IQGAPs are differentially expressed and regulated in polarized gastric epithelial cells

Catherine S. Chew, Curtis T. Okamoto, Xunsheng Chen, Hai Yan Qin


IQGAPs, GTPase-activating proteins with an IQ motif, are thought to regulate many actin cytoskeleton-based activities through interactions with Cdc42 and Rac. Recently, Cdc42 was implicated in regulation of gastric parietal cell HCl secretion, and IQGAP2 was immunolocalized with Cdc42 to F-actin-rich intracellular canalicular membranes of isolated gastric parietal cells in primary culture. Here we sought to define distribution and localization of IQGAP1 and IQGAP2 in major oxyntic (acid-secreting) gastric mucosal cell types and to determine whether secretory agonists modulate these proteins. Differential staining protocols were used to identify different cell populations (parietal, chief, surface/pit, and mucous neck cells) in semi-intact glands isolated from rabbit gastric mucosae and to characterize these same cells after dispersion and fractionation on isopycnic density gradients with simultaneous staining for F-actin, H+-K+-ATPase, and GSII lectin-binding sites. There was a pronounced increase in intracellular F-actin staining in dispersed chief cells, apparently from internalization of F-actin-rich apical membranes that normally abut the gland lumen. Therefore, other membrane-associated proteins might also be redistributed by disruption of cell-cell contacts. Western blot analyses were used to quantitate relative concentrations of IQGAPs in defined mucosal cell fractions, and gastric glands were used for in situ localizations. We detected uniform levels of IQGAP2 expression in oxyntic mucosal cells with predominant targeting to regions of cell-cell contact and nuclei of all cell types. IQGAP2 was not detected in parietal cell intracellular canaliculi. IQGAP1 expression was variable and targeted predominantly to the cortex of chief and mucous neck cells. Parietal cells expressed little or no IQGAP1 vs. other mucosal cell types. Phosphoprotein affinity chromatography, isoelectric focusing, and phosphorylation site analyses indicated that both IQGAP1 and IQGAP2 are phosphoproteins potentially regulated by [Ca2+]i/PKC and cAMP signaling pathways, respectively. Stimulation of glands with carbachol, which elevates [Ca2+]i and activates PKC, induced apparent translocation of IQGAP1, but not IQGAP2, to apical poles of chief (zymogen) and mucous neck cells. This response was mimicked by PMA but not by ionomycin or by elevation of [cAMP]i with forskolin. Our observations support a novel, PKC-dependent role for IQGAP1 in regulated exocytosis and suggest that IQGAP2 may play a more general role in regulating cell-cell interactions and possibly migration within the gastric mucosa.

  • parietal cell
  • chief cell
  • protein kinase C
  • carbachol
  • actin

the rho gtpases Cdc42 and Rac1 regulate actin cytoskeletal dynamics by interacting with a number of effectors, including the IQGAPs (5, 15, 43), which are so named because they possess IQ domains, which are tandem repeats of four IQ motifs (tandem isoleucine and glutamine residues), and Ras GTPase-activating protein (GAP)-related domains (43). There are two known IQGAP family members, IQGAP1 and IQGAP2, with a putative third member, IQGAP3 (4, 32). Human IQGAP1 and IQGAP2 share 62% identity. IQGAP1 binds and cross-links actin filaments in vitro (1, 12) and has been implicated in Ca2+/calmodulin signaling (16, 31), E-cadherin-dependent cell adhesion (18, 26, 27, 29), cell motility, and invasion (30, 40) (see Refs. 4 and 32 for recent reviews). IQGAP1 overexpression has also been detected in gastric and colorectal carcinomas and gastric cancer cell lines (35, 41).

Less is known about the cellular distribution profiles and functions of IQGAP2, which was originally thought to be liver specific (5) but has since been localized in other cell types, including HCl-secreting gastric parietal cells (46) and blood platelets (39). Like IQGAP1, IQGAP2 binds to Cdc42 in its GTP-bound state and inhibits intrinsic and RhoGAP-stimulated GTP hydrolysis [although the nucleotide dependence of the interaction between IQGAP2 and Cdc42 is not as well established as for IQGAP1 (5, 25, 33, 34)]. Recently, IQGAP2 was proposed to be a Cdc42-dependent regulator of parietal cell HCl secretion (46). In this same report, in parietal cells maintained in primary culture, IQGAP2 and Cdc42 were immunolocalized to apically directed intracellular canalicular membranes, which are the site of active HCl secretion (36), whereas IQGAP1 was localized to the basolateral membranes of parietal cells.

In the present study, we show that IQGAP1 and IQGAP2 are differentially expressed and localized within the major epithelial cell types in the oxyntic gastric mucosa. In polarized gastric fundic mucosal cells, IQGAP2 is uniformly expressed and is localized at cell-cell contacts and within nuclei rather than intracellular canaliculi of parietal cells. In contrast, IQGAP1 is expressed in nonparietal cells and is concentrated within cortical membrane regions. Both IQGAPs appear to be phosphoproteins with the capacity to be regulated by the PKC- and cAMP-dependent signaling pathways. When glands are stimulated with the cholinergic agonist carbachol, IQGAP1 becomes concentrated at apical cell membranes of chief and mucous neck cells. This effect is mimicked by PMA, a phorbol ester that activates PKC, but apparently not by the calcium ionophore ionomycin. Together, these results suggest a novel role for PKC-dependent agonists in modulating IQGAP1-related functions in exocytotic cells but do not support a role for either IQGAP isoform in the regulation of parietal cell HCl secretion at the level of the intracellular canalicular membrane.


