Glucocorticoid activation of Na(+)/H(+) exchanger isoform 3 revisited. The roles of SGK1 and NHERF2.

The stimulative effect of glucocorticoids on intestinal salt and water absorption has been known for more than two decades. However, molecular mechanisms underlying this activation remain elusive. Previous studies showed that methylprednisolone specifically increased Na(+)/H(+) exchanger isoform (NHE) 3 mRNA in ileum and kidney without affecting NHE1 mRNA levels. These results suggest that glucocorticoids activate NHE3 activity by induction of NHE3 transcripts. We recently found in PS120 and opossum kidney cells that chronic incubation with dexamethasone activated NHE3 independent of gene induction, indicating that the transcriptional activation may not be the only determining factor in the NHE3 activation. Furthermore, dexamethasone activated NHE3 activity only in the presence of a NHE3 regulatory protein, NHERF2, which was previously shown to confer cAMP-dependent inhibition of NHE3. This activation of NHE3 could not be duplicated by NHERF1. We identified serum- and glucocorticoid-induced protein kinase, SGK1, as the protein interacting with PDZ domains of NHERF2 to regulate NHE3 activity. The expression of SGK1 enhanced NHE3 transport in PS120 fibroblasts. In addition, the "kinase-dead" SGK1 blocked activation of NHE3 by dexamethasone in opossum kidney cells. These data demonstrated that glucocorticoid activation of NHE3 requires the activation of SGK1 and the presence of NHERF2 acting as a scaffold protein.

The stimulative effect of glucocorticoids on intestinal salt and water absorption has been known for more than two decades. However, molecular mechanisms underlying this activation remain elusive. Previous studies showed that methylprednisolone specifically increased Na ؉ /H ؉ exchanger isoform (NHE) 3 mRNA in ileum and kidney without affecting NHE1 mRNA levels. These results suggest that glucocorticoids activate NHE3 activity by induction of NHE3 transcripts. We recently found in PS120 and opossum kidney cells that chronic incubation with dexamethasone activated NHE3 independent of gene induction, indicating that the transcriptional activation may not be the only determining factor in the NHE3 activation. Furthermore, dexamethasone activated NHE3 activity only in the presence of a NHE3 regulatory protein, NHERF2, which was previously shown to confer cAMP-dependent inhibition of NHE3. This activation of NHE3 could not be duplicated by NHERF1. We identified serum-and glucocorticoid-induced protein kinase, SGK1, as the protein interacting with PDZ domains of NHERF2 to regulate NHE3 activity. The expression of SGK1 enhanced NHE3 transport in PS120 fibroblasts. In addition, the "kinase-dead" SGK1 blocked activation of NHE3 by dexamethasone in opossum kidney cells. These data demonstrated that glucocorticoid activation of NHE3 requires the activation of SGK1 and the presence of NHERF2 acting as a scaffold protein.
Adrenal steroids play a central role in the maintenance of fluid and electrolyte balance (1)(2)(3). Glucocorticoids such as prednisolone and methylprednisolone have been used to treat inflammatory bowel diseases such as ulcerative colitis and Crohn's disease (4,5). Although in some of these circumstances the antidiarrheal effects of glucocorticoids may result from the direct effect on processes underlying the diseases, it is widely accepted that glucocorticoids have a direct effect on intestinal salt and water absorption. The fact that glucocorticoids have a direct effect on intestinal salt and water absorption has been known for more than two decades (2,6). Pharmacologic doses of methylprednisolone have been shown to increase Na ϩ , Cl ϩ , and water absorption in vivo in both small intestine and colon of the rat. However, the molecular mechanisms by which this homeostasis is achieved at the cellular level remain to be elucidated.
The small intestine and colon are major sites for NaCl absorption, and much of the Na ϩ absorption in the gastrointestinal tract appears to be mediated by coupled Na ϩ /H ϩ exchange and Cl Ϫ /HCO 3 Ϫ exchange (7). In mammalian intestine, Na ϩ /H ϩ exchanger isoform (NHE) 1 3 is a component of neutral NaCl absorption, which is a major constituent in basal Na ϩ absorption and a frequent target of inhibition in many diarrheal diseases (7). NHE3 also plays an essential role in the maintenance of systemic pH homeostasis via Na ϩ and HCO 3 Ϫ reabsorptive processes in renal proximal tubules (8).
Several years ago, we demonstrated that 1-3 days of treatment with methylprednisolone doubled rabbit ileal neutral NaCl absorption by specifically stimulating NHE3 mRNA expression by 4 -6-fold without affecting NHE1 (9). Glucocorticoids also stimulate NHE3 in renal proximal tubules by increasing NHE3 mRNA level (10). Subsequent genomic cloning of rat NHE3 revealed the presence of glucocorticoid response elements in the 5Ј-flanking promoter region (11).
