Effect of β1 integrin deletion on adhesion of primary keratinocytes. (A-C) Surface expression of integrins in freshly isolated keratinocytes from 2-day-old K5β1-null mice (β1 Δ/Δ) and control littermates that were homozygous for the floxed β1 integrin allele but did not express Cre recombinase (β1 fl/fl), as determined by FACS analysis. Orange lines represent the second antibody control. Keratinocytes were stained with antibodies against β1, β4 and αv integrins. (D,E) Loss of focal adhesions and actin stress fibres in β1-null keratinocytes. Wild-type (D) and β1-deficient (E) keratinocytes were cultured for 2 days and examined for focal adhesions (green) and F-actin (red) by immunofluorescence staining with an antibody against paxillin or by staining with phalloidin, respectively. Note that the β1-null cells are completely rounded and that the green fluorescence is cytoplasmic paxillin. (F) Adhesion of wild-type and β1-deficient keratinocytes. Keratinocytes were plated onto 96-well plates (5×10⁴ cells/well), pre-coated with fibronectin (FN), laminin (LN), poly-D-lysine (PDL), collagen type I (COLL 1) and collagen type IV (COLL 4). Adhesion was quantified using a CytoTox 96 colorimetric kit. Error bars represent standard deviation of the mean of triplicate samples within one experiment.

Effect of β1 integrin deletion on keratinocyte migration in vitro. (A) β1 fl/fl keratinocytes were infected with an EGFPCre-expressing retrovirus and examined 5 days later. Flow cytometry demonstrates that the β1 integrin subunit is missing in 50% of EGFP-positive keratinocytes. (B) The GFP positive keratinocytes fail to spread. (C) β1 fl/fl keratinocytes transduced with chick β1 integrin. Red, second antibody control; blue, chick β1 integrin-positive cells. (D) Chick transduced keratinocytes remain spread after infection with the EGFPCre retrovirus. (E) Phase and photograph of β1 fl/fl keratinocytes infected with EGFPCre at t=0 hours of the movie. In colour are the tracks of two EGFPCre infected (dark and light blue tracks) and uninfected (red and green tracks) keratinocytes. For detailed migration movies, see http://dev.biologists.org/supplemental/.

Impaired re-epithelialisation in K5β1-null mice. Haematoxylin and Eosin staining of sections from 1-day wounds reveals similar histology between wounds of control (A) and K5β1-null (B) mice (arrows indicate the epithelial margin). By 5 days post-wounding, 60% of wounds to control mice had re-epithelialised, with the hyperproliferative epithelium clearly covering the wound bed (C), whereas those of their K5β1-null littermates (D) had never closed, with the hyperproliferative epithelium remaining almost static at the wound margins (n=10 for each). The tissue covering the wound bed in D is eschar, which will eventually be shed when the epithelial fronts meet several days later. D, dermis; E, epidermis; Es, eschar; F, fatty tissue; G, granulation tissue; HE, hyperproliferative epithelium; HF, hair follicle; M, muscle. Scale bars: 200 μm.

Cell proliferation in 5-day wounds. Mice were injected with BrdU 2 hours before sacrifice. Sections of 1-day (A,B) and 5-day (C,D) wounds from control mice (A,C) and K5β1-null mice (B,D) were stained with an anti-BrdU antibody (stained nuclei are indicated by arrowheads and the epithelial wound edge by arrows in A,B). There was an obvious impairment of keratinocyte proliferation in the hair follicles of K5β1-null mice (A-D; compare asterisks in A and B). G, granulation tissue; HE, hyperproliferative epithelium. Scale bars: 200 μm.

