Krüppel-like Factor 5 Controls Keratinocyte Migration via the Integrin-linked Kinase*

Migration of epithelial cells is critical for normal homeostasis in gut and skin, but the factors regulating this process are not completely understood. The zinc finger transcription factor Klf5 (IKLF; BTEB2) is highly expressed in proliferating cells of esophagus, skin, and other organs. We hypothesized that Klf5 regulates keratinocyte migration via the integrin-linked kinase (ILK), which, like Klf5, is localized to basal keratinocytes. We stably transduced mouse primary esophageal keratinocytes to overexpress Klf5 or small interfering RNA against Klf5. Klf5 overexpression in keratinocytes increased migration and correlated directly with ILK expression and activation. ILK expression restored migratory capacity in keratinocytes with suppression of Klf5, whereas ILK small interfering RNA blocked the increased migration resulting from Klf5 overexpression. By chromatin immunoprecipitation, electromobility shift assay, and luciferase reporter assays, we confirmed that ILK was a direct target for Klf5. In addition, Klf5 induced the activation of the ILK targets Cdc42 and myosin light chain, which are critical for cell migration and motility but not Rac1, AKT, or GSK3β. Overall, these results demonstrate that Klf5 is a key regulator of cell migration via ILK and provide new insight into the regulation of epithelial cell migration.

Spatial separation of proliferating and differentiating cells is a common theme throughout the epithelia of the luminal gastrointestinal tract, the skin, and several other organs (1)(2)(3). Migration of cells within these epithelia is critical for differentiation and normal homeostasis. A number of different epithelial types are found throughout the human body, of which the stratified squamous epithelium, which lines the esophagus, skin, oral cavity, and several other organs, is the most common. In stratified squamous epithelia, the basal layer is composed of both epithelial stem cells and transit-amplifying cells (4 -6). These transit-amplifying cells are partially committed, rapidly proliferating cells that originate from asymmetric divisions of the stem cells and then migrate, both within the basal layer and from the basal layer, toward the surface (7,8). While migrating within the basal layer, the transit-amplifying cells undergo several rounds of rapid proliferation and clonal expansion. Keratinocytes then undergo additional differentiation during migration through the suprabasal and superficial layers, losing their proliferative capacity.
A number of pathways are known to be critical for keratinocyte migration (reviewed in Ref. 9). Interactions of keratinocytes with the extracellular matrix are especially important for migration, and the integrins and the integrin-linked kinase (ILK) 2 play key roles in transducing signals from the extracellular matrix. ILK is an adaptor protein that couples integrins and growth factor receptors to a variety of downstream signaling events, influencing cell adhesion, proliferation, migration, differentiation, and survival (10 -12). ILK binds to the cytoplasmic tails of integrin ␤1 and ␤3 and forms a complex with a number of other proteins, including PINCH and members of the Parvin family (12)(13)(14). Through interactions with these and other proteins, ILK mediates the actions of a large number of downstream effectors, including Cdc42, Rac, myosin light chain (MLC 20 ), GSK-3␤, and AKT. Of note, activation of the small Rho GTPases Cdc42 and Rac is critical for cell migration and motility (11,15), and phosphorylation of MLC 20 , the 20-kDa regulatory light chain of myosin II, is important for cell motility and contractility (16,17). Nonetheless, the targets for ILK are clearly cell type-and tissue-specific; for example, the well established ILK targets GSK-3␤ and AKT are not ILK targets in chondrocytes (18). Thus, the role of ILK in regulating its downstream effectors must be understood within a specific context. In squamous epithelia, such as those of the skin and esophagus, ILK is localized to basal cells, and expression is lost as cells enter the suprabasal layer (19). Within keratinocytes, ILK function is critical and is required for normal skin morphogenesis (20).
Although many substrates for ILK have been identified (14), the transcriptional regulation of ILK is still not well established. Moreover, many of the key transcriptional regulators of keratinocyte migration have not been identified. In stratified squamous epithelia such as skin and esophagus, the zinc finger tran-scription factor Krüppel-like factor 5 (Klf5; IKLF; BTEB2), like ILK, localizes to basal keratinocytes (21). Members of the KLF family are linked to the regulation of cell migration (22), as well as numerous other key cellular processes (23). Klf5 was initially identified in proliferating crypt cells of the intestine (24). In the developing gut and epidermis, Klf5 is expressed beginning at embryonic day 10.5 and persists in the adult in cells within the proliferative compartments (25). In vascular cells, Klf5 mediates the response to injury, and heterozygous knock-out mice have abnormalities in cardiovascular remodeling in response to stress (26,27). Klf5 also regulates adipocyte differentiation (28). In keratinocytes, Klf5 transcriptionally regulates EGFR, activates MAPK signaling, and promotes rapid cell proliferation, such as that seen in transit-amplifying cells of the basal layer (21,29). Thus Klf5 exhibits many features necessary for the transcriptional control of normal epithelial homeostasis.
