Abstract

Polyamines are required for the early phase of mucosal restitution that occurs as a consequence of epithelial cell migration. Our previous studies have shown that polyamines increase RhoA activity by elevating cytosolic free Ca2+ concentration ([Ca2+]cyt) through controlling voltage-gated K+ channel expression and membrane potential (E m) during intestinal epithelial restitution. The current study went further to determine whether increased RhoA following elevated [Ca2+]cyt activates Rho-kinase (ROK/ROCK) resulting in myosin light chain (MLC) phosphorylation. Studies were conducted in stable Cdx2-transfected intestinal epithelial cells (IEC-Cdx2L1), which were associated with a highly differentiated phenotype. Reduced [Ca2+]cyt, by either polyamine depletion or exposure to the Ca2+-free medium, decreased RhoA protein expression, which was paralleled by significant decreases in GTP-bound RhoA, ROCK-1, and ROKα proteins, Rho-kinase activity, and MLC phosphorylation. The reduction of [Ca2+]cyt also inhibited cell migration after wounding. Elevation of [Ca2+]cyt induced by the Ca2+ ionophore ionomycin increased GTP-bound RhoA, ROCK-1, and ROKα proteins, Rho-kinase activity, and MLC phosphorylation. Inhibition of RhoA function by a dominant negative mutant RhoA decreased the Rho-kinase activity and resulted in cytoskeletal reorganization. Inhibition of ROK/ROCK activity by the specific inhibitor Y-27632 not only decreased MLC phosphorylation but also suppressed cell migration. These results indicate that increase in GTP-bound RhoA by polyamines via [Ca2+]cytcan interact with and activate Rho-kinase during intestinal epithelial restitution. Activation of Rho-kinase results in increased MLC phosphorylation, leading to the stimulation of myosin stress fiber formation and cell migration.

  • mucosal injury
  • intracellular calcium
  • Cdx2 gene
  • dominant negative mutant RhoA
  • cytoskeleton

the small GTPase Rho functions as a molecular switch of various cellular processes by shuttling between the inactive GDP-bound form and the active GTP-bound form (11, 14). Rho exerts its distinct actions through interactions with specific targets. Recently, numerous effector molecules of Rho have been identified, including protein kinase N (PKN) (3, 55), ROK/ROCK (20, 40), the myosin-binding subunit of myosin phosphatase (15), p140mDia (56), citron and citron-kinase (19), rhophilin (55), rhotekin (35), and rectifier potassium channel (4). Among these effectors, ROK/ROCK is characterized as a downstream target of Rho and has been implicated in the regulation of cell shape and dynamic reorganization of cytoskeletal proteins (1, 13, 25, 37, 44, 47). ROK/ROCK is commonly divided into two isoforms, ROCK-I and ROCK-II, corresponding to ROKβ and ROKα, respectively. The active form of Rho interacts with the COOH-terminal portion of the putative coiled-coil domain of ROK/ROCK and activates its phosphotransferase activity (20). Increasing evidence indicates that activated ROK/ROCK regulates the phosphorylation of myosin light chain (MLC) by the direct phosphorylation of MLC and by the inactivation of myosin phosphatase through the phosphorylation of myosin binding subunit (2, 15). MLC phosphorylation is crucial for the actin-myosin interaction for the formation of stress fibers and contractile rings in nonmuscle cells, thus resulting in cell migration (2, 25).

Early mucosal restitution is a primary repair modality in the gastrointestinal tract and occurs as a consequence of epithelial cell migration to resealing of superficial wounds after injury (5, 6,27, 41, 50). This rapid mucosal reepithelialization following superficial wounding is a complex process that includes the flattening, spreading, migrating, and repolarizing of differentiated columnar epithelial cells. Our previous studies (10, 30, 54) and others (5, 6, 38) have examined this process of intestinal epithelial cell migration in an in vitro system that resembles the early stage of mucosal healing in vivo, such as independence from cell division, complete dependence on cytoskeletal reorganization, and absolute requirement of polyamines. Although most of these studies employ undifferentiated intestinal epithelial cells (IEC-6 line) as a model, we (31) have recently demonstrated that differentiated intestinal epithelial cells (IEC-Cdx2L1 line) induced by forced expression of the Cdx2 gene, which encodes a transcription factor controlling intestinal epithelial cell differentiation (42, 46), migrate over the wounded edge much faster than undifferentiated parental IEC-6 cells. Increased migration of differentiated IEC-Cdx2L1 cells after wounding results, at least partially, from the activation of voltage-gated K+(Kv) channels and the increase in the driving force for Ca2+ influx during restitution (32). Because the early rapid mucosal repair is function of differentiated intestinal epithelial cells from the surface of the mucosa in vivo, these differentiated IEC-Cdx2L1 cells provide an excellent in vitro model for restitution.

The natural polyamines spermidine and spermine and their precursor, putrescine, are organic cations found in all eukaryotic cells (43). It has been recognized for some time that the control of cellular polyamines is a central convergence point for the multiple signaling pathways driving different epithelial cell functions including cell motility, proliferation, and apoptosis (17, 18, 50). Studies from our laboratory (30, 31,33, 54) and others (18, 21, 22) have shown that polyamines are necessary for the stimulation of cell migration after wounding and play a critical role in the maintenance of intestinal mucosal integrity. The process of early mucosal restitution is associated with a dramatic increase in polyamine synthesis both in vivo (17, 50, 51) and in vitro (30, 31, 54) after wounding, and depletion of cellular polyamines by inhibition of ornithine decarboxylase (ODC), the rate-limiting enzyme in polyamine biosynthesis, withd,l-α-difluoromethylornithine (DFMO) inhibits cell migration and delays mucosal healing. Although little is known about specific functions of polyamines in the regulation of cell migration, these compounds have been shown to stimulate the expression of Kv channel genes, induce membrane hyperpolarization, and increase Ca2+ influx during restitution in intestinal epithelial cells that do not express voltage-dependent Ca2+ channels (54).

