved in the apical expression of Npt2 and the presence of other Npt2-binding proteins does not compensate totally for the absence of NHERF-1 in NHERF-1-null mice. Although NHERF-1 links NHE3 to the actin cytoskeleton through ezrin, the absence of NHERF-1 does not result in a generalized disruption of the architecture of the cell. Thus the mistargeting of Npt2 seen in NHERF-1-null mice likely represents a specific disruption of pathways mediated by NHERF-1 to achieve targeting of Npt2. These findings suggest that the organized subcellular distribution of the NHERF isoforms may play a role in the specific interactions mediating physiological control of transporter function.
- PDZ domains
the nherf family of proteins has been found to bind to over 30 proteins, including specific transporters and ion channels, receptors, signaling proteins, structural proteins, and transcription factors (23, 24, 27). Na/H exchanger regulatory factor (NHERF)-1 and NHERF-2 (also known as E3KARP) are distinct gene products, and although there is considerable overlap in the binding targets of these proteins, recent experiments have identified some specific interactions unique to the individual isoforms. For example, NHERF-2 but not NHERF-1 is involved in calcium-mediated internalization of NHE3 (12). Nonetheless, it remains unknown at this time whether NHERF-1 and NHERF-2 represent a redundant regulatory system, a cooperative system whereby both proteins are required for normal regulation of cell function, or a system where each isoform carries out distinct functions independent of the other. Evidence has been advanced to indicate that both NHERF-1 and NHERF-2 bind to the renal proximal tubule transporter NHE3 and, when expressed in PS120 cell fibroblasts, both can facilitate cAMP-mediated inhibition of NHE3 activity (13, 30, 31). Yeast 2-hybrid screening indicates that NHERF-1 and NHERF-2 also interact with Npt2 (NaPi IIa), the major regulated renal brush border membrane (BBM) sodium phosphate cotransporter (8). Recent evidence in native kidney tissue, however, indicates that NHERF-1 is uniquely required for cAMP regulation of NHE3 and for apical membrane targeting of Npt2 (22, 28). These results would not be consistent with NHERF-1 and NHERF-2 having redundant roles with respect to these two transporters.
The specificity of binding to PDZ sites usually resides in the COOH-terminal amino acids of the target proteins (21). Despite the similarities in the PDZ domains of NHERF-1 and NHERF-2, some target proteins are able to distinguish one protein from the other (7, 10, 12, 16, 29). We speculate that in addition to the requirement for unique amino acid sequences in the COOH termini, there may be another level of specificity related to differences in the cellular distribution of NHERF-1 compared with NHERF-2. The present experiments in wild-type and NHERF-1-null mice were designed, therefore, to study the subcellular localization of the NHERF isoforms and selected known target proteins in the mouse proximal tubule and to determine whether the absence of NHERF-1 results in mislocalization of NHERF-1-binding partners or loss of cell architecture.
BBM were isolated from mouse kidney using previously described methods (28).
All animal procedures were approved by the Institutional Animal Care and Use Committee of the University of Maryland School of Medicine.
Antibodies. New antipeptide antibodies were made against NHERF-1 (amino acids 298-314): NH2-CSQDSPKKEDSTAPSSTS-COOH and NHERF-2 (amino acids 110-126) NH2-CRGLPPAHDPWEPKPDWA-COOH in both chickens and rabbits. Western immunoblots (22) used antiserum from these immunizations, whereas antibody prepared by affinity purification with immobilized peptide was used for immunocytochemistry (25). Commercially available anti-NHE3 monoclonal and polyclonal antibodies (Chemicon International) and an anti-Npt2 polyclonal (L697) antibody (11) provided by M. A. Knepper (National Heart, Lung, and Blood Institute, Bethesda, MD) were also used.