Na ϩ /H ϩ exchangers display sensitivity to changes in intracellular pH and cell volume and are modulated by agents primarily targeting serine/threonine and tyrosine kinases. Despite a great deal of effort to understand protein kinase/growth factor-dependent regulation of Na ϩ /H ϩ exchangers, the mechanisms underlying the modulation of transport activity are only partially defined. We have previously identified NHERF (also known as EBP50 and NHERF1) and E3KARP (also known as NHERF2 and TKA-1), each containing a pair of PDZ protein interaction modules (12). Because NHERF and E3KARP are closely related in structure and sequence and appear to belong to the same family of proteins, E3KARP was also referred to as NHERF2. To be consistent with the nomenclature of the family, E3KARP will be referred to as NHERF2 in this report. We demonstrated that these proteins confer modulation of NHE3 activity in response to increased intracellular cAMP level. NHERF1/NHERF2 function as scaffold proteins clustering NHE3 and cytoskeletal ezrin (13)(14)(15). Ezrin itself is capable of binding filamentous actin, serving as a bridge between membrane-associated proteins and cytoskeletons. The importance of the linkage to cytoskeletal network by NHERF proteins was also demonstrated in a recent report that NHERF2 is necessary in maintenance of foot processes in glomerular epithelial cells by linking podocalyxin to ezrin-actin network (16).
Serum-and glucocorticoid-inducible kinase 1 (SGK1) is a serine/threonine kinase, which was originally identified through subtractive cloning of a serum and glucocorticoid-induced mammary tumor cell cDNA library (17). In addition to serum and glucocorticoids, SGK1 is induced by various stimuli such as follicle-stimulating hormone, aldosterone, hyperosmolarity, protein kinase A, expression of p53, and injury to the brain (18 -20). Insulin and insulin-like growth factors stimulate SGK activity by a mechanism requiring the participation of phosphatidylinositol 3-kinase (PI 3-kinase) (19 -21). SGK1 is ubiquitously expressed in a wide variety of tissue, including intestine and kidney. Recently two additional SGK isoforms, termed SGK2 and SGK3, have been cloned (19).
In this study, we present new evidence that glucocorticoids result in NHE3 activation via a mechanism other than NHE3 gene activation. We show that glucocorticoid-dependent activation is mediated through activation of SGK1 and this activation requires the presence of NHERF2.
Expression of SGK1-To express SGK1 in PS120/NHE3V/NHERF2 cells, SGK1 was cloned into the mammalian glutathione S-transferase (GST) fusion protein vector, pBC (12). The "kinase-dead" form of SGK1/ K127Q was constructed by mutating Lys-127 to Gln, and the NH 2 terminus was tagged with a hemagglutinin epitope.
Northern Blot Analysis-Twenty g of total RNA isolated from PS120 fibroblasts, OK cells, or Caco-2 cells were hybridized at 42°C in 5ϫ SSPE, 10ϫ Denhardt's, 2% SDS, 50% formamide, and 100 g/ml salmon sperm DNA. DNA probes used for NHE3 in PS120 fibroblasts and OK cells correspond to rabbit NHE3 aa 475-832 and OK NHE3 aa 490 -777, respectively. For detection of SGK1 mRNA, entire human SGK1 was used. Blots were washed at room temperature in 2ϫSSC, 0.5% SDS, followed by high stringency washes at 50°C in 0.1ϫSSC, 0.1% SDS.
Na ϩ -dependent Intracellular pH Recovery-To study intracellular pH (pH i ) using the ratio-fluorometric, pH-sensitive dye 2Ј,7Ј-bis-(2carboxyethyl)-5(6)-carboxyfluorescein acetoxymethyl ester (BCECF-AM), cells were seeded on coverslips, grown to confluence and then serum starved for 1-2 days. Cells were washed in 130 mM NaCl/pH i buffer (20 mM HEPES, 5 mM KCl, 1 mM tetramethylammonium-PO 4 , 2 mM CaCl 2 , 1 mM MgSO 4 ) and then dye-loaded by incubation for 20 min with 6.5 M BCECF-AM in the same solution as described previously (15). Cells were acidified by ammonium prepulse in 40 mM NH 4 Cl, 90 mM NaCl/pH i buffer and subsequently perfused with 130 mM tetramethylammonium-Cl/pH i buffer. 130 mM NaCl was then reintroduced, and the sodium-dependent pH i recovery was recorded as described previously (15). At the end of each experiment the fluorescence ratio was calibrated to pH i using the high potassium/nigericin method using 130 mM KCl/pH i buffer plus 10 M nigericin titrated to pH 6.0, 6.3, and 7.2. For kinetic analysis of sodium-dependent pH i recovery, H ϩ efflux rate or Na ϩ /H ϩ exchange rate was calculated as (the rate of intracellular pH ϫ buffering capacity) as previously described (15), and data analysis was performed by a nonlinear regression using Origin (Microcal Software). In some cases, Na ϩ /H ϩ exchange rate was described by the rate of intracellular pH recovery rate, which was calculated by determining slopes along the pH i recovery by linear least square analysis over a minimum of 9 s. Statistical significance was assessed using independent Student t test for paired samples. Results were considered statistically significant when p Ͻ 0.05. Otherwise specified, results are presented as the mean Ϯ standard error of the mean (S.E.).