We (30) and others (34) have recently found that polyamines are necessary for the activity and synthesis of RhoA by raising cytosolic free Ca2+ concentration ([Ca2+]cyt) and that activation of RhoA plays an important role in polyamine-dependent intestinal epithelial cell migration after wounding. The aim of the current study was to further determine whether activated RhoA increases ROK/ROCK activity resulting in MLC phosphorylation. First, we examined whether manipulating RhoA activity by altering cellular polyamines, [Ca2+]cyt, or overexpression of the dominant negative mutant RhoA affected the expression of ROK/ROCK proteins and the Rho-kinase enzyme activity in differentiated IEC-Cdx2L1 cells. Second, we determined whether observed activation of ROK/ROCK regulated MLC phosphorylation in the presence or absence of polyamines. Third, we determined whether inhibition of ROK/ROCK activity by treatment with the specific inhibitor Y-27632 [(+)-R-trans-4-(1-aminomethyl)-N-(4-pyridyl) cyclohexanecarboxamide] altered cellular distribution of actomyosin stress filaments and decreased cell migration. Some of these data have been published in abstract form (29).

MATERIALS AND METHODS

Materials.

Disposable culture ware was purchased from Corning Glass Works (Corning, NY). Tissue culture media and dialyzed fetal bovine serum (dFBS) were obtained from GIBCO-BRL (Gaithersburg, MD), and biochemicals were from Sigma (St. Louis, MO). The primary antibody, an affinity-purified mouse monoclonal antibody against ROKα or ROCK-1, was purchased from BD Biosciences (San Jose, CA). Specific antibodies against RhoA and MLC kinase (MLCK) were from Santa Cruz Biotechnology (Santa Cruz, CA). The specific rabbit polyclonal antibody against nonmuscle myosin II was obtained from Biomedical Technologies (Stoughton, MA). Rhotekin Rho binding domain assay kit and dominant negative RhoA cDNA in pUSEamp were purchased from Upstate Biotechnology (Lake Placid, NY). Y-27632 was purchased from Calbiochem (San Diego, CA). Adeno-X Expression system was obtained from Clontech Laboratories (Palo Alto, CA). [32P]orthophosphate and [γ-32P]ATP were obtained from Amersham (Arlington Heights, IL). DFMO was purchased from Ilex Oncology (San Antonio, TX).

Cell culture and general experimental protocol.

The stable Cdx2-transfected IEC-6 cells were developed and characterized by Suh and Traber (42) and were a kind gift from Dr. Peter G. Traber (University of Pennsylvania, Philadelphia, PA). The expression vector, the LacSwitch System (Stratagene, La Jolla, CA), was used for directing the conditional expression ofCdx2, and isopropyl-β-d-thiogalactopyranoside (IPTG) served as the inducer for the gene expression (46). IEC-6 cells, derived from normal rat intestinal crypts, were transfected with pOPRSVCdx2 by electroporation technique, and clones resistant to selection medium containing 0.6 mg/ml G418 and 0.3 mg/ml hygromycin B were isolated and screened for Cdx2 expression by Northern blot, RNase protection assays, and electrophoretic mobility shift assay. Stock-stable Cdx2-transfected IEC-6 (IEC-Cdx2L1) cells were grown in DMEM supplemented with 5% heat-inactivated FBS, 10 μg/ml insulin, and 50 μg/ml gentamicin sulfate. Before experiments, cells were grown in DMEM containing 4 mM IPTG for 16 days to induce cell differentiation as described in our previous publications (31, 32) and by others (42,46).

In the first series of experiments, we examined whether manipulating RhoA activity by altering cellular polyamines, [Ca2+]cyt, or infection of the adenoviral vector containing a dominant negative mutant RhoA (AdDNMRhoA) affected expression of ROKα and ROCK-1 and the Rho-kinase enzyme activity in differentiated IEC-Cdx2L1 cells. The cells were grown in control cultures and in cultures containing either 5 mM DFMO or DFMO plus 5 μM spermidine (SPD) for 4 days. The Ca2+ ionophore ionomycin (1 μM) was used to increase [Ca2+]cyt, whereas Ca2+-free medium was employed to decrease [Ca2+]cyt. Levels of GTP-bound RhoA, ROKα, and ROCK-1 were measured by immunoprecipitation and Western blot analysis after various treatments, and the Rho-kinase enzyme activity was determined on immunoprecipitates by the in vitro assay. Cells were washed three times with ice-cold Dulbecco's PBS (D-PBS), and different solutions were then added according to the assays to be conducted.

In the second series of studies, we investigated whether RhoA-mediated ROK/ROCK activation induced the MLC phosphorylation. Function of ROK/ROCK was inhibited by either treatment with its specific inhibitor Y-27632 (7, 40, 49) or reduction of [Ca2+]cyt through polyamine depletion. The MLC phosphorylation was measured by using [32P]orthophosphate assays after various treatments.

In the third series of studies, we determined whether ROK/ROCK activation played a role in polyamine-dependent intestinal epithelial cell migration after wounding. Differentiated IEC-Cdx2L1 cells were grown in control cultures and in cultures containing DFMO and SPD for 4 days. Y-27632 at various concentrations was added immediately after wounding. The rates of cell migration and cellular distribution of myosin II were assayed 6 h after treatments.

Recombinant adenovirus construction and infection.

Adenoviral vectors were constructed using the Adeno-X Expression system (Clontech) according to the protocol recommended by the manufacturer (23). Briefly, the cDNA of human dominant negative mutant RhoA (DNMRhoA) was cloned into the pShuttle by digesting the pUSEamp(+)/DNMRhoA (T19N) withEcoR1/Xho1 and ligating the resulting fragments into the Xba1 site of the pShuttle vector. pAdeno-X/DNMRhoA (AdDNMRhoA) was constructed by digesting pShuttle constructs with PI-SceI/I-CeuI and ligating the resulting fragments into the PI-SceI/I-CeuI sites of the pAdeno-X adenoviral vector. Recombinant adenoviral plasmids were packaged into infectious adenoviral particles by transfecting human embryonic kidney (HEK)-293 cells by using LipofectAMINE PLUS reagent. The adenoviral particles were propagated in HEK-293 cells and purified on cesium chloride ultracentrifugation. Titers of the adenoviral stock were determined by standard plaque assay. Recombinant adenoviruses were screened for expression of the introduced genes by fluorescence microscopy and Western blot analysis using anti-RhoA antibody. pAdeno-X, which was the recombinant replication-incompetent adenovirus carrying no cDNA insert (AdNull), was grown and purified as described above and served as a control adenovirus. Cells were infected by various concentrations of AdDNMRhoA or AdNull, and cell samples were collected for various measurements 72 h after the infection.

Measurement of [Ca2+]cyt.

