Nedd4-1 and β-Arrestin-1 Are Key Regulators of Na⁺/H⁺ Exchanger 1 Ubiquitylation, Endocytosis, and Function

Abstract

The ubiquitously expressed mammalian Na⁺/H⁺ exchanger 1 (NHE1) controls cell volume and pH but is also critically involved in complex biological processes like cell adhesion, cell migration, cell proliferation, and mechanosensation. Pathways controlling NHE1 turnover at the plasma membrane, however, are currently unclear. Here, we demonstrate that NHE1 undergoes ubiquitylation at the plasma membrane by a process that is unprecedented for a mammalian ion transport protein. This process requires the adapter protein β-arrestin-1 that interacts with both the E3 ubiquitin ligase Nedd4-1 and the NHE1 C terminus. Truncation of NHE1 C terminus to amino acid 550 abolishes binding to β-arrestin-1 and NHE1 ubiquitylation. Overexpression of β-arrestin-1 or of wild type but not ligase-dead Nedd4-1 increases NHE1 ubiquitylation. siRNA-mediated knock-down of Nedd4-1 or β-arrestin-1 reduces NHE1 ubiquitylation and endocytosis leading to increased NHE1 surface levels. Fibroblasts derived from β-arrestin-1 and Nedd4-1 knock-out mice show loss of NHE1 ubiquitylation, increased plasmalemmal NHE1 levels and greatly enhanced NHE1 transport compared with wild-type fibroblasts. These findings reveal Nedd4-1 and β-arrestin-1 as key regulators of NHE1 ubiquitylation, endocytosis, and function. Our data suggest a broader role for β-arrestins in the regulation of membrane ion transport proteins than currently known.

Introduction

Na⁺/H⁺ exchangers (NHEs)² are ion transporters catalyzing the exchange of sodium with protons in prokaryotes and eukaryotes (1). The ubiquitous mammalian NHE isoform 1 (NHE1) controls cellular volume and pH and is therefore often referred to as the “housekeeping” NHE (2). In recent years, however, it has become clear that NHE1 is a highly dynamic protein at the plasma membrane with pivotal importance for mammalian biology that extends beyond ion translocation (3, 4). NHE1 regulates cell shape, adhesion, proliferation and migration (5). Moreover, NHE1 senses mechanical stress directly, thereby serving as a cellular mechanosensor (6). In resting cells, NHE1 is found at sites of focal adhesions in nonpolarized cells and at basolateral membranes in polarized epithelial cells, where it is required for assembly of stress fibers and focal adhesions. In migrating cells, NHE1 is endocytosed at the rear end of the cell and enriched at the leading edge of the cell, promoting cell migration by the development of pseudopodial protrusions and retraction at the rear end (7, 8). NHE1-deficient cells display reduced adhesion, loss of polarity and greatly diminished motility and chemotaxis (7). Conversely, up-regulation of NHE1 is associated with increased tumor growth and tumor cell invasion (9, 10). Regulation of these complex biological processes is thought to require both NHE transport directly as well as anchoring of cytoskeletal elements and scaffolding of signaling molecules by the large, intracellular C terminus of NHE1 (4). Clearly, the dynamics of NHE1 at the plasma membrane mandate rapid and tight regulation of NHE1 turnover. The molecular mechanisms underlying plasmalemmal NHE1 turnover are not known.

Turnover of eukaryotic ion transport proteins often involves the posttranslational conjugation of one or several ubiquitin residues to target proteins by E3 ubiquitin ligases (11). This process, called ubiquitylation, typically targets plasmalemmal proteins for internalization by endocytosis, sorting into multivesicular bodies and delivery to lysosomes. So far, involvement of the ubiquitin pathway in the regulation of NHEs has not been explored.

In some instances, downregulated proteins interact directly with their cognate E3 ubiquitin ligases through PY motifs (PPXY or LPXY). However, many proteins (including NHE1) lack PY motifs and are not capable of directly binding their E3 ubiquitin ligases (11). In the case of G protein-coupled receptors (GPCRs), receptor tyrosine kinases (RTKs) and cytokine receptors, β-arrestins function as adaptors linking receptors with their cognate E3 ubiquitin ligases (12–15).

Here we find that NHE1 turnover at the plasma membrane is regulated by ubiquitylation through a process that is unprecedented for mammalian ion transport proteins, requiring the concerted action of both β-arrestin-1 and Nedd4-1. Our data suggest that Nedd4-1 and β-arrestin-1 are key regulators of NHE1 ubiquitylation, endocytosis, and function.

EXPERIMENTAL PROCEDURES

Unless specified otherwise, all chemicals and reagents were obtained from Sigma. Densitometric quantifications were performed using National Institutes of Health Scion Image software. Statistical analysis was done using Student's t test or one-way analysis of variance (ANOVA) to correct for multiple comparisons, as appropriated. All statistical tests were two-sided and a p value <0.05 was considered statistically significant.

Cell Culture, DNA Transfection, and siRNA Transfection Experiments

HEK293 cells were obtained from ATCC. β-arrestin-1 single knock-out mouse embryonic fibroblasts (MEFs) and corresponding wild-type MEFs derived from littermate mice were obtained from R. Lefkowitz. Generation and characterization of β-arrestin-1 knock-out and wild-type MEFs was described previously (16). Nedd4-1 wild-type and knock-out MEFs were a gift of H. Kawabe and N. Brose and were described in two recent publications (17, 18). Generation and characterization of immortalized skin fibroblasts derived from NHE1 wild-type and knock-out mice was described (19). Transient cDNA and siRNA transfections were performed with Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. Octameric, N-terminally tagged HA-ubiquitin and His₆-ubiquitin constructs were gifts of M. Treier and described in detail (20). N-terminally HA-tagged ubiquitin K/R mutants were generated by site-directed mutagenesis of a single wild-type HA-ubiquitin template in pcDNA3.1. Myc-ubiquitin and untagged wild-type and ligase-dead (C867S) Nedd4-1 constructs were a gift of O. Staub (21). The Myc-tagged wild-type Nedd4-1 construct was a gift from H. Abriel and J. Rougier. β-arrestin-1 cDNA was obtained from Imagenes and cloned into pMH vector (Roche). Wild-type and truncated versions of NHE1 were cloned into p3xFLAG-CMV-14. All constructs were verified by sequencing. Double-stranded siRNAs targeting human Nedd4-1, Nedd4-2, and β-arrestin-1 were purchased from Qiagen and have been validated in HEK293 cells previously (15, 22, 23). The siRNA sequences (5′-3′) used were: Nedd4-1: UAGAGCCUGGCUGGGUUGUUU; Nedd4-2: AACCACAACACAAAGUCACAG; β-arrestin-1: AGCCUUCUGCGCGGAGAAU. The control non-targeting sequence used was: UCAUCUAAGCUGGCUUUGCTT.