In Vitro Interaction-Various domains of NHERF2 generated by PCR were cloned into pGEX-4T (Amersham Biosciences, Inc.) and expressed as GST fusion proteins in Escherichia coli. These domains include the first PDZ domain (P1: aa 9 -128), the second PDZ domain (P2: aa 92-270), and the COOH-terminal domain (C: aa 232-337). Fidelity of the PCR products was confirmed by nucleotide sequencing.
For in vitro binding assays, SGK isoform were labeled with [ 35 S]Met by using the TNT in vitro transcription-translation system (Promega). Four g of GST fusion proteins immobilized on glutathione-agarose beads were incubated with 5 l of 35 S-labeled SGK isoforms for 4 h at 4°C in 10 mM Tris, pH 7.2, 150 mM NaCl, 0.2% Tween 20. At the end of the incubation period, the complexes were washed three times with the same buffer and bound proteins were visualized by autoradiography.

RESULTS
NHERF2 Is Necessary for the Activation of NHE3 by Dexamethasone-Glucocorticoids have been demonstrated to exert a direct effect on NHE3 activity (9,10,24). The current perception of this stimulation is that glucocorticoids increase gene expression of NHE3. To retest this hypothesis, we studied the effect of synthetic glucocorticoid dexamethasone in the human colonic carcinoma cell line Caco-2, which express both the apical NHE3 and the basolateral NHE1. Confluent monolayers of Caco-2 cells were deprived of serum for at least 6 h, and, at the end of the serum starvation, 1 M dexamethasone was added. After 4 or 24 h of incubation with dexamethasone, NHE3 activity was measured fluorometrically using the pH-sensitive dye, BCECF-AM, in the presence of 1 M amount of an inhibitor of Na ϩ /H ϩ exchanger 5-N-(methylpropyl)amiloride to inhibit NHE1 activity. K i values for NHE1 and NHE3 are 0.08 and 10 M, respectively (25). Fig. 1A shows that 24 h of dexamethasone treatment increased NHE3 activity in Caco-2 cells by 110%. The Na ϩ recovery of intracellular pH following an ammonium prepulse was 0.051 Ϯ 0.010 pH unit/min for control versus 0.107 Ϯ 0.024 pH unit/min for cells treated with dexamethasone for 24 h (⌬pH i 6.5 are shown, p Ͻ 0.05). Surprisingly, even a 4-h incubation with dexamethasone led to significant increase in NHE3 activity (0.075 Ϯ 0.014 pH unit/min, p Ͻ 0.05).
To determine whether there is concomitant induction of NHE3 transcripts, NHE3 mRNA levels were determined in these Caco-2 cells. Because Northern blot analysis failed to detect NHE3 transcripts resulting from low mRNA levels in Caco-2 cells, semiquantitative RT-PCR was performed on total RNA isolated from Caco-2 cells. Fig. 1B shows that NHE3 level was not significantly changed at 4 h but increased at 6 h by 45%, which further increased to 80% at 24 h. This suggested that the increase in NHE3 activity at 4 h of incubation with dexamethasone might be independent of transcriptional activation of NHE3.
To further confirm the dissociation of gene activation of NHE3 and the activation of transport activity, we studied the effect of dexamethasone in PS120/NHE3V cells. Because NHE3 in these cells is under the control of a heterologous promoter, its expression should be independent of dexamethasone. PS120/NHE3V fibroblasts were serum-starved for 8 h. Cells were then treated with 1 M dexamethasone for 18 -24 h prior to measuring Na ϩ /H ϩ activity. The same amount of ethanol was added to the control cells. As expected, the apparent maximal Na ϩ /H ϩ exchange rate (V max ) was not altered by incubation with dexamethasone (1410 Ϯ 52 M/s for control versus 1363 Ϯ 103 M/s with dexamethasone) (Fig. 2B). Northern analysis confirmed that NHE3 mRNA level was not altered by dexamethasone in PS120/NHE3V cells ( Fig. 2A). Human phosphoprotein 36B4 (26) and rRNA are shown as loading controls. We did not use the commonly used D-glyceraldehyde-3-phosphate dehydrogenase as a loading control based on previous reports that expression of D-glyceraldehyde-3-phosphate dehydrogenase could be affected by dexamethasone (24,27).