Details of the digital imaging methods employed for measuring [Ca2+]cyt are described in our previous publications (30, 32, 54). Briefly, IEC-Cdx2L1 cells were plated on 25-mm coverslips and incubated in culture medium containing 3.3 μM fura 2-AM for 30–40 min at room temperature (22–24°C) under an atmosphere of 10% CO2 in air. Fluorescent images were obtained by using a microchannel plate image intensifier (Amperex XX1381; Opelco, Washington, DC) coupled by fiber optics to a Pulnix charge-coupled device video camera (Stanford Photonics, Stanford, CA). Image acquisition and analysis were performed with a Metamorph Imaging System (Universal Imaging). The concentration of [Ca2+]cyt was calculated from fura 2 fluorescence emission excited at 380 and 360 nm using the ratio method (28).

Determination of cellular GTP-bound RhoA.

The GST-tagged fusion protein, corresponding to residues 7–89 of mouse rhotekin Rho binding domain, was used to determine the cellular GTP-bound RhoA by the affinity precipitation (pull down) protocol according to the manufacturer's instructions with slight modifications (36). Briefly, cells were lysed with cold Rho binding lysis buffer containing 50 mM Tris, pH 7.2, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 500 mM NaCl, 10 mM MgCl2, 10 μg/ml leupeptin, 10 μg/ml aprotinin, and 1 mM phenylmethylsulfonyl fluoride (PMSF). After brief sonication, cell lysates were clarified by centrifugation at 14,000 g at 4°C for 10 min. A small portion of supernatants (40 μg) was taken for determinations of protein concentrations by BCA reagent and total RhoA by Western blot analysis. Equal amounts of (500 μg) supernatants were incubated with 30 μg of glutathione-agarose slurry of rhotekin Rho binding domain for 45 min at 4°C on a rocker platform. The beads were then washed three times with a washing buffer comprising 50 mM Tris, pH 7.2, 1% Triton X-100, 150 mM NaCl, 10 mM MgCl2, 0.1 mM PMSF, 10 μg/ml aprotinin, and 10 μg/ml leupeptin. RhoA bound to agarose beads was solubilized in Laemmli's SDS sample buffer and boiled for 5 min. Each sample was analyzed by 15% SDS-PAGE, followed by Western immunoblotting with the specific antibody against RhoA. Specific bands were visualized with a chemiluminescence detection system (NEL-100; Du Pont NEN). The amount of GTP-bound RhoA was normalized for the total amount of RhoA (1/12.5) in each sample.

Measurement of ROK/ROCK enzyme activity.

ROK/ROCK enzyme activity was determined on immunoprecipitate from cell extracts as described previously (24). Cells were rinsed with ice-cold PBS containing 1 mM vanadate and 1 μM microcystin-LR, and cells lysates were sonicated at 4°C for 20 s and centrifuged for 10 min at 10,000 rpm. The supernatant was incubated for 1 h with 100 μl of protein A/G plus agarose beads, and, after centrifugation, the amount of protein was determined in the supernatant. Equal amounts of proteins were immunoprecipitated with 2 μg of the antibody against ROKα for 2 h and then with 100 μl of protein A/G-agarose beads for another 1 h at 4°C. After the incubation, supernatant was carefully removed, and the beads were washed for four times with lysis buffer and two times with kinase buffer. The ROK/ROCK kinase activity was measured in a reaction mixture (final volume 50 μl) containing 20 mM Tris · HCl, pH 7.5, 100 mM KCl, 0.1 mM DTT, 5 mM MgCl2, 1 mM EDTA, 1 μM microcystin-LR, 1 mM ATP, and 10 μM myosin light chain-20 (MLC20). Reactions were initiated by the addition of [γ-32P]ATP to a final concentration of 100 μM. After incubation at 30°C for 10 min, 25-μl aliquots were removed and added to phosphocellulose disks. The disks were washed immediately for 10 min with 75 mM phosphoric acid and then dried, and the 32P radioactivity was determined by adding 5 ml of scintillation fluid to vials containing paper. ROK/ROCK enzyme activity was expressed as counts per minute per milligram of protein per minute.

Western blot analysis.

Cell samples, placed in SDS sample buffer, were sonicated and then centrifuged (12,000 rpm) at 4°C for 15 min. The supernatant from cell samples was boiled for 5 min and then subjected to electrophoresis on 7.5% or 12.5% SDS-PAGE gels according to Laemmli (16). After the transfer of protein onto nitrocellulose filters, the filters were incubated for 1 h in 5% nonfat dry milk in 1× phosphate-buffered saline/Tween 20 [PBS-T: 15 mM NaH2PO4, 80 mM Na2HPO4, 1.5 M NaCl, pH 7.5, and 0.5% (vol/vol) Tween 20]. Immunological evaluation was then performed for 1 h in 1% BSA/PBS-T buffer containing the specific antibody (1 μg/ml) against RhoA, ROKα, ROCK-1, or MLC protein. The filters were subsequently washed with 1 × PBS-T and incubated for 1 h with the second antibody conjugated to peroxidase by protein cross-linking with 0.2% glutaraldehyde. After extensive washing with 1× PBS-T, the immunocomplexes on the filters were reacted with chemiluminescence reagent and then exposed to autoradiography film.

Assay for MLC phosphorylation.

The MLC phosphorylation was assessed by using a method described previously (39). Before the addition of [32P]orthophosphate, cells were rinsed with balanced salt solution without phosphate (BSS: 10 mM HEPES buffered with Tris, pH 7.4, 140 mM NaCl, 4.5 mM KCl, 1 mM MgCl2, and 1.5 mM CaCl2) and then incubated with 4 ml of BSS containing 0.2 mCi/ml [32P]orthophosphate at 37°C for 3 h. After the incubation, cells were lysed with RIPA buffer containing protease inhibitors. Equal amounts of proteins (500 μg) were taken from each sample and immunoprecipitated with anti-MLC antibody for 2 h, followed by the addition of protein L-agarose beads, the incubation was continued for another hour. Beads were washed with lysis buffer and the proteins suspended in Laemmli buffer, followed by the separation of proteins on 15% SDS-PAGE gel. Phosphorylated MLC (P-MLC) bands were analyzed by autoradiography. Densities of bands corresponding to P-MLC were measured by a densitometric analysis.

Nonmuscle myosin II staining.

