α6β4 Integrin, a Master Regulator of Expression of Integrins in Human Keratinocytes*

Background: Keratinocyte migration involves the coordinated expression of various integrin heterodimers. Results: Loss of α6β4 integrin expression impairs cell migration and decreases α2 and α3 integrin subunit expression via transcriptional and translational mechanisms. Conclusion: Migration of human keratinocytes requires α6β4 integrin-dependent regulation of integrin subunit expression. Significance: α6β4 integrin controls integrin expression profiles and thereby regulates migration. Three major laminin and collagen-binding integrins in skin (α6β4, α3β1, and α2β1) are involved in keratinocyte adhesion to the dermis and dissemination of skin cells during wound healing and/or tumorigenesis. Knockdown of α6 integrin in keratinocytes not only results in motility defects but also leads to decreased surface expression of the α2, α3, and β4 integrin subunits. Whereas α2 integrin mRNA levels are decreased in α6 integrin knockdown cells, α3 and β4 integrin mRNAs levels are unaffected. Expression of either α6 or α3 integrin in α6 integrin knockdown cells restores α2 integrin mRNA levels. Moreover, re-expression of α6 integrin increases β4 integrin protein at the cell surface, which results in an increase in α3 integrin expression via activation of initiation factor 4E-binding protein 1. Our data indicate that the α6β4 integrin is a master regulator of transcription and translation of other integrin subunits and underscore its pivotal role in wound healing and cancer.

Three major laminin and collagen-binding integrins in skin (␣6␤4, ␣3␤1, and ␣2␤1) are involved in keratinocyte adhesion to the dermis and dissemination of skin cells during wound healing and/or tumorigenesis. Knockdown of ␣6 integrin in keratinocytes not only results in motility defects but also leads to decreased surface expression of the ␣2, ␣3, and ␤4 integrin subunits. Whereas ␣2 integrin mRNA levels are decreased in ␣6 integrin knockdown cells, ␣3 and ␤4 integrin mRNAs levels are unaffected. Expression of either ␣6 or ␣3 integrin in ␣6 integrin knockdown cells restores ␣2 integrin mRNA levels. Moreover, re-expression of ␣6 integrin increases ␤4 integrin protein at the cell surface, which results in an increase in ␣3 integrin expression via activation of initiation factor 4E-binding protein 1. Our data indicate that the ␣6␤4 integrin is a master regulator of transcription and translation of other integrin subunits and underscore its pivotal role in wound healing and cancer.
In intact skin, ␣6␤4 integrin mediates the interaction of basal keratinocytes to laminin-332 located at the interface of the epidermis and dermis. The importance of ␣6␤4 integrin in maintaining skin integrity has been demonstrated in mouse knockout models and human studies. In mice, the loss of ␣6␤4 integrin leads to separation of the epidermal from the dermal layers of the skin at the site of the basement membrane zone, skin blistering, and post-natal lethality (1)(2)(3)(4)(5). In humans, mutations in either the ␣6 or ␤4 integrin genes is the cause of junctional epidermolysis bullosa (JEB), 2 characterized by skin blistering and pyloric atresia (6 -14).
In addition to its role in stable adhesion, ␣6␤4 integrin is up-regulated during wound healing and tumorigenesis where it determines skin cell motile behavior. Several other integrins enriched in the basal cells of the epidermis, including ␣3␤1 integrin, a receptor for laminin-332, and ␣2␤1 integrin, a receptor for collagen I, have also been implicated in regulating wound healing of the skin and/or cell migration during tumorigenesis (15)(16)(17). Specifically, ␣3␤1 integrin enhances matrix proteolysis during tumorigenesis and, possibly, wound healing (18). Moreover, blocking ␣3␤1 integrin function inhibits skin cell migration in vitro (19). In the case of ␣2␤1 integrin, the story is more complex. There is evidence that ␣2␤1 integrin regulates cell migration by promoting matrix proteolysis (20). In contrast, in the complete absence of ␣2␤1 integrin, tumor metastasis is enhanced, most likely as a result of an inhibition of cancer cell adhesion to collagen (21). Indeed, the latter result emphasizes that a precise regulation of expression of integrins in skin cells is a key regulator of migration in wound healing and metastasis, yet we know little about how such regulation is accomplished.