Cell Surface Biotinylation, Endo-, and Exocytosis Assays

Cell surface biotinylation, endo-, and exocytosis assays were essentially conducted as described (24). For surface biotinylation, cells were rinsed with PBS and surface proteins were biotinylated by incubating cells with 1.5 mg/ml sulfo-NHS-LC-biotin in 10 mm triethanolamine (pH 7.4), 1 mm MgCl₂, 2 mm CaCl₂, and 150 mm NaCl for 90 min with horizontal motion at 4 °C. After labeling, plates were washed with quenching buffer (PBS containing 1 mm MgCl₂, 0.1 mm CaCl₂, and 100 mm glycine) for 20 min at 4 °C, then rinsed once with PBS. Cells were then lysed in RIPA buffer (150 mm NaCl, 50 mm Tris·HCl (pH 7.4), 5 mm EDTA, 1% Triton X-100, 0.5% deoxycholate, and 0.1% SDS), and lysates were cleared by centrifugation. Cell lysates of equivalent amounts of protein were equilibrated overnight with streptavidin-agarose beads at 4 °C. Beads were washed sequentially with solutions A (50 mm Tris·HCl (pH 7.4), 100 mm NaCl, and 5 mm EDTA) three times, B (50 mm Tris·HCl (pH 7.4) and 500 mm NaCl) two times, and C (50 mm Tris·HCl, pH 7.4) once. Biotinylated proteins were then released by heating to 95 °C with 2.5× Lämmli buffer.

For sequential cell surface biotinylation/immunoprecipitation, monomeric avidin-agarose beads were used which allow, in contrast to streptavidin-agarose beads, elution of biotinylated proteins by excess of free biotin. For this purpose, monomeric avidin beads were pre-washed according to the manufacturer's instructions and subsequently used to precipitate surface membrane proteins that had been biotinylated before as described above. Beads were washed with solutions A–C as described above with additional two washes with a buffer containing PBS/1% Triton X-100 at the end. Biotinylated proteins were then eluted at 4 °C for 1 h in a buffer containing PBS/1% Triton X-100, 20 mm biotin, fresh protease inhibitors (Roche), 2.5 mm N-ethylmaleimide and 10 μm MG132. The resulting bead supernatant was then used for immunoprecipitation experiments.

To measure NHE1 endocytosis, cells were surface-labeled with sulfo-NHS-SS-biotin instead of sulfo-NHS-LC-biotin and quenched as described above. Cells were then warmed to 37 °C for 3 h to allow protein trafficking to occur, control cells were kept at 4 °C. After twice washing with TBS, surface biotin was cleaved with the small cell-impermeant reducing agent sodium 2-mercaptoethane sulfonate (10 mm MesNa, 1 mm EDTA, 0.2% BSA in 50 mm Tris, pH 8.6). The biotin bound to endocytosed proteins is protected from MesNa cleavage. After MesNa quenching with iodoacetamide and one wash with TBS, cells were then solubilized in RIPA buffer and biotinylated proteins were retrieved with streptavidin-agarose affinity precipitation as described above.

For measurement of NHE1 exocytosis, cells were rinsed with PBS thrice and the apical surface was then exposed to 1.5 mg/ml sulfo-NHS-acetate in 0.1 m sodium phosphate (pH 7.5) and 0.15 m NaCl (3 times 40 min at 4 °C) to saturate NHS-reactive sites on the cell surface. After quenching for 20 min (see above for quench conditions), cells were warmed to 37 °C for 3 h to permit protein trafficking, control cells were kept at 4 °C. Cells were then surface-labeled with 1.5 mg/ml sulfo-NHS-LC-biotin and lysed with RIPA buffer. The biotinylated fraction, which represents newly inserted surface proteins, was then affinity-precipitated with streptavidin-agarose, and the precipitate was subjected to SDS-PAGE and immunoblotting. MesNa, sulfo-NHS-acetate, sulfo-NHS-SS-biotin, sulfo-NHS-LC-biotin, monomeric avidin agarose beads, streptavidin-agarose beads and biotin were obtained from Pierce.

Isolation of His₆-ubiquitylated Proteins

Purification of His₆-ubiquitylated proteins was essentially performed as described by Treier et al. (20). Briefly, 24 h after transfection, cells grown on 100-mm dishes were lysed in 2 ml of 6 m guanidium-HCl, 0.1 m Na₂HPO₄/NaH₂PO₄ (pH 8.0), 1% Triton X-100, and 5 mm imidazole and sonicated to reduce viscosity. Lysate was mixed on a rotor with 50 μl (settled volume) of Ni-NTA-agarose (Qiagen) for 4 h at room temperature. The slurry was applied to Bio-Rad chromatography columns and successively washed with the following: 2 ml of 6 m guanidium-HCl, 0.1 m Na₂HPO₄/NaH₂PO₄ (pH 8.0), 1% Triton X-100; 4 ml of 6 m guanidium-HCl, 0.1 m Na₂HPO₄/NaH₂PO₄ (pH 5.8), 1% Triton X-100; 2 ml of 6 m guanidium-HCl, 0.1 m Na₂HPO₄/NaH₂PO₄ (pH 8.0), 1% Triton X-100; 4 ml of (6 m guanidium-HCl, 0.1 m Na₂HPO₄/NaH₂PO₄ pH 8.0, 1% Triton X-100:protein buffer) 1:1; 4 ml of (6 m guanidium-HCl, 0.1 m Na₂HPO₄/NaH₂PO₄ pH 8.0, 1% Triton X-100:protein buffer) 1:3; 4 ml of protein buffer; 2 ml of protein buffer plus 10 mm imidazole. Elution was 1 ml of protein buffer plus 200 mm imidazole. Protein buffer was 50 mm Na₂HPO₄/NaH₂PO₄ (pH 8.0), 100 mm KCl, 20% glycerol, and 0.2% Nonidet P-40. The eluate was then trichloroacetic acid/acetone precipitated for further analysis.