The immunofluorescence procedure was carried out according to the method of Vielkind and Swierenga (48) with minor changes (30, 32). The primary antibody recognizes the 200-kDa nonmuscle myosin II in immunoblots of IEC-Cdx2L1 cell extracts and does not cross-react with other cytoskeletal proteins (31). Nonspecific slides were incubated without antibody to nonmuscle myosin II. Slides were viewed through a Zeiss confocal microscope (model LSM410).

Measurement of cell migration.

The migration assays were carried out as described in our earlier publications (30-32). Cells were plated at 6.25 × 104/cm2 in DMEM-dFBS on 60-mm dishes thinly coated with Matrigel according to the manufacturer's instructions. To initiate migration, we scratched the cell layer with a single-edge razor blade cut to ∼27 mm in length. The scratch began at the diameter of the dish and extended over an area 7–10 mm wide. The migrating cells in six contiguous 0.1-mm squares were counted at ×100 magnification beginning at the scratch line and extending as far out as the cells had migrated. All experiments were carried out in triplicate, and the results are reported as the number of migrating cells per millimeter of scratch.

Statistics.

All data are expressed as means ± SE from six dishes. Autoradiographic and immunofluorescence labeling experiments were repeated three times. The significance of the difference between means was determined by analysis of variance. The level of significance was determined using Duncan's multiple-range test (12).

RESULTS

Effects of cellular polyamines and [Ca2+]cyt on GTP-bound RhoA in differentiated IEC-Cdx2L1 cells.

Our previous studies have shown that reduced [Ca2+]cyt concentration by either polyamine depletion or the removal of extracellular Ca2+significantly decreased expression of total RhoA protein in undifferentiated intestinal epithelial cells (parent IEC-6 line) and that elevation of [Ca2+]cyt by the Ca2+ ionophore ionomycin increased RhoA expression regardless of the presence or absence of cellular polyamines (30,32). Because cellular RhoA has the GDP-bound inactive form and the GTP-bound active form (14, 26), the current study was further to determine whether alteration of cellular polyamines and [Ca2+]cyt affected levels of GTP-bound RhoA in differentiated intestinal epithelial cells (IEC-Cdx2L1 line).

Exposure of differentiated IEC-Cdx2L1 cells to 5 mM DFMO for 4 days, which totally inhibited ODC activity (31, 32), almost completely depleted cellular polyamines (data not shown), which was associated with a decrease in K+ channel activity, depolarized E m, and reduction of [Ca2+]cyt (data not shown). The reduction of [Ca2+]cyt in polyamine-depleted cells was accompanied by a significant decrease in RhoA expression (Fig. 1,A and B). Levels of GTP-bound RhoA and total RhoA in DFMO-treated cells were decreased by ∼55% and ∼40%, respectively. Addition of 5 μM spermidine to the cultures containing DFMO not only prevented the reduction of [Ca2+]cyt (data not shown) but also restored GTP-bound RhoA and total RhoA to normal. Levels of [Ca2+]cyt, GTP-bound RhoA, and total RhoA in cells treated with DFMO plus spermidine were similar to those of control cells. On the other hand, exposure to 1 μM ionomycin increased [Ca2+]cyt in DFMO-treated cells (data not shown). Consistent with the effect on [Ca2+]cyt, treatment with ionomycin also increased levels of GTP-bound RhoA in polyamine-deficient cells (Fig.1, A and B). Furthermore, removal of extracellular Ca2+ from the culture medium completely prevented the restoration of GTP-bound RhoA by exogenous spermidine. Neither the Ca2+-free medium nor ionomycin altered cell attachment and cell viability in control and DFMO-treated cells (data not shown). These results indicate that polyamines increase levels of GTP-bound RhoA, at least partially, through Ca2+ in differentiated intestinal epithelial cells.

Fig. 1.

Changes in levels of GTP-bound RhoA (GTP-RhoA) and total RhoA proteins in control and polyamine-deficient cells. Before experiments, IEC-Cdx2L1 cells were grown in DMEM containing 5% FBS in the presence of 4 mM isopropyl β-thiogalactopyranoside (IPTG, the inducer for gene expression) for 16 days to induce cell differentiation. These differentiated cells were grown in control cultures and in cultures containing 5 mM α-difluoromethylornithine (DFMO) with or without 5 μM spermidine (SPD) for 4 days.A: representative Western blots for GTP-RhoA and total RhoA. In experiments dealing with manipulation of cytosolic free Ca2+ concentration ([Ca2+]cyt), DFMO-treated cells and cells treated with DFMO plus SPD were exposed to the Ca2+-free medium or 1 μM ionomycin (Iono) for 6 h. Whole cell lysates (500 μg) were incubated with 30 μg of GST-RBD beads and then subjected to electrophoresis on 15% acrylamide gel. GTP-RhoA (∼21 kDa) was identified by probing nitrocellulose with the specific antibody against RhoA. Total RhoA levels in whole cell lysates (40 μg/lane) were measured by standard Western blot analysis, and actin (∼45 kDa) immunoblotting was performed as an internal control for equal loading. B: quantitative ratio analysis of GTP-RhoA/total RhoA immunoblots by densitometry from cells described inA. Values are means ± SE from 3 separate experiments. *P < 0.05 compared with control; +P < 0.05 compared with DFMO+SPD.

Effect of induced RhoA by polyamines on ROK/ROCK activity.

It has been shown that ROK/ROCK proteins in epithelial cells are isoforms ROKα and ROCK-1 (2, 13, 20). To determine the potential downstream effector of RhoA during intestinal epithelial restitution, we tested the possibility that induced RhoA by polyamines via Ca2+ increased ROKα and ROCK-1 expression. As shown in Fig. 2, treatment of cells with DFMO decreased ROKα and ROCK-1 protein expression by ∼60% and decreased the Rho-kinase activity by ∼45%. This inhibitory effect of polyamine depletion on ROK/ROCK was completely prevented by the activation of RhoA through exogenous spermidine and elevation of [Ca2+]cyt concentration by ionomycin. Expression of ROKα and ROCK-1 proteins and the Rho-kinase enzyme activity in cells treated with DFMO plus either spermidine or ionomycin were similar to those observed in control cells. In contrast, decreased RhoA activity caused by exposure to the Ca2+-free medium prevented the restoration of ROKα and ROCK-1 levels and the Rho-kinase activity in cells treated with DFMO plus spermidine.

Fig. 2.