Cell-cell and cell-matrix interactions during wound repair. Immunostaining on sections from the middle of 10-day wounds, with an anti-β1 integrin antibody reveals expression in basal keratinocytes of the healed epidermis of control mice and confirms the absence of β1 in the epidermis of K5β1-null mice (arrows in A,B). Lack of β1 integrins also results in a disturbance in hemidesmosome localisation, as detected by immunostaining on sections from 5-day wounds for β4 integrin (red; arrows in C,D). The hyperproliferative epithelium is stained with an anti-keratin 6 antibody (green). G, granulation tissue; HE, hyperproliferative epithelium. This difference in hemidesmosome staining is recapitulated in the staining for the laminin 5 substrate (magenta) to which they bind (arrows in E,F). By contrast, expression of E-cadherin (magenta) is similar in the hyperproliferative epithelium of 5-day wounds of control and K5β1-null mice (G,H). G, granulation tissue; HE, hyperproliferative epithelium. Scale bars: 50 μm.

Structural analysis of the epithelial tongue. Light microscopical (A-D) and ultrastructural (E-H) views of the epidermal tongue at the margin of 5-day wounds as it cuts a path between the granulation tissue and the overlying eschar. The migrating keratinocytes of the hyperproliferative epithelium in control mice are relatively loosely associated with each other, showing extensive intercellular spacing (compare asterisk in E with that in F). By contrast, wound margin keratinocytes of K5β1-null mice are densely packed (compare A,C,E with B,D,F). The boxes represent the approximate areas shown in the higher magnification figures. The number of tonofilaments attached to the desmosomes was significantly reduced in K5β1-null mice (H) compared with controls (G) (indicated by arrows). Es, eschar; G, granulation tissue; HE, hyperproliferative epithelium; N, nucleus. Scale bars: 50 μm in A,B; 20 μm in C,D; 1 μm in E-H.

Histology of 15-day wounds reveals marked differences in the outcome of the repair process. Masson trichrome staining of 15-day wounds of control mice reveals complete re-epithelialisation (A). However, in several wounds of K5β1-null mice, the epithelial tongues fail to fuse and detach from the underlying granulation tissue (B) and, where fusion does occur, the epidermis is far thicker than that seen in control mice (compare A with C). G, granulation tissue; HE, hyperproliferative epithelium. Scale bars: 200 μm.

Repair of incisional wounds is delayed K5β1-null mice. Haematoxylin and Eosin staining reveals that full-thickness incisional wounds on the backs of 10 d.p.p. control mice had healed by day 3 post-injury (A; arrows indicate where the panniculus carnosus has been cut), but wounds of K5β1-null mice are wide open at the same stage (B). By 6 days after wounding, wounds of both control and K5β1-null mice have healed (C,D). Es, eschar; F, fatty tissue; G, granulation tissue; HE, hyperproliferative epithelium; HF, hair follicle; M, muscle. Scale bars: 200 μm in A-D.

Increased inflammation but normal angiogenesis in the wounds of K5β1-null mice. F4/80 staining for monocytes/macrophages (brown stain indicated by arrows in A and B) at day 2 post-wounding showed no difference. In 5-day wounds (C-F) there was a clear increase in neutrophil presence (as detected by staining for Ly-6G – brown stain indicated by arrows in C,D) in the wounds of K5β1-null mice (D) compared with control littermates (C). No difference was observed in angiogenesis as detected by staining for PECAM1 – magenta stain indicated by arrows in E,F). Es, eschar; F, fatty tissue; G, granulation tissue; HE, hyperproliferative epithelium; HF, hair follicle. Scale bars: 200 μm in A-D; 40 μm in E,F.

Gene expression during the repair process. Expression levels of genes encoding key players during wound healing were determined by RNase protection assay on 20 μg RNA samples from non-wounded back skin from 10 and 20 d.p.p. mice plus 1, 5 and 10 day wounds from control and K5β1-null mice. 1000 cpm of the hybridisation probes were loaded in the lanes labelled ‘probe’ and used as size markers. tRNA (20 μg) was used as a negative control. RNA (1 μg of each) was loaded on a 1% agarose gel and stained with Ethidium Bromide to control for sample integrity and concentration (bottom panel). The intensity of the signals as determined by phosphorimaging is shown schematically on the right-hand side. All protection assays were repeated with a separate pool of RNA samples from an independent skin/wound series.