Antibodies and fluorescently tagged reagents.

Monoclonal antibodies directed against IQGAP1 and IQGAP2 were purchased from BD Transduction Labs and Upstate Biochemicals (Lake Placid, NY), respectively. Monoclonal anti-Cdc42 antibodies were obtained from BD Transduction Labs (clone 44), Santa Cruz Biotechnology (Santa Cruz, CA) (clone B8), and Upstate. Sheep polyclonal anti-Cdc42 was from Cytoskeleton (Denver, CO). Monoclonal anti-drebrin antibody (clone M2F6) was from Medical and Biological Laboratories (Nagoya, Japan) or Stressgen (Victoria, BC, Canada). Anti-H+-K+-ATPase α-subunit (clone HK 12.18) was from Calbiochem (EMD Biosciences, La Jolla, CA). Secondary horseradish peroxidase-tagged donkey anti-rabbit Ig was from Amersham-Biosciences (Piscataway, NJ). Fluorescently tagged probes used for subcellular immunolocalizations included Cy5-tagged donkey anti-mouse Ig (Jackson ImmunoResearch Laboratories, West Grove, PA), Oregon green phalloidin, Alexa Fluor 568 phalloidin, and Alexa Fluor 488 GSII lectin (Molecular Probes, Eugene, OR). [14C]aminopyrine (AP) was purchased from Amersham.

Isolation of gastric glands and oxyntic mucosal cell populations.

Gastric glands were isolated from nembutal-anesthetized, male New Zealand White rabbits as previously described (10). Parietal cells (75–90% purity) were isolated by using sequential collagenase/pronase digestion followed by isopycnic density gradient separation with Optiprep [GIBCO/Invitrogen Technologies, Carlsbad, CA (6)]. Oxyntic mucosal cell fractions enriched for the other major cell types were obtained from the same density gradients by collecting fractions sedimenting at different densities (6). In some experiments, parietal cell fractions were placed in primary culture for up to 3 days as previously described (9, 37). Parietal cell acid-secretory responses in isolated gastric glands were assessed by measuring the accumulation of the weak base AP (10).

Western blot analyses.

Western blots were generated and analyzed with enhanced chemiluminescence detection as previously described (8). Cells or immunoprecipitates were electrophoresed on SDS-PAGE gels (7%) or on isoelectric focusing strips (pH 3–10, nonlinear; Amersham-Pharmacia) (8) then transferred to PVDF membranes (Amersham-Pharmacia). Antibodies used in these analyses included anti-IQGAP1 (1:250), anti-IQGAP2 (1:1,000), and anti-Cdc42 monoclonal antibodies (1:500, Upstate; 1:250, Santa Cruz Biotechnology; and 1:250, BD Transduction Labs); anti-Cdc42 sheep polyclonal antibodies (1:250, Cytoskeleton); anti-Cdc42 rabbit polyclonal antibodies (1:100–1:200, Santa Cruz Biotechnology); and anti-drebrin (1:1,000). Cross-reacting bands on membranes were quantified with a Syngene GeneGnome charge-coupled device-based system (8)

Subcellular immunolocalization of IQGAPs using a triple-labeling protocol to identify gastric mucosal cells.

Confocal microscopic analyses were performed with a Zeiss LSM 510 equipped with Meta 3.2 software. Cells were fixed with 4% paraformaldehyde immediately after isolation or after primary culture (8, 9). Glands were similarly fixed or preincubated for 30 min at 37°C, then exposed to one of the following agonists or the appropriate vehicle control: carbachol, 15 min; PMA, 15 min; ionomycin, 5 min; or forskolin, 15–30 min. DMSO (0.1%) was the vehicle control for all agents, with the exception of carbachol, which was dissolved in cellular incubation medium. Dual labeling was performed using monoclonal antibodies directed against IQGAP1 (1:25) or IQGAP2 (1:50) paired with donkey anti-mouse Cy5-tagged secondary antibody (1:100) and Oregon green phalloidin for detection of F-actin (8). In lectin-based triple-labeling studies, 3% BSA-PBS was substituted for 5% nonfat milk-PBS and cells/glands were stained with either anti-H+-K+-ATPase α-subunit (1:1,000) or the IQGAP1 or IQGAP2 antibodies plus Alexa Fluor 568 phalloidin, Alexa Fluor 488 GSII lectin (5 μg/ml, to stain mucous neck cells), and the Cy5-tagged secondary antibody. Independent controls for nonspecific binding and fluorescent emission crossover were included in each experiment as previously described (10). In brief, glands stained with primary antibodies and glands stained with secondary antibodies alone were analyzed on the same slides to confirm the absence of nonspecific staining by secondary antibodies. In multiple-labeling experiments, laser excitation, gains, and photomultiplier tube settings were adjusted such that there was no fluorescent emission crossover. This was accomplished by directly comparing emission signals between multiply labeled glands with glands that were singly labeled with each fluorophore. For primary antibodies, the efficiency of blocking of nonspecific binding was confirmed by Western blot analysis using the same blocking conditions employed in the immunolocalization analyses.

Phosphoprotein isolation and mass spectrometry analysis.