Changes in ROKα and ROCK-1 protein expression and the Rho-kinase enzyme activity in cells described in the Fig 1.A: representative Western blots for ROKα and ROCK-1. Whole cell lysates (40 μg/lane) were subjected to electrophoresis on 7.5% acrylamide gel, and ROKα (∼180 kDa) or ROCK-1 (∼160 kDa) proteins were identified by probing nitrocellulose with the specific antibody against ROKα or ROCK-1. After the blot was stripped, actin immunoblotting was performed as an internal control for equal loading. Three separate experiments showed similar results. B: ROK/ROCK activity in cells described in A. ROK/ROCK activity was measured in immunoprecipitates as described in materials and methods. The MLC20 served as the substrate, and the kinase assays were initiated by the addition of 10 μCi of [γ-32P]ATP and 20 μM ATP. The extent of phosphorylation was determined from the radioactivity on phosphocellulose disks by liquid scintillation. Values are means ± SE from 3 separate experiments. *P < 0.05 compared with control; +P < 0.05 compared with DFMO+SPD.

To further determine the relationship between RhoA and ROK/ROCK activation in intestinal epithelial cells, we examined the effect of inhibition of wild-type RhoA by ectopic expression of the dominant negative mutant RhoA (DNMRhoA) on ROK/ROCK in differentiated IEC-Cdx2L1 cells. We constructed the adenoviral vector containing theDNMRhoA cDNA under the control of the human cytomegalovirus immediate-early gene promoter. The reason we have chosen adenoviral vectors (rendered replication incompetent by deletion of E1 sequences) over other methods of transfection is that adenoviral vectors have been shown to infect a variety of cultured rat and human epithelial cells with nearly 100% efficiency. We demonstrated that >95% of IEC-Cdx2L1 cells were positive when they were infected with the adenoviral vector encoding GFP served as the marker for 24 h. As shown in Fig. 3, infection of differentiated IEC-Cdx2L1 cells with the AdDNMRhoA inhibited expression of the wild-type RhoA protein in amounts increasing with the viral load, probably through the process of negative feedback inhibition. Importantly, inhibition of the wild-type RhoA with the DNMRhoA significantly decreased ROKα protein expression and suppressed the Rho-kinase activity (Fig.3, A and B). Levels of ROKα protein and the Rho-kinase activity were decreased by ∼60% and ∼55% when the adenovirus at the concentration of 2 plaque-forming units per cell was used, respectively. Although the exact mechanism responsible for the reduction of ROKα in cells expressing DNMRhoA is unclear, it is possible that RhoA has a positive feedback to its downstream target protein ROKα and that the reduction of ROKα expression results from a decrease in this positive feedback stimulation following inhibition of wild-type RhoA expression. An adenovirus that lacked exogenous DNMRhoA cDNA was used as the negative control and did not alter levels of the wild-type RhoA, ROKα, and the Rho-kinase activity.

Fig. 3.

Effect of ectopic expression of the dominant negative mutant RhoA (DNMRhoA) on ROKα protein levels, ROK/ROCK activity, and cellular distribution of nonmuscle myosin II.A: representative Western blots for wild-type RhoA and ROKα. Differentiated IEC-Cdx2L1 cells were infected with the recombinant adenoviral vector encoding human DNMRhoA cDNA (AdDNMRhoA) or an adenovirus lacking DNMRhoA(AdNull) at a multiplicity of infection of 0.5–2 plaque-forming units per cell (pfu/cell). Levels of RhoA and ROKα proteins were analyzed 72 h after the infection. RhoA and ROKα proteins were identified with a specific antibody against RhoA or ROKα, and actin immunoblotting was performed as an internal control for equal loading. Three separate experiments showed similar results. B: ROK/ROCK enzyme activity from cells described in A. The ROK/ROCK enzyme activity was measured as described in materials and methods. Values are means ± SE from 3 separate experiments. *P < 0.05 compared with control.C: image of cellular distribution of nonmuscle myosin II in cells described in A: a, AdNull (2 pfu/cell);b, AdDNMRhoA (2 pfu/cell). Cells were fixed, permeabilized, and incubated with the specific antibody against nonmuscle myosin II antibody and then with anti IgG conjugated with FITC. Original magnification, ×1,000. Three separate experiments showed similar results.

Furthermore, inhibition of RhoA/Rho-kinase signaling by theDNMRhoA resulted in reorganization of cytoskeletal proteins. Nonmuscle myosin II stress fibers were sparse and devoid of long stress fiber formation in the cells infected with the AdDNMRhoA at the concentration of 2 plaque-forming units/cell (Fig. 3 Cb). In contrast, long stress fibers traversed through the cytoplasm, and a thick network of cortical myosin II fibers was just beneath the plasma membrane in cells that were infected with the AdNull at the same concentration (Fig. 3 Ca). Together, these results clearly indicate that activation of ROK/ROCK as a result of Ca2+-induced RhoA plays an important role in the control of cytoskeletal organization in intestinal epithelial cells.

Effect of ROK/ROCK activity on MLC phosphorylation.

To define the mechanism through which ROK/ROCK activation regulates cellular distribution of cytoskeleton, we examined the role of ROK/ROCK in MLC phosphorylation. Levels of MLC and P-MLC proteins were measured when ROK/ROCK activity was manipulated in differentiated IEC-Cdx2L1 cells. Consistent with the inhibitory effect on ROK/ROCK activity, polyamine depletion by DFMO decreased levels of MLC protein and phosphorylation of MLC (Fig. 4). Expression of MLC protein was decreased by ∼30% in DFMO-treated cells (Fig. 4 A), and levels of P-MLC were decreased by ∼45% (Fig. 4, B and C). Elevation of [Ca2+]cyt induced by ionomycin or exogenous spermidine given with DFMO not only reversed the inhibitory effect on polyamine depletion on MLC expression but also restored MLC phosphorylation to normal levels. There were no significant differences in P-MLC levels between controls and cells treated with DFMO plus spermidine or ionomycin. On the other hand, decreased [Ca2+]cyt by removal of extracellular Ca2+ from the culture medium prevented the restoration of P-MLC by spermidine in polyamine-deficient cells. Levels of P-MLC in cells treated with DFMO plus spermidine and then exposed to the Ca2+-free medium were similar to those observed in cells treated with DFMO alone. These data suggest that activation of RhoA/Rho-kinase signaling pathway following elevation of [Ca2+]cyt enhances MLC phosphorylation, thus resulting in actomyosin stress fiber formation.

Fig. 4.