We hypothesized that in addition to promoting proliferation in transit-amplifying cells, Klf5 might also regulate migration of these cells, a critical step for keratinocyte differentiation. We further hypothesized that Klf5 might regulate keratinocyte migration via ILK. Here, using mouse primary esophageal keratinocytes as a model, we identify a critical positive regulatory role for Klf5 in keratinocyte migration. We also demonstrate that ILK is a transcriptional target of Klf5, that the effect of Klf5 on cell migration occurs via ILK, and that Cdc42 and MLC 20 are key specific mediators of the effects of Klf5 and ILK on keratinocyte migration. Overall, these studies provide important new insights into the regulation of epithelial cell migration and the mechanisms of normal epithelial homeostasis.

EXPERIMENTAL PROCEDURES
Cell Culture and Treatment-The isolation and culture of mouse primary esophageal keratinocytes have been described elsewhere (29), and all animal studies were approved by the Institutional Animal Care and Use Committee at the University of Pennsylvania. Once isolated, mouse esophageal epithelial cells were grown in keratinocyte serum-free medium (K-SFM, Invitrogen) supplemented with 40 g/ml bovine pituitary extract (Invitrogen), 1.0 ng/ml epidermal growth factor (EGF) (Invitrogen), 100 units/ml penicillin, and 100 g of streptomycin (Invitrogen). All experiments were begun when cells reached 70% confluence except as noted.
Retroviral Vectors and Infection-For retroviral expression of Klf5, we utilized full-length mouse Klf5 cDNA subcloned into the pFB-neo retroviral vector (Stratagene, La Jolla, CA), as described previously (29). To suppress Klf5 in mouse esophageal keratinocytes, we employed the pSR-siKlf5 retroviral construct (29), which generates double-stranded RNA directed against nucleotides 676 -694 of mouse Klf5. Empty vector served as control for Klf5 overexpression. For siRNA experiments, we generated a mismatch control (30), pSR-siKlf5⌬, which is identical to pSR-siKlf5 except for mutations in 6 of the 19 targeted nucleotides (ACATTATCATATTACTACC, mismatches underlined) and contains no similarities to other mouse sequences by BLAST analyses. The constructs were packaged in Phoenix-Ampho cells (Stanford University, Stanford, CA) by transfection with Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions, and mouse pri-mary keratinocytes were infected with culture supernatants from individual Phoenix-Ampho cells at a 1:6 dilution in K-SFM. Cells were passaged after 24 h and then selected with 300 g/ml G418 for 14 days.
Cell Migration and Adhesion Assays-Cell migration assays were performed as described previously (31). In brief, after culturing cells without EGF for 48 h, single cell suspensions containing 4 ϫ 10 4 cells/well in 0.5 ml of K-SFM (without EGF) were plated in triplicate into 24-well inserts (Falcon cell culture inserts, 8 m pore size). The lower chamber was filled with 0.5 ml of K-SFM containing 0.5 g/ml hydrocortisone (Sigma) and 10 ng/ml EGF (BD Biosciences). After incubation for 16 h at 37°C, the cells on the upper side of the transwell membrane were removed by cotton swab and rinsed with PBS. Cells migrating to the lower side of membrane were fixed in 4% paraformaldehyde for 20 min at room temperature, stained with crystal violet (Sigma), and counted. To perform migration assays in cells with ILK overexpression, we utilized an ILKexpressing adenovirus, Ad-ILK (gift of Dr. Gregory Hannigan, University of Toronto) or an empty vector control (32). Adenovirus was amplified in 293T cells and used to infect 1 ϫ 10 6 primary esophageal keratinocytes per well in each well of a 6-well plate (BD Biosciences) at a multiplicity of infection of 1:15. After 48 h of infection, migration assays were performed as described above. For studies of cell adhesion, 24-well plates coated with Matrigel (BD Biosciences) were seeded with 6 ϫ 10 4 cells/well in K-SFM and incubated for 2 h at 37°C. After washing three times with PBS to remove nonadherent cells, the remaining cells were dissociated with 0.05% trypsin/EDTA and counted. Data represented the average number of cells from triplicate wells.