In the current study, we analyzed the consequences of a targeted knockdown in expression of ␣6 integrin. Keratinocytes deficient in ␣6 integrin not only exhibit the same pattern of aberrant motility that we previously observed in cultures of ␤4 integrin-deficient cells (22), but they also show a loss in ␣2␤1 and ␣3␤1 integrin expression. The current data indicate that ␣6␤4 integrin regulates the transcription of ␣2 integrin and the translation of ␣3 integrin.
Lentiviral and Adenoviral Constructs-To express shRNA targeted against ␣6 integrin expression, the BLOCK-iT TM lentiviral RNAi expression system was used (Invitrogen). Two complementary single-stranded DNA oligonucleotides (21mers) derived from the human ITGA6 gene were synthesized, annealed, and cloned into the pENTR TM /U6 entry vector (Invitrogen). A LR recombination was performed between the entry construct and the pLenti6/BLOCK-iT TM -DEST vector to generate an expression construct. To produce lentivirus, the expression construct was transfected into the 293FT packaging cell line. The lentiviral stock was titered, and keratinocytes were infected at a multiplicity of infection of 1:10 in cell medium. To generate stable clones lacking ␣6 integrin expression, infected keratinocytes were selected in 1.75 g/ml of blasticidin. To reexpress ␣6 integrin in the knockdown clones, adenovirus encoding ␣6 integrin mRNA refractory to the shRNA was generated. cDNA encoding ␣6 integrin with 2 kb of its 3Ј-untranslated region was subcloned into the pEGFP-N1 vector (Clontech, Palo Alto, CA). The ␣6 cassette was subsequently subcloned into the polylinker of the pENTR4 vector (Invitrogen). Four point mutations were generated within the ␣6 integrin target shRNA sequence using the QuikChange II XL sitedirected mutagenesis kit (Agilent Technologies, Santa Clara, CA). These point mutations conserved the amino acid sequence of ␣6 integrin and prevented the refractory construct from being targeted by RNAi machinery. The entry vector containing the refractory ␣6 sequence was used in a LR recombination reaction with the pAD/CMV/V5-DEST vector (Invitrogen) to generate an expression clone. The expression clone was transfected into 293A cells using Lipofectamine 2000 (Invitrogen). After 10 days, the crude viral lysate was harvested and used to amplify the adenovirus. The amplified viral stock was titered, and keratinocytes were infected at a multiplicity of infection of 1:50 in cell medium. The cells were used for various analyses 48 -72 h after infection. To express ␣3 integrin in the cells, we used either adenovirus encoding CFP-tagged ␣3 integrin or retrovirus encoding GFP-tagged ␣3 integrin. The retrovirus was a generous gift of Dr. Michael DiPersio (Albany Medical College). To generate the ␣3 integrin adenovirus, the coding sequence of ␣3 integrin and 1.2 kb of its 3Ј-untranslated region were amplified from wild-type keratinocytes by PCR and subcloned into the pECFP-N1 vector (Clontech, Palo Alto, CA). The ␣3 integrin cassette was subsequently subcloned into the polylinker of the pENTR4 vector (Invitrogen), and adenovirus was prepared as described above. Adenovirus encoding myris-toylated AKT was a generous gift of Dr. Navdeep Chandel (Northwestern University).
Polysome Isolation-Polysomal RNA was isolated from wildtype keratinocytes and ␣6 integrin knockdown cells as previously described (26). Total polysomal RNA was quantitated for each cell type, and 1 g of RNA was used for cDNA synthesis. Real time PCR for the ITGA3 or ITGB4 genes was performed. The ribosomal protein S26 was used for normalization.
Quantitative RT-PCR, SDS-PAGE, and Immunoprecipitation-Total RNA was extracted from keratinocytes using the Gene Elute TM Mammalian total RNA Miniprep kit (Sigma-Aldrich) or RNeasy Plus Mini kit (Qiagen). cDNA synthesis was performed using 1 g of total RNA and the qScript cDNA synthesis kit (Quanta Biosciences, Inc., Gaithersburg, MD). Quantitative analysis of ␣2 integrin, ␣3 integrin, ␣6 integrin, or ␤4 integrin mRNA expression was performed using an Applied Biosystems 7000 sequence detection system with SYBR green master mix kit (Invitrogen). GAPDH or S26 mRNA levels served as normalization controls. The ⌬⌬C t method was used to evaluate transcript levels in the various samples.
