Na+/H+ exchanger 3 (NHE3) is expressed in the brush border (BB) of intestinal epithelial cells and accounts for the majority of neutral NaCl absorption. It has been shown that the Na+/H+ exchanger regulatory factor (NHERF) family members of multi-PDZ domain-containing scaffold proteins bind to the NHE3 COOH terminus and play necessary roles in NHE3 regulation in intestinal epithelial cells. Most studies of NHE3 regulation have been in cell models in which NHERF1 and/or NHERF2 were overexpressed. We have now developed an intestinal Na+ absorptive cell model in Caco-2/bbe cells by expressing hemagglutinin (HA)-tagged NHE3 with an adenoviral infection system. Roles of NHERF1 and NHERF2 in NHE3 regulation were determined, including inhibition by cAMP, cGMP, and Ca2+ and stimulation by EGF, with knockdown (KD) approaches with lentivirus (Lenti)-short hairpin RNA (shRNA) and/or adenovirus (Adeno)-small interfering RNA (siRNA). Stable infection of Caco-2/bbe cells by NHERF1 or NHERF2 Lenti-shRNA significantly and specifically reduced NHERF protein expression by >80%. NHERF1 KD reduced basal NHE3 activity, while NHERF2 KD stimulated NHE3 activity. siRNA-mediated (transient) and Lenti-shRNA-mediated (stable) gene silencing of NHERF2 (but not of NHERF1) abolished cGMP- and Ca2+-dependent inhibition of NHE3. KD of NHERF1 or NHERF2 alone had no effect on cAMP inhibition of NHE3, but KD of both simultaneously abolished the effect of cAMP. The stimulatory effect of EGF on NHE3 was eliminated in NHERF1-KD but occurred normally in NHERF2-KD cells. These findings show that both NHERF2 and NHERF1 are involved in setting NHE3 activity. NHERF2 is necessary for cGMP-dependent protein kinase (cGK) II- and Ca2+-dependent inhibition of NHE3. cAMP-dependent inhibition of NHE3 activity requires either NHERF1 or NHERF2. Stimulation of NHE3 activity by EGF is NHERF1 dependent.
- Na+/H+ exchanger 3
- Na+/H+ exchanger regulatory factor 1
- Na+/H+ exchanger regulatory factor 2
- intestinal sodium absorption
na+/h+ exchanger 3 (NHE3) is present in the brush-border (BB) membrane of small intestine, colon, and renal proximal tubule and plays a major role in neutral NaCl and NaHCO3 absorption (14, 53). Regulation of NHE3 occurs during normal digestion by neural and humoral substances, which affect its function and membrane trafficking (27). Short-term regulation of NHE3 activity is achieved through a variety of factors that affect NHE3 turnover number and/or surface expression and often involves a role for the cytoskeleton and accessory proteins, including the multi-PDZ domain-containing proteins Na+/H+ exchanger regulatory factor (NHERF)1 and NHERF2 (8, 10, 11, 29–32, 47, 49).
NHERF1 and NHERF2 are scaffold proteins that are made up of two related PDZ domains and a COOH-terminal ERM binding domain (8, 10, 45, 46). They are highly homologous and are both present in the apical domain of epithelial cells including renal proximal tubule, small intestine and colon, and airway. Consequently, their relative functional role in single polarized epithelial cells has been questioned. Some information concerning the specificity or lack thereof of their function has been determined. This information, based primarily on studies in murine proximal tubule and small intestine, includes the following: 1) They are not located in precisely the same apical subdomains, with NHERF1 being in the outer aspects of the microvillus while NHERF2 is at the inner aspects of the microvillus and in the intervillus clefts and subapical domain (41, 42, 56). 2) While they share some binding partners, in addition, some of their binding partners are unique for one or another of these NHERFs (6, 16, 17, 20–22). This specificity has been correlated with NHE3 regulation by single NHERF family members. 3) The regulation of the NHERF proteins as part of signaling seems different as well. For instance, NHERF1 is phosphorylated under basal conditions and undergoes increased phosphorylation by cAMP and PKC. Increased phosphorylation of NHERF1 Ser77 correlates with reduced binding to its ligands. NHERF2 has not been shown to be phosphorylated under any conditions (40, 44). 4) While NHERF1 and 2 regulate transporters that traffic from and to the apical membrane in polarized epithelial cells, as well as being necessary for recycling of several G protein-coupled receptors (GPCRs), NHERF1 and 2 themselves do not appear to traffic (5, 12). 5) NHERF family members hetero- or homomultimerize and thus may function together in overlapping locations to carry out their scaffolding and NHE/regulatory activities.
