Targeted Ablation of the Murine Involucrin Gene

The protective function of the mammalian epidermis depends largely on the presence of the chemically resistant envelope of the corneocytes. The envelope consists of proteins stabilized by ε(γ-glutamyl)lysine cross-links formed under the action of transglutaminase 1, an enzyme specific to keratinocytes (Rice and Green 1978; Simon and Green 1985; Thacher and Rice 1985). The protein envelope is covered with a monomolecular layer of N-(ω-hydroxyacyl)sphingosine bound to protein by ester bonds (Swartzendruber et al. 1987; Wertz and Downing 1987; Marekov and Steinert 1998).

Involucrin is synthesized in the outer living layers of terminally differentiating keratinocytes of all stratified squamous epithelia (Banks-Schlegel and Green 1981). Human involucrin has been shown to be a preferred substrate of keratinocyte transglutaminase and a precursor of the cross-linked envelope (Rice and Green 1979; Eckert et al. 1993; Yaffe et al. 1993). Expression of human involucrin in transgenic mice results in abnormalities of the skin and hair (Crish et al. 1993).

The nucleotide sequence of the involucrin gene has been determined in a large number of anthropoid and nonanthropoid mammals. In all species examined, the coding region of the gene contains a segment of short tandem repeats, which accounts for one half to two thirds of the coding region. The entire involucrin molecule, and particularly its segment of repeats, contains numerous glutamine residues (Eckert and Green 1986), clearly relating to the function of the protein as an amine acceptor in transglutaminase-catalyzed cross-linking (Tseng and Green 1988, Tseng and Green 1990; Phillips et al. 1990, Phillips et al. 1997; Djian and Green 1991; Djian et al. 1993).

Other precursors of the cross-linked envelope are loricrin (Mehrel et al. 1990), the small proline-rich proteins (SPRRs) (Kartasova and van de Putte 1988), and two membrane-bound proteins, periplakin and envoplakin (Simon and Green 1984; Ruhrberg et al. 1996, Ruhrberg et al. 1997). Recently, additional envelope constituents including S100 proteins have been described (Robinson et al. 1997). Whereas the role of transglutaminase 1 in the synthesis of the cornified envelope is clear (Jeon et al. 1998), the relative importance of the various substrates has not been established. Therefore, we decided to generate mice lacking involucrin by targeted ablation of the gene. Surprisingly, the absence of involucrin had no obvious effect on the cornified envelope, the morphology of the epidermis, or the hair follicles.

The whole involucrin coding region, which is confined to a single exon, was replaced with a neomycin phosphotransferase gene. We had isolated previously the involucrin genes of the NIH Swiss and BALB/c mice (Djian et al. 1993). To transform J1 embryonic stem (ES) cells, which are derived from a male agouti 129/SV mouse (Li et al. 1992), with isogenic DNA, we isolated the involucrin gene of a 129/SV mouse genomic library in the λFIXII vector (Stratagene). Screening of 10⁶ phage DNA plaques (Sambrook et al. 1989) with a 1.2-kb HindIII fragment containing most of the involucrin coding region identified two positive clones, of which clone λ17B was used to prepare the targeting vector. This clone contained the coding region flanked by 1.5 kb of upstream sequence and 11.5 kb of downstream sequence. The entire insert of λ17B was excised with Not1 (whose site is present in the λFIXII vector, on both sides of the insert) and subcloned into pBluescript. The resulting plasmid (psv6) was cut with Not1, Xba1, and Pst1 to excise most of the coding region and to subclone the flanking sequences. These sequences were isolated as a 1.7-kb Not1–Pst1 fragment (upstream) and a 10.6-kb Xba1–Not1 fragment (downstream), thus generating psv18 and psv26, respectively. Since psv18 retained 95 codons, including the initiator ATG, and since we were not certain that the encoded peptide could not act as a substrate in transglutaminase-catalyzed cross-linking, we deleted those codons with BAL31. The resulting clones were sequenced and one clone (psv23) was used in subsequent experiments; it lacked the entire coding region, the 3′ splice site, and a small 3′ part of the intron (see Fig. 1). The neomycin phosphotransferase gene under the control of the phosphoglucokinase promoter and followed by the phosphoglucokinase 3′ end, including the polyA addition site, was excised from pGKneo (a gift from Dr. Fred Alt, Children's Hospital, Boston, MA) with EcoR1 and HindIII. An Xho1 site was added immediately 3′ of the HindIII site by subcloning the EcoR1–HindIII fragment into pBluescript and reexcising it with EcoR1 and XhoI. The resulting fragment was cloned directly into the linker of psv23, in the right orientation and downstream of the involucrin 5′ sequence. The insert of the resulting plasmid was excised with Not1 and Xho1, blunt-ended with the Klenow fragment, and cloned with Xba1 linkers upstream of the Xba1–Not1 insert of psv26. The sequence flanking the Xba site of the linkers was such as to regenerate the Not1 site that had been partially abolished by treatment with the Klenow enzyme. The orientation of the Xba insert in the clones obtained was determined, and one of the constructs, psv29, was used to transfect ES cells. To facilitate homologous recombination, we excised the insert with Not1 and removed the pieces of linker on both ends by a brief treatment with BAL31, before transfection of the ES cells.

