Physiological Reviews

Regulatory Binding Partners and Complexes of NHE3

Mark Donowitz, Xuhang Li


NHE3 is the brush-border (BB) Na+/H+ exchanger of small intestine, colon, and renal proximal tubule which is involved in large amounts of neutral Na+ absorption. NHE3 is a highly regulated transporter, being both stimulated and inhibited by signaling that mimics the postprandial state. It also undergoes downregulation in diarrheal diseases as well as changes in renal disorders. For this regulation, NHE3 exists in large, multiprotein complexes in which it associates with at least nine other proteins. This review deals with short-term regulation of NHE3 and the identity and function of its recognized interacting partners and the multiprotein complexes in which NHE3 functions.


NHE3 (SLC26A3) is one of nine isoforms of the mammalian Na+/H+ exchanger (NHE) gene family (60, 62, 121, 357, 358, 476, 537). The mammalian NHEs consist of isoforms that occur primarily in the plasma membrane and those that appear to primarily reside in intracellular organelles (60). However, it may be that all NHEs have at least transient plasma membrane localization (62). The evolutionary history of the NHEs showed that the intracellular NHEs arose in a setting in which their transport was driven by H+ gradients generated by V-ATPases (62, 182, 342, 353, 354). The earliest evolving plasma membrane NHEs exist both in the plasma membrane and on the membranes of intracellular organelles (22, 33, 48, 255, 359, 448). These NHEs traffick continually between the plasma membrane and intracellular organelles (62, 226, 255, 434). They arose at a time of evolutionary appearance of the Na+-K+-ATPase, which creates a Na+ gradient and provides the energy for their exchange of extracellular Na+ for intracellular H+. The most recent NHEs to evolve appear to be the NHEs that are present predominantly on the plasma membrane (359, 401, 450, 451). The plasma membrane and intracellular NHEs can be separated by differences in the amino acid residues in their membrane-spanning domains, especially the putative P-loop (244, 335, 339, 417, 475). NHE1-5 are the plasma membrane NHEs. Of these, NHE3 and NHE5 traffic between the recycling endosomes and the plasma membrane under basal conditions, while NHEs 1, 2, 4 appear to be statically present in the plasma membrane. The intracellular NHEs include NHE6, -7, and -9. It is not known whether the intracellular NHE isoforms exist statically in specific organelles or whether individual intracellular NHEs exist in multiple organelles (328, 342, 505). NHE8 resembles the intracellular NHEs based on amino acid composition but appears to be present primarily on the plasma membrane, being found in the brush border of the renal proximal tubule and small intestine (20, 181, 182, 513). The driving force for the NHEs on the plasma membrane is the Na+ gradient generated by Na+-K+-ATPase (123). For resident organellar NHEs, which appear to generally exchange cytosolic K+ (perhaps also Na+) for intraorganellar H+, the driving force is primarily the H+ gradient generated by the organellar V-ATPase (61, 465). These NHEs transport H+ out of the compartment in exchange for cytosolic K+ or Na+ and thus help set the organellar pH (61, 465). NHE3 and NHE5 also appear to function on intracellular organelles (8, 134, 163, 434). In a subset of early endosomes, extracellular Na+ appears to provide the driving force for acidification (162165). All plasma membrane members of the NHE gene family transport Na+ and H+ with a stoichiometry of 1:1 (20, 21, 476). The relative affinity for transport of Na+ versus K+ by the organellar NHEs in exchange for intraorganellar H+ is not known (60, 61). Please note, a distantly related sperm transporter is not considered a member of the SLC26A family. Also, a proposed colonic Cl-dependent NHE isoform (384, 385, 400) is believed to be a cloning artifact (403), and rather, the suggestive physiological studies are suspected of representing the now recognized Cl dependence of multiple NHE isoforms (4).

NHE3 is present in the brush border (BB) of small intestine (duodenum, jejunum, ileum), colon, gallbladder, parietal cells of some species but not others, cholangiocytes, some pancreatic duct cells, proximal tubule of the kidney, thick ascending limb of Henle and proximal portion of the long descending thin limb of Henle, a few neurons in the cerebellum and respiratory neurons of the ventrolateral medulla (where it is suggested as being involved in respiratory rhythm generation, neuronal chemosensitivity, and the central CO2 respiratory response), in perineuronal cells (not nerves) of the laryngeal nerve, chondrocytes, parotid duct, and epidydimus duct (13, 20, 24, 25, 30, 47, 48, 56, 58, 60, 62, 64, 98, 99, 101, 138, 154, 159, 200, 207, 210, 226, 236, 254, 255, 259, 282, 291, 295, 313, 326, 345, 365, 375, 378, 383, 389, 394, 395, 414, 424, 438, 508, 517). In addition, NHE3 is present in gills of fish (86, 141, 509) and is involved in mouse blastocyst formation (237).

The major recognized functions of NHE3 are in the apical membrane of the proximal tubule and small intestine and colon (20, 121, 160, 403, 452, 511, 537). NHE3 is responsible for the majority of NaCl absorption in the intestine and kidney, acting predominantly in their proximal portions (explaining proximal tubule and small intestinal linked Na+ and Cl absorption). In addition, it accounts for the majority of HCO3 absorption in the kidney since the amount of NHE3 appears to exceed the amount of apical anion exchanger, with the residual Na+/H+ exchange activity being equivalent to HCO3 absorption (89, 121, 123, 184, 313, 314, 330, 512, 537). At least in the gastrointestinal (GI) tract and kidney, NHE3 is functionally linked to a BB Cl/HCO3 exchanger and takes part in neutral NaCl absorption (123, 259, 260, 275). In this process one molecule of Na+ and one molecule of Cl are absorbed together with no net movement of charge. This process depends on intracellular carbonic anhydrase to generate the H+ and HCO3 (123). The Cl/HCO3 exchanger involved probably varies among intestinal segments and appears to include PAT1 (putative anion transporter or SLC26A6) or DRA (downregulated in adenoma, SLC26A3) (20, 35, 159, 223, 268, 275, 406, 482, 483). The current suggestion is that PAT1 may dominate in Na+ absorptive cells in the duodenum and in jejunum, while DRA is involved in ileum and colon, although this appears to be species dependent. For instance, in some species, DRA is the duodenal villus apical anion exchanger. In addition, some tissues and even some cells may have both DRA and PAT1; for instance, in mouse duodenum, DRA is in crypt cells and PAT1 is in the villus cells (415). In the kidney, the involved anion exchanger also appears to be PAT1, although SLC26A7, another anion exchanger, is also reported to be in the proximal tubule BB apical membrane.

The neutral NaCl absorptive process is highly regulated (121, 123, 537). It is both rapidly (over minutes) stimulated and/or inhibited as part of digestion in the GI tract and in response to neurohormal stimulation in both GI tract and kidney proximal tubule. NHE3 is the part of neutral NaCl absorption that has been shown to be directly regulated, although similar detailed studies have not been reported for the BB Cl/HCO3 exchanger. In fact, in preliminary studies, the apical membrane anion exchange may be inhibited by protein kinase C (397, 398). In colon, in addition to the NHE3/PAT1 or DRA-linked NaCl absorption, there is an additional neutral Na+-linked anion absorptive process in which NHE3 or NHE2 is linked to a short-chain fatty acid/anion exchanger (Fig. 1). (176, 267, 466, 467). A congenital diarrheal disease due to lack of BB NHE activity (almost certainly NHE3) occurs (55, 238, 336). It is not due to a structural or localization abnormality in NHE3.

FIG. 1.

Two forms of neutral Na+ anion absorption in mammalian colon. The neutral NaCl absorptive processes involve NHE3 linked to either SLC26A3 or -6 based on cells involved (A) and the short-chain fatty acid (SCFA)/anion exchanger linked to either NHE2 or NHE3, with NHE2-SCFA uptake active even in the presence of cAMP, which inhibits NHE3 (B).

In the kidney and intestine and in all cell culture models in which NHE3 is expressed endogenously or exogenously, NHE3 cycles between the plasma membrane or BB and the recycling compartments under basal conditions. The compartment involved has not been characterized biochemically, but is assumed to be early endosomes perhaps before incorporation of the V-ATPase (8, 134, 162165, 481). NHE3 functions to acidify the early endosomes. The driving force in this vesicular acidification appears to be the concentration gradient of Na+, which depends on extracellular Na+ that is trapped in the vesicle as it is endocytosed from the plasma membrane (134). In several cell types, including fibroblasts and the opossum kidney (OK) renal proximal tubule cell line, as first shown by D'Souza et al. (134), NHE3 functions to acidify early endosomes (134). This acidification has been suggested by Geckle and co-workers (162, 163) to be necessary for receptor-mediated endocytosis of albumin in renal proximal tubule cells. The effect of the NHE3 inhibitor S3226 on receptor-mediated/clathrin-dependent endocytosis is additive with inhibition of endocytosis caused by knock out of ClC-5 (404, 481). This suggests two separate uptake processes are involved in proximal tubule endocytosis (481). Whether this also occurs in the intestine or is relevant to intestinal function is not known, nor is it known what replaces this process in cells lacking NHE3.

Short-term regulation of NHE3 mimics digestive and renal physiology or regulation by changes in the dietary or neurohumoral environment of the intestine and kidney and also models pathological states, including diarrhea (13, 121, 123, 330, 537). The mechanisms involved in short-term NHE regulation include changes in plasma membrane turnover number, changes in amount of NHE3 expressed on the plasma membrane by altering rates of endocytosis and/or exocytosis, and changes in NHE3 half-life (50, 70, 97, 121, 130, 212, 226, 330, 358, 537). We have found that NHE3 simultaneously exists in multiple large multiprotein complexes particularly on the plasma membrane under basal conditions and that the NHE3 complexes change as part of some NHE3 regulation in terms of complex size and associating proteins (8, 71, 296, 298, 340). There appears to be coordinated regulation of NHE3 and anion secretion in murine jejunum, which would allow coordinated movement of intestinal Na+, Cl, HCO3, and water transport (158, 159).

This review examines the current understanding of the mechanisms of NHE3 regulation, the identified NHE3 complexes, and the proteins which directly and/or indirectly interact with NHE3, with an emphasis on the NHE3 binding partners and their role in acute NHE3 regulation.


NHE3 is one of the most regulated transport proteins. It is both rapidly stimulated and inhibited as part of normal digestive physiology, and it contributes to multiple pathophysiological states when it is downregulated for a prolonged period (121, 130, 330, 537). Rapid regulation of NHE3 occurs over minutes in the intestine as part of digestion either in response to change in luminal end products of digestion (at least, generation of d-glucose and short-chain fatty acids) or to changes in the neurohumoral mileu of the intestine. Changes in NHE3 activity in the kidney occur with a somewhat slower time frame (many minutes to a few hours) in response to drugs, dietary changes (especially change in Na or K intake), or changes in blood pressure/volume (13, 321). While the agonists that alter NHE3 in the GI tract and kidney are somewhat different, the response of NHE3 to changes in second messengers is the same, i.e., NHE3 inhibition induced by parathyroid hormone (PTH) may occur over many minutes in the kidney proximal tubule via activation of the adenylate cyclase/cAMP/protein kinase A (PKA) II plus protein kinase C (PKC) systems, while similar effects on NHE3 activity occur in the small intestine over a few minutes in response to elevation of vasoactive intestinal polypeptide (VIP), secretin, or cholingergic agonists and accompanying changes in cAMP or Ca2+.

Transport kinetics for NHEs, including NHE3, include Michaelis-Menton kinetics for extracellular Na+ uptake (kinetics of K+ transport by intracellular organellar NHEs has not been defined as yet), with 1:1 exchange of extracellular Na+ for H+ (20, 21, 252, 358, 449, 452). However, intracellular H+ export is more complex, with H+ both being transported and activating the exchanger allosterically through the so-called H+ modifier site (21, 198, 422, 471, 472, 476). This leads to the K′(H+)i being a complex term, with a Hill coefficient of ∼2 (21). Regulation of NHE3 generally involves changes in NHE3 Vmax and often, but not always, changes in K′(H+)i (121, 357, 537). However, usually this is not associated with changes in the NHE3 Math extracellular. Also, occasionally there are changes in the NHE3 set point (73). Two basic regulatory mechanisms are involved in short-term NHE3 regulation: 1) changes in turnover number with no changes in trafficking; and 2) changes in trafficking (121, 330, 537). The latter involves changes in regulated exocytosis and/or endocytosis, which can include effects in both by the same agonist.

Some of the agonists that rapidly regulate NHE3 in the intestine or renal proximal tubule are listed in Table 1. Categories of ligands include neurohumoral substances produced in the GI tract or systemically, neural mediators most made in the GI tract, kinase regulators, phosphatase inhibitors, changes in osmolarity, and end products of digestion.

View this table:

Short-term regulation of NHE3

The stimulation of NHE3 by an end product of digestion was identified by Turner and co-workers (Fig. 2) when they showed that NHE3 was rapidly stimulated in response to luminal d-glucose by a process involving SGLT1 (213, 410, 458460, 542). This occurred by activation of an apical membrane initiated linear signaling pathway that includes early activation of p38 mitogen-activated protein (MAP) kinase, followed by activation of MAPKAPK2 (MAP kinase activated kinase 2), phosphatidylinositol 3-kinase (PI 3-K), Akt2, and ezrin, which lead to stimulation of NHE3 activity due to increased exocytosis. The model used to identify this pathway is the Na+ absorptive colon cancer cell line Caco-2, which is a small intestinal Na+ absorptive cell model. The importance of the identification of this pathway is that it demonstrates that NHE3 is an intermediate in the increased Na+ absorption that occurs in the postprandial state by two mechanisms. 1) It is an intermediate in end product of digestion initiated stimulation of intestinal Na+ absorption, where it couples to other mechanisms of stimulated postprandial Na+ absorption. However, it is not known how much of this NHE3 activation accounts for d-glucose-driven jejunal Na+ absorption versus the solvent drag that was thought to be the magnifying effect to Na+/d-glucose uptake across the jejunal apical membrane (149, 309, 310, 460). 2) Regulation of NHE3 is also initiated by neurohumoral ligands that are released during digestion, as described below. Important unanswered questions about the linkage of NHE3 stimulation to end products of digestion in the lumen of the small intestine include the following: 1) What is the initial intracellular signal? Does it occur in intact intestine in addition to in a cell culture model, and if so, is the signal transduction involved the same as in cell culture models and how is it altered by changes in the digestion-induced alteration in the neurohumoral milieu? This is important given the recognition that even major diarrhea-related secretagogues, such as cholera toxin, which alter ion transport in epithelial cells that can be examined in cell culture models, exert approximately half of their effects in intact intestine via neurally mediated mechanisms that involve the enteric nervous system (308). 2) Do other end products of digestion, such as l-amino acids and di- and tripeptides, have a similar role? If so, is it via the same signaling pathway as d-glucose? 3) How far inside the cell away from the apical domain are signals from this pathway activated and do they change signaling in the entire cell?

FIG. 2.

Putative signaling pathway by which luminal d-glucose acts via SGLT1 to stimulate NHE3 by increasing its exocytosis. RE, recycling endosomes. [Adapted from work of Turner and co-workers (213, 410, 458460, 542).]

Short-chain fatty acids take part in linked neutral Na+ absorption in the colon (Fig. 1) (176, 267, 466, 467). NHE3 is the BB NHE linked to BB Cl/HCO3 exchange; however, there is an additional neutral Na+ absorptive process linked to a BB short-chain fatty acid/anion exchange (Fig. 1). The latter leads to neutral Na+/short-chain fatty acid uptake. Both NHE3 and NHE2 are able to link to colonic short-chain fatty acid uptake (467). The regulation of this process is assumed to be driven by the high luminal concentration of short-chain fatty acids, plus the Na+ gradient. This probably accounts for some basal colonic Na+ absorption given the high luminal concentration of short-chain fatty acids in the normal colon. cAMP elevation only inhibits NHE3 and thus blocks the neutral NaCl absorptive process but does not inhibit the NHE2-linked SCFA absorptive process. Of note, this process may be stimulated since cAMP stimulates NHE2 (54, 232, 323, 347). This is the theoretical basis for the development of an oral rehydration solution (ORS) that increases colonic luminal short-chain fatty acid concentration by using relatively amylase-resistant carbohydrates, such as corn starch (386) In an early report, this ORS shortened the duration and volume of cholera diarrhea, the first ORS reported to have both these effects (386).

As examples of the rapid neurohumoral ligand regulation of NHE3 (Table 1), epidermal growth factor and α2-adrenergic agonists stimulate ileal Na+ absorption, and lysophosphatidic acid increases NHE3 in OK renal proximal tubule cell line. Stimulation of renal proximal tubule BB NHE3 also occurs with endothelin-1, angiotensin, and exposure to acidosis. Most of these agonists stimulate NHE3 exocytosis without changing the rate of endocytosis. These agonists act to increase BB NHE3 amount, with comparable increases in NHE3 activity and percent of NHE3 on the apical membrane (285, 295).

cAMP, cGMP, and elevated intracellular Ca2+ and the neurohumoral substances and bacterial toxins that cause these second messengers to increase all inhibit neutral NaCl absorption and its component BB NHE3 (121, 197, 330, 537). This occurs in both the GI tract and renal proximal tubule. The regulated mechanisms vary widely. At one extreme, cAMP inhibits NHE3 in PS120 fibroblasts by changing the NHE3 turnover number without affecting the amount of plasma membrane NHE3 (130, 277). In contrast, in OK cells, cAMP initially decreases the turnover number, but in ∼30 min, it is associated with an increase in rates of endocytosis, followed by inhibition of exocytosis (60 min) and a decrease in total NHE3 half-life (506, 543). This demonstrates highly coordinated NHE3 regulation in response to a single second messenger.

A. NHE3 Phosphorylation

NHE3 is phosphorylated under basal conditions (525), and that phosphorylation is often affected as part of acute NHE3 regulation (97, 115, 212, 271, 330, 331, 372, 453, 507, 547). Changes in NHE3 phosphorylation are necessary for some acute regulation of NHE3 but apparently not for others. There are examples of NHE3 regulation by both changes in trafficking and not involving trafficking that are dependent on changes in NHE3 phosphorylation. This basic regulatory mechanism is partially understood with identification of some phosphorylation sites of NHE3, some roles of phosphatases in NHE3 regulation, the role of kinase localization to specific subcellular domains, and involvement of NHE3 complexes all being partially defined. In contrast, further NHE3 phosphorylation sites, additional roles of phosphatases, and restricted spatial and temporal phosphorylation of NHE3 as part of regulation are all predicted, and cAMP has even been shown to act by a non-PKA-dependent mechanism that inhibits NHE3 in some cells without causing a change in NHE3 phosphorylation (EPAC, exchange protein directly activated by cAMP) (209). This area was reviewed with great insight by Moe (330), whose laboratory has provided many of the studies in this area.

NHE3 is phosphorylated under basal conditions in several cell culture models as well as in intact tissue. These include fibroblasts, OK proximal tubule, and Caco-2 cells as well as in renal proximal tubule and ileal BB. The great majority of basal and stimulated phosphorylation is on Ser with minor phosphorylation of Thr (212, 232, 330, 331, 372, 507, 525, 547). Tyr phosphorylation of NHE3 has not been identified. Of interest is that at least one Tyr residue (Y323 of rabbit NHE3) is predicted to be phosphorylated with a 95% probability by Netphos2.0. The phosphorylated sites have only been partially defined, and use of mass spectroscopic approaches has revealed previously unidentified phosphorylated residues of NHE3 (X. Li, A. Pandey, and M. Donowitz, unpublished data). The role of basal phosphorylation is unclear since mutagenesis which eliminated up to six putative PKC phosphorylation sites and sites phosphorylated by cAMP and other putative Ser phosphorylation sites did not alter basal NHE3 activity (507). This led to the conclusion that some basal phosphorylation does not have a major role in basal NHE3 activty. In contrast, the protein phosphatase inhibitor okadaic acid acutely stimulated NHE3 activity, demonstrating that at least some kinase(s) stimulate NHE3 activity under what is considered to be basal conditions (292).