Studies have begun to define the roles of NHERF1 and 2 in regulation of transport proteins in mouse intestinal and renal proximal tubules and to a lesser extent in airway cells. Regulation studied to date includes acute changes in transport rates of Na+, Cl−, HCO3−, and phosphate (3, 4, 12, 35, 38, 45, 46). NHERF1 and 2 are required for regulation of several murine small intestinal and renal proximal tubule apical domain transport proteins including the Na+/H+ antiporter NHE3, the Cl− channel CFTR, and to some extent the phosphate transporter NaPi2a. Multiple aspects of specific roles of the NHERFs have been defined in regulation of these transporters, which include specificity, relating to cells, agonists, and the transporters involved. The relevant studies are incomplete, but several conclusions have already been established: 1) The NHERFs are involved in setting basal transport rates of NHE3, CFTR, and NaPi2a as well as their regulated activity. 2) Each of the NHERFs affects NHE3 regulation, but in intestine the role of each NHERF seems to differ based on which intestinal segment is involved (1–3, 31, 35–37). NHERF1 is required for cAMP inhibition of NHE3 in murine renal proximal tubule but not in distal ileum (3, 31, 35, 46, 48), murine duodenal CFTR-dependent HCO3− secretion but not ileal anion transport (4), and cAMP inhibition of NHE3 via exchange protein directly activated by cAMP (EPAC) in kidney, while EPAC does not act on NHE3 in murine small intestine (13, 32). The purpose of these studies was to further our mechanistic studies of how NHERF1 and NHERF2 contribute to acute regulation of NHE3 in the intestine. To do so, we developed a cell model to carry out more mechanistic studies than have proven possible with the NHERF knockout models in mouse intestine. Our strategy was to knock down NHERF1 and NHERF2 separately and together in Caco-2/bbe cells and to correlate the changes in basal and regulated NHE3 activity with what was found in our studies of acute NHE3 regulation in NHERF1 and NHERF2 knockout models in mouse jejunum, ileum, and colon. Moreover, we pursued mechanistic studies here and will use the information generated from the knockdown (KD) models to further examine mechanisms by which the NHERF proteins regulate NHE3 in mammalian intestine.
MATERIALS AND METHODS
Chemicals, reagents, and antibodies.
Reagents were from Sigma-Aldrich (St. Louis, MO) unless otherwise stated. HOE-694 was a kind gift from Sonafi-Aventis. 8-(4-chloro-phenylthio)- 2′-O-methyladenosine-3′,5′-cAMP (8-pCPT-2′-O-Me-cAMP) and 8-p-chlorophenylthio-cGMP (8-pCPT-cGMP) were from BioLog (Hayward, CA). EZ-Link Sulfo-NHS-SS-biotin and streptavidin-agarose were from ThermoPierce Chemical (Rockford, IL). Restriction endonucleases were from New England Biolabs (Ipswich, MA). 2′,7′-Bis(2-carboxyethyl)-5(6)-carboxyfluorescein acetoxymethyl ester (BCECF-AM) was from Invitrogen (Carlsbad, CA). Mouse monoclonal antibodies to the hemagglutinin (HA) epitope were from Covance Research Products (Princeton, NJ). Rabbit polyclonal anti-NHERF1 and NHERF2 antibodies were a generous gift from Dr. Chris Yun (Department of Medicine, Emory University School of Medicine) (53) as well as being prepared by us. cGK II rabbit polyclonal antibodies were as described (37).
Anti-NHE3 polyclonal antibodies were raised against a synthetic peptide corresponding to amino acids 809–831 (NH2-DSFLQADGHEEQLQPAAPESTHM-COOH) of the COOH terminus of mouse NHE3 modified from a peptide used to make an antibody to rat NHE3 (16). GAPDH mouse monoclonal antibodies were from U. S. Biological (Swampscott, MA). Alexa 488 mouse monoclonal and Alexa 568 rabbit polyclonal secondary antibodies were from Invitrogen. DNA primers were from Operon Biotechnologies (Huntsville, AL).
The Caco-2/bbe cell line originally derived from a human adenocarcinoma was obtained from M. Mooseker (Yale University, New Haven, CT) and grown on membranes (Transwell or “filterslips”) in DMEM containing 25 mM NaHCO3 supplemented with 0.1 mM nonessential amino acids, 10% fetal bovine serum, 4 mM glutamine, 50 U/ml penicillin, and 50 μg/ml streptomycin, pH 7.4, in 5% CO2-air at 37°C.
Adenoviral HA-NHE3 preparation, purification, and expression.
Triple HA-tagged rabbit NHE3 was cloned into the adenoviral shuttle vector ADLOX.HTM under control of a cytomegalovirus (CMV) promoter. Virus was generated by transfection of CRE8 cells with ψ5 viral DNA and ADLOX.HTM/3HA-NHE3 using Lipofectamine 2000 (Invitrogen). The crude adenoviruses were then propagated by infection in HEK 9-11 cells. Adenovirus was separated by CsCl gradient centrifugation and purified with a Sephadex G-25 column. Viral particle numbers were calculated as (A260 value) × (1.1 × 1012) × dilution. To test the expression of adenovirus (Adeno)-HA-NHE3, Caco-2/bbe cells on Transwell filters were treated with 6 mM EGTA in serum-free Caco-2 medium for 2 h at 37°C before viral infection to allow maximum virus exposure to both the apical and basolateral surfaces. Viral particles were diluted in serum-free Caco-2 medium, and the cells were infected by incubating at 37°C for 6–7 h, after which the growth medium was replaced. For transport assays or Western blot analyses, cells were used ∼44 h after infection. Viral infection of Caco-2/bbe cells was typically at day 11–12 after reaching confluence, and study was at approximately day 14. The cGK II adenovirus was as described previously (39).
NHERF1 and NHERF2 knockdown by short hairpin RNA or small interfering RNA.
