A Novel Role for Villin in Intestinal Epithelial Cell Survival and Homeostasis*

Apoptosis is a key regulator for the normal turnover of the intestinal mucosa, and abnormalities associated with this function have been linked to inflammatory bowel disease and colorectal cancer. Despite this, little is known about the mechanism(s) mediating intestinal epithelial cell apoptosis. Villin is an actin regulatory protein that is expressed in every cell of the intestinal epithelium as well as in exocrine glands associated with the gastrointestinal tract. In this study we demonstrate for the first time that villin is an epithelial cell-specific anti-apoptotic protein. Absence of villin predisposes mice to dextran sodium sulfate-induced colitis by promoting apoptosis. To better understand the cellular and molecular mechanisms of the anti-apoptotic function of villin, we overexpressed villin in the Madin-Darby canine kidney Tet-Off epithelial cell line to demonstrate that expression of villin protects cells from apoptosis by maintaining mitochondrial integrity thus inhibiting the activation of caspase-9 and caspase-3. Furthermore, we report that the anti-apoptotic response of villin depends on activation of the pro-survival proteins, phosphatidylinositol 3-kinase and phosphorylated Akt. The results of our studies shed new light on the previously unrecognized function of villin in the regulation of apoptosis in the gastrointestinal epithelium.

Apoptosis is a key regulator for the normal turnover of the intestinal mucosa, and abnormalities associated with this function have been linked to inflammatory bowel disease and colorectal cancer. Despite this, little is known about the mechanism(s) mediating intestinal epithelial cell apoptosis. Villin is an actin regulatory protein that is expressed in every cell of the intestinal epithelium as well as in exocrine glands associated with the gastrointestinal tract. In this study we demonstrate for the first time that villin is an epithelial cell-specific anti-apoptotic protein. Absence of villin predisposes mice to dextran sodium sulfate-induced colitis by promoting apoptosis. To better understand the cellular and molecular mechanisms of the anti-apoptotic function of villin, we overexpressed villin in the Madin-Darby canine kidney Tet-Off epithelial cell line to demonstrate that expression of villin protects cells from apoptosis by maintaining mitochondrial integrity thus inhibiting the activation of caspase-9 and caspase-3. Furthermore, we report that the anti-apoptotic response of villin depends on activation of the pro-survival proteins, phosphatidylinositol 3-kinase and phosphorylated Akt. The results of our studies shed new light on the previously unrecognized function of villin in the regulation of apoptosis in the gastrointestinal epithelium.
The homeostatic balance between proliferation and apoptosis is essential for the intestinal epithelium to function as a physiological and structural barrier. Intestinal epithelial cells have a high rate of cell turnover accompanied by an equally high rate of apoptosis (1). In human intestinal epithelium, it is estimated that ϳ10 10 cells are shed per day by a mechanism that involves apoptosis (2,3). This normal apoptosis is essential for the hierarchical organization of the intestinal epithelium, and apoptotic epithelial cells have been detected both at the base of the crypt as well as at the villus tips of the small and large intestine (3)(4)(5)(6). Defects in apoptosis are also associated with several gastrointestinal conditions, including villus atrophy, epithelial hyperplasia, loss of normal absorptive function, and increased risk of tumorigenesis. It is well documented that cells derived from colorectal cancer are more resistant to apoptosis, and transformation of colorectal epithelium to carcinoma is associated with progressive inhibition of apoptosis (4,5). A role for apoptosis in intestinal and renal epithelial cell injury is also well documented. In fact, an intestinal ischemia-reperfusion model is often used to study apoptosis in vivo (7,8).
Inflammatory bowel disease (IBD), 2 which represents ulcerative colitis (UC) and Crohn's disease (CD), is a progressive and relapsing disease whose etiology and pathogenesis remain unsolved. Disruption of epithelial barrier function is associated with the pathogenesis of IBD, although it is not clear whether this is a causal defect or a result of severe inflammation. Evidence from animal and human studies has suggested that increased apoptosis in the intestinal epithelium contributes to the tissue damage and the severity of colonic inflammatory response (6). Increased apoptosis and oxidative stress are also evident in colon of patients with UC and CD (9 -12). Furthermore, decreased colonocyte apoptosis is associated with increased mucosal healing during recovery from colitis (13). These studies suggest that strengthening the epithelial resistance to injurious stimuli and enhancement of intestinal repair could provide novel and effective approaches to treat IBD (14,15).
Villin is a major actin-modifying protein that is associated with the microvillar actin filaments and is expressed in most significant amounts in renal and gastrointestinal epithelial cells. We and others have previously demonstrated that villin regulates epithelial cell morphology, actin reorganization, and cell motility (16 -19). A functional abnormality in villin gene expression has been associated with patients with progressive cholestasis and hepatic failure (20). Patients with inflammatory gastrointestinal and renal diseases have been shown to have higher levels of serum villin as well as autoantibodies specific for villin (21). More importantly, a decrease in the levels of villin expression in enterocytes from CD and UC relative to healthy controls has also been reported (22). In this study, using immunoelectron microscopy and immunogold labeling, the authors demonstrated a significant reduction in villin-labeling density in the microvilli of patients with CD and UC. The authors suggested that decreased villin levels in CD and UC relative to healthy controls may be related to disturbances in differentiation and maturation process in the gastrointestinal epithelium of patients with CD and UC. Likewise, in other inflammatory diseases such as chronic pancreatitis decreased villin expression has been reported (23). In one previous study it has been reported that the death probability was twice as high in dextran sodium sulfate (DSS)-treated villin-null mice compared with their wild-type littermates (24), although no explanation was provided for this observation nor was the specific molecular mechanism identified. In this study we demonstrate for the first time that the villin-null mice have higher levels of apoptosis compared with their wild-type littermates that correlate with the severity of colitis induced in DSS-treated villin-null mice. Furthermore, in vitro experiments allowed us to characterize villin as an epithelial cell-specific anti-apoptotic protein.
