E2F-1 Is Essential for Normal Epidermal Wound Repair*

E2F factors are involved in proliferation and apoptosis. To understand the role of E2F-1 in the epidermis, we screened wild type and E2F-1−/− keratinocyte mRNA for genes differentially expressed in the two cell populations. We demonstrate the reduced expression of integrins α5, α6, β1, and β4 in E2F-1−/− keratinocytes associated with reduced activation of Jun terminal kinase and Erk upon integrin stimulation. As a consequence of altered integrin expression and function, E2F-1−/− keratinocytes also show impaired migration, adhesion to extracellular matrix proteins, and a blunted chemotactic response to transforming growth factor-γ1. E2F-1−/− keratinocytes, but not dermal fibroblasts, exhibit altered patterns of proliferation, including significant delays in transit through both G1 and S phases of the cell cycle. Recognizing that proliferation and migration are key for proper wound healing in vivo, we postulated that E2F-1−/− mice may exhibit abnormal epidermal repair upon injury. Consistent with our hypothesis, E2F-1−/− mice exhibited impaired cutaneous wound healing. This defect is associated with substantially reduced local inflammatory responses and rates of re-epithelialization. Thus, we demonstrate that E2F-1 is indispensable for a hitherto unidentified cell type-specific and unique role in keratinocyte proliferation, adhesion, and migration as well as in proper wound repair and epidermal regeneration in vivo.

The E2F family of transcription factors is pivotal in the control of cell proliferation and is involved in patterning during early embryogenesis (1)(2)(3). E2F belongs to a network that also includes cyclins, cyclin-dependent kinases (cdk), cdk inhibitors, and the retinoblastoma family of proteins (for review, see Refs. 1 and 4). The E2F multigene family consists of two subgroups termed E2F and DP. Functional E2F units are heterodimers composed one E2F and one DP protein. Mammalian E2F-1-5 also have a C-terminal transactivation domain, which mediates binding to retinoblastoma family proteins (for review, see Refs. 1 and 4).
The precise biological roles played by each E2F protein are poorly understood. To address this issue, mice with targeted inactivation of E2F genes have been generated. These studies have shown that E2F proteins can fulfil tissue-specific functions. For example, E2F-1 is necessary for thymocyte apoptosis and selection (5,6), E2F-4 is indispensable for hematopoietic and intestinal epithelial cell maturation (7,8), and E2F-5 is critical for choroid plexus function (9).
E2F-1 is one of the most extensively characterized members of the E2F family. E2F-1 is a strong activator of transcription and DNA synthesis (10,11) and can promote S phase entry and induce apoptosis in a p53-dependent or -independent manner (12,13). E2F-1 Ϫ/Ϫ mice are viable and fertile (5,6). However, they exhibit testicular atrophy and exocrine gland dysplasia, and one of the two lines reported develops tumors with multiple tissues of origin (6). Notably, abnormalities in stratified epithelia have yet to be reported in these animals.
To begin to define the biological functions of E2F factors, we analyzed their patterns of expression during development. We have previously shown that E2F-1 is expressed in undifferentiated cells in tissues of neuroectodermal origin, including neuronal progenitors and epidermal keratinocytes (14,15). Terminal differentiation of epidermal keratinocytes results in pronounced down-regulation of E2F-1 transcripts and protein, and ectopic expression of E2F-1 is sufficient to promote entry into S phase in keratinocytes at any stage of differentiation (14). These properties are consistent with a role for E2F-1 in maintenance of the proliferative state in these cells.
A fundamental property of the epidermis is its constant state of self-renewal (for review, see Ref. 16). This is accomplished through keratinocyte stem cells, which divide to give rise to new stem cells or to differentiating keratinocytes. The epidermis also has the ability to regenerate upon injury through a dynamic process that involves multiple cell types and responses (for review, see Refs. 17 and 18). Cutaneous repair requires blood clotting, inflammation, re-epithelialization, and tissue remodeling. Multiple signals control these processes, including cell interactions with other cells or with the extracellular matrix as well as secretion of and responses to a variety of cytokines.