Lentivirus-based (Lenti) and adenovirus-based (Adeno) short hairpin RNA (shRNA) [RNA interference (RNAi)] were used to knock down NHERF1 and/or NHERF2. Initially we used Adeno-small interfering RNA (siRNA) to knock down NHERF2 transiently (used for studies of cGMP inhibition of NHE3). NHERF2 siRNA constructs were designed with the program provided by OligoEngine (OligoEngine, Seattle, WA). DNA sequences used were GCTGGCAAGAAGGATGTCA (construct 1) and GCAAGATCCCTTCCAGGAG (construct 2). The hairpin constructs (containing TTCAAGAGA as the loop sequence) were transferred into the RNAi/adenovirus transfer plasmid pPHA74243, linearized with BglI and HindIII, and used for adenovirus production in CRE8 cells as described for HA-tagged NHE3. As negative controls for the siRNA studies, a scrambled siRNA construct (catalog no. D01200-05, Dharmacon Research, Lafayette, CO) was used to make the adenovirus.
In most of our studies, we generated stable cell lines of Caco-2/bbe with NHERF1 and/or NHERF2 knocked down with Lenti-shRNA constructs. In brief, gene sequence-specific shRNA clones were constructed within the lentivirus plasmid vector pLKO.1-puromycin, which were obtained through the Johns Hopkins High Throughput Biology (HiT) Center from Open Biosystems (Huntsville, AL).
The constructs used to generate lentiviral transduction particles are listed in Table 1. Stable cell lines of Caco-2/bbe with expression of NHERF1 or NHERF2 knocked down were generated by infecting cells with the respective gene-specific Lenti-shRNA particles, and selection was achieved by inclusion of 10 μg/ml puromycin in the culture medium. KD of protein expression was verified by Western blot analysis using specific antibodies against NHERF1 and NHERF2. The transduction negative control was a Lenti-shRNA construct specific for green fluorescent protein (GFP), which is not expressed in mammalian cells endogenously. In all experiments, control and KD were matched to receive a similar number of both adenoviral and lentiviral particles to control for viral effects.
NHE3 activity measurement.
Monolayers of polarized Caco-2/bbe cells were grown on polycarbonate membranes (0.4-μm pore size) glued to plastic coverslips (filterslips) for 12 days and infected with Adeno-HA-NHE3. Approximately 40 h after infection, cells were serum starved for at least 4 h and Na+/H+ exchange activity (NHE3) was determined with the intracellular pH (pHi)-sensitive dye BCECF-AM (60-min loading) as described previously (15, 25, 33) in the presence of HOE-694 (50 μM) to inhibit the contributions of NHE1 and NHE2. Exposure to 8-bromoadenosine 3′,5′-cyclic monophosphate (8-BrcAMP; 100 μM) or forskolin (20 μM), 8-pCPT-cGMP (100 μM), or epidermal growth factor (EGF; 0.2 μg/ml) was during the final 30 min of dye loading/NH4Cl prepulse. In some experiments, 50 μM carbachol was added to the TMA perfusion after dye loading/NH4Cl prepulse with cells exposed to carbachol only during the 5-min tetramethyl ammonium (TMA) and Na+ perfusion. For transport assays, filterslips were mounted in a cuvette, placed in a fluorometer (Photon Technology International), and perfused at both monolayer surfaces with TMA medium. Na+/H+ exchange was started by replacing the “TMA medium” with “Na+ medium” at the apical monolayer surface while TMA perfusion continued basolaterally. Changes in pHi were monitored by recording the emission signal at 530 nm after excitation alternating between 440 and 490 nm. The fluorescence ratio was calibrated to pHi with the high potassium/nigericin method, as we have adapted (6, 25). Rates of Na+-dependent intracellular alkalinization (efflux of H+, in μM/s) were calculated for a given pHi (within the linear phase ∼1 min for the initial rate of intracellular alkalinization and reported as ΔpH/Δt).
Caco-2/bbe cells 2 days after infection were serum starved for 4 h and washed three times with cold PBS. Cells were scraped in PBS and collected in 1.5-ml Eppendorf tubes. Cells were mixed with lysis buffer (in mM: 20 HEPES, pH 7.4, 1 Na3VO4, 150 NaCl, 50 NaF) containing protease inhibitors and homogenized by sonication. After removal of insoluble cell debris by low-speed centrifugation (2,000 g × 10 min), the protein concentration was measured with the bicinchoninic acid (BCA) method. Proteins were separated by SDS-PAGE (10%), transferred onto nitrocellulose membranes, and immunostained with primary antibodies to HA (1:1,000), NHERF1 (1:5,000), NHERF2 (1:3,000), and cGK II (1:2,000). Fluorescently labeled IR-Dyes 800 and 680 conjugated with rabbit polyclonal or mouse monoclonal antibodies were used as secondary antibodies (1:10,000). Protein bands were visualized and quantitated with the Odyssey system and Lycor software for the IR-Dye secondary antibodies, as described previously (34).
Caco-2/bbe cells were seeded on Anopore filters (0.02 μm, Nunc) coated with type I collagen. On day 12 after confluence, cells were infected with Adeno-HA-NHE3 as described above. Two days after infection, cells were serum starved for 4 h, kept at 4°C for 30 min, and then fixed with 3% paraformaldehyde (PFA) in cold PBS for 1 h. Cells were neutralized in PBS with 20 mM glycine, pH 7.4 for 10 min and then permeabilized and blocked in PBS containing 1% BSA plus 0.075% saponin 45 min. α-HA Alexa Fluor 488-conjugated antibody (1:100) and rabbit polyclonal antibodies against NHERF1 (1:500) or NHERF2 (1:300) were incubated for 1 h at room temperature in blocking solution. Cells were then washed two times with 0.1% BSA-PBS containing 0.075% saponin and once with PBS for 10 min for each wash. Cells were incubated with anti-rabbit Alexa Fluor 568 (1:100 dilution) secondary antibody (Invitrogen) and Hoechst 33342 (Invitrogen) for 1 h at room temperature. Cells were washed three times with PBS, mounted with Gel Mount (Sigma, St. Louis, MO), and imaged with a Zeiss LSM410 confocal fluorescence microscope using a ×63 water immersion lens. Serial xy-sections were obtained at 0.5-μm steps, and z images were reconstructed with MetaMorph image analysis software (Molecular Devices).