Together, these studies identify villin as a new epithelial cellspecific anti-apoptotic gene and provide a molecular mechanism for the role of villin in regulating epithelial cellular plasticity related to cell injury.

EXPERIMENTAL PROCEDURES
Materials-Monoclonal antibody against villin was purchased from Transduction Laboratories; polyclonal antibodies against cleaved caspase-3 and VDAC were purchased from Santa Cruz Biotechnology and against cleaved PARP (Asp-214) and phospho-Akt from Cell Signaling. Antibody to cytochrome c was purchased from Pharmingen. The caspase-3 substrate Ac-DEVD-pNA, the caspase-9 substrate Ac-LEHD-pNA, the caspase-8 substrate Ac-IETD-pNA, the pan-caspase inhibitor Z-VAD-fmk, and the caspase-9 inhibitor Ac-LEHD-CHO were purchased from Biomol. In situ Cell Death Detection kit and the DNA fragmentation kit were purchased from Roche Applied Science. Hoechst 33258 was purchased from Sigma. Tetramethylrhodamine methyl ester (TMRM) and Alexa Fluor 555 were purchased from Molecular Probes. CellTiter Glo luminescent cell viability kit was purchased from Promega. HeLa cells stably transfected with cytochrome c-GFP were a kind gift from Dr. D. R. Green (St. Jude Children's Research Hospital, Memphis, TN). The villin knock-out mice were a generous gift from Dr. D. Gumucio (University of Michigan, Ann Arbor, MI).
Villin Knock-out Mice and DSS Treatment-Villin null mice were generated as described previously by injecting C57BL/6J blastocysts, and male offspring with Ͼ70% agouti coat color were bred to C57BL/6J female mice (25). Genotypes of pups were determined by PCR also as described previously (25). A 3% (w/v) solution of DSS (M r 40,000 -50,000) was prepared in water. DSS was administered either for 3 consecutive days or 7 consecutive days followed by euthanasia. Control mice received distilled water ad libitum. Because of the severity of colitis in the villin-null group, our study was limited to a maximum of 7 days of DSS treatment. Eight wild-type or villin-null mice were analyzed in each group. The damage and inflammation induced by DSS treatment were evaluated in a blinded fashion. Three sections were evaluated, each at a distance of ϳ100 m from the distal third of the colon. Mice were scored individually, and each score represented the mean of three sections. Colons were assigned a grade according to the percentage of colon affected with crypt damage as described before and as follows: grade 0 corresponds to normal epithelial morphology and no infiltrate; grade 1 corresponds to 1-25% of the colon affected with loss of the bottom fourth of the crypts and focal inflammatory cell infiltration; grade 2 corresponds to 26 -50% of the colon affected with loss of the bottom half of crypts and inflammatory cell infiltration; grade 3 corresponds to 51-75% of the colon affected with loss of the bottom two-thirds of the crypt and greater than 50% of the epithelium with significant inflammatory cell infiltration, gland dropout, and crypt abscess; grade 4 corresponds to Ͼ75% of the colon affected with loss of the entire crypt and surface epithelium (26 -29). Apoptosis was measured by counting TUNEL-positive nuclei as well as histologically in hematoxylin and eosin-stained sections of the colon in 100 epithelial cells per high power field, and a total of three fields were counted per section of mouse colon (i.e. cecum, proximal, and distal colon).
Cell Culture and Treatment-Full-length human villin was cloned in pTRE-HA as described previously (17,30). MDCK or HeLa Tet-Off cells were stably transfected with HA-tagged wild-type villin protein as described previously (16). To repress the expression of villin gene, cells were cultured in the presence of 10 ng/ml doxycycline. We have previously demonstrated the induction of villin following doxycycline withdrawal and shown the absence of villin in the MDCK and HeLa Tet-Off cells in the presence of doxycycline (16,17). In addition, IEC-6 cells were infected with recombinant adenovirus expressing full-length human villin protein as described previously (19). To induce cell death, cells were treated with camptothecin (20 M), staurosporine (1 M), a combination of TNF-␣ (10 ng/ml) and cycloheximide (10 ng/ml), or serum-starved for 72 h.