To determine the role that E2F-1 plays in epidermal homeostasis, morphogenesis, and repair, we examined the proliferative and regenerative properties of keratinocytes isolated from E2F-1 Ϫ/Ϫ mice. Our studies demonstrate that E2F-1 is indispensable for normal keratinocyte proliferation in culture and cutaneous wound repair in vivo. Our studies also show a novel, previously unidentified biological role of E2F-1 beyond cell proliferation, which includes modulation of cell adhesion, migration, and chemotaxis through integrin expression and function. dinium thiocyanate method (19) and reverse-transcribed with Superscript II (Invitrogen) following the manufacturer's recommendations. Representational difference analysis was performed on the cDNAs prepared as described (20). E2F-1 Ϫ/Ϫ keratinocyte cDNA was subtracted three times against wild type cDNA (1:100, 1:400, and 1:800) to generate the final difference products. Integrin mRNA levels were measured using a quantitative reverse transcription-PCR method (21) from total RNA prepared from freshly isolated or from cultured keratinocytes. Reverse transcription followed by polymerase chain reaction amplification was conducted on at least two different RNA samples (1 g of RNA was reverse-transcribed, and cDNA equivalent to 0.1 g of RNA was used in each PCR reaction). The primers used were: ␣ 5 integrin, 5Ј-C-ATTTCCGAGTCTGGGCCA and 5Ј-TGGAGGCTTGAGCTGAGCTT (22); ␣ 6 integrin, 5Ј-TAGAGCCAGCATCAGAATCCC and 5Ј-GCTAAG-CCTTCGCAGGTGTAT; ␤1 integrin, 5Ј-TGTTCAGTGCAGAGCCTTCA and 5Ј-CCTCATACTTCGGATTGACC (22); ␤4 integrin, 5Ј-GAGGATC-TCCTGCCTAACTAC and 5Ј-ACTGTTGGTCCATATGAGTGC; glyceraldehyde-3-phosphate dehydrogenase, 5Ј-CAAAGTTGTCATGGATG-ACC and 5Ј-GTTGCCATCAACGACCCCTT or 5Ј-GCTTCACCACCTT-CTTGATGTCATC and 5Ј-GTTGCCATCAACGACCCCTT. PCR fragments obtained at 18, 20, 22, and 24 amplification cycles were resolved by electrophoresis, transferred to nylon membranes, and hybridized to appropriate 32 P-labeled probes. The signals were detected and quantified using a Canberra-Packard phosphorimager, as described (21,23), and normalized to glyceraldehyde-3-phosphate dehydrogenase products amplified in the same reactions. For each cDNA, amplification was quantified by phosphorimaging analysis at multiple PCR cycles (18 -24 cycles). Amplification of all cDNAs tested was logarithmic within these parameters.
In Situ Hybridization-In situ hybridization experiments were conducted as described (15) using 8-m frozen sections from wound tissue samples obtained after debridement.
Primary Keratinocyte Cultures and Proliferation Measurements-Primary keratinocytes were isolated from 1-3-day-old E2F-1 Ϫ/Ϫ , E2F-1 ϩ/ϩ , or E2F-1 ϩ/Ϫ mice (5) and cultured under non-differentiating conditions as described (14). Unless otherwise indicated, experiments were conducted on triplicate samples 3-5 days after initial plating in cultures that were 70 -80% confluent by the end of the experiment. For proliferation experiments, 2 ϫ 10 5 cells/well were plated in 12-well culture dishes, and cell numbers were determined daily. To measure DNA synthesis, keratinocyte cultures were incubated with 1.5 Ci/ml [ 3 H]dThd for 3 h, and incorporation into DNA was measured by liquid scintillation counting of trichloroacetic acid-insoluble cell fractions. The 3 H activity was normalized to cell numbers. For flow cytometric analyses, 5 ϫ 10 6 cells/10-cm tissue culture dish were plated and collected at the indicated intervals. Each experiment was conducted on triplicate samples. Pharmacological treatments included culture with human TGF-␤1 (Life Science Technologies) at 1, 5, or 10 ng/ml (final), as indicated in individual experiments. Chemicals were purchased from Sigma, unless otherwise indicated Keratinocyte Migration and Adhesion Assays-Primary keratinocytes cultured for 2 days in low calcium medium (0.05 mM CaCl 2 ) were trypsinized, washed, resuspended at 1.5 ϫ 10 6 cells/ml in serum-free calcium-free Eagle's minimum essential medium, and 200 l of this cell suspension were added to tissue culture inserts (8-m pore size, Becton and Dickinson). Test medium in the lower chamber included EMEM, conditioned medium from E2F-1 Ϫ/Ϫ or E2F-1 ϩ/ϩ keratinocytes, or TGF-␤1 1 (1 ng/ml, final). Cells were cultured at 37°C for 7 h, and those cells that migrated through the membrane were stained with Hemacolor (EM Science) and counted from microscope images obtained with a Leica DMIRBE photomicroscope. Adhesion experiments were conducted on keratinocytes briefly trypsinized and re-plated (100,000 cells/cm 2 ) for 1 h in non-coated culture dishes or on dishes coated with fibronectin, collagen IV (Becton-Dickinson), or laminin as described (24). All experiments were conducted on triplicate samples.