Immunocytochemistry. Mice were anesthetized with metofane and perfused through the left ventricle of the heart via a blunted 21-gauge needle. Perfusion was for 2 min in PBS to clear the kidneys of blood and 5 min in 2% paraformaldehyde. Kidneys were sliced and further fixed for 60 min in 2% paraformaldehyde followed by 60 min in a cryoprotectant of 10% EDTA in 0.1 M Tris. Tissue was then wrapped in aluminum foil and frozen on dry ice. Cryostat sections 8 μm thick were made and picked up on coverslips coated with HistoGrip (Zymed, San Francisco, CA). Sections were then treated with either 1% SDS (6) or 6 M guanidine (18) for 10 min to unmask antigenic sites. Sections were washed three times with high-salt buffer (50 ml PBS, 0.5 g BSA, 1.13 g NaCl) and incubated in blocking agent (50 ml PBS, 0.5 g BSA, 0.188 g glycine, pH 7.2) for 20 min, followed by incubation with primary antibody overnight at 4°C. Primary antibodies were diluted to 10 μg/ml with incubation medium (50 ml PBS, 0.05 g BSA, 200 μl 5% NaN3). After this incubation, sections were rinsed five times with high-salt buffer over the course of 1 h. Appropriate species-specific secondary antibodies coupled to Alexa 488 or 568 dyes (Molecular Probes, Eugene, OR) were diluted 1:200 with incubation medium and then incubated with the tissue sections for 2 h at 4°C. These samples were again washed five times with high-salt buffer over the course of 1 h and then in PBS to remove the excess salt before mounting and confocal microscopy.
EM immunolocalizations. Cryostat sections of 20 μm thickness were prepared from frozen renal tissue as described above and treated in suspension on coverslips. Antigen retrieval treatment was omitted to preserve morphology, but blocking and primary antibody incubation and washes followed the schedule described above. As a secondary antibody, Alexa 488-labeled goat anti-rabbit (1:100) was applied for 2 h. After washes, tissue sections were fixed in 2% glutaraldehyde and then in 1% OsO4 according to standard protocols and embedded in Epon. Thin sections (80 nm, mounted on bare 300-400 mesh Ni grids) were prepared from suitable cortical regions. To allow labeling of sections, Epon was etched with sodium ethoxide (5). A stock of sodium ethoxide was made by allowing a saturated solution of NaOH in ethanol to age several days in the dark until a burgundy color developed. Saturated sodium ethoxide was diluted to 2% of saturation with ethanol and was used fresh. Grids were treated for three different etching times (from 20 s to 10 min) to achieve optimal etching. Sections were then rinsed in ethanol and water and then blocked as above before the application of the third antibody, which was rabbit anti-Alexa 488 diluted 1:200 (Molecular Probes) for at least 2 h. Use of this antibody takes advantage of the fact that the Alexa 488 fluorophore withstands fixation and embedding in Epon. This labeling was detected by using goat anti-rabbit-gold diluted 1:25 and incubated overnight at 4°C. After being washed in PBS, grids were fixed in 2% glutaraldehyde and stained with uranyl acetate and lead citrate.
Coimmunoprecipitation procedures. Renal cortices from wild-type and NHERF-1 (-/-) mice (22) were hand dissected and homogenized in 1 ml of IP buffer, a solution containing 100 mM NaCl and 10 mM sodium phosphate buffer (pH 7.4) with Complete Protease Inhibitor cocktail (Roche Molecular Biochemicals, Indianapolis, IN). After homogenization, Triton X-100 was added to a final concentration of 1% and the samples were lysed by being drawn through a 27-gauge needle and rocked at 4°C for 30-60 min, followed by centrifugation at 12,000 g for 30 min to remove insoluble cellular debris. The supernatants were precleared with protein GSepharose fast flow beads (prewashed in IP buffer containing 1% Triton X-100) by rocking for 1 h. The beads were then spun down, and the supernatants were aliquoted into 2 samples and incubated overnight with polyclonal antibodies to NHERF-1 or NHERF-2 in the case of wild-type kidneys and NHERF-2 and NHE3 in the kidneys from null animals. Protein A-Sepharose CL4B beads (prewashed in IP buffer containing 1% Triton X-100) were added and allowed to rock for an additional 2 h. Beads were washed several times in IP buffer, and proteins were eluted from beads by being boiled in 200 μl of Laemmli SDS-sample buffer.