Flow Cytometry-The cells were trypsinized, resuspended in a 1:1 mix of phosphate-buffered saline containing normal goat serum, and incubated with mouse monoclonal antibodies against ␣2, ␣3, ␣6, or ␤4 integrins at room temperature for 45 min. The cells were washed with phosphate-buffered saline and incubated with FITC-conjugated or Cy5-conjugated goat antimouse IgG for 45 min at room temperature in the dark. Integrin cell surface expression was analyzed using a Beckman Coulter CyAn flow cytometer (Beckman Coulter Inc., Brea, CA). As a negative control, primary antibody was omitted.
Fluorescence Microscopy-The cells were plated onto glass coverslips and processed for microscopical analyses as detailed previously (27). All of the immunofluorescence preparations were viewed with a Zeiss laser scanning 510 confocal microscope (Zeiss Inc., Thornwood, NY), equipped with a Plan-Apochromat 63ϫ 1.4 oil immersion objective lens. The samples were analyzed with Zen software. Images were exported as TIFF files, and figures were generated using Adobe Photoshop software.
Rac Activity Assay-The level of Rac1 activity in keratinocytes was measured using the G-LISA Rac activation assay (Cytoskeleton, Denver, CO) according to the manufacturer's protocol. Cell extracts were prepared from cells, and total protein for each sample was diluted to a concentration of 0.5 mg/ml. Duplicate samples were prepared for each assay. Rac1 activity was measured using an ELx808 ultramicroplate spectrophotometer (Bio-Tek Instruments, Winooski, VT). Data analyses were performed using Microsoft Excel.
Attachment Assay and Strength of Adhesion Assay-Keratinocyte adhesion to matrix was measured by an adhesion assay described previously (28). Briefly, wells of a 96-well non-tissue culture-treated plate (Sarstedt, Newton, NC) were coated with 200 l of laminin-332-rich conditioned medium from the rat bladder cell line, 804G, or with 10 g/ml rat tail collagen I (BD Biosciences) for 2 h at 37°C. Each well was rinsed three times with phosphate-buffered saline and blocked with 1% bovine serum albumin in phosphate-buffered saline for 1 h at 37°C. Various keratinocyte populations (wild-type or ␣6 shRNA clones) were added to the coated wells at 1 ϫ 10 5 cells/well. After 1 h at 37°C, the wells were washed with phosphate-buffered saline to remove nonadhering cells. The adherent cells were fixed in 3.7% formaldehyde in phosphate-buffered saline for 15 min at room temperature. The fixed cells were incubated with crystal violet in 20% methanol for 15 min at room temperature and then solubilized with 1% SDS. Absorbance at 570 nm was measured with a V max plate reader (Molecular Devices, Menlo Park, CA).
Strength of adhesion was measured by a trypsinization assay in which 1 ϫ 10 5 cells were plated onto 60-mm tissue culturetreated dishes and allowed to adhere for 18 -24 h. The cells were exposed to diluted trypsin (Invitrogen) at a concentration of 0.001% trypsin-EDTA for 2 and 5 min or 0.05% trypsin-EDTA for 15 min. The detached cells were collected and counted using a hemocytometer.
Cell Motility-Single cell motility was measured as described previously (23). Briefly, the cells were plated onto uncoated 35-mm glass bottomed culture dishes (MatTek Corp., Ashland, MA) 18 -24 h prior to motility assays. The cells were viewed on a Nikon TE2000 inverted microscope (Nikon Inc., Melville, NY). Images were taken at 2-min intervals over 2 h, and cell motility behavior was tracked using the MetaMorph Imaging System (Universal Imaging Corp., Molecular Devices, Downingtown, PA). Keratinocyte matrix was prepared as previously described (29). For migration on preformed matrices, the cells were allowed to adhere to the matrix for 2 h and imaged over an additional 2 h.