Phosphoproteins were isolated by using purification kits obtained from Qiagen (Valencia, CA) and Clontech (BD Biosciences). In each case, cellular extracts (2.5–6 mg protein) were fractionated on manufacturer-supplied phosphoprotein-affinity columns. Peak fractions were pooled, concentrated with Centricons (Mr cutoff, 10 kDa; Amicon) or TCA precipitation, and analyzed on SDS-PAGE gels (7–10%). The gels were stained with Colloidal Coomassie blue G-250 (Sigma, St. Louis, MO) for mass spectrometry (MS) analyses and band quantification. In the latter case, a portion of each sample was similarly electrophoresed and transferred to PVDF membrane for Western blotting (see above). Bands of interest were excised, destained, and sent to the Microchemical Facility at Emory University, where they were subjected to overnight “in-gel” tryptic digestion, extracted, and desalted with C18 zip tips. The resulting peptides were analyzed by liquid chromatography-MS/MS with an Applied Biosystems Qstar XL operating in information-dependent acquisition mode. Data were analyzed with ProID using All Taxa and Mascot using Mammalia.

Cloning strategy for rabbit IQGAP2.

Rabbit IQGAP2 was cloned by RT-PCR [oligo(dT)-generated cDNA from mucosal cell mRNA with a gene-specific primer that included the stop codon (5′-TCA CTT TCC ATA GAA CTT CTT GTT CAG-3′) followed by PCR with nested gene-specific primers and 3′- and 5′-RACE]. The initial cloning strategy was based on peptide sequences obtained from MS analyses and identification of highly conserved regions of the molecule identified by aligning human (NM_006633), mouse (BC_052916, BC_042790), and rat (XM_226697) IQGAP2 DNA sequences in GenBank. The complete open reading frame (ORF) of rabbit IQGAP2 was deduced from overlapping sequence analyses (MCG Molecular Biology Core).

Statistical analyses.

For confocal microscopic analyses, duplicate slides for each treatment (4–6 animals) were prepared and scanned at random. Images were acquired from a minimum of 25 glands/slide and analyzed as dichotomous observations using a nominal 0–1 scale. Significance was determined using the χ-square test. In other experiments, values were expressed as means ± SE with n equaling the number of cellular isolates from different animals. Where appropriate, paired comparisons were analyzed with Student's t-test and multiple treatments were analyzed with analysis of variance and Duncan's multiple range test.


Characterization of a triple-labeling technique for identification of major cell types in the oxyntic gastric mucosa.

The oxyntic gastric mucosa contains columnar surface epithelial cells that interface with tubular invaginations consisting of short pits followed by long glands. The main anatomic demarcations of gastric glands in this region are the pit, isthmus, neck, and base. Surface and pit cells contain mucous granules, as do a portion of the cells present in the neck region (mucous neck cells). Chief or zymogenic cells are localized at the base, whereas parietal cells are distributed throughout the base, neck, and isthmus of glands. These cells are thought to arise within the gland isthmus, which contains immature versions of all major cell types (20). In mucosal tissue sections, specific cell types can be identified based on their morphologies, staining characteristics, and relative positions within the gland units. In isolated cells and glands, specific stains and morphological criteria may also be used as differential markers. Parietal cells in isolation and within glands uniquely express the H+-K+-ATPase and possess complex intracellular canaliculi that are rich in F-actin. Chief cells contain the enzyme pepsinogen and can also be identified by their basophilic staining, granular appearance, and basal position within glands. In mouse, Gordon and colleagues (11) showed that mucous neck cells in the gland isthmus stain strongly with GSII lectin, whereas mucous cells in the surface/pit region do not. To confirm a similar lectin-staining pattern in rabbit, semi-intact gastric glands were triple labeled for fluorescently tagged GSII lectin, F-actin, and H+-K+-ATPase. Representative data in Fig. 1 confirm that rabbit mucous neck cells also bind GSII lectin. GSII staining was also evident at the apex of surface epithelial cells but not in pit cells. This latter staining pattern appeared to be nonspecific, resulting from the presence of adherent mucous on cell surfaces; however, this could not be unequivocally confirmed with confocal microscopy. In contrast to mucous neck cells, chief cells did not bind GSII lectin. As expected, parietal cells were exclusively labeled with the H+-K+-ATPase antibody. Occasionally, H+-K+-ATPase-staining parietal cells that stained moderately for F-actin and weakly for GSII lectin were identified in the isthmus. This staining pattern is expected in immature parietal cells (19). The cellular origin of parietal cells has not been definitively established in any species, but these cells have been postulated to arise from pre-neck and pre-pit cell precursors in humans (22). Thus the presence of GSII lectin-staining, mucous-like granules in a subpopulation of parietal cells suggests that these cells are immature and may have pre-neck cell origins.

Fig. 1.

Semi-intact, isolated gastric glands triple labeled for F-actin, mucous granules, and H+-K+-ATPase to show localization and staining characteristics of major gastric oxyntic epithelial cell types. Glands were isolated by brief collagenase digestion, fixed with 4% paraformaldehyde, and stained as described in materials and methods. Left: staining in regions near the surface. Right: staining from neck region to a region near the base. F-actin, red; GSII lectin, green; H+-K+-ATPase, blue. PC, parietal cell; IPC, immature parietal cell; SC, surface cell; MNC, mucous neck cell; CC, chief cell.

Cellular distribution of IQGAP1 and IQGAP2 as defined by Western blot analyses of gastric mucosal cell fractions from density gradients.