To confirm whether the dexamethasone stimulation of NHE3 is truly dependent on the presence of NHERF2 and not an artifact of the non-epithelial environment of PS120 fibroblasts, we repeated the study in OK cells, which are a well proven renal proximal tubule cell line for characterization of Na ϩ /H ϩ exchange and endogenously express apical Na ϩ /H ϩ exchanger NHE3 (15). Treatment of OK cells with dexamethasone resulted in almost 2-fold increase in NHE3 mRNA level regardless of the presence of NHERF2 (Fig. 3A). In contrast, NHE3 activity in OK cells was not affected by the dexamethasone treatment despite the 2-fold increase in NHE3 mRNA level (Fig. 3B). The maximum rate of pH i recovery was 0.015 Ϯ 0.001 pH unit/s for OK versus 0.017 Ϯ 0.001 pH unit/s with dexamethasone. Dexamethasone was similarly ineffective in cells overexpressing NHERF1 (Fig. 3C). The pH recovery rate was 0.015 Ϯ 0.007 pH unit/s for control OK/NHERF versus 0.020 Ϯ 0.002 pH unit/s for dexamethasone-treated cells. In contrast, OK/NHERF2 cells (15) responded within 24 h to 1 M dexamethasone resulting in nearly 2-fold increase (Fig. 3D). The pH recovery rate was 0.021 Ϯ 0.004 pH unit/s for control OK/ NHERF2 versus 0.040 Ϯ 0.004 pH unit/s for cells treated with dexamethasone. The magnitude of stimulation was similar whether the cells were treated with 1 M or 100 nM dexamethasone (Fig. 3E). These studies confirm the dependence on the presence of NHERF2 for enhanced NHE3 transport by dexamethasone.
NHERF2 Is Expressed in Caco-2 Cells-To determine whether the stimulation of NHE3 activity by dexamethasone in Caco-2 cells (Fig. 1A) is consistent with the presence of NHERF2, Western blot was performed. Fig. 4 shows that NHERF2 with an apparent molecular mass of 48 kDa is present in PS120/NHERF2, HT29, T84, Caco-2 cells, and rabbit intestinal brush border membrane, but absent in OK cells and PS120 fibroblasts. In contrast, NHERF1 with a molecular mass of 50 kDa was expressed in OK cells and PS120 fibroblasts. We initially reported that NHERF1 was absent in PS120 fibroblasts based on Northern blot analysis (12). The presence of NHERF1 in PS120 fibroblasts is in contrast to our initial report on the absence of NHERF1 based on Northern blot analysis. The failure to detect NHERF1 mRNA by Northern analysis is probably due to the lack of sensitivity to detect a small quantity of messages. Other have reported the presence of NHERF1 in PS120 fibroblasts by RT-PCR (28).
SGK1 Interacts with NHERF2 through the Carboxyl Terminus-Dependence on the presence of NHERF2 for dexamethasone stimulation of NHE3 suggested the presence of a protein that might interact with NHERF2 to facilitate activation of NHE3. To date, there are more than 10 proteins interacting with NHERF/NHERF2 via the COOH-terminal motif interacting with the PDZ domains. Comparison of the protein sequences of these proteins yielded a consensus COOH-terminal sequence of -D-(S/T)-(R/A/F/L)-(L/V) (29,30). We, therefore, searched GenBank for proteins with the COOH-terminal sequences of -D-(S/T)-(R/A/FL)-(L/V), which may interact with the PDZ domains of NHERF proteins. One of the candidate proteins was SGK1 with the COOH-terminal sequence of DSFL. In addition, two additional SGK isoforms, termed SGK2 and SGK3, have recently been cloned (19). These SGK isoforms, which share ϳ80% identity in the catalytic domain, differ in the COOH-terminal sequences: -DSFL for SGK1, -ILDC for SGK2, and -DLFL for SGK3.
We amplified COOH-terminal domains of all three human SGK isoforms from Caco-2 cells by RT-PCR. These correspond to aa 277-431, 203-367, and 200 -429 for SGK1, SGK2, and SGK3, respectively. To test interaction between the SGK isoforms and NHERF2, SGK1, SGK2, and SGK3 were labeled with [ 35  interaction against NHERF2 was tested under in vitro conditions. Fig. 5A shows that SGK1 bound GST-NHERF2, whereas SGK2 showed no interaction. The interaction between SGK3 and GST-NHERF2 was significantly weaker than SGK1. Consistent with the lack of dexamethasone-induced activation of NHE3 in cell lines co-expressing NHERF1, NHERF1 did not interact with SGK isoforms. To demonstrate that the interaction between the SGK isoforms and NHERF2 occurred via the COOH-terminal sequences, we generated SGK1-3 lacking the COOH-terminal 4 aa (SGK1⌬4, SGK2⌬4, SGK3⌬4). Fig. 5B shows that the deletion of the COOH-terminal 4 aa in SGK1 and SGK3 completely abolished their interaction with NHERF2. Although SGK isoforms show a high homology, only SGK1 is activated by dexamethasone and serum (19). Based on its regulation by glucocorticoids and its expression in intestinal villi and renal proximal tubules (19,31), only SGK1 was pursued for further studies, and SGK3 was excluded from further analyses despite its interaction with NHERF2.