Cell surface biotinylation.
Caco-2/bbe cells (control, GFP-KD, Lenti-NHERF1-KD, and Lenti-NHERF2-KD) were grown on 0.4-μm Transwell polycarbonate semipermeable supports, and cell monolayers 12 days after confluence were infected with Adeno-HA-NHE3 as above. Approximately 40 h later, the cells were serum starved for 4 h and then kept at 4°C for 30 min. The Caco-2/bbe cell surface biotinylation protocol was modified from that used for PS120 fibroblast surface biotinylation published previously (6, 17, 33). Cells were rinsed three times with ice-cold PBS and incubated for 10 min with borate buffer (in mM: 154 NaCl, 1.0 boric acid, 7.2 KCl, and 1.8 CaCl2, pH 8.0). For apical surface labeling of NHE3, cells were incubated with 1.0 mg/ml NHS-SS-biotin in borate buffer added at the apical side and only borate buffer alone at the basolateral surface for 40 min and repeated once. Cells were then treated with quenching buffer (in mM: 20 Tris, 120 NaCl, pH 7.4) to scavenge the unreacted biotin, three times for 5 min each. Cells were washed three times with ice-cold PBS and solubilized with 0.8 ml of N+ buffer (in mM: 60 HEPES, pH 7.4, 150 NaCl, 3 KCl, 5 Na3EDTA, and 3 EGTA, with 1% Triton X-100). Protein concentrations of cell lysates were measured with the BCA method. Approximately 0.5–1.0 mg of cell proteins was incubated with streptavidin-agarose beads for 3 h at 4°C, and biotinylated proteins were isolated after precipitations of beads. The supernatant was retained as the intracellular fraction. The surface protein-bound streptavidin-agarose beads were washed five times in N+ buffer (containing 0.1% Triton X-100) to remove nonspecifically bound proteins, and bound proteins were eluted with 70–100 μl of loading buffer (5 mM Tris·HCl, pH 6.8, 1% SDS, 10% glycerol, and 1% 2-mercaptoethanol). Two dilutions of total cell lysate and surface and intracellular proteins for each group were loaded and size-fractionated by SDS-PAGE (10% gels). Proteins transferred to nitrocellulose membrane were blotted with monoclonal anti-HA and anti-GAPDH antibody. Leaking of biotin to the basolateral surface was considered by failure to detect biotinylated Na+-K+-ATPase (α-subunit) (data not shown). Western blot analysis and the quantification of the surface fraction were performed with the Odyssey system and Odyssey software (Li-COR, Lincoln, NE) as described previously (33).
Results are expressed as means ± SE. Statistical evaluation was performed by analysis of variance (ANOVA) when more than three conditions were compared in a single experiment or by Student's t-test (paired and unpaired).
Expression of NHE3 in Caco-2/bbe cells.
Regulation of NHE3 has been studied mostly in PS120 fibroblasts in which recombinant NHERF1 and NHERF2 have been overexpressed and in the OK proximal tubule cell line (opossum kidney) in which NHERF2 is overexpressed (OK cells express endogenous NHERF1 and a very small amount of endogenous NHERF2) (50). To further study NHE3 regulation in polarized intestinal epithelial cells, we chose Caco-2/bbe cells (human colonic cancer cell line). Caco-2/bbe cells express all members of the NHERF family and have a low level of endogenous NHE3 (15, 50). To allow for accurate measurements of NHE3 activity and biochemical study of NHE3, HA-tagged NHE3 was transiently expressed with recombinant adenovirus (9, 33, 52). Caco-2/bbe cells are difficult to transfect (transfection rate is ∼5% with Lipofectamine 2000), and, more importantly, once transfected the cells lose NHE3 expression rapidly even under selection with hygromycin or G418. Concentration dependence of the efficiency of infection of purified Adeno-HA-NHE3 was tested in polarized and confluent Caco-2/bbe cells. Approximately 44 h after infection, cell lysates were prepared and analyzed for NHE3 expression by immunoblot assay using HA antibody. Figure 1A shows that HA-NHE3 increased with increase in the number of viral particles exposed. To determine the localization of expressed HA-NHE3 in Caco-2/bbe cells, Adeno-HA-NHE3-infected cells were immunostained under the permeabilized condition with anti-HA-antibody conjugated with Alexa 488. x-y, x-z, and y-z confocal sections of stained Caco-2/bbe cells revealed mostly apical localization of NHE3 (Fig. 1B). To quantitate the level of expression of HA-NHE3 in this model, we compared 1) endogenous NHE3 (hNHE3) expression in the same Caco-2/bbe cells used for viral infection, 2) Adeno-HA-NHE3 expression (2 × 1010 particles/ml) in Caco-2/bbe cells, and 3) endogenous NHE3 expression in mouse jejunum, all with the use of cell lysates with normalization to β-actin. As shown in Fig. 1C, with a polyclonal antibody raised against mouse NHE3, HA-NHE3 expression in adenovirus-infected Caco-2 cells was ∼10 fold higher than endogenous hNHE3 expression in the same cells and was very similar to the level that occurred endogenously in mouse jejunum. To further determine the percent efficiency of NHE3 expression after infection of Adeno-HA-NHE3, Caco-2 cells in which the nuclei were stained with Hoechst 33342 were immunostained with anti-HA antibody and then examined with a confocal microscope. As shown in Fig. 1D, in this model the number of cells expressing NHE3 correlated with the number of viral particles used, with as high as 75% positive cells found when the cells were infected with 2 × 1010 particles/ml. NHE3 exchange activity in Caco-2/bbe cells infected with Adeno-HA-NHE3 was measured by using the same number of adenoviral particles used to test NHE3 expression. Rates of NHE3 activity correlated with the number of viral particles used as well as with the level of NHE3 expression and the percentage of cells expressing NHE3 (Fig. 1E). Thus we have established an adenovirus-based HA-NHE3 expression model in Caco-2/bbe cells to study NHE3 regulation in intestinal epithelial cells.