Statistical Analysis-Statistical significance was determined by two-tailed Student's t test. A p value equal to or less than 0.05 was considered statistically significant.

RESULTS
Knockdown of ␣6 Integrin in Human Keratinocytes-In keratinocytes, ␣6 integrin preferentially pairs with ␤4 integrin to form ␣6␤4 integrin, a receptor for laminin-332 (30). However, in ␤4 integrin-deficient keratinocytes, expression of ␣6␤1 integrin is induced. To assess the consequences of loss of ␣6␤4 integrin expression, without subsequent ␣6␤1 integrin expression, we generated keratinocyte cell lines with reduced expression of the ␣6 integrin subunit by infecting cells with a lentivirus that encodes shRNA targeted against the ␣6 integrin coding sequence. Because ␣6 integrin undergoes alternative splicing to yield two ␣6 integrin subunits (␣6A and ␣6B) that contain distinct cytoplasmic domains (31), we designed the shRNA to target both ␣6A and ␣6B subunits to prevent any complicating effects from expression of, or compensation by, ␣6B integrin.
Knockdown of ␣6 integrin expression was evaluated in three randomly selected clones expressing ␣6 integrin shRNA (Fig. 1,  A-D). ␣6 integrin protein level was decreased between 65 and 85% in these three ␣6 integrin shRNA clones, and all three showed decreased ␣6 integrin immunostaining with little or no localization of ␤4 integrin at the basal surface of the cells (Fig.   1C). The latter is in contrast to the localization of ␣6␤4 integrin along the substratum-attached surface of control (wild-type) keratinocytes (Fig. 1C). Surface expression of ␣6 integrin was decreased between 80 and 90% compared with parental keratinocytes, whereas ␤4 integrin surface expression was almost completely absent in the three shRNA-expressing clones (Fig.  1D). The latter was not due to an off target effect of the ␣6 integrin shRNA because mRNA levels of ␤4 integrin were not decreased in any of our ␣6 shRNA clones (Fig. 1A). Furthermore, we "rescued" expression of both ␣6 and ␤4 integrin, as assessed by immunofluorescence staining, FACS analysis, and immunoblotting, by infecting ␣6 shRNA-expressing cells with adenovirus expressing ␣6 integrin mRNA that is refractory to the shRNA (Fig. 1, C-E).
Knockdown of ␣6 integrin, and the resulting depletion of ␤4 integrin, did not significantly alter total protein expression of ␣6␤4 integrin-associated proteins, including the transmembrane protein BP180 (type XVII collagen) and BPAG1e (Fig.  1B). Instead, the loss of ␣6␤4 integrin altered their localization (supplemental Fig. S1). Staining of both BP180 and BPAG1e appeared diffuse in the ␣6 shRNA-expressing cells, whereas in control keratinocytes, these proteins were localized to the basal cell surface (supplemental Fig. S1). This is consistent with previous published data in which ␣6␤4 integrin is required for assembly of ␣6␤4 integrin-BP180-BPAG1e complexes (32,33).
␣2 and ␣3 integrin subunit surface expression was also measured in all three ␣6 integrin shRNA-expressing clones (Fig. 2,  A and B). Surface expression of ␣3 and ␣2 integrin was decreased in the clones, averaging 42 and 26%, respectively, of the levels expressed in wild-type cells (Fig. 2, A and B). Similar to ␤4 integrin mRNA, ␣3 integrin mRNA levels were not significantly decreased in the ␣6 shRNA-expressing clones (Fig.  2C). Interestingly, in contrast to ␣3 and ␤4 integrin mRNA, ␣2 mRNA levels were decreased in the cells (Fig. 2D). This effect was not due to nonspecific targeting of our shRNA because re-expression of ␣6 integrin in the knockdown cells increased ␣2 integrin mRNA levels and restored ␣2 integrin surface expression (Fig. 2, E and G; results for one clone only are shown). In addition, overexpression of ␣3 integrin in the ␣6 integrin shRNA-expressing cells restored ␣2 integrin mRNA levels (Fig. 2E), suggesting a hierarchy of expression whereby ␣6␤4 integrin regulates ␣3 integrin subunit expression that in turn regulates ␣2 integrin expression.