Apoptosis Assay-An enzyme-linked immunosorbent assay was used to quantitate DNA fragmentation in MDCK cells as described before (16). Cells were cultured in 12-well plates, lysed, and centrifuged to isolate nuclei. Nuclei-free supernatant was incubated with anti-histone biotin and anti-DNA peroxidase-conjugated antibody in 96-well streptavidin-coated plates for 2 h. 100 l of peroxidase substrate (2,2Ј-azino-di(3-ethylbenozothiazoline sulfonate) was added and incubated for an additional 10 min at room temperature. Absorbance was recorded at 405 nm using a microplate reader. DNA fragmentation was measured as absorbance units per mg of protein per min and expressed as fold increase in apoptotic cell death compared with cells expressing no villin (control). Cell morphologic analysis was done using bright field images. To assess DNA cleavage, hypotonic propidium iodide (PI) method was used as described elsewhere (31). The PI fluorescence intensity was measured using an LSRII flow cytometer (BD Biosciences) equipped with an Enterprise II dual argon/UV laser. Cell viability was also measured using the CellTiter-Glo luminescent cell viability assay by quantitation of ATP levels as a measure of metabolically active cells. An ATP standard curve was generated using 1 M ATP in culture medium.
The assay is based on the spectrophotometric detection of the chromophore pNA after cleavage from the labeled substrate p-nitroanilide (pNA) (Ac-DEVD-pNA for caspase-3, Ac-IETD-pNA for caspase-8, and Ac-LEHD-pNA for caspase-9). The light emission of pNA was quantified using a microplate reader (Bio-Rad) at 405 nm. To confirm that caspases were involved in camptothecin-induced cell death, caspase inhibition assay was performed using pan-caspase inhibitor, Z-VAD-fmk (50 M), and a caspase-9-specific inhibitor, Ac-LEHD-CHO (50 M), solubilized in Me 2 SO.
Measurement of Mitochondrial Membrane Potential-TMRM was used to obtain semi-quantitative estimates of changes in mitochondrial membrane potential (⌬⌿ m ) as described previously (32). Villin-null and villin-expressing MDCK cells cultured in chamber slides at a density of 3 ϫ 10 5 cells and treated without or with camptothecin (20 M) were incubated with TMRM (150 M, 10 min). Time lapse images were collected every 5 min for a total of 3 h using a confocal laser scanning microscope (LSM 5 PASCAL; Carl Zeiss, Thornwood, NY) with excitation at 558 nm and emission at 583 nm. Carbonyl cyanide-4-(trifluoromethoxy)-phenylhydrazone (FCCP, 1 M) was used as a positive control to depolarize mitochondria following camptothecin treatment. Loss of ⌬⌿ m was visualized as a reduction in the fluorescence signal.
Live Cell Imaging of Cytochrome c Release-HeLa cells were stably transfected with cytochrome c-GFP as described before (33). Cerulean-tagged villin was cloned as described previously (34). HeLa GFP-cytochrome c cells were transiently transfected with cerulean-tagged villin, treated with 1 M staurosporine for 0 -5 h, and live cell images captured at different time points (0 -5 h) using confocal laser scanning microscope and a 63ϫ oil objective (LSM 5 PASCAL; Carl Zeiss, Thornwood, NY). For GFP-cytochrome c an excitation wavelength of 488 nm and emission wavelength of 505 nm and for cerulean-villin an excitation wavelength of 458 nm and emission wavelength of 474 nm were used. Representative individual cell images were processed using MetaMorph software.
Preparation of Mitochondria-free Cytosolic Fractions-This assay was performed essentially as described by Waterhouse et al. (35). Cells were incubated for 10 min in 200 l of ice-cold buffer containing 250 mM sucrose, 80 mM KCl, and 200 g/ml digitonin. The cells were pelleted (1000 ϫ g for 10 min at 4°C), and the supernatant containing the cytosolic protein was obtained. The pellets were incubated in buffer containing 50 mM Tris/HCl, pH 7.4, 150 mM NaCl, 2 mM EDTA, 2 mM EGTA, 0.2% Triton X-100, 0.3% Nonidet P-40 and protease inhibitor mixture at 4°C for 10 min and centrifuged (10,000 ϫ g for 10 min at 4°C), and the supernatant containing the mitochondrial fraction was obtained. 45 g of protein from each sample was separated by 15% SDS-PAGE and subjected to Western blotting using antibodies against cytochrome c and VDAC. Equal loading was ensured by probing the same blot with monoclonal anti-actin antibody.
Immunohistochemical Staining-Mice were sacrificed, and the colon was removed and fixed in 3.7% formalin. Paraffinembedded sections of 5-m thickness were prepared, and hematoxylin and eosin staining was performed. Endogenous peroxidase was inactivated by 0.3% hydrogen peroxide in meth-anol for 30 min. TUNEL staining was performed on paraffin sections using an in situ cell death detection kit according to the manufacturer's instructions. A negative control lacking the TUNEL reaction mixture was included in each assay. For villin staining, paraffin-embedded tissues were also deparaffinized in xylene for 10 min, followed by 5 min each in serial dilutions of ethanol (100, 95, and 75%). After antigen retrieval, sections were incubated for 1 h at room temperature with anti-villin antibody at 1:1000 dilution. For the amplification of signals, Alexa Fluor 555 goat anti-mouse antibody was applied to the slides for 1 h at room temperature, followed by three washes in TBS (100 mM Tris-Cl, 150 mM NaCl, pH 8.0). Hoechst 33258 was used for nuclear staining. Negative control studies were done by incubating samples without the primary antibody.