In analyzing the role of stimulated phosphorylation of NHE3, Moe (330) appropriately pointed out that most rapid stimulation and inhibition of NHE3 are coupled to protein kinases, and pharmacological inhibitors of kinases prevent many examples of NHE3 regulation. He categorized the questions related to the role of phosphorylation of NHE3 in its regulation into 1) Is the site phosphorylated in vivo? 2) Is phosphorylation of a site of NHE3 necessary for a regulatory function? 3) Is phosphorylation of a site of NHE3 sufficient for a regulatory function? The overview, not surprisingly, given that complexes and associated regulatory proteins are known to be involved in NHE3 regulation and that the endocytic machinery itself is known to be regulated by phosphorylation, is that while phosphorylation of NHE3 appears to be necessary for some but not all its acute regulation, NHE3 phosphorylation may not be sufficient for either stimulation or inhibition (330, 507, 525). Interestingly, Netphos 2.0 software predicted 19 COOH-terminal Ser/Thr phosphorylation sites of rabbit NHE3 with probabilities of phosphorylation of >92%. All currently known phosphorylated residues of NHE3 are among the predicted list. This suggests that the role of Ser/Thr phosphorylation of NHE3 is only partially defined.

1. Evidence that NHE3 is phosphorylated as part of its acute regulation


In vitro, purified cAMP-dependent protein kinases phosphorylate NHE3 in the COOH terminus. More importantly, in intact cells exposed to elevated cAMP, immunoprecipitated NHE3 showed increased phosphorylation by specific phosphopeptide mapping, direct tandem mass spectroscopic analysis, and anti-phospho antibodies to specific NHE3 amino acids (97, 212, 264, 271, 330, 506, 525, 547). Moreover, a back-phosphorylation approach showed that PKA elevation in vivo led to occupation of PKA sites in NHE3 and inhibited subsequent in vitro phosphorylation (497, 500). The exact amino acid residues that are phosphorylated by cAMP elevation remain somewhat controversial. The most definitive studies showed that cAMP caused concentration-dependent in vivo phosphorylation of rat NHE3 Ser605 and Ser552, which are in PKA consensus sequences (330, 543). Both were necessary for cAMP to maximally inhibit NHE3, since mutation of these Ser individually decreasing and both together preventing NHE3 inhibition by cAMP. In contrast, Kurashima et al. (271) showed that NHE3 truncated to amino acid (aa) 585 lost all cAMP inhibition and that Ser605 and Ser634 of rat NHE3 were necessary for NHE3 inhibition by cAMP (271). They showed that the Ser605 was phosphorylated in the presence of PKA. However, although Ser634 was not phosphorylated (either basal or with cAMP), it was also necessary for cAMP to fully inhibit NHE3. Each of these Ser contributed ∼50% to cAMP inhibition of NHE3, with mutation of both totally preventing cAMP-induced changes in NHE3 activity in fibroblasts. In these studies, Ser552 was not phosphorylated. The reason for the discrepancy is unclear, although phosphopeptide mapping demonstrates that NHE3 that is immunoprecipitated from cells with elevated cAMP has changes in multiple phosphopeptides, suggesting changes in phosphorylation at multiple sites (272, 525, 547). It is of note that mutation of Ser from other putative PKA phosphorylation sites in the NHE3 COOH terminus did not alter cAMP effects on NHE3 activity (Ser-330, -514, -576, -662, -691, -692, -805). It remains to be determined whether phosphorylation of the Ser residues occurs as part of acute NHE3 regulation by mechanisms that do not involve changes in cAMP. As described in more detail below, cAMP-induced phosphorylation of NHE3 is necessary for all functional effects of cAMP on NHE3 described. Moreover, in OK renal proximal tubule cells, elevating cAMP with dopamine inhibited NHE3 over the same time and concentration range that increased NHE3 phosphorylation (212).


cGMP kinase II (cGKII) is a BB protein that is involved in acute NHE3 inhibition in response to the luminally released gut humoral substances guanylin and uroguanlyin. Guanylin is released as part of digestion. Heat-stable Escherichia coli enterotoxin is similar structurally to guanylin/uroguanylin and binds to the same BB receptor, guanylate cyclase C. It increases the cGMP content in the cell at the apical domain and activates cGKII. This activation leads to acute inhibition of NHE3 by a process in which NHE3 is phosphorylated (B. Y. Cha, H. Kochinsky, P. Aronson, H. de Jonge, and M. Donowitz, unpublished data).


Elevated Ca2+ inhibits NHE3 by a PKC-α-dependent process (250, 286). PKC-α becomes part of the NHE3 complexes with elevated Ca2+ in both PS120 cells and ileal Na+ absorptive cell BB (286, 298). However, PKC phosphorylation of NHE3 could not be demonstrated in PS120 cells or in ileal BB, and mutagenesis studies of NHE3 suggest that PKC phosphorylation of NHE3 does not correlate with NHE3 regulation (507, 525). Wiederkehr et al. (507) showed that purified PKC phosphorylated similar phosphopeptides in the NHE3 COOH terminus in vitro that occurred with phorbol ester exposure to intact cells in vivo. However, Yip et al. (526) did not see consistent changes caused by phorbol ester exposure on NHE3 phosphopeptides from NHE3 transfected PS120 cells [although in retrospect this may not have occurred due to lack of expression of sufficient Na+/H+ exchanger regulatory factor (NHERF) proteins in these cells; see below]. In detailed studies using AP-1 cells, mutation of six Ser in the NHE3 COOH terminus which were in putative PKC consenses sequences prevented phorbol ester regulation of NHE3 in vivo (Ser-13, -552, -575, -661, -690, -804) (507). However, PKC has been associated with NHE3 stimulation, inhibition, or no effect depending on the cell model studied. Also, Wiederkehr and Moe and co-workers (507) importantly showed that when NHE3 was expressed in AP-1 cells and individual clones selected, phorbol esters stimulated, inhibited, or had no effect on NHE3 activity depending on the specific clone studied. In spite of this, phorbol esters stimulated NHE3 phosphorylation with identical phosphopeptide maps, and this was regardless of whether phorbol esters stimulated, inhibited, or had no effect on NHE3 activity (507). Thus identical changes in NHE3 phosphorylation were associated with the full spectrum of changes in NHE3 activity. Moreover, in preliminary studies of effects of elevated Ca2+ on NHE3 regulation in PS120 cells and OK cells in which NHERF2 complexes are required for inhibition, we could not demonstrate changes in phosphorylation of IP NHE3. Thus the conclusion for Ca2+/PKC inhibition of NHE3 is that while PKC (PKC-α) is involved and joins the NHE3 complexes, the PKC phosphorylated substrate does not appear to be NHE3, although at least one of the Ser that can be phosphorylated by PKC is necessary for the inhibition of NHE3.

That NHE3 phosphorylation is necessary for some acute regulation but not others eliminates the possibility that phosphorylation is sufficient for all NHE3 regulation. Even for cAMP inhibition, phosphorylation is not sufficient. For instance, in OK cells, dopamine elevated cAMP content acting both by DA-1 and DA-2 receptors, with the effects on NHE3 being synergistic. DA-1 inhibited NHE3 activity, while DA-2 itself had no effect on NHE3 activity but DA-1 plus DA-2 caused a greater effect than DA-1 alone. In spite of this, DA-1 and DA-2 individually increased phosphorylation of NHE3 on the same sites as when they were studied together (212, 330). These findings, plus the failure of PKC phosphorylation to consistently affect NHE3 activity (507), demonstrate that phosphorylation of NHE3 is not sufficient to explain NHE3 regulation. In addition, the quantitative role of NHE3 phosphorylation in NHE3 regulation needs further examination.

In spite of the correlation of phosphorylation of specific amino acids of NHE3 as part of its regulation with changes in turnover number, for instance, with cAMP inhibition, the series of molecular events that lead to the regulation downstream from the changes in phosphorylation are not understood. The changes in NHE3 trafficking are partially understood at the level of changes in associating proteins and changes in associating with some parts of the trafficking machinery, although again molecular details are lacking (see below).


Several other kinases also either associate with or regulate NHE3 activity. PI 3-K, a major NHE3 regulating enzyme, is discussed below. A kinase which binds directly to NHE3 is calmodulin (CaM) kinase II (545), while other kinases which join NHE3 are part of complexes involved in regulation. These include PKA, PKG, and PKC, all of which associate with NHE3 via BB PDZ proteins of the NHERF family. Understanding how kinases regulate proteins now includes understanding how kinases are activated locally in cells with the consequence that specific signaling pathways appear to involve a few substrates, rather than all putative cell substrates. This specificity is attributed, at least partially, to activation of kinases at specific locations in the cells as part of specific receptor/signaling pathways. The location of NHE3 complexes with BB PDZ proteins and their association with kinases may partially explain this phenomenon for NHE3 at the BB. In addition, multiple other kinases are involved in acute NHE3 regulation, although apparently several steps away from NHE3. For instance, already mentioned is that d-glucose activates Caco-2 BB NHE3 by a process that involves the p38 MAP kinase, MAPAKP2, PI 3-K, and Akt2. Also, endothelin both stimulates and inhibits NHE3 in a dose-dependent manner by a process that involves the Ca2+-sensitive kinase PYK2 (274). Thus phosphorylation of NHE3 by kinases that associate with it via its signaling complexes or serve as intermediates in the signaling networks are necessary for some but not all aspects of acute NHE3 regulation both via changes in turnover number and by trafficking. This incompletely defined area requires much further study to define other associating kinases and phosphatases and how changes in phosphorylation leads to changes in NHE3 complex formation, trafficking, turnover number, as well as how the NHE3 signal complexes are involved in controlling NHE3 phosphorylation as well as in phosphorylation of the NHE3 regulatory machinery. NHE3 stimulation by SGK1 is described in section vB8cVI.


Another understudied area is the role of regulated phosphatase activity in NHE3 regulation. Protein phosphatase (PP) 2A was reported as associating with NHE3 when cAMP was elevated acutely (380). The role of this association has not been defined. However, the PP1 and PP2A inhibitor okadaic acid stimulates NHE3 activity (292). This suggests that basal phosphatase activity inhibits NHE3 activity, although it was not established whether the phosphatase involved physically associated with NHE3 or whether it was changes in NHE3 phosphorylation or in NHE3 regulatory proteins that were responsible for changes in NHE3 activity. Given the large number of kinases shown to associate with NHE1, at least calls for further examination of kinase and phosphatase binding partners of NHE3, both directly and as part of its signaling complexes.

In addition, not yet explored is the role of localization of NHE3 into lipid rafts (LR) in the BB in determining the state of NHE3 phosphorylation under basal or stimulated conditions. Some NHE3 is in LR and is involved in basal cycling and endocytosis and growth factor stimulation of NHE3 (295, 339). While NHERF1 and NHERF2 in the BB do not appear to be LR associated (295), the state of the other NHERF family members is not known, nor is it known whether the kinases involved with NHE3 regulation are in apical membrane LR.

B. Acute Regulation of NHE3 Can Occur by Changes in Trafficking or Changes in Turnover Number

In all cells and tissues in which NHE3 has been studied, NHE3 stimulation and inhibition are at least partially due to changes in trafficking, with separate endocytic and exocytic processes that can be regulated independently (88, 97, 121, 226, 270, 537). In many cases there is also regulation by changes in NHE3 turnover number, which in some cases occurs faster than changes in rates of trafficking (121, 537).

In evaluating mechanisms of regulation of NHE3, the model used determines some of the results. Of the models used, changes in turnover number as well as regulated endocytosis and exocytosis have been demonstrated in the renal proximal tubule cell line, OK cells, the intestinal Na+ absorptive cell line, Caco-2 cells, intact small intestine, and NHE3 null fibroblasts transfected with NHE3. In all cells/tissues, NHE3 exists both on the plasma membrane and in intracellular vesicles, which are probably endosomes, under basal conditions.

The tissue which NHE3 seems to be differently regulated from others is intact renal proximal tubule (45, 288290, 321, 322, 519, 521, 522, 526, 539541). This in spite of the fact there is an intracellular pool of NHE3, similar to other cells, which is probably endosomal, as has been demonstrated by electron microscopy (47, 48). In intact proximal tubule, NHE3 seems to cycle predominantly between microvilli and intervillus areas. In the apical membrane of proximal tubule cells, Biemesderfer et al. (45) showed that NHE3 exists in two pools, thought to be in the microvilli and in the intervillus spaces (45). In the former, NHE3 is in lighter complexes (9. 2s), does not associate with megalin, and appears to be active. In contrast, in the latter, NHE3 is in heavier complexes (18s), associates with megalin, and appears to be less active or inactive. As part of signaling, NHE3 in proximal tubule partially moves from a lighter distribution on sucrose density gradients that includes apical membrane markers, to heavier fractions which have more markers of the intervillus compartment but have some overlap with both BB and intracellular endosomal markers. In these density gradient studies initiated by Mircheff and pursued by McDonough, NHE3 was shown to be present in three fractions. The debate is what they represent and how distinct they are: alkaline phosphatase is enriched in the microvillus fraction, galactosyltransferases are enriched in microvillus clefts, and acid phosphatase is enriched in endosomes. As an example, with PTH treatment, ∼20% of total NHE3 shifts from low-density membranes enriched in microvillar markers (541) to mid-density membranes enriched in intermicrovillar cleft markers and later to heavier density membrane with some endosomal markers. Models used to evaluate changes in NHE3 trafficking in proximal tubule have included acute hypertension-induced natriuresis, PTH, and an angiotensin-converting enzyme inhibitor, all of which caused rapid and reversible inhibition of NHE3 activity and redistribution of NHE3 in the sucrose gradients from lighter to heavier density fractions interpreted as away from BB to intervillus clefts. McDonough and co-workers (519, 526, 539, 540) showed that 20 min of acute hypertension led to redistribution of NHE3 from BB to the base of the BB, while a control protein, NaPi2a, internalized into endosomes. However, how much of each fraction is BB versus microvillus clefts versus endosomes was not resolved.

Of importance, the Biemesderfer studies characterized the Triton X-100 soluble pools of NHE3 (45, 46). Only ∼50% of total proximal tubule NHE3 was Triton X-100 soluble. Thus this model does not consider ∼50% of BB NHE3. They also demonstrated that anti-NHE3 antibodies did not recognize the entire NHE3 pool and showed that at least part of what was not recognized was bound to megalin (46). The role of megalin specifically in NHE3 regulation has not been resolved (see below).

In ileal Na+ absorptive cells, epidermal growth factor (EGF) and clonidine stimulated active NaCl absorption and BB NHE3 activity, which correlated with the increase in BB NHE3 amount (130, 246, 295). This stimulation was LR dependent, being inhibited by disrupting LR with methyl-β-cyclodextrin (MβCD) (295, 340). Basal endocytosis was also LR dependent with the amount of NHE3 coimmunopurified with EEA1 containing vesicles decreasing with MβCD and with cytoskeleton disruption with cytochalasin D (295, 340). Thus, in ileum, unless antibodies are blocked from recognizing large populations of NHE3, the addition of NHE3 to the BB as part of rapid stimulation is not associated with shifts within the BB of NHE3 pools characterized by differences in detergent solubility. Whether Biemesderfer's and McDonough's careful studies in proximal tubule and our ileal studies of the EGF and clonidine induced increases in BB NHE3 are describing comparable pools of NHE3 is not known. However, the current view is that ileal NHE3 but not renal proximal tubule NHE3 traffics between endosomes and BB. It is also not known whether ileal NHE3 exists in a large intermicrovillar membrane pool, as occurs in proximal tubule. In addition, since a significant portion of renal NHE3 is Triton X-100 insoluble, the role of LR in renal NHE3 trafficking should be determined.

Oppositely, carbachol rapidly decreased the amount of NHE3 in the BB and increased the amount coimmunopurified with EEA1, suggesting regulation by trafficking (298). However, whether the regulated process is endocytosis, exocytosis, or both in intact tissue is still not established. Also, in rabbit ileum, the amount of NHE3 in BB increases over minutes with EGF and clonidine and decreases with carbachol (295, 298). This supports trafficking of NHE3 as a mechanism involved with acute NHE3 regulation. This is specific for ileum, and mechanisms in the ileum and proximal tubule could be separate. A portion of ileal Na+ absorptive cell NHE3 is in the same fractions of sucrose density or OptiPrep gradients as EEA1 (30%), and some NHE3 can be coimmunopurified with endosomal vesicles by using EEA1 antibody coupled to beads (295). The amount of NHE3 copurified with EEA1 vesicles based on immunopurification and colocalized with EEA1 based on sucrose density gradient was reduced by cytochalasin D exposure. These results strongly indicate that there is an early endosomal pool of NHE3.

In epithelial cell culture models, basal NHE3 activity is regulated by a balance of basal endocytosis and exocytosis. A major difference that has been recognized among cell and tissue models of NHE3 is the percent of NHE3 on the plasma membrane under basal conditions. This includes ∼15% in fibroblasts (PS120 and AP-1 cells), 15–50% in OK cells based on the application of both cell surface biotinylation and confocal microscopy, ∼80% in Caco-2 cells measured both by cell surface biotinylation and confocal microscopy, ∼80% in renal proximal tubule, and ∼80% in rabbit ileal BB (7, 8, 116, 197, 371, 523). In OK cells, NHE3 is in BB and an intracellular pool that colocalizes with Rab 11 (8). In BB, the amount of NHE3 in the central portion of BB over the recycling compartment may be increased compared with in other parts of the apical membrane (8). Based on fluorescence recovery after photobleaching (FRAP) and use of cross-linkers, it appears that basal trafficking is preferentially to this area of the BB where NHE3 density is increased (71). This suggests the presence of a targeting signal for NHE3 in this localized domain in the BB.

Acute regulation of NHE3 in OK cells initially involves changes in turnover number or amount based on the stimulus followed by changes in trafficking. Endothelin and metabolic acidosis increased the surface amount of NHE3 and in parallel increased NHE3 activity (371, 523). PTH inhibited NHE3 acting by cAMP. At 30 min NHE3 activity was decreased, but surface amount of NHE3 was not affected. Amount of BB NHE3 was reduced 4–12 h after PTH by a mechanism that involves increased endocytosis and no change in exocytosis (97). This process was dynamin dependent, being inhibited by a dominant negative dynamin mutant. Increased endocytosis of NHE3 was associated with increased association of NHE3 with clathrin and AP-2. How this time-dependent regulation of NHE3 occurs is postulated to be either initial inhibition of all the affected NHE3 molecules followed by their removal from the BB versus initial inactivation of NHE3 with subsequent removal of only some of the BB NHE3 and reactivation of the remaining. Unknown also is whether NHE3 is active or not at the site of removal and intracellularly. There is evidence that some of the recycling NHE3 is functionally active (8, 134, 481), but it is not clear if this represents the pool removed from the BB. Thus, in cell culture models, both changes in turnover number and trafficking on and off the BB has been shown to contribute to NHE3 regulation in multiple cell types from different tissues, using different ligands, and by different techniques.

A myriad of signaling molecules are linked to plasma membrane receptors, the activation of which affect NHE3 activity. In addition, however, there are signaling molecules that are involved with NHE3 regulation at or close to the apical membrane. Most clearly identified as having a role in basal NHE3 activity is PI 3-K (78, 134, 136, 178, 224, 246, 270, 284, 295, 296). In all cells and most tissue models of NHE3, basal NHE3 activity is under control of PI 3-K, with PI 3-K inhibitors decreasing basal transport rate and percent of NHE3 on the plasma membrane. In rabbit ileum, wortmannin did not alter basal NaCl absorption, which must be confirmed as this is the only example of PI 3-K not being involved in basal regulation of NHE3 (246). In contrast, in NHE null fibroblasts in which NHE3 is expressed (PS120 and AP-1), Caco-2 cells, and OK cells, wortmannin inhibits basal NHE3 activity by ∼50% (8, 130, 224, 270). PI 3-K is also involved in stimulated NHE3 activity. Constitutive activation of PI 3-K or Akt stimulates NHE3 activity and increases plasma membrane amount in fibroblasts and OK cells, while growth factor stimulation of NHE3 in fibroblasts is blocked by ∼50% by PI 3-K inhibitors (224, 284), and lysophosphatidic acid (LPA) stimulation of NHE3 in OK cells is 100% blocked by wortmannin (285).