In this Caco-2/bbe cell model, we compared the degree of colocalization of NHERF1 and NHERF2 with HA-NHE3. Adeno-HA-NHE3-infected Caco-2/bbe cell monolayers were immunostained with monoclonal HA and polyclonal anti-NHERF1 and 2 antibodies. Under basal conditions, most of the NHERF1 overlapped with NHE3 at the BB membrane (Fig. 2, A and B). NHERF2 was also localized to the apical pole of Caco-2/bbe cells, in both the BB and a subapical region immediately below the region where NHE3 is localized (Fig. 2, C and D).
Effect of NHERF1 or NHERF2 knockdown in Caco-2/bbe cells on basal NHE3 activity.
NHERF1 and NHERF2 were shown previously to be necessary for some aspects of acute NHE3 regulation in intestine and proximal tubules. To better understand the specific involvement of NHERF1 and NHERF2 in NHE3 regulation in intestinal epithelial cells, a protein KD approach was used. NHERF1 and NHERF2 were knocked down separately and together in polarized Caco-2/bbe cells. Initially, an adenoviral approach was taken for transient expression of human NHERF2 gene-specific siRNA to knock down NHERF2 protein expression (6). Later, a lentiviral shRNA specific to NHERF1 or NHERF2 gene sequence became available to knock down the NHERF proteins, and lentiviral shRNA specific for the GFP gene was used as a viral transduction control. Three different Lenti-shRNA constructs were tested for each protein knocked down (Table 1). Caco-2/bbe cells at earliest possible passages were infected with lentivirus prepared from each construct. Selection was made with puromycin (10 μg/ml) after 3 days of infection. After two or three passages, KD of proteins was verified by immunoblot analysis using antibodies specific to NHERF1 and NHERF2. As shown in Fig. 3A, >90% of NHERF1 protein was knocked down by construct 1-1 and >85% of NHERF2 protein was knocked down by construct 2-3 compared with control (Lenti-shRNA for GFP). KD of NHERF1 had no effect on the expression of NHERF2 and vice versa (data not shown). To evaluate the effect of NHERF1 or NHERF2 KD on basal NHE3 activity in Caco-2/bbe cells, NHE3 activity was measured in these cells after infection with Adeno-HA-NHE3. Caco-2/bbe cells with stable expression of GFP shRNA or empty vector were used as controls. NHE3 activity in Caco-2/bbe cells with NHERF1-KD cells showed 30–40% less transport activity compared with control (Fig. 3B). In contrast, NHERF2-KD cells had increased NHE3 activity (Fig. 3B). To investigate the cause for the changes in NHE3 activity in NHERF1-KD and NHERF2-KD Caco-2/bbe cells, we performed apical cell surface biotinylation and measured the total and surface NHE3 expression in control, NHERF1-KD, and NHERF2-KD Caco-2/bbe cells (Fig. 3, C–E). The level of total NHE3 expression in NHERF1-KD and NHERF2-KD cells was 30–40% lower compared with control cells (significantly reduced for NHERF1 but not NHERF2). On the other hand, NHE3 expressed on the apical surface normalized to cell β-actin was slightly but not significantly increased for NHERF1 KD and significantly increased for NHERF2 KD (Fig. 3E).
These results indicate that there is some reduced total NHE3 expression in NHERF1-KD cells but with normal surface NHE3 expression, while there is increased NHE3 surface expression in NHERF2-KD cells. Thus NHERF1 and NHERF2 KD affected NHE3 activity differently. This unexpected reduction in NHE3 activity in NHERF1-KD Caco-2 cells corresponds to a smaller turnover number for NHE3, while in NHERF2-KD Caco-2 cells there was an increase in NHE3 activity due to more surface NHE3.
In Caco-2/bbe cells, cAMP inhibition of NHE3 is dependent on either NHERF1 or NHERF2.