The J1 ES cell line was cultivated on a feeder layer of irradiated embryonic fibroblasts (6,000 rads). The fibroblasts were prepared from day 14 mouse embryos homozygous for the neomycin resistance gene. ES cells were grown in bicarbonate buffered DMEM supplemented with 15% FCS (heat inactivated at 56°C for 30 min), nonessential amino acids, 500 U/ml of murine leukemia inhibitory factor (ESGRO; GIBCO BRL), penicillin, streptomycin, and 0.11 mM 2-mercaptoethanol. For electroporation of DNA, ES cells were trypsinized, washed with electroporation buffer, and resuspended at 2 × 10⁷ cells/ml. The cells were mixed with 120 μg of DNA and electroporated at 0.4 kV and 25 μF, at room temperature, using a gene pulser apparatus (Bio-Rad Laboratories). After electroporation, the cells were cooled on ice for 10 min, and then plated on a feeder layer. After 24 h, the transfectants were selected in the presence of 0.4 mg/ml of G418 (GIBCO BRL). After 9 d of selection, individual drug-resistant clones were picked and transferred to 96-well plates. When the cells were about half confluent, they were trypsinized; half was left in the well and frozen at −120°C in the presence of 20% DMSO, whereas the remaining half was transferred to 24-well plates, for screening by Southern blotting. Positive clones were injected into BALB/c and C57BL/6 blastocysts at the Brigham and Women's Hospital Core Transgenic Mouse Facility, Boston, MA (by Dr. Arlene Sharpe).

DNA was prepared from ES cells or mouse tails by lysis in the presence of 100 mM Tris-HCl, pH 8.5, 200 mM NaCl, 5 mM EDTA, 0.2% SDS, and 0.1 mg/ml proteinase K. The lysate was incubated at 55°C for at least 6 h, and the DNA was precipitated in the presence of isopropanol, spooled out, and resuspended in 10 mM Tris-HCl, pH 8.0, 1 mM EDTA (Laird et al. 1991). Southern blots were performed as described in Tseng and Green 1988. The probe was a 1.0-kb PstI–HindIII fragment located immediately 5′ to the genomic fragment used to prepare the targeting vector (see Fig. 1). To clone the probe, we first constructed a library from size-fractionated genomic DNA. After digestion with Pst1 and electrophoresis of the digest, fragments of ∼2.6 kb were excised from the agarose gel and ligated to pBluescript. The ampicillin-resistant bacteria were divided into 22 pools, each containing ∼7,000 primary transformants. Screening by Southern blotting with the 1.25-kb Not1–Pst1 fragment located immediately upstream of the involucrin coding region revealed two positive pools. The fragment of interest was amplified by PCR, using an upstream primer located in the plasmid, and a downstream primer corresponding to codons 2–8 of the involucrin gene. The PCR product was cut with Pst1 and Hind III and the mixture of fragments was ligated directly into pGEM3z cut with PstI and HindIII. All recombinants obtained contained the fragment of interest (psv35).