Immunoblotting. Proteins and cell lysates (15 μg/lane) were resolved using 10% SDS-polyacrylamide gels electro-phoretically transferred to nitrocellulose and analyzed by Western immunoblotting. The immune complexes were detected by enhanced chemical luminescence (ECL) (Amersham, Arlington Heights, Il).
Monospecific antipeptide antibodies were raised to NHERF-1 in rabbit and to NHERF-2 in both rabbit and chicken. Figure 1 demonstrates that each of these antibodies are isoform specific. Lane 1 of Fig. 1A shows that although an antibody to NHERF-1 peptide recognized a single band in PS120 cells transfected with rabbit NHERF-1, antibodies raised in rabbit to a NHERF-2 peptide recognized a band at a slightly lower molecular weight in PS120 cells transfected with NHERF-2 (lane 2) but not NHERF-1 (lane 3). Similarly, antibodies raised in chicken to the same NHERF-2 peptide recognized the same band in PS120 cells transfected with NHERF-2 (lane 4) but not NHERF-1 (lane 5). When these antibodies were tested on kidney homogenates (Fig. 1B), a single band was labeled by the NHERF-1 antibody in mice with wild-type NHERF-1 (WT), but no bands were labeled in homogenates from NHERF-1-null mice (KO). The NHERF-2 antibodies, on the other hand, each recognized a single band at the lower molecular weight of NHERF-2. The same NHERF-2 band was observed in heterozygous and homozygous NHERF-1-null mice. All bands were eliminated by preincubation of the antibodies with immunizing peptide (not shown).
To define the subcellular location of NHERF-1 and NHERF-2 in the proximal tubule, wild-type mouse kidney was stained using rabbit antibody to NHERF-1 and chicken antibody to NHERF-2 (Fig. 2). NHERF-1 was very abundant in the BBM (Fig. 2A), whereas labeling by the NHERF-2 antibody raised in chicken was strongest at the base of the BBM (Fig. 2B). Although NHERF-2 could be detected in the BBM with this NHERF-2 antibody, its abundance there was low compared with that of NHERF-1. When labeling by the two different antibodies to NHERF-2 were compared, the antibody raised in rabbit (Fig. 3A) labeled the BBM more strongly, whereas the antibody raised in chicken (Fig. 3B) labeled the region at the base of the BBM more strongly. Although these antibodies were raised to the same peptide and both recognize NHERF-2 only as shown above, these immunolocalizations suggest that the two antibodies may recognize different epitopes on NHERF-2 and that the major epitope recognized by the rabbit antibody is more exposed in the BBM, whereas the major epitope recognized by the chicken antibody is more exposed at the base of the BBM. When identical procedures were used to test these same antibodies on rat sections, labeling was seen in the glomeruli and in the vasculature of the kidney, including the descending vasa rectae, but no labeling was detected in the cells of the proximal convoluted tubule, which is consistent with previous reports that NHERF-2 is not expressed proximally in the rat (25). The absence of NHERF-2 in rat proximal tubule was confirmed by RT-PCR of proximal tubules and descending vasa recta hand dissected from collagenase-treated rat kidneys. With the use of nested primers for NHERF-2, an mRNA of predicted size was detected in descending vasa recta but not in proximal tubules. RT-PCR for GAPDH was used as an internal control and was detected in both the proximal tubule and in the descending vasa rectae (data not shown).