Immunoprecipitation and Immunoblotting

Immunoprecipitation and immunoblotting have been described (25). For chemical cross-linking, cells grown on 6-cm dishes were washed 3× with PBS and incubated with 1 mm DTME (Pierce) in PBS containing 10 mm Hepes pH 7.5 at room temperature for 30 min. Cells were then washed 3× with PBS to remove unreacted DTME, lysed with RIPA buffer, and then used for immunoprecipitation experiments. Antibodies were obtained from the following sources: α-β-arrestin-1, α-Nedd4-1 and α-NHE1 polyclonal (Santa Cruz Biotechnology), α-Nedd4-2 (Abcam), α-ubiquitin P4D1 (Cell Signaling, recognizes monoubiquitin and polyubiquitin chains (26, 27)), α-ubiquitin FK1 (BioMol, recognizes polyubiquitin chains only (26, 27)), α-NHE1 monoclonal (Chemicon), secondary HRP-coupled antibodies and avidin-HRP (Bio-Rad). Cellular lysis buffers always contained fresh protease inhibitors (Roche), 2.5 mm N-ethylmaleimide and 10 μm MG132 to prevent ex vivo artifacts. The Lys⁶³-linked polyubiquitin ladder was obtained from Boston Biochem.

Immunofluorescence and Confocal Microscopy

Cells were fixed in 4% paraformaldehyde in PBS for 10 min, permeabilized in 0.1% Triton X-100 in PBS for 3 min and blocked by 1.5% BSA and 10% goat serum in PBS for 1 h. Fixed monolayers were incubated with primary antibodies in 1.5% BSA and 5% goat serum overnight at 4 °C. Then, after 3× washing in PBS, cells were incubated with the appropriate Alexa 405-, 488-, or 568-conjugated secondary antibodies (Invitrogen) for 1 h at room temperature. After 3× washing in PBS, fixed and stained monolayers were mounted on glass slides. Images were obtained through a Nikon C1 confocal microscope.

NHE1 Activity Assay

NHE1 activity was measured fluorometrically using the intracellularly trapped pH-sensitive dye BCECF and intracellular acid loading was achieved using the NH₄Cl prepulse technique as described previously (28–30). Cells grown on glass coverslips were loaded with 1 μm BCECF-AM and exposed to 25 mm NH₄Cl for 15 min in a buffer containing in mm: 120 NaCl, 5 KCl, 2 CaCl₂, 1.5 MgCl₂, 25 NH₄Cl, 30 Hepes titrated to pH 7.4 with NMDG. Then, cells were washed 3× with and incubated in a buffer containing in mm: 120 tetramethylammonium-Cl (TMACl), 5 KCl, 2 CaCl₂, 1.5 MgCl₂, 30 Hepes titrated to pH 7.4 with NMDG. Recording was started and after 60 s cells were rapidly exposed to a buffer containing in mm: 120 NaCl, 5 KCl, 2 CaCl₂, 1.5 MgCl₂, 30 Hepes titrated to pH 7.4 with NMDG. BCECF fluorescence signals (λ excitation: 490 and 440 nm, λ emission: 535 nm) were recorded in a computer-controlled spectrofluorometer (Fluoromax-2, Photon Technology International). The 490/440 nm fluorescence ratio was calibrated to pH_i using K⁺/nigericin (30, 31). Intrinsic buffer capacity β_intrinsic at pHi 6.0 was determined as described (29, 30) and was not different between β-arrestin-1 wild-type and knock-out MEFs (23.97 ± 4.94 mm H⁺/ΔpH versus 23.78 ± 4.26 mm H⁺/ΔpH) and between Nedd4-1 wild-type and knock-out MEFs (19.16 ± 2.74 mm H⁺/ΔpH versus 18.18 ± 0.33 mm H⁺/ΔpH). All steps of incubation, recording and calibration were performed at 37 °C. Proton flux was calculated as follows: J_H+ = β_intrinsic × ΔpH/Δt (30). Comparisons were always made between cells studied on the same day.

RESULTS

NHE1 Is Ubiquitylated at the Plasma Membrane

As it is currently unknown if NHE1 or any other NHE are ubiquitylated at the plasma membrane, we first tested if plasmalemmal NHE1 is ubiquitylated. NHE1–3xFLAG and HA-ubiquitin constructs or corresponding empty plasmids (pCMV and pMH, respectively) were transiently transfected in HEK293 cells. Then, after labeling of plasma membrane proteins by sulfo-NHS-LC-biotin followed by affinity isolation with monomeric avidin agarose beads, biotinylated proteins were eluted with excess of free biotin (for details see supplemental Fig. S1 and “Experimental Procedures”). Eluates were then denatured by boiling in 1% SDS to remove putative ubiquitylated NHE1-associated proteins (17, 32). After this denaturation step, immunoprecipitation was performed with the indicated antibodies in Fig. 1. Using this approach, the mature, plasmalemmal form of NHE1 (∼90–95 kDa) could be specifically isolated and enriched (Fig. 1A, upper left panel). No NHE1 was present in the immunoprecipitate if the initial biotinylation step was omitted (Fig. 1A, upper left panel). After immunoprecipitation of NHE1, ubiquitylation of NHE1 was detected by blotting with an α-HA antibody recognizing transfected HA-ubiquitin (Fig. 1A, lower left panel). Conversely, a high molecular weight smear of NHE1 was detected when ubiquitin was immunoprecipitated followed by immunoblotting for NHE1 (Fig. 1A, upper right panel).

FIGURE 1.