Quantitative PCR-Total RNA was isolated with the RNeasy micro kit (Qiagen, Valencia, CA), and cDNA was synthesized with Superscript II reverse transcriptase (Invitrogen). Quantitative real time PCR was performed in triplicate on three samples for each experimental condition using an ABI Prism 7000 sequence detection system (Applied Biosystems) and SyBr Green PCR master mix (Applied Biosystems). TATA box-binding protein was used as the internal control. Primer sequences are available on request.
ILK Kinase Assay-ILK kinase assays were performed as described previously (32). In brief, equivalent protein concentrations of cell lysates were pre-cleared with nonspecific IgG bound to protein A-Sepharose, and supernatants were immunoprecipitated with 1.5 g of rabbit anti-ILK antibody (Millipore) overnight at 4°C with rotation. Protein A-Sepharose (Sigma), pre-swollen in Nonidet P-40 lysis buffer, was added for 2 h at 4°C to capture the antibodies. Following two washes with Nonidet P-40 lysis buffer and two washes with kinase wash buffer (10 mM MgCl 2 , 10 mM MnCl 2 , 50 mM HEPES, pH 7.5, 0.1 mM sodium orthovanadate, 1 mM dithiothreitol), kinase assays were performed directly on the protein A beads in a 25-l volume containing 10 mM MgCl 2 , 10 mM MnCl 2 , 50 mM HEPES, pH 7.5, 1 mM sodium orthovanadate, 2 mM sodium fluoride, 5 mCi of [␥-32 P]ATP (Amersham Biosciences), and 12.5 g of myelin basic protein as substrate (Millipore). The reaction was incubated for 30 min at 30°C and stopped with 10 l of SDS-PAGE nonreducing stop buffer and heated for 5 min at 95°C. Phosphorylated myelin basic protein bands were visualized by 10% SDS-PAGE and autoradiography with analysis of five independent experiments on a Storm 840 PhosphorImager (GE Healthcare).
Cdc42 and Rac1 Activation Assay-Cdc42 and Rac1 activation assays were performed with the Rac/Cdc42 assay reagent (Millipore) following the manufacturer's instructions. Briefly, cells were lysed with a magnesium-containing lysis buffer (MLB), and active complexes were immediately precipitated with 10 g of PAK-1 PBD using 300 g of protein lysate in a 1-ml total volume at 4°C for 60 min with rotation. PAK-1 PBD corresponds to the p21 binding domain (PBD, residues 67-150) of human PAK-1. After collecting and washing with MLB three times with pulsing centrifugation, agarose beads were resuspended in 20 l of Laemmli sample buffer and boiled for 5 min. Rac1-GTP and Cdc42-GTP were detected by Western blotting with Rac1 or Cdc42 antibodies as described above.
Chromatin Immunoprecipitation (ChIP) Assay-ChIP assays were performed in triplicate with the ChIP assay kit (Millipore) as described previously (29). Cells were treated with 1% formaldehyde for 10 min to cross-link associated protein to DNA, lysed, and sonicated. After a 10-fold dilution, the samples were pre-cleared with protein A-agarose/salmon sperm DNA for 30 min at 4°C and incubated overnight at 4°C with 1:500 anti-Klf5 or 1:500 anti-mouse IgG (Sigma), as a negative control. Cells were then precipitated with protein A-agarose for 1 h, heated at 65°C for 4 h, treated with proteinase K, and DNA-extracted with phenol/chloroform. Primers were designed to amplify the region from Ϫ322 to Ϫ82 and the region from Ϫ829 to Ϫ671 in the 5Ј regulatory region of the ILK gene, and PCR was performed with puReTaq Ready-to-Go PCR beads (Amersham Biosciences) for 25 cycles at 95°C for 1 min, 55°C for 1 min, and 72°C for 1 min. PCR products were separated on a 2% agarose gel and visualized by a Gel Doc XR system (Bio-Rad).