For preparation of total skin RNA, mice were first anesthetized by intraperitoneal injection of tribromoethanol (Sigma-Aldrich). Hairy mice were depilated with a commercial preparation containing thioglycolic acid. The skin was then removed, frozen immediately in liquid nitrogen, pulverized in a mortar, resuspended in Tripure isolation reagent, and purified according to the manufacturer's recommendation (Boehringer). Total RNA was analyzed by electrophoresis in agarose gels containing formaldehyde, according to Sambrook et al. 1989, and RNA was transferred to nylon without prior treatment of the gel. Prehybridization, hybridization, and washing conditions were exactly as those described for Southern blotting. The probes consisted of a 1.2-kb fragment containing most of the involucrin coding region (Djian et al. 1993), a 0.9-kb BamHI–PstI fragment containing the 3′ half of the loricrin coding region (Mehrel et al. 1990), and a 0.7-kb EcoR1 fragment derived from the SPRR1 cDNA (Kartasova et al. 1996).

Protein extracts containing involucrin were prepared from mouse skin. Pulverized skin was prepared as described above for RNA. Approximately 400 mg of tissue was resuspended in 2 ml of a buffer containing 50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 2 mM EDTA, 0.1 mM leupeptin, 0.01 mM pepstatin, and 1 mM PMSF. The tissue was homogenized (Potter Elvekjem), clarified by two centrifugations at 120 g for 15 min and 48,000 g for 30 min, and the supernatant was collected. The protein concentration of the supernatant was determined with a protein assay kit using IgG as standards (Bio-Rad Laboratories). Protein extracts were stored at 4°C in the presence of 1% SDS. Partially purified mouse involucrin was prepared according to Etoh et al. 1986.

The proteins were resolved by SDS-PAGE, using 4 and 6% acrylamide plus bisacrylamide for the stacking and resolving gels, respectively (a ratio of 29:1 for both stacking and resolving gel). Immunoblot analysis was carried out according to Kahlem et al. 1998. The antiinvolucrin (Djian et al. 1993) and the secondary antibody were both added at a 1:4,000 dilution.

For sections stained with hematoxylin and eosin, the skin samples were first fixed in 10% buffered formalin and embedded in paraffin then stained. For immunofluorescence studies of involucrin, skin samples were snap frozen in OCT (Miles) and isopentane and cut with a cryomicrotome at 5 mm. Sections were fixed in acetone/methanol (1:1) at −20°C and stained with rabbit anti–mouse involucrin (Covance), then were detected with goat antiserum to rabbit IgG coupled to Alexa 488 (Molecular Probes). DNA was stained using Vectashield mounting medium containing DAPI (Vector Laboratories).

For preparation of envelopes, the tip of the ear was cut, placed in water containing 25 mM DTT and 2% SDS, heated to 100°C for 15 min, and centrifuged. The pellet was resuspended in 10 mM Tris-HCl, pH 8.0, and 1 mM EDTA. Envelopes were examined in a hemocytometer under phase microscopy.