Once the staining characteristics of cells in rabbit gastric glands were defined, the triple-labeling technique was used to identify major cell types fractionated on isopycnic density gradients. As previously shown (6) and confirmed in Fig. 2, this procedure produces specific enrichment of parietal cells in the top gradient fraction with a progressive disappearance of this cell type in denser fractions. Interestingly, ∼10–15% of parietal cells in the first fraction that cross-reacted with the H+-K+-ATPase antibody also exhibited weak punctate staining with GSII lectin (Fig. 2, arrowheads). Similar lectin staining was rarely observed in parietal cells in other gradient fractions, which suggests that the top fraction contains the majority of immature parietal cells. Fractions 3 and 4 contained a mixture of cells, with mucous neck cells and small-diameter cells that are presumably pit cells (based on their relatively small size and absence of GSII labeling) predominating over other cell types. Fraction 5 was highly enriched in chief cells [based on the rarity of H+-K+-ATPase and GSII lectin staining (Fig. 2, A and B), basophilic cell staining, and high pepsinogen concentration (as assessed by SDS-PAGE analysis and pepsinogen assay; Ref. 6 and unpublished observations)]. F-actin staining was unexpectedly prominent in the interior of many chief cells (arrows, Fig. 2C). This presumably reflects an internalization of F-actin-rich apical membrane that normally abuts the glandular lumen in intact tissue and in isolated glands.

Fig. 2.

Gastric mucosal cells from isopycnic density gradient fractions triple labeled as in Fig. 1 to identify major cell types in these fractions. Top: from left to right, columns show H+-K+-ATPase, F-actin, GSII lectin, and merged images for each group. Numbers indicate gradient fractions of increasing density sampled from top (#1) to bottom (#5) of the gradient. Note progressive disappearance of parietal cells with increasing density. Arrowheads point to parietal cells that stained weakly with GSII lectin. On the basis of differential counting of cells dual labeled for H+-K+-ATPase and F-actin, the parietal cell content in fractions 1–5 was 86, 48, 32, 8, and 5%, respectively. Bar, 20 μm. Bottom: higher-magnification view of cells from fraction 5 stained for H+-K+-ATPase (A), mucous granules (GSII lectin, B), and F-actin (C) showing that isolated chief cells do not stain for H+-K+-ATPase or GSII lectin but frequently possess prominent intracellular pools of F-actin. These pools are not present in chief cells in gastric glands where the most prominent F-actin signal is present on apical membranes that line the gland lumen (see Fig. 8 for example). Bar, 10 μm.

Western blot analyses of the same density gradient fractions demonstrated an inverse correlation between parietal cell enrichment and IQGAP1 expression. In contrast, there was a similar level of expression of IQGAP2 and Cdc42 in all major mucosal cell types (Fig. 3). In these experiments, both H+-K+-ATPase and the actin-associated protein drebrin, which also appears to be parietal cell specific (23, 42), were used as internal controls for comparing the relative parietal cell content in each fraction. There was similar enrichment with both proteins; however, only drebrin was analyzed because of the difficulty in quantitating H+-K+-ATPase, which migrates as a diffuse band under conditions most ideal for SDS-PAGE analyses of other proteins. The high pepsinogen content in fraction 5 interfered with SDS-PAGE analyses, precluding an accurate determination of protein expression levels by Western blot analysis, but immunolocalization experiments detected both IQGAP1 (Fig. 3, AD) and IQGAP2 (not shown) on chief cell membranes.

Fig. 3.

Comparison of expression levels for IQGAP1, IQGAP2, Cdc42, and drebrin in gastric mucosal cell fractions 1–5 as determined by Western blot analysis and IQGAP1 staining patterns of chief cells in density gradient fraction 5. Top left: images of Western blots. Top right: phosphoimager quantitation of data shown at left. Samples of the cell fractions characterized in Fig. 2 were used in these analyses. Chief cells in fraction 5 contain high levels of the enzymatic secretory product pepsinogen, which interferes with the resolution of other cellular proteins in SDS-PAGE analyses. Therefore, density values for this fraction were not included and IQGAP1 expression was analyzed by immunolocalization (bottom). Results shown are representative of data from 3–5 independent experiments. Bottom: chief cells from fraction 5 triple labeled for IQGAP1 (A), GSII lectin (B), and F-actin (C). D: merged images, with IQGAP1 in blue, GSII lectin in green, and F-actin in red. Bar, 20 μm.

Subcellular localization of IQGAP1 and IQGAP2.

Confocal microscopy was used to define the subcellular distribution of IQGAPs in polarized epithelial cells within gastric glands. As shown in Fig. 4A, IQGAP1 was immunolocalized mainly to cellular cortexes. In some glands, a more prominent IQGAP1 signal was apparent in F-actin-enriched regions of cell-cell contact (arrows, Fig. 4, A and B). The peripheral localization of IQGAP1 within regions of cell-cell contact made it impossible to establish whether or not this protein is localized exclusively within nonparietal cells. However, IQGAP1 immunoreactivity was not detected in isolated parietal cells (not shown), and there was a negative correlation between the IQGAP1 signal and parietal cell enrichment in Western blot analyses (Fig. 3). Thus we concluded that there is little to no IQGAP1 expression in parietal cells relative to the other major oxyntic mucosal cell types. In contrast to IQGAP1, IQGAP2 was prominently immunolocalized in the nuclei as well as to sites of cell-cell contact of all major cell types present in glands (Fig. 4, C and D). The intensity of the nuclear signal varied depending on the level at which the confocal section was acquired. There was also variation in the degree of nuclear labeling from preparation to preparation (see Fig. 7D for example). In a total of eight independent gland isolates (≥50 glands imaged/isolate), neither IQGAP1 nor IQGAP2 was detected in the intracellular canalicular region of parietal cells in gastric glands; however, in parietal cells in primary culture, IQGAP2 was localized within the this region (Fig. 4, E and F) as previously reported (46). It was not possible to define the subcellular localization of Cdc42 because none of antibodies tested produced a specific, detectable signal in glands fixed with paraformaldehyde (2–8%) and permeabilized with Triton X-100 (0.1–0.5%). In cultured parietal cells transfected with a myc-tagged Cdc42 construct, Cdc42 was localized mainly to cell cortexes and was not present on intracellular canaliculi (not shown).