Electrophoretic Mobility Shift Assay-A double-stranded DNA probe with the sequence 5Ј-ATAGGAGCTGGCCGG-GCGGGCCGGGCGGGGCCGGGCGGCGGGCGCGGCC-CGGA-3Ј, which corresponded to a region of triple Klf5binding sites in the 5Ј regulatory region of ILK, was labeled with [ 32 P]dCTP, as was a similar probe containing mutations in 12 bases of the triple Klf5-binding sites (5Ј-ATAGGAGC-TGGTCGAGTGAGTCAGGTGAGACTGAGTGGCGGGC-GCGGCCCGGA-3Ј, mutated bases are underlined). Klf5 in vitro translated protein was made using the TNT Quick-Coupled transcription/translation kit (Promega), and 60,000 cpm of probe was incubated with 3.0 g of nuclear extract or differing amounts of in vitro translated Klf5 protein in a 25-l volume of binding buffer, containing 20 mM Tris-HCl, pH 7.2, 10 mM MgCl 2 , 150 mM NaCl, 20 M KCl, 5 M ZnCl 2 , 0.5 mM dithiothreitol, 5% glycerol, and 1 g of poly(dI-dC), for 20 min at 4°C. Supershifts were performed with 5 g of anti-Klf5 antibody per lane. The DNA-protein complexes were separated on a native 6% polyacrylamide gel and visualized on a Storm 840 PhosphorImager.
Cell Transfection and Reporter Assays-A DNA fragment corresponding to the 5Ј regulatory region of the mouse ILK gene from Ϫ938 to ϩ496, immediately upstream of the translational start site, was amplified by PCR and subcloned into a pGL3-Luc basic reporter vector (Promega). We constructed a reporter vector with mutation of 12 bases in the 5Ј regulatory region of ILK, between Ϫ240 and Ϫ206 (GGTCGAGTGAGT-CAGGTGAGACTGAGTGGCGGGC, mutated bases underlined), the region corresponding to the putative triple Klf5binding site, using the GeneEditor in vitro site-directed mutagenesis system (Promega). Cells were co-transfected with pIRES-Klf5-GFP, containing the complete Klf5 coding sequence in pIRES-GFP (Clontech) or with pIRES-GFP empty vector at 50% confluence in triplicate on 24-well plates using FuGENE 6 transfection reagent (Roche Applied Science). After 48 h, cells were lysed with Cell Lysis Buffer (Pharmingen), and luciferase reporter activity was analyzed using luciferase assay reagent (Promega) with a microtiter plate luminometer (Dynex Technologies). Luciferase activity was normalized to ␤-galactosidase and expressed as relative luciferase activity.

RESULTS
Klf5 Increases Keratinocyte Migration-Cell migration is important in such diverse processes as embryonic development, wound healing, and tumor progression and metastases (34) and in the maintenance of normal epithelial homeostasis (1). To evaluate the role of Klf5 in cell migration, we stably transduced mouse primary esophageal keratinocytes with retroviral vectors to overexpress Klf5 or to suppress Klf5 with

Klf5 Controls Keratinocyte Migration via ILK
siRNA (29). We quantitated the effects of Klf5 on keratinocyte migration using Transwell assays. Cells migrating across the membrane were stained purple ( Fig. 1A and supplemental Fig.  1B) and counted. As demonstrated in Fig. 1B and supplemental Fig. 1C, Klf5 overexpression increased keratinocyte migration across the membrane, whereas Klf5 suppression inhibited cell migration significantly. Notably, suppression of Klf5 did not significantly alter apoptosis in primary esophageal keratinocytes (supplemental Fig. 2).
The early stages of cell migration require adhesion and the formation of cell protrusions, the broad lamellipodia or the "spike-like" filopodia, which provide the traction sites for cells (34). To understand the effects of Klf5 on the processes of migration, we examined the formation of lamellipodia and filopodia. Grossly, cells with Klf5 overexpression appeared to have more of these cell protrusions, whereas keratinocytes with Klf5 inhibition by siRNA had fewer protrusions (Fig. 1C). As indicated in Fig. 1D, Klf5 expression increased the percentage of primary esophageal keratinocytes with lamellipodia and filopodia, whereas inhibition of Klf5 markedly decreased lamellipodia and filopodia formation. Adhesion of cells to the extracellular matrix, another key component of migration, is mediated in large part by integrins and integrin signaling (35). Overexpression of Klf5 resulted in a statistically significant increase in cell adhesion, whereas Klf5 suppression inhibited adhesion (Fig. 1E). Fig. 1F and supplemental Fig. 1A confirm the successful overexpression or suppression of Klf5 in these stably infected cell lines. Overall, these data support an important role for Klf5 in the regulation of keratinocyte migration.