The entire coding region of the involucrin gene was replaced with the neomycin phosphotransferase gene. ES cells were transformed with the construct and neomycin-resistant clones were screened by Southern analysis, using psv35 as probe (Fig. 1). Analysis of 297 neomycin-resistant clones disclosed 3 in which the involucrin gene had been ablated by homologous recombination (Fig. 2 A). All three clones (D₂D₂, D₃D₅, and A₁C₄) were injected into C57BL/6 and BALB/c blastocysts to generate chimeric animals. Chimeric males derived from clone A₁C₄ were backcrossed with either C57BL/6 or BALB/c females; both backcrosses yielded progeny of which 70–80% were agouti. As expected, probing of a Pst1 digest of genomic DNA with psv35 showed that agouti mice contained one ablated allele (2.15-kb fragment) and one normal allele (2.6-kb fragment) (Fig. 2 A). Reprobing of the same blot with psv18, which hybridizes to the construct itself (Fig. 1), showed that heterozygous mice contained no additional copy of a transgene that might have inserted randomly by nonhomologous recombination. Heterozygous mice were intercrossed to generate mice homozygous for the mutation. Absence of the involucrin coding region in the inv^−/− mice was confirmed by Southern blotting. Absence of involucrin mRNA and protein in the skin was confirmed by Northern and Western blot analyses (Fig. 2 B). Because nude mice have a thicker epidermis and more involucrin than hairy mice, we transferred the ablated allele to a nude background. nu/nu males (BALB/c) were crossed with inv^−/− females (C57BL/6), thereby generating F1 progeny that were inv^+/− and nu/+. The F2 progeny of F1 intercrosses yielded nu/nu, inv^−/− at the expected frequency of 1:16. Frozen sections of mouse skin (nu/nu, inv^−/− and nu/nu, inv^+/+) were stained with a rabbit antiserum prepared against recombinant involucrin (Covance). In the inv^+/+ mice, involucrin had the typical distribution in the outer third to half of the living layers, in the stratum corneum, and in the hair follicles; no involucrin was detected in the mice lacking the involucrin gene (Fig. 3).

Whether produced from backcrosses with C57BL/6 or BALB/c, the inv^−/− mice appeared normal and showed no obvious difference from their heterozygous or wild-type littermates. Inv^−/− females intercrossed with inv^−/− males produced litters of normal size. Among the 177 offspring examined from the heterozygous intercrosses, the numbers of wild-type, heterozygous, and inv^−/− mice were 54 (30.5%), 87 (49.2%), and 36 (20.3%), respectively, or close to the expected Mendelian ratio of 0.25:0.5:0.25.

Histological sections of the back skin of adult wild-type and mice lacking involucrin were compared. No differences were observed either in the epidermis or in the hair follicles (Fig. 4A and Fig. B). As mouse epidermis of the trunk is very thin and contains only two to three layers, we also examined epidermis of the tail, whose thickness is comparable with that of the human epidermis, and which possesses clearly distinguishable basal, spinous, granular, and cornified layers. Neither in the spinous and granular layers, where free involucrin is found, nor in the stratum corneum, where the involucrin is cross-linked in the envelopes, was there a perceptible difference between wild-type (Fig. 4C and Fig. E) and inv^−/− (Fig. 4D and Fig. F) littermates. The healing time of the wounds resulting from the skin biopsies used for histological analysis was compared in mice lacking involucrin and normal mice and was not found to be significantly longer in the former. The gross appearance of the esophageal, conjunctival, and corneal epithelia, which also contain involucrin, appeared unchanged in its absence.

Envelopes were prepared by removing the tips of ears of inv^−/− and wild-type mice, heating them to 100°C for 10 min in the presence of 2% SDS and 25 mM DTT, and examining the envelopes by phase–contrast microscopy. The envelopes prepared from inv^−/− mice were indistinguishable from those of wild-type mice (Fig. 5). Both strains of mice possessed a mixture of balloon-shaped and angular envelopes. Balloon-shaped envelopes probably originate from the outermost corneocytes of the epidermis, which have lost their desmosomes, whereas angular envelope are likely to have originated from more internal corneocytes, which were still attached to each other by the remains of desmosomes.

We compared the resistance of wild-type envelopes and envelopes lacking involucrin to various physical and chemical agents. We tested the effects of sonication, detergents (SDS, guanidine, and urea), proteases (trypsin and pronase), and grinding with a motor-driven homogenizer. Although all of these treatments caused progressive disruption of the envelopes, we could detect no difference in the resistance of wild-type envelopes and envelopes lacking involucrin.

In view of the lack of any detectable phenotype produced by the absence of involucrin, we thought that increasing levels of other envelope precursor proteins might have compensated for the absence of involucrin. The levels of the mRNAs encoding loricrin (Mehrel et al. 1990) and SPRR1 (Kartasova and van de Putte 1988; Kartasova et al. 1996) were found to be identical in the mice lacking involucrin and in the wild-type mice (Fig. 6). Therefore, the absence of involucrin is not compensated by increased levels of either loricrin or SPRR1 mRNAs.