Fig. 4.

Subcellular localization of IQGAP1 and IQGAP2 in isolated gastric glands and IQGAP2 in parietal cells in primary culture. Gastric glands were isolated, paraformaldehyde fixed, and dual labeled for either IQGAP1 (A) and F-actin (B) or IQGAP2 (C) and F-actin (D) as described in materials and methods. Primary cultures of parietal cells were prepared and fixed, then stained for IQGAP2 (E) and F-actin (F) as described in materials and methods. Parietal cells in glands are readily identified on the basis of their size and position as well as the presence of F-actin-rich intracellular canaliculi (arrowheads, for example). Note absence of IQGAP1 signal in basal regions of parietal cells and general cortical staining of nonparietal cells in A. Arrows in A and B indicate F-actin-rich regions of cell-cell contact in which IQGAP1 is concentrated. IQGAP2 staining is not detectable in the apical membrane region of nonparietal cells but is prominent in regions of cell-cell contact and within nuclei. The strong pattern of nuclear staining with IQGAP2, which is apparent in most cells in C, was not as obvious when glands were scanned at a different focal plane (see Fig. 7D for example). In contrast to polarized parietal cells in glands, IQGAP2 staining is prominent within F-actin-rich canalicular regions in cultured parietal cells (arrows in E and F). Bars, 10 μm.

Evidence for IQGAP phosphorylation.

When gastric mucosal cell lysates were fractionated on phosphoprotein affinity columns and analyzed on SDS-PAGE gels, several Coomassie blue-stained bands were detected, including a major band of ∼190 kDa (Fig. 5A). MS analyses of the 190-kDa band tentatively identified IQGAP2 (11 peptide matches with human; 5 with mouse). The presence of IQGAP2 in column eluates was confirmed by Western blot (Fig. 5B) and a comigrating band stained with a specific phosphoprotein stain from Molecular Probes (not shown). IQGAP2 also migrated as multiple-charged species on isoelectric focusing gels, which is consistent with the addition of negatively charged phosphate ions (Fig. 5C). The known phosphoproteins, lasp-1 (8) and ezrin (45), were also detected in Western blots of phosphoprotein column eluates (not shown). Subsequent Western blot analyses also confirmed the presence of IQGAP1 in phosphoaffinity column eluates in which 30 ± 5% (n = 3) of the total IQGAP1 in cellular extracts was recovered.

Fig. 5.

Characterization of IQGAP2 recovered in phosphoaffinity column eluates and gastric mucosal cells. A: Coomassie blue-stained band of ∼190 kDa (arrow) was subjected to mass spectrometry analysis. Prestained Bio-Rad Precision Mr standards of 200 and 175 kDa are at left. B: Western blot analysis showing the presence of IQGAP2 in mucosal cell extracts after phosphoprotein affinity column purification. Data are representative of 4 similar experiments. C: Western blot of an IEF gel showing the migration pattern of IQGAP2 in a mucosal cell extract. This migration pattern is typical for a phosphoprotein with charge variants that are generated by consecutive additions of negatively charged phosphate groups. At least two major and two minor bands (arrows) could be detected in these experiments, but only the major bands produced sufficient signals for accurate quantitation. +, anode; −, cathode.

Cloning, sequencing, and phosphorylation site analysis of rabbit IQGAP2.

Because we were unable to locate the complete ORFs for rat or mouse IQGAP2 in BLAST searches of GenBank and EMBL databases, a combination of RT-PCR, 5′-RACE, and 3′-RACE (see materials and methods) was used to clone the DNA encoding for rabbit IQGAP2 and to deduce the amino acid sequence to confirm that rabbit IQGAP2 is a true homolog of the human protein. An alignment of human, mouse (partial), and rabbit IQGAP2 amino acid sequences is shown in Fig. 6A. The complete rabbit IQGAP2 ORF contained 1,572 amino acids that were 84% identical and 91% conserved compared with human IQGAP2 (NP_006624.1; Q13576; AAB37765.1). As in the human protein, the rabbit protein contains a calponin homology domain (residues 42–156, 99% conserved), a RasGAP domain (residues 938–1150, 100% conserved), and a RasGAP COOH-terminal domain (residues 1363–1498, 100% conserved) based on a BLAST CD-Search. Numerous conserved consensus phosphorylation sites were detected in rabbit and human IQGAP2 sequences in PhosphoBase version 2.0 analyses (24). These included 13–14 serine and 2–3 threonine residues (depending on the species) with P values >0.9. Five of these sites were conserved PKA phosphorylation sites, including S16, P = 0.99; S24, P = 0.99; S1455, P = 0.99; S1458, P = 0.99; and T1055, P = 0.98. Of these, four were also consensus sites for other protein kinases as follows: S16, p70s6 kinase, calmodulin kinase II (CAMKII), casein kinase (CK) II; S24, CKII; S1455 (T1458 in human), CAMK II/protein kinase G (PKG); and S1458 (S1461 in human), CKI. There was one conserved PKC consensus site (S911, P = 0.92 in human; P = 0.99 in rabbit) and an independent CAMKII site (T881). Upstream protein kinases were not identified for six of the total sites with high phosphorylation potentials (S245, S660, S742, S875, S1047, T98). The remaining sites were linked to either CKI (S595, S621) or CKII (S208, S432). The locations of PKA and PKC sites are shown schematically in Fig. 6B. Of potential significance is that these particular sites are localized within or very close to regulatory domains that have been shown to be important in IQGAP1 function. Thus S16 and S24 are upstream of the calponin homology domain, a region that has been shown to enhance actin polymerization and to be regulated by Cdc42 and Ca2+/calmodulin binding (4); S911 is upstream of the RasGAP domain, a region that is thought to promote cytoskeletal dynamics, possibly by associating with active Cdc42/Rac1 and with Ca2+/calmodulin binding, exerting a negative regulatory influence, and T1055 falls within this domain. Finally, both S1461 and S1455/T1458 are in the COOH-terminal RasGAP domain (Fig. 6B), which is thought to regulate cell adhesion by binding to E-cadherin and β-catenin and to be negatively regulated by calmodulin/Cdc42 (17, 26). This region also binds cytoplasmic linker protein (CLIP)-170, a protein that associates with the ends of growing microtubules (14).