To further characterize the site of NHERF-2 labeling, we carried out colabeling localizations with antibody to the coated pit/coated vesicle protein clathrin (Fig. 4). Whereas there was no significant overlap of NHERF-1 labeling with clathrin at the base of the BBM (Fig. 4, A-C), there was significant overlap of NHERF-2 labeling with that of clathrin using both the rabbit (Fig. 4, D and F) and the chicken (Fig. 4, G-I) antibodies. Thus both antibodies indicate that NHERF-2 occurs not only in the BBM with NHERF-1 but also at an adjacent site below the BBM, where NHERF-2 but not NHERF-1 colabels with clathrin. This distribution was confirmed by electron microscopic localizations that showed that NHERF-1 was exclusively detected in the brush border microvilli (Fig. 5A). NHERF-2, on the other hand, was largely localized to an apical band of small vesicles and minimally associated with microvilli (Fig. 5B).
We next sought to define the cellular distribution of NHE3, Npt2, and ezrin because these three proteins have been identified as targets of both NHERF isoforms. NHE3, Npt2, and ezrin all localized strongly to the BBM where NHERF-1 labeling was strong. Modest labeling of NHE3, Npt2, and ezrin was also detectable at the base of the BBM where NHERF-2 is strong (arrows, Fig. 6). We also determined that the distribution of NHERF-2 with respect to NHE3 is unaltered in NHERF-1 null mice (Fig. 7).
To extend these findings, NHERF-1 or NHERF-2 was immunoprecipitated from renal cortical lysates of wild-type mice and NHERF-1-null animals. As shown in Fig. 8, NHERF-1 coimmunoprecipitated NHE3, Npt2, and ezrin from lysates of wild-type mice. NHERF-1 was detected in these immunoprecipitates, as expected. In addition, NHERF-2 was recovered in the NHERF-1 immunoprecipitates. With the use of the rabbit antibody, NHERF-2 coimmunoprecipitated the same proteins from lysates of wild-type mice. In NHERF-1-null animals, however, immunoprecipitation of NHERF-2 coimmunoprecipitated NHE3 and ezrin but Npt2 could not be detected. Immunoprecipitation of NHE3 coimmunoprecipitated ezrin and NHERF-2 but also failed to bring down Npt2.
The NHERF proteins, by binding to ezrin, link the actin cytoskeleton of the cell to their integral membrane targets. The absence of the NHERF proteins, therefore, has the potential to disrupt this connection and the organization of proximal tubular microvilli. However, evaluation of fluorescent phalloidin labeling in NHERF-1-null animals showed that the amount and cellular location of actin was the same in wild-type and NHERF-1-null animals (Fig. 9). The ezrin localization was also unaltered (data not shown). Electron microscopy was used to examine the fine structure of the BBM. As shown in Fig. 10, the substructure of the BBM appears the same in wild-type and NHERF-1-null mice.
NHERF-1 and NHERF-2 are a family of modular proteins containing two tandem PDZ protein interaction domains and a COOH-terminal ezrin/radixin/moesin (ERM)-binding domain. A large number of proteins have been found to interact with the NHERF proteins, including transporters and ion channels, signaling proteins, transcription factors, receptors, and structural proteins. Although some studies have demonstrated preferential interaction with one or the other isoform, most identified targets can interact with both NHERF-1 and NHERF-2 proteins. Of particular interest, both NHERF-1 and NHERF-2 bind NHE3 and ezrin in heterologous expression systems and in other assays, and both isoforms support cAMP-mediated inhibition of NHE3 when expressed in PS120 cell fibroblasts (13, 30, 31). Screening of a yeast 2-hybrid library identified both NHERF-1 and NHERF-2 as interactive proteins for Npt2 (8). Thus, with respect to NHE3 and Npt2, two transporters localized to apical membrane of the renal proximal tubule, it is not clear whether the two NHERF isoforms function cooperatively or independently in the hormonal regulation of renal proximal tubular function. Despite the ability of both transport proteins to interact with NHERF-1 and NHERF-2 when assayed in test systems, recent evidence suggests that both proteins require NHERF-1 for at least some of their functions. The regulation of NHE3 by cAMP is absent in NHERF-1-(-/-) null mice (28). Also, these mice manifest increased rates of phosphate excretion and decreased expression of Npt2 in the BBM of the renal proximal tubule despite the presence of normal amounts of NHERF-2 (22). These results suggest a unique requirement for NHERF-1 in regulation by PKA in the case of NHE3 and apical membrane targeting in the case of Npt2. This specificity may derive, in part, from the interaction of the COOH termini of the target proteins with the PDZ domains of the NHERF proteins. Although the PDZ domains of the two NHERF proteins are similar, they are not identical, and some target proteins can distinguish between them (7, 10, 12, 16, 29). In the present experiments, we explored the possibility that specificity may also arise from the unique cellular distribution of each of the NHERF isoforms by examining the subcellular location of each isoform in the mouse kidney and their association with NHE3, Npt2, and ezrin.
