gap junctions are intercellular conduits through which small

The Connexin31 F137L mutant mouse as a modelfor the human skin disease Erythrokeratodermiavariabilis (EKV)

Marc Schnichels1, Philipp Worsdorfer1, Radoslaw Dobrowolski1, Christian Markopoulos1,

Markus Kretz1, Gabriele Schwarz1, Elke Winterhager2 and Klaus Willecke1,*

1Institut fur Genetik, Abteilung Molekulargenetik, Universitat Bonn, Bonn 53117, Germany and 2Institut fur Anatomie,

Universitat Duisburg-Essen, Essen 45122, Germany

Received October 27, 2006; Revised February 21, 2007; Accepted March 16, 2007

Erythrokeratodermia variabilis (EKV) is a rare autosomal dominant human genodermatosis. Its clinicalappearance varies from transient, fast moving erythemas to persistent brown hyperkeratoses. So far, severalmutations in the Cx31 or Cx30.3 gene have been reported to cause EKV in humans. We have generated a con-ditional mouse mutant that carries the human F137L mutation in the Cx31 gene which was described to act ina transdominant negative manner. The phenylalanine residue at position 137 is highly conserved in severalhuman and mouse connexin genes. Mouse embryonic stem (ES) cells expressing one allele of the Cx31F137Lmutation were stable but showed �30% decreased transfer of neurobiotin. This is probably due to dominantnegative effects of the Cx31F137L protein on wild type Cx31 and Cx43 protein expressed in ES cells.Surprisingly, the healing process of tail incision wounds in Cx311/F137L mice was shortened by 1 day, i.e.very similar as previously reported for mice with decreased expression of Cx43 in the epidermis. Thissuggests again that Cx31 and Cx43 proteins functionally interact, possibly by forming heteromeric channelsin the epidermis. Heterozygous Cx311/F137L mice are viable and fertile, in contrast to homozygous Cx31F137L/F137L

mice that die around ED 7.5. In Cx311/F137L mice, the epidermal expression pattern and level of Cx26, Cx30,Cx30.3 and Cx43 proteins were not altered compared with wild-type mice. No erythemas were detected inyoung C311/F137L mice before 2 weeks of age. In contrast to human EKV patients, hyperproliferation of thestratum germinativum was found in only 5% of the analyzed skin area.

INTRODUCTION

Connexins (Cx) are subunit proteins of gap junction channelswhich allow the intercellular diffusion of ions and metabolitesbelow a molecular mass of about 1.500 Dalton. Two hemi-channels, also called connexons, in contacting plasma mem-branes can dock to each other to form a gap junctionchannel. In heterotypic gap junction channels, each hemichan-nel is composed of a different connexin isoform. Heteromerichemichannels express more than one type of connexins (1).Different connexin channels can have distinct molecularpermeabilities (2).

During recent years, several human genodermatoses havebeen shown to be caused by mutations in Cx26, Cx30,Cx30.3 or Cx31 (3). Among these inherited diseases areErythrokeratodermia variabilis (EKV) which is characterized

by an unspecific appearance of transient erythemas andlocally defined or general hyperkeratosis. Sometimes, one ofthese features prevails or is completely absent. The variabilityof the erythematous patches regarding number, size and shapeis indicated by the name of the disease. So far, connexinmutations in EKV patients were described in 18 unrelatedfamilies, mainly from northern Europe. Seven mutations inthe Cx31 gene and five mutations in the Cx30.3 gene werefound to cause EKV. In one case, the appearance of EKVcould not be correlated with mutations in connexin genes(4). The clinical symptoms caused by these different connexinmutations are generally not distinguishable. All mutationsalter amino acid residues highly conserved in the beta sub-group of connexins, but up to now only mono-allelicmutations have been found. Dominant and transdominant

# The Author 2007. Published by Oxford University Press. All rights reserved.For Permissions, please email: [email protected]

*To whom correspondence should be addressed. Tel: þ49 228734210; Fax: þ49 228734263; Email: [email protected]

Human Molecular Genetics, 2007, Vol. 16, No. 10 1216–1224doi:10.1093/hmg/ddm068Advance Access published on April 19, 2007

Downloaded from https://academic.oup.com/hmg/article-abstract/16/10/1216/627846by gueston 04 March 2018

negative effects were discussed with Cx30.3 and Cx31 (5,6). Arecessive variant of EKV was also reported (7).