Fig. 6.

Alignment of mammalian IQGAP2 sequences and comparison of structural domains and potential phosphorylation sites in IQGAP1 and IQGAP2. A: human, rabbit, and partial mouse IQGAP2 protein sequences. Regions sequenced from mass spectrometric analyses of the rabbit protein are underlined. †, conserved PKA phosphorylation consensus sites; *, conserved PKC site. B: diagrammatic comparisons of conserved domains and PKA and PKC phosphorylation consensus sites in IQGAP2 vs. IQGAP1 show similar domain structures but major differences in predicted phosphorylation patterns. Top: IQGAP2 contains 5 conserved PKA consensus sites and 1 PKC site. Bottom: IQGAP1 contains 8 PKC sites and two PKA sites. Black rectangles, WW domains; †, PKA; *, PKC; boxed regions, consensus site for multiple protein kinases (see text for details).

Phosphorylation site analysis of IQGAP1.

The rabbit IQGAP1 homolog was not cloned because there was 95% identity between the human (P46940) and mouse (AAH46385) amino acid sequences and 80% identity between the human and Xenopus (BAC75407) sequences. PhosphoBase analyses of human and mouse IQGAP1 sequences identified numerous conserved phosphorylation consensus sites in this protein as well, including the following (positions from the human IQGAP1 sequence): eight PKC consensus sites (S91, S280, S934, S998, S1134, S1545, S1556, S1560), two PKG sites (S360, S1560), and two PKA sites (T101, S360) (P > 0.9) (Fig. 6B). Of these, S91 falls within the calponin homology domain. S934 falls within the IQGAP1 domain immediately downstream of the WW domain, which also binds calmodulin, thereby possibly decreasing the binding of other targets (4). S1134 falls within the RasGAP domain. S1545, S1556, and S1560 (P = 0.99) are clustered within the RasGAP COOH-terminal domain. Interestingly, S1560 is a potential phosphorylation site not only for PKC and PKG but also for glycogen synthetase kinase-3 and CKI (in this latter case, only if S1556 is phosphorylated), and T101 is also a CAMKII consensus phosphorylation site. Of the PKC sites, all but S998 are also conserved in the Xenopus IQGAP1 homolog (BAC75407) and have similar P values. Both PKA consensus sites are also conserved (T99, S322; P = 0.99 and 0.83, respectively).

Effects of elevation of intracellular cAMP concentration and cholinergic stimulation on the distribution of IQGAP1 and IQGAP2 in gastric glands.

Stimulation of glands with the adenylyl cyclase activator forskolin induced an expansion of parietal cell actin-rich intracellular canaliculi as previously reported (8) [compare the canaliculi of forskolin-stimulated parietal cells in Fig. 7, A, C, and E (arrows) with the unstimulated parietal cell in Fig. 7F (arrows) and with unstimulated parietal cells in Fig. 4, B and D]. In the same experiments, forskolin stimulation did not detectably alter the distribution of IQGAP1 (Fig. 7B) or IQGAP2 (Fig. 7D). In addition to morphological changes (as evidenced by intracellular canalicular expansion), robust parietal cell responses to forskolin were confirmed by measurement of AP accumulation, a well-established index of acid-secretory responses in these cells (2). Stimulation with 10 μM forskolin for 15 min induced a 10-fold increase in AP accumulation (AP accumulation ratios were as follows: control, 22.7 ± 1.1; forskolin, 227 ± 15.6; n = 4). In similar experiments, stimulation of glands with 10 μM carbachol in the presence of the histamine H2 receptor blocker cimetidine (10 μM) for 15 min induced a 4.8-fold increase in AP accumulation (cimetidine control, 20.2 ± 1.2; carbachol, 96 ± 19; n = 4). In contrast to forskolin, carbachol stimulation induced an apparent translocation of IQGAP1 to the apical pole of cells lining the gland lumen (compare images in Fig. 8, AC with DF and Fig. 9). In these same experiments, the subcellular localization of IQGAP2 was not altered (compare images in Fig. 8, GI with Fig. 4, C and D and Fig. 7, CE). Differential staining with F-actin and GSII lectin indicated that IQGAP1 translocation occurs mainly in chief cells and mucous neck cells (Fig. 9). The PKC activator PMA (100 nM, 15 min) induced a similar translocation (Fig. 10, AC), whereas the calcium ionophore ionomycin (3 μM, 5–15 min) had no apparent effect (Fig. 10, DF).

Fig. 7.