Klf5 Up-regulates ILK Expression and Kinase Activity-As Klf5 is a DNA-binding transcriptional regulator (36), we sought to identify the mediators of the effects of Klf5 on keratinocyte migration. Signals from the extracellular matrix are critical for cell migration (37), and we hypothesized that Klf5 might impact integrin signaling, specifically by regulating the key adaptor protein ILK (10 -12). Following scratch wounding, Klf5 and ILK are both expressed in keratinocytes migrating across the wound, consistent for a role for both these proteins in keratinocyte migration (supplemental Fig. 3). In addition, both Klf5 and ILK are expressed in the basal layer of the esophagus in vivo (Fig. 2A). Cells of the basal layer include the stem cells and transit-amplifying cells, daughters of the stem cells that migrate within the basal layer, undergoing several rounds of rapid proliferation before migrating further, undergoing additional differentiation, and losing their proliferative capacity (6,38).
Utilizing mouse primary esophageal keratinocytes with overexpression of Klf5 or Klf5 suppression with siRNA, we examined whether Klf5 altered ILK mRNA and protein expression.  . C FIGURE 1. Klf5 increased keratinocyte migration and adhesion. A, Transwell assays demonstrated that Klf5 increased migration of mouse primary esophageal keratinocytes across the membrane. Cells migrating across the membrane were stained purple. B, overall, Klf5 overexpression increased keratinocyte migration 2-fold whereas Klf5 suppression with siRNA decreased migration by 70% (n ϭ 5). Results were expressed as mean Ϯ S.E. C, keratinocytes with overexpression of Klf5 appeared larger with more prominent cellular protrusions (arrows) compared with control cells. In contrast, keratinocytes with Klf5 suppression by siRNA appeared more compacted and had fewer protrusions. Cells were stained with phalloidin for F-actin. D, overexpression of Klf5 resulted in a 75% increase in the number of keratinocytes with filopodia or lamellipodia, whereas inhibition of Klf5 produced a 50% decrease in the number of keratinocytes with these protrusions, which are characteristic of migrating cells (n ϭ 5). Results were expressed as mean Ϯ S.E. E, Klf5 increased adherence of keratinocytes to Matrigel by more than 60%, and Klf5 suppression resulted in a 60% decrease in cell adherence. The number of adherent cells was expressed as a percentage of total cells (n ϭ 3), and results represent the mean Ϯ S.E. F, Western blots confirmed that retroviral infection of primary esophageal keratinocytes with pFB-Klf5 increased Klf5 expression, compared with pFB-neo control, whereas infection with pSR-siKlf5 decreased Klf5 expression, compared with pSR-siKlf5⌬ mismatch control. Student's t test was used for statistical analyses.
Quantitative real time PCR revealed that Klf5 overexpression increased ILK mRNA levels, whereas siRNA against Klf5 reduced ILK expression (Fig. 2B). By Western blot, overexpression of Klf5 increased ILK protein levels, whereas ILK protein was reduced by inhibition of Klf5 (Fig. 2C). Although some have argued that ILK is a "pseudokinase" (39), a number of studies support an important role for the kinase activity of ILK (40,41). To ensure that this increased ILK expression correlated with an increase in ILK kinase activity, we performed ILK kinase assays (32). A representative blot is shown in Fig. 2D. Overall, Klf5 overexpression produced a 2.5-fold increase in ILK kinase activity, whereas Klf5 inhibition yielded a 60% decrease in activity (Fig. 2E). Although we had shown previously that Klf5 upregulates MEK/ERK signaling (29), inhibition of MEK with the MEK inhibitor PD98059 did not alter ILK expression or ILK kinase activity (supplemental Fig. 4), indicating that the regulation of ILK by Klf5 was not mediated by MEK/ERK signaling.
ILK Is a Transcriptional Target for Klf5-To determine whether ILK was a direct transcriptional target for Klf5, we examined the 5Ј regulatory region of ILK for putative Klf5binding sites, using the computational program TESS (56). We identified a putative triple Klf5-binding site from Ϫ240 to Ϫ206 and hypothesized that Klf5 might bind to ILK in this region.
Using ChIP assays, we demonstrated binding of Klf5 to ILK in the vicinity of the triple Klf5 site (Fig. 3A). Next, to confirm that Klf5 bound to the region between Ϫ240 and Ϫ206, we performed electromobility shift assays (EMSA) with an oligonucleotide probe corresponding to the triple Klf5 site. By EMSA, Klf5 bound to the triple Klf5 site on ILK (Fig. 3B). Addition of increasing amounts of unlabeled probe competed away this binding, and the Klf5 band was shifted upward with anti-Klf5 antibody. Klf5 did not bind to a probe containing mutations in 12 bases of the triple Klf5 site, and inclusion of 200ϫ cold mutant probe did not compete away binding of Klf5 to labeled probe that did not contain the mutations (Fig. 3C). These results demonstrated the sequence specificity of Klf5 binding to this region of ILK.