Loss of ␣6 Integrin Alters Keratinocyte Cell Adhesion and Motility-Each of the ␣6 integrin shRNA clones adhered less robustly to laminin-332-and collagen-coated substrates and exhibited low strength of attachment to their own matrix, when compared with control keratinocytes (Fig. 3A and supplemental Fig. S2). Results for clone 1 cells are shown in this and subsequent figures because all clones behaved similarly. Supplemental Fig. S2 shows results from the other ␣6 shRNA-expressing clones.
We have previously demonstrated that ␤4 integrin-deficient keratinocytes display aberrant motility compared with normal keratinocytes, implicating ␣6␤4 integrin in mediating migration behavior (22). Furthermore, we have also presented evidence that ␣6␤4 integrin, through the plakin molecule BPAG1e, regulates Rac1 activity, which in turn signals to the Slingshot phosphatase proteins to dephosphorylate and activate the actin-severing protein cofilin (23,34). Therefore, to further characterize the ␣6 integrin shRNA-expressing clones, we evaluated their motility phenotype and the level of their Rac1 activity. Whereas control keratinocytes moved primarily in linear tracks, the ␣6 shRNA-expressing cells moved in circles, had decreased Rac1 activity, and migrated significantly more slowly than their wild-type counterparts (Fig. 3, B-D, and   FIGURE 1. Stable knockdown of ␣6 integrin expression in human keratinocytes. Human keratinocytes were infected with lentivirus encoding ␣6 integrin shRNA and stably selected for the loss of ␣6 integrin expression. ␣6 and ␤4 integrin expression was measured in three knockdown clones (1-3, as indicated) by quantitative RT-PCR (A). The bars of the graph represent the relative expression (ϮS.D.) of ␣6 and ␤4 integrin mRNA in the shRNA clones normalized to GAPDH and compared with HEK. Samples were measured in triplicate for two individual experiments. B, ␣6 integrin, ␤4 integrin, actin, BP180, mTOR, and lamin were evaluated in the same clones by immunoblot. ␤-Actin, mTOR, and lamin were used as loading controls. C, ␣6 and ␤4 integrin localization was determined by confocal immunofluorescence (␣6 shRNA clone 1 only; bar, 10 m). D, surface expression of ␣6 and ␤4 integrins in the three knockdown clones was evaluated by FACS (top two panels). The black curves represent secondary antibody alone. In the lower two panels of D and E, wild-type cells, ␣6 shRNA clone 1 cells, and ␣6 shRNA clone 1 cells infected with adenovirus encoding a refractory ␣6 integrin construct to re-express ␣6 integrin (ϩref␣6) were analyzed by FACS and immunoblot. In E, the graph represents the relative level of ␤4 integrin protein expression in cells compared with parental HEK, quantified from immunoblots as shown to the left. The individual bars represent the means Ϯ S. E. (n ϭ 3). The p values were generated by Student's t test. **, p Յ 0.001. The results using control (wild-type) keratinocytes (HEK) are presented in each assay. FIGURE 2. Loss of ␣6␤4 integrin leads to decreased ␣2 and ␣3 integrin expression. ␣2 and ␣3 integrin expression was measured in control cells (HEK), ␣6 shRNA clones 1-3, and ␣6 shRNA clone 1 cells induced to re-express GFP-tagged ␣6 integrin (ϩref␣6) by FACS (A, B, F, and G) and quantitative RT-PCR (C-E). The individual bars on the graphs represent the relative levels of ␣2 or ␣3 integrin mRNA normalized to GAPDH and compared with HEK (ϮS.D.). The samples were measured in triplicate in two experiments. In E, ␣2 integrin mRNA expression was also measured in ␣6 shRNA clone 1 keratinocytes infected with adenovirus expressing ␣3 integrin (ϩ␣3). In H, ␣3 integrin surface expression was measured by FACS analysis in ␤4 integrin-deficient keratinocytes (JEB) and JEB cells re-expressing ␤4 integrin. The black curves on the FACS analyses represent secondary antibody alone.
supplemental Fig. S3). Interestingly, in prior studies we did not observe a comparable difference in velocity between wild-type keratinocytes and skin cells deficient in BPAG1e or BP180 (34,35). Again the current result was not due to nonspecific effects of the ␣6 integrin shRNA because velocity and normal migratory behavior was restored in the clones following expression of ␣6 integrin mRNA that is refractory to the ␣6 integrin shRNA (Fig. 3, B and D).