Measurement of Integrin-induced Jun N-terminal Kinase (JNK) or Erk Activity-To obtain ligation of integrins in the absence of other stimuli, keratinocytes were cultured in 0.1% serum and growth factorfree medium for 24 h, detached, and resuspended in serum-free medium containing 1% bovine serum albumin. The cells were incubated in suspension for 45 min and plated onto dishes previously coated with fibronectin (0.5 mg/ml stock) and collagen I (3 mg/ml stock) or with laminin 5-enriched matrix, prepared as described (25). The cells attached and spread equally well on all matrices. Cell lysates from ad-herent and non-adherent cells were prepared 30 min after plating and subjected to biochemical analysis. Lysis buffer contained 20 mM Tris⅐HCl, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM ␥-glycerophosphate, 1 mM Na 3 VO 4 , 1 g/ml leupeptin and pepstatin, and 1 mM phenylmethylsulfonyl fluoride. Positive controls for JNK and Erk activation were prepared using cell lysates from serum-free adherent cultures exposed to UV irradiation, as described (26). JNK activity was assessed by measuring phosphorylation of glutathione S-transferase-Jun fusion proteins with a stress-activated protein kinase/JNK assay kit (New England Biolabs) following the manufacturer's instructions or from immunoblots prepared from 50 g of cell lysates/sample and probed with an antiphospho-JNK antibody (New England Biolabs). Erk activity was measured from immunoblots using a phospho-Erk antibody (New England Biolabs). Signals from the blots were normalized with anti-tubulin or with anti-Erk (total) antibodies (New England Biolabs). The results shown are representative of results obtained from experiments conducted on duplicate samples and repeated four times.
Wound-healing Experiments and Histological Analysis-Groups of sixteen 6-week-old mice (8 males and 8 females; E2F-1 ϩ/ϩ or E2F-1 Ϫ/Ϫ ) were used for each time analyzed after wounding. For wound-healing measurements, tails were clipped with a single stroke of a scalpel blade in halothane-anesthetized animals, following a described procedure (27,28). The animals were sacrificed at the indicated times after tail amputation, and tail stumps containing the healing areas were removed, embedded in OCT compound (Miles), and frozen. The tissue was then used to prepare frozen histological sections. Tissue sections (8 m) were fixed with 4% paraformaldehyde and stained with hematoxylin and eosin following standard procedures. Measurements of cell numbers/ unit area or wound area (measured below the clot) and re-epithelialization were done using Openlab software (Improvision). Representative photographs of sections traversing the tissue in its mid-portion were recorded. Following this approach, % re-epithelialization was calculated from the percentage of distance migrated by newly formed epidermis relative to upper wound width. The values shown in Table I for re-epithelialization at any given time represent the minimum and maximum fraction of epithelialized tissue found per animal in each group of 16 mice. A single value of 100% indicates full re-epithelialization in all 16 animals. Inflammation in tissue sections was assessed by determining the number of inflammatory cells (macrophages, neutrophils) per unit area in Giemsa-stained tissue from wild type and mutant animals, as described (29). We scored as (ϩϩϩ) the maximum number of cells observed in wild type wounds (505 Ϯ 79 cells/unit area, mean Ϯ S.E., which were observed at days 1 and 2 post-wounding). Average cell number values in the range 50 -166 cells/unit area and 167-333 cells/ unit area were scored, respectively, as (ϩ) and (ϩϩ) in Table I.