Elevation of intracellular cAMP concentration does not alter the subcellular distribution of IQGAP1 or IQGAP2 in gastric glands. Glands were stimulated with the adenylyl cyclase activator forskolin (10 μM, 15 min), paraformaldehyde-fixed, and dual labeled for F-actin (A) and IQGAP1 (B) or F-actin (C) and IQGAP2 (D) as described in materials and methods. E: higher-magnification merged image of a parietal cell in C (F-actin, green; IQGAP2, red). F: comparable high-magnification image of a parietal cell in an unstimulated gland included to show that the expected canalicular expansion, a hallmark effect of this secretory agonist, occurred in forskolin-stimulated parietal cells. Arrows indicate the positions of several intracellular canaliculi in these cells.

Fig. 8.

Cholinergic stimulation induces the translocation of IQGAP1, but not IQGAP2, to apical poles of cells in isolated gastric glands. Control, AC; carbachol (10 μM, 15 min), DI. Glands were fixed and dual labeled for IQGAP1 (A, D; red) and F-actin (B, E; green) or IQGAP2 (G) and F-actin (H) as in Fig. 4. C and F are merged images of A and B and D and E, respectively; I is a merged image of a higher magnification of a parietal cell in G (red) and H (green). Note distinct IQGAP1 staining on basal membranes of nonparietal cells (arrows) and absence of detectable staining on basal membranes of parietal cells (arrowheads). Images are representative of those obtained from 4 independent gland isolates with 25–50 glands scanned/isolate. P < 0.001, IQGAP1 distribution significantly different with carbachol vs. control. Bars, 10 μm.

Fig. 9.

Cholinergically induced redistribution of IQGAP1 occurs in mucous neck and in chief cells in gastric glands. Images are of representative glands from 5 independent experiments that were stimulated with carbachol (15 min) in the presence of the histamine H2 receptor blocker ranitidine (10 μM), paraformaldehyde fixed, and triple labeled for IQGAP1 (A, E), F-actin (B, F), and GSII lectin (C, G) as in Fig. 2. Merged images are shown in D and H with IQGAP1 (blue), F-actin (red), and GSII lectin (green). GSII staining identifies mucous neck cell (MNC). Parietal cells (PC) are identified by the presence of F-actin-rich intracellular canaliculi, large size, and triangular-to-rounded (in less mature cells) morphology. Chief cells (CC) are interspersed among these two cell types. They are typically round-to-oval-shaped cells that stain strongly for apical and weakly for basolateral F-actin. CC do not contain intracellular canaliculi and do not stain with GSII lectin. Controls for these experiments are not shown but were characteristic of the images in Fig. 7, A and B. P < 0.001, IQGAP1 distribution significantly different with carbachol vs. paired controls. Bar, 20 μm.

Fig. 10.

IQGAP1 redistribution is mimicked by PKC activation but not by elevation of [Ca2+]i. Glands in AC were incubated with the PKC activator PMA (100 nM, 15 min). Glands in DF were incubated with the calcium ionophore ionomycin (3 μM, 5 min). Controls (not shown) were characteristic of the images in Fig. 8, A and B. After incubation, glands were fixed and dual labeled for IQGAP1 (A, D) and F-actin (B, E), as described in materials and methods. Merged images are shown in C and F in which IQGAP1 labeling is represented in red and F-actin in green. Data are representative of 3–8 independent experiments with 50–75 glands scanned/isolate; P < 0.001, IQGAP1 distribution is different with PMA vs. control. Significance was not tested for ionomycin because changes in fluorescence patterns were not sufficiently pronounced to be analyzed as dichotomous observations. Thus a subtle effect of ionomycin on the subcellular distribution of IQGAP1 cannot be ruled out.


The present results provide the first evidence that IQGAPs 1 and 2 are phosphorylated and differentially expressed and regulated in the major epithelial cell types in the oxyntic gastric mucosa. IQGAP2 appears to be expressed uniformly in different cell types within the gastric oxyntic mucosa and to be localized mainly at cell-cell contacts and in nuclei. In contrast, IQGAP1 expression is largely restricted to nonparietal cells and is targeted to lateral cortical regions. The divergence in localization and expression patterns of IQGAP family members suggests the possibility that these proteins modulate distinct cellular activities. In addition, the pronounced differences in phosphorylation consensus sites, with a preponderance of PKA consensus sites in IQGAP2 and a preponderance of PKC sites in IQGAP1, suggests that these proteins may also be differentially regulated by phosphorylation-dependent mechanisms. The predominance of conserved PKC phosphorylation sites in IQGAP1 is particularly intriguing given that both cholinergic stimulation and PKC activation induce the apparent translocation of IQGAP1, but not IQGAP2, to the luminally facing regions of chief and mucous neck cells. It is also interesting to note that the majority of potential PKC phosphorylation sites are located within important regulatory domains. Several of the predicted phosphorylation sites in IQGAP1 and IQGAP2 are consensus sites for multiple protein kinases and are located within or near the NH2-terminal calponin homology domain region and the COOH-terminal RasGAP domain. Thus these regions may serve as important targets for agonist-dependent regulation. Additional studies are required to establish if this is indeed the case.