Comparison of NHE3 trafficking mechanisms should be made with GLUT4, the best studied transporter that trafficks to the plasma membrane, which occurs through insulin-stimulated exocytosis (80, 222, 233, 420, 461, 486, 487, 544). The GLUT4 studies are reviewed to serve as background to consider the various pathways and regulatory proteins known to be involved in regulation of trafficking of a transport protein, although it must be pointed out that NHE3 trafficking between the plasma membrane and intracellular pools is much less dramatic than regulated GLUT4 trafficking. A minimal amount of GLUT-4 is on the plasma membrane under basal conditions, with minimal basal exocytosis from the recycling compartment. In contrast, after insulin stimulation, GLUT-4 traffics via both a storage vesicular pool and increased exocytosis from the recycling compartment. In NHE3, Alexander and Grinstein (11) performed a detailed kinetic analysis of NHE3 expressed exogenously in the distal renal tubule cell line MDCK (11). They identified four pools of NHE3: 1) a mobile apical membrane population, the trafficking of which could be rapidly stimulated; 2) an immobile apical membrane pool that binds the cytoskeleton by interaction with Rho GTPases; and 3 and 4) two intracellular pools with a fast and less rapid exchange rate with the apical membrane. The relevance of this distal tubule model to normal physiology of the proximal tubule is unknown. Also unknown is how similar is the mechanism of regulation of exocytosis of NHE3 to the GLUT-4 model, which predicts a storage pool that as yet is not well characterized biochemically and the recycling endosomes as distinct, although perhaps interacting pools.


Does intracellular NHE3 transport Na+ and H+? Our interpretation is that the answer is yes, although intracellular NHE3 function and its regulation have not been studied in detail. Some of the reasons to suspect that this pool of NHE3 is functional include the following. 1) In several cell culture models, inhibition of NHE3 alkalinizes an intracellular compartment thought to be early endosomes (8, 134, 163, 164). This indicates NHE3 normally acidifies this compartment. 2) In proximal tubule of wild-type mice, NHE3 inhibition by cAMP increased early endosomal pH within 1 min, as marked by FITC-BSA, implying cAMP inhibits both BB and intracellular NHE3 activity and demonstrates that intracellular NHE3 is active. Note this effect is too fast to be explained by changes in trafficking (164). 3) In NHE3 knock-out mouse, there is reduced renal proximal tubule albumin uptake (165). Whether this involves intracellular NHE3 is not known. 4) In proximal tubule of ClC-5 knock-out mouse, NHE3 has primarily an intracellular location (373), which has been assumed to be in some pool of endosomes and does not resemble the distribution of Golgi. This intracellular pool is still able to allow albumin endocytosis, which is strong evidence that intracellular NHE3 functions.


The NHE3 COOH terminus is necessary for all identified examples of acute NHE3 regulation (293, 534). This generally has been examined by truncation or point mutants of the NHE3 COOH terminus. Truncation of the NHE3 COOH terminus alters basal as well as regulated NHE activity. Truncation to aa 756 increases NHE3 activity twofold, while truncation to aa 690 increases it to sixfold of basal (7). This is accompanied by similar increases in the amount of NHE3 on the plasma membrane in PS120 fibroblasts (2 × for 756, 6 × for 690) with the transport per plasma membrane exchanger remaining constant. Further truncating NHE3 to aa 585 did not further alter the percent of NHE3 on the plasma membrane or the activity per plasma membrane exchanger, although total expression was diminished. This suggested that there are two endocytosis signals in the NHE3 COOH terminus, one between aa 756–832 and the other dominant signal between aa 690 and 756. Additional truncation of NHE3 to aa 509, 475, and 455 decreased NHE3 activity, with NHE3 445 activity minimally detectable in the absence of stimulation with serum. As shown in Figure 3, study of truncation mutants demonstrated that the multiple acute regulators of NHE3 activity require specific domains of the COOH terminus to act. The sole exception identified in any mammalian NHE is the domain required for NHE response to hyperosmolarity. While not studied in NHE3, the effect of hyperosmolarity to stimulate NHE1 and inhibit NHE2 was shown to involve the first extracellular loop between aa 41–53 of NHE1 (aa 31–43 of NHE2) (422). This is an area of great divergence among the NHEs. This domain is believed to represent the first extracellular loop for NHE1, which is not believed to contain a cleavable signal peptide, while we found that NHE3 did have a cleavable signal peptide, thus making the initial expressed part of NHE3 extracellular (475, 546). Similar results were found for Nhx1, the yeast intracellular NHE (505). Wakabayashi and co-workers also reported similar suggestive evidence for a cleaved signal peptide in NHE6, although his other evidence does not support this view (328, 475). This hyperosmolar response of NHE1/NHE2 is the only NHE regulation that does not depend entirely on the intracellular COOH-terminal domain (422). Not only is the NHE3 COOH terminus necessary for rapid regulation of NHE3, but it defines the nature of the regulation. Swapping the COOH terminus of NHE3 with that of NHE1 or NHE2 led to regulation consistent with the NHE providing the COOH terminus (534).

FIG. 3.

A: NHE3 organization including COOH-terminal domains involved in acute regulation and binding partners. The NHE3 NH2 and COOH termini are illustrated. Shown above the COOH terminus in blue are sites of action of regulatory stimuli defined by the truncations which abolish the specific regulatory effects. The direction of the arrow indicates acute stimulation (↑) or inhibition (↓) of NHE3 activity. Below the COOH terminus in yellow are listed associating proteins and the areas of their binding sites on NHE3. B: classification of NHE3 COOH terminus into 4 domains for discussion of binding partners.

A. Regulatory Domain

The NHE3 (rabbit) COOH terminus is predicted to begin at aa 455 based on the assumption of similarity in structure to the crystallized bacterial Na+/H+ exchanger, NhaA, which lacks any significant intracellular COOH-terminal domain (216). The NHE3 COOH terminus from aa 455–832 (378 aa) almost certainly has tertiary structure, although this has not been determined using physical techniques. Programs that predict secondary structure indicate that the first ∼130 aa after the transport domain is predominantly α-helical (74). In contrast, the NHE1 COOH terminus between aa 503–545 was crystallized (cocrystallized with its binding partner CHP2) and shown to be α-helical as was predicted from secondary structure predicting programs (18, 40).

The NHE COOH terminus has multiple functional subdomains. Initial hints of the existence of functional subdomains in the COOH terminus come from COOH-terminal truncation studies of NHE1. Based on the intracellular pH (pHi) sensitivity and changes in acute NHE1 regulation, Ideka et al. (219) divided the NHE1 COOH terminus into four functional domains, and these are discussed on the assumption that all NHEs probably have some similarities in the organization of their intrinsic regulatory mechanisms (219). These four domains of NHE1 are as follows: 1) aa 515–595, 2) aa 596–635, 3) aa 636–659, and 4) aa 660–815 (219). Subdomain 1 had a pHi maintenance function, preserving pHi sensitivity in the physiological range, whereas subdomain 2, overlapped with the high-affinity CaM-binding site in subdomain 3, and exhibited an autoinhibitory function. Subdomains 2 and 4 had no function for pHi sensitivity. Deletion of subdomain 1 abolished the decrease of pHi sensitivity induced by cellular ATP depletion, suggesting that this subdomain is involved in ATP regulation of NHE1. Deletion of subdomain 3 rendered the inhibition by ATP depletion less efficiently, suggesting a possible functional interaction between subdomains 1 and 3. However, deletion of subdomain 2 partially abolished the inhibitory effect of subdomain 3. Therefore, subdomain 2 was proposed to function as a mobile “flexible loop,” permitting the CaM-binding subdomain in subdomain 3 to exert its normal function. These findings further showed that the COOH-terminal domain of the NHEs plays a key role in regulating NHE activity. Fliegel and associates (294, 297) examined the structure of NHE1 aa 516–815 by circular dichroism spectroscopy and concluded that it was 35% α-helix, 17% β-turn, and 48% random coil. They concluded that NHE1 was largely α-helical near the NH2-terminal transport domain (aa 516–632), while aa 633–815 was largely β-sheet. Moreover, elevating Ca2+ decreased the amount of α-helix and increased the amount of β-sheets in the COOH terminus.

The COOH terminal 377 aa of NHE3 are intracellular, based on accessibility to antibodies in the unpermeabilized compared with permeabilized state. A single study, using monoclonal antibodies, suggested that some COOH-terminal epitopes were accessible to extracellular antibodies, and by implication that this part of NHE3 contained extracellular epitopes (48). There has been no explanation provided for this finding. The cytosolic NHE3 COOH terminus interacts with proteins, which play essential roles in regulating NHE3 activity, trafficking, and phosphorylation. For NHE1, at least for CHP, this association is complicated and involves a structural contribution that is necessary for NHE1 to have any transport activity (see below). Several NHE3 COOH-terminal interacting proteins have been reported, and domains required for specific regulatory effects are beginning to be defined (Fig. 3B). Four subdomains of the NHE3 COOH terminus are arbitrarily defined in Figure 3. Detailed description of NHE3 associating proteins that interact with each subdomain are described later.

B. NHE3 Complexes

What are the mechanisms by which the NHE3 COOH terminus takes part in the acute regulation of NHE3? The identified mechanisms include 1) acting as a kinase substrate and undergoing changes in Ser/The phosphorylation, 2) interacting with the NH2 terminus, and 3) interacting with other proteins or lipids to change the location or signaling molecules associating with NHE3. This section deals with NHE3 interacting proteins and putative lipid interactions. Consistent with its association with multiple binding partners, the estimated size of NHE3 on density gradient centrifugation (OptiPrep or sucrose density gradients) indicates that NHE3 rarely exists as a monomer or dimer (∼90 or ∼180 kDa), but after solubilization in 1% Triton X-100 exists simultaneously in complexes between 350 and at least 1,000 kDa (Fig. 4A) (8, 295, 296, 298, 340). This observation was made in multiple intact tissue types and cell culture models, including mouse and rabbit small intestine, mouse proximal tubule, the intestinal epithelial cell line Caco-2, the proximal tubule cell line OK, and PS120 fibroblasts. Several characteristics of these complexes have been established. 1) In multiple cell types, the size of NHE3 complexes has been compared on the apical surface, identified by cell surface biotinylation versus the total cell pool (8, 298; Li and Donowitz, unpublished data). The plasma membrane NHE3 complexes were larger than the total pool in OK cells (the majority of the total NHE3 in OK cells is intracellular with estimates from 50 to 85% of NHE3 being intracellular) (8, 196) and PS120 cells (∼85% of total NHE3 is intracellular). 2) These complexes are dynamic with acute regulation of NHE3 (298). Evidence that the NHE3 complexes were dynamic comes from studies in which ileum was exposed in vitro to carbachol, which inhibited NHE3 by elevating intracellular Ca2+ and stimulated endocytosis, and in which OK cells were exposed to LPA, which stimulated NHE3 by increasing exocytosis. BB membranes were prepared from ileal Na+ absorptive cells, and total membrane was prepared from the OK cells; in both cases, NHE3 shifted into larger complexes as part of acute regulation of NHE3 (see Fig. 4B for carbachol).

FIG. 4.

A: NHE3 is present simultaneously in large, multiprotein complexes. Triton X-100 solubilized cells were separated on discontinuous sucrase gradients with molecular weight standards run on parallel gradients, as shown below. B: carbachol increases the size of NHE3 complexes in the ileal brush border. Ileum was exposed to carbachol (1 μM) for 10 min. Brush border was prepared and studied as in A. [From Li et al. (298), with permission from Blackwell Publishing.]

Thus it appears that NHE3 exists simultaneously as multiprotein complexes in both the apical plasma membrane domain and intracellularly. The size of these complexes is larger on the plasma membrane than those inside the cell. This indicates that the complexes are not all formed in the endoplasmic reticulum/Golgi but probably are modified on arriving at or leaving the apical domain under basal and regulated conditions. Moreover, these complexes are dynamic and are further changed as part of signal transduction that acutely regulates NHE3 activity. It should be considered whether the association or dissociation of NHE3-interacting proteins, which determine the size of NHE3 complexes, may define the specificity of the different modes of NHE3 regulation described above. Of interest is that NHE1 and NHE2 also form complexes (38, 378). With the use of identical conditions to compare complexes of NHE1, -2, and -3 expressed in PS120 cells, NHE1 and -2 formed much smaller complexes with smaller ranges in size than did NHE3 (Li and Donowitz, unpublished data).

1. How many proteins are there in the NHE3 complexes?

With the assumption that the average size of an NHE3-interacting protein is 70 kDa, the number of NHE3 proteins associating with a dimer of NHE3 (NHE3 like all mammalian NHEs characterized exists as a dimer under basal conditions) at any point in time is estimated to be ∼3 to ∼10 (based on complex sizes ranging from 350 to 1,000 kDa minus the size of a dimer of NHE3 divided by 70 kDa). It is still to be determined why NHE3 complexes exhibit such a large range in size, the meaning of simultaneous isolation of many sized complexes containing NHE3, and how these complexes are regulated. Our current view is that different combinations of known and yet to be identified NHE3-interacting proteins exist simultaneously in NHE3 complexes producing different sizes of complexes. In fact, given the possibility of some complex degeneration during processing, we hypothesize that even more proteins associate with NHE3.

C. Intramolecular Interactions

Does dimerization or oligomerization of NHE3 occur in vivo and contribute to the formation of the large complexes of NHE3? Disuccinimidyl suberate (DSS) cross-linking showed that homodimerization exists for NHE1 and NHE3 (146). The NH2-terminal transmembrane domain is sufficient for this dimerization since deletion of the 300 aa COOH terminus of NHE1 did not alter dimerization. However, the COOH terminus appears to take part in dimerization. Domain 1 of the NHE1 COOH terminus, aa 562–579, was identified as the “dimerization domain” (208). Deletion of this region disrupted disulfide cross-linking between the COOH termini and functionally markedly reduced the intracellular pH sensitivity of the exchanger, suggesting that this aspect of the dimer might be necessary for the pH-dependent regulation of NHE1 (208). Multimerization of NHE3 was also reported when NHE3 was analyzed by SDS-PAGE and Western blotting under denaturing conditions (250); however, whether this represents a true picture of in vivo complex formation is not known.

Using cross-linking reagents such as the Ca2+ phenanthroline or the bifunctional sulfhydryl reagent methanethiosulfonate, Wakabayashi et al. (208) suggested that NHE1 forms a dimer between the COOH-terminal domain intrinsic Cys794 and Cys561. This was the first evidence of the existence of intermolecular interactions between the COOH terminus of NHE1. However, because NHE3 does not contain Cys residues corresponding to Cys794 and Cys561 of NHE1, similar intramolecular interactions are not expected to occur in NHE3. If there were a Cys-mediated COOH-terminal dimerization in NHE3, the only potential residue would be the sole Cys595 in the COOH terminus of NHE3. It is conserved in NHE3 and NHE5 across mammalian species including human, rat, and rabbit. Whether intermolecular interaction of the NHE3 COOH terminus occurs by other mechanisms should be considered as well. Oligomerization of NHE3 could occur indirectly via its COOH-terminal binding proteins. Since the NHE3 COOH terminus interacts with PDZ domain containing proteins, including NHERF1, NHERF2, NHERF3, and NHERF4, and each has two to four PDZ domains, NHE3 could oligomerize indirectly through different PDZ domains bound to its COOH terminus. It remains to be determined whether such oligomerization of NHE3 occurs in cells or intact tissue models.

D. Dynamic Aspects With Signaling

Our previous studies in rabbit ileal villus cells demonstrated that the NHE3 complex sizes were dynamically regulated (298). Exposure to carbachol, which inhibits NHE3 activity and LPA, which stimulates NHE3, both resulted in an increase in sizes of NHE3 complexes compared with control. We have to emphasize that, when no change in complex size is observed upon specific treatments, it does not necessarily mean there is no change in the number or composition of proteins in the complexes. The sucrose gradient fractionation approach can only reliably detect relatively large changes in complex size that resist exposure to 1% Triton X-100 (≥100 kDa), not the subtle changes that may result from addition and/or subtraction of small proteins or change in protein composition (i.e., addition of one protein and dissociation of another with similar molecular weight). We hypothesize that 1) the dynamic nature of NHE3 complexes is consistent with NHE3 trafficking under basal and stimulated/inhibited conditions that occur by different mechanisms, with separate signaling events. 2) The large range in complex sizes may reflect dynamic NHE3 cellular distributions, where the number and composition of NHE3-interacting proteins in one location (such as the plasma membrane) is different from those in other subcellular compartments (such as recycling endosomes). An example is the difference in complex size of NHE3 on the apical membrane of OK cells (larger) compared with that intracellularly, suggesting different NHE3 interacting proteins are involved in NHE3 regulation in those two domains (8). 3) Upon changes in signaling, the number/composition of NHE3-interacting proteins changes, and such changes may lead to either changes in surface amount of NHE3 or direct modulation of the transport domain, thereby resulting in the changes in transport activity.


To allow easier molecular manipulation, we divided the NHE3 COOH-terminal domain into four subdomains and will describe interacting proteins with each (using rabbit NHE3 as an example): 1) aa 455–585 (131 aa), 2) aa 586–660 (75 aa), 3) aa 661–756 (96 aa), and 4) aa 757–832 (76 aa). These four NHE3 COOH-terminal subdomains were not selected by function or structure, although there are some relationships with the subdomains of NHE1 characterized by Wakabayashi and co-workers (219). Study of the regulation of NHE3 mutants with serial COOH-terminal truncations defined some of the roles of NHE3 COOH-terminal subdomains in regulating NHE3 stimulation and inhibition (7, 293). There appears to be at least three stimulatory domains in the COOH terminus of NHE3, including aa 455–475 for fetal bovine serum (FBS), aa 475–509 for fibroblast growth factor (FGF), and aa 509–543 for okadaic acid (OA) (293). There are also two inhibitory domains, including aa 585–689 for PKA and PKC, and aa 661–757 for CaM kinase II (271, 292, 545). Although it is not yet known how these examples of regulation are executed at a molecular level, we hypothesize that they may be exerted through regulatory proteins that interact with the corresponding regions. Such NHE3 regulatory proteins have been demonstrated by several groups to bind specific COOH-terminal subdomains and are either required for basal or regulated NHE3 activity. Caution has to be taken to interpret the results of studies on COOH-terminal truncation or deletion mutants since it is possible that a deleted region may be required for the integrity of a tertiary structure that supports a particular regulatory function but is not itself physically involved in such regulation.

So far, at least nine different proteins have been reported to directly interact with the NHE3 COOH terminus. These include CHP, ezrin, CaM, CaM kinase II, NHERF family members, dipeptidyl peptidase IV (DPPIV) (may be indirect), PP2A, megalin, and Shank2. The list is expected to grow. Also, it has been suggested that NHE3 binds to phosphoinositol phosphates.

A. COOH-Terminal Domain 1, Amino Acids 455–585: CHP, Ezrin, and Phosphoinositol Phosphates

This domain is important for NHE3 basal function, since multiple point mutations in this domain decrease NHE3 function. The identified binding partners for NHE3 in this part of the molecule include CHP and ezrin. In addition, it has been suggested that signaling phosphatidylinositides bind here as well (5). This NHE3 COOH-terminal domain is a highly hydrophobic region and contains multiple areas predicted to be α-helical, including a comparable area to NHE1 aa 503–545, which was shown to be α-helical by crystallization to 2.7 Å when in a complex with CHP2 (18, 40). This domain has been shown to be necessary for NHE3 regulation by FBS, FGF, and OA based on truncation studies (293) and ATP (108) (Fig. 3A).

1. CHP

CHP, calcineurin homologous protein (p22), is a calcium binding, multi-EF hands domain containing protein that binds to and regulates trafficking and function of multiple classes of proteins including NHE1 to NHE5 (304, 318, 344, 363). CHP binding to NHE3 is necessary for NHE activity. There are three CHP isoforms (CHP1 to CHP3, with CHP3 also called tescalcin) (318, 344). CHP1 and CHP2 have been crystallized and have an α-helical structure with 4-EF hand domains and resemble the calcineurin regulatory subunit (18, 40, 344). EF hands domain 1/2 and 3/4 pair with each other, although only EF hands 3 and 4 bind calcium via a hydrophobic pocket on the opposite side of the molecule from that which binds ligands (18, 40).