Identification of Genes Differentially Expressed in E2F-1 Ϫ/Ϫ
Keratinocytes-To gain insight into the role of E2F-1 in epidermal morphogenesis, we employed representational difference analysis (20) to identify genes differentially expressed in E2F-1 Ϫ/Ϫ versus wild type undifferentiated primary keratinocytes. We identified 11 distinct products and compared their sequences with those available in GenBank TM (NCBI) using the FASTA program. The isolated products correspond to nine known and two potentially novel cDNAs down-regulated in mutant keratinocytes. 2 Four of these cDNAs correspond to integrin gene products, specifically integrins ␣ 5 , ␣ 6 , ␤ 1 , and ␤ 4 .
Decreased Integrin Expression and Signaling in E2F-1 Ϫ/Ϫ Keratinocytes-To confirm that the above described integrins are indeed down-regulated in mutant keratinocytes, we isolated RNA from wild type and E2F-1 Ϫ/Ϫ primary undifferentiated keratinocytes and epidermis and estimated integrin mRNA abundance using a quantitative reverse transcription-PCR approach (21,23). The results of these experiments confirmed that the transcript levels of integrins ␣ 5 , ␣ 6 , ␤ 1 , and ␤ 4 are 2-3-fold lower in mutant keratinocytes (Fig. 1A).
We next examined whether decreased integrin transcript levels resulted in altered signaling via integrin heterodimers. Undifferentiated keratinocytes express several integrin het-erodimers, including ␣ 6 ␤ 4 and ␣ 5 ␤ 1 integrins, which mediate adhesion to and are stimulated by laminin-5 and fibronectin or collagen, respectively.
Stimulation of ␣ 6 ␤ 4 integrin by laminin-5 induces activation by phosphorylation of JNK and is key for normal keratinocyte proliferation (26). To begin to understand the functional consequences of reduced ␣ 6 and ␤ 4 integrin expression in mutant keratinocytes, we first analyzed the degree of activation of JNK by laminin stimulation of integrin ␣ 6 ␤ 4 in wild type and E2F-1 Ϫ/Ϫ cells. We first compared the abundance of phosphorylated JNK in wild type or E2F-1 Ϫ/Ϫ keratinocyte lysates prepared from suspended keratinocytes (to preclude integrin engagement) or from cells plated on laminin-5. We found that phospho-JNK levels were substantially decreased in laminin-5-stimulated mutant keratinocytes, relative to wild type cells (Fig. 1B). To confirm these observations, we measured the ability of JNK immunoprecipitates prepared from these lysates to phosphorylate in vitro a glutathione S-transferase-Jun fusion protein. We also found correspondingly similar decreases in phosphorylated glutathione S-transferase-Jun in mutant lysates (data not shown). Thus, the reduced expression of ␣ 6 and ␤ 4 integrin subunits in mutant keratinocytes is associated with decreased signaling through ␣ 6 ␤ 4 dimers and activation of the JNK pathway. Notably, this phenotype is specific to JNK activation via the integrin pathway, as UV irradiation of quiescent, adherent wild type, and mutant cells resulted in similar degrees of JNK activation (Fig. 1B).
Similar to the activation of JNK by laminin, ligation of ␣ 5 ␤ 1 integrin in keratinocytes by fibronectin and/or collagen results in activation of the mitogen-activated protein kinase cascade, with phosphorylation and consequent activation of Erk (p42 and p44). We measured the levels of active phospho-Erk p42/ p44 in lysates from keratinocytes in suspension or plated on fibronectin/collagen I-coated dishes (Fig. 1C). Consistent with the alteration in integrin ␣ 5 and ␤ 1 levels, phospho-Erk abundance in E2F-1 Ϫ/Ϫ keratinocytes was substantially lower than in wild type cells. To determine whether the reduced Erk activation was specifically associated with integrin signaling, we measured phosphorylated Erk abundance in quiescent, adherent cells exposed to UV radiation. Irradiated mutant cells exhibited similar levels of Erk phosphorylation than those observed in normal keratinocytes, indicating a selective reduction in Erk activation associated with integrin stimulation.