Mutations in the human transglutaminase 1 gene result in lamellar ichthyosis (Huber et al. 1995) due to the absence of cross-linked envelopes (Jeon et al. 1998). Targeted disruption of the transglutaminase 1 gene in mice also causes the absence of envelopes, a severe disruption in barrier function, and early lethality (Matsuki et al. 1998). Therefore, it is clear that the cornified envelope is essential for the protective function of the epidermis, and that only transglutaminase 1 is essential for the synthesis of the envelope, since the other two transglutaminases present in keratinocytes (types 2 and 3) do not compensate for its absence.

A large body of data supported the earlier conclusion that involucrin was an essential component of the cross-linked envelope. The synthesis of involucrin is tightly regulated: it is restricted to terminally differentiated keratinocytes of stratified squamous epithelia. Human involucrin is a preferred substrate of transglutaminase 1, since after incubation of a crude extract of cultured epidermal cells in the presence of labeled putrescine and transglutaminase 1, involucrin was labeled at least 80 times more intensely than the average of the other cytosolic protein (Simon and Green 1985). Involucrin was also shown to promote the cross-linking of the particulate fraction containing membrane proteins. In other respects, involucrin is thought suitable as a protein able to cross-link other proteins (Yaffe et al. 1992). Involucrin is rich in glutamine residues (Djian et al. 1993; Eckert and Green 1986), a property in keeping with its function as an amine acceptor in transglutaminase-catalyzed cross-linking; other envelope precursors, such as loricrin, envoplakin, periplakin, or the small proline-rich proteins, are not glutamine-rich. For all of these reasons, it is surprising that envelopes should form normally in mice lacking involucrin. The presence of these cross-linked envelopes in inv^−/− mice clearly establishes that contrary to earlier suggestions (Steinert and Marekov 1997), whatever properties might be conferred on envelopes by involucrin, this protein is not essential for the assembly of envelopes, which in its absence consist exclusively of other proteins.

It remains to be seen whether more detailed examination of the skin of inv^−/− mice will reveal a detectable difference from that of wild-type mice. We have provided the inv^−/− mice to Dr. Peter Elias (University of California at San Francisco, San Francisco, CA) and to Drs. M. Jensen and E. Proksch (University of Kiel, Kiel, Germany) for detailed examination of the water vapor barrier and ultrastructural studies of the envelopes.

The disruption of the loricrin gene, which encodes another envelope precursor (Mehrel et al. 1990), also does not prevent envelope assembly, although it results in transitory redness of skin immediately after birth (Koch et al. 2000, this issue). It seems that whereas the transglutaminase is required in order to make the cross-linked envelope, neither involucrin nor loricrin is individually required. Mutations in the COOH-terminal part of human loricrin cause Vohwinkel syndrome, but the mutated alleles are dominant and the disease is likely to result from a gain of function (Maestrini et al. 1996).

The involucrin gene of the mouse is very polymorphic and this in itself would be compatible with the absence of strong selective pressure. However, this polymorphism includes the addition of numerous repeats that increase the size of the protein from 448 to 695 residues (Delhomme and Djian 2000). The involucrins of the BALB/c and C57BL/6 strains used in this study contain 448 and 468 residues, respectively. It is therefore necessary to suppose either that the increases in size occur in the absence of any demonstrated necessity for the protein, or that evolutionary development of a larger involucrin will eventually render the protein indispensable.

We thank Drs. Dennis Roop (Baylor College of Medicine, Houston, TX) and Tonja Kartasova (National Institutes of Health, Bethesda, MD) for their gifts of plasmids encoding loricrin and SPRR1. We are grateful to Dr. Chen Su (of Fred Alt's laboratory) for his advice and for his gift of pGKneo and to Dr. Arlene Sharpe for the injections and implantation of the blastocysts.

We are indebted for research support from the National Cancer Institute, the Centre National de la Recherche Scientifique, and the European Community (Biomed 2).