Changes in myosin light chain (MLC) and phosphorylated MLC (P-MLC) in cells described in the Fig. 1. A: representative Western blots for MLC. Protein levels of MLC (∼20 kDa) were measured with a specific antibody against MLC. Equal amounts of proteins were monitored by reprobing the member with anti-actin antibody. Three separate experiments showed similar results.B: representative autoradiograms for P-MLC as measured by the [32P]orthophosphate assays described inmaterials and methods. Cells were incubated with 200 μCi/ml [32P]orthophosphate for 3 h, followed by the preparation of cell lysates. After cell lysates in the amount of 500 μg had been immunoprecipitated using 2 μg of the anti-MLC antibody, they were separated on SDS-PAGE gel and then followed by autoradiography. C: quantitative analysis of P-MLC autoradiograms by densitometry from cells described as in B. Values are means ± SE from 3 separate experiments. *P < 0.05 compared with control and DFMO+SPD; +P < 0.05 compared with DFMO alone.

Association of observed changes in RhoA/Rho-kinase activity and rates of cell migration.

Polyamine depletion by DFMO significantly inhibited cell migration (from 398 ± 8 to 93 ± 4 cells/mm, n = 6,P < 0.05) in differentiated IEC-Cdx2L1 cells, which was completely prevented by spermidine given with DFMO (from 93 ± 4 to 393 ± 10 cells/mm, n = 6, P< 0.05). Removal of extracellular Ca2+ from the culture medium blocked the restoration of cell migration by spermidine in DFMO-treated cells (from 393 ± 10 to 90 ± 3 cells/mm,n = 6, P < 0.05), whereas an increase in [Ca2+]cyt induced by ionomycin promoted cell migration in the absence of cellular polyamines (from 86 ± 3 to 138 ± 5 cells/mm, n = 6, P < 0.05). These results indicate that alterations in RhoA/Rho-kinase signaling are associated with changes in intestinal epithelial cell migration.

Effect of inhibition of ROK/ROCK activity by its specific inhibitor Y-27632 on cell migration.

To elucidate the role of ROK/ROCK activation in the process of polyamine-dependent cell migration during restitution, we carried out three complementary relevant experiments using the specific ROK/ROCK inhibitor Y-27632 (49). The first study was performed to confirm the specific inhibitory effect of Y-27632 on ROK/ROCK in differentiated IEC-Cdx2L1 cells. Figure5, A and B, shows that administration of Y-27632 at the concentration of 50 μM for 6 h not only decreased levels of ROKα protein but also inhibited the Rho-kinase activity in control cells and cells treated with DFMO plus spermidine. In contrast, treatment with Y-27632 at the same dose did not inhibit expression of MLCK. Consistently, inhibition of ROK/ROCK by Y-27632 was associated with a significant decrease in MLC phosphorylation (Fig. 5 C). These results clearly indicate that Y-27632 is a specific inhibitor for ROK/ROCK in intestinal epithelial cells.

Fig. 5.

Effect of the ROK/ROCK inhibitor Y27632 on expression of ROKα, the Rho-kinase activity, and P-MLC in differentiated IEC-Cdx2L1 cells. A: representative Western blots for ROKα and myosin light chain kinase (MLCK). After cells were grown in control culture and in culture containing DFMO+SPD for 4 days, they were exposed to Y27632 at the concentration of 50 μM for 6 h. Levels of ROKα protein were measured by Western blot analysis, while MLCK expression was examined by immunoprecipitation assays. Three separate experiments showed similar results. B: summarized data showing changes in the ROK/ROCK enzyme activity from cells described in A. The enzyme activity was measured in immunoprecipitates and the MLC20 served as the substrate. Values are means ± SE from 3 separate experiments. *P < 0.05 compared with control and DFMO+SPD. C: representative autoradiograms for P-MLC (left) and their quantitative analysis by densitometry (right). Controls cells were treated with 50 μM Y27632 for 6 h, and P-MLC was measured by using [32P]orthophosphate as described in materials and methods. Values are means ± SE from 3 separate experiments. *P < 0.05 compared with control.

The second study examined whether inhibition of ROK/ROCK by Y-27632 altered the rate of cell migration after wounding in control cells (without DFMO). Y-27632 was added immediately after wounding and was present during the period of cell migration (6 h). As shown in Fig.6 A, exposure to Y-27632 significantly decreased the rate of cell migration. When various doses of Y-27632 were tested, cell migration was inhibited dose dependently, with concentrations ranging from 25 to 75 μM. Maximum inhibition of cell migration occurred at 75 μM, where the rate of cell migration was decreased by ∼60%. There was no apparent loss of cell viability in cells exposed to Y-27632 (data not shown).

Fig. 6.

Effect of inhibition of ROK/ROCK activity on cell migration in controls and polyamine-deficient cells. A: summarized data showing the rates of cell migration in normal cells (without DFMO) when different concentrations of Y27632 were given immediately after wounding. Cell migration was assessed 6 h after treatments. Values are means ± SE from 6 dishes. *P < 0.05 compared with control (Con). B: summarized data showing the effect of Y27632 on cell migration in cells treated with DFMO+SPD. Cells were grown in the presence of DMEM containing DFMO or DFMO+SPD for 4 days. Various concentrations of Y27632 were given immediately after wounding, and cell migration was assayed 6 h after treatment. Values are means ± SE from 6 dishes. *P < 0.05 compared with control; +P < 0.05 compared with DFMO+SPD.

The third study examined the effect of inhibition of ROK/ROCK by Y-27632 on the restoration of cell migration by exogenous spermidine in polyamine-deficient cells. As shown in Fig. 6 B, the migration was significantly decreased in DFMO-treated cells. When spermidine was added concomitantly with DFMO, it was able to maintain cell migration at near-normal levels. Treatment with Y-27632 at different concentrations during the period of cell migration prevented restoration of cell migration by spermidine in DFMO-treated cells. When various doses of Y-27632 were given immediately after wounding, rates of cell migration were inhibited by ∼20%, ∼40%, and ∼55% at 25, 50, and 75 μM, respectively. Exposure of cells grown in the cultures containing DFMO plus spermidine to Y-27632 at the concentration of 75 μM for 6 h did not alter cell viability (data not shown). These results strongly suggest that activation of ROK/ROCK signaling following elevation of [Ca2+]cyt plays a major role in the stimulation of cell migration by polyamines after wounding.

Effect of inhibition of ROK/ROCK on cellular distribution of nonmuscle myosin II.