NHE1 is ubiquitylated. A, HEK293 cells were co-transfected with NHE1–3xFLAG and HA-ubiquitin or empty vectors (pCMV and pMH, respectively). Plasma membrane proteins were isolated by surface biotinylation followed by pull-down with monomeric avidin-agarose beads. After release of biotinylated proteins from monomeric avidin with excess free biotin, eluates were boiled at 95 °C for 15 min in 1% SDS. Samples were then diluted with RIPA buffer containing no SDS to a final SDS concentration of 0.1% followed by immunoprecipitation and immunoblotting by both α-FLAG and α-HA antibodies, respectively (upper panels). Lower panels show expression levels of constructs in cell lysates of HEK293 cells. B, HEK293 cells were transiently transfected with indicated constructs. 24 h after transfection, cells were lysed in a buffer containing 6 m guanidium-HCl and His₆-ubiquitylated proteins were isolated using Ni-NTA-agarose beads. After several washing steps as described under “Experimental Procedures,” His₆-ubiquitylated proteins were eluted from Ni-NTA-agarose beads by 200 mm imidazole, TCA-precipitated, separated by SDS-PAGE and analyzed by immunoblotting (Eluate, upper panel). Bottom panels show expression of constructs in cell lysates prior to Ni-NTA-agarose bead addition (Input, lower panel). Data are representative of two independent experiments.

To further confirm that NHE1 and not an NHE1-associated protein was ubiquitylated, His₆-ubiquitin and NHE1–3xFLAG constructs were co-expressed in HEK293 cells. His₆-ubiquitylated proteins were then isolated by Ni-NTA-agarose beads under denaturing conditions in a buffer containing 6 m guanidium-HCl (20). Fig. 1B demonstrates that a high molecular weight smear of NHE1 could only be detected if NHE1 and ubiquitin constructs were co-expressed but not if the empty vector, NHE1 or ubiquitin constructs were expressed alone. Together, these findings clearly indicate that NHE1 itself and not an NHE1-associated protein is subject to ubiquitylation.

The 76 amino acid protein ubiquitin contains seven lysine residues (Lys⁶, Lys¹¹, Lys²⁷, Lys²⁹, Lys³³, Lys⁴⁸, Lys⁶³) all capable of conjugating ubiquitin (33, 34). It is the current belief that Lys⁴⁸-linked chains primarily serve as recognition signal for proteasomal degradation whereas monoubiquitylation and Lys⁶³-linked polyubiquitylation seem to be involved in cellular trafficking (34, 35). Recent evidence, however, suggests that all non-Lys⁶³-linked chains can target proteins for degradation in eukaryotes (33). We next generated several HA-ubiquitin constructs with arginine replacement of key lysine residues and assessed plasmalemmal NHE1 ubiquitylation. As shown in Fig. 2A, cellular expression levels of the different K/R HA-ubiquitin constructs in HEK293 cells varied greatly. Clearly, however, NHE1 ubiquitylation seemed to be unaffected by arginine replacement of key ubiquitin lysine residues, including Lys¹¹, Lys⁴⁸, or Lys⁶³ (Fig. 2A, left panel). Although poorly expressed when compared with the wild-type ubiquitin construct, even a ubiquitin construct lacking all seven lysine residues led to a NHE1 ubiquitylation signal (Fig. 2A, right panel). These results suggested that NHE1 is either only multi-monoubiquitylated or both multi-monoubiquitylated and polyubiquitylated. To further differentiate between multi-mono- and polyubiquitylation, we employed two monoclonal α-ubiquitin antibodies with different affinities to monoubiquitin and tested native NHE1 ubiquitylation in untransfected mouse skin fibroblasts derived from NHE1 wild-type and knock-out animals, respectively (19, 26, 27). As shown in Fig. 2C, both α-ubiquitin antibodies recognize ubiquitylated proteins in a Western blot of cell lysates. However, whereas the α-ubiquitin antibody P4D1 recognizes both monoubiquitin and polyubiquitin chains of a ubiquitin ladder, the α-ubiquitin antibody FK1 only recognizes the polyubiquitin chains (Fig. 2B). As shown in Fig. 2D, both the P4D1 and FK1 α-ubiquitin antibody detected endogenous ubiquitylated NHE1 in NHE1 wild-type fibroblasts compared with NHE1 knock-out fibroblasts. Together with results obtained with mutant ubiquitin constructs, these findings suggest that NHE1 is subject to both multi-mono- and polyubiquitylation.

FIGURE 2.

NHE1 is both polyubiquitylated and multi-monoubiquitylated at the plasma membrane. A, HEK293 cells were co-transfected with NHE1–3xFLAG and wild-type or indicated mutant HA-ubiquitin K/R constructs. After isolation of surface proteins as described in Fig. 1A, immunoprecipitation was performed using a polyclonal α-FLAG antibody. Immunoblotting was then performed with either a monoclonal α-FLAG antibody (upper panel) or α-HA (lower panel). Note that cellular expression levels of HA-ubiquitin K/R constructs in HEK293 cells were not identical. Data are representative of three individual experiments. B, immunoblotting of a Lys⁶³-linked (1–7) polyubiquitin ladder by monoclonal P4D1 and FK1 α-ubiquitin antibodies and corresponding Coomassie (CM)-stained gel. C, detection of ubiquitylated proteins in a native HEK293 cell lysate (20 μg of lysate protein loaded per lane) by P4D1 and FK1 α-ubiquitin antibodies. D, ubiquitylation of endogenous NHE1 in skin fibroblasts derived from NHE1 wild-type (+/+) and knock-out (−/−) mice. Endogenous NHE1 was immunoprecipitated with a polyclonal α-NHE1 antibody after denaturing samples in 1% SDS for 15 min at 95 °C. Immunoblotting was performed with monoclonal α-NHE1 and monoclonal P4D1 or FK1 α-ubiquitin antibodies. Data are representative of three individual experiments.

NHE1 C Terminus Is Required for NHE1 Ubiquitylation

The large, ∼320 amino acid spanning intracellular C terminus of the 820 amino acid long NHE1 protein contains 27 conserved lysine residues making it a likely target for the attachment of ubiquitin moieties. Whereas the short intracellular NHE1 N terminus does not contain any lysine residues, several short intracellular loops between transmembrane domains also contain conserved lysine residues that would theoretically be accessible to the ubiquitin machinery. To map regions important for NHE1 ubiquitylation, we expressed wild-type (NHE1 WT) or C-terminally truncated (NHE1Δ747, NHE1Δ675, and NHE1Δ550, respectively) constructs in HEK293 cells and assessed surface ubiquitylation (Fig. 3A). NHE1Δ550 lacks most of the intracellular C terminus, including docking sites of a myriad of associated proteins and all known phosphorylation sites (5). As depicted in Fig. 3B, whereas ubiquitin signals of appropriate size could be detected for wild-type NHE1, NHE1Δ747, and NHE1Δ675, no specific ubiquitin signal was detected for NHE1Δ550 compared with control, even after long exposure of the blot (Fig. 3B). Together, these findings indicate that the NHE1 C terminus is required for NHE1 ubiquitylation at the plasma membrane.