Villin-null Mice Demonstrate Increased Epithelial Cell Apoptosis and Are More Sensitive to DSS-induced Injury-
In an effort to examine the in vivo function of villin in intestinal epithelial cell injury, we elected to treat villin-null mice and their wild-type littermates with established protocols to induce DSSinduced colitis, namely 3% DSS dissolved in sterile, distilled drinking water fed ad libitum. Control mice were fed sterile, distilled drinking water. This procedure is known to induce colonic epithelial injury in mice following five cycles of 7 days each of DSS treatment (36,37). Fig. 1A shows confocal analysis of immunohistochemical staining of ileal tissue from wild-type (WT) and villin Ϫ/Ϫ (VKO) mice with anti-villin antibody. Villin expression is noted in epithelial cells along the crypt-villi axis in the WT mice but not in the VKO mice. Before DSS treatment, both wild-type as well as villin-null mice had normal stools that were heme-negative (data not shown). Occult blood and rectal bleeding was first detected in some villin-null mice as early as 4 days post-treatment and in all villin-null mice by day 7. Rectal bleeding was never seen in the wild-type mice by day 7 of DSS treatment. Our studies were limited to a maximum of 7 days of DSS treatment because of the severe injury and rectal bleeding noted in the villin-null mice by day 7. In control mice, in the absence of any treatment, the morphology of the colon was very similar in both villin-null mice as well as their wild-type littermates (supplemental Fig. 1A). In villin-null mice receiving DSS, the most severe ulcerative lesions were consistently found in the distal colon and very often in the proximal colon (data not shown). In wild-type mice, the lesions were more restricted, confined only to the distal colon and were generally much less severe (Fig. 1B). It may be noted that more severe lesions were noted in the wild-type littermates following 13 consecutive days of DSS treatment, consistent with this model of injury (data not shown). However, the villin-null mice do not survive such long exposures to DSS. There was some loss of the epithelial layer and limited infiltration of inflammatory cells in the wild-type mice. In contrast, in the villin-null cells there was almost complete loss of the epithelium by day 7 of DSS treatment, disappearance of mucosal crypts, and significant increase in infiltration of inflammatory cells, including neutrophils and lymphocytes, thickening and severe edema of the muscularis, glandular atrophy, and fibrosis (Fig. 1B). The colons were also assigned a grade consistent with the percentage of colon affected with crypt damage as described under "Experimental Procedures." As shown in Fig. 1C most villin-null mice demonstrated a grade close to 3, indicating that ϳ75% of the crypts was damaged in these mice by day 7 of DSS treatment. Body weight changes (shown as percentage change from base-line body weight on day 0) are shown in Fig. 1D. There was significant weight loss between days 5 and 7 in villin-null mice compared with wild-type littermates following DSS treatment. The onset of rectal bleeding corresponded closely to the development of weight loss in villin-null mice. In contrast, the wild-type mice showed very limited clinical symptoms of UC on day 7.
We hypothesized that the absence of villin may result in increased apoptosis in DSS-treated mice and further that this increase in apoptosis possibly preceded the development of ulcerative lesions in the villin-null mice. To assess the levels of apoptosis in wild-type and villin-null mice after DSS treatment, we used the TUNEL assay to measure the response to DSS treatment. As shown in Fig. 2, villin-null mice showed a significant increase in apoptosis relative to littermate control wildtype mice at day 3 in mice receiving DSS. The absence of villin was associated with an enhanced DNA fragmentation as recorded by the TUNEL assay (Fig. 2, A and B; p Ͻ 0.01, n ϭ 5, denoted by arrowheads) as well as enhanced activation of caspase-3 in the epithelium following DSS treatment ( Fig. 2C; denoted by arrowheads). Apoptotic cells were also noted histologically in hematoxylin and eosin sections of proximal and distal colon of wild-type and villin-null mice and the data cor-FIGURE 1. Histological analysis of DSS-induced colitis shows increased severity of colitis in villin knock-out mice compared with wild-type littermates. A, immunohistochemistry of ileal tissue from WT and villin Ϫ/Ϫ (VKO) mice shows apical staining of epithelial cells along the crypt-villi axis in WT but not VKO mice. Villin staining is shown in red (Alexa Fluor 555), and nuclear staining is shown in blue (Hoechst 33258). Bar, 100 m. B, representative histological specimens of the colon from WT and VKO mice treated with 3% DSS for 7 days and stained with hematoxylin and eosin, n ϭ 8. Upper panel, bar, 500 m; lower panel, bar, 250 m. C, histological score for VKO and WT mice treated with distilled water or 3% DSS for 7 days. A score of 0 -4 was used for each animal as described under "Experimental Procedures." The values represent the average of five mice in each group. Asterisk denotes statistically significant values compared with wild-type mice, p Ͻ 0.01. D, age-, weight-, and sex-matched villin Ϫ/Ϫ (VKO) and WT mice were given 3% DSS for 7 days, and changes in body weight were recorded and expressed as the percentage change in basal body weight. A minimum of eight animals per group was studied. Asterisk denotes statistically significant values compared with wildtype mice, p Ͻ 0.01. related with the TUNEL assay. There was no significant difference in the number of apoptotic cells between villin-null and wild-type mice in untreated control animals (data not shown). This marked increase in apoptosis in villin-null mice occurred early in the course of DSS treatment (day 3). The increased apoptotic signal was localized largely to the surface epithelial cells, which include mostly differentiated epithelial cells. This is consistent with the features of human IBD where the gross abnormality is associated with mucosal inflammation and damage to the mucosal epithelial layer. It is noteworthy that the most significant levels of villin expression are also in the terminally differentiated surface epithelial cells of the small and large intestine, although villin is expressed in undifferentiated cells of the crypt as well. Apoptotic rates in the epithelial layer could not be determined accurately at the final time point (day 7) because of severe epithelial disintegration in the villin-null mice. These data demonstrate for the first time that in the absence of villin the colonic epithelium is sensitized to DSSinduced apoptosis and hence DSS-induced colitis.