The aberrant motile behavior of ␤4 integrin-deficient cells and cells exhibiting inhibition in Rac1 signaling to cofilin because of BPAG1e or BP180 loss is reversed by plating the cells onto the laminin-332-rich matrix deposited by wild-type keratinocytes (22,23). Thus, we tracked migration patterns of individual ␣6 integrin shRNA-expressing cells after plating on matrix deposited by wild-type keratinocytes. As expected, the ␣6 integrin shRNA cells now migrated in a more linear pattern on this matrix compared with on their own matrix, yet they failed to increase their speed of migration (Fig. 3, E-G). These results suggested that the ␣6 shRNA-expressing cells lacked a key regulator of cell velocity. In this regard, we and others have reported that migration of keratinocytes on laminin-332-rich matrix is positively regulated by ␣3␤1 integrin, which, like ␣6␤4 integrin, is a receptor for laminin-332 (16, 22, 36 -38). Because our ␣6 shRNA-expressing cells exhibited a loss in ␣3␤1 integrin surface expression, we induced expression of ␣3 integrin in the cells and tracked the motile behavior of the ␣3-expressing cells following plating onto matrix deposited by control keratinocytes. These cells migrated in a similar pattern and with a similar rate of migration as control keratinocytes plated onto the same matrix (Fig. 3, E-G).
␣6␤4 Integrin Regulates Translation of ␣3 Integrin-We next measured the relative amount of ␣3 integrin mRNA associated with the polysomes of control and ␣6 integrin shRNA-expressing cells. Keratinocytes lacking ␣6␤4 integrin expression had significantly less ␣3 integrin mRNA associated with the polysome fraction compared with control cells (Fig. 4A). Thus, these data suggest that ␣3 integrin is translationally regulated by ␣6␤4 integrin. In support of this conclusion, re-expression of the refractory ␣6 integrin mRNA in ␣6 knockdown cells rescues ␣3 integrin expression. Likewise, expression of ␤4 integrin in ␤4 integrin-deficient JEB cells also resulted in an increase in ␣3 integrin surface expression (Fig. 2, F and H).
Others have demonstrated that ␣6␤4 integrin regulates the translation of several genes important for cell survival and migration by mediating the phosphorylation of initiation factor FIGURE 3. Human keratinocytes lacking ␣6␤4 integrin expression display adhesion and migration defects. In A, control keratinocytes (HEK) and ␣6 shRNA clone 1 cells were assayed for adhesion to laminin-332-or collagen I-coated dishes at 1 h after plating. The number of adherent cells was measured by reading the absorbance at 570 nm. The average absorbance for HEK was taken as 100%. The individual bars on the graph represent the means Ϯ S.E. (n ϭ 3) with samples measured in triplicate for each experiment. p values were derived by comparing samples to HEK (Student's t test). *, p Յ 0.05. B, vector diagrams depict the individual migration patterns of wild-type keratinocytes (HEK) (n ϭ 11), ␣6 shRNA clone 1 cells (n ϭ 12), or ␣6 shRNA clone 1 cells induced to re-express GFP-tagged ␣6 integrin (ϩref␣6) (n ϭ 12). The cells were plated on glass and tracked over a 2-h period. C, Rac1 activity in control HEK and ␣6 shRNA clone 1 cells was measured by the G-LISA Rac1 activation assay. The individual bars of the graph represent the means Ϯ S.E. 4E binding protein 1 (4EBP1) through signaling via the PI3K pathway (39,40). 4EBP1 binds and represses the activity of the translation initiation factor 4E (eIF-4E) (41). Phosphorylation of 4EBP1 disrupts this interaction and releases eIF-4E, which can then recruit translational machinery and initiate translation (42). Therefore, we compared the level of phosphorylated 4EBP1 in extracts of wild-type keratinocytes with that in the ␣6 integrin knockdown cells. Significantly less 4EBP1 was phosphorylated in the ␣6 shRNA-expressing keratinocytes compared with wild-type cells (Fig. 4B). Interestingly, total 4EBP1 expression was also decreased in the ␣6 shRNA-expressing cells (Fig. 4, B and C). Although re-expression of the refractory ␣6 integrin construct in the shRNA cells failed to restore total 4EBP1 levels, ␣6 integrin re-expression led to a significant increase in phosphorylation of 4EBP1, such that the level of phosphorylated 4EBP1 mirrored that of control keratinocytes (Fig. 4B). Consistent with these data, re-expression of ␣6 integrin in the shRNA-expressing cells led to an increase in the level of phosphorylated AKT, suggesting that PI3K signaling was deficient in ␣6 integrin knockdown cells (supplemental Fig. S4). Moreover, activation of the AKT-mTOR pathway, by inducing the expression of myristoylated AKT, also increased the levels of phosphorylated 4EBP1 in the ␣6 shRNA-expressing cells without inducing an increase in total 4EBP1 levels (Fig. 4C). Expression of myristoylated AKT also increased the level of ␣3 integrin protein in the ␣6 shRNA-expressing cells, supporting the notion that ␣6␤4 integrin signaling to 4EBP1 regulates ␣3 integrin expression (Fig. 4D).