The F137L mutation was found both in the Cx31 and theCx30.3 gene of EKV patients. This mutation was caused bya single point mutation of T to C at position 409 which ledto an altered base triplet coding for leucine (L) instead ofphenylalanine (F). The phenylalanine residue at position 137is highly conserved throughout different connexins and differ-ent species. Patients suffering from EKV caused by theCx31F137L mutation show more pronounced phenotypicalterations of the skin (6).

Expression of the Cx31F137L mutated form in HeLa cellsled to the death of transfected cells (8). In contrast, no effectof the Cx30.3F137L mutation was found after transfection,although this mutation was also discussed to have transdomi-nant negative effects on Cx31 (5,8). Cx31 and Cx30.3 proteinscan build up heteromeric hemichannels which were shown tobe more stable and functional more efficient in dye couplingthan corresponding homomeric channels (5).

Previously, we had found that Cx31 deficient embryosshowed a reduced survival probability in utero associatedwith placental dysmorphogenesis but adult Cx31 deficientmice did not exhibit any epidermal abnormalities (9). Inorder to investigate the effects of the Cx31F137L mutationin living animals, we generated transgenic mice that expressmutated Cx31 instead of wild-type Cx31 under control ofthe endogenous Cx31 promoter.

Here, we describe that Cx31þ/137L mice show largely normalepidermal morphology. Surprisingly, the healing of incisionwounds in the tail of these mice was shortened by 1 day—similar as previously found with Cx43 deficient mice (10).This suggests that the mutated Cx31 protein can transdominantlyaffect the wild-type Cx43 protein in the epidermis.

RESULTS

Figure 1A illustrates the location of the F137L mutation in thethird transmembrane region of the Cx31 protein. Thismutation, which was identified in a human patient sufferingfrom EKV, was inserted into the corresponding mouse Cx31gene. Mouse and human Cx31 coding DNAs show 83%sequence identity. As shown in Figure 1B, the phenylalanineresidue at position 137 is highly conserved among several con-nexins of the beta subgroup.

Figure 2A illustrates schematically the construction of thetargeting vector and the recombined Cx31 locus after intro-duction of the Cx31F137L mutation. The mutation was gener-ated by overlapping PCRs and the mutated Cx31 DNA wasinserted into a targeting vector which contained a 50- (about2 kb) and a 30untranslated region (about 5 kb) of the Cx31locus flanking the endogenous Cx31 coding region (CR), aNeomycin (Neo) selection cassette and the mutated Cx31coding region (F137L). Since the Neo cassette was flankedby frt sites, it could be removed by Flp recombinase activity.The wild-type Cx31 coding region and the frt flanked Neo cas-sette were enclosed by loxP sites and thus could be deleted byCre recombinase activity. Thus, our cloning strategies allowedconditional expression of the Cx31F137L mutation if the cor-responding embryos turned out to be lethal.

Using the targeting vector for electroporation of mouseembryonic stem (ES) cells (11), recombinant clones were iso-lated at a frequency of �15%. ES cells from these clones were

Figure 1. (A) Schematic transmembrane topology of a connexin protein withits amino terminus (NT), cytoplasmatic loop (CL), the two extracellular loops(E1, E2), four transmembrane domains (M 1–M 4) and the carboxy terminus(CT). The positions of the mutations that cause EKV are indicated for Cx31(in black color) and Cx30.3 (in grey color). At positions 12 and 137, thesame mutations both in Cx31 and Cx30.3 cause EKV (6). (B) Sequence hom-ology of different connexin coding regions. The phenylalanine residue at pos-ition 137 of Cx30.3, Cx31 and Cx31.1 is highly conserved in different human(hCx) and mouse (mCx) connexins.

Figure 2. (A) Scheme for generating the Cx31F137L mutated allele by homolo-gous recombination: NdeI and SmlI are restriction sites, 50-Hom and 30-Hom: 50

and 30 homologous regions, respectively. (B) Southern blot hybridization ofdifferent mouse genotypes. The wild-type allele was detected at 15.8 kb. TheCx31F137L allele showed no difference in comparison to the wild-type allele,because the locus was only altered by one loxP site and a single base exchangecausing the mutation. The floxed allele was detected at 11.1 kb and the flox delNeo allele (Dneo) was found at 18.4 kb. (C) PCR, demonstrating the presence ofthe F137L mutation in the coding region of the Cx31 gene, following DNAdigestion with the endonuclease SmlI. The wild-type Cx31 allele yielded anamplicon of 809 bp. This amplicon could be digested with SmlI to yield twofragments of 401 and 408 bp, if the F137L mutation was present.