FIGURE 3.

Intracellular NHE1 C terminus is required for NHE1 ubiquitylation. A, scheme of C-terminally truncated NHE1 constructs used in experiments. NHE1 transmembrane domain is shown in gray and intracellular NHE1 C terminus in white color. B, HEK293 cells were transiently co-transfected with HA-ubiquitin and either wild-type (NHE1WT) or C-terminally truncated (NHE1Δ747, NHE1Δ675, and NHE1Δ550, respectively) NHE1–3xFLAG constructs. Plasma membrane protein isolation and subsequent immunoprecipitation were performed as described in the legend to Fig. 1A. Asterisks mark mature, fully N- and O-glycosylated NHE1 forms. Data are representative of three independent experiments.

Nedd4-1 Mediates NHE1 Ubiquitylation

Nedd4-1 and Nedd4-2 are E3 ligases shown previously to be responsible for the ubiquitylation of several membrane transport proteins (11). We thus hypothesized, that Nedd4-1 or Nedd4-2 could be mediating plasmalemmal ubiquitylation of NHE1. For this purpose, HEK293 cells were transfected with siRNAs targeting Nedd4-1, Nedd4-2, or control siRNA and plasmalemmal NHE1 ubiquitylation was then assessed (Fig. 4A). Silencing of endogenous Nedd4-1 significantly decreased NHE1 ubiquitylation by ∼ 40% whereas knock-down of the closely related Nedd4-2 isoform led to a small albeit insignificant increase in NHE1 ubiquitylation when compared with control treated cells (Fig. 4B). Typically, > 70% knock-down of Nedd4-1 and Nedd4-2 was achieved by the respective siRNA treatments compared with control siRNA-treated cells (Fig. 4C). Residual NHE1 ubiquitylation despite knock-down of Nedd4-1 suggests that small amounts of E3 ligase activity are sufficient for partial NHE1 ubiquitylation, an observation that was also made in the case of Nedd4-1-mediated ubiquitylation of the β2-adrenergic receptor (15). To confirm results obtained with siRNA targeting Nedd4-1, wild-type, and ligase-dead human Nedd4-1 were transiently overexpressed and NHE1 ubiquitylation was assessed. Despite significant amounts of endogenous Nedd4-1 in HEK293 cells, overexpressed wild-type but not ligase-dead Nedd4-1 further enhanced NHE1 ubiquitylation compared with cells transfected with the empty vector (Fig. 4D). In a next step we aimed to validate these findings in the recently generated Nedd4-1 wild-type and knock-out mouse embryonic fibroblasts (MEFs) (17, 36). As shown in Fig. 4E, whereas NHE1 is ubiquitylated in Nedd4-1 wild-type MEFs, Nedd4-1 knock-out MEFs show loss of plasmalemmal NHE1 ubiquitylation. Taken together, our data suggest that the ubiquitous E3 ligase Nedd4-1 is responsible for NHE1 ubiquitylation at the plasma membrane.

FIGURE 4.

The E3 ligase Nedd4-1 ubiquitylates NHE1. HEK293 cells were transfected with indicated siRNAs at a final concentration of 100 nm at day 0, co-transfected with NHE1–3xFLAG and HA-ubiquitin at day 2 and used for experiments at day 3. Plasma membrane protein isolation and subsequent immunoprecipitation were performed as described in the legend to Fig. 1A. A, typical experiment. B, summary of individual experiments (n = 7 for Nedd4-1 versus control, n = 5 Nedd4-2 versus control). For quantification, ubiquitin signal was normalized to plasmalemmal NHE1 amount and then to control in each individual experiment. Data are expressed as the mean ± S.D. (*, p < 0.05, Nedd4-1 versus control) C, levels of knock-down of endogenous Nedd4-1 and Nedd4-2. 50 μg of lysate protein were loaded per lane. D, HEK293 cells were co-transfected with NHE1–3xFLAG, HA-ubiquitin, and WT or ligase-dead (CS) Nedd4-1 or the empty vector pcDNA3.1 (n = 3). To detect endogenous and transfected Nedd4-1, 20 μg of lysate protein were loaded and probed with a polyclonal α-Nedd4-1 antibody. Data are representative for three individual experiments. E, NHE1 ubiquitylation in Nedd4-1 WT or knock-out (KO) MEFs (n = 2). Immunoblotting was performed with monoclonal α-NHE1 or α-ubiquitin (P4D1) antibodies.

β-Arrestin-1 Is Essential for NHE1 Ubiquitylation

Nedd4 ligases typically interact directly with targets via PY motifs, but many membrane proteins, including NHE1, lack PY motifs (11). β-Arrestins are a family of versatile adapter proteins that were shown to bind to several E3 ligases, including Nedd4-1 (14, 15, 37). Moreover, β-arrestins were shown to interact with the NHE5 isoform, and their overexpression reduced NHE5 surface levels (38).

We thus hypothesized that β-arrestins could play a role in the ubiquitylation process of NHE1. To test this, we used siRNA targeting endogenous β-arrestin-1 in HEK293 cells and assessed plasmalemmal NHE1 ubiquitylation (Fig. 5A). siRNA targeting β-arrestin-1 led to a ∼ 50% inhibition of NHE1 ubiquitylation (Fig. 5B). Typically, > 80% knock-down of β-arrestin-1 was achieved compared with control siRNA treated cells (Fig. 5D). In addition, both Nedd4-1 and β-arrestin-1 together seem to be required for NHE1 ubiquitylation as simultaneous knock-down of β-arrestin-1 and Nedd4-1 together did not lead to a further decrease in NHE1 ubiquitylation (Fig. 5, B and C). In contrast, overexpression of β-arrestin-1 but not of the empty vector enhanced NHE1 ubiquitylation (Fig. 5E). In addition, the importance of β-arrestin-1 for NHE1 ubiquitylation is further supported by the loss of NHE1 ubiquitylation in MEFs derived from β-arrestin-1 KO mice (Fig. 5F).