Overexpression of Villin Delays Apoptosis in Epithelial Cells-To assess the role of villin in apoptosis, we elected to overexpress villin in MDCK Tet-Off cells. Doxycycline treatment was used to suppress villin expression but had no effect on cell apoptosis (supplemental Fig. 1B) (16). Cells were treated with camptothecin (20 M) and monitored 0 -14 h post-treatment. Camptothecin (CPT) is a topoisomerase I inhibitor that is widely used as an antitumor drug that induces DNA strand breaks and triggers apoptosis in postmitotic cells. In villin-expressing cells evidence of significant cell death was not appreciable up to 6 h post-treatment (Fig. 3A). The majority of villinnull cells were dead by 8 h of treatment, whereas the majority of villin-expressing cells were dead 14 h post-treatment. Fig. 3B shows expression of villin in MDCK Tet-Off cells cultured in the absence of doxycycline but not in cells cultured in the presence of doxycycline (10 ng/ml). Quantitative analysis showed a significant increase in the number of propidium iodide-positive villin-null cells as measured by flow cytometry but no significant change in the villin-expressing cells 5 h post-treatment (Fig. 3C). Quantitative analysis to measure cell viability by measuring total ATP levels, as shown in Fig. 3D, demonstrated that compared with villin-expressing cells, camptothecin treatment resulted in a significant decrease (46%, p Ͻ 0.01) in total cellular ATP levels and hence in the number of viable villin-null cells 6 h post-treatment. There was a similar significant decrease in viable villin-null cells at 4 and 5 h post-treatment (26 and 33%, p Ͻ 0.01, respectively).
To determine whether the decrease in viable cells was mediated by apoptosis, we elected to examine morphological changes in the nucleus. Hoechst 33258 staining showed a significant increase in fragmented and condensed nuclei characteristic of apoptotic cells in villin-null cells 5 h post-treatment (Fig. 3E). Genomic DNA fragmentation is another distinctive feature of caspase-dependent apoptosis. Treatment of villinnull cells with camptothecin at a concentration of 20 M for 5 h resulted in a significant increase in DNA fragmentation compared with villin-expressing cells (Fig. 3F).
In addition to camptothecin, villin-null and villin-expressing MDCK, HeLa, and IEC-6 cells were exposed to other apoptotic insults such as TNF-␣/cycloheximide (10 ng/ml each, 5 h), staurosporine (l M, 5 h), or serum-starved (for 72 h). As shown in Fig. 3G and supplemental Figs. 2 and 3, A and B, all four treatments induced significant apoptosis in villin-null relative to villin-expressing cells. Thus, the anti-apoptotic effects of villin are not cell type-specific or determined by the type of apoptotic stimuli. Because all treatments/toxic insults resulted in a similar response in villin-expressing cells, we elected to limit our study henceforth to the use of camptothecin and the renal epithelial MDCK cells for reasons described here. (i) Most gastrointestinal cell lines that express villin are transformed cell lines (Caco-2 and HT-29), and although they express villin, they are not appropriate for studies related to apoptosis because most of them are resistant to cell death. (ii) Down-regulation of villin in Caco-2 cells has been shown previously to alter brush border and cell morphology as well as targeting of apical membrane proteins, so short interfering RNA and similar studies are not possible with this or similar cell lines (38). (iii) Nontransformed gastrointestinal cell lines are crypt-like (IEC-6) rather than terminally differentiated epithelial cell lines. There are several reports that have clearly established that intestinal epithelial cell survival and death are subjected to the differentiation state-specific control mechanisms, and undifferentiated crypt cells exhibit a distinct susceptibility to apoptosis/anoikis from differentiated villus cells (39,40). (iv) We and others have previously reported that overexpression of villin and regulation of its function do not depend on the cell type (19,41); moreover, villin expressed in the renal epithelial cell line, MDCK, behaves functionally similar to villin in native mouse intestinal epithelium (41). (v) MDCK cells have the morphological characteristics of a differentiated enterocyte and are nontransformed "enterocyte-like" cells.
Villin Delays Apoptosis by Inhibiting Caspase-9 and Caspase-3-To characterize the anti-apoptotic function of villin, we analyzed the cleavage of PARP, a substrate of activated caspases in MDCK cells with or without villin expression following camptothecin treatment. PARP cleavage generated an 80-kDa fragment that was detected only in villin-null cell extracts 5 and 6 h post-treatment (Fig. 4A). Densitometric analysis gave a 2.18-fold increase of cleaved PARP between 5 and 6 h of CPT treatment in villin-null cells (p Ͻ 0.05, n ϭ 3). Likewise, we observed an increase in the activated caspase-3 fragment in villin-null but not villin-expressing cells 5 and 6 h post-treatment (Fig. 4A). Densitometric analysis gave a 2.41-fold increase of activated caspase-3 fragment between 5 and 6 h of CPT treatment in the villin-null cells (p Ͻ 0.05, n ϭ 3). These data strongly suggest that the caspase cascade is inhibited during camptothecin treatment thus delaying apoptosis in villin-expressing cells.