Human Molecular Genetics, 2007, Vol. 16, No. 10 1217


injected into mouse blastocysts, which were then transferredinto pseudo pregnant mice. Chimeric progeny due to agoutifur colors were bred to C57BL/6 mice and genomic DNA,digested with the restriction endonucleases NdeI and AgeI,was analyzed by Southern blot hybridization (Fig. 2B). Thefloxed Cx31 allele upstream (11.1 kb) and downstream(18.4 kb) of the Neo selection cassette could be distinguishedfrom the wild-type Cx31 allele (15.8 kb).

The presence of the Cx31F137L mutation was demonstratedby PCR of the Cx31 coding region following DNA digestionwith the endonuclease SmlI. An SmlI restriction site ispresent only in the mutated Cx31F137L but not in theendogenous Cx31 coding region (Fig. 2C). This mutationwas also verified by direct sequencing (data not shown).

Stably transfected Cx31F137L HM1 ES cells were furthertransiently transfected with a Cre encoding vector, in orderto excise the floxed endogenous Cx31 coding region withthe Neo selection cassette. After transfection, the mutationwas heterozygously expressed in ES cells. These cells weregrown to confluency and single cells were injected with neu-robiotin. These studies revealed that tracer coupling of cells,which heterozygously expressed this mutation Cx31þ/F137L,was decreased by �30% in comparison to untransfected ESCells (Cx31þ/flox) (Student’s t-test: P ¼ 0.0013) (Fig. 3).Since it had been suggested that the Cx31F137L mutationleads to the formation of hemichannels (12) in Ca2þ freemedium, hemichannel activity of mouse ES cells expressingCx31F137L was evaluated in Ca2þ free medium using theATP release assay. No significant change of the ATP concen-tration in the extracellular solution could be detected inCx31F137L expressing ES cells (Supplementary Material,Figure S1). Furthermore, the ATP release from Cx31þ/F137L

ES cells and wild-type ES cells was not significantly differentin Ca2þ containing medium.

Heterozygous Cx31þ/flox mice were bred with deleter flpmice (13) in order to obtain Cx31þ/floxdelNeo mice. Addition-ally, Cx31þ/flox mice were crossed with PGK-Cre mice to gen-erate Cx31þ/F137L mice. Mice of these genotypes were fertileand showed no obvious phenotypic abnormalities. However,as listed in Table 1, no Cx31F137L/F137L mice were born. Inorder to determine the time of death during embryonic

development, we analyzed the genotypes of embryos afterinterbreeding of Cx31þ/F137L mice. Living homozygousCx31F137L/F137L embryos were found on ED 4.5 and 7.5 butnot on ED 8.5 or later until ED 12.5. Thus, the presence oftwo Cx31F137L alleles appears to be lethal after ED 7.5. Incontrast, Cx312/F137L mice survived to adulthood.

Regarding investigations of mouse skin alterations, wefocused first on changes in connexin expression pattern andlevel. Immunofluorescence analyses of adult Cx31þ/F137L

mice revealed the normal expression pattern of Cx26, Cx30,Cx31 and Cx43 in mouse tail skin, suggesting that intracellulartransport of these connexins is not altered by the expression ofthe Cx31F137L mutation (Fig. 4). Furthermore the proteinlevels of Cx26, Cx30, Cx30.3, Cx31 and Cx43 were notchanged in Cx31þ/F137L mouse skin as demonstrated byimmuno blot analyses (Fig. 5). Double bands obtained withCx30.3 and Cx31 antibodies are probably due to unspecificimmunoreactivity, whereas the triple bands with Cx43 anti-bodies represent different states of phosphorylation. As acontrol, the level of tubulin was analyzed, in order tocompare the amount of protein loaded on each electrophoreticlane. The specificity of the antibodies was tested in wild typeand homozygously connexin deficient mice. As additionalcontrols, HeLa wild-type cells and the corresponding HeLa-connexin transfected cells were tested (Fig. 5), except ofHeLa-Cx30.3 cells where the Cx30.3 antibodies appeared toreact unspecifically with human proteins.

Tail incision wounds in Cx31þ/F137L mice were closedabout 1 day earlier than in wild-type mice (Cx31þ/þ).Fusion of the epidermis in the mutated animals led to distinctstratification at day 3 of the wound healing process (Fig. 6),whereas in wild-type mice this state was only reached at day4 after wounding. Alterations in the level of other epidermalconnexins during the healing of tail incision in wounds wereassayed and found to be similar as described by Kretz et al.(10). Hyperproliferation of epidermal keratinocytes wasobserved in only 5% of the mutated (Cx31þ/F137L) tail skin,but never in wild-type skin (Fig. 7). Thickening of the epider-mis was due to an extended stratum germination suggestinglocally enhanced proliferation of keratinocytes in this epider-mal area.