To examine the transcriptional regulation of ILK by Klf5, we cloned a DNA fragment of ILK from Ϫ938 to ϩ496, immediately upstream of the translational start site, into the pGL3-Luc basic reporter vector to create the pILK-Luc reporter. We co-transfected mouse primary esophageal keratinocytes with pILK-Luc and with the bicistronic Klf5 expression plasmid pIRES-Klf5-GFP or the pIRES-GFP empty vector. As shown in Fig. 3D, expression of Klf5 increased ILK promoter activity by 4-fold. To determine whether Klf5 regulated ILK transcription via the Klf5-binding site identified on ChIP and EMSA, we created a mutant reporter construct, pILK-LucMT, containing mutations in the same 12 nucleotides within the triple Klf5-binding site of pILK-Luc as for the EMSA probe. When pIRES-Klf5-GFP was cotransfected with pILK-LucMT, the responsiveness of the reporter to Klf5 was lost. Moreover, basal ILK promoter activity, likely related to endogenous Klf5 expression in primary esophageal keratinocytes, was reduced by 50%. These results confirm that Klf5 transcriptionally regulates ILK by binding to the triple Klf5-binding site in the 5Ј regulatory region of ILK.
ILK Mediates the Effects of Klf5 on Keratinocyte Migration-To demonstrate that ILK was downstream of Klf5 and mediated the effects of Klf5 on cell migration, we infected primary esophageal keratinocytes stably infected with pSR-siKlf5⌬ or pSR-siKlf5 using an adenovirus that expresses ILK (Ad-ILK) or an empty vector control (32). Infection with Ad-ILK produced markedly increased expression of ILK, both in control cells and in cells with Klf5 suppression with siRNA (Fig. 4A). In contrast, infection with pSR-ILK, which generates siRNA against ILK, reduced ILK levels both in control and Klf5-overexpressing keratinocytes, compared with infection with pSR-siILK⌬ control ( Fig. 4B and supplemental Fig. 1D). As seen in Fig. 4, C and D, transwell assays revealed that ILK overexpression increased keratinocyte migration in control cells. Moreover, in cells with Klf5 suppression, ILK overexpression produced a 7-fold increase in migration of cells across the membrane, thus restoring migratory capacity in primary esophageal keratinocytes with suppression of Klf5. Next, to confirm these findings, we performed the opposite experiment, inhibiting ILK in primary esophageal keratinocytes with overexpression of Klf5. When we examined keratinocyte migration by transwell assay, we observed that ILK inhibition significantly reduced migration of keratinocytes with Klf5 overexpression ( Fig. 4E and supplemental Fig. 1, E and F). These results demonstrated that Klf5 regulated cell migration in primary esophageal keratinocytes via ILK.
Cdc42 and MLC 20 Are Targets for ILK in Keratinocytes-To identify the mechanism by which ILK regulated migration in primary esophageal keratinocytes, we investigated the expression of several known ILK targets. Akt and GSK3␤, as well as the small Rho GTPases Rac1 and Cdc42, which are critical for cell motility, contractility, and migration, have all been shown to be ILK targets in other cell types (40,42,43). Klf5 overexpression activated the ILK target Cdc42, and Klf5 inhibition decreased Cdc42 activation (Fig. 5, A and B), whereas activation of Rac1 was not affected. In addition, levels of phosphorylated Akt and GSK3␤ (Fig. 5, C and D) as well as total Cdc42, Rac1, Akt, and GSK3␤ were unchanged. Confirming that increases in Cdc42 upon induction of Klf5 were mediated by ILK, suppression of ILK with siRNA in Klf5-overexpressing cells markedly decreased Cdc42 activation (Fig. 5E). MLC 20 , which also plays a critical role in cell motility, has been identified both as a direct target for ILK (41) and as a downstream target of Cdc42 (44). MLC 20 phosphorylation at Thr 18 and Ser 19 was increased in keratinocytes with Klf5 overexpression and decreased by Klf5 suppression (Fig. 5F). Suppression of Cdc42 in Klf5-overexpressing keratinocytes produced a significant decrease in cell migration (supplemental Fig. 5). Thus, Cdc42 and MLC 20 were downstream targets of ILK in primary keratinocytes, providing a link between Klf5 and two critical regulators of cell migration.