To further describe the mechanisms by which villin delays epithelial cell apoptosis, we sought to determine which specific caspase cascades are activated in the absence of villin. We measured the activity of caspase-8, caspase-9, and caspase-3 in MDCK cells expressing villin or not and treated without or with camptothecin. For these experiments, epithelial cells were exposed to camptothecin for varying times following which cells were lysed and analyzed for the cleavage products of caspase-8, caspase-9, and caspase-3. There was no significant difference in caspase-8 activation in villin-null and villin-expressing cells treated with camptothecin (Fig. 4B). In contrast, the mitochondrially activated initiator caspase-9 was significantly inhibited in the villinexpressing cells compared with villin-null cells following camptothecin treatment (Fig. 4C). Likewise, there was a significant inhibition of the executor caspase-3 in villin-expressing cells as early as 4 h posttreatment (Fig. 4D). These data were confirmed by treatment of villin-null cells with the pan-caspase inhibitor Z-VAD-fmk and the caspase-9 inhibitor Ac-LEHD-CHO, both of which prevented camptothecin-induced cell death ( Fig. 4E; supplemental Fig. 3C). These observations are consistent with alterations in mitochondrial membrane integrity and indicate that villin may regulate this cell function.
Overexpression of Villin Maintains Mitochondrial Integrity during Apoptosis-Mitochondrial dysfunction is an early event, preceding nuclear and plasma membrane alterations that directly regulate activation of caspase-9. It is characterized by an increase in mitochondrial membrane permeability and loss of membrane potential. The release of cytochrome c has been linked to loss of mitochondrial transmembrane potential (⌬⌿ m ) and is controlled by the pro-and anti-apoptotic members of the Bcl-2 family of proteins. As an independent assessment of alterations of mitochondrial integrity, we assessed changes in mitochondrial permeability as well as cytochrome c release into the cytosol from villinnull and villin-expressing MDCK and HeLa cells following camptothecin and staurosporine treatment.
In these experiments MDCK villinnull and villin-expressing cells were labeled with the potentiometric probe TMRM, which accumulates within polarized mitochondria (28). At low, nonquenching concentrations of the dye (150 M), depolarization of ⌬⌿ m is accompanied by both a decrease in fluorescence and a loss of punctate localization of the dye to the mitochondria. Time-lapse images were collected between 0 and 3 h following camptothecin treatment (Fig. 5A; supplemental videos 1 and 2). From these studies, it can be seen that the addition of camptothecin was followed by significant depolarization of the mitochondria in villin-null but not villin-expressing cells. Furthermore, after 3 h of camptothecin exposure, the mitochondria in villin-null cells were not responsive to FCCP (1 M), whereas the villin-expressing cells were still responsive to depolarization by FCCP (data not shown). We quantified mitochondrial transmembrane potential release from the mitochondria by measuring the fluorescence intensity, which showed significant mitochondrial depolarization in villin-null cells compared with villin-expressing cells at 1.5 and 3 h post-treatment (p Ͻ 0.01, n ϭ 25).
To confirm these observations, we elected to use live cell imaging with HeLa cells stably transfected with GFP-tagged cytochrome c and transiently transfected with cerulean-tagged villin. Cells were treated with staurosporine (1 M; this choice was based on a more significant effect of staurosporine compared with camptothecin on HeLa cell survival (see Fig. 3G)) and imaged in real time (Fig. 5B). Please note, Fig. 5B, upper panel, shows HeLa cells transiently transfected with cerulean-villin, whereas the lower panel shows the expression of GFP-cytochrome c in the same field in both villin-transfected and -untransfected HeLa cells. Treatment of villin-null cells with staurosporine resulted in the release of cytochrome c from the mitochondria, which was seen as a change from the punctate mitochondria to a more diffuse cytoplasmic signal (Fig. 5B,  white arrowheads). This change was seen as early as 2 h post-treatment. In contrast, cells expressing villin maintained the punctate mitochondrial staining of cytochrome c up to 5 h post-treatment. As noted earlier, HeLa cells overexpressing villin behave like MDCK cells expressing villin (Fig. 3G). Epithelial monolayers (MDCK cells) were also treated with camptothecin (20 M), and the amount of cytochrome c present in mitochondria-depleted cytosol was assessed using Western blot analysis (Fig. 5C). The presence of cytoplasmic cytochrome c in villin-null cells was significantly higher than villin-expressing cells 5 h post-treatment. In control samples, purified mitochondrial fractions contained both cytochrome c as well as VDAC. The enhanced cytosolic cytochrome c suggests increased mitochondrial release and availability for activation of procaspase-9.