Many human diarrheal diseases are caused by elevated tissue cAMP content, the effect of which is to inhibit Na+ absorption and stimulate Cl− secretion, resulting in net fluid loss in the gut. Previously, we showed (51) in PS120 cells overexpressing NHERF1 or NHERF2 that cAMP inhibition of NHE3 could be reconstituted by either NHERF1 or NHERF2. To investigate whether NHERF1 and/or NHERF2 are required for cAMP-dependent NHE3 inhibition in polarized intestinal epithelial cells, we used the Caco-2/bbe cell model. The cAMP/forskolin effect on NHE3 activity was examined in Caco-2/bbe cells with NHERF1 or NHERF2 stably knocked down with Lenti-shRNA. Control cells (Lenti-shRNA GFP) demonstrated ∼30–40% inhibition of NHE3 activity with forskolin, and NHERF1-KD and NHERF2-KD cells also showed very similar inhibition of NHE3 activity (Fig. 4A). This result suggested that either NHERF1 or NHERF2 might be sufficient for the cAMP effect on NHE3 in Caco-2/bbe cells. To confirm this, we knocked down both NHERF1 and NHERF2 simultaneously in Caco-2/bbe cells. Western blot analysis confirmed the double KD of >80% of NHERF1 and NHERF2 (Fig. 4B). We measured NHE3 activity in the absence and presence of forskolin in these double-KD cells expressing Adeno-HA-NHE3. Knocking down both NHERF1 and NHERF2 did not alter basal NHE3 activity but totally eliminated the inhibitory effect of forskolin on NHE3 activity (Fig. 4C). This result indicates that cAMP inhibition of NHE3 is dependent on either NHERF1 or NHERF2 in Caco-2/bbe cells.
Furthermore, it suggests that NHERF3 plus NHERF4 are not able to mediate the cAMP inhibitory effect since they are present after NHERF1 or NHERF2 KD in these cells (data not shown).
Carbachol inhibition of NHE3 activity in Caco-2/bbe cells is NHERF2- but not NHERF1 dependent.
Elevation of free intracellular Ca2+ by ionophore (4-bromo-A23187) or induced by carbachol (M3 cholinergic receptor mediated) inhibits NHE3 activity (9, 17, 24, 26, 28, 52, 54). Previous studies have shown that NHE3 inhibition by elevated Ca2+ in PS120 cells requires NHERF2 but not NHERF1 (17). Recent studies additionally showed that NHERF3 could reconstitute Ca2+-dependent NHE3 inhibition by a different mechanism than NHERF2 (52). In the present study, we investigated the effects of carbachol on NHE3 activity in the stable NHERF1/NHERF2 KD Caco-2/bbe cell model. Treatment of Caco-2/bbe cells expressing Adeno-HA-NHE3 with Lenti-shRNA-GFP with carbachol for 5 min demonstrated ∼35–40% inhibition of NHE3 activity compared with untreated cells (Fig. 5). Thus in this model carbachol effectively inhibits NHE3 basal activity, and the effect is rapid in onset. Using similar conditions, we measured NHE3 activity in Caco-2/bbe cells in which NHERF1 or NHERF2 was knocked down. Lentivirus-NHERF1 KD cells also exhibited 35–40% inhibition of NHE3 activity with carbachol, while KD of NHERF2, in contrast, resulted in total absence of carbachol-induced NHE3 inhibition (Fig. 5). This result was similar to the previous studies done in PS120 fibroblasts and showed that in polarized intestinal epithelial cells NHE3 inhibition by elevated Ca2+ requires NHERF2 but cannot be reconstituted by NHERF1 plus NHERF3.
Knocking down NHERF2 but not NHERF1 abolishes cGMP-dependent inhibition of NHE3 in Caco-2/bbe cells.
In our present study, we intended to confirm and extend previous observations on the role of NHERF2 in cGK II regulation of NHE3 in a more physiologically relevant epithelial cell model, i.e., Caco-2/bbe cells, compared with our previously reported studies in fibroblasts and OK cells (6). Initial experiments showed that the addition of a cGMP analog had no effect on NHE3 activity in Caco-2/bbe cells expressing Adeno-HA-NHE3, presumably because cGK II expression had been lost under cell culture conditions (Fig. 6, A and B). However, when the cGMP-dependent regulation of NHE3 was studied after adenoviral infection with cGK II (Fig. 6B), the Caco-2/bbe cells demonstrated a significant (∼36%) cGMP-dependent inhibition of NHE3 activity (Fig. 6C).
To study the role of NHERF2 in cGMP-dependent NHE3 regulation, we initially used an adenovirus-based siRNA construct specific for NHERF2. Adenoviral siRNA was available prior to lentiviral shRNA. Rapid and selective KD of NHERF2 was achieved by infecting with adenovirus-containing pSUPER and hairpin siRNA directed against NHERF2. The siRNA constructs used dramatically reduced the level of NHERF2 expression (Fig. 7A). The combination of both siRNA construct 1 and construct 2 resulted in the strongest reduction of NHERF2 expression (>85%) (Fig. 7A). Reduced NHERF2 abundance in the Caco-2/bbe cells expressing cGK II resulted in complete loss of inhibition of NHE3 in response to 8-pCPT-cGMP compared with a scrambled siRNA (Fig. 7B). This indicates that NHERF2 plays an essential role in the cGMP/cGK II-mediated regulation of NHE3. We next investigated the effect of cGMP on NHE3 activity in Caco-2/bbe cells after stable KD of NHERF1 or NHERF2 and expressing cGK II (via adenovirus infection). Knocking down NHERF2 in Caco-2/bbe cells expressing cGK II resulted in complete loss of inhibition of NHE3 activity in response to 8-pCPT-cGMP, with the cGMP inhibition occurring in the presence of the control GFP-shRNA. On the other hand, NHERF1-KD cells expressing cGK II showed cGMP inhibition of NHE3 activity similar to that in control cells (Fig. 7C). These results support and strengthen the previous finding that showed that NHERF2 (and not NHERF1, 3, or 4) reconstitutes GMP inhibition of NHE3.
Epidermal growth factor stimulation of NHE3 activity is NHERF1- but not NHERF2 dependent in Caco-2/bbe cells.