Association of NHE1 with CHP1 is needed for maintenance of the pHi sensitivity of NHE1 (362364). This led to the suggestion that CHP with tightly bound Ca2+ may serve as an important structural element in the “H+ modifier” of NHE1 (40, 362). The “H+ modifier,” a putative allosteric NHE regulatory site, was proposed to contain two binding sites for internal H+ based on kinetic studies of renal BB NHE and is suggested to include NHE1 aa 433–450 [intracellular loop 5′ (IL5)] and aa 451–470 [transmembrane domain 11 (TM11)] of the NHE1 NH2 terminus (363, 471, 472, 473). Extensive mutational analysis showed that mutation of Arg440 (R440D) in IL5 caused a large acidic shift of the pHi profile for 22Na+ efflux. The equivalent “H+ modifier” in NHE3 is located at aa 389–405 (IL5) and aa 406–425 (TM11), with 41 and 62% aa identity, respectively (Fig. 5). This suggests the “H+ modifier” is structurally and functionally very similar in NHE1 and NHE3. Both CHP1 and CHP2 bind in this area to NHE1–NHE5. NHE3 or NHE1 mutants, in which hydrophobic residues were altered to disrupt NHE3-CHP interaction, have little Na+/H+ exchange activity, suggesting the CHP binding is required for basal NHE activity (304, 362). It is suggested that CHP1 always contains two Ca2+ with high affinity (Kd = 90 nM) when associated with NHE1 in cells. Overexpression of green fluorescent protein (GFP)-tagged CHP1 with mutations in EF3 or EF4 that reduce Ca2+ binding significantly reduced the exchange activity in the neutral pHi range and partly impaired the activation of NHE1 in response to various stimuli, such as growth factors and osmotic stress (362).

FIG. 5.

Sequence alignment of NHE1 and NHE3 in putative H+ modifier site and CHP-binding domains.

These studies were supported and extended by structural studies. CHP2-NHE1 binding has been characterized by crystallization by Wakabayashi's group (18, 40). The CHP binding domain of NHE1 is located at aa 515–542, a hydrophobic aa-enriched area. A similar CHP-binding domain also exists in NHE2, -3, -4, and -5, with corresponding areas of aa 495–522 (for NHE2), aa 473–500 (for NHE3), aa 486–513 (for NHE4), and aa 465–492 (for NHE5) (18). The His6-tagged human CHP2 (aa 1–196) complexed with its binding domain-containing sequence of NHE1 (aa 503–545) was crystallized in the presence of yttrium (Ca2+-bound state), with a resolution of 2.7 Å (18, 40). The NHE1-binding sequence of CHP2 is folded into two globular domains (N- and C-lobes) consisting of 12 α-helices and four β-sheets. The CHP-interacting domain of NHE1 forms a α-helix, which is oriented into the hydrophobic cleft encompassing the N- and C-lobes of CHP2. The stability of the complex is maintained by a large area of hydrophobic interactions between NHE1 and CHP, including amino acids of NHE1: Ile-518, Ile-522, Phe-526, Leu-527, Leu-530, Leu-531, Ile-534, and Ile-527. There is also potential interaction between a unique region of CHP2 (aa 94–104) and IL5 of NHE1. From the crystal structure, the CHP2-unique region (the 7th α-helix) is oriented far outside of the hydrophobic cleft, leading to the possibility that this α-helix interacts directly with IL5 of NHE1. Indeed, it was observed that NHE1 and CHP cross-linked with each other through IL5 of NHE1 and the unique region of CHP2. Furthermore, either deletion of the CHP-unique region (18) or mutation of Arg-440 in IL5 of NHE1 (472) resulted in similar large acidic shifts of the pHi-dependent Na+ uptake and efflux. On the basis of these observations, two important roles of CHP1/2 in NHE1 function were proposed (18). 1) CHP functions as an obligatory subunit (of NHE1), which is tightly associated with the COOH terminus of NHE1 and is essential for NHE1 activity in both Vmax and H+ affinity. 2) CHP may interact with intracellular H+ regulatory sites of NHE1 (such as IL5 or other ILs), through the CHP-unique region, and is involved in modulation of the H+ affinity, but not Vmax.

Recently, using nonoverexpressing NHE1 cell lines, CHP1 was shown to have an additional and perhaps primary role in stabilizing NHE1 on the plasma membrane (320a). We speculate that the role of CHP in the function of NHE3 is likely to be similar to that of NHE1 based on the following analysis. First, among the eight hydrophobic residues (7 Leu/Ile and 1 Phe of the NHE1 COOH terminus as discussed above between aa 515–542) that are thought to be involved in the hydrophobic interactions between NHE1 and CHP, seven are present in similar locations in the CHP-binding domain of the NHE3 COOH terminus. Second, IL5 of NHE3 is highly homologous to that of NHE1 (75%) and contains Arg-396 that is equivalent to Arg-440 of NHE1. Thus we suggest that IL5 of NHE3 is potentially a H+ modifier site that interacts with the CHP-unique region to modulate the H+ affinity.

The relationship between CHP1, CHP2, and CHP3 in NHE function is unknown. CHP1 is ubiquitous. In contrast, CHP2 is present mostly in apical domain of intestinal epithelial cells and has minimal expression in kidney (221). This suggests that 1) CHP2 is the primary isoform for intestinal NHE function, while 2) CHP1 is a major isoform for the function of kidney-expressed NHEs. CHP2 interestingly was also expressed in some tumor cells including hepatoma and cervical cancer, tissues that normally do not express it (364). CHP2 protects cells from serum deprivation-related cell death by increasing pHi, essentially mimicking the phenotype of malignant cells. The role of this function in physiology/pathophysiology is not understood.

Tescalcin/CHP3, a 24-kDa Ca2+ and Mg2+ binding proteins with four EF hand domains which bind one molecule of Ca2+, has been called CHP3 based on structural similarities to CHP1 and -2 (CHP1 and -2 are ∼70% identical and CHP1 and -2 and tescalcin are ∼28% identical at amino acid level) (18, 297). There are no studies of the association of tescalcin and NHE3, but Mailander et al. (318) reported the tescalcin binds to the COOH-terminal NHE1, and Fliegel and co-workers (297) then showed this was to the COOH-terminal 182 aa, specifically to His182. That is, it binds the COOH terminus of NHE1 at a site separate from that which binds CHP1 and -2. This part of NHE1 has minimal α-helical structure, and Ca2+ causes a shift to an even more B-sheet structure. The NHE1/tescalcin association was predicted to influence the pH sensitivity of NHE. An interesting but untested possibility is that CHP1 and CHP3 binding sites might interact to explain effects on pH sensitivity of NHE1. CHP was also reported to be involved in acute ligand-initiated regulation of NHE3. The specific agonist involved was adenosine A(1) receptors in opossum kidney (OK) and Xenopus laevis uroepithelial (A6) cells (116). Activation of adenosine A(1) receptor [A(1)R] by N6-cyclopentyladenosine (CPA) inhibits the NHE3 via a phospholipase C/Ca2+ PKC signaling pathway. Mutation of PKC phosphorylation sites on NHE3 does not affect regulation of NHE3 by CPA (consistent with PKC regulation of NHE3 not involving phosphorylation of NHE3 by PKC), but amino acid residues 462 and 552 are essential for A(1)R-dependent control of NHE3 activity. Yeast two-hybrid assay identified a minimal interacting region with CHP localized to the COOH-terminal juxtamembrane region of NHE3 (amino acids 462–552 of opossum NHE3). CPA (10−6 M) increased the CHP-NHE3 interaction by 30–60%. Overexpression of the polypeptide from the CHP binding region (aa 462–552) interfered with the ability of CPA to inhibit NHE3 activity and to increase the interaction between CHP and full-length NHE3, while decreasing native CHP expression by siRNA abolished the ability of CPA to inhibit NHE3 activity. Therefore, it was concluded that CHP-NHE3 interaction was regulated by A(1)R activation, and this interaction is a necessary and integral part of the signaling pathway between adenosine and NHE3.

2. Direct binding of ERM proteins

The ERM proteins ezrin, radixin, and moesin bind to NHE3 both directly (74) and indirectly via the NHERF family (the latter interactions were the first functional effects on regulation of NHE activity recognized and are discussed below in sect. vB). Thus NHE3 associates with the cytoskeleton using at least two COOH-terminal domains. Of note, some regulation of NHE3 which involves the cytoskeleton has been identified but not attributed to either of these interactions (11, 71, 75, 135, 155, 197, 269, 431433, 488). This represents an area requiring future study.

ERMs are present in the ileal and proximal tubule BB (59). These proteins link the actin cytoskeleton to intrinsic membrane proteins that have cytoplasmic domains. The ERM proteins bind multiple plasma membrane NHEs with direct physical interactions with NHE1 identified first (112, 527). These occur via two positively charged amino acid clusters in the most NH2-terminal first subdomain of the NHE1 COOH terminus. These appear to be the same RK clusters that are involved in phosphatidylinositol 4,5-bisphosphate (PIP2) binding to NHE1. As elegantly described by Barber and colleagues (38, 110112, 300), NHE1 binding to the ERM proteins in fibroblasts is necessary to locate NHE1 in lamellipodia. Importantly, ERM binding to NHE1 appears to be involved in organizing the cytoskeleton in symmetrical cells. This NHE1-ezrin interaction involves NHE1 acting as an anchor in the lamellipodia for actin via ERM binding. The fibroblast functions dependent on this NHE1 binding include localization of ERM proteins and determination of fibroblast mobility and polarity of migration as well as coordination of motility in a manner that is independent of the transport activity of NHE1. Structurally, disruption of NHE1-ERM binding is associated with abnormal stress fibers and focal adhesions and results in lack of polar migration of fibroblasts which are attempting to fill in a wound. This striking abnormality in fibroblasts is in contrast to NHE1 knockout mice, which not only are viable, but have a phenotype that includes seizures (39, 100). This suggests redundancy in the function of this NHE1-ERM interaction. The NHE1 knockout mice appear to have normal intestinal and renal morphology and have decreased salivary secretion.

The direct NHE3-ERM interaction appears quite separate from the type of interaction described for NHE1, although whether, similarly to the role of NHE1, NHE3 binding to ERM proteins is important for the organization of the BB is unknown. Direct ERM binding to NHE3 occurs in domain 1 of the COOH terminus (aa 515–595) and involves a positive cluster of amino acids that include R520, R527, and possibly K516 in an area of high pI (estimated to be 12.1) (74). Ezrin binding to this area is further evidence (along with programs predictive of secondary structure) that this area of NHE3 is α-helical, as this is the only way that these positive amino acids would form a cluster, which is needed for ezrin binding. The ezrin domain involved in NHE3 binding is the third “clover leaf” of the NH2-terminal FERM domain (59, 140, 193, 253, 367, 401, 440, 441, 527). This is the same part of the ERM protein involved in binding to the NHERF family (74, 366, 402, 418). Functionally, direct ERM binding to NHE3 is necessary for NHE3 trafficking (74). When these positive amino acid mutants were expressed in PS120 fibroblasts and OK cells (R520F, R527F referred to as NHE3 F1D), there was a marked reduction in the transport activity of NHE3, due to much less NHE3 on the plasma membrane. The specific aspects of BB trafficking that depend on direct ERM binding to NHE3 and thus are inhibited by these mutations include endocytic recycling (exocytosis from the recycling system) and delivery to the plasma membrane of newly synthesized NHE3. Initial rates of endocytosis were less affected, being only slightly reduced (note this would increase the amount of BB NHE3). In addition to inhibited basal NHE3 activity, Ca2+ inhibition of NHE3 was reduced in the ezrin binding mutant. Since Ca2+ inhibits NHE3 by stimulating endocytosis, what explains this effect? The possible explanation comes from 1) NHE3 plasma membrane half-life studies demonstrating a normal initial decrease in plasma membrane NHE3 followed by a prolonged plasma membrane presence, plus 2) FRAP studies with confocal microscopy examining OK cell apical membranes of wild-type and ezrin binding mutant NHE3, which showed that the mutant had a markedly reduced mobile fraction. We suggest that Ca2+ inhibition of NHE3 normally involves complex formation followed by endocytosis of NHE3. The ezrin mutant has a normal rate of endocytosis of the NHE3 present under basal conditions in the vicinity of the intervillus clefts, where endocytosis initiates. However, we speculate that direct ezrin binding to NHE3 is necessary to move microvillus NHE3 to the intervillus cleft area for endocytosis, and this is defective in the ERM binding mutant (74). Whether the increased NHE3 complex formation and its subsequent endocytosis that occurs with elevated Ca2+ regulation of NHE3 is affected in the direct ezrin binding mutant has not been examined experimentally. Thus NHE3 is one of a growing family of ERM binding proteins in which one molecule of the protein binds two ERM molecules, one direct and one indirect via NHERF family members (402, 484). Ezrin is now recognized to bind to proteins either directly or indirectly via NHERF proteins. Substrates that only directly bind ezrin include intracellular adhesion molecule (ICAM)-1, ICAM-2, ICAM-3, CD43, CD44, CD95, syndecan2, and NHE1 (110, 202, 306, 454, 527). Indirect associations have been described with the β2-adrenergic receptor, cystic fibrosis transmembrane regulator (CFTR), podocalyxin, and now NHE3 (191, 324, 428, 454, 479). The renal podocyte plasma membrane protein podocalyxin appears most similar to NHE3 in ERM binding (402). Podocalyxin binds ezrin both directly and indirectly via NHERF1 or -2; however, the function of the direct ezrin binding has not been discovered. Interestingly, it appears that podocalyxin may directly bind ezrin using an α-helical structure since there is not three or more consecutive positively charged amino acids in the domain of podocalyxin in which it binds ezrin.

3. Phosphoinositol phosphates and ATP

All NHEs require ATP for function, although the mechanism is not clear (4, 5, 65, 68, 73, 108, 155, 234, 292). Domain 1 in the COOH terminus is the region responsible for the cellular ATP dependency of NHE activity (5, 65, 108). Although ATP is not directly involved in the Na+/H+ exchange, cellular ATP has been known to be necessary for NHE activity since 1986 (68). Subsequently, this phenomenon was observed in several different cells, including type II alveolar epithelial cells (64); PS120 cells transfected with NHE1 (292); AP1 cells transfected with NHE1, NHE2, and NHE3 (234); and red blood cells (108). In NHE1, the 80-aa subdomain 1 region (aa 515–595) adjacent to the last transmembrane domain is involved in ATP depletion-induced acidic shift of pHi-dependent activity. It was proposed that this domain of NHE1 might interact with the H+-modifier site, leading to its increased affinity to H+ and that ATP depletion might prevent such an interaction (3). The mechanism for the ATP dependency was further elucidated by Aharonovitz et al. (5), who showed that PIP2 might be directly involved in the ATP-dependent NHE activity. PIP2 was shown to bind to the positive charged amino acids (Arg or Lys) at aa 513–564 in NHE1. Mutant NHE1 lacking the putative PIP2 binding motif had greatly reduced transport capability, implying that association with PIP2 is required for optimal activity. On the basis of these experiments, it was proposed that ATP depletion diminished the cellular PIP2 content and thereby inhibited NHE activity (5). More recently, roles of phospholipids in regulating NHE activity were further demonstrated by Fuster et al. (155) through measuring extracellular protein (H+) gradients during whole cell patch-clamp studies with perfusion of the cell interior in NHE1- or NHE3-expressing cells. In their experiments, NHE1 activity is stable with perfusion of nonhydrolyzable ATP [adenosine 5′-(β,γ-imido)triphosphate], suggesting Na+/H+ exchange does not require direct ATP hydrolysis. However, NHE1 activity was abolished by ATP depletion (2-deoxy-d-glucose with oligomycin or perfusion of apyrase) but could be restored with PIP2 and was unaffected by actin cytoskeleton disruption. These data support 1) the previous reports that cellular ATP is required to generate sufficient PIP2 for NHE activity and 2) the PIP2 effect on NHE activity is directly on NHE1 itself (through direct NHE binding). Importantly, NHE3 activity was not affected by intracellular perfusion with PIP2 under conditions that activated NHE1. However, NHE3 was reversibly activated by phosphatidylinositol 3,4,5-trisphosphate, suggesting that the role of phospholipids in modulating NHE activity occurs in an NHE isoform-specific fashion.

B. COOH-Terminal Domain 2, Amino Acids 586–660: Calmodulin, Calmodulin Kinase II, and NHERF Family

1. CaM and CaM kinase II

Ca2+-dependent activation of NHE1 is well documented (470, 473, 474, 476). NHE1 contains a CaM-binding domain (CBD) at aa 637–656. Deletion or mutation in this area induces a constitutive activation of NHE1, resulting in an increase in the affinity for intracellular H+ (an alkaline shift of pHi-dependent exchange activity) and loss of responsiveness to a rise in intracellular Ca2+ concentration (41, 219, 470, 476). Therefore, it was concluded that the CBD reduces the affinity of NHE1 for intracellular H+ by exerting an autoinhibitory function in quiescent cells (473, 474, 476). This effect was replicated in an NHE1 chimera, in which the CBD of NHE1 was replaced with the corresponding domain in NHE2 (but not NHE4), suggesting that both NHE1 and NHE2 contain a functionally similar CaM-binding domain (473). Moreover, photoaffinity cross-linking between immunoaffinity-purified NHE1 and 125I-labeled CBD of NHE1 showed that the CBD was efficiently cross-linked to NHE1, suggesting that CBD physically interacts with an unspecified domain(s) of NHE1 (473). Whatever the unspecified domain(s) might be, its interaction with the autoinhibitory domain (CBD) might be to alter the H+ affinity of the H+-modifier site either directly or indirectly through a conformational change of NHE1. Also, CaM is involved as an intermediate in regulatory volume increase (RVI) stimulation of NHE1 activity, which occurs downstream of Janus kinase 2. The latter phosphorylates CaM (157).

Fliegel et al. (294) recently showed that optimal CaM binding to NHE1 requires a COOH-terminal conformation that appears to be maintained at least in part by a stretch of six or seven acidic residues (aa 753–759). Mutation of these resulted in diminished and slower NHE1 activity and >50% reduction of CaM-NHE1 binding without significantly affecting the total and surface expression of NHE1. These data suggest that a proper conformation of the cytosolic COOH terminus of NHE is necessary for a fully functional NHE.

CaM and CaM kinase II are present in the ileal and renal proximal tubule BB and regulate neutral NaCl absorption in rabbit ileum (95, 128, 132, 195, 300, 545). Basal NaCl absorption and BB NHE3 activity are inhibited by CaM and CaM kinase II since 1) inhibiting CaM (with W13) or CaM kinase (with H-7) stimulates neutral NaCl absorption and 2) including CaM and CaM kinase II in the presence of Ca2+ in ileal BB vesicles inhibited BB NHE3 activity (95, 122, 132, 144, 391, 545). In PS120 fibroblasts, both CaM and CaM kinase II directly bind to the NHE3 COOH-terminal domain between aa 586–660 (545). Similar to ileal Na+ absorptive cells, a CaM kinase II inhibitor, KN-62, stimulates basal NHE3 activity in PS120 fibroblasts (545), by a NHERF2-dependent mechanism. Thus CaM kinase II binds NHE3 directly and inhibits NHE3 activity under basal Ca2+ conditions. CaM binds to the COOH termini of NHE1, -2, and -3 and is involved in basal regulation in these three isoforms. While the mechanism of this regulation is unknown, NHE3 is phosphorylated under basal conditions by CaM kinase II (M. Zizak and M. Donowitz, unpublished data).

2. NHERF family of multiple PDZ domain proteins

Multi-PDZ domain containing proteins bind to NHE3 at domain 2 and are involved in NHE3 regulation. There are multiple controversial aspects of these observations, and this is an area of active study. A role for PDZ domain-containing proteins in NHE3 regulation was first established by observations in PS120 fibroblasts that elevated cAMP failed to inhibit NHE3 (534). These cells lack endogenous NHERF2 and have only a small amount of NHERF1 (529, 532, 534). However, when these cells stably expressed NHERF1 or NHERF2, cAMP inhibited NHE3, as occurs in epithelial cells (277, 532, 534). Since this observation, multiple aspects of stimulation and inhibition of NHE3 activity have been shown to depend on the presence of PDZ domain containing proteins. Similar observations were subsequently established for CFTR and other transport proteins. Not known is whether all acute stimulation and/or inhibition of NHE3 requires or is influenced by NHERF family binding.