To elucidate the signaling events upstream of villin-induced inhibition of cytochrome c release, we investigated the Bcl-2 family members that have been implicated in the mitochondrial regulation of apoptosis. Bax is a proapoptotic protein that induces cytochrome c release (42,43),although Bcl-2 and Bcl-xL regulate apoptosis by inhibiting the release of cytochrome c from mitochondria (44). The assessment of both proapoptotic and antiapoptotic cellular proteins showed no significant change in Bcl-xL, but a significant decrease (2-fold) in Bcl-2 expression in villin-null cells compared with villin-expressing cells 5 h post-treatment ( Fig. 5D; p Ͻ 0.05, n ϭ 3). In contrast, there was a significant increase (2-fold) in Bax expression in villin-null cells 5 h post-treatment compared with villinexpressing cells (Fig. 5D; p Ͻ 0.05, n ϭ 3).
Activation of PI 3-Kinase/Akt Is Required for Anti-apoptotic Function of Villin-To better elucidate the mode of action of villin during the induction of apoptosis, we investigated the involvement of PI 3-kinase and Akt, the pro-survival proteins. This was based on the assumption that villin, which is a phosphatidylinositol 4,5-bisphosphate-binding protein, could regulate PI 3-kinase activity. For these studies, we pretreated villin expressing MDCK cells with specific PI 3-kinase inhibitors LY-294002 (10 nM for 1 h) or wortmannin (1 M for 1 h), before incubation with camptothecin (20 M, 6 h). Pretreatment of cells with the PI 3-kinase inhibitors abolished the protective effect of villin in cells exposed to camptothecin (Fig. 6A). Analysis of the percentage of apoptotic cells by Hoechst staining (Fig.  6A) and propidium iodide staining (Fig. 6B) showed camptothecin-induced apoptosis in ϳ70% of the villin-expressing cells pretreated with either LY-294002 or wortmannin. LY-294002 and wortmannin by themselves had no effect on apoptosis at the concentrations used in this study (supplemental Fig. 3D). Thus, both PI 3-kinase inhibitors completely blocked the protective effects of villin. Other reagents were used to examine if villin regulates the ATP/ADP homeostasis during apoptosis or uncoupling of actin from Ras-cAMP signaling. None of these had any effect on the anti-apoptotic function of villin in vitro (data not shown).
One of the main targets of PI 3-kinase is the serine-threonine kinase Akt, which has been reported to mediate cell survival in a wide range of cell types (45). Akt has been shown to phosphorylate several cellular proteins in vivo that are known to play an important role in the induction of apoptosis, including activation of caspase-9 (46). Two serine-threonine sites (Thr-308 in the catalytic domain and Ser-473 in the COOH terminus) are phospho-rylatedtoactivateAktinaPI3-kinasedependent manner. We investigated whether the Akt signaling pathway was involved in villininduced delay in epithelial cell apoptosis, by examining the activation of Akt phosphorylation at Ser-473. For these studies villin-expressing cells were either pretreated or not with the PI 3-kinase inhibitors LY-294002 or wortmannin followed by camptothecin treatment (20 M, 5 h). Camptothecin treatment in villin-expressing cells resulted in significant increase in phosphorylation of Akt compared with untreated villin-expressing cells (Fig. 6C). In control cells in the absence of camptothecin treatment, no Akt phosphorylation was detected. Likewise, in villin-null cells treated with camptothecin, no Akt phosphorylation was detected (supplemental Fig. 3E). Pretreatment of villin-expressing cells with LY-294002 or wortmannin completely abolished the phosphorylation of Akt in response to camptothecin treatment, suggesting that PI 3-kinase signaling and Akt phosphorylation are critical for villin-induced delay in apoptosis.

DISCUSSION
DSS-induced colitis resembles human UC in several important clinical and histopathological features (37,47). Using homozygous villin-null mice, we show unequivocally that the histopathological and clinical changes induced in vivo following DSS treatment were dependent on villin expression. By following the development of the ulcerative lesions over time, throughout the early phases of DSS-induced colitis, we were able to demonstrate that a sharp and early increase in apoptosis preceded the development of severe mucosal damage in villinnull mice. Our data indicate that the degree of apoptosis early in the DSS treatment in the villin-null mice correlated with the severity of colitis subsequently. These data agree with the observations in UC patients that increasing rates of colonic epithelial apoptosis correlate with increasing severity of the disease (11). Our study also potentially provides an explanation for the higher death probability noted previously in DSS-treated villin knock-out mice compared with their wild-type littermates (24). No differences in the expression of brush border or tight junction proteins have been noted in the villin-null mice (24,25). No significant changes in the other actin-binding proteins were noted either (24,25). Hence, changes in the permeability or junctional complexes are unlikely to contribute to the severity of colitis in villin-null mice. It is also noteworthy that the villin-null mice provide a model with an epithelial cell-specific defect that could be used to dissect the specific contributions of the mucosal immune defect versus the epithelial defect in the pathogenesis of IBD. Thus, we would suggest that another major finding of our inquiry is the identification of an animal model that may prove to be an important tool for investigating the mechanisms of IBD pathogenesis and for detection of potential therapeutic agents to treat this disease.