Interestingly, significantly less ␤4 integrin mRNA was associated with the polysome fraction from ␣6 integrin shRNAexpressing cells (Fig. 4A), implicating a feedback loop in which ␤4 integrin translation is regulated by phosphorylation of 4EBP1. Consistent with this hypothesis, expression of myristoylated AKT increased the level of ␤4 integrin expression in ␣6 shRNA-expressing keratinocytes (Fig. 4D).

DISCUSSION
␣6 Integrin and Motility-Although a central role for laminin-332 in regulating wound healing through its effects on keratinocyte migration is generally agreed upon, the functions of its two integrin receptors (␣6␤4 and ␣3␤1) in migration are much debated (19,22,36,37,(43)(44)(45)(46)(47)(48)(49)(50)(51)(52). In this study we demonstrate that a loss of ␣6 integrin expression in human keratinocytes leads to the same aberrant circular migratory phenotype and decrease in Rac1 activation as is exhibited by keratinocytes deficient in ␤4 integrin (22). However, knockdown of ␣6 integrin in keratinocytes also results in lower motility rates when the cells migrate on preformed laminin-332-rich matrices. We demonstrate that this is due to decreased expression of ␣3␤1 integrin, because induced up-regulation of ␣3 integrin is sufficient to rescue the motility rates of ␣6 integrin knockdown cells moving on preformed keratinocyte matrix.
In contrast to our results, other workers have presented evidence that ␣6␤4 and ␣3␤1 integrins impede migration of keratinocytes and/or wound closure in vitro (37,(47)(48)(49). The latter evidence comes primarily from studies using keratinocyte cultures generated from the skin of various integrin knock-out mice. Therein may lay the explanation for the apparent contradiction in results. Whereas human keratinocytes deposit and move over a matrix rich in laminin-332, mouse keratinocytes deposit a matrix rich in both laminin-332 and fibronectin. Recent evidence suggests that fibronectin in mouse skin cell matrix impedes laminin-332-driven migration (52), most likely by enhancing cell adhesion to substrate, such that mouse keratinocytes move much more slowly than human keratinocytes in vitro. The presence of fibronectin in mouse keratinocyte matrix complicates studies where the goal is to analyze integrin function in laminin-332-driven migration/adhesion and invalidates a direct comparison of migration results derived from work FIGURE 4. ␣6␤4 integrin regulates the translation of ␣3 integrin through 4EBP1. A, polysome fractions from control keratinocytes or ␣6 shRNA clone 1 keratinocytes were isolated, and the levels of ␣3 integrin (left), ␤4 integrin (middle), and GAPDH (right) mRNA were determined by quantitative RT-PCR, as indicated. The graphs represent the relative levels of ␣3, ␤4, or GAPDH mRNA normalized to the ribosomal protein S26 and compared with HEK. The samples were measured in triplicate, and the graphs represent the means Ϯ S.E. of three independent experiments. B, whole cell extracts from wild-type (HEK), ␣6 shRNA clone 1 cells, or ␣6 shRNA cone 1 cells induced to re-express ␣6 integrin (ϩref␣6) were probed for levels of phosphorylated 4EBP1 and total 4EBP1. C and D, whole cell extracts from wild-type (HEK), ␣6 shRNA clone 1 cells, or ␣6 shRNA clone 1 cells infected with adenovirus encoding myristoylated AKT were probed for levels of phosphorylated 4EBP1 and total 4EBP1 (C) or ␤4 integrin and ␣3 integrin (D). Actin and lamin A/C were used as loading controls in B-D. Graphs in B-D represent the means expression (normalized to lamin A/C) ϩ S.E. of three independent experiments, quantified from immunoblots.