Inactivation of E2F-1 Affects Keratinocyte Chemotactic Responses to TGF-␤1 and Adhesion-Integrins are central for cell adhesion and migration. Having determined alterations in integrin signaling pathways consequent to E2F-1 Ϫ/Ϫ inactivation, we then examined the consequences of reduced integrin levels and function on keratinocyte migration in response to chemotactic agents. Thus we measured the ability of cells to migrate through culture inserts placed over wells in the presence or absence of chemoattractants ( Fig. 2A). Wild type keratinocytes showed the same extent of increased migration toward conditioned media from normal or from E2F-1 Ϫ/Ϫ cultures, indicating that disruption of the E2F-1 gene does not appreciably alter the ability of keratinocytes to secrete chemoattractant factors. TGF-␤1 was strongly chemotactic toward normal cells. In contrast, E2F-1 Ϫ/Ϫ keratinocytes exhibited Integrin levels were estimated from quantitative reverse transcription-PCR assays, detected by hybridization to appropriate 32 P-labeled probes. For each cell population, band intensities were developed using phosphorimaging normalized to those obtained from glyceraldehyde-3phosphate dehydrogenase and are expressed as the mean Ϯ S.D. (n ϭ 6). For all data shown, p Ͻ 0.05 (Student's t test). B and C, activation of JNK and Erk by integrin stimulation is impaired in E2F-1 Ϫ/Ϫ keratinocytes. Keratinocytes were serum-and growth factor-starved for 24 h and re-plated on dishes coated with the indicated matrix protein to stimulate integrin signaling or maintained in suspension (Sus) for 30 min. Activated, phospho-JNK (P-JNK) or phospho-Erk (P-Erk), total Erk, and tubulin were detected from total cell lysates by immunoblotting. Positive controls were prepared from serum-starved adherent cells irradiated with 50 J/m 2 UV radiation. greatly reduced migration both toward TGF-␤1 and toward growth factors present in keratinocyte-conditioned media ( Fig.  2A). To investigate whether other responses of E2F-1 Ϫ/Ϫ keratinocytes to TGF-␤1 were affected, we also measured the ability of this cytokine to inhibit proliferation of cultured mutant and wild type keratinocytes. Treatment with TGF-␤1 for 24 or 48 h inhibited DNA synthesis by 80 -90% in both cell types (Fig. 2B). Thus, E2F-1 inactivation and integrin signaling defects in keratinocytes specifically affect chemotaxis but not proliferative responses to TGF-␤1.
Keratinocyte migration and adhesion to the extracellular matrix are jointly regulated by integrins. Therefore, we determined the integrin-mediated adhesion properties of wild type and E2F-1 Ϫ/Ϫ keratinocytes toward various extracellular matrix proteins using conventional short term adhesion assays. Primary keratinocytes maintained as basal cells in 0.05 mM extracellular Ca 2ϩ for 3 days were re-plated onto untreated plates or plates coated with fibronectin, collagen IV, laminin-1, or laminin-5. Wild type cells adhered more efficiently to noncoated plates than E2F-1 Ϫ/Ϫ keratinocytes (Fig. 3). The numbers of wild type cells that adhered extracellular matrix-coated plates were substantially higher than those on uncoated culture dishes (Fig. 3). In contrast, E2F-1 Ϫ/Ϫ keratinocytes exhibited significantly smaller increases in adhesion in the presence of the extracellular matrix proteins (Fig. 3). Thus, decreased integrin expression appears to be a contributing factor to the impaired migration and adhesion exhibited by mutant keratinocytes.
Epidermis-specific E2F-1 Requirement for Normal Cell Cycle Progression-One of the most thoroughly characterized functions of E2F-1 is its ability to promote transit to S phase and cell proliferation (for review, see Ref. 1). In addition, integrin stimulation is necessary for normal keratinocyte growth (26,30). Hence, we next characterized the proliferative capacity of primary epidermal keratinocytes as well as dermal fibroblasts isolated from wild type, E2F-1 ϩ/Ϫ , and E2F-1 Ϫ/Ϫ mice.