To determine which parts of NHERF2 interact with SGK1, we tested the interaction between SGK1 and various portions of NHERF2, which were bacterially expressed as GST fusion proteins. Fig. 5C shows that SGK1 interacted with both PDZ domains but the interaction with the second PDZ (P2) was severalfold stronger than that with the first PDZ (P1).
We have previously shown that NHE3 interacts with NHERF2 and NHERF1 via the second PDZ domain. The current studies showed that SGK1 also interacts with the second PDZ domain. Although it is possible that a single PDZ domain might interact with two proteins via different modes of interaction, it is plausible that NHERF2 forms multimers to facilitate interaction with more than one protein. We hence investigated whether NHERF2 forms homodimers by in vitro assay. Fig. 5D shows that 35 Slabeled NHERF2 was able to interact with GST-NHERF2 and GST-P1 but not with GST-P2 or GST-C, indicating that dimerization occurs via the first PDZ domain of NHERF2. In contrast to our results, a recent report showed that NHERF2 forms homodimers through interaction involving both PDZ domains (32). The reason for the discrepancy is not known.
The presence of SGK1 in PS120 and OK cells was not previ-ously documented. We hence determined the presence of SGK1 and its inducibility by dexamethasone in these cells. Fig. 6 (A-C) shows that SGK1 is present in PS120, OK, and Caco-2 cells, and that SGK1 is transcriptionally induced by dexamethasone. There was a 50 -60% increase in SGK1 expression after 24 h with dexamethasone in PS120 fibroblasts, whereas SGK1 increased by 2-fold in the same time frame in OK cells.
In Caco-2 cells, SGK1 was induced within as early as 4 h by 30%, and was further increased at 24 h. The time course of SGK1 induction in Caco-2 cells is consistent with NHE3 activation shown in Fig. 1A. SGK1 Directly Stimulates NHE3 Activity-To determine whether SGK1 can directly activate NHE3, SGK1 fused with glutathione S-transferase (GST-SGK1) was stably expressed in PS120/NHE3V/NHERF2 cells. The expression of GST and GST-SGK1 in these cells was confirmed by Western immunoblot using anti-GST antibodies (Fig. 7A). To confirm the interaction between NHERF2 and SGK1, cell lysates from PS120/ NHE3V/NHERF2 cells co-expressing either GST or GST-SGK1 were used for in vivo interaction study. As shown in Fig. 7B, NHERF2 was co-purified with GST-SGK1 but not with the GST control, demonstrating that NHERF2 and SGK1 specifically interact in vivo.
SGK1 is known to be activated by receptor tyrosine kinases including insulin receptor (19,20). We next determined NHE3 activity following activation of GST-SGK1 by insulin. Fig. 7B shows that NHE3 activity in PS120/NHE3/NHERF2/GST-SGK1 is increased by 43% upon pretreatment with insulin (1733 Ϯ 180 M/s for control versus 2487 Ϯ 93 M/s with insulin). Over four sets of independent experiments, the stimulation of NHE3 activity ranges between 23 and 83% (data not shown). In contrast, insulin had no significant effect on NHE3 in the presence of GST (1885 Ϯ 140 M/s versus 2112 Ϯ 123 M/s with insulin) (Fig. 7C). Basal activities of NHE3 were not significantly different, suggesting that the basal activity of GST-SGK1 is minimal. These studies demonstrate that SGK1 is able to directly stimulate NHE3 activity.
We have thus far shown that activation of NHE3 activity by dexamethasones can occur independent of NHE3 gene induc- FIG. 2. Effects of dexamethasone on NHE3 in PS120 fibroblasts. A, total RNA was isolated from PS120/NHE3V or PS120/NHE3V/NHERF2 under control condition or treated with 1 M dexamethasone for 24 h. Northern analysis was performed using NHE3 cDNA cloned from OK cells as a probe under the conditions described under "Experimental Procedures." Quantification was made against both phosphoprotein 36B4 and rRNA. PS120 fibroblasts were treated with 1 M dexamethasone for 24 h prior to determining sodium-dependent pH recovery. B, PS120/NHE3V; C, PS120/NHE3V/ NHERF; D, PS120/NHE3/NHERF2. n ϭ 4 or more. q, control; OE, dexamethasonetreated.
tion. The necessity of SGK1 is also suggested based on its interaction with NHERF2 and temporal correlation in induction of SGK1 and stimulation of NHE3 activity. However, the results do not conclusively demonstrate the involvement of SGK1 in the regulation of NHE3 by dexamethasone. To reconfirm the role of SGK1 in dexamethasone-induced activation of NHE3, we expressed a "kinase-dead" SGK1/K127Q in OK/ NHERF2 cells. Treatment of the cells with dexamethasone for 24 h markedly decreased the stimulative effect of dexamethasone in the presence of hemagglutinin-SGK1/K127Q compared with the mock-transfected control (Fig. 8A). These results demonstrate that the dexamethasone stimulation of NHE3 is mediated through the activation of SGK1.