3. Overview of PDZ domain proteins (49, 63, 113, 156, 196, 215, 248, 352, 411, 455, 538)

PDZ domains (named after the three initially identified proteins which contained these domains, PSD-95, Discs large, and ZO1) are ∼90-aa modular protein-protein interacting domains. Structurally they are made up of six β-sheets and two α-helices, with ligand binding in a linear grove between β-sheet 2 and α-helix 2. PDZ domains are commonly used in the human genome for protein-protein interactions, and currently >600 PDZ domains in human proteins have been identified. Since many proteins contain multiple PDZ domains, ∼400 mammalian proteins have been identified as containing PDZ domains. PDZ domain containing proteins often have more than a single PDZ domain (as well as other protein interacting domains), with the structural relationship among the PDZ domains fixed, i.e., the PDZ domains are not organized like their stick diagrams, but their relationships among themselves confers structure to their binding partners.

PDZ domain interactions with their ligands generally occur at the ligand COOH terminus and primarily involve amino acids at the 0 and −2 positions, although up to at least the last six amino acids can affect binding. Similar internal sequences, called β-fingers, have been identified in nNOS, PSD-95, syntrophin, and NHE3 and are used to bind PDZ domains. There are several categories of PDZ binding domains based on the ligand binding sites. Class I consists of X S/T X hydrophobic, with I/V/L/M reported at aa 0. Class II consists of X hydrophic, X hydrophobic. Class III consists of X E/D X hydrophobic. The reported affinities of PDZ domain-ligand interactions appear low, 1 nM-10 μM, which seems appropriate for dynamic interactions (196).

PDZ domain containing proteins are generally cytoplasmic and connect the plasma membrane to the cytoskeleton via binding to the cytoplasmic tails of membrane proteins. Identified binding partners are of many categories and include growth factor receptors (EGFR, platelet-derived growth factor receptor), G protein-coupled receptors, neurotransmitter receptors, transport proteins, adhesion molecules, cytoskeletal elements, transcription factors, and PIP2 (248). Understanding functions of PDZ domain proteins must consider not only their ligands but also their location, which may be dynamic. Their location is generally at specific membrane domains (196). In epithelial cells there are separate PDZ domain proteins restricted to the BB, basolateral (BLM) and tight junctions (TJ), as well as cell-cell and cell-matrix contacts (63); and in polarized neuronal cells, PDZ proteins are at multiple depths from the surface (156, 455).

The major function of PDZ domains is as scaffolds, using their ability to homo- or heteromultimerize and form cysteine bonds (156, 428, 455). While defining functions of PDZ proteins has been complicated by redundant expression of multiple family members in the same location at the same time, specific functions identified include 1) complex formation, in most cases on the plasma membrane but also at intracellular sites. The site of assembly of these complexes is not known, although for NHE3 the complexes on the BB and plasma membrane are larger than intracellular forms, suggesting at least part of the formation is at the plasma membrane (8, 298). 2) They organize plasma membrane domains, including LR. This is critical for cell signaling, in which certain interacting signaling molecules are included, while others are excluded. This allows increased efficiency, speed, and specificity in signaling (248). 3) They establish and maintain polarity in embryos as well as epithelial and neuronal cells (49). 4) They regulate trafficking-sorting of proteins, moving from TGN to the BB, BLM, or TJ, and endocytic recycling versus lysosomal targeting (185, 248). 5) They cluster or retain proteins in the plasma membrane. For instance, PSD-95 clusters ion channels and InaD retains TRP channels, PKC, and PLC in the rhabdomere (455). Also, PDZ proteins have been suggested as retaining NaPi2a and CFTR in the BB rather than the BLM (43, 186, 205, 206, 276, 409). The latter is controversial (418). 6) They affect ion channel gating. CFTR gating is altered by binding to tandem PDZ domains of NHERF1 or PDZK1, although the mechanism is not understood (381, 480).

Understanding the roles of PDZ domain proteins in polarized cells has evolved from the initial perception that they served a strictly scaffolding role to allow complexes to form. As recently reviewed by Sheng and co-workers (215, 248), these proteins are abundant, often an order of magnitude greater than the transporters they help regulate. It appears that these proteins may have some highly cell specific effects. Their recognized epithelial actions include (63) 1) assembly of specific proteins into large static molecular complexes at specific locations in a cell. This includes forming clusters of transporters at the plasma membrane. Both plasma membrane and cytoplasmic proteins can be brought into the same complex, including signaling molecules (36, 190). 2) They allow dynamic trafficking of proteins by bringing molecular motors (for instance, bringing both kinesins and myosins, including myosin VI) into apical membrane complexes (430, 522). The complexes can move around the cell as preassembled packages or from one site such as the apical membrane to others. Both microtubule tracks and actin-myosin motors are used. PDZ proteins on the surface of the cargo vesicles can act as receptors for motors and lead to movement to targeted sites. 3) They deliver and/or stabilize plasma membrane proteins including transport proteins. This includes regulation of exocytosis, endocytosis, subcellular localization, determination of subunits present, and control of intrinsic activity.

The PDZ proteins are often also highly regulated (43, 248). They bind to multimers of themselves and other PDZ proteins. Myristoylation or palmitoylation are used to link the PDZ proteins to plasma membrane or intracellular sites, and binding to tetraspannin proteins can be involved in surface expression. There is generally dynamic regulation of the PDZ proteins and thus of their complexes, which involve changes in subcellular distribution, phosphorylation, and degradation of both the PDZ domain protein and the PDZ binding ligand (3, 91, 92, 192, 265, 280, 348, 444, 463). The regulation of actions of PDZ proteins can be affected by phosphorylation of the ligand COOH-terminal amino acids, especially the S/T at the −2 position. Most commonly, disassembly of complexes occurs with PDZ ligand phosphorylation. For instance, Weinman showed that β2-adrenergic receptor binding to NHERF1 was decreased by receptor phosphorylation (unpublished data).

4. Epithelial apical membrane PDZ domain proteins relevant to NHE3 regulation

The BB of mammalian small intestine and renal proximal tubule and cell culture models of intestinal Na+ absorptive cells (Caco-2, HT29, T84 cells) and renal proximal tubule cells (OK cells) contain varying amounts (from none to amounts up to 0.1% of the total cell protein) of four related PDZ domain proteins in or near the apical membrane which are members of a gene family, called the NHERF family (126, 205, 276, 405, 442, 492). There are additional unrelated PDZ domain proteins in the BB which affect NHE3, such as Shank2 (194). These NHERF family members include NHERF1 [which is also called NHERF or EBP50 ezrin binding protein 50 kDa in size (EBP50)], NHERF2 [also called NHE3 kinase A regulatory protein (E3KARP), SIP-1, TKA-1], NHERF3 or PDZK1 (also called clamp, CaP70, diphor-1 or NaPi-cap1), and NHERF4 or IKEPP (intestinal and kidney enriched PDZ domain protein, also called PDZK2 diphor-2 and NaPi-cap2) (Fig. 6). These four proteins are scaffolds and consist of multiple PDZ domains (2 for NHERF1 and -2 and 4 for NHERF3/PDZK1 and NHERF4/IKEPP). The only other recognized domain that any of the four proteins contain is an ERM (ezrin, radixin, moesin) binding domain at the COOH terminus of NHERF1 and -2 (533).

FIG. 6.

NHERF family members showing number of PDZ domains, other protein-protein interacting domains, and number of amino acids.

These four members of the NHERF family are closely related. Mammalian PDZ proteins originated in worms and flies rather than in bacteria, plants, or yeast (126, 442). In fact, bacteria, plants, and yeast lack true PDZ proteins in which the ligand binding grove is linear and rather have related proteins that have a circular binding cleft. NHERF family origins can be traced to C. elegans and D. melanogaster (126, 442; A. Sharma, C. Brett, and M. Donowitz, unpublished data). Blast analysis across species of human NHERF1 identified 400 related proteins. Clustal W analysis was used to align these sequences, and they were organized into a Claudogram using PAUP v4.1, which showed that these four PDZ proteins were the most related proteins in multiple genomes. The 12 individual PDZ domains from these 4 proteins, as defined by Scansite (PFAM), had their origins traced to either worms or flies. Eleven of 12 NHERF family PDZ domains appear to have evolved from 2 worm proteins that are probably related to each other by gene duplication. Similarly, two fly PDZ domain-containing proteins are highly related to the human NHERF family PDZ domains and have a closer similarity than with the worm proteins for some of these PDZ domains but with no relationship with others of those PDZ domains. However, one of these fly proteins has homology to a single NHERF family PDZ domain that lacks homology to the worm proteins (126). Thus we suggested that the NHERF family evolved from gene duplication in the worm, which evolved into a protein in fly which also evolved into a separate PDZ domain that via evolution produced the human NHERF family.

Why there are multiple similar, related PDZ domain proteins in the apical domain in the same epithelial cell is not understood, although some hints have emerged recently. Two areas of divergence have been identified among the four NHERF family members: a) subtle differences in location in the apical surface of kidney and intestine (469, 537) and b) somewhat different binding partners among proteins including those that have been shown to be involved in NHE3 regulation (72, 87, 217, 229, 230, 250, 355, 402, 423).


All NHERF family proteins are in or near the BB. Wade and Weinman (469) demonstrated in mouse renal proximal tubule that NHERF1 was in the BB while NHERF2 was also in the BB but located mostly just below the BB in an area consistent with the intervillus clefts (where clathrin-coated pits form) (469). Murer and co-workers (311) further demonstrated that NHERF3/PDZK1 had a similar distribution to NHERF1 in the renal proximal tubule BB, while NHERF4/IKEPP was apical plus subapical (172, 467). Our results in mouse ileum and Caco-2 cells expressing HA-NHE3 support this differential distribution. In both mouse ileal Na+ absorptive cells and Caco-2 cells, NHERF1 and NHERF3/PDZK1 are largely in the BB, with NHERF1 also appearing to be present in goblet cells (535). NHERF2 and NHERF4/IKEPP are distributed in the BB and subapically, although NHERF4/IKEPP also is in the supranuclear cytosolic area. Thus BB location of these proteins appears similar under basal conditions in ileum and renal proximal tubule although NHERF2 is also in the ileal microvilli, perhaps to a greater extent than in the proximal tubule. How the apical membrane and intracellular localization of these proteins change with conditions which regulate NHE3 is not known.

Also, not all species have all NHERF family members or have them in the same locations. For instance, while human, rabbit and mouse renal proximal tubules contain NHERF2, it appears to be lacking in rat proximal tubules (469). Moreover, different epithelial cell culture models express different amounts of each NHERF family member. For instance, the renal proximal tubule cell line OK has NHERF1, but generally lacks or has low expression of NHERF2. Caco-2 cells have little NHERF2 while another colon cancer cell lines HT-29 contains an abundant amount of NHERF2 (529).


These are only partially known, although ∼40 binding partners have already been identified (see recent review in Ref. 408). Moreover, there are few organized comparisons reported of whether binding partners of one member of this family also bind other members, and some studies are contradictory, which is not unexpected given their generally low binding affinity. Common binding partners of NHERF1 and NHERF2 (listed by category of ligand) include the following (109, 408): tyr kinase receptors (PDGFR), G protein-coupled receptors (β2-adrenergic receptors, PTH1R), transporters (NHE3, CFTR, Na+-K+-ATPase), enzymes (PLC-β1 and -β2, PKC-α, EP164), and cytoskeleton linkers (ezrin). A recent review listed ligands of NHERF1 that do not bind NHERF2. These include TRP4 and -5, NBC3, H+-ATPase, Kir1.1, κ-opiod receptor, LPA2R, GRK6A, inducible nitric oxide synthase, and YAP65 (192, 214, 332, 377, 408, 437), although most studies were not performed in detail enough to be definitive. Ligands that appear to bind NHERF2 and not NHERF1 include TAZ (a transcription factor) (230), α-actinin-4 (250), adenosine 2β receptor (416), PLC-β3 (217), DRA (275), SKG1 (529), podocalyxin (data contradictory; Refs. 300, 402, 436), SRY-1, cGKII (72) TRPV5 (143), Ca2+-ATPase 2b (107), and ClC-5 (211). No specificity in binding has been established as yet for NHERF3/PDZK1 (binds NHE3, CFTR, NaPi2a, ezrin, DRA, d-AKAP2 and α-actinin-4) (173, 174, 205, 393, 408, 479) and NHERF4/IKEPP (binds guanylate cyclase C, NHE3, TRPV5, and TRPV6) (405, 442, 464). While these studies are generally preliminary, the concept of specificity among binding partners of the NHERF family is established.

5. “Hand-off” hypothesis

The possible role of these multiple NHERF family proteins may relate to the “hand-off hypothesis” first suggested by the Murer group (27, 42, 43, 204, 206) and supported by the studies of Guggino and co-workers for CFTR (8183). Under basal conditions in Caco-2 cells, NHE3 and NHERF3/PDZK1 overlap in the BB (535). We found that with elevated intracellular Ca2+, NHERF3/PDZK1 stays in the BB but forms aggregates and separates from NHE3, which in 10 min appeared to move to the area of the intervillus clefts (536). We suggest that this may represent separation of BB NHE3 from NHERF3/PDZK1 before initiation of endocytosis. Studies with CFTR and a Golgi-resident single PDZ domain-containing protein called Cal supports this concept. Cal associates with CFTR under basal conditions intracellularly, but under conditions in which CFTR trafficks to the apical membrane, there is less Cal-CFTR association and more NHERF1-CFTR association (8183). This supports sequential interaction of transport proteins with individual NHERF family members or other intracellular PDZ domain containing proteins, based on states of regulated stimulation versus inhibition. However, support for this hypothesis is very preliminary, and details of transfer and reasons for changes in NHE3/individual NHERF protein interactions are unknown.

6. Model systems used to understand the cell specific role for NHERF proteins in NHE3 regulation

The roles of each NHERF family member in NHE3 regulation have been approached at multiple levels, although conclusions are still preliminary. These approaches include 1) expression of individual NHERF family proteins in null simple cell lines, for instance, PS120 fibroblasts. The strength of this approach is that these cells largely lack NHERFs (529, 533), although small amounts of NHERF1 have been reported (6) and our laboratory also is able to identify small amounts of NHERF2 in these cells using more sensitive antibodies than initially reported. 2) There is expression in polarized epithelial cell lines either that lack a specific NHERF family protein (i.e., OK cells contain NHERF1 but generally lack or have very low levels of NHERF2) (529) or use of siRNAi in these cell lines. 3) Knockout mice models have been reported for NHERF1 (93, 102104, 243, 276, 333, 409, 495, 500) and NHERF3/PDZK1 (67, 93, 262, 263, 413, 445, 478).

Most relevant is understanding of how the NHERF family proteins take part in NHE3 regulation in intact intestine. To date, only a few studies have been reported. The conclusions from studies of the renal proximal tubule and intestinal knockout mouse models that have emerged are that multiple NHERF family members are involved in cAMP, cGMP, and Ca2+ inhibition of NHE3. These results are strongly suggestive that interactions among the NHERF family members are probably involved. Also, it appears that there may be major differences in the role of these proteins in renal proximal tubule and intestine (338, 409). It remains unknown whether all acute regulation of NHE3 by extracellular ligands requires the NHERF family. There are no studies indicating a role for the NHERF family in long-term regulation of NHE3. A single report of abnormal phosphate handling in NHERF1 knockout mouse supports a role of NHERF1 in chronic phosphate transport (495). This complexity in intact tissue, already recognized, supports the study of simpler systems to identify potential roles for each member of the NHERF family in NHE3 regulation. However, while reductionist studies must be used to understand how the NHERF proteins can carry out specific regulatory functions, they must be studied together in the context of native intestine and proximal tubule in NHE3 regulation to allow dissection of the complexity that almost certainly occurs in intact tissue.


Regulation of NHE3 by cAMP, cGMP, or elevated Ca2+ has been characterized based on inhibition of NHE3 by all three second messengers in ileum and colon as well as in renal proximal tubule (Table 2). NHERF1 reconstitutes only cAMP inhibition of NHE3 and is not able to reconstitute cGMP (in presence of cGKII transfected into the cells) or elevated Ca2+ inhibition of NHE3 (72, 250, 533). NHERF2 reconstitutes cAMP, cGMP, and elevated Ca2+ inhibition of NHE3 (72, 250, 533). PDZK1 reconstitutes cAMP and Ca2+ but not cGMP inhibition of NHE3 (535, 536; N. Zachos and M. Donowitz, unpublished data). There are only preliminary studies of IKEPP, but it seems to mediate stimulation of NHE3 by elevated Ca2+. This is not what occurs in intact intestinal and renal epithelial cells but might mimic part of stimulatory mechanisms of NHE3 that are Ca2+ dependent. cGMP inhibition of NHE3 is not reconstituted by IKEPP expression (Zachos and Donowitz, unpublished data).

View this table:

NHERF family dependence of second messenger inhibition of NHE3 in PS120/NHE3 cells


OK cells express NHERF1 and NHERF3 endogenously but have none or very low amounts of NHERF2 (211, 529, 533). cAMP inhibits NHE3 in OK cells by a complicated stepwise process with rapid inhibition which is not associated with changes in surface amount (turnover number change), which by 30 min is associated with increased endocytosis and decreased surface NHE3 amount. By 60 min it is also associated with decreased exocytosis. All aspects require cAMP phosphorylation of NHE3 (28, 97, 212, 330, 331, 380, 506, 543). In OK cells, elevating cGMP did not affect NHE3 activity (72). However, in OK cells expressing NHERF2, cGMP inhibited NHE3, an effect that was dependent on presence of both NHERF2 and cGKII (latter transfected in but not expressed endogenously) (72). This demonstrated that NHERF2 was necessary for cGMP regulation in polarized epithelial cells as well as in fibroblasts.

In addition to being needed for acute inhibition of NHE3, NHERF2 but not NHERF1 is necessary for LPA stimulation of NHE3 (285). This is discussed below.


These mice have been created by homologous recombination and characterized by Shenilokar, Weinman and colleagues primarily for effects on renal function and structure (93, 102104, 243, 276, 409, 491, 495, 500). A second NHERF1 knockout model has been reported by Georgescu and co-workers (333). Structurally, the BB of renal proximal tubule cells and the small intestine of the Shenilokar/Weinman mouse are normal by electron microscopy (409; Hubbard, Murtazina, Kovbasnjuk, and Donowitz, unpublished data). However, in the Georgescu NHERF1 knockout mouse, small intestine was reported to have abnormal ileal villi with less and shorter microvilli and thickened and disorganized terminal webs (333). What explains this difference is not known, and both NHERF1 knockout mice were on the same C57Bl/6 background. Functionally, basal transport of NHE3 was similar in wild-type and NHERF1 knockout proximal tubule based on studies of BB vesicles, proximal tubule cells in primary culture, and two-photon microscopy of kidney cortical slices (338, 409, 500). However, cAMP failed to alter NHE3 activity in proximal tubule characterized by all three approaches (338, 409, 500). In contrast, in ileum, basal transport was also normal, but cAMP inhibited NHE3 at 30 min by ∼50%, which is similar to what occurs in wild-type NHE3 (338). What explains these differences is not clear, although cAMP failed to increase NHE3 phosphorylation in proximal tubule (409, 500), which had been shown to be necessary for cAMP inhibition of NHE3 in both fibroblasts and proximal tubule cell models (OK cells). Biochemical studies in kidney showed that amounts of NHE3, NHERF2, and NHERF3/PDZK1 were normal (increased NHERF2 in one of two knockout mouse models). In contrast, the amount of BB ezrin and phosphoezrin were decreased, suggesting that failure to form the NHE3/NHERF1/ezrin/PKAII complex might explain the lack of effect of cAMP in kidney (333). What the differences in the NHE3 complexes are that allow cAMP to inhibit NHE3 in the ileum are not understood, although it is possible that NHERF2 has different functions (or locations) in proximal tubule and ileum.