We first measured the growth rates of primary cultured keratinocytes and dermal fibroblasts. Wild type and heterozygote E2F-1 ϩ/Ϫ keratinocytes proliferated at similar rates throughout the experiment (8 days), with a mean doubling time of about 33 h (Fig. 4A). In contrast, E2F-1 Ϫ/Ϫ keratinocytes did not complete a population doubling within 8 days after plating (Fig. 4A). Based on the experimental data obtained, the estimated doubling time of these cells in culture is 250 h, about 7-fold greater than that in wild-type cells. This failure of E2F-1 Ϫ/Ϫ keratinocytes to divide was accompanied by a reduced capacity to synthesize DNA (Fig. 4B). DNA synthetic capacity was not altered in E2F-1 ϩ/Ϫ keratinocytes, indicating the lack of gene dosage effects. The low proliferation rates and DNA synthetic capacity in E2F-1 Ϫ/Ϫ keratinocytes did not arise from increased cell loss, as evidenced by the ability to exclude trypan blue of Ͼ95% of cells in wild-type and E2F-1 Ϫ/Ϫ cultures (data not shown). Furthermore, this proliferation defect was specific to epidermal keratinocytes, as the doubling times and DNA synthetic capacity of E2F-1 Ϫ/Ϫ dermal fibroblasts were similar from those of heterozygous or wild type cells (Fig. 4, C  and D). Thus, E2F-1 plays a tissue-specific role in keratinocyte proliferation that other E2F forms cannot fulfil.
Prolonged Transit through G 1 and S Phases of the Cell Cycle in E2F-1 Ϫ/Ϫ Keratinocytes-To better understand the delayed proliferation response in E2F-1 Ϫ/Ϫ keratinocytes, we conducted flow cytometric analysis on wild type, E2F-1 ϩ/Ϫ , and E2F-1 Ϫ/Ϫ keratinocytes at various times after isolation. We reasoned that this would allow us to identify those phases of the cell cycle during which E2F-1 Ϫ/Ϫ cells transit abnormally and, consequently, the times during which E2F-1 activity is essential.
Immediately after isolation, about 85% of wild-type cells were in the G 0 /G 1 phase, constituting an almost fully synchronized population. This synchronized cell population entered the cell cycle when placed in culture, as indicated by progressive reduction in the G 0 /G 1 fraction until day 4 after isolation. After this time and probably coincident with the development of asynchronous growth, the keratinocytes adopted a profile in which the G 0 /G 1 fraction was maintained at about 35-40% (Fig. 5). The early decrease in G 0 /G 1 cells was accompanied by corresponding increases in cells traversing the S phase, which was eventually maintained at about 25%. Finally, wild type cells reached the G 2 /M phases about 4 days after plating. E2F-1 ϩ/Ϫ keratinocytes exhibited a profile indistinguishable from that described for wild type cells (Fig. 5).
The majority of E2F-1 Ϫ/Ϫ keratinocytes were also in the G 0 /G 1 phase at the time of isolation. In contrast with wild type and heterozygous null cells, E2F-1 Ϫ/Ϫ required about 4 days to reach the beginning of S phase. Moreover, having reached S-phase, they required a substantially extended period (Ն3 days) to transit to G 2 . Our flow cytometry analysis did not show cells with a DNA content lower than that of G 0 /G 1 cells, consistent with the concept that most of these keratinocytes are not undergoing apoptosis. Thus, the increased doubling time in E2F-1 Ϫ/Ϫ keratinocytes appears due to delayed exit from G 0 and/or transit through G 1 as well as a lengthening in the S phase.
Regulation of E2F Factors during Wound Repair-We have shown that maintenance of the proliferative state in primary keratinocytes is associated with the expression of E2F-1, E2F-2, and E2F-3. Keratinocyte entry into quiescence by differentiation stimuli or by growth inhibitory factors such as TGF-␤1 occurs with a concomitant down-regulation of these three proteins (14). Increases in keratinocyte proliferation and the ability to migrate are essential in vivo for epidermal regeneration after injury (17,18). Thus, by focusing on cutaneous wound repair, we examined in vivo the consequences of decreased keratinocyte proliferation and integrin signaling resulting from E2F-1 inactivation. To this end, we first determined by in situ hybridization the patterns of E2F gene expression in keratinocytes at wound margins and compared them to expression in adjacent resting skin. Similar to E2F expression in murine epidermis (31), very low levels of E2F-1 and E2F-2 mRNA, essentially confined to the basal layer, are detected in resting human epidermis (Fig. 6). In stark contrast, E2F-1 and E2F-2 mRNA levels increase dramatically in migrating keratinocytes at the wound edge, suggesting a possible involvement of these factors in epidermal repair. Notably, expression of E2F-3 and E2F-4 transcripts does not appear to change substantially in the migrating epithelium relative to resting epidermis (Fig. 6).