Dexamethasone-induced Activation of NHE3 Is Dependent on PI 3-Kinase-Activation of SGK1 requires phosphorylation at Thr 256 and Ser 422 by 3-phosphoinositide-dependent protein kinase (PDK) I and II, respectively, which are in turn activated by PI 3-kinase (19,20). Therefore, we sought to determine whether PI 3-kinase-dependent signaling is required for NHE3 activation by dexamethasone. OK/NHERF2 cells were treated with 1 M dexamethasone for 18 h in the presence or absence of a PI 3-kinase inhibitor, LY294002. Dexamethasone alone stimulated NHE3 activity by 2-fold compared with control as shown earlier (Fig. 8B). Incubation of the cells with LY294002 alone had a significant inhibitory effect on NHE3 activity (41%). This is in agreement with an earlier report that PI 3-kinase inhibitors reduced NHE3 activity by decreasing the amount of surface NHE3 through redistribution of NHE3 into endosomal compartments (33). This redistribution of surface NHE3 into endosomal compartments was seen as early as 10 min after an addition of wortmannin (33). In the presence of LY294002, dexamethasone exhibited insignificant effect on NHE3 activity. This shows that dexamethasone-induced stimulation of NHE3 is dependent on PI 3-kinase, and inhibition of PI 3-kinase blocked dexamethasone-dependent activation of NHE3.

DISCUSSION
Glucocorticoids chronically stimulate sodium absorption in ileum, proximal colon and renal proximal tubule. Treatment of animals with synthetic glucocorticoids led to a 2-fold increase in rabbit ileal electroneutral NaCl absorption (9,10,24). Induction of NHE3 transcripts following methylprednisolone administration in small animals suggested that NHE3 is the likely candidate responsible for the increased sodium absorption (9). These observations prompted the hypothesis that glucocorticoids increased ileal Na ϩ /H ϩ exchange by increasing the NHE3 amount. Cho et al. (24) also found that dexamethasone had a stimulatory effect on NHE3 in rat ileum, and proximal colon but not in jejunum, and distal colon. Unlike adult rats, sucking rats showed responsiveness to glucocorticoids in jejunum but not in ileum, and this difference in the glucocorticoid induction was attributed to differential expression of glucocorticoid receptors in adult and suckling rats (34). Wormmeester et al. (35) showed that methylprednisolone treatment in rabbit ileum led to an increased NHE3 activity by more than 4-fold by increasing both total amount of NHE3 in ileal brush border membrane and NHE3 turnover rate. Whether the increase in NHE3 activity is solely due to an increase in NHE3 in the brush border membrane or other mechanism such as phosphorylation of NHE3 protein is unclear. A similar mechanism appears to be responsible for renal proximal tubular sodium absorption, where administration of dexamethasone in rabbit for 1-2 days resulted in up to 2.5-fold increase in NHE3 transcript level, without an effect on NHE1 (10). As in ileum, glucocorticoids increased sodium absorption in renal brush border vesicles by increasing V max without altering the affinity for H ϩ or the Hill coefficient (36). Interestingly, incubation of isolated proximal tubules in dexamethasone for 3 h resulted in a 33% increase in NHE3 activity without a significant change in NHE3 or NHE1 transcript levels. This temporal separation of NHE3 mRNA abundance and Na ϩ /H ϩ transport was also FIG. 3. Effects of dexamethasone on NHE3 in OK cells. A, OK and OK/NHERF2 cells were pretreated with 1 M dexamethasone for 24 h prior to RNA preparation. Total RNA was prepared, and Northern blot analysis was performed to determine NHE3 transcript level. Quantification was made against both phosphoprotein 36B4 and rRNA. B, OK; C, OK/NHERF; D, OK/NHERF2. Effect of dexamethasone on NHE3 activity was determined fluorometrically. n ϭ 4 or more. q, control; OE, dexamethasone-treated. E, OK/NHERF2 cells were treated with 1 M or 100 nM dexamethasone. The rates of pH recovery at pH i 6.4 are shown. For determination of pH i recovery rate (⌬pH i /min), slopes were calculated along the pH i recovery by linear least square analysis over a minimum of 9 s. n ϭ 3. *, p Ͻ 0.05 versus control.