To investigate the mechanism by which ROK/ROCK activation mediates cell migration, we examined the effects of changes in ROK/ROCK activity on cellular distribution of nonmuscle myosin II in control and polyamine-deficient IEC-Cdx2L1 cells. In control migrating cells (Fig.7 A), there were long stress fibers that traversed the cytoplasm, and a thick network of cortical myosin II fibers was observed just inside the plasma membrane. Inhibition of ROK/ROCK activity by Y27632 at the concentration of 75 μM during the period of cell migration significantly inhibited the formation of myosin II stress fibers in control cells (without DFMO) (Fig. 7, A vs. B). Myosin II stress fibers were sparse and devoid of long stress fiber formation. Consistent with the inhibitory effect on ROK/ROCK, polyamine depletion by DFMO also inhibited formation of myosin stress fibers in IEC-Cdx2L1 cells (Fig.7, A vs. C), and features of myosin distribution were similar to those observed in control cells exposed to Y27632 (Fig. 7, B vs. C). On the other hand, restoration of ROK/ROCK activity by exogenous spermidine given together with DFMO reversed the inhibitory effect of polyamine depletion on myosin stress fiber formation. The distribution of nonmuscle myosin II in cells grown in the presence of DFMO plus spermidine was indistinguishable from that in control cells (Fig. 7,A vs. D). In contrast, inhibition of ROK/ROCK by either Y27632 or removal of extracellular Ca2+ after wounding completely prevented the restoration of the distribution of nonmuscle myosin II by exogenous spermidine in polyamine-deficient cells (Fig. 7, D vs. E and F). These results indicate that ROK/ROCK activation following elevation of [Ca2+]cyt induced by polyamines regulates cellular distribution of nonmuscle myosin II and enhances actomyosin stress fiber formation during restitution after wounding.

Fig. 7.

Effect of inhibition of ROK/ROCK by Y27632 on cellular distribution of nonmuscle myosin II in the presence or absence of polyamines. Differentiated IEC-Cdx2L1 cells were grown in control medium and medium containing DFMO alone or DFMO + SPD for 4 days and were then fixed after wounding. Cells were permeabilized and incubated with the antibody against nonmuscle myosin II and then with anti-IgG conjugated with FITC. A: 6 h after wounding in control cells. B: 6 h in control cells immediately exposed to Y27632 after wounding. C: 6 h after wounding in DFMO-treated cells. D: 6 h after wounding in cells treated with DFMO+SPD. E: 6 h in cells treated with DFMO+SPD for 4 days and then immediately exposed to Y27632 after wounding. F: 6 h in cells treated with DFMO+SPD for 4 days and then immediately exposed to the Ca2+-free medium after wounding. Original magnification, ×2,000. Three separate experiments showed similar results.

DISCUSSION

Precise regulation of epithelial restitution to reseal superficial wounding is critical for the maintenance of gastrointestinal mucosal integrity under physiological and pathological conditions. Cellular polyamines are known to be an important regulator of intestinal mucosal restitution, but specific functions of polyamines in epithelial cell migration are largely undefined. We (30, 54) have recently reported that polyamines increase RhoA activity by altering [Ca2+]cyt concentration through regulation of K+ channels and that depletion of cellular polyamines by DFMO inhibits K+ channel expression, reduces [Ca2+]cyt by downregulating the driving force for Ca2+ influx, decreases RhoA activity, and suppresses cell migration in undifferentiated parent IEC-6 cells. However, the exact mechanisms by which activated RhoA regulates epithelial cell migration during restitution remain to be demonstrated. In particular, whether downstream effectors of Ca2+-induced RhoA after activation of K+ channels by increased polyamines are important in this process is unclear. Because ROK/ROCK is identified as one of the downstream targets of RhoA and is involved in Rho-induced formation of actomyosin stress fibers and focal adhesions (13,25, 37), we have elucidated the role of this molecule in the regulation of polyamine-dependent intestinal epithelial cell migration after wounding.

To extend our previous findings, we employed differentiated IEC-Cdx2L1 cells as a model in the current study and have demonstrated that RhoA is also implicated in the signaling pathway of differentiated intestinal epithelial cell migration after wounding. Differentiated IEC-Cdx2L1 cells highly express Kv1.1 and Kv1.5 channels, which is associated with an increase in whole cell K+ currents, membrane hyperpolarization, and a rise in [Ca2+]cyt (32). The migration rates in differentiated IEC-Cdx2L1 cells are about four times those of parental IEC-6 cells. Basal levels of total RhoA and GTP-bound RhoA in differentiated IEC-Cdx2L1 cells are higher than those observed in parental IEC cells (data not shown). Consistent with the observations in parental IEC-6 cells (30, 34), the activation of RhoA also absolutely requires cellular polyamines in these differentiated epithelial cells. As shown in Fig. 1, reduction of [Ca2+]cyt by polyamine depletion with DFMO decreased levels of GTP-bound RhoA. Elevation of [Ca2+]cyt levels in polyamine-deficient cells by the Ca2+ ionophore ionomycin prevented the inhibitory effect of DFMO on RhoA activity. Although polyamines regulate intestinal epithelial cell migration through multiple signaling pathways (52, 53), the current findings indicate that the polyamine's action is mediated, at least partially, by [Ca2+]cyt.

The most important findings reported in this article are that Ca2+-induced RhoA following activation of K+channels by increased polyamines activates ROK/ROCK activity in differentiated IEC-Cdx2L1 cells. An increase in GTP-bound RhoA by elevation of [Ca2+]cyt induced ROKα and ROCK-1 protein expression and stimulated the Rho-kinase activity, whereas inhibition of RhoA by either reduced [Ca2+]cyt (Fig. 2) or a dominant negative mutant RhoA (Fig. 3) decreased ROK/ROCK activity. Although the precise mechanisms that account for activation of ROK/ROCK by Rho are unclear in general, GTP-bound RhoA has been shown to recognize at least two different types of target interfaces, the coiled coil domain of ROK/ROCK and the NH2-terminal regulatory domain of PKN (3, 20, 40, 55). Because GTP-bound RhoA slightly induces autophosphorylation of ROK/ROCK but dramatically increases the Rho-kinase activity (1), it is unlikely that GTP-bound RhoA stimulates ROK/ROCK activity by altering its autophosphorylation. The Ca2+-induced RhoA following activation of K+ channels by increased polyamines might directly interact with the COOH-terminal portion of the coiled domain of ROK/ROCK and result in a conformational change of ROK/ROCK, leading to activation of the Rho-kinase toward selective substrates such as MLC or the myosin binding subunit of myosin phosphatase. Further studies are clearly necessary to understand the exact mechanisms involved in the RhoA-induced ROK/ROCK activation in intestinal epithelial cells.