Our observations in glands conflict with an earlier report in which IQGAP1 was localized to the basolateral membrane and IQGAP2 was localized along with Cdc42 to the canalicular membrane of parietal cells in primary culture (46). The disparities between these studies likely reflect, at least partly, differences in cellular models. Cells within gastric glands are polarized and retain normal cell-cell contacts as in intact tissue. When parietal cells are isolated from glands, there is a loss of polarity and their canalicular membranes no longer communicate directly with the extracellular milieu (7). There may also be some internalization of junctional complexes and membrane in parietal cells in primary culture, as is the case with chief cell isolates (See Fig. 2B, for example). Such internalization could explain why IQGAP2 is detected within F-actin-rich intracellular compartments of cultured parietal cells (46) but not in parietal cells that retain their normal cell-cell contacts as is the case in glands. With respect to Cdc42, we were unable to specifically immunolocalize this protein within any of the glandular gastric cell types using several different antibodies under a range of conditions, including one that was used previously to localize this protein to the canalicular membrane (46). One possibility is that fluorescent “bleedthrough” in samples dual labeled for F-actin generated an artifactual Cdc42 signal in the earlier study (46). Western blot analyses presented herein indicate that Cdc42 is uniformly distributed in the various cell types present within the oxyntic gastric mucosa, and overexpression studies indicate that Cdc42 is localized mainly at the cell cortex, at least in cultured parietal cells overexpressing Cdc42 (data not shown). Because IQGAP2 is not localized to the intracellular membranes of normally polarized parietal cells, it seems unlikely that this protein links Cdc42 to the canalicular membrane or that, as previously proposed (46), IQGAP2 is directly involved in the regulation of canalicular membrane remodeling that occurs on cAMP-dependent stimulation.

The targeting of IQGAP2 to lateral cortical regions of gastric epithelial cells suggests a role for this protein in regulating cell-cell associations and/or migration rather than in the regulation of HCl secretion. Additional studies are necessary to determine if Cdc42 modulates HCl secretion through an actin-dependent mechanism, independent of IQGAP2. Although there is currently no defined function for IQGAP2 in a nuclear context, present findings are in agreement with the nuclear localization of XIQGAP2 in cultured Xenopus cells and embryos (44). The fact that two different IQGAP2 antibodies exhibited similar nuclear cross-reactivities supports the relevance of this observation. It is not yet known whether or not IQGAP2 plays a role in transcriptional regulation and/or gastric epithelial cell migration. A positive finding in this regard may help to explain previous findings in which mice deficient in IQGAP1 showed no obvious developmental defects and tumor development and progression was not altered (28). Thus IQGAP2 may play a more prominent role in maintaining gastric mucosal integrity than does IQGAP1.

The observation that cholinergic stimulation induces the apparent translocation of IQGAP1 to the apical poles of zymogen and mucous-secreting cells suggests a potential new role for this protein in mediating regulated exocytotic events. Thus far, the majority of functional studies of IQGAP1 have focused on the regulation of cell adhesion (13, 26, 27, 29) or on regulatory interactions between IQGAP1 and the small GTPases Cdc42 and Rac1 in the context of actin cytoskeletal signaling in cultured cells (4). However, IQGAP1 is expressed in nonadherent blood cells (39) and thus presumably possesses functions in addition to those associated with cell adhesion. Cytoskeleton-associated interactions that appear to be unrelated to cell adhesion have also been reported. For example, IQGAP1 is present in complexes that contain the cytoskeletal proteins actin, ezrin, α-actinin, and gelsolin, as well as chloride intracellular channel protein 5 (3). There is also some evidence that IQGAP1 interacts with CLIP-170, a protein that binds the growing ends of microtubules and plays a role in maintaining cell polarization (14).

To our knowledge, only one other study has examined the effects of cholinergic stimulation on IQGAP1 distribution. In this case, cholinergic activation of stably transfected M3 muscarinic receptors in Chinese hamster ovary cells induced the translocation of IQGAP1 and Rac1 to cell junctions (38). These findings were interpreted in the context of cell adhesion, and it was suggested that muscarinic receptor activation might enhance cell adhesion by increasing the level of GTP-bound Rac1, which would then bind to IQGAP1, causing the dissociation of this inhibitory protein from E-cadherin complexes (38). In the case of gastric epithelial cells, the rapid translocation of IQGAP1 to the apical pole of chief and mucous neck cells is temporally more in keeping with the known exocytotic response of these cells to cholinergic stimulation rather than cell adhesion, which is likely to be altered more slowly as these cells migrate downward through the glands (21). The finding that IQGAP1 redistribution is mimicked by PKC activation further suggests that IQGAP1 may be regulated not only by Ca2+/calmodulin (32) but also by PKC-dependent mechanisms. If, as proposed for NIH/3T3 cells (31), elevation of [Ca2+]i reduces the interaction between IQGAP1 and actin filaments in gastric mucosal cells, this interaction could serve to release IQGAP1 from the cell cortex. Elevated Ca2+/calmodulin might exert a similar effect by suppressing the association between IQGAP1 and active Cdc42 (16, 17, 31). This interaction could be further modulated by the activation of PKC and subsequent PKC-dependent phosphorylation of IQGAP1 or some associated regulatory protein. Thus cholinergically induced translocation of IQGAP1 in gastric epithelial cells may be a multistep process in which elevated intracellular Ca2+ concentration destabilizes the association between IQGAP1, active Cdc42, and F-actin and a PKC-dependent phosphorylation event induces translocation. Future studies are needed to determine whether or not translocated IQGAP1 modifies Cdc42 or Rac1-dependent functions of the actin cytoskeleton or if other regulatory proteins are involved. It will also be important to determine whether IQGAP1 is directly involved in mediating actin-associated exocytotic functions.


This work was supported by National Institutes of Health grants DK-31900 (to C. S. Chew) and DK-51588 (to C. T. Okamoto).


We thank Dr. Jan Pohl and Matthew Reed at the Emory University Microsequencing Facility for their assistance in the mass spectrometry analyses and sequencing of the rabbit IQGAP2 protein. We also thank John Nechtman and Michelle Glendenning in the Molecular Biology Core at Medical College of Georgia for their assistance in the DNA sequencing of IQGAP2.

The nucleotide sequence reported in this paper has been submitted to GenBank with accession no. AY390427.


  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


View Abstract