The current thinking about IBD identifies three principal factors that influence the pathogenesis of this rather complex disease, namely bacterial infection, a hyper-immune response, and a persistent epithelial defect. Although the first two components of this hypothesis have received a lot of attention, in comparison relatively little is known about the role of the epithelium in the development of IBD. Our study confirms previous suggestions that epithelial defects play a major role in the pathogenesis of IBD. Because many clinical conditions that affect the intestinal surface are manifested by recurrent erosions and persistent epithelial defects, there are compelling reasons to understand endogenous mechanisms that sustain mucosal integrity, modulate vulnerability to cytotoxic injury, and modulate the repair response. We suggest that a better knowledge of the complex effects of villin on epithelial cells will allow the development of new therapeutic strategies to maintain a healthy intestinal epithelium. It has been suggested that it is not only the amount or type of damage that determines whether cells will undergo apoptosis, but also the patterns of gene expression that determine a so-called "survival threshold." This survival threshold integrates both cell survival as well as cell damage signals, and the balance between these two opposing features resolves the ultimate decision about cell fate. Our study suggests that villin may be one of several pro-survival genes in intestinal epithelial cells that determine cell fate during injury.
The finding that Bax expression increased and Bcl-2 expression was decreased in villin-null cells suggests that Bcl-2 family members could also be one of the mediators of the observed decrease in epithelial cell apoptosis in the mouse colitis model. It may be noted that homozygous Bcl-2-null mice demonstrate elevated levels of spontaneous apoptosis in the colon (48). Bcl-2 antagonist genes have also been reported to be up-regulated in enterocytes from patients with UC (49). In this study we also demonstrate that epithelial injury during acute inflammation may result from the dysregulation of the Akt-dependent cellular survival pathways. Consistent with our data, Akt has been shown to regulate cell survival and apoptosis in human colonocytes (13).
Most recent studies suggest that the mitochondria is an important cell survival switch. By maintaining membrane permeability, the mitochondria also decides the response of the FIGURE 6. Anti-apoptotic function of villin is regulated by PI 3-kinase and Akt. Villin-expressing MDCK cells were pretreated (1 h) with PI 3-kinase inhibitors LY-294002 (10 nM) or wortmannin (1 M) followed by CPT treatment (20 M, 5 h). A, apoptosis was determined by phase-contrast microscopy and by staining with Hoechst 33258. Bar, 50 m. B, cell viability was measured by PI staining and analysis of positively stained cells using flow cytometry. C, villinexpressing MDCK cells untreated or pretreated with LY-294002 or wortmannin were exposed to CPT (20 M, 5h) or not, and cell extracts were immunoblotted for phospho-Akt (pAkt). Actin staining was done in parallel for quantitative analysis. This Western blot is representative of three other experiments with similar results. IB, immunoblot.
cell to a multitude of physiological and genetic stresses. Our data support the concept that villin provides protection against apoptosis by maintaining the mitochondrial integrity, a possibility confirmed by direct measurement of mitochondrial membrane potential using TMRM and by measuring cytochrome c translocation from the mitochondria to the cytosol in both living cells as well as in cell extracts in response to camptothecin or staurosporine exposure. Gelsolin is a related protein of the villin superfamily and the best studied regulator of actin that also has a role in apoptosis. Although gelsolin is ubiquitously expressed, low levels of gelsolin are detected in differentiated intestinal and renal epithelial cells. Like villin, gelsolin is believed to regulate the mitochondrial integrity; however, the molecular mechanism of gelsolin's anti-apoptotic function are controversial. Gelsolin has been shown to localize to the mitochondria and to interact with the VDAC thus preventing loss of mitochondrial permeability (50,51). However, a recent study by Baines et al. (52) clearly established that VDACs are not required for mitochondria-dependent cell death. Unlike its homologous partner gelsolin, villin does not target to the mitochondria. 3 Also unlike gelsolin, villin can assemble as well as disassemble actin filaments. Villin is also a phospholipid-binding protein that binds phosphatidylinositol 4,5-bisphosphate with very high affinity (53). We suggest that these properties of villin could determine its anti-apoptotic function.
Apoptosis is accompanied by a dramatic reorganization of the actin cytoskeleton. Some of the most comprehensible data supporting links between actin and apoptosis pathways come from studies using drugs that affect actin turnover. Actin and actin-binding proteins are also substrates of activated caspases, although it remains to be established if such changes are important in the progression of the apoptotic process. Most of these studies fail to provide a causal relationship between actin reorganization and cell death. Our study demonstrates that loss of an actin-binding protein, more importantly absence of a gastrointestinal epithelium specific gene, results in increased apoptosis both in vitro and in vivo. Actin filaments can regulate mitochondrial structure and function during apoptosis by regulating mitochondrial motility, mitochondrial morphology, and/or transport of pro-apoptotic cargo to the mitochondria (54 -59). The relation between F-actin and mitochondrial membrane potential has been elucidated most convincingly in studies with yeast (60). Although the evidence from yeast studies shows that changes in actin as well as stabilization of actin can induce cell death that is mediated through changes in the mitochondria, the case in mammalian cells is less clear, and some of the data present a conflicting picture. Although open questions remain as to the precise mechanism(s) by which alterations in actin dynamics regulate apoptosis in higher eukaryotes, the present work confirms that a physiological link exists between actin regulation and cell death, providing new leads for the understanding of apoptosis and actin remodeling in higher eukaryotes. Our studies reinforce and expand our appreciation of the diversity of functions controlled by villin and the actin cytoskeleton to include epithelial cell survival.
Together, our findings identify the actin cytoskeleton as a target to protect from apoptosis and thus may have implications for therapeutic approaches to digestive and renal diseases.