DISCUSSION

We chose the human Cx31F137L mutation to introduce it intothe Cx31 gene of a transgenic mouse, in order to generate amouse model for the human genodermatosis EKV. The samepoint mutation had been found in the Cx30.3 gene of apatient who also suffered from EKV. In order to explain theautosomal dominant inheritance of these mutations, it hadbeen suggested that the mutated Cx31 protein (or the

Table 1. Genotype frequencies of born mice. Wild type and Cx31þ/F137L ani-mals were born at the expected Mendelian ratios, but no Cx31F137L/F137L micewere found

Genotype þ/þ þ/F137L F137L/F137LNumber of animals 35 69 0Expected ratio 1/4 2/4 1/4

Figure 3. Transfer of neurobiotin in Cx31þ/F137L ES cells in comparison toCx31þ/flox ES cells. The columns represent the number of neurobiotin-coupledcells. Neurobiotin transfer is decreased by about one third when one allele ofthe Cx31F137L mutation is present (Student’s t-test: P ¼ 0.0013).

1218 Human Molecular Genetics, 2007, Vol. 16, No. 10


mutated Cx30.3 protein) interacts with and functionally inhi-bits wild type Cx31 and/or Cx43 protein (6). When humanCx31 EKV mutations were attempted to be expressed inHeLa cells, no transfectants carrying the point mutationcould be isolated (12,14). It was suggested that the mutatedCx31 protein can form hemichannels in transfected HeLacells, leading to extracellular leakage of ions and metabolitesand finally to lysis of the affected cells. Obviously, this resultcontrasted to the situation in the skin of EKV patients. By gen-erating a mouse model of EKV, we wanted to study theCx31F137L mutation in ES cells and in epidermal keratino-cytes of transgenic mice.

Our results with heteromeric mouse ES cells, which carryone allele of the Cx31F137L mutation, indicated that thecells are viable under cell culture conditions. They cancontribute to the germ line when introduced via manipulatedblastocysts into transgenic mice. Perhaps other connexins

(i.e. Cx31, Cx43 and Cx45), which are also expressed in EScells, may prevent the lytic effect of the single Cx31F137L

allele found in HeLa cells. All HeLa cells were transfectedwith vectors containing strong viral promoters (i.e. Cytome-galo Virus or Simian Virus 40) so that the expression levelmight have been higher than with the corresponding endogen-ous Cx31 promoter. Interestingly, Cx31þ/F137L ES cells show�30% less neurobiotin transfer to neighboring cells thancontrol ES cells (i.e. Cx31þ/flox).

No difference in ATP release, indicating hemichannelactivity, could be observed between the wild-type ES cellsor cells heterozygously expressing Cx31F137L besides Cx43and Cx45. Thus, we think it unlikely that phenotypic abnorm-alities observed in Cx31F137L expressing mice can beexplained by altered hemichannel activity. We suggest thatthe Cx31F137L mutated protein likely interacts with wild-type Cx31 and Cx43 protein, probably leading to closed

Figure 5. Several immunoblot analyses of Cx31þ/F137L skin in comparison to wild-type (þ/þ) mice revealed similar amounts of all five connexins investigatedin this study. The first lane of each set contained lysates of HeLa cells transfected with the corresponding connexin. The second lane was a HeLa wild-typecontrol to demonstrate that the antibodies did not react with lysates of non-transfected HeLa cells. The next two lanes contained lysates of tail skin of wildtype (third lane) or Cx31þ/F137L mice (fourth lane). In case of Cx30.3 only lysates of wild-type (þ/þ) and Cx31þ/F137L skin were analyzed.

Figure 4. Expression pattern of connexins in mouse tail skin based on immunofluorescence analyses (A–D): wild type, (E-H) Cx31þ/F137L. Cx26 and Cx30 wereexpressed in the upper-most layers of the epidermis. Occasionally, slight unspecific background fluorescence was noticed at the border between dermis and epi-dermis. Cx31 immunosignals were also mainly found in the upper epidermal layers. In contrast, Cx43 was expressed in each of the epidermal layers. In severalexperiments, no significant differences between connexin expression in Cx31þ/þ (A–D) and Cx31þ/F137L (E–H) epidermis could be observed. Immunofluor-escence was visualized in green and nuclei were stained with propidium iodide (red). Bar represents 20 mm.