Serum, EGF, and glucocorticoids (dexamethasone) have all been shown to acutely stimulate NHE3 activity (2, 9, 23, 26, 49, 50). Stimulation of NHE3 by dexamethasone has been shown to be NHERF2 dependent. Previously, we showed (26) that EGF stimulated NHE3 activity by increasing surface expression of NHE3 in Caco-2/bbe cells and rabbit ileal BB membranes. It is not known whether EGF regulation of NHE3 is dependent on NHERF proteins. Thus NHE3 activity was measured in Caco-2/bbe cells after KD of either NHERF1 or NHERF2 (49, 50), with Lenti-empty vector as a control. Control Caco-2/bbe cells exhibited ∼40% stimulation of NHE3 activity after EGF treatment (0.2 μg/ml for 30 min) compared with untreated cells (Fig. 8). KD of NHERF2 in Caco-2/bbe cells had no effect on stimulation of NHE3 by EGF, but NHERF1-KD cells failed to stimulate NHE3 activity in response to EGF. This internal comparison between NHERF1 and NHERF2 KD makes up another negative control since use of the empty vector is not an ideal negative control as shRNA GFP. This result indicates that NHERF1 (but not NHERF2, 3, or 4) plays a necessary role in EGF stimulation of NHE3.
The studies described here resulted in the following findings. 1) Basal NHE3 activity in Caco-2 cells was dependent on both NHERF1 and NHERF2, although by very different mechanisms. NHERF1 was necessary for total NHE3 expression but did not appear to affect trafficking under basal conditions. In contrast, NHERF2 appears to be involved with defining the basal pool of NHE3 in the apical domain, and when NHERF2 was knocked down there was increased BB NHE3 with correspondingly more NHE3 activity. This presumably was due to NHE3 moving from the storage pool to the microvilli. The nature of this storage pool of NHE3 is under study. 2) cAMP inhibition of NHE3 could be mediated by either NHERF1 or NHERF2, and dependence on either could only be demonstrated by simultaneously knocking down both. 3) cGMP and elevated Ca2+ inhibition of NHE3 were mediated only by NHERF2 and not NHERF1. 4) EGF stimulation of NHE3 from the basolateral surface required NHERF1 and not NHERF2.
The adenoviral infection model used produced NHE3 expressed to levels similar to that found endogenously in mouse jejunum and allowed quantitation of basal and regulated (inhibition and stimulation) NHE3 activity in the BB of polarized Caco-2 cells. This model used up to four simultaneous viral infections/transductions in a single experiment (2 Adeno- and 2 Lenti-shRNA: Adeno: NHE3, cGK II; Lenti: shRNA NHERF1 and/or NHERF2). In all studies, control and test cells were exposed to approximately the same total number of adenoviral and lentiviral particles to control for the effects of infection. Also, in transduction with Lenti-shRNA constructs, we used a GFP shRNA construct as negative control. GFP is not endogenously expressed in mammalian cells, but the GFP shRNA construct should activate the RNAi pathways without targeting any human genes. Thus we are studying a model of cell infected with multiple viruses, with the effects of viral infection controlled for. In the few comparisons we made of basal NHE3 activity in uninfected Caco-2 cells versus cells infected with the maximum number of adenoviruses and lentiviruses used as controls for the KD studies, no change in basal NHE3 activity occurred (data not shown), but our studies must be interpreted recognizing the viral infection aspect of these studies.
The decrease in basal NHE3 activity in NHERF1-KD Caco-2/bbe cells was due to a smaller amount of expressed NHE3. What happens to NHE3 activity in NHERF1-null mouse small intestine is somewhat different. NHE3 activity was reduced in jejunum and proximal colon but not proximal ileum of NHERF1-null mice (3). When studied in more detail in jejunum, this change in transport was associated with reduced NHE3 BB percent expression but normal total expression, consistent with altered NHE3 trafficking or retention. Thus in both Caco-2 cells with NHERF1 KD and mouse jejunum with NHERF1 knockout there is reduced BB amount/reduced NHE3 activity, but the mechanisms by which this occurs are different. The explanation for the reduced amount of NHE3 in Caco-2 cells with NHERF1 knocked down has not been evaluated.
The mechanism by which cAMP acutely inhibits NHE3 activity requires phosphorylation of NHE3 at both amino acids 552 and 605 and involves an initial change in turnover number followed by increased endocytosis (18, 21). In addition, there is an additional postphosphorylation, cAMP-dependent step in NHE3 inhibition, which can be separated temporally from the phosphorylation and which was demonstrated to be necessary in both polarized renal proximal OK cells and isolated BB from renal proximal tubules (18). In PS120 cells, either NHERF1 or NHERF2 was sufficient for cAMP to inhibit NHE3 (1, 53). In PS120 cells, there is a small amount of NHERF1 but no NHERF2 (1, 53). In our original studies, this small amount of NHERF1 was not sufficient to allow cAMP to inhibit NHE3 (51). However, in our subsequent studies, basal NHERF1 was sufficient to allow cAMP inhibition of NHE3. cAMP-induced inhibition of NHE3 in Caco-2 cells is similar to that in PS120 cells, with both NHERF1 and NHERF2 having to be knocked down at the same time to prevent cAMP inhibition of NHE3. It is not known whether the step after cAMP-induced phosphorylation of NHE3 that is required for cAMP-induced inhibition of NHE3 is necessary for the cAMP-induced change in turnover number and/or the change in trafficking of NHE3, nor is it known whether this process requires either NHERF1 or NHERF2 or both (18).