In more preliminary studies of mouse colon using the Ussing chamber/voltage-clamp technique, comparing wild-type to NHERF1, NHERF2, and NHERF3/PDZK1 knockouts, all three NHERF knockouts affected regulation of Na+ absorption that was attributed to NHE3 (93). Most significantly, NHERF3/PDZK1 knockout had markedly reduced basal NHE3 activity and lacked cAMP, cGMP, or Ca2+ inhibition of NHE3. NHERF2 knockout had reduced basal Na+ absorption and absent Ca2+ inhibition of NHE3. NHERF1 knockouts had the least effect with slightly reduced basal cAMP, cGMP, and Ca2+ inhibition of NHE3.

7. NHERF-interacting domains of NHE3

NHE3 and NHERF1–4 directly interact. In multiple cell types, NHE3 interacts with NHERF1, NHERF2, and NHERF3/PDZK1 based on yeast two-hybrid assays (173, 174, 205, 277, 407, 532, 536). This is true when NHE3 and the NHERF proteins are both expressed in fibroblasts, OK cells stably transfected with NHERF2, Caco-2 cells, mouse ileum, and rabbit ileal BB. Thus they are in the same complex, although the amount of coprecipitation is relatively small and surprisingly variable. Whether this is due to the detergents used for solubilization or reflects the low affinities of the interactions is not known. That at least some of these interactions are direct has been established using fusion proteins of the NHE3 COOH terminus and recombinant NHERF proteins. Direct protein-protein interactions of the NHE3 COOH terminus with each member of the NHERF family has been demonstrated (445, 532, 533, 536; and Thelin, Hodson, Zachos, Donowitz, and Milgram, unpublished data). NHE3 binds via an internal sequence to NHERF proteins (532, 533). The domain in the NHE3 COOH terminus which binds to the NHERFs is between aa 586–660. In multiple pulldown and/or overlay studies, rabbit NHE3 only bound to intact NHERF1–4 via this internal domain, specifically not binding with its very COOH terminus, that includes aa 757–832 (532, 533). When recombinant NHERF2 was overlaid with 35S-NHE3 COOH-terminal constructs, binding occurred with full-length NHE3 COOH terminus (aa 475–832), aa 475–711, and aa 475–660 but not to aa 475–585 (532). This also suggested that binding occurred between aa 585–660 in the NHE3 COOH terminus. IP of NHE3/585 (NHE3 truncated at the COOH terminus to aa 585) failed to coprecipitate NHERF1 or NHERF2, and vice versa, when both were expressed in PS120 cells, supporting that aa between 585–660 were necessary for the NHE3-NHERF interactions. In contrast, NHERF3 was reported to bind in vitro to rabbit NHE3 COOH-terminal 131 aa but not when the four COOH-terminal amino acids of this construct were removed (445). Is there further evidence that the COOH-terminal four amino acids of NHE3 are involved in binding to NHERF members? The data are inconsistent. On initial yeast two-hybrid screening, truncating the NHE3 COOH terminus to examine binding to NHERF1 and NHERF2 resulted in autoactivation, and thus the question could not be addressed (533). This was repeated by Weinman (504) with mutations of the last four amino acids with changes in the 0 and −2 and −3 positions eliminating binding (504). In addition, using a membrane-friendly variant of the yeast two-hybrid system based on ubiquitin, NHERF1 and NHERF2 binding to the NHE3 COOH terminus was also demonstrated with evidence that in addition to binding the last four amino acids, NHERF1 but not NHERF2 bound NHE3 internally (Murer, Gisler, Moe, and Biber, unpublished data). Of note, while these yeast interactions suggested COOH-terminal binding, they are prone to false positives. Moreover, when the COOH-terminal truncations of NHE3 were examined for physical interactions with NHERF1 by pull-down assays, the COOH-terminal four amino acids were not necessary for the physical association (504). Thus the bulk of the evidence supports that NHE3 binds NHERF1, NHERF2, NHERF3/PDKZ1, and NHERF4/IKEPP in a NHE3 COOH-terminal domain between aa 586–660. The interaction with the COOH-terminal four amino acids of NHE3 in mammalian cells is not proven, though still possible, even though the last four amino acids of NHE3 are a classic COOH-terminal class I PDZ binding domain (STHM). We speculate that NHE3 does not bind the NHERFs with its last 4 aa because they are buried in its tertiary structure.

8. Role of NHERF family members in NHE3 regulation (compared with functions of other PDZ domain containing proteins in polarized epithelial cells)


While multiple PDZ domain proteins are present at the tight junctional complexes and are necessary for establishment of epithelial polarity (these include mammalian Scrib-1, PAR-3, and Pals-1) (63), this is not a role of NHERF proteins, and these proteins are not junctional.


The NHERF family is not involved in determining the percent of NHE3 on the plasma membrane (71, 126, 250, 537). In NHE3 truncation mutants which have had the NHERF binding domains removed, the same percent of NHE3 is present on the plasma membrane or BB as that in wild-type NHE3 and in NHE3 truncations which bind the NHERF members. Moreover, PS120 cells, which contain no NHERF proteins (no endogenous proteins and use of NHERF1 siRNA to remove the small amount of endogenous NHERF1 expression), have the same plasma membrane percentage of NHE3 as cells in which multiple NHERF proteins were expressed (M. Cavet, R. Sharker, and M. Donowitz, unpublished data). Moreover, in NHERF1 null mice, NHE3 is present in a normal amount and location in the renal proximal tubule (409). This makes it unlikely that there is a role for NHERF members in basal trafficking of NHE3 to the plasma membrane and/or membrane retention, at least in PS120 fibroblasts. Of interest, NHERF3/PDZK1 requires the presence of a membrane-associated protein MAP17 for it to be in the BB of OK cells (261, 376, 413). In contrast to changes in NHERF members in which NHERF1/2 bind to actin via ezrin, disruption of microtubules with colchicine in rat renal proximal tubule led to NHE3 and NHERF1 appearing on the BLM in addition to the BB (396).

I) cAMP inhibition of NHE3 activity.

This was the initial model developed in which NHERF1 or NHERF2 appeared necessary for cAMP to inhibit NHE3 activity. These studies were based on models of cholera toxin inhibition of small intestinal NaCl absorption by elevation of cAMP. Similar inhibition by elevated cAMP occurred in renal proximal tubule of BB Na+/H+ exchange activity, and this was shown by Weinman, Shenolikar, and co-workers to require a soluble 55-kDa protein which was separate from the Na+/H+ exchangers (493, 496, 498, 499, 501, 503). They named this protein NHERF (now called NHERF1) and cloned it by initial partial purification via column chromatography, determination of a partial amino acid sequence, design of an oligonucleotide probe based on this sequence, and screening of a rabbit renal library (501). That NHERF1 was necessary for cAMP inhibition of NHE3 was initially demonstrated in PS120 fibroblasts (533). These cells lack endogenous NHERF proteins, but when stably transfected with NHE3, it was assumed they would respond to elevated cAMP with NHE3 inhibition and serve as a model for cAMP regulation of NHE3. However, acute elevation of cAMP in these cells had only very small effects on NHE3 activity until these cells were transfected with either NHERF1 or NHERF2 (533). NHERF1 or NHERF2 was necessary to allow cAMP to phosphorylate NHE3 (547). In fact, all described aspects of cAMP inhibition of NHE3 require NHE3 phosphorylation (330, 493, 494, 496, 500, 506, 543). PKA II was the kinase involved (277). PKAs exert many of their effects by binding to A kinase anchoring proteins (AKAPs), which place the kinase at specific cell locations close to PKA substrates and also to other proteins involved in regulation of the ligand. The nature of the NHE3 complex involved in this cAMP effect evolved from recognition by Goldenring and associates (133, 387) that NHERF1 was a human ezrin binding protein (also called EBP50). Ezrin was shown to be a low-affinity AKAP (133). NHERF2 was also shown to bind ezrin, while neither NHERF1 nor NHERF2 were AKAPs (277, 532). The model that has evolved and stood up to multiple tests indicated that there was a NHE3 complex involved that consisted of NHE3 bound to the PDZ domain proteins NHERF1 or NHERF2. These NHERFs also bound to ezrin, which acted as an AKAP to allow cAMP-activated catalytic subunit of PKA II to be freed up close to NHE3, which was then phosphorylated to lead to NHE3 inhibition (Fig. 7) (121, 126, 205, 492, 494, 497, 498, 502, 537, 547). Thus the role of NHERF1/NHERF2 was suggested as being to form the NHE3 complex by binding to both NHE3 and ezrin and in this way exposing NHE3 to activated PKA II catalytic subunit. That this binding of NHERF1 to ezrin was required for cAMP inhibition of NHE3 was demonstrated in PS120 and OK cells. In OK cells, in which wild-type NHERF1 was present, expression of NHERF1 lacking the COOH-terminal 30 amino acids did not reconstitute cAMP inhibition of NHE3 or NHE3 phosphorylation, acting in a dominant negative manner to prevent cAMP inhibition of NHE3 and phosphorylation of NHE3 (497, 502). The domains of NHERF1/NHERF2 involved in this complex formation are the PDZ2-COOH terminus or a somewhat extended PDZ2 which binds the NHE3 COOH terminus plus the COOH-terminal ERM binding domain (502, 532). The ERM binding domain of NHERF1 includes the COOH-terminal 23 amino acids which is predicted to be α-helical and contains 4 positively charged amino acids on side of the α-helix. When three of these positive charged amino acids are mutated, NHERF1 is unable to reconstitute cAMP inhibition of NHE3 and fails to bind ezrin (532). NHERF2 has a similar structure with its COOH-terminal 30 amino aicds being α-helical with four grouped positively charged amino acids on one side of the helix with mutation of three amino acids blocking ezrin binding and reconstitution of cAMP inhibition of NHE3 (B.Y. Cha, C. Yun, and M. Donowitz, unpublished data).

FIG. 7.

Model of putative NHE3 complexes containing NHERF1 or NHERF2 in epithelial cell brush border involved in acute cAMP inhibition of NHE3. The essential features are a 5-component complex that leads to PKA phosphorylation of NHE3. The proteins involved are NHE3, NHERF1 or -2, ezrin, actin, and PKAII.

There are many aspects of understanding this NHE3 complex involved in cAMP inhibition of NHE3, which are only partially defined, in addition to the specific site of NHE3 involved in NHERF binding described above.

  1. ) Blocking the NHE3 COOH terminus has been reported to reduce the cAMP inhibition by ∼25%, although we have not found any significant effect (504). Basal and growth factor stimulation of NHE3 activity and percent on the plasma membrane do not appear affected.

  2. ) Whether the NHE3/NHERF1 or NHERF2/ezrin/PKAII complex (Fig. 7) is static has not been definitively established. Studies in epithelial cells are required to address this issue, since trafficking is only part of NHE3 regulation by cAMP in a polarized epithelial cell model and does not contribute at all in PS120 fibroblasts. In fibroblasts, Yun et al. (532) showed that NHERF2 and NHE3 colocalize in the plasma membrane similarly both under basal conditions and after cAMP exposure but do not colocalize in the recycling juxtanuclear compartment under either basal conditions or after cAMP exposure.

  3. ) Given that direct ezrin binding to NHE3 occurs, why does this not appear to allow cAMP phosphorylation and regulation of NHE3 (74)? It is possible that the location of the direct ezrin/PKAII is not correct to allow the catalytic subunit of PKA II to phosphorylate NHE3. However, given that it is freed up catalytic subunit that is involved, this does not appear to be a likely explanation.

  4. ) Does the regulation of NHE3 via BB PDZ proteins apply to NHE2, which is also present in the small intestinal and colonic BB? While studies have not been done in adequate detail with NHE2 to comment definitively, there is no evidence of an interaction with NHERF proteins. Also, in Caco-2 cells and PS120 cells, cAMP stimulates rather than inhibits NHE2 (52, 53, 232, 292, 323, 347). Moreover, NHE2 appears to be present mostly in the plasma membrane, does not appear to traffick on and off the plasma membrane under basal conditions, is not affected by inhibitors of PI 3-K under basal conditions, and does not appear to bind NHERF1 or NHERF2, all characteristics different from NHE3 (69, 70, 347).

  5. ) Do these cell culture models of cAMP inhibition of NHE3 apply to intact models of renal proximal tubule and small intestine/colon? These questions have begun to be addressed in knockout models of NHERF1. It must be remembered that the initial suggestion for involvement of an accessory regulatory protein in BB Na+/H+ exchange activity came from studies in BB of renal proximal tubule. In NHERF1 knockout mice, cAMP (true for forskolin, cAMP, and PTH) failed to inhibit BB NHE3 in the renal proximal tubule. This was due to failure of cAMP to increase phosphorylation of NHE3 (409, 500). The failure of cAMP to inhibit NHE3 in NHERF1 knockout mice was true in BB vesicles made from kidney cortex, primary cultures of proximal tubules, and renal cortical slices acutely studied using a two-photon microscope using SNARF-4F to measure intracellular pH (338). The explanation is only partially determined, but in a second NHERF1 knockout mouse in the same C57Bl/6 background, changes in proteins involved in the NHE3 complexes identified in PS120 cells occurred (333). Specifically, NHERF1 knockout mice had less renal proximal tubule BB membrane ezrin and p-ezrin and increased NHERF2 (changes in NHERF2 and ezrin in renal proximal tubule BB were not seen in the Shenilokar and Weinman knockout mouse). The explanation for the failure for renal proximal tubule BB NHERF2 to reconstitute cAMP inhibition of NHE3 is not known, but we hypothesize that it is the lack of phosphorylated ezrin in the BB that explains lack of cAMP effect due to failure to provide AKAP function for NHE3 phosphorylation. This suggests that other AKAPs that are present in the BB cannot substitute for ezrin, perhaps because of failure to closely associate with NHE3 via NHERF family binding. Another suggested possibility is that NHERF2 may not directly bind NHE3 with sufficient affinity to allow NHE3 phosphorylation or may be in too low a concentration at the microvilli to allow this to occur. In contrast, in ileum of NHERF1 knockout mice, cAMP inhibits NHE3 similarly to wild-type mouse (338). Interestingly, ezrin and p-ezrin were also decreased in the ileal BB, suggesting that in ileum another AKAP is able to allow cAMP to inhibit NHE3. There are multiple AKAPs in the ileal BB and at least NHERF3/PDKZ1 is known to associate with d-AKAP2 (173, 174).

Regulation of NHE3 has been described in small intestine of the NHERF3/PDZK1 knockout mouse in preliminary form (93, 276). Amounts of NHERF1 and NHE3 in the small intestinal BB were normal as was the apparent NHE3 location. The kidney of these NHERF3/PDZK1 knockout mice has a normal-appearing BB, and the amount of NHERF1 is normal or there may be a small increase in NHERF1 amount when the animals are fed a high-protein diet (67). In contrast, in colon of NHERF3/PDZK1 knockout mice, not only is NHE3 activity decreased to 33% of wild type, but cAMP, cGMP, and elevated Ca2+ fail to affect colonic NHE3 activity (93). This identifies a major role for NHERF3/PDZK1 in NHE3 regulation in mouse colon, although mechanistic understanding is lacking.

6) Regulation of NHE3 by cAMP should be analyzed in the presence of additional proteins that also bind NHERF proteins, including DRA and CFTR? This is discussed in section vi.

II) cGMP inhibition of NHE3 activity.

cGMP amount in small intestinal Na+ absorptive cells is increased by luminal exposure to the intrinsic gut peptides guanylin or uroguanylin or the heat-stable E. coli enterotoxin by binding to their common receptor, the BB guanylate cyclase C (121, 123, 405). This is associated with inhibition of small intestinal NaCl absorption. Unexpectedly, in PS120 fibroblasts and OK cells, cGMP failed to affect NHE3 activity (72). Most cells in culture lack cGKII, which is normally present in the BB of epithelial cells (319, 462). However, cGMP failed to affect NHE3 even in cells transfected with cGKII (72). However, NHERF2 expression in both PS120 and OK cells reconstituted cGMP inhibition of NHE3 (72). NHERF1 or NHERF3/PDZK1 was unable to reconstitute cGMP inhibition of NHE3 (72; N. Zachos and M. Donowitz, unpublished data). NHERF2 and cGKII directly interacted as shown by in vitro pull-down/overlay assays and coprecipitated when both were expressed in fibroblasts and from ileal BB in which they were expressed endogenously (72). The specificity of NHERF2 but not NHERF1 to reconstitute cGMP inhibition of NHE3 may be due to failure of NHERF1 to bind cGKII as studied by in vitro overlay and pull-down assays (72). The recognized mechanism of the NHERF2 reconstitution of cGMP inhibition of NHE3 (Fig. 8) included that NHERF2 acted as a cGMP kinase anchoring protein (GKAP) by binding cGKII (72). In the NHE3/NHERF2/cGKII complex, which was involved in cGMP inhibition of NHE3, cGKII had to be anchored in the BB by both binding to the PDZ2-COOH terminus of NHERF2 and directly to the BB by myristoylation. In preliminary studies, cGMP inhibition of NHE3 was associated with cGMP-dependent phosphorylation of NHE3 (B.Y. Cha, H. Kocinsky, P. Aronson, and M. Donowitz, unpublished data). The source of elevated cGMP at the BB in intestine and kidney is BB guanylate cyclase, which is membrane anchored by binding to IKEPP (405, 442). A function of NHERF4/IKEPP appears to be to reduce the level of generated BB guanylate cyclase (405). Unknown is whether the entire signaling process (shown in Fig. 8) is part of a single NHERF family scaffolded complex.

FIG. 8.

Model of putative NHE3 complexes involving NHERF2 in epithelial cell brush border involved in acute cGMP inhibition of NHE3 and NHERF4/IKEPP in inhibition of guanylate cyclase C (gcc). The essential features are a 3-component complex that leads to PKG phosphorylation of NHE3. The proteins involved are NHE3, NHERF2, and cGKII. Ezrin is not necessary for this process, perhaps because the complex is further fixed to the BB by myristoylation of cGKII.

III) Elevated Ca2+ inhibition of NHE3 activity.