Targeted Inactivation of the E2F-1 Gene Impairs Wound
Healing in Vivo-We next compared healing rates after partial tail amputation in E2F-1 Ϫ/Ϫ and wild type mice (Fig. 7, Table  I). One day after surgery, extensive infiltration of inflammatory leukocytes was evident in normal mice, whereas E2F-1 Ϫ/Ϫ wounds showed reduced cell infiltration despite apparently normal clot formation. Only on day 3 after surgery, inflammatory cells in mutant wounds reached levels similar to those observed in wild type wounds on day 1 (Fig. 7).
Re-epithelialization in wild-type wound edges was first evident at day 3 post-surgery. By day 4, most of the wound was re-epithelialized. Complete re-epithelialization with stratification occurred 5 days after wounding (Fig. 7). Similar to the delayed initial inflammatory response, E2F-1 Ϫ/Ϫ wounds showed the first evidence of re-epithelialization only by day 4 or 5 and achieved full re-epithelialization by day 8 post-wounding (Fig. 7, Table I). This represents a delay as high as 60% compared with wild type animals. Significantly, no delays in stratification were apparent in E2F-1 Ϫ/Ϫ wounds once re-epithelialization had occurred. In addition, the observed healing impairment was similar in males and females. Thus, E2F-1 plays a novel, unique, and essential role in cutaneous wound healing.

DISCUSSION
The E2F family of transcription factors is tightly regulated during epidermal morphogenesis. Indeed, the transition from undifferentiated to terminally differentiated murine keratinocytes is accompanied by a switch in E2F gene expression, with down-regulation of E2F-1, -2, and -3 and up-regulation of E2F-5 (14,31). This pattern of E2F expression suggests that E2F-1 may be involved in maintaining a proliferative and/or undifferentiated state in epidermal keratinocytes.
To better define the biological role of E2F-1 in the epidermis, we first designed a screen to identify genes that specifically down-regulated E2F-1 Ϫ/Ϫ keratinocytes. Among the tran- scripts identified in this screen were integrins ␣ 5 , ␣ 6 , ␤ 1 , and ␤ 4 . Keratinocytes in the undifferentiated, basal layer of the epidermis attach to the basement membrane through interactions involving integrin receptors, including ␣ 6 ␤ 4 (a receptor for laminin), and several ␤ 1 integrins (receptors for fibronectin and collagen) (25,(32)(33)(34). In addition to their role in keratinocyte adhesiveness and migration, integrins play a role in keratinocyte stem cell maintenance, signaling via mitogen-activated protein kinase pathways (35), and ␣ 6 ␤ 4 and ␤ 1 integrins control keratinocyte proliferative potential (26,35). We demonstrate that abnormal integrin expression in E2F-1 Ϫ/Ϫ keratinocytes is accompanied by reduced extracellular matrix-stimulated signaling through mitogen-activated protein kinase pathways, including JNK and Erk activation. Notably, this defect is restricted to integrin-signaling pathways, given that E2F-1 Ϫ/Ϫ keratinocytes maintain normal JNK and Erk activation in response to stress, such as UV irradiation. Integrins are involved in spreading and migration of keratinocytes in response to extracellular matrix proteins (36), and alterations in integrin expression and signaling were associated with blunted chemotactic responses of E2F-1 Ϫ/Ϫ keratinocytes toward conditioned media or TGF-␤1. Again, this is not a generalized defect in TGF-␤ signaling given that mutant keratinocytes retained a normal proliferative response toward this growth inhibitor.
The mechanisms of down-regulation of integrins in E2F-1 Ϫ/Ϫ keratinocytes remain to be established. Our analysis of the literature and available sequence data bases indicate the involvement of various transcription factors in integrin gene expression (e.g. Pax-6 for integrins ␣ 5 and ␤ 1 , retinoic acid receptors for ␣ 6 and ␤ 4 , and Sp1 for ␣ 6 ) (37)(38)(39). This analysis failed to reveal the existence of E2F binding elements in the known 5Ј-flanking regions of the affected integrin genes. E2F-1 may regulates the expression of other transcription factors involved in integrin expression, although cDNA microarray analyses of cell lines overexpressing E2F-1 has failed to reveal any integrin gene responses (40,41).