FIGURE 5.

β-Arrestin-1 is required for NHE1 ubiquitylation. HEK293 cells were transfected with indicated siRNAs at a final concentration of 100 nm at day 0, co-transfected with NHE1–3xFLAG and HA-ubiquitin at day 2 and assayed at day 3. Plasma membrane protein isolation and subsequent immunoprecipitation were performed as described in Fig. 1A. A and B, typical experiments. C, summary of individual experiments (n = 7 for β-arrestin-1 versus control, n = 3 β-arrestin-1 and Nedd4-1 versus control). For quantification, ubiquitin signal was normalized to plasmalemmal NHE1 amount and then to control in each individual experiment. Data are expressed as the mean ± S.D. (*, p < 0.05, β-arrestin-1 versus control; **, p = not significant, β-arrestin-1 and Nedd4-1 versus β-arrestin-1 and p < 0.05, β-arrestin-1 & Nedd4-1 versus control). D, levels of knock-down of endogenous β-arrestin-1, 50 μg of lysate protein were loaded per lane. E, β-arrestin-1 overexpression increases NHE1 ubiquitylation. Myc-ubiquitin and indicated constructs or empty vectors (pMH and pCMV, respectively) were co-expressed and NHE1 was immunoprecipitated with polyclonal α-FLAG antibody (n = 3). F, NHE1 ubiquitylation in β-arrestin-1 wild-type (WT) or knock-out (KO) MEFs (n = 2). Immunoblotting was performed with monoclonal α-NHE1 or α-ubiquitin (P4D1) antibodies.

NHE1, Nedd4-1, and β-Arrestin-1 Interact with Each Other

We next tested if NHE1 associates physically with β-arrestin-1 by co-immunoprecipitation experiments of overexpressed NHE1 and β-arrestin-1. As shown in Fig. 6A, wild-type NHE1 co-precipitated with β-arrestin-1 and truncation of NHE1 C terminus to amino acid 550 (NHE1Δ550) led to loss of β-arrestin-1-NHE1 interaction. To exclude overexpression artifacts, we also performed the experiment with endogenous proteins in HEK293 cells. As depicted in Fig. 6B, endogenous NHE1 also co-precipitated with endogenous Nedd4-1 and endogenous β-arrestin-1 in HEK293 cells. In addition, we also observed an interaction of Nedd4-1 and β-arrestin-1 in HEK293 cells, confirming results described recently by others (supplemental Fig. S2) (15). We next studied the subcellular distribution of NHE1, Nedd4-1 and β-arrestin-1 by confocal imaging in transfected HEK293 cells. Both Nedd4-1 and β-arrestin-1 partially co-localized with NHE1 in the cell (Fig. 6, C–E), supporting the idea of a physical interaction of these three proteins as suggested by functional ubiquitylation studies and co-immunoprecipitation experiments.

FIGURE 6.

β-arrestin-1 and Nedd4-1 interact with NHE1. A, HA-tagged β-arrestin-1 or empty vector (pMH) and wild-type (WT) or C-terminally truncated (Δ550) NHE1–3xFLAG were transfected in HEK293 cells, followed by immunoprecipitation 2 days after transfection. B, endogenous Nedd4-1 and β-arrestin-1 co-immunoprecipitate with native NHE1. HEK293 cells were grown to confluence, subjected to chemical cross-linking (DTME) as described under “Experimental Procedures” and then solubilized by RIPA buffer. Immunoprecipitation was performed with 1 mg of cell lysate protein by either a polyclonal α-Nedd4-1 or a monoclonal α-β-arrestin-1 antibody or equal amounts of rabbit or mouse IgG, respectively. For immunoblotting, a monoclonal α-NHE1 antibody was used. C–E, HEK293 cells were co-transfected with NHE1–3xFLAG and β-arrestin-1-HA (C), NHE1–3xFLAG and Nedd4-1-myc (D), or NHE1–3xFLAG, β-arrestin-1-HA, and Nedd4-1-myc (E). 24 h after transfection, cells were fixed, permeabilized, and immunostained. The white scale bar represents 20 μm. Data are representative for three independent experiments.

Knock-down of Nedd4-1 or β-Arrestin-1 Increases Plasmalemmal NHE1 Levels by Inhibiting NHE1 Endocytosis without Affecting NHE1 Exocytosis

We next studied the consequences of Nedd4-1 or β-arrestin-1 deficiency on NHE1 abundance at the plasma membrane in HEK293 cells. As shown in Fig. 7, A and B, knock-down of Nedd4-1 or β-arrestin-1 led to an increase of NHE1 levels at the plasma membrane, while total cellular NHE1 levels were unaffected (Fig. 7, C and D). The observed increase of plasmalemmal NHE1 levels could be due to decreased NHE1 internalization from the plasma membrane or increased NHE1 exocytotic insertion into the plasma membrane or a synergistic combination of both.

FIGURE 7.

β-arrestin-1 or Nedd4-1 deficiency increases NHE1 steady-state surface levels but not total cellular NHE1 levels. HEK293 cells were transfected with indicated siRNAs at a final concentration of 100 nm. 4 days after transfection, NHE1 surface levels were assessed by immunoblotting of the surface biotinylated fraction. A, typical experiment. B, combined results of four individual experiments are shown in the bar graph where data are expressed as the mean ± S.D. (*, p < 0.05, Nedd4-1 versus control and β-arrestin-1 versus control). For quantification, surface NHE1 levels were normalized to the loading control (avidin-HRP). No actin signal was detected in the biotinylated fraction. C and D, HEK293 cells were transfected with indicated siRNAs at a final concentration of 100 nm. 4 days after transfection, total NHE1 levels were assessed by immunoblotting of cellular lysates. C, typical experiment. D, combined results of four individual experiments are shown in the bar graph. Total NHE1 levels were normalized to the loading control (actin) and then compared with the control siRNA condition in each experiment (NS, not significant, Nedd4-1 versus control and β–arrestin-1 versus control). E, levels of knock-down of endogenous Nedd4-1 and β-arrestin-1, 50 μg of lysate protein were loaded per lane.

To measure NHE1 internalization, we prelabeled membrane proteins with cleavable sulfo-NHS-SS-biotin and measured during 3 h the amount of NHE1 that moved to a location inaccessible to cleavage by the membrane impermeant reagent sodium 2-mercaptoethane sulfonate (MesNa, Fig. 8A). As shown in Fig. 8B, both knock-down of Nedd4-1 or β-arrestin-1 led to a reduction of NHE1 internalization compared with cells transfected with control siRNA. This assay, however, measures net internalization. The observed reduction of NHE1 internalization could thus be either due to a reduction of NHE1 endocytosis or due to an increase in NHE1 recycling. To differentiate between the two possibilities, we measured NHE1 internalization in the presence of 25 μm monensin, a well documented inhibitor of endosomal recycling (39, 40). As demonstrated in Fig. 8B, NHE1 internalization was not significantly increased by monensin treatment, indicating that endocytosis itself was inhibited. To confirm that the monensin dose used was indeed effective in inhibiting endosomal recycling in our cells, we assessed Alexa 594-coupled transferrin uptake in HEK293 cells by confocal microscopy in the presence or absence of 25 μm monensin. In the absence of monensin, endocytosed transferrin is found in the typical perinuclear recycling endosomal compartment (supplemental Fig. S3). In the presence of 25 μm monensin, transferrin accumulates in large cytoplasmic vesicles and labeling of the perinuclear compartment is lost (supplemental Fig. S3). Thus, under the conditions employed, NHE1 trafficking through the recycling pathway does not seem to be an important route for NHE1 turnover at the plasma membrane. In contrast to NHE1 endocytosis, exocytotic insertion of NHE1 into the plasma membrane seemed not to be affected by siRNA-mediated knock-down of Nedd4-1 or β-arrestin-1 in HEK293 cells (Fig. 8, C and D).

FIGURE 8.

Deficiency of Nedd4-1 and β-arrestin-1 affect NHE1 endocytosis but not exocytosis. HEK293 cells were transfected with indicated siRNAs or control siRNA. For NHE1 endocytosis (A and B), cells were surface biotinylated at day 3 after transfection by using sulfo-NHS-SS-Biotin at 4 °C. The cells were then either kept at 4 °C for 3 h (background) or put back to a 37 °C incubator for 3 h to allow active protein trafficking (Fig. 8A). After 3 h, where indicated, surface biotin was cleaved by the reducing agent MesNa. Surface protein was then isolated using streptavidin-agarose beads and protected NHE1 protein abundance was compared with cells not treated with MesNa and is expressed as % NHE1 internalized. Background signal at 4 °C was subtracted for each condition from NHE1 internalization in cells incubated at 37 °C. A, typical blot. B, summary of results. Results of three individual experiments per condition (25 μm monensin or vehicle) are shown in the bar graph where data are expressed as the mean ± S.D. (*, p < 0.05, versus no monensin control; **, p = not significant, versus no monensin treatment condition; #, no net internalization at 37 °C compared with background detected). For NHE1 exocytosis (C and D), sulfo-NHS-reactive sites on the cell surface were saturated with sulfo-NHS-acetate at day 3 after transfection at 4 °C. The cells were then either kept at 4 °C (negative exocytosis control; Fig. 8C, lower panel) or put back to a 37 °C incubator for 3 h (Fig. 8C, upper panel) to allow active protein trafficking. After 3 h, surface proteins were biotinylated. The biotinylated fraction which represents newly inserted surface proteins were then isolated using streptavidin-agarose beads. Background signal at 4 °C was subtracted for each condition from NHE1 exocytosis in cells incubated at 37 °C. C, typical blots. D, summary of results. Results of three individual experiments per condition are shown in the bar graph where data are expressed as the mean ± S.D. (NS, not significant, versus control condition).

Increased Plasmalemmal NHE1 Levels and NHE1 Transport in Nedd4-1 and β-Arrestin-1 Knock-out MEFs

Given the profound effects of Nedd4-1 and β-arrestin-1 knock-down on NHE1 ubiquitylation, surface levels and endocytosis in HEK293 cells, we next utilized MEFs derived from Nedd4-1 and β-arrestin-1 knock-out mice to further test our findings obtained in HEK293 cells (16, 17). Plasmalemmal NHE1 levels were increased in β-arrestin-1 (Fig. 9, A and B) and Nedd4-1 (Fig. 9, D and E) knock-out MEFs by ∼50% compared with MEFs derived from wild-type littermates. These findings support our observations made in HEK293 cells with siRNA-mediated knock-down of Nedd4-1 and β-arrestin-1. When tested for NHE1 transport activity under V_max conditions employing maximal sodium and proton gradients (19), we observed a robust increase of NHE1-mediated proton flux in β-arrestin-1 (Fig. 9, C and G) and Nedd4-1 (Fig. 9F) knock-out MEFs compared with their wild-type counterparts. Enhanced NHE1-mediated transport under V_max conditions is compatible with the observed increase of NHE1 at the cell surface in knock-out MEFs. Thus, collectively, our data indicate that both Nedd4-1 and β-arrestin-1 are key regulators of NHE1 ubiquitylation, endocytosis, and function.

FIGURE 9.

Increased NHE1 surface levels and enhanced NHE1 transport in β-arrestin-1 and Nedd4-1 knock-out MEFs. Plasmalemmal NHE1 surface levels were determined by surface biotinylation in β-arrestin-1 and in Nedd4-1 wild-type (WT) and knock-out (KO) MEFs. A and D, typical experiments. B and E, combined results of three individual experiments are shown in the bar graphs where data are expressed as the mean ± S.D. (*, p < 0.05, β-arrestin-1 or Nedd4-1 KO MEFs versus β-arrestin-1 or Nedd4-1 WT MEFs). For quantification, surface NHE1 levels were normalized to the loading control (avidin-HRP). No actin signal was detected on the biotinylated fractions. C and F, measurement of NHE1 transport activity. NHE1-induced, sodium-dependent proton flux (J_H+) at pH_i 6.0 was determined as described in detail under “Experimental Procedures.” Results are shown in the bar graphs where data are expressed as the mean ± S.D. (n = 6 for each condition; *, p < 0.05, β-arrestin-1 KO versus β-arrestin-1, WT MEFs or Nedd4-1 KO versus Nedd4-1 WT MEFs). G, overlay of typical recordings for β-arrestin-1 KO MEFs (gray) or β-arrestin-1 WT MEFs (black). Arrow indicates switch from TMACl to NaCl containing solution.