Ca2+ is acutely elevated in intestinal Na+ absorptive cells by the cholinergic agonist carbachol, exposure to which mimics part of digestive physiology (129, 388). Similar elevation of intracellular Ca2+ in these cells occurs with serotonin, mimicking part of the pathophysiology of the diarrhea of carcinoid syndrome (168). Ca2+-dependent inhibition of NHE3 has been demonstrated in PS120 fibroblasts and rabbit ileal Na+ absorptive cell BB to involve a two-step process in which NHE3 complexes rapidly form at least partially on the plasma membrane, enlarge, and then internalize (Fig. 9) (250, 286, 298). This decreases the amount of plasma membrane NHE3, with inhibition of NHE3 activity being due to the decrease in amount of surface NHE3 (250, 286). In PS120 cells, elevation of Ca2+ failed to affect NHE3 activity unless the cells were transfected with NHERF2 but not NHERF1 (250). Ca2+ elevation in NHERF2-containing cells was associated with formation of complexes that included NHE3, α-actinin-4 (non-muscle actinin), and PKC-α, with effects on NHE3 activity and on complex formation occurring within minutes of initial Ca2+ elevation (250, 286, 298). Actinin-4 involvement was determined using a proteomic approach in which glutathione-S-transferase (GST)-NHERF2 but not GST-NHERF1 pull-downs of ileal lysate isolated a protein of ∼100 kDa which was identified by mass spectroscopy as α-actinin-4 (250). Ability of NHERF2 but not NHERF1 to bind actinin-4 appears to explain the specificity of NHERF2 in reconstituting Ca2+ inhibition of NHE3 (250). NHERF2 coprecipitated NHE3 and actinin-4 under basal and elevated Ca2+ conditions. While the association of NHERF2 and actinin-4 increased with elevated Ca2+, that with NHE3 was Ca2+ independent. Oppositely, IP actinin-4, coimmunoprecipitated NHERF2 and NHE3 in a Ca2+-dependent manner. The explanation appears to be that, with elevated Ca2+, the association of actinin-4 with NHERF2 was increased, and since the ratio of NHERF2/NHE3 binding was Ca2+ independent, this led to more NHE3 coprecipitation as well. That actinin-4 was involved in Ca2+ inhibition of NHE3 was demonstrated by expressing the different actinin-4 domains separately in cells containing NHE3 and NHERF2 and searching for dominant negative effects (250). Actinin-4 is made up of actin binding, spectrin homology, and EF hand domains. The actinin-4 construct containing the actin binding plus spectin homology domains but lacking the EF hands domains prevented elevated Ca2+ from inhibiting NHE3, while the other actinin-4 constructs, including full length, increased the Ca2+ inhibition of NHE3. It was concluded that not only did α-actinin-4 join NHE3 complexes with elevated Ca2+, but was necessary for Ca2+ inhibition of NHE3 via its EF hand domain. We speculate that local Ca2+ elevation involving the EF hand domains of actinin-4 is involved in NHE3 inhibition, although the specific step involved is unknown. The Ca2+ inhibition of NHE3 was associated with NHE3 entering larger complexes, which was both observed on SDS-PAGE gels and visualized as clumping in PS120 cells by confocal light microscopy. These complexes contained at least NHE3, NHERF2, actinin-4, and PKC-α (Fig. 9). On the basis of cell surface biotinylation, some of these complexes could be seen forming on the plasma membrane, but it is not known whether they also form intracellularly (250). Temporally, cell surface NHE3 complexes formed before an increase in intracellular NHE3 was demonstrable, suggesting that complex formation preceeded internalization (250). That over time intracellular clumps containing NHERF2, NHE3, actinin-4, and PKC-α were visible suggested that the complexes were taken up together or formed intracellularly. However, trafficking of the complexes as a unit has not been demonstrated, and it is not known whether the large complexes are present intracellularly in epithelial cells of intact ileum or proximal tubules. PKC is involved in elevated Ca2+ inhibition of NHE3 in intact rabbit ileal Na absorptive, but neither calmodulin nor calmodulin kinase appears to be involved (96, 121, 123, 125, 127, 128, 131). This was also the case in PS120 cells expressing NHE3 and NHERF2. Based on biochemical and inhibitor studies (isoform specific pseudosubstrate inhibitors), PKC-α was shown to be the PKC isoform that joined the NHE3 complexes and was necessary for Ca2+ inhibition of NHE3 (286). PKC-α directly bound NHERF2 based on pull-down assays. PKC-α bound to PDZ domain 1 of NHERF2 coprecipitated with NHERF2 in a Ca2+-dependent manner and colocalized with NHE3, NHERF2, and actinin-4, based on confocal microscopy, after Ca2+ elevation both on the plasma membrane and within the cell (PS120 cells). The specificity of NHERF2 in Ca2+ inhibition of NHE3 was not related to PKC-α binding, since PKC bound both to NHERF1 and NHERF2 in a Ca2+-dependent manner (286). The role of PKC was not in formation of the NHE3 complexes, but rather in internalization of NHE3. PKC did not appear to affect the phosphorylation status of NHE3, suggesting that its control of endocytosis was via phosphorylation of components of the endocytosis machinery, as has been demonstrated previously (286). Of interest is that the PKC appears to be able to affect phosphorylation of the endocytosis machinery from the cytosolic face.

FIG. 9.

Model of putative NHE3 complexes involving NHERF2 in elevated Ca2+ inhibition of NHE3. Complexes initially form consisting of NHE3, NHERF2, α-actinin-4, and protein kinase C (PKC)-α. NHE3 is endocytosed. It is not known whether NHE3 is endocytosed with the entire complex or with any of its associating proteins. The essential features are a 4-component complex that leads to Ca2+-dependent/PKC-dependent inhibition of NHE3. The proteins involved are NHE3, NHERF2, α-actinin-4, and PKC-α.

Many aspects of Ca2+ inhibition of NHE3 in overexpressing PS120 fibroblasts were duplicated by studies in rabbit ileal Na+ absorptive cell BB and mouse ileal villus Na+ absorptive cells (93, 122, 123, 125, 128, 298; R. Murtazina, O. Kovbasnjuk, and M. Donowitz, unpublished data). Acute Ca2+ elevation with carbachol or serotonin inhibited BB NHE3 activity and surface amount to similar degrees. This was associated with an increase in amount of NHE3 in endosomes, consistent with increased endocytosis, although a role for decreased exocytosis has not been excluded (298). This inhibition of NHE3 activity was associated with an increase in size of NHE3 complexes in the ileal BB based on sucrose density gradient fractionation (298). Actinin-4 similarly shifted into larger complexes of similar size to those of NHE3 (298). In addition, after Ca2+ elevation, the same proteins that appeared to form NHE3 complexes in PS120 cells also were found in NHE3 complexes in ileal BB. Immunoprecipitated NHERF2, coprecipitated actinin-4, PKC-α, and activated PKC with increased association with Ca2+ elevation (298). The amount of actinin-4 that was immunoprecipitation from the BB was increased with elevated Ca2+ consistent with the freed up actinin-4 in PS120 cells. However, different sources are likely, with actinin-4 moving from the adhesion plaques to the NHE3 complexes in PS120 cells, while it probably is mobilized from the tight junctions in ileal Na+ absorptive cells. In ileal Na+ absorptive cells, the carbachol effects on Na+ absorption were c-Src dependent, with carbachol stimulating c-Src phosphorylation and the Src family blocker PP2 preventing carbachol inhibition of Na+ absorption and the increase in size of complexes containing NHE3 and actinin-4 (298). In preliminary studies, in mouse ileum, serotonin inhibition of NHE3 is NHERF2 dependent (R. Murtazina, O. Kovbasnjuk, R. Sharker, H. deJonge, B. Hogema, U. Seidler, E. Weinman, and M. Donowitz, unpublished data).

IV) LPA stimulation of NHE3.

In addition to second messenger inhibition of NHE3, an example of acute stimulation that is also dependent on NHERF2 but not NHERF1 was demonstrated (87, 285). LPA, considered a mediator of restitutive aspects of inflammation, stimulates NHE3 activity, as studied in the apical membrane of polarized OK cells. LPA has no effect on NHE3 activity in these cells, which lack endogenous NHERF2, unless cells transfected with NHERF2 are studied (285). This stimulation of NHE3 activity was associated with and was equivalent in magnitude to the increase in amount of BB NHE3 and was due to an increase in rate of exocytosis. There was no change in the rate of endocytosis of surface NHE3. The LPA stimulation of NHE3 was PI 3-K dependent (285). The mechanism of the NHERF2 dependence of this LPA effect was not in the degree of PI 3-K activation, which was similar in OK cells that contained or lacked NHERF2. Surprisingly, the mechanism appears to involve the previously reported stimulation of intracellular Ca2+ in response to LPA (119). LPA increased intracellular Ca2+ in OK cells, and the magnitude and time of elevation was increased in OK cells expressing NHERF2 (87). Moreover, the LPA stimulation of NHE3 was Ca2+ dependent, being prevented by BAPTA-AM (87). This effect further involved phospholipase C (PLC) activity, since the PLC inhibitor U73122 prevented LPA stimulation of NHE3, with likely involvement of PLC-β3, the only PLC-β isoform expressed in these cells (87). In the presence of NHERF2, LPA caused a larger and more prolonged elevation of intracellular [Ca2+] (87, 423). Thus increased PLC activity in OK cells occurred only with NHERF2 and not NHERF1, which is consistent with the previous demonstration that NHERF2 but not NHERF1 binds PLC-β3, although both bind the B1 and B2 isoforms (356). In contrast to the role of elevated Ca2+ in inhibition of NHE3, which is mediated via PKC-α, a general PKC inhibitor did not alter LPA stimulation of NHE3. The role of the elevated intracellular Ca2+ in LPA stimulation of NHE3 is not known, although the final steps of exocytosis and Golgi plasma membrane trafficking, including V and T SNARE function are Ca2+ dependent, including the very distal step in exocytosis that involves synaptotagmin.

While the involvement of elevated Ca2+ in NHE3 stimulation via exocytosis initially was not expected, given the above description of the role of elevated Ca2+ acting via PKC in NHE3 inhibition by carbachol and serotonin, another almost identical example of stimulation of NHE3 by a PLC-dependent mechanism has been reported. Adenosine acting via A1 receptors at a low concentration (∼10−9 M) rapidly stimulates NHE3 activity (15 min), while at higher concentrations (maximum effect at 10−6 M) it inhibits NHE3 (114, 117). The adenosine stimulation of NHE3 was Ca2+ dependent, being prevented by BAPTA-AM, involved PLC, since the PLC inhibitor U73122 inhibited the effect, was not dependent on PKC activity, and appeared dependent on an increase of Ca2+ from intracellular stores (117). The mechanism of the elevated Ca2+ stimulation of NHE3 in OK cells was postulated to involve calcineurin homologous protein and also inhibition of adenylate cyclase (115, 117). This was based on the adenosine agonist CPA opposing the effects on NHE3 activity of 8-bromo-cAMP (117). However, no biochemistry was provided, and the effects of CPA and cAMP were additive, consistent with separate mechanisms rather than CPA acting by inhibiting adenylate cyclase to decrease the cAMP content. Thus the paradox of elevated Ca2+ under some circumstances stimulating NHE3 occurs with 1) LPA exposure (87, 285), 2) low concentrations of adenosine (114, 117), and 3) IKEPP expression in PS120 cells expressing NHE3 is associated with ionomycin (Ca2+)-induced stimulation of NHE3 (N. Zachos and M. Donowitz, unpublished data). We hypothesize that the more common response of Ca2+-induced NHE3 inhibition by stimulated endocytosis of NHE3 and consequent inhibition of activity requires formation of a NHERF family complex that involves NHE3 and leads to changes in trafficking. However, in the absence of any necessary component of the complex, the Ca2+ stimulation of exocytosis that occurs in every cell and appears to involve synaptotagmin as a final step that is responsive to elevated Ca2+ (51, 349) is unmasked. We suggest that this functions even during the Ca2+ inhibition, since exocytosis continues.

V) Luminal PTH receptor-NHERF complex as a molecular switch.

BB NHE3 is inhibited by both elevation in cAMP and Ca2+, although the NHE3 complexes involved appear to be distinct. This concept is supported by additivity of NHE3 inhibition with elevation of cAMP plus Ca2+ in PS120/NHE3 cells expressing NHERF2 (B.Y. Cha and M. Donowitz, unpublished data). This interaction of cAMP and Ca2+ is relevant for PTH regulation of NHE3. The PTH receptor has a COOH-terminal class I PDZ domain binding motif which is necessary for its binding to NHERF2 (153, 316). In PS120 cells expressing the PTH receptor and NHERF2, PTH increased inositol phosphates 20- to 25-fold and had a minimal effect (2-fold) on cAMP content, while in similar cells lacking NHERF2, PTH increased cAMP 10- to 15-fold and minimally increased inositol phosphates (2–4 times) (316). Pertussis toxin treatment both inhibited PTH activation of PLC and also partially restored PTH activation of adenylate cyclase. Thus it was suggested that NHERF2 and perhaps NHERF1 (426) act as a molecular switch allowing PTH to activate a PLC and to inhibit activation of adenylate cyclase to increase cAMP (316). NHERF2 not only is involved with activation of a PLC, but is associated with inhibition of adenylate cyclase probably via activation of a Gi/o. NHERF2 binds PLC-β3 at its in PDZ domain 1, while it interacts with the PTH receptor with its PDZ2-COOH terminus, suggesting that even though PTH added to the apical surface inhibited NHE3, NHE3 and the PTH receptor were probably not physically associated with the same NHERF2 molecule. It is not known whether they were in the same large complexes. This model held true in an epithelial cell (315, 317). In OK cells, NHERF1 assembles an apical signaling complex that includes the PTH1 receptor PLC-β3, NHE3, and actin. Also, in mouse proximal tubule from NHERF1 knockout versus wild-type mice, PTH receptor coupling to PLC activity was NHERF1 dependent (66). In contrast, activation of Gi/o to inhibit adenylate cyclase activity did not occur. This leaves open the question of how widespread is the action of NHERF members as a molecular switch between adenylyl cyclase and Ca2+ signaling.

VI) Glucocorticoids and serum and glucocorticoid-inducible kinase 1.

Glucocorticoids stimulate intestinal and renal proximal tubule NaCl absorption (13, 15, 37, 50, 77, 79, 85, 188, 231, 247, 360, 361, 399, 435, 528531), an effect previously thought to start in ∼18 h, but recently shown to occur even faster (477, 528, 529). Until recently, the mechanism was unclear with stimulation of Na+-K+-ATPase and transcriptional stimulation of NHE3 through increased NHE3 mRNA (NHE3 has a glucocorticoid response element in the 5′-flanking region), previously shown to be involved, although what percent of this response they provided was unknown. Recently, Yun, Lang, and co-workers (477, 528, 529) showed that glucocorticoids also stimulated NHE3 activity within 3–4 h by a process that did not require stimulation of transcription, although some of the stimulation of NHE3 activity may have been transcription dependent. This process required NHERF2 but not NHERF1 and was associated with binding of serum- and glucocorticoid-inducible kinase 1 (SGK1) to NHERF2 but not to NHERF1 in vivo and in vitro (529). SGK1 is an Akt-related, Ser/Thr kinase. With the use of in vitro interaction assays, it was suggested that the NHERF2-SGK1 interactions predominantly used the NHERF2 PDZ-2 domain, although data were contradictory (90). Glucocorticoids increased the amount of SGK1 by transcription activation, providing a mechanism by which glucocorticoids could affect NHE3 activity separately from the documented increase in NHE3 mRNA that occurs with a slightly later time course (531). Glucocorticoid stimulation of NHE3 activity was PI 3-K dependent, being inhibited by the PI 3-K inhibitor LY294002, and also required PDZK1 (399, 477, 529). The 3-phosphoinositide-dependent protein kinase 1 (PDK1), which phosphorylates and activates SGK1, bound the COOH terminus of NHERF2 which contained a PIF sequence (FXXF=FSNF). PIF sequences have been shown to bind PDK1. In fact, PDK1 and NHERF2 coprecipitated, indicating that they interacted in vivo. That SGK1 (but not SGK2 or SGK3) was involved with stimulation of NHE3 activity was demonstrated by studies using kinase dead SGK1 as a dominant negative inhibitor of glucocorticoid stimulation of NHE3 activity. PDK1 hypoexpressing mice had marked reduction of glucocorticoid stimulation of NHE3 (399)

The mechanism of the NHERF2 effect in SGK1-dependent stimulation of NHE3 in epithelial cells in response to glucocorticoids is not known, although it is assumed that it involves anchoring of SGK1 to the BB in close apposition to PDZK1 or NHE3 and perhaps provides the PIF sequence to place PDK1 near SGK1 so that the phosphorylated and activated SGK1 can activate NHE3. SGK1 phosphorylates NHE3 at S663, mutation of which blocks the glucocorticoid stimulation of NHE3 (477). Also, SGK1 was shown to phosphorylate T102 of NHERF2 in vitro, the first suggestion that NHERF2 is phosphorylated (529). However, whether this occurs in vivo or is an in vitro artifact is not known, and NHERF2 has not been shown to be phosphorylated under any conditions in vivo. Thus glucocorticoid stimulation of NHE3 activity is a complex combination of transcriptional stimulation of NHE3 and SGK1, with SGK1 binding to NHERF2 where it is activated by PDK1 to stimulate NHE3 separately from transcription by a PI 3-K-dependent process which would be predicted to involve increased NHE3 exocytosis and requires phosphorylation of NHE3.

9. Other BB PDZ proteins in NHE3 regulation

In addition to the four NHERF family proteins, Shank2 is another BB PDZ domain containing protein involved in NHE3 regulation, although where it binds NHE3 is not known (194). Shank2, a single PDZ domain containing protein, is present in the BB of some epithelial cells, including Caco-2, T84, and rat colon surface and crypt cells. Shank2 or CortBP (SH3 and ankyrin repeats containing protein 2/cortactin binding protein) is a member of another gene family of PDZ proteins with three members (Shank 1–3). The role of Shank2 in regulation of transporters includes the following: 1) binds NHE3, CFTR, and NaPiIIa in a yeast two-hybrid screen (251) and 2) decreases the effect on NHE3 of cAMP when studied in fibroblasts. Thus Shank2 has an effect different from the NHERF family and suggests that up- and downregulation of NHE3, which is part of physiological regulation of NHE3, involves different classes of PDZ domain proteins. 3) Shank2 has a similar function in CFTR regulation, opposing the NHERF stimulation of CFTR phosphorylation by cAMP (251) with effects on both basal and cAMP stimulation. Lack of ability to quantitate extent of association of Shank2 with substrates has limited most studies of NHERF family and Shank2 interaction, although NHE3 and Shank2 have been shown to coprecipitate (194). Shank2 was expressed in PS120 cells with NHE3; Shank 2 increased the membrane expression and basal activity of NHE3 but prevented the cAMP inhibition of NHE3. Oppositely, knockdown of endogenous Shank2 in Caco-2 cells decreased the expression of NHE3 protein with an accompanying decrease in NHE3 activity and an increased cAMP inhibitory response (194). 4) Thus, in addition to the apical membrane NHERF family, an additional apical membrane PDZ protein is involved in regulation of regulated NHE3 activity and amount. Where NHE3 binds Shank2 and how Shank2 interacts with NHERF family proteins and coordinates their effects in NHE3 regulation in intact tissue is an unknown but important area for future study.

10. Regulation of NHERF proteins

Thus scaffolding by multiple NHERF family members of NHE3 complexes is involved in rapid changes in NHE3 activity. Regulation of the NHERF proteins also occurs but is only beginning to be understood. This involves at least agonist binding-related multimerization and phosphorylation (150, 192, 199, 501, 547). Binding to substrate by the NHERFs is thought to be constitutive, although with low affinities. Interaction with ligands in some cases depends on activation of the ligand (191, 343). For instance, NHERF1 only binds activated β2-adrenergic receptor (191). The Ser/Thr kinase GRK-5 phosphorylates the β2-adrenergic receptor −2 position Ser, which leads to uncoupling from NHERF1 binding. However, while not well defined, some regulatory functions of NHERF1 do not depend on changes in its phosphorylation (277, 501, 547).

NHERF1 (106, 150, 199, 277, 280, 386, 547) and NHERF3/PDZK1 (106, 205, 206) are phosphoproteins under basal conditions, while NHERF2 has not been shown to be phosphorylated in vivo under basal or any regulated condition. However, all the NHERF proteins, including NHERF2, have phosphorylation consensus sequences (Scansite). Of importance, NHERF1 is phosphorylated by multiple kinase including GRK-6A (S289), which increases its dimerization (192). It is also phosphorylated by cyclin-dependent kinase II and dephosphorylated by PP1 and PP2A at S279 and S301 (199). cAMP inhibition of NHE3 is not associated with changes in NHERF1 phosphorylation (277). Moreover, mutation of a phosphorylated amino acid of NHERF1 did not alter cAMP regulation of NHE3, demonstrating that the NHERF1 phosphorylation was not necessary for cAMP regulation of NHE3 or coprecipitation with NHE3 (277, 547). This is based on studies in PS120 cells, and it is not established whether this holds true for polarized cells. PKC also phosphorylates NHERF1 in its PDZ2 domain (150). This is associated with a lowering of the affinity of NHERF1 for CFTR and disrupting NHERF1-CFTR binding and the NHERF1 increase in CFTR-related current (103, 302, 381). PDZK1 phosphorylation at S509 has been demonstrated, and there is suggestion of stimulation of the extent of PDZK1 phosphorylation in response to glucagon in hepatocytes (343).

11. Homo- and heteromultimerization of NHERF family

Multimerization of NHERF1, NHERF2, and NHERF3/PDZK1 all occur mostly based on in vitro studies. However, reports of whether these interactions depend on the phosphorylation status of the NHERF protein and which of their PDZ domains are involved are not consistent from study to study. The NHERF protein multimerization suggests that all (or most) members of this family might be included in the same NHE3 complexes and bring different ligands into the complexes, akin to the large complexes at different distances from the plasma membrane that occurs with the postsynaptic density complexes scaffolded by INAD (455).

C. COOH-Terminal Domain 3, Amino Acids 661–756

There are no identified binding partners.