The observations from the current study further indicate that changes in ROK/ROCK activity alter MLC phosphorylation in differentiated IEC-Cdx2L1 cells. Inhibition of ROK/ROCK by either polyamine depletion with DFMO or treatment with its specific inhibitor Y27632 was associated with a significant decrease in levels of P-MLC, whereas the restoration of ROK/ROCK activity in DFMO-treated cells by exogenous spermidine or ionomycin returned P-MLC levels to near normal (Figs. 4and 5). MLC has been identified as a substrate of ROK/ROCK in muscle and nonmuscle cells, and activation of ROK/ROCK phosphorylates MLC at Ser-19 and Thr-18, the same sites that are phosphorylated by MLC kinase (2). It has been shown that phosphorylation of MLC at Ser-19 stimulates actin-activated myosin ATPase and plays an important role in the regulation of cellular motility, cellular contraction, and cytokinesis (2, 9, 14). When constitutively active forms of Rho or ROK/ROCK are introduced into nonmuscle cells, an enhancement of MLC phosphorylation is detected, which is associated with increases in the stress fiber formation and focal adhesions (2, 15). Because recent studies have demonstrated that ROK/ROCK also regulates MLC phosphorylation through the inactivation of myosin phosphatase, the exact contribution of these two pathways to the elevation of P-MLC following RhoA-induced ROK/ROCK activation by increased polyamines in intestinal epithelial cells is not clear at present.

The RhoA-induced ROK/ROCK activation plays a critical role in the process of intestinal epithelial cell migration during early epithelial restitution. Decreased ROK/ROCK activity by Y27632 not only decreased MLC phosphorylation (Fig. 5) but also inhibited normal cell migration after wounding (without DFMO) (Fig. 6 A). These findings are consistent with data from others (8, 45), who have found that ROK/ROCK activation is necessary for hepatic stellate cell migration after wounding and that expression of the dominant negative mutant ROK/ROCK inhibits wound-induced cell migration in Madin-Darby canine kidney (MDCK) epithelial cells. An interesting and extended finding obtained from the current study is that decreased ROK/ROCK activity by Y27632 also prevents the restoration of intestinal epithelial cell migration by exogenous spermidine in polyamine-deficient cells (Fig. 6 B). Our previous studies (30) have demonstrated that RhoA expression is regulated by cellular polyamines and that Ca2+-induced RhoA activation is necessary for polyamine-dependent intestinal epithelial cell migration. Together, the current observations and our previous findings (30-32, 54) strongly support the possibility that polyamines are required for the stimulation of cell migration after wounding in association with their ability to activate RhoA/Rho-kinase signaling pathway through control of [Ca2+]cyt concentration.

Results presented in Fig. 7 show that ROK/ROCK activation stimulates intestinal epithelial cell migration at least partially by altering cellular distribution of the cytoskeletal proteins. It is well known that cellular distribution and formation of cytoskeletal proteins are highly regulated by Rho and its downstream effectors in different types of cells (11, 25, 26). In the current study, decreases in ROK/ROCK activity by either treatment with Y27632 or exposure to the Ca2+-free medium in control cells and cells treated with DFMO plus spermidine resulted in reorganization of actomyosin filaments. The numbers of long stress fibers of myosin II decreased significantly, and in some cells they disappeared completely from cytoplasm, as observed in cells treated with DFMO alone (Fig. 7). Although the exact mechanisms by which inactivation of ROK/ROCK by polyamine depletion results in reorganization of myosin II are obscure, the MLC phosphorylation seems to play an important role in this process. Inhibition of ROK/ROCK activity was associated with a decrease in levels of P-MLC (Figs. 4 and 5), and the MLC phosphorylation has been shown to induce myosin-actin interaction in certain type of cells (2). It is possible that decreased MLC phosphorylation in polyamine-deficient cells reduces the interaction between myosin and actin, thereby leading to inhibition of stress fiber formation.

On the basis of current findings and our previous studies (30-32, 54), we propose a model delineating the role of ROK/ROCK activation following increased polyamines in the process of intestinal epithelial cell migration after wounding (Fig.8). In this model, increased polyamines enhance K+ channel expression, cause membrane hyperpolarization, raise [Ca2+]cytconcentration by enhancing the driving force for Ca2+influx, and increase RhoA activity, leading to ROK/ROCK activation. The resultant activation of ROK/ROCK increases MLC phosphorylation, resulting in stimulation of stress fiber formation and cell migration during restitution. In contrast, depletion of cellular polyamines inactivates ROK/ROCK by reducing [Ca2+]cytthrough downregulation of K+ channel activity, decreases MLC phosphorylation, and inhibits stress fiber formation, thus leading to inhibition of cell migration.

Fig. 8.

Schematic diagram depicting the proposed role of ROK/ROCK activation in the stimulation of intestinal epithelial cell migration by polyamines after wounding. Polyamines stimulate K+channel activity, cause membrane hyperpolarization, and raise [Ca2+]cyt by increasing the driving force for Ca2+ influx, leading to activation of RhoA. Increased GTP-bound RhoA can interact with ROK/ROCK and activate it. Activated ROK/ROCK increases MLC phosphorylation and actomyosin stress fiber formation, leading to stimulation of cell migration during restitution.

Acknowledgments

This work was supported by National Institutes of Health Grants DK-57819 and DK-61972 (to J.-Y. Wang), DK-28300 (to K. S. Murthy), and HL-54043 and HL-64945 (to J. X.-J. Yuan); by a Department of Veterans Affairs Merit Review Grant (to J.-Y. Wang); and by a Baltimore Research Education Foundation Pilot Grant (to J. N. Rao). J.-Y. Wang is a Research Career Scientist for the Department of Veterans Affairs Medical Research Service.

Footnotes

  • Address for reprint requests and other correspondence: J.-Y. Wang, Dept. of Surgery, Baltimore Veterans Affairs Medical Center, 10 North Greene St., Baltimore, MD 21201 (E-mail:jwang{at}smail.umaryland.edu).

  • 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.

  • First published December 4, 2002;10.1152/ajpcell.00371.2002

REFERENCES

View Abstract