DISCUSSION
Cell migration is a fundamental process for organogenesis during development, for normal homeostasis of adult tissues, and for tissue regeneration and repair after injury (34). In the past few years, enormous advances have been achieved in our understanding of the complexities and subtleties underlying the regulation of cell migration, which includes the spatio-temporal controls of cell adhesion, asymmetric polarization, and individual or layered cell motility. Using primary esoph-

T A T A T A T A A T A T
Protein -  . ILK was a direct transcriptional target for Klf5. A, 5Ј regulatory region of ILK contained a putative triple Klf5-binding site from Ϫ240 to Ϫ206. ChIP assays (n ϭ 3) demonstrated that Klf5 bound to ILK within the region Ϫ322 to Ϫ82, which includes the triple Klf5-binding site, but not to the region from Ϫ829 to Ϫ671, which was used as a control. Input was total DNA prior to immunoprecipitation. DNA was precipitated with anti-IgG antibody as an additional negative control. B, EMSA revealed binding of Klf5 (black arrows) to a probe corresponding to the triple Klf5-binding site from ILK. This binding could be competed away with increasing amounts of unlabeled probe. Upon addition of 5 g of anti-Klf5 antibody (Ab), a new band (white arrowhead) appeared at an apparent higher molecular mass, and the lower Klf5 band disappeared, confirming that this band represented Klf5 binding to the probe. C, by EMSA, Klf5 binding to the probe (black arrow) was not altered by the addition of 200ϫ mutant probe, which contained mutations in 12 bases within the triple Klf5-binding site from ILK (mutated bases are shown below the sequence from Ϫ240 to Ϫ206 in A). In addition, Klf5 did not bind to labeled mutant probe, confirming the sequence-specific binding of Klf5 to this region of ILK. D, cotransfection of mouse primary esophageal keratinocytes with the bicistronic Klf5 expression vector pIRES-Klf5-GFP (or pIRES-GFP control) and an ILK reporter, pILK-Luc, which contained the region Ϫ938 to ϩ496, revealed a 4-fold induction of ILK transcriptional activity with Klf5. No promoter activation was seen when a 12-nucleotide mutation (pILK-LucMT) was introduced in the triple Klf5-binding site of ILK (p ϭ 0.004 for mutant versus control, n ϭ 4). Results were normalized to ␤-galactosidase and expressed as mean of relative luciferase activity ϫ 10 4 Ϯ S.E. Student's t test was used for statistical analyses.
ageal keratinocytes, we now identify a role for the pro-proliferative Krüppel-like factor Klf5 as a regulator of cell migration. Epithelial stem cells often reside within a stem cell "niche," and migration of epithelial cells out of this niche, even within the proliferative compartment, is critical for differentiation into transit-amplifying cells (5,7). In stratified squamous epithelia, such as those of the skin and esophagus, proliferative cells in the basal layer periodically detach from the underlying basement membrane, move outward, undergo terminal differentiation, and eventually die (1). Loss of contact with the extracellular matrix contributes to the terminal differentiation of keratinocytes and is mediated by changes in integrin signaling (45). Thus, migration of keratinocytes within the basal layer and out of the basal layer is essential for proliferation and differentiation and for the maintenance of epithelial integrity (8,38,46). Perturbations of the processes of keratinocyte migration result in abnormal wound healing, as is seen in gastroesophageal reflux disease, which affects nearly one-half of the population of the United States (47), and esophageal cancer, the sixth most common cause of cancer deaths worldwide (48).
Here, we identify Klf5, which is expressed within the basal layer of the esophagus and skin, as a novel transcriptional acti-vator of ILK. Furthermore, we describe an important role for Klf5 in the regulation of normal keratinocyte migration via the adaptor protein ILK, a central mediator of signaling from the extracellular matrix via the integrins (11,14), and we identify Cdc42 and MLC 20 as downstream targets. Normally, ILK forms a complex at focal adhesions, binding to the cytoplasmic tail of integrin ␤1, as well as a number of other adaptor proteins, such as PINCH, paxillin, and the parvins CH-ILKBP/␣-parvin and affixin/␤parvin (14). Yet the role of ILK in the activation of its targets is controversial. Although a number of studies implicate the kinase activity of ILK in the phosphorylation of target proteins (40,41), others argue that ILK is a pseudokinase (39). Thus, it is not clear whether Cdc42 and MLC 20 are directly phosphorylated by ILK in keratinocytes. Other proteins, such as the focal adhesion kinase, are recruited to focal adhesions at the leading edge of migrating cells (49), and Rac and Cdc42 have been reported to be regulated by ILK via the guanine-nucleotide exchanger ␣-PIX (43). Thus it may be valuable in the future to identify additional intermediates for the activation of Cdc42 and MLC 20 by Klf5 in keratinocytes. Although the inhibition of both cell adhesion and cell migration by suppression of Klf5 may seem counterintuitive, these findings are consistent with the established role for adhesion and especially integrin-mediated adhesion in cell migration (34,35). Moreover, the cell adhesion assays that we performed measure adhesion of cells to the matrix, rather than cell-cell adhesion. Of note, similar findings of both decreased adhesion and decreased migratory capacity were seen in keratinocytes from ILK knock-out mice (20).