FIG. 4. Protein expression of NHERF2 and NHERF1. Ten g of lysates were loaded, and the presence of NHERF2 and NHERF1 were detected using Ab2570 and Ab5199 raised against the COOH termini of NHERF2 and NHERF1, respectively. The migration of NHERF1 in PS120/NHERF1 slight retarded because of the tagging of NHERF1 with His 6 . observed at 4 h after intravenous administration of dexamethasone in rabbit (10). In addition, a number of findings in the current studies show that NHE3 gene induction cannot fully explain activation of NHE3. 1) NHE3 was activated by dexamethasone in the absence of NHE3 mRNA induction in PS120 fibroblasts. 2) Dexamethasone failed to activate NHE3 in OK cells even though the NHE3 mRNA level was doubled. The absence of dexamethasone-induced activation of NHE3 in OK cells is in contrast to a previous report that 1 M dexamethasone resulted in 40% stimulation of NHE3 activity in OKP cells (37). The reasons for the discrepancy are not clear, but the studies were done using OKP cells, which are a subset of OK cells, and the difference in effect of dexamethasone may be the result of cell type difference. 3) In all cells used in the current studies, NHERF2 was necessary for dexamethasone-dependent activation of NHE3, and the NHE3 gene induction alone cannot explain the necessity of NHERF2 in NHE3 activation. Although the present studies suggest that NHE3 activation by glucocorticoids occur in the absence of NHE3 gene induction, our data do not completely exclude the NHE3 gene induction as a part of the activation. In OK/NHERF2 and Caco-2 cells, where the NHE3 transcript levels were increased 2-fold by dexamethasone, NHE3 activity was up-regulated by more than FIG. 5. In vitro interaction between SGK1 and NHERF2. Full-length and subdomains of NHERF2 expressed as GST fusion proteins were purified from E. coli. 4 g of GST fusion proteins immobilized on glutathione-agarose beads were incubated with 5 l of [ 35  FIG. 6. Induction of SGK1 by dexamethasone. A, total RNA was isolated from PS120/NHE3V or PS120/NHE3V/NHERF2 under control condition or treated with 1 M dexamethasone for 24 h. Northern analysis was performed using SGK1. Incubation with dexamethasone significantly increased SGK1 transcript levels in PS120 (A), OK (B), and Caco-2 (C) cells. In Caco-2 cells, SGK1 was induced at 4 h, consistent with stimulation of NHE3 activity. Quantification was made against both human phosphoprotein 36B4 and rRNA.
FIG. 7. SGK1 directly stimulates NHE3 activity. A, SGK1 was expressed in PS120/NHE3/NHERF2 cells as GST fusion protein. 20 g of lysate was resolved by 12% SDS-PAGE, and expression of GST and GST-SGK1 was detected by Western immunoblot (IB) using anti-GST antibody. B, SGK1 interacts with NHERF2 in vivo. Upper panel, lysates from PS120/NHE3V/NHERF2 expressing either GST-SGK1 or GST as control was prepared as described under "Experimental Procedures." GST fusion proteins were affinity-purified against glutathione-agarose, and co-purifed NHERF2 was detected by immunoblot using Ab2570. Bottom panel, 10 g of lysates were immunoblotted with Ab2570. C, PS120/NHE3V/NHERF2/GST-SGK1 cells were treated with 10 g/ml insulin for 20 min prior to determining NHE3 activity. SGK1 significantly activates NHE3 activity in the presence of GST-SGK1. D, in contrast, insulin has no effect on NHE3 in the presence of GST. n ϭ 4 or more. q, control; OE, insulin-treated.
100% compared with 40 -70% in PS120 cells, where NHE3 is ectopically expressed. Although quantitative comparison of NHE3 activation in different cell lines is not possible, it is plausible that the difference in the magnitude of the NHE3 activation between Caco-2/OK cells and PS120 might arise from the NHE3 gene induction.
Recent studies have demonstrated that PDZ domains play an essential role in signal transduction by targeting of proteins and compartmentalization/assembly of signaling proteins (38). By organizing assembly of signaling proteins, PDZ-containing proteins play a fundamental role in enhancing the specificity and efficiency of signal transduction. NHERF1 and NHERF2 were initially described as NHE regulatory proteins, which appear to be necessary for cAMP-mediated inhibition of NHE3 (12,39). Recent works showed that NHERF1 and NHERF2 interact with a number of proteins, including the ␤ 2 -adrenergic receptor, the platelet-derived growth factor receptor, and the cystic fibrosis transmembrane conductance regulator (29,40,41). Many of the studies indicated that NHERF1 and NHERF2 seemingly have the same functions, although potential differences in functions of NHERF1 and NHERF2 was suggested primarily based on difference in cellular localization of these proteins (15,16,23,42,43). The current studies show that SGK1 binds only NHERF2 but not NHERF1, despite the high identity between these two proteins. Consistent with the differential interaction, only NHERF2 was able to reconstitute glucocorticoid-induced activation of NHE3. SGK1 shows 54% homology in its catalytic domain to the proto-oncogene Akt/PKB (17). The current consensus view of activation of Akt involves the recruitment of Akt to the plasma membrane through its NH 2 -terminal lipid-binding pleckstrin homology domain (19,20). Once properly positioned, Akt is phosphorylated by PDK, resulting in activation of Akt. Unlike Akt, SGK1 lacks a pleckstrin homology domain at its NH 2 terminus and the mechanism recruiting SGK1 to the plasma membrane remains undefined. The interaction between SGK1 and NHERF2 raises the possibility that NHERF2 may play a role in anchoring SGK1 to the plasma membrane.