With the use of confocal microscopy, NHERF-1 was detected strongly in the BBM of the mouse proximal tubule as previously described in rat (4, 25) and mouse (9). Although previous studies had indicated that NHERF-2 does not occur in proximal tubule cells of rat (25) and mouse (9), our current work shows that NHERF-2 does occur in the proximal tubule of mouse. Other work indicates that NHERF-2 is also expressed in the human proximal tubule (26). With the use of an antibody raised in chicken, NHERF-2 intensely localized to the base of the microvilli where it colabeled with clathrin. Immunogold electron microscopy confirmed the presence of NHERF-1 in the microvilli and the greater abundance of NHERF-2 in the region at the base of the microvilli. Distinct microdomains of the renal brush border and associated clathrin labeling have been characterized at the base of the microvilli with a coated-pit intermicrovillar region (1-3, 19). Our finding that NHERF-2 but not NHERF-1 is strongly localized to this site suggests that the latter may be important in vesicular traffic.
Modest labeling of NHERF-2 could also be detected in the microvilli with the chicken antibody. To extend these findings, we also employed an anti-NHERF-2 antibody raised in rabbit to the same peptide used for the chicken antibody. The rabbit antibody showed more labeling of the microvillar region but confirmed the presence of NHERF-2 in the clathrin-rich microdomain. Both antibodies indicated that the abundance of NHERF-2 in the microvilli is significantly less than that of NHERF-1. It is of interest, however, that two antibodies to the same protein provided different localizations. The best explanation for the different staining patterns is that antibodies recognizing largely different epitopes were generated in the different animals and that these antigenic sites were differentially occluded in the different microvillar regions. Biemesderfer and colleagues (1) reached a similar conclusion based on experience with multiple antibodies to NHE3.
Given the localization of NHERF-1 in the BBM and NHERF-2 at the base, additional studies were performed to determine their association with some known target proteins. NHE3, Npt2, and ezrin strongly colocalize with NHERF-1 in the microvilli. Lesser amounts of these proteins were also identified at the base of the microvilli where NHERF-2 predominates. In wild-type mice, NHERF-1 coimmunoprecipitated NHE3, Npt2, and ezrin. The immunoprecipitates also contained NHERF-2, indicating the formation of heterodimers of the two isoforms. Evidence for such a direct interaction between the two NHERFs has recently been advanced by Lau and Hall (14). The current results, however, do not exclude the possibility that the interaction may be indirect via other proteins present in the complex. With the use of the rabbit antibody to NHERF-2, the antibody that appeared to detect NHERF-2 most broadly by confocal microscopy, the same proteins were coimmunoprecipitated from the wild-type kidney. In the NHERF-1-null kidneys, however, NHERF-2 antibody coimmunoprecipitated NHE3 and ezrin but not Npt2. This would suggest that the association of NHERF-2 and Npt2 seen in wild-type animals derives from the association between NHERF-1 and NHERF-2 and that NHERF-1 is the major isoform interacting with Npt2. This conclusion would be consistent with the physiological defects in the NHERF-1-null animals (22). It should also be noted that in the rat, in contrast to the mouse, only NHERF-1 is detectable in the renal proximal tubule (25), and in this species, Npt2 is normally targeted and regulated (15, 17) even in the absence of NHERF-2. When NHE3 antibody was used to immunoprecipitate proteins from the kidney of NHERF-1 (-/-) mice, ezrin and NHERF-2 were brought down but not Npt2. This indicates that NHE3 interacts with both NHERF-1 and NHERF-2 and, given that NHERF-1 is solely responsible for cAMP regulation, suggests that NHERF-1 and NHERF-2 subserve unique roles in the regulation of NHE3. A working model of the association among these proteins is shown in Fig. 11.