homomeric and heteromeric channels, respectively. Alterna-tively, a trafficking defect of Cx31F137L containing connex-ons appears possible (12,14) which could lead to theobserved decrease in neurobiotin transfer. This conclusion isin accordance with our observations that Cx31þ/F137L EScells do not show increased secretion of ATP (throughhemichannels) in Ca2þ free extracellular medium relative towild-type ES cells (Supplementary Material, Figure S1).Previously, Elfgang et al. (15) had reported that mouseCx31 did not form heterotypic gap junction channels withCx43 when expressed in transfected HeLa cells. However,the formation of heteromeric channels was not investigatedin this study. Plantard et al. (5) have demonstrated that

Cx31 and Cx30.3 protein can stabilize each other in hetero-meric channels of HeLa cell transfectants. In order toexclude that wild-type Cx31 protein can stabilize theCx31F137L protein, we raised Cx312/F137L mice. Thesemice were viable and fertile. Interestingly, homozygousCx31F137L/F137L mice died around ED 7.5. Since Cx312/F137L

mice survived to adulthood, it cannot be the wild-type Cx31protein that stabilizes the Cx31F137L allele, but presumablyother connexin isoform(s) which are expressed in the epider-mis. The molecular reasons for the early embryonic death ofCx31F137L/F137L mice are not known. We speculate that theincreased dosage of the mutated protein in homozygousCx31F137L/F137L embryos may lead to more extensive func-tional inhibition of heteromeric channels.

Since hyperkeratosis of EKV is characterized in humans byerythemas and hyperproliferation of the skin, we carefullyinspected the Cx31þ/F137L mice for these symptoms. Inyoung mice (up to two weeks after birth), no erythemaswere found. However, during inspection of tail skin, rarecases of hyperproliferation in the stratum germinativum (in�5% of the inspected skin area) were seen in Cx31þ/F137L

mice, but not in Cx31þ/þ control mice. Thus, the hyperproli-ferative effect of the Cx31F137L mutation in the skin of trans-genic mice appears to be largely compensated by effects ofother genes. No effect of the Cx31F137L protein on thelevel and localization of other epidermal connexin isoformswas seen in the skin of Cx31þ/F137L mice. This is in accord-ance with the interpretation that only the activity of

Figure 6. Closure of tail wounds in Cx31þ/F137L mice (A), in comparison to wild-type Cx31þ/þ mice (B). On day 3, the wound area in Cx31þ/F137L skin wasalready multilayered with punctate Cx31 immunosignals in the reformed epidermis. In contrast, the stratification of the wounded epidermis in wild-type skin hadnot yet occurred on day 3, and Cx31 immunosignals were sparse. In Cx31þ/F137L mice, wound healing occurred about 1 day faster, i.e. on day 3 it was equivalentto Cx31þ/þ wounds on day 4 after wounding. Nuclei and coagulum in the wound were stained with propidium iodide in red. Bar represents 20 mm.

Figure 7. Hyperproliferation of keratinocytes was observed in mutated(Cx31þ/F137L) tail skin (A), but not in wild-type skin (B). The stratum germi-nativum was extended in �5% of tail skin areas in Cx31þ/F137L mice. Micro-graphs show immunofluorescence analyses using anti-Cx31 (green) andpropidium iodide staining of nuclei (red). Bar: 20 mm.



Cx31F137L containing heteromeric gap junction channelsis affected, but not their intracellular transport and locali-zation or that of connexons composed of other epidermalconnexins (16).

Previously, a transgenic mouse line had been described (17)that expressed a mutant connexin26 (D66H). The correspond-ing mutation in the human Cx26 gene causes a dominant gen-odermatosis, the Vohwinkel syndrome. This disease and thephenotype of the transgenic mouse model are characterizedby thickening of the epidermal cornified layers and accumu-lation of the mutated Cx26 protein together with wild-typeCx30 protein in the cytoplasm of keratinocytes. Furthermore,increased apoptosis was found in suprabasal keratinocytes ofthe mouse mutant. In Cx31þ/F137L mice of this study, no gen-eralized thickening of the stratum corneum was seen, althoughthis is commonly found in patients suffering from erythroker-atodermia variabilis. The Cx31þ/F137L mouse mutant shouldbe genetically very similar as the corresponding humanpatients, since the transgene is expressed at the homologousgenomic position from the endogenous Cx31 promoter, likethe Cx31þ/F137L mutation in patients. Thus, the relativelyweak keratodermic abnormalities in Cx31þ/137L mice mayreflect a difference in the function of the Cx31 protein inhumans and mice. Furthermore, it is possible that theCx31þ/F137L mutation shows stