AKT and GSK-3 Are Necessary for Direct Ezrin Binding to NHE3 as Part of a C-terminal Stimulatory Complex

Background: Phosphoinositides and ezrin form a complex with NHE3, which stimulates basal transport activity. How this complex is regulated is unknown. Results: GSK-3 and Akt stimulate NHE3 activity by changing ezrin binding. Conclusion: Sequential effects of two kinase regulate the association of NHE3 with ezrin to affect basal activity. Significance: A novel Ser cluster in the NHE3 C-terminal domain regulates its ezrin binding. Basal activity of the BB Na+/H+ exchanger NHE3 requires multiprotein complexes that form on its C terminus. One complex stimulates basal NHE3 activity and contains ezrin and phosphoinositides as major components; how it stimulates NHE3 activity is not known. This study tested the hypothesis that ezrin dynamically associates with this complex, which sets ezrin binding. NHE3 activity was reduced by an Akti. This effect was eliminated if ezrin binding to NHE3 was inhibited by a point mutant. Recombinant AKT phosphorylated NHE3 C terminus in the domain ezrin directly binds. This domain (amino acids 475–589) is predicted to be α-helical and contains a conserved cluster of three serines (Ser515, Ser522, and Ser526). Point mutations of two of these (S515A, S515D, or S526A) reduced basal NHE3 activity and surface expression and had no Akti inhibition. S526D had NHE3 activity equal to wild type with normal Akti inhibition. Ezrin binding to NHE3 was regulated by Akt, being eliminated by Akti. NHE3-S515A and -S526D did not bind ezrin; NHE3-S515D had reduced ezrin binding; NHE3-S526D bound ezrin normally. NHE3-Ser526 is predicted to be a GSK-3 kinase phosphorylation site. A GSK-3 inhibitor reduced basal NHE3 activity as well as ezrin-NHE3 binding, and this effect was eliminated in NHE3-S526A and -S526D mutants. The conclusions were: 1) NHE3 basal activity is regulated by a signaling complex that is controlled by sequential effects of two kinases, Akt and GSK-3, which act on a Ser cluster in the same NHE3 C-terminal domain that binds ezrin; and 2) these kinases regulate the dynamic association of ezrin with NHE3 to affect basal NHE3 activity.

PI3K is a cytoplasmic signaling molecule that in response to stimulation is recruited to activated receptors. Activation of PI3K results in increased intracellular levels of 3Ј-phosphorylated inositol phospholipids and induction of signaling responses, including the activation of the protein kinase Akt. Basal and some examples of stimulated NHE3 activity are PI3K-dependent (1)(2)(3)(4). Likewise, constitutively active PI3K and Akt are known to stimulate plasma membrane NHE3 activity and amount (3). Akt2 is necessary for NHE3 translocation and activation after initiation of Na ϩ -glucose cotransport, a process that requires ezrin phosphorylation by Akt (5). Although multiple studies have implicated PI3K in regulation of NHE3, evidence for a direct role of Akt on NHE3 activity is limited, and mechanisms have not been explored.
NHE3 is one of the most regulated transport proteins. It can be both rapidly stimulated and inhibited in minutes as part of normal digestive physiology, and it also contributes to multiple diarrheal conditions, in which it is down-regulated (6 -9). Ezrin plays an important role in acute regulation of NHE3 (10). This protein links NHE3 with the actin cytoskeleton in the microvillus and is necessary for changes in rates of regulated endocytosis and some examples of stimulated exocytosis (10 -12). Our recent studies showed that NHE3 associates with ezrin, by at least two different protein-protein interaction sites: 1) ezrin directly binds to NHE3 at its C-terminal domain between amino acids (aa) 3 590 and 667 (13), and 2) ezrin binds to NHERF1 and NHERF2, which then bind the NHE3 C terminus (14,15). The direct binding of ezrin to NHE3 is necessary for: 1) the exocytic trafficking of and plasma membrane delivery of newly synthesized NHE3, which determines the amount of plasma membrane NHE3 and partially determines basal NHE3 activity, and 2) BB mobility of NHE3, which appears to be necessary to increase NHE3 delivery from microvilli to the intermicrovillus space as part of stimulated endocystosis and from recycling system to the intermicrovillar space as part of stimulated exocytosis (13). Our previous reports have shown that the indirect interaction of NHE3 with ezrin through NHERF1/NH-ERF2 is regulated by kinases (14,16,17). However, so far nothing is known about the regulation of direct ezrin binding to NHE3 or in fact whether it is altered as part of signaling that regulates NHE3.
The current study was designed to explore the role of Akt in regulation of direct ezrin binding to NHE3 and thereby its role in regulating NHE3 activity by effects on basal and stimulated exocytosis. This report is the first demonstration of direct Aktmediated stimulation of NHE3 activity, which is independent of changes in ezrin phosphorylation and suggests a complex functional association between these critical signaling molecules. In addition, direct binding of ezrin to NHE3 is shown to be a dynamic process that changes with signaling that stimulates NHE3 activity.
Cell Culture, Plasmid Construction, and Transfection-PS120 fibroblasts and human embryonic kidney cells (HEK293) were used for stable expression of rabbit NHE3-WT, NHE3-S515A, NHE3-S515D, NHE3-S522A, NHE3-S522D, NHE3-S526A, NHE3-S526D, and NHE3-S515A,S526D double mutant, all with the C-terminal vesicular stomatitis virus glycoprotein (VSV-G) epitope tag. The NHE3 mutations were made using the QuikChange site-directed mutagenesis kit (Stratagene). The template for mutagenesis was the pcDNA3.1/hygroϩ vector (Invitrogen) containing VSV-G-tagged rabbit NHE3, which had an epitope tag derived from the VSV-G at the C-terminal end, as described previously (18). DMEM supplemented with 25 mM NaHCO 3 , 10 mM HEPES, 600 g/ml hygromycin, 50 units/ml penicillin, 50 g/ml streptomycin, and 10% fetal bovine serum were used in PS120 and HEK293 cell culture in a 5% CO 2 , 95% O 2 incubator at 37°C. OK-Tina cells, which are OK kidney proximal tubule cells, were previously selected by acid suicide to lack endogenous NHE3 activity (19,20). These cells were cultured in DMEM/ F-12 medium (Invitrogen), supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin at 37°C in a 5% CO 2 , 95% air atmosphere. Weekly "acid suicide" times three was applied to OK-Tina cells to markedly decrease the intrinsic NHE3 activity. This resulted in NHE3 activity after transient transfection of NHE3 nearly all being due to the transfected and not to endogenous NHE3. Post-confluent OK cells were treated with 3 mM EGTA for 30 min before transfection and then transiently transfected with NHE3 WT or mutations using 10 l of Lipofectamine 2000 (Invitrogen) and maintained in serum-free medium overnight. These cells were then used without passaging.
Expression and Purification of Fusion Proteins-The His 6tagged fusion proteins made in the pET30a vector included four fragments of the NHE3 C terminus: F1 (aa 475-589), F2 (aa 590 -667), F3 (aa 668 -747), and F4 (aa 748 -832) (21) and different Ser to Ala/Asp mutants in the F1 fragment. His 6 -tagged fusion proteins were expressed in Escherichia coli Rosetta 2 (DHE-3) cells, and crude cell extract was prepared following the Invitrogen protocol under native conditions. Proteins were affinity-purified with Ni 2ϩ -nitrilotriacetic acid resin as suggested by the manufacturer (Qiagen). Point mutations in fulllength NHE3 were prepared using QuikChange site-directed mutagenesis kit according to the manufacturer's protocol. All the cDNAs were fully sequenced to ensure proper sequence, orientation, and reading frame.
In Vitro Akt Phosphorylation Assay-In vitro phosphorylation was carried out with the Pro Q Diamond phosphoprotein gel stain kit (Invitrogen). Six synthetic peptides based on rabbit NHE3 F1-3 domains (NHE3-F1, aa 475-589; NHE3-F2, aa 590 -667; NHE3-F3, aa 668 -75; NHE3-F1-S515A; NHE3-F1-S522A; and NHE3-F1-S526A) were used in the experiments. 2 g of purified fusion proteins, 0.2 mM ATP, 10 mM MgCl 2 , 100 ng of active recombinant Akt-Ser 473 (Millipore), and 5 l of 10ϫ reaction buffer (final volume, 50 l) were incubated for 30 min at 30°C. Samples were loaded in 14% SDS-PAGE minigels. The minigel was immersed in 100 ml of the Pro Q Diamond phosphoprotein gel stain kit "fix solution" and incubated at room temperature with mild agitation for 45 min, repeating the fixation step once. The gel was washed in 100 ml of ultrapure water with mild agitation for 15 min for a total of three washes. Then the gels were incubated in 100 ml of Pro Q Diamond phosphoprotein gel stain with mild agitation in the dark for 90 min, followed by incubation in destaining solution with agitation for 30 min three times. After three more washes, the gels were scanned with visible light-based scanners (Molecular Imager; Bio-Rad). The gel was further stained with Coomassie Blue.
Overlay Assay-An overlay approach was used to examine the direct interaction of purified recombinant His-tagged F1 fragments (wild type, S515A, S515D, S522A, S522D, S526A, S526D, and S515A,S526D) of the NHE3 C terminus on blots with a purified GST fusion protein encoding the N terminus of ezrin (aa 1-309). Recombinant His-tagged F1 proteins (3 g) were loaded in 14% SDS-PAGE, transferred to nitrocellulose, and blocked with 5% nonfat dry milk in PBS for 1 h. The blots were incubated with 0.8 l of N-terminal ezrin (2 g/l) in 5% milk/PBS overnight at 4°C and then gently washed with 0.05% PBS-Triton three times for 5 min each. The blots were identified with anti-His and anti-GST antibodies (1:1000) and incubated with fluorescently labeled goat anti-mouse secondary antibody (Rockland Immunochemicals, Gilbertsville, PA) for 30 min. Finally, they were washed three times for 5 min each with 0.05% PBST. Protein bands were detected by the Odyssey system (LI-COR).
Cell Surface Biotinylation-The percentage of total NHE3 on the plasma membrane was determined by cell surface biotinylation with NHS-SS-biotin. PS120 cells stably expressing NHE3-VSV-G and the mutants S515A, S515D, S522A, S522D, S526A, S526D, and S515A,S526D were grown to 80% confluence in 10-cm Petri dishes. The cells were serum-starved for ϳ4 h and then treated with Akt i for 60 min. The reaction was stopped by rinsing three times with ice-cold PBS and once in borate buffer (154 mM NaCl, 1.0 mM boric acid, 7.2 mM KCl, and 1.8 mM CaCl 2 , pH 9.0). NHS-SS-biotin (0.5 mg/ml for 20 min twice; Pierce) was used for surface labeling of NHE3. After labeling, cells were washed (three times) with quenching buffer (20 mM Tris and 120 mM NaCl, pH 7.4). Cells were then washed three times with ice-cold PBS and then solubilized with 0.8 ml of N ϩ buffer (60 mM HEPES, pH 7.4, 150 mM NaCl, 3 mM KCl, 5 mM Na 3 EDTA, 3 mM EGTA, and 1% Triton X-100). Cell lysate was incubated with streptavidin-agarose beads overnight. The supernatant was stored as the intracellular fraction. The avidinagarose beads were washed five times in N ϩ buffer to remove nonspecifically bound proteins. Then the avidin-agarose beadbound plasma membrane NHE3 was solubilized in equivalent volumes of 2ϫ SDS loading buffer, boiled for 5 min, separated by SDS-PAGE (10% gels), and then transferred to nitrocellulose. Western analysis and the quantification of NHE3 surface expression were performed using the Odyssey system (LI-COR). Multiple volumes for each total and surface were used with analysis by linear regression with intensity of signal, and the slope was considered n ϭ 1 for each experiment.
Coimmunoprecipitation of NHE3 and Ezrin-Immunoprecipitation was performed using total lysates from PS120 cells stably overexpressing HA-tagged rabbit NHE3 (wild type) and VSV-G-N-terminal ezrin; PS120 cells with no NHE3 were used as a negative control. At ϳ80% confluency, cells were prepared and treated with Akt i or GSK-3 i as described in the surface biotinylation methods. Aliquots (2 mg of protein in 1 ml) of lysate were incubated with 50 l of monoclonal anti-HA affinity agarose or 6 l of monoclonal ezrin antibody at 4°C overnight in a rotator. Bound proteins were eluted with 2ϫ SDS sample buffer, and proteins were separated by SDS-PAGE and transferred to nitrocellulose. The blots were probed with primary antibody (anti-HA and anti-VSV-G or anti-ezrin), followed by fluorescently labeled secondary antibody, scanned, and analyzed using Odyssey system and Odyssey software (LI-COR).
Statistical Analyses-The results are presented as means Ϯ S.D. or S.E. as stated. Comparisons were performed by unpaired Student's t tests or analysis of variance for multiple comparisons.

Akt i Decreases Basal NHE3 Activity in a Concentration-dependent Manner, Which Is a NHERF2-independent Effect-To
understand the function of the Akt signaling pathway in maintaining NHE3 activity under basal conditions, we studied the effect of an Akt i in the NHE null PS120 fibroblast model stably expressing NHE3 and in polarized renal proximal tubule OK cells. The Akt inhibitor VIII used in this study is an isozymeselective, Akt i -1/2 inhibitor, which blocks both basal and stimulated phosphorylation/activation of Akt1/2 (22). This inhibitor has no effects on other closely related AGC family members (23). As shown in Fig. 1, Akt i 10 M treated for 60 min inhibited basal NHE3 activity. This occurred in a concentration-dependent manner with maximum effect at 20 M (Fig. 1A). Because Akt i precipitated at 20 M, 10 M Akt i was studied. This concentration inhibited the initial rate of NHE3 activity (⌬pH/min) by ϳ40%. Phosphorylation of Akt at aa Thr 308 is known to activate Akt kinase. As shown in Fig. 1B, Akt i (10 M) at a concentration that reduced basal NHE3 activity lowered the phosphorylation of Akt on aa Thr 308 .
NHERF2 is a PDZ domain-containing protein that simultaneously binds NHE3 and links it to the actin cytoskeleton via binding to ezrin (13,14). It plays an important role in cGMP and Ca 2ϩ inhibition of basal NHE3 activity and in LPA and dexamethasone stimulation of NHE3 (18,24,25). To determine whether NHERF2 is also involved in Akt-mediated regulation of basal NHE3 activity, transport activity in NHERF2 null PS120/NHE3 and PS120/NHE3/NHERF2 cells treated with 10 M Akt i for 60 min was determined. The presence of NHERF2 did not alter the effect of Akt i on basal NHE3 activity (Fig. 1C). These results indicate that NHERF2 is not required for Akt to stimulate basal NHE3 activity. Similarly, Akt i inhibition of NHE3 in OK cells was not NHERF2-dependent (data not shown).
NHE3 Double Mutant R520F,R527F, Which Has Minimal Direct Binding to Ezrin, Is Not Affected by Akt i :Akt Stimulation of NHE3 Activity Requires Direct Ezrin Binding to NHE-Because both direct ezrin binding to NHE3 and Akt activity are necessary for basal NHE3 activity, we hypothesized that the Akt effect and ezrin binding might be connected. Initially the requirement of direct ezrin binding to NHE3 for Akt regulation of NHE3 activity was determined in PS120 cells. NHE3 and the NHE3 double mutant R520F,R527F were stably transfected in PS120 cells and clones selected that expressed similar amounts of NHE3. Fig. 2A shows that the NHE3 double mutant R520F,R527F had less activity than wild type NHE3 (12), and the effect of Akt i on NHE3 activity was completely lost. Akt2 has previously been shown to be necessary for D-glucose stimulation of NHE3 activity and a parallel increase in plasma membrane expression of NHE3. Both effects required ezrin phosphorylation/activation (10). Hence to separate the effect of Akt on NHE3 activity as being exerted directly on NHE3 and not by effects to activate/phosphorylate ezrin, pulldown assays were performed in PS120 cells, stably expressing HA-NHE3 and transiently expressing N-terminal ezrin (aa 1-309), the active form of ezrin that links target molecules to the actin cytoskeleton, the ezrin domain that binds NHE3 but does not require phosphorylation for activation (26). These cells were treated with Akt i (or DMSO) for 60 min, and NHE3 was pulled down using HA beads. As shown in Fig. 2B, HA-NHE3 pulls down N-terminal ezrin. This effect was prevented by pretreatment with the Akt i . This result shows that Akt activity is necessary for the association of NHE3 and ezrin. Please note that this result also shows that Akt has an effect separate from phosphorylating/activating ezrin that is involved in the NHE3/ezrin association.
A Putative Akt Phosphorylation Site in NHE3 Is Located in the Domain That Directly Binds Ezrin-The above functional effects are consistent with the Akt effect on NHE3 activity involving Akt phosphorylation of NHE3. Based on truncation and point mutation studies, it is known that the NHE3 C terminus is necessary for all identified acute regulation of NHE3 (27,28). The NHE3 (rabbit) C terminus has been divided into four functional domains from which fusion proteins were prepared (Fig. 3A).
To determine which site(s) (if any) of NHE3 was phosphorylated by Akt, an in vitro phosphorylation assay was performed in which purified NHE3-F1-3 domain fusion proteins were exposed to active recombinant Akt-Ser 473 , in the presence of Mg/ATP for 30 min at 30°C. The Pro-Q Diamond phosphoprotein gel stain kit was used to detect the phosphorylation state of each His-tagged fusion protein. As shown in Fig. 3 (B and C), Akt phosphorylates the NHE3 F1 domain but not the F2 or F3 domains, even though the loading of all three purified proteins was similar.
Akt is a Ser/Thr kinase. There are 15 Ser/Thr in the NHE3 F1 domain (aa 475-589). We used a bioinformatics approach (analysis of NHE3 primary sequence via Scan site/Pfam and the Human Proteome Reference Database) to identify potential protein-protein and protein-lipid interaction sequences in the NHE3 F1 domain. NHE3 aa 509 -527, which is in the F1 domain, is predicted to be ␣-helical, as described (Fig. 3D). This putative ␣-helix contains: 1) the demonstrated ezrin direct NHE3-binding site (12) and 2) a demonstrated phosphatidylinositol 4,5-bisphosphate/phosphatidylinositol 3,4,5-trisphosphate and putative PDK1-binding site (aa Arg 511 and Arg 512 ), the latter a Ser/Thr kinase activated by PI3K that activates Akt (Fig. 3E). This NHE3 domain also contains a cluster of three serines (Ser 515 , Ser 522 , and Ser 526 ) modeled to be isolated at one side of a putative ␣-helix (Fig. 3D). One of these is an Akt phosphorylation consensus site (RRXX(S 515 /T)). All of these characteristics of this NHE3 domain are highly conserved across species (Fig. 3F).
We hypothesized that these three serines (Ser 515 , Ser 522 , and Ser 526 ) in the putative aa 509 -527 ␣-helix were part of or helped organize a PI3K-Akt signaling complex on NHE3 that is involved in PI3K-Akt stimulation of NHE3 activity. To test this hypothesis, each of these serines was mutated to Ala (nonphosphorylated) and to Asp (mimics phosphorylation). Purified fusion proteins of NHE3-F1-S515A, F1-S522A, and F1-S526A were used in comparison with wild type NHE3-F1 to identify the Akt phosphorylation site(s) in NHE3. The in vitro phosphorylation of the NHE3-F1 peptide with Akt plus Mg/ATP exposure was eliminated (S515A), significantly decreased (S526A), or not significantly altered (S522A) (Fig. 3, G and H). This indicates that Ser 515 is necessary for the Akt-dependent basal phosphorylation of the NHE3-F1 domain and that Ser 526 but not Ser 522 is involved in this regulation. How Ser 515 is involved in the Akt regulation of basal NHE3 activity was evaluated next. In addition, the potential involvement of Ser 522 and Ser 526 was considered further.
The Ser Cluster Located in NHE3 aa 509 -527 Is Necessary for NHE3 Regulation by Akt-The functional role of this Ser cluster on NHE3 transport activity was determined in PS120 and OK cells, in which WT NHE3 and all the Ser to Ala and Ser to Asp mutants were stably expressed (Fig. 4). NHE3 activities under basal conditions and as altered in response to Akt i were similar in PS120 (Fig. 4A) and OK cells (Fig. 4B) for all mutants. Under basal conditions, all Ser to Ala mutant cell lines had reduced NHE3 activity. NHE3 515 and 526 Ser to Asp mutant cell lines had increased NHE3 activity compared with the comparable Ser to Ala mutant cell lines, with the effect much greater for S526D. NHE3-S515D was still greatly reduced compared with wild type NHE3, but NHE3-S526D had activity that was not significantly different from WT NHE3 (Fig. 4, A and B). All of the mutant cell lines expressed more NHE3 than WT, while being similar to each other (Fig. 4, A and B, insets). These results suggest that all three Ser play an important role in basal NHE3 activity. The only Ser to Asp mutation that increased basal NHE3 activity to wild type level was S526D.
The Akt i inhibition of NHE3 activity was abolished in both NHE3-S515A and -S515D. This is consistent with Ser 515 being an Akt phosphorylation site (phosphomimetic mutation would not be expected to be inhibited by a kinase inhibitor such as Akt i ). Both NHE3-S522A and S522D maintained Akt i inhibition. This indicates that Ser 522 is not the site at which Akt phosphorylates/regulates NHE3 activity.
NHE3 activity in NHE3-S526A and -S526D cell lines responded differently to Akt i . NHE3-Ser 526 was inhibited by Akt i , but NHE3-S526A was not affected. This suggests that NHE3-Ser 526 phosphorylation is required for Akt to stimulate NHE3, but NHE3-Ser 526 itself is not an Akt phosphorylation site.
Akt Regulates NHE3 Activity by Changing Its Trafficking, Which Involves the Ser Cluster-To evaluate whether Akt stimulates NHE3 activity by regulating NHE3 trafficking, we measured the percentage of NHE3 in plasma membrane compared with total NHE3 in the absence or presence of Akt i using cell surface biotinylation. As shown in Fig. 5A, in PS120 cells in which actin was used as the loading control and GAPDH was used to determine whether cytosol was appropriately excluded from surface preparations, surface biotinylation experiments demonstrated that Akt i dramatically decreased the amount of wild type NHE3 on the cell surface. NHE3-S515A and -S515D had less surface expression than wild type under basal conditions, and these two mutants did not significantly change surface expression with Akt i . Together, these data indicate that NHE3-Ser 515 is important for NHE3 trafficking, which is Akt-dependent.  FEBRUARY 28, 2014 • VOLUME 289 • NUMBER 9

JOURNAL OF BIOLOGICAL CHEMISTRY 5453
The NHE3-S526A mutant had a lower percentage of NHE3 on the cell surface than WT NHE3, and this returned to WT levels in the S526D mutant. Akt i did not alter surface expression of NHE3-S526A. However, NHE3 surface expression in the NHE3-S526D cell line was significantly decreased by Akt i . This indicates that Ser 526 is involved in regulation of NHE3 acting at a step prior to or separate from the Akt effect. A similar pattern of change in cell surface expression of NHE3 and these mutants in response to Akt i was obtained in the OK cell lines (data not shown). We conclude that both NHE3-Ser 515 and -Ser 526 play a role in NHE3 regulation by Akt via trafficking and that NHE3-Ser 526 phosphorylation is needed to enable Akt to stimulate NHE3 activity, although this amino acid is not phosphorylated by Akt. NHE3-S515A, -S515D, and -S526A-The pool of ezrin that directly binds to NHE3 is necessary for basal and stimulated exocytic trafficking and BB mobility of NHE3 (13,29). The Ser cluster studied here is required for Akt-stimulated NHE3 activity and is also located in the same putative ␣-helix to which ezrin directly binds. We thus hypothesized that this Ser cluster in the NHE3 C terminus might regulate direct ezrin binding to NHE3 and might also be necessary for plasma membrane distribution of NHE3, which is regulated by Akt.

NHE3 Direct Binding to GST-N-terminal Ezrin in Vitro through Its Cytoplasmic C-terminal F1 Fragment Is Inhibited in
To test whether this Ser cluster has a role in the direct binding of NHE3 and ezrin, epitope-tagged fusion proteins of the NHE3-F1 domain and N-terminal ezrin were studied by an in FIGURE 3. Identification of potential Akt phosphorylation sites in NHE3. A, amino acids defining the F1-4 domains of the rabbit NHE3 C terminus. B, purified fusion proteins were phosphorylated by active Akt-Ser 473 in vitro and then detected by a Pro-Qா Diamond phosphoprotein gel stain kit. The Akt phosphorylation site is located in the NHE3-F1 domain (aa 475-589), but not in F2 or F3. C, densitometry for F1,2,3 phosphorylation with a Pro-Q diamond phosphorylation stain kit. The results of densitometry were normalized to amount of fusion protein with F1 fusion protein set to 100 for each experiment. The results are means Ϯ S.E. of three identical experiments. D, the illustration presents a cluster of three serines (Ser 515 , Ser 522 , and Ser 526 ) located in the same putative NHE3-F1 domain ␣-helix (aa 509 -527) as the previously demonstrated direct ezrin-binding domain (Lys 516 , Arg 520 , and Arg 527 ) (12). E, illustrates the presence in the same NHE3-F1 domain of an established phosphatidylinositol 4,5-bisphosphate/phosphatidylinositol 3,4,5-trisphosphate-binding domain (20) and of a putative PDK1-binding site, as well as an Akt consensus phosphorylation sequence. F, sequence alignment of NHE3 putative ␣-helix region showing amino acids conserved across species, including the serine cluster. G, Akt phosphorylation of NHE3-F1 fusion protein is abolished with NHE3-S515A mutation, significantly reduced with S526A and not significantly altered with S522A mutation. A single experiment is shown. H, densitometric analyses as in G for NHE3-F1 wild type, -S515A, -S522A, and -S526A were normalized to Coomassie Blue signal with NHE3 wild type F1 set to 100% for each experiment. The results are means Ϯ S.E. of at least five separate studies. p values are in comparison to NHE3 wild type F1.
Role of the Phosphorylation of NHE3-Ser 515 and -Ser 526 in Akt Regulation of NHE3 Activity: GSK-3 Phosphorylates NHE3-Ser 526 to Stimulate NHE3-The above results that NHE3-S526D (but not -S526A) was inhibited by Akt i indicate that S526D phosphorylation by a non-Akt kinase is necessary for Akt to stimulate NHE3-Ser 515 . Supporting that phosphoryla-tion of Ser 526 is necessary for Akt to act on NHE3-Ser 515 is that Akt i failed to alter NHE3-S515A,S526D (double mutant) activity (Fig. 7, A and B), and this double mutant had reduced basal activity (Fig. 7, A and B) and surface expression (Fig. 6B). These results support that phosphorylation of Ser 526 precedes the NHE3 effects of Akt.
Based on a bioinformatics approach, NHE3-Ser 526 is predicted as being part of a GSK-3 substrate phosphorylation motif (SXXXS). GSK-3 is a kinase that often acts downstream of PI3K and Akt (30), is often inactivated by Akt (31), and is constitutively active in cells. GSK-3 activity has been shown to be inhibited through phosphorylation of Ser 21 in GSK-3␣ and Ser 9 in GSK-3␤. SB-216763 has been shown to be a potent inhibitor of GSK-3, while having no significant influence on the activity of other kinases (32). To test the efficacy of this inhibitor, PS120/ HEK293 cells were treated with GSK-3i (10 M) for 60 min. A, PS120 cell lines stably expressing wild type NHE3 or mutated NHE3 (S515A/D, S522A/D, and S526A/D) were generated. NHE3 activity with or without Akt i was measured with BCECF/fluorometry. Expression of each cell line is shown in the inset with actin for normalization. NHE3-S515A and -S515D single mutants had lower activity than wild type NHE3 and did not respond to Akt i . Both NHE3-S522A and -S522D had lower basal activity but maintained Akt i inhibition. NHE3-S526D had higher activity than NHE3-S526A. The Akt i inhibition of NHE3 activity was not present in NHE3-S526A cells but occurred similarly to wild type in NHE3-S526D cells. B, studies as in A (including the level of expression shown in inset) were performed in OK cells stably transfected with wild type and mutated NHE3. The results in OK cells gave the same pattern of results described in PS120 cells in A. The means Ϯ S.E. are from at least six separate studies. p values represent Akt i effect compared with untreated control and are shown only when p Յ 0.05. IB, immunoblot.
Representative Western blots are shown in Fig. 8A. Consistent with previous reports, Fig. 8A shows that SB-216763 increases the phosphorylation of GSK-3␤ on Ser 9 , suggesting decreased GSK-3␤ activity. Exposure to a GSK-3 inhibitor significantly reduced basal NHE3 activity (GSK-3i (10 M) for 60 min (p Ͻ 0.05) (Fig. 8B). This GSK-3 inhibitor effect occurred in a concentration-and time-dependent manner, with 5 M having no effect; 10 M GSK i had no effect with 30 min of exposure, but with 60 -90 min of exposure it inhibited NHE3 similarly to the 20 M effect with 30 min of exposure.
Because Akt i altered NHE3/ezrin binding, we also tested the effect of GSK i on the NHE3/ezrin association. GSK-3 i decreased ezrin binding to WT NHE3 (Fig. 8C). This suggests that GSK-3 is also involved in regulation of NHE3 by a process that involves ezrin.
Next the effects of GSK-3 i on NHE3-S522A and -S522D and -S526A and -S526D activities were determined. As shown in Fig. 8D NHE3-S522A and S522D were inhibited by GSK-3 i similarly to wild type NHE3. In contrast neither the NHE3-S526A nor -S526D mutant was inhibited by GSK-3 i (Fig. 8E). To reconfirm the effect shown by GSK-3 i on NHE3 activity, a siRNA-based GSK-3␣/␤ knockdown was created in HEK293 cells. A representative Western blot is shown in Fig. 8F. siRNA transfection decreased the protein expression of both GSK-3␣ (ϳ60%) and ␤ (ϳ40%). The GSK-3 knockdown cells had reduced NHE3 basal activity (Fig. 8G). Also in the knockdown cells NHE3-S522A, -S522D, -S526A, and -S526D mutants had the same qualitative effects on NHE3 activity as the results with GSK-3 i inhibitor (data not shown). We therefore conclude that GSK-3 kinase is acting through NHE3-Ser 526 phosphorylation and not Ser 522 to stimulate NHE3 activity by increasing ezrin binding to NHE3; these events are necessary for Akt to stimulate NHE3 by a process requiring NHE3-Ser 515 .

DISCUSSION
NHE3 is both stimulated and inhibited as part of acute physiologic regulation in the intestine and in the kidney. This regulation involves multiprotein signaling complexes that form on the NHE3 C terminus (6,(33)(34)(35). Two separate complexes have been identified, both of which involve ezrin, although with different functions. One complex, which is the topic of this study, is present nearer the N terminus and predominantly regulates basal and stimulated NHE3 activity (13,21,29). The second complex includes the major site at which the scaffolding NHERF family of multi-PDZ proteins attach to NHE3 along with two active protein kinases; CaMKII, which inhibits NHE3 activity; and CK2, which simulates NHE3 under basal conditions (14,36,37). This study further explores the nature of the more N-terminal stimulatory signaling complex that includes direct ezrin binding to the NHE3 C terminus and also the role of two additional protein kinases, Akt and GSK-3, that stimulate NHE3 by acting through this complex.

NHE3 Stimulatory Complex
Akt2 is present in the apical domain of polarized intestinal epithelial cells. We previously confirmed the studies of Turner and co-workers (10,38) in Caco-2 cells showing that D-glucose stimulation of NHE3 involves apical domain initiated signaling that involves activation of p38 MAP kinase, PI3K, and Akt2, with Akt2 phosphorylating ezrin at aa Thr 567 to activate ezrin by freeing up its N-and C-terminal domains to allow more association with the microvillar actin cytoskeleton as well as with its substrates, including NHE3. We demonstrate here that Akt is involved in stimulation of basal NHE3 activity by mechanisms involving ezrin that occur separately from activation of ezrin by phosphorylaton of Thr 567 (39), although we assume that the ezrin pool involved in this effect of Akt is already phosphorylated at Thr 567 . That the Akt stimulation of basal NHE3 activity did not involve NHERF2 was established, because an Akt i decreased basal NHE3 activity in cells that do not express NHERF2 (PS120 and OK cells). Although Akt may stimulate basal NHE3 activity by affecting multiple sites on NHE3 directly or indirectly, we identified a major role via stimulation of NHE3 phosphorylation in the F1 domain (aa 456 -586). This domain contains a single putative Akt phosphorylation consensus sequence (recognized Akt phosphorylation sites include RXRXX(S/T) or RRXXS 515 ; the latter is present in the NHE3 F1 domain), and our studies strongly suggest that this site is phosphorylated by Akt because recombinant Akt phosphorylated a fusion protein of the NHE3-F1 domain, but this phosphorylation did not occur with mutation of Ser 515 in the consensus site. In addition, the Akt i reduction of NHE3 activity was abolished by mutation of both NHE3-S515A and -S515D with the phosphomimetic not expected to be affected by an Akt i . This Akt phosphorylation consequence sequence was conserved across species (Fig. 4), as was the rest of the putative ␣-helix, in which it occurs, supporting the importance of this domain. However, the activity and surface expression of NHE3-S515D were not quantitatively reconstituted to resemble wild type NHE3. There are at least two possible explanations for this: 1) Ser 515 is important for the regulation of NHE3 activity by Akt, but it is not directly phosphorylated by Akt; and 2) the phosphomimetic mutant NHE3-S515D does not adequately mimic the in vivo phosphorylation. Attempts to demonstrate Akt-dependent   FEBRUARY 28, 2014 • VOLUME 289 • NUMBER 9 phosphorylation of NHE3-Ser 515 could not be accomplished via LC-MS/MS/TiO2, and thus that this site of NHE3 is phosphorylated by Akt under basal conditions remains a hypothesis. It also remains possible that Akt stimulates NHE3 by phosphorylating a non-NHE3 substrate(s), which then requires NHE3-Ser 515 to stimulate NHE3 activity.

NHE3 Stimulatory Complex
A second Ser in the F1 domain was also involved in Akt stimulation of basal NHE3 activity. NHE3-Ser 526 also had reduced basal activity when mutated to Ala and did not exhibit any effect of exposure to the Akt i , demonstrating involvement in Akt stimulation of basal NHE3 activity. This mutant was also phosphorylated significantly less than wild type by Akt studied as part of an NHE3-F1 fusion protein, although the reduction was not as large in magnitude as with NHE3-S515A. The activity of NHE3-S526D was similar to wild type, and it was inhibited by the Akt i similarly to wild type NHE3. This strongly suggests that Ser 526 is not the site at which Akt phosphorylates NHE3 and that it is phosphorylated by a non-Akt kinase, which is necessary for Akt to regulate NHE3 activity by effects at Ser 515 .
The kinase that phosphorylates NHE3 at Ser 526 appears to be GSK-3 (glycogen synthase kinase-3), the initial suggestion of the involvement of GSK-3 in regulation of NHE3 activity. NHE3 activity was reduced by a GSK-3 inhibitor, an effect eliminated with mutations NHE3-S526A and NHE3-S526D. This identifies NHE3 as new substrate for GSK-3 to add to the large number already identified, including other transport proteins. The consideration of GSK-3 as a kinase that phosphorylates NHE3 by an effect at Ser 526 was first based on a bioinformatics approach, which identified NHE3-Ser 526 as having a high probability of being a GSK-3 phosphorylation consensus sequence (SXXXS). GSK-3 has been shown to take part in microtubule dynamics and organization (40,41) and to be involved in both vesicle trafficking (42) and cell polarity (43,44). GSK-3 is ubiquitously expressed and constitutively active, consistent with its having an effect on basal NHE3 activity, and recently it was shown to play a major role in phosphorylation-dependent scaffold regulation in the Wnt signaling pathway (45). Despite compelling evidence of a GSK-3 role in stimulation of NHE3 that requires Ser 526 but not Ser 522 , there are some aspects of this role suggested for GSK-3 that are atypical. GSK-3 often acts downstream of PI3K and Akt as is suggested in this case, but it generally is inhibited by Akt. This is not what appears to be true in the case of NHE3. Also, GSK-3 has a preference for target proteins that are prephosphorylated at a "priming" residue located C-terminal to the site of GSK-3 phosphorylation (46), although this is not always required (47). The consensus sequence for GSK-3 substrates is (S/T)XXX(S/T)-P where the N-terminal S/T usually is the target residue phosphorylated and the C-terminal S/T is the site of priming phosphorylation. Although mutation of both serines in this putative GSK-3 site, Ser 522 and Ser 526 , mildly lowered Akt phosphorylation of the NHE3-F1 domain, NHE3-Ser 522 mutated to both Ala and Asp did not alter the Akt i reduction of NHE3 activity, suggesting that Ser 522 does not have a major role in the Akt or GSK-3 regulation of NHE3. These results suggest that NHE-Ser 526 and not Ser 522 is the site phosphorylated by GSK-3 to allow Akt to regulate NHE3 by acting at Ser 515 . Thus GSK-3 is suggested as acting at a nonconventional GSK-3 phosphorylation site in NHE3 to stimulate it under basal conditions in which it is required for the Akt stimulation/phosphorylation of NHE3.
Although different kinases affect individual Ser in the NHE3-F1 domain Ser cluster, they are in the same putative ␣-helix at which NHE3 directly binds ezrin, which has been shown to be involved in regulation of basal and stimulated NHE3 activity (13,21). Consequently, we tested the hypothesis that the Akt and GSK-3 effects and direct ezrin binding to the NHE3-C terminus were related. The NHE3 direct ezrin binding mutant NHE3-R520F,R527F lost the effect of the Akt i on NHE3 activity. Conversely, pulldown assays using the N terminus of ezrin (aa 1-309) showed that inhibition of Akt and of GSK-3, as well as mutation of the Ser that are phosphorylated by these kinases in the NHE3 F1 domain, separately decreased the coprecipitation of NHE3 with ezrin. Inhibition of Akt and GSK-3 alters basal NHE3 activity, indicating that both are acting under basal conditions to stimulate NHE3 activity, and inhibition of both reduced NHE3 association with ezrin, demonstrating that NHE3-direct ezrin binding is dynamically regulated with signaling that regulates NHE3 activity. These studies used a fusion protein of the N terminus of ezrin such that all interactions between NHE3 and ezrin occurred independently of a previously demonstrated role for Akt in phosphorylation of the C terminus and activation of ezrin (5). This demonstration that there are rapid dynamic, signaling related changes in direct binding of ezrin to NHE3 suggests the possibility that this represents a process that may be important in acute regulation of NHE3. We reported that reducing direct ezrin binding to NHE3 reduced both basal and stimulated NHE3 activity (13,29). This result has been further studied by Hayashi et al. (48). They confirmed dependence of basal NHE3 activity in ileum on the presence of ezrin, although comparable studies in mouse colon, Madin-Darby canine kidney, and OK cells did not demonstrate this dependence. Surprisingly their results showed dependence of cAMP inhibition of NHE3 on the presence of ezrin at a step separate from NHE3 phosphorylation. Although the explanation for the differences and similarities of these results with our studies is not understood, the role of direct ezrin binding of NHE3 in regulation of NHE3 may be futher revealed by the studies of the F1 signaling complex.
Despite the description of the NHE3 regulation that involves Akt and GSK-3, how the Ser cluster functions has not been clarified beyond showing that its effect on NHE3 activity involves regulation of direct binding of ezrin to NHE3. However, some information about the relationship of a Ser cluster and ezrin binding in the same ␣-helix of a protein is available from studies of ICAM-3. ICAM-3 previously was shown to have an ␣-helix that bound both ERM proteins directly and had a Ser cluster in another part of the same helix (49). When this Ser cluser of ICAM-3 was mutated, ICAM-3 binding to ERM protein was prevented, as was the polar distribution of ICAM-3. Although there was no evidence of changes in polar distribution of NHE3 in OK cells in which NHE3-Ser 515 and -Ser 526 were mutated, our studies have established a role for the Ser cluster in regulation of NHE3 activity by effects on trafficking of NHE3 under basal conditions. Taken together, these studies demonstrate that a Ser cluster in the putative NHE3 ␣-helix (aa 509 -527) acts to organize an Akt and GSK-3 signaling complex that forms on the NHE3 C terminus, which is involved in the stimulation of basal NHE3 activity by affecting direct ezrin binding to NHE3. This putative ␣-helix is the site for convergence of multiple proteins involved in basal and stimulated NHE3 activity. In addition to ezrin binding, this putative helix also contains aa shown to be necessary for NHE3 to bind the phosphoinositides phosphatidylinositol 4,5-bisphosphate and phosphatidylinositol 3,4,5trisphosphate and a putative PKD-binding domain. Phosphatidylinositol 4,5-bisphosphate and phosphatidylinositol 3,4, 5-trisphosphate have been shown to be necessary for basal and regulated NHE3 activity (21). PDK is necessary for normal NHE3 activity, which was reduced when hypomorphic PDK1 was present (50). In addition PDK is necessary for short term dexamethasone stimulation of NHE3 activity and for SGK3-dependent stimulation of NHE3 activity (3). Further studies are needed to identify additional proteins that take part in this NHE3-F1 multiprotein complex, to identify how they are affected by Akt and GSK-3, and to further define the role of this complex in NHE3 regulation.