Although our data demonstrate that SGK1 activates NHE3 activity, the mechanistic nature of this activation remains to be determined. The optimal consensus sequences for phosphorylation by SGK are RXRXX(S/T) and RRX(S/T), where S/T is the site of phosphorylation (20). Within NHE3, there are three RRXS motifs at aa 555, 607, and 693, and a single RXRXXS motif at aa 663, all of which are conserved in all NHE3 cloned from rabbit, rat, human and opossum. It is plausible that SGK1 phosphorylates NHE3 at one or more of these sites to either directly increase its transport activity or to increase the plasma membrane distribution of NHE3. Preliminary studies showed that NHE3 is phosphorylated by SGK1 under in vitro conditions. 2 SGK1 is recently shown to activate the epithelial sodium channel (ENaC) in response to aldosterone (44). Co-injection of ENaC and SGK1 mRNA into Xenopus laevis oocytes led to a significant increase in sodium conductance. Attempts to demonstrate direct phosphorylation of ENaC by SGK1 have failed (45). SGK1 has been shown to associate with the ␤-subunit of the channel (46). However, this subunit does not bear a consensus sequence for SGK1. Instead, the ␣-subunit contains the only intracellular SGK1 consensus sequence within the ENaC protein. Replacement of the serine at this consensus sequence with alanine did not prevent the activation of ENaC by SGK1, suggesting that SGK1 regulates ENaC not by direct phosphorylation (45). Taken together, these experiments suggest that SGK1 binds to the ␤-subunit but activates the channel by phosphorylation of a regulatory protein (45,47). A recent report showed that SGK1 might modulate ENaC activity via phosphorylation of Nedd4 (48).
In addition to glucocorticoids, SGK1 is transcriptionally activated by aldosterone. However, the transcriptional activation of SGK1 by aldosterone shows a different time course than by glucocorticoids. SGK1 mRNA is rapidly activated by aldosterone, peaking at 1 h and declining thereafter (49). Glucocorticoids also activate SGK1 mRNA as early as 30 min, but its expression is maintained at least for 24 h (50). It should be noted that neutral NaCl absorption in the colon is also stimulated by aldosterone (51)(52)(53). The aldosterone level has been increased chronically by several different methods including chronic sodium depletion and chronic potassium depletion (52,53). Chronic aldosterone elevation stimulates NaCl absorption in the rat proximal colon (52), but its effect in the distal colon remains unclear (52,53). Turnamian and Binder (51) using aldosterone or glucocorticoid receptor-specific steroid RU28362 showed that aldosterone inhibited electroneutral NaCl absorption but activated electrogenic sodium absorption in the rat distal colon, whereas dietary depletion showed no change in electroneutral NaCl absorption in the distal colon (53). In addition, Ikuma et al. (52) showed that, in distal colon, NHE3 activity and mRNA levels decreased by dietary sodium depletion.
The present data suggest that NHE3 induction by dexa-2 C. C. Yun and Y. Chen, unpublished data.

FIG. 8.
A, SGK1 is necessary for the glucocorticoid-dependent activation of NHE3. Hemagglutinin-SGK1/K127Q is expressed in OK/ NHERF2 cells and the cells were treated with dexamethasone for 24 h. Dexamethasone activated NHE3 by more than 100% in the control cells transfected with a carrier. In contrast, the dexamethasone effect on NHE3 was largely blocked by the presence of SGK1/K127Q. B, dexamethasone-dependent activation of NHE3 is blocked by PI 3-kinase inhibitor LY294002 in OK/NHERF2 cells. PI 3-kinase activity was blocked with 50 g of LY294002 for 24 h. To study the effect of dexamethasone in the presence of LY294002, OK/NHERF2 cells were preincubated with 50 g of LY294002 for 30 min followed by incubation with 1 M dexamethasone for 24 h. The rates of pH recovery at pH i 6.4 are shown. pH i recovery rate were calculated along the pH i recovery by linear least square analysis over a minimum of 9 s. Results are presented as the mean Ϯ S.D. n ϭ 4. *, p Ͻ 0.05 compared with control. methasone is not the sole determinant in the glucocorticoidinduced activation of NHE3. Fig. 9 depicts a putative model of glucocorticoids induced activation of NHE3 based on the current findings. Glucocorticoids transcriptionally activate NHE3 and increase the amount of NHE3. At the same time, glucocorticoids rapidly activate SGK1. The novel finding in the current work is that SGK1 interacts with NHERF2 via the PDZ domains to activate NHE3, and this is an essential step in stimulation of NHE3 activity by glucocorticoids. SGK1 may then facilitate translocation of NHE3 to the plasma membrane by a mechanism yet to be defined. Because NHERF2 is associated with the plasma membrane through the interaction with membrane proteins including transport proteins and receptor tyrosine kinases, the association of SGK1 with NHERF2 also suggests that NHERF2 or a protein similar to NHERF2 may be needed to anchor SGK1 to the plasma membrane.