D. COOH-Terminal Domain 4, Amino Acids 757–832: DPPIV(?), PP2A, and Megalin

A review of phosphatase regulation of NHE3 activity serves as an introduction. NHE3 is phosphorylated under basal conditions in all cells examined. Okadaic acid, a PP2A inhibitor, stimulates activity of rabbit NHE3 expressed in fibroblasts (293). It requires aa 509–585 in the NHE3 COOH terminus. This indicates that PP2A under basal conditions inhibits NHE3 activity (293) and that an unknown kinase(s) stimulates NHE3 activity under basal conditions. In contrast, the tyrosine kinase inhibitor genistein inhibits NHE3 activity, and this effect requires the COOH-terminal 96 aa of NHE3 (245, 246; Donowitz, unpublished data). This is further evidence for a role for a balance of kinases/phosphatases in NHE3 regulation. However, that NHE3 is not phosphorylated on Tyr (525) suggests this effect occurs via an intermediate which is Tyr phosphorylated.


DPPIV (also called CD26) is a BB serine protease which cleaves NH2-terminal dipepides from peptides with a penultimate Pro or Ala residue. It is present in large amounts in the BB of renal proximal tubule and small intestine (241). The recent demonstration that the epithelial Na+ channel ENaC is regulated by an extracellular protease has increased interest in regulation of transport protein function by proteases (392). The functional association of DPPIV with NHE3 was examined after demonstration that NHE3 physically interacted with DPPIV based on use of monoclonal antibodies to immunoprecipitate NHE3 from OK cells and search for interacting proteins (170). The association involves microvillus NHE3 but appears to be indirect. DPPIV effects on NHE3 require aa 702–756, although there is no evidence that this is the site at which DPPIV binds NHE3 (170, 171). DPPIV activity was necessary for NHE3 activity, since DPPIV inhibitors inhibited NHE3 activity (171). This inhibition was not associated with a change in the amount of BB NHE3 nor with a change in the size of NHE3 complexes. Furthermore, the DPPIV inhibitor effect was not altered by the PKA inhibitor H-89, but was inhibited by the tyrosine kinase inhibitor genistein. Not only did genistein inhibit basal NHE3 activity, as we previously reported, but the effects of the DPPIV inhibitor and genistein were not additive (171). Thus it appears that a tyrosine kinase pathway is involved in the steps in signal transduction that are affected by DPPIV which are involved in NHE3 stimulation. A protein of 212 kDa changed in Tyr phosphorylation in parallel with the effects of genistein and the DPPIV inhibitor, suggesting that it might be involved in regulation of NHE3 (171). The mechanism of DPPIV in regulated NHE3 activity as well as the signaling pathways involved in its effects are unknown.

A recent study has called into question the specificity of the DPPIV-NHE3 interactions (177). DPPIV is a receptor in the invasive human prostate tumor cell line 1-LN for plasminogen type II as well as angiostatin. These cells were suggested as expressing NHE1, -2, and -3. Plasminogen type II alkalinized these cells by an amiloride-sensitive process, which also inhibited their invasiveness, suggesting involvement of an NHE. Moreover, NHE1, -2, and -3 all coprecipitated DPPIV and plasminogen II. While these studies require confirmation, especially the specificity of the NHE involved, a strong functional link between DPPIV and Na+/H+ exchange activity is supported (177).

2. PP2A

It was expected that phosphatases might directly bind to NHE3 since 1) phosphorylation is important for regulation of NHE3; 2) kinases have been shown to associate with multiple NHE isoforms, including NHE1 and NHE3; and 3) kinase inhibitors and phosphatase inhibitors alter NHE3 regulation. Indeed, PP2A binds to the COOH-terminal 50 aa of NHE3 (aa 783–832) following dopamine exposure, presumably via the increased cAMP (380), which inhibits NHE3 activity. This binding was demonstrated in vivo based on coprecipitation studies in OK renal proximal tubule cells and was direct based on yeast two-hybrid and in vitro (pull-down) studies. The PP2A-B isoform, γ-subunit was involved. Surprisingly, okadaic acid blocked the dopamine effect on NHE3 activity. This indicated that the PP2A effect on dopamine inhibition of NHE3 was due to effects of PP2A separate from dephosphorylation of the increases in NHE3 phosphorylation that occurs with dopamine activation by PKA. Moreover, the demonstration that the domain of the NHE3 COOH terminus to which PP2A binds is not necessary for effects of okadaic acid on basal NHE3 activity shows that another pool of PP2A is involved with regulation of basal NHE3 activity.

While not shown for NHE2, PP1 was demonstrated to be involved in NHE1 regulation by Fliegel and co-workers (327). NHE1 was dephosphorylated in vitro by PP1. Overexpression of PP1 decreased NHE1 activity, while PP1 inhibitors increased NHE1 activity. PP1 binds NHE1 both in vivo and in vitro.

3. Megalin

Megalin binds NHE3 and may be involved in its regulation (45, 46). Megalin is a large protein (gp330) and a member of the low-density lipoprotein receptor gene family (242), as reviewed in Reference 44. It is thought to be the major endocytic receptor of renal proximal tubule and binds albumin, lipoproteins, vitamin binding proteins including vitamin D binding protein, insulin, and drugs as well as NHE3 (44). In addition to its role in endocytosis, knockout of megalin in mice reduced the proximal tubule BB size and led to low-molecular-weight proteinuria (44, 287). Studies have largely been based in the renal proximal tubule, in which megalin is present in large amounts. Megalin also is present in the distal half of the rat small intestine and colon but not the proximal small intestine, and even in the distal small intestine the amount of message and protein present is a small fraction of that present in rat renal cortex (518). Megalin is also expressed in the mouse gallbladder (145). The proximal tubule and ileum also contain cubulin, which binds to and interacts with megalin (145). Cubulin is the intrinsic factor receptor. No studies have been reported of megalin interaction with NHE3 in the ileum. In the rabbit renal proximal tubule, NHE3 tightly binds megalin, and NHE3 exists as megalin-bound and megalin-free forms (45, 46). This was reviewed under trafficking of NHE3 above. The interrelationships among these processes have important implications. For instance, decreased albumin uptake in proximal tubule activates NHE3 production, among other genes (256), and increased distal kidney albumin uptake causes renal damage (44). Also, megalin knockout mice have abnormal NaPi2a trafficking and increased amounts of NaPi2a in the proximal tubule BB (26). As recently determined by Biemesderfer (44), megalin undergoes regulated intramembrane proteolysis, perhaps similar to notch, and is implicated in a PKC-metalloprotease-γ-secretase process which is involved in transcriptional regulation of additional proteins. This is a fertile area for study because of the implications that the loss of BB megalin in knockout mice might lead to changes in gene transcription, as well as abnormalities in NHE3 endocytosis. Neither of these has been adequately studied as yet.



The BB di/tripeptide transporter PEPT1 (SLC15A1) is present primarily in small intestinal BB, and to a lesser extent in proximal tubule BB (152, 351). It is H+ coupled, being stimulated by a transapical membrane H+ gradient (usually outside acidic corresponding to the luminal acid microclimate, although luminal acidity is not necessary), but is Na+ independent. This H+ gradient normally depends on BB NHE activity in small intestine and Caco-2 cells, requires extracellular Na+, is inhibited by a specific NHE3 inhibitor (S1611) and by elevating cAMP (which also inhibits NHE3), and is calmodulin dependent (19, 57, 148, 239, 240, 337, 446, 447, 485). Increase of Gly-Sar uptake via PEPT1 was proportional to NHE3 expression in a transfection system, showing NHE3 can drive PEPT1 function (485).

These effects are consistent with the involvement of NHE3 and not NHE2, since NHE3 is inhibited and NHE2 stimulated by cAMP. The dependence of hPEPT1 transport activity on NHE3 is consistent with direct or indirect linkage of BB NHE3 activity with PEPTI transport of di/tripeptides across the small intestinal BB. It is unknown whether the linkage of NHE3 and hPEPT1 involves other than the apically generated acid microclimate. The implications of this linkage are that in diarrheal diseases, in which there is inhibition of NHE3 activity (elevated cAMP, cGMP, Ca2+), it would be predicted that there would be inhibition of di/tripeptide uptake by PEPT1, although Na+-dependent l-amino acid uptake should be intact.

Similar to PEPT1, PEPT2 (SLC15A2) is also a renal proximal tubule apical membrane electrogenic H+-coupled peptide cotransporter that unlike PEPT1 is a high-affinity, low-capacity transporter (445, 446). It also uses an inward-directed H+ gradient to energize uptake of di- and tripeptides. It can couple to apical membrane NHEs, using their locally generated apical H+ microclimate and creating uptake equivalent to Na+/di-tripeptide cotransport. NHERF3 (PDZK1) binds PEPT2 in the proximal tubule apical membrane and increases its surface expression and its activity (351). It is not known whether this represents increased trafficking to the membrane or stabilization in the membrane. There is no evidence of direct binding of either PEPT1 or PEPT2 to NHE3.


In polarized epithelial cells, NHE1 is often on the BLM and NHE3 on the apical surface, and thus there is no question of physical association. Is there any evidence of functional association? In the rat renal medullary thick ascending limb, inhibiting NHE1 by amiloride, Na+ removal, or nerve growth factor (NGF) inhibits apical NHE3 activity (179, 488). This inhibitory effect on NHE3 did not occur in NHE1 knockout mice. This demonstrates the dependence of this interaction on NHE1 (180). The NHE1 inhibition-related downregulation of NHE3 was dependent on an intact cytoskeleton and was inhibited by both jasplakinolide and latrunculin B (488). That latrunculin disrupts F-actin while jasplakinolide stabilizes and prevents actin remodeling indicates that this cytoskeletal linkage of NHE1 activity with BB NHE3 activity is complicated, and no mechanisms for the linkage have been established. Similar inhibition of BB NHE3 by inhibition of BLM NHE1 has not been reported in either small intestine or proximal tubule.

C. SLC26A3 (DRA) and SLC26A6 (PAT1)

In small intestine and colon, part of intestinal Na+ absorption is via functionally linked Na+/H+ exchange and Cl/HCO3 exchange (20, 260). As elegantly shown, in BB vesicle studies, linkage was by intravesicular pH (259, 260). However, because BB vesicles do not contain all components of the apical signaling/regulatory mechanism, how linkage of Na+/H+ and the BB Cl/HCO3 exchangers (DRA and/or PAT1 as reviewed sect. ii) occur in intact intestine is not known. Both DRA (EKF) and PAT1 (TRL) end in amino acids resembling type 1 PDZ domain binding motifs. It has recently been shown that DRA like NHE3 bind to the second PDZ domain of NHERF2 and PDZ domains 2 and 3 of NHERF3/PDZK1, while PAT1 binds in vitro to NHERF3?PDZK1 PDZ domains 1 and 2 (275, 276). It is not known whether NHE3 and DRA or PAT1 are physically linked or whether they are present in the same BB complexes. However, some coordinated regulation is indicated by upregulation of DRA mRNA in NHE3 knockout colon (325).

SLC26A6 functions as the small intestinal Cl/oxalate transporter (151, 227). Mice lacking SLC26A6 develop Ca oxalate kidney stones, apparently due to inability to secrete oxalate in ileum; effects on Na+ absorption in this model have not been described. In congenital chloridorrhea, in addition to diarrhea since birth, there is decreased fertility. In the efferent duct of the epidydimus, NHE3, CFTR, and SLC26A3 normally are all present, but in congenital diarrhea, there is absence of CFTR in addition to SLC26A3 (207). The association of the same three proteins as is present in some intestinal Na+ absorptive cells suggests a functional interaction preserved across tissues as well as a role in male reproduction.


In small intestine and colon, neutral NaCl absorption, which occurs largely in the villus Na+ absorptive cells, occurs in parallel with Cl secretion, which occurs largely in the crypt. In the Cl secretory cell, apical Cl channels include CFTR and ClC-2. Given that the GI tract is involved in volume homeostasis and attempts to minimize loss of stool water and Na+, coordinated regulation of Na+ absorption and Cl secretion has been predicted. That this occurs both by changes in acute signaling and control at the transport protein level has been indicated by recent studies using knockout mice (159). In small intestine of NHE3 knockout mice, maximal cAMP-induced anion secretion was decreased, and this was associated with a decreased amount of CFTR protein and mRNA expression. This result was unexpected, since cAMP inhibits NHE3 in cells that lack CFTR (533). In small intestine of CFTR knockout mice, Na+ absorption was reduced in parallel with a decrease in NHE3 protein expression (94, 158, 159). The mechanism of coordinated regulation is not known. Of note, the opposite regulation of Na+ absorption with increased secretion occurred as well. In a mouse model of osmotic diarrhea, expression of both NHE3 and DRA mRNA increased (421).

Is there further evidence for coordinated control of these two regulated processes, and do NHE3 and CFTR physically interact? In the intestine, CFTR and NHE3 are mostly not in the same cells, although some CFTR appears to be in some villus Na+ absorptive cells (159). However, in pancreatic duct cells, which contain both NHE3 and CFTR, and in PS120 fibroblasts stably transfected with NHE3 and CFTR, there was physical association of NHE3 and CFTR based on coprecipitation (6). This required the COOH-terminal four amino acids (PDZ binding domain) of CFTR. In the pancreatic duct, there was similar parallel regulation of NHE3 and CFTR as occurred in mouse intestine, with the Δ508 CFTR mutant associated with decreased NHE3 amount and activity. Similarly in PS120 cells, the presence of CFTR but not CFTR lacking the COOH-terminal four amino acids increased the magnitude of cAMP inhibition of NHE3 (6). CFTR without the COOH-terminal four amino acids had no change in Cl channel activity but eliminated coprecipitation of NHE3 and reduced cAMP inhibition of NHE3. Thus CFTR regulates NHE3 activity, which appears to be another example of the regulatory function of CFTR, altering activity and/or amount of an associated transporter. In this case, CFTR appears to stabilize NHE3 in the brush border or plasma membrane and acutely increases the cAMP inhibition of NHE3. How much of the effects of Δ508 CFTR and reduced NHE3 expression in parallel in pancreatic duct are due to effects during development is unknown.

What about the converse, Does NHE3 regulate CFTR? In a renal cell model (A6 cell) which expressed endogenous CFTR and NHERF1, and in which NHE3 was stably expressed, the magnitude of PKA-dependent inhibition of CFTR was decreased with no change in CFTR amount (32). This effect required PKA increased phosphorylation at NHE3 Ser552 or Ser605 and required an AKAP, since it was inhibited by the AKAP competitive inhibitor S-HT31. Also, antisense suppression of CFTR resulted in less cAMP inhibition of NHE3. Thus CFTR and NHE3 physically associate with each other, at least when exogenously expressed in the same cell, by a complex that includes a NHERF family member, and they affect each other's regulation. A recent observation suggested that the NHE3/CFTR interaction (147) involving NHERF2 was similar to the β2-adrenergic receptor/NHE3 involving NHERF1 (191) in that the NHERF amount seems to be rate limiting. When NHE3 was expressed in cells that contain CFTR, it reduced PKA-dependent apical CFTR expression/activation by a process that required NHE3 be occupying its NHERF binding domain. This suggests a competition for locally expressed NHERF. However, the importance of this effect is not established physiologically, primarily because NHE3 and CFTR are usually not in the same cell. Concerning other potential mechanisms of linkage of CFTR/NHE3 regulation, changes in cell volume have been suggested (158). In the mouse jejunal mid villus area, both NHE3 and CFTR are present in the same cells (158). Here cAMP-activated anion secretion leads to cell shrinkage, with cell shrinkage known to inhibit NHE3 activity. This also indicates why lack of CFTR prevents cAMP effects on NHE3 activity, although it would not explain why cAMP phosphorylation of NHE3 is required (since it is not needed for shrinkage-induced inhibition of NHE3 activity). Our interpretation is that the data favor a functional linkage rather than a physical linkage of CFTR and NHE3, although the mechanism(s) for interactions are not adequately understood as yet.


This work was supported in part by National Institute of Diabetes and Digestive and Kidney Diseases Grants RO1-DK-26523, RO1-DK-61765, KO1-DK-62264, PO1-DK-44484, PO1-DK-72084, and R24-DK64388 (The Hopkins Basic Research Digestive Diseases Development Core Center) and by the Hopkins Center for Epithelial Disorders.


We thank Peter Aronson, Orson Moe, and Ursula Seidler for access to data prior to appearance in print.

We acknowledge the expert editorial assistance of H. McCann.

Address for correspondence: M. Donowitz, Ross 925, Johns Hopkins University School of Medicine, 720 Rutland Ave., Baltimore, MD 21205 (e-mail: mdonowit{at}


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
  65. 65.
  66. 66.
  67. 67.
  68. 68.
  69. 69.
  70. 70.
  71. 71.
  72. 72.
  73. 73.
  74. 74.
  75. 75.
  76. 76.
  77. 77.
  78. 78.
  79. 79.
  80. 80.
  81. 81.
  82. 82.
  83. 83.
  84. 84.
  85. 85.
  86. 86.
  87. 87.
  88. 88.
  89. 89.
  90. 90.
  91. 91.
  92. 92.
  93. 93.
  94. 94.
  95. 95.
  96. 96.
  97. 97.
  98. 98.
  99. 99.
  100. 100.
  101. 101.
  102. 102.
  103. 103.
  104. 104.
  105. 105.
  106. 106.
  107. 107.
  108. 108.
  109. 109.
  110. 110.
  111. 111.
  112. 112.
  113. 113.
  114. 114.
  115. 115.
  116. 116.
  117. 117.
  118. 118.
  119. 119.
  120. 120.
  121. 121.
  122. 122.
  123. 123.
  124. 124.
  125. 125.
  126. 126.
  127. 127.
  128. 128.
  129. 129.
  130. 130.
  131. 131.
  132. 132.
  133. 133.
  134. 134.
  135. 135.
  136. 136.
  137. 137.
  138. 138.
  139. 139.
  140. 140.
  141. 141.
  142. 142.
  143. 143.
  144. 144.
  145. 145.
  146. 146.
  147. 147.
  148. 148.
  149. 149.
  150. 150.
  151. 151.
  152. 152.
  153. 153.
  154. 154.
  155. 155.
  156. 156.
  157. 157.
  158. 158.
  159. 159.
  160. 160.
  161. 161.
  162. 162.
  163. 163.
  164. 164.
  165. 165.
  166. 166.
  167. 167.
  168. 168.
  169. 169.
  170. 170.
  171. 171.
  172. 172.
  173. 173.
  174. 174.
  175. 175.
  176. 176.
  177. 177.
  178. 178.
  179. 179.
  180. 180.
  181. 181.
  182. 182.
  183. 183.
  184. 184.
  185. 185.
  186. 186.
  187. 187.
  188. 188.
  189. 189.
  190. 190.
  191. 191.
  192. 192.
  193. 193.
  194. 194.
  195. 195.
  196. 196.
  197. 197.
  198. 198.
  199. 199.
  200. 200.
  201. 201.
  202. 202.
  203. 203.
  204. 204.
  205. 205.
  206. 206.
  207. 207.
  208. 208.
  209. 209.
  210. 210.
  211. 211.
  212. 212.
  213. 213.
  214. 214.
  215. 215.
  216. 216.
  217. 217.
  218. 219.
  219. 220.
  220. 221.
  221. 222.
  222. 223.
  223. 224.
  224. 225.
  225. 226.
  226. 227.
  227. 228.
  228. 229.
  229. 230.
  230. 231.
  231. 232.
  232. 233.
  233. 234.
  234. 235.
  235. 236.
  236. 237.
  237. 238.
  238. 239.
  239. 240.
  240. 241.
  241. 242.
  242. 243.
  243. 244.
  244. 245.
  245. 246.
  246. 247.
  247. 248.
  248. 249.
  249. 250.
  250. 251.
  251. 252.
  252. 253.
  253. 254.
  254. 255.
  255. 256.
  256. 257.
  257. 258.
  258. 259.
  259. 260.
  260. 261.
  261. 262.
  262. 263.
  263. 264.
  264. 265.
  265. 266.
  266. 267.
  267. 268.
  268. 269.
  269. 270.
  270. 271.
  271. 272.
  272. 273.
  273. 274.
  274. 275.
  275. 276.
  276. 277.