DISCUSSION

In this study, we demonstrate the first time that a member of the large NHE family, the ubiquitous NHE1 isoform, is ubiquitylated. Using various ubiquitin K/R mutants and α-ubiquitin antibodies with differential sensitivity to monoubiquitin residues suggest that NHE1 is both poly- and multi-monoubiquitylated. Experiments including: 1) siRNA-mediated knock-down of endogenous Nedd4-1 in HEK293 cells, 2) overexpression of wild-type and ligase-deficient Nedd4-1 constructs in HEK293 cells and 3) analysis of NHE1 ubiquitylation in Nedd4-1 knock-out and wild-type MEFs suggest that the E3 ubiquitin ligase Nedd4-1 is critical for NHE1 ubiquitylation. As shown in Fig. 4A, we occasionally observed diminished plasmalemmal NHE1 expression when Nedd4-1 was knocked down in cells transiently transfected with NHE1–3xFLAG, which was never the case for endogenous NHE1 (Fig. 7, A and B). Possible explanations for this discrepancy include a decrease of protein synthesis/maturation or altered protein trafficking during forced expression of NHE1 with simultaneous Nedd4-1 deficiency. Quantification of NHE1 ubiquitylation thus always included normalization to surface NHE1 levels. Also, effects of Nedd4-1 or β-arrestin-1 deficiency on NHE1 trafficking and surface levels were always performed on endogenously expressed proteins to avoid overexpression artifacts.

In addition to Nedd4-1, the intracellular NHE1 C terminus seems to be required for the ubiquitylation process as truncation of the C terminus to amino acid 550 leads to loss of NHE1 ubiquitylation at the plasma membrane. Whereas the NHE1 C terminus constitutes the most likely site of direct NHE1 ubiquitylation, intracellular loops between transmembrane segments are theoretically also accessible to the ubiquitylation machinery. Thus, it is also conceivable that the NHE1 C terminus plays only an indirect role in NHE1 ubiquitylation e.g. by harboring regulatory sites necessary for NHE1 ubiquitylation. That the NHE1 C terminus is at least modifying the ubiquitylation process is supported by the finding that two partial C-terminal truncations (NHE1Δ747 and NHE1Δ675) are more intensively ubiquitylated than wild-type NHE1, as shown in Fig. 3B. Because of the known redundancy of ubiquitylation sites, however, definitive mapping of NHE1 ubiquitylation sites in the C terminus will require site-specific mutations of all conserved 27 lysine residues in future studies (41).

In addition to Nedd4-1, presence of the adapter protein β-arrestin-1 seems to be essential for NHE1 ubiquitylation. The functional importance of β-arrestin-1 for NHE1 ubiquitylation as revealed by silencing and overexpression experiments and β-arrestin-1-deficient MEFs is supported by the finding that β-arrestin-1 co-immunoprecipitates with NHE1 and Nedd4-1. Both Nedd4-1 and β-arrestin-1 are cytosolic proteins, but both proteins were shown previously to also interact with and regulate plasma membrane proteins (13, 36, 42, 43). In support of this, we observed partial co-localization of NHE1 with Nedd4-1 and β-arrestin-1 by confocal imaging at or close to the plasma membrane (Fig. 6, C–E). We also observed intracellular co-localization of the three proteins, the reasons of which are unclear at the moment. This could be due to an overexpression artifact or indicate the possibility that once internalized, NHE1 remains associated with Nedd4-1 and β-arrestin-1 for a certain time period.

Given experimental evidence provided within this study and the documented scaffolding function of β-arrestins for E3 ligases, β-arrestin-1 most likely recruits Nedd4-1 to the NHE1 C terminus for subsequent NHE1 ubiquitylation (14, 15, 44). However, other scenarios are theoretically also possible, we cannot exclude the possibility that β-arrestin-1 affects NHE1 ubiquitylation independently of Nedd4-1.

siRNA-mediated knock-down of either Nedd4-1 or β-arrestin-1 in HEK293 cells increases NHE1 levels at the plasma membrane by attenuating NHE1 endo- but not exocytosis. These findings are in agreement with our observation of increased plasmalemmal NHE1 levels and increased NHE1 transport activity in MEFs deficient in either Nedd4-1 or β-arrestin-1. The magnitude of the increase in transport activity in knock-out MEFs, however, clearly exceeds the observed increase of surface NHE1 levels. Nedd4-1 and β-arrestin-1 may also influence the balance of active and inactive NHE1 transporters at the plasma membrane, e.g. by altering NHE1 distribution between plasma membrane lipid microdomains or interaction with associated proteins (3, 45). Alternatively, Nedd4-1 and β-arrestin-1 may influence sodium and/or proton kinetics of NHE1 (6, 19). These hypotheses will need to be addressed in detail in future studies.

β-Arrestins have been studied extensively in the regulation of GPCRs, RTKs and cytokine receptors, but few reports demonstrating interaction of β-arrestins with plasmalemmal transport proteins exist (38, 46–48). In the cases of NHE5, the Na/K-ATPase and the calcium channel Ca_v1.2, β-arrestins affected steady-state surface levels of these ion transport proteins. It is conceivable that the regulation also involves ubiquitylation comparable to NHE1.

The NHE family consists of at least 12 members in mammals (1), of which NHE1 is the first member now shown to be regulated by ubiquitylation. Given the known homology of NHEs, β-arrestin-mediated ubiquitylation may serve as a common theme for NHE regulation. Furthermore, our study assessed the mechanisms of NHE1 ubiquitylation under physiological conditions. Future studies will also have to shed light on the impact of pathophysiologically relevant conditions on NHE1 ubiquitylation, e.g. during altered intracellular proton or sodium concentrations. Our findings highlight the importance of β-arrestins in the ubiquitylation of eukaryotic ion transport proteins and suggest that β-arrestin-mediated ubiquitylation of eukaryotic ion transport proteins maybe more common than currently anticipated.