using human keratinocytes versus those using mouse keratinocytes. In this regard, our results emphasize the differential roles of ␣6␤4 and ␣3␤1 integrin in human skin cell motility under conditions where laminin-332 is the predominant matrix secreted by the cells. Thus, we conclude that ␣6␤4 integrin regulates directed migration, whereas ␣3␤1 integrin determines migration velocity.
␣6␤4 Integrin Regulates Transcription or Translation of Other Keratinocyte Integrins-Our analyses of ␣6 integrin-deficient keratinocytes indicate that ␣6␤4 integrin is a regulator of the transcription and translation of other major skin cell integrin subunits. We demonstrate here that the loss of ␣6 integrin leads to a concomitant decrease in surface expression of the ␣3 and ␣2 integrin subunits. This was observed in all of our clones and is not the consequence of viral-mediated delivery of shRNA, because we fail to observe a comparable decrease in keratinocytes infected with adenovirus encoding BPAG1e shRNA (34). Our finding is particularly intriguing because ␣2␤1, ␣3␤1, and ␣6␤4 integrins are among the most abundant integrins expressed in the keratinocytes of the epidermis (15). Thus, we propose that ␣6␤4 integrin is a master regulator of the expression levels of the other major integrins in the epidermis. Such a functional role for ␣6␤4 integrin has not been previously described, although it is consistent with the previous results of Chung et al. (39), who established that ␣6␤4 integrin-dependent signaling regulates the translation of mRNA encoding genes required for cell survival. Here, we present evidence that ␣6␤4 integrin-dependent signaling, via phosphorylation of 4EBP1 and activation of PI3K, regulates the translation of ␣3 integrin. The expression of ␣3 integrin in turn determines the speed of keratinocytes moving over laminin-332. Moreover, our results also indicate that the expression of ␣3 integrin regulates either the transcription of the ␣2 integrin gene or the stability of ␣2 integrin mRNA (Fig. 5). Future experiments are needed to discern the exact mechanism by which ␣3 integrin regulates ␣2 integrin expression. Nevertheless, there appears to be a hierarchy in integrin expression in the epidermis, with ␣6␤4 integrin orchestrating the coordinated expression of other epidermal integrins.
In summary, our data support the notion that ␣6␤4 integrin is involved in skin cell migration, an area of recent controversy. Moreover, it does so not only via an ability to activate specific signaling pathways, but also by determining the expression of ␣3␤1 integrin, which we show regulates cellular velocity. In addition, our study has uncovered a novel role for ␣6␤4 integrin in regulating the transcription and translation of the subunits of two major laminin-and collagen-binding integrin heterodimers expressed in skin cells in vivo. Indeed, we suggest that this new paradigm for integrin cross-talk plays a central role in determining tissue organization and remodeling in normal, developing, and healing skin, as well as mediating the dissemination of skin cells in cancer by fine tuning matrix interactions and matrix-mediated signal transduction. FIGURE 5. Schematic of ␣6␤4 integrin regulation of ␣3 and ␣2 integrin expression. ␣6␤4 integrin signaling to PI3K leads to activation of AKT and its downstream effectors. Activation of this pathway leads to phosphorylation/ inactivation of 4EBP1, dissociation of the complex containing initiation factor 4E (eIF4E) and 4EBP1, and ultimately, the translation of ␣3 integrin. Expression of ␣3 integrin in turn mediates the expression of ␣2 integrin by regulating the transcription of the ␣2 integrin gene and/or ␣2 integrin mRNA stability.