Integrin signaling is involved in maintenance of keratinocyte stem cells, and keratinocyte differentiation is accompanied by down-regulation of integrin levels and signaling (35). The stem cell status of E2F-1 Ϫ/Ϫ keratinocytes has yet to be assessed. One possibility would be that reduced integrin function might alter keratinocyte differentiation at the expense of the undifferentiated cell population. However, E2F-1 Ϫ/Ϫ keratinocytes cultured under non-differentiating conditions do not express higher levels of the differentiation markers involucrin and keratin 1 relative to wild type cells. 3 Furthermore, induction of differentiation by treatment with 1 mM Ca 2ϩ induces similar increases in involucrin expression in mutant and wild type cells, 3 indicating that inactivation of E2F-1 and reduced integrin signaling are not sufficient to promote keratinocyte differentiation. An important role for E2F-1 in induction of keratinocyte proliferation has been demonstrated in transgenic mice overexpressing E2F-1 in basal keratinocytes in the epidermis. In addition to substantially increased keratinocyte apoptosis, these animals have hyperproliferative keratinocytes, which 3 A. Pajak and L. Dagnino, unpublished observations. FIG. 7. Impaired cutaneous wound healing in E2F-1 ؊/؊ mice. At day 2 post-wounding, wild type, but not E2F-1 Ϫ/Ϫ , wounds show extensive monocyte infiltration. At day 4, wild type wounds have substantially re-epithelialized, whereas E2F-1 Ϫ/Ϫ wounds show only the beginning of re-epithelialization. The extent of monocyte infiltration in day 4 E2F-1 Ϫ/Ϫ wounds is similar to that observed in wild type wounds at day 2. By day 5, wild type wounds have completely healed, whereas E2F-1 Ϫ/Ϫ wounds have only partially re-epithelialized. The arrows indicate the hyperproliferative front at the wound edges. E, eschar; HE, hyperproliferative epithelium; SE, stratified epithelium. Bar, 250 m. Inflammation (ϩ) Inflammation (ϩϩϩ) Granulation tissue (ϩϩ) Partial re-epithelialization 40-55% a Granulation tissue (ϩ) Granulation tissue (ϩ) Slight re-epithelialization 10-15% No re-epithelialization No re-epithelialization a Full re-epithelialization at days 7-8. eventually give rise to epidermal tumors (42)(43)(44). Consistent with these observations, we have demonstrated that ectopic expression of E2F-1 by adenovirus-mediated gene transfer increases both apoptosis and DNA synthesis in undifferentiated and differentiated cultured keratinocytes (14).
Another alteration consequent to E2F-1 inactivation is a decrease in proliferation specific to epidermal keratinocytes without effects on dermal fibroblasts. This demonstrates that different members of the E2F family have tissue-specific functions. Of note, E2F-1 Ϫ/Ϫ keratinocytes exhibited marked delays not only entering the cell cycle immediately after isolation but also traversing the S phase, suggesting that E2F-1 may play a direct role in DNA replication, as shown for example in the fruit fly (45).
Under normal conditions, other E2F factors such as E2F-2 and E2F-3 may primarily control morphogenesis and keratinocyte renewal in resting epidermis. However, cutaneous wound healing is a dynamic process that requires activation of epidermal and dermal cells for regeneration as well as stimulation of immune function. Multiple cell types participate in this process, including keratinocytes that re-epithelialize the wound, dermal fibroblasts that provide the extracellular matrix to support the keratinocytes, and monocytes, which infiltrate the wound and secrete stimulatory cytokines. Our studies indicate that E2F-1 is not essential for normal epidermal function. In contrast, E2F-1 is indispensable for skin regeneration. The delay in wound healing in E2F-1 Ϫ/Ϫ mice appears to occur mainly during the early stages or repair, namely during the inflammatory, proliferative, and migratory phases. Once the wound is re-epithelialized, epidermal stratification and remodeling in mutant animals is indistinguishable from that in normal mice. The signals induced by monocytes and macrophages are central for wound repair (18), and the chemotactic responses of these cells also appear to be diminished in the mutant mice. Therefore, the observed delay in wound repair in these animals may also involve hematopoietic tissues.
Generally, cutaneous wound healing occurs readily. However, deficient repair in cases of extensive wound burden or ulcerative situations can become a critical problem. The essential requirement for E2F-1 activity in epidermal repair makes the idea of pharmacological manipulation of this factor a novel, potentially effective approach in the management of poor wound healing.