By linking to the scaffold protein ezrin, NHERF-1 forms a part of the actin cytoskeleton of the cell. Sabolic and colleagues (20) have recently reported that disruption of the microtubule network with colchicine resulted in the trafficking of NHERF-1 and at least one of its targets, NHE3, to the basolateral surface of rat renal proximal tubules. By routine histology, the kidneys of NHERF-1 (-/-) mice appeared normal (22). To refine these initial observations, we used confocal and electron microscopy to define the role of NHERF-1 in maintaining the structure of renal cells. The abundance and distribution of actin, ezrin, as well as NHERF-2, appear to be the same in wild-type and NHERF-1 (-/-) mice. By electron microscopy, the structure of the BBM appeared normal in the null mice. Thus, although there is a potential for NHERF-1 to play a critical role in maintenance of the cytoskeleton, these observations suggest that its role is not essential or that other proteins can compensate for the absence of NHERF-1. Thus it seems unlikely that the mislocalization of Npt2 from the apical membrane of the proximal tubule seen in NHERF-1-null animals is due to a gross disruption of the cellular architecture. A more likely explanation for the defect is the absence of NHERF-1 and, as shown in the present experiments, the apparent absence of direct interaction between Npt2 with NHERF-2 in vivo.
These studies, then, extend prior observations and provide new information pertinent to the possible specific functional roles of the two NHERF isoforms. Initial studies showed an overlapping specificity of NHERF-1 and NHERF-2 with respect to target proteins. A number of recent papers, however, have described NHERF-2-mediated interactions that cannot be duplicated by NHERF-1. Our observations indicate that there is not only biochemical specificity but also an organizational specificity with NHERF-2 localizing to a cell region where NHERF-1 is undetectable and the predominant expression of NHERF-1 is relative to NHERF-2 in the brush border microvillar membrane. The distinct cellular locations of the two NHERF isoforms provide another mechanism allowing for specific interaction of each protein with its targets. This is consistent with the hypothesis that cell types expressing both NHERFs utilize the two isoforms for distinct as well as overlapping functions. The presence of NHERF-2 in the mouse proximal tubule and the ability of NHERF-2 to bind and stabilize NHE3 may explain the rather mild phenotype of NHERF-1-null mice. The apparent inability of NHERF-2 to interact with Npt2 in the absence of NHERF-1 could contribute to the mislocalization of Npt2 and elevated phosphate excretion seen in these animals (22). Finally, recent studies have expanded the biological targets of the NHERF proteins to include not only transporters and channels but also receptors and signaling proteins. Similar specific localization studies are needed to determine whether differential localization of the NHERF proteins affects the function of these target proteins. NHERF-1- and NHERF-2-null mice should be valuable models allowing study of each NHERF in these diverse biological processes.
These studies were supported by National Institutes of Health Grants DK-32839 (to J. B. Wade), DK-55881 (to E. J. Weinman and S. Shenolikar), HL-62220 and DK-42495 (to T. L. Pallone), and grants from Research Service and Department of Veterans Affairs (to E. J. Weinman).
We acknowledge the expert technical assistance of Feng Ying Wang. Dr. Mark Donowitz provided the lysates from the NHERF-2-expressing PS120 cells.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2003 the American Physiological Society