Overexpression of ILK restores migratory capacity in keratinocytes with suppression of Klf5, and siRNA against ILK decreases keratinocyte migration by 80% in cells with Klf5 overexpression, indicating that the primary means of regulation of keratinocyte migration by Klf5 is directly through ILK. Nonetheless, other factors might also be involved in the regulation of keratinocyte migration by Klf5 and ILK. For example, Klf5 may regulate other binding partners for ILK, such as the integrins, PINCH, ␣-PIX, or the parvins. Notably, the adaptor protein Nck-2 associates with PINCH and is regulated by EGF receptor signaling (50), providing a potential link between another Klf5 target, EGF receptor (29), and the ILK complex. In addition, other Rho GTPases, in addition to Cdc42, may contribute to the   processes of keratinocyte migration. In fact, Rho-associated kinase and RhoA have been implicated in ILK-mediated regulation of osteosarcoma cell spreading and motility (51). Activation of pathways such as these may be downstream or may occur in parallel to ILK signaling in keratinocytes.
We have previously identified Klf5 as part of a feedback loop critical for the induction of rapid keratinocyte proliferation (29). By identifying a role for Klf5 as an important mediator of keratinocyte migration, we demonstrate that Klf5 confers two essential properties of transit-amplifying cells. In contrast to its effects in primary esophageal keratinocytes, KLF5 inhibits proliferation and invasion in esophageal squamous cancer cells and also promotes anoikis (31). A context-dependent effect of KLF5 has also been described in intestinal cells (52) and has been reported for the KLF family member KLF4 (53), so this may be a common theme both for KLF5 and for other KLF family members.
Interestingly, ILK has been shown to suppress anoikis through Akt and to promote an invasive phenotype by up-regulation of MMP-9 via GSK3␤ and AP-1 (54,55). Thus, it will be interesting to examine the effects of Klf5 on the regulation of ILK, Cdc42, MLC 20 , and other ILK targets such as AKT and GSK3␤, in nontransformed versus transformed esophageal epithelial cells, as well as to determine the role of Klf5 on migration and proliferation in other epithelial cells.
In sum, we have identified a novel role for the Krüppel-like factor Klf5 in regulating keratinocyte migration via the integrin-linked kinase, ILK. Klf5 binds to and transcriptionally activates ILK, as shown by quantitative PCR, ChIP assays, EMSA, and promoter reporter constructs. In addition, we have identified Cdc42 and MLC 20 as specific targets for Klf5 via ILK in keratinocytes. These studies provide a means to understand the processes of cell migration in skin, esophagus, and other stratified squamous epithelia, as well as offering a potential mechanism for migration in other epithelia. Moreover, these data establish a framework for future investigations of altered migration in diseases such as gastroesophageal reflux disease and of tumor invasion and metastases in esophageal squamous cell cancers.  . MLC 20 and the small Rho GTPase Cdc42 were targets for ILK in keratinocytes. A, by Western blot, Cdc42 activation was increased by Klf5 expression in mouse primary esophageal keratinocytes and decreased by Klf5 inhibition, whereas activation of Rac1 was not changed. B, quantitation of assays performed in triplicate confirmed a statistically significant effect of Klf5 on Cdc42 but not on Rac1. Results were normalized to controls and expressed as mean Ϯ S.E. C, activation of Akt and GSK3␤ was also unchanged by overexpression or suppression of Klf5 in mouse primary esophageal keratinocytes. ␣-Tubulin served as a loading control. Notably, Rac1, Akt, and GSK3␤ have been identified as ILK targets in other contexts. D, quantitation of triplicate assays on Akt and GSK3␤ demonstrated no change in expression or activation. Results were normalized to controls and expressed as mean Ϯ S.E. E, suppression of ILK with siRNA blocked the activation of Cdc42 by Klf5 overexpression in keratinocytes, indicating that Klf5 activated Cdc42 via ILK. F, phosphorylation of MLC 20 , which plays a key role in cell motility, was increased by Klf5 overexpression and decreased by Klf5 suppression. ␣-Tubulin was a loading control. Student's t test was used for statistical analyses.