The cAMP dependence on NHERF1 and 2 is somewhat different in Caco-2 cells and mouse intestine and proximal tubule. cAMP inhibition of NHE3 depends on NHERF1 in mouse renal proximal tubule (based on study of NHERF1 homozygous null mice in which neither cAMP-induced changes in NHE3 phosphorylation nor transport activity occurs), with NHERF2 not being sufficient to allow cAMP inhibition of NHE3. In mouse distal ileum, cAMP inhibition of NHE3 is dependent on NHERF2 and not NHERF1 (32, 38, 48). In contrast, in mouse jejunum, ileum, and colon, cAMP inhibits NHE3 in NHERF1-null mice, with the role of NHERF2 knockout alone and in combination with NHERF1 knockout not yet evaluated (3). Why renal proximal tubule and distal ileum seem to be regulated differently with a specific requirement for NHERF1 to allow cAMP inhibition of NHE3 in the former and for NHERF2 to allow cAMP inhibition of NHE3 in the latter is not yet understood. Moreover, cAMP acts by both a PKA- and a EPAC-dependent mechanism in mouse renal proximal tubule but not in mouse jejunum and ileum (13, 32).
It is not known whether EPAC is involved in PKA inhibition of NHE3 in Caco-2 cells, although it is not involved in forskolin-induced Cl− secretion in these cells (K. Hoque and C. M. Tse, personal communication). The results shown here have shown that NHERF1 and NHERF2 alone do not have a necessary role in cAMP inhibition of NHE3 in Caco-2 cells, with one or the other being able to entirely reconstitute cAMP-dependent regulation of NHE3 in these cells.
Elevated Ca2+ and cGMP/cGK II inhibition of NHE3 both depend on NHERF2 but not NHERF1 in Caco-2 cells. This finding was similar to results in PS120 cells and for cGMP was similar to results in OK cells as well (6, 17). Similarly, in NHERF2 knockout mouse jejunum and ileum, cGMP failed to alter NHE3 activity (6a). In addition, in mice with NHERF1 knocked out, cGMP inhibited NHE3 similarly to wild type in jejunum and proximal ileum.
These results are compatible with NHERF2 allowing formation of a cGK II/NHE3/NHERF2 complex in which it acts as a G kinase anchoring protein (GKAP) while NHERF1 does not bind cGK II and does not allow cGMP to inhibit NHE3 (6).
Elevated Ca2+ models in Caco-2 (carbachol) showed that NHERF2 but not NHERF1 reconstituted elevated Ca2+ inhibition of NHE3 as occurs in PS120 cells transfected with NHERF1 versus NHERF2 (17). Elevated Ca2+ (ionophore, serotonin, apical UTP) similarly inhibited NHE3 in ileum of wild-type mice but not in NHERF2-null ileum (R. Murtazina, M. Donowitz, unpublished observations) and similarly, carbachol and A23187 inhibited NHE3 in the jejunum and colon of wild-type but not NHERF2 null mice (6a). NHERF3 alone, which is present in Caco-2 cells, in the absence of NHERF2, was not able to reconstitute Ca2+ inhibition of NHE3 in Caco-2 cells. However, elevated Ca2+ failed to inhibit NHE3 activity in NHERF3-null mouse colon (34, 35). Thus, while NHERF2 is necessary for elevated Ca2+ inhibition of NHE3, the specific mechanism for that effect remains incompletely understood and the interactions with NHERF3 in this regulation require further study as well.
Least comparable to defined studies in mouse intestine and kidney is NHERF1 dependence of EGF (basolateral membrane receptor) stimulation of NHE3. EGF increases NaCl absorption and NHE3 activity in intact ileum and Caco-2/bbe cells through a phosphatidylinositol 3-kinase (PI3-kinase)/Akt pathway (9, 25, 26). Mechanistically, neither the EGF stimulation of NHE3 activity nor the NHERF1 dependence of EGF stimulation in Caco-2 cells has been explored, although other models of linkage of NHERF1 to EGF receptor (EGFR) have been identified. For instance, in nonintestinal cells, NHERF1 binds at the COOH terminus of the EGFR and restricts receptor downregulation (22). KD of NHERF1 might affect the EGFR, but the mechanism by which NHERF1 is necessary for EGF stimulation of NHE3 activity is currently unknown.
In summary, a value of the studies reported here is that they define the role of NHERF1 and NHERF2 in NHE3 regulation in an absorptive cell model and should allow us to use this model, when correlated with studies in intact intestine using NHERF1 and NHERF2 knockout mice, for detailed mechanistic studies to further understand regulation of NHE3. In these studies, we will be able to consider both compensatory changes that occur in the knockout mice as well as issues of overexpression in cell culture models to better understand the role of the NHERF scaffolding molecules in regulation of epithelial cell Na+ absorption.
These studies were supported in part by National Institute of Diabetes and Digestive and Kidney Diseases Grants RO1-DK-26523, RO1-DK-61765, PO1-DK-072084, and R24-DK-64388 (Hopkins Basic Research Digestive Diseases Development Core Center) and T32-DK-2007632 and by the Hopkins Center for Epithelial Biology. R. Lin was supported by an American Gastroenterological Association Jon Isenberg Fellowship Award.
M. Donowitz is a part owner of Tranzmembrane, which owns the patent for human NHE3 gene.
Present address of R. Lin: Wuhan Union Hospital, Tongji Medical School, Huazhong University of Science and Technology, Wuhan, China 43002.
- Copyright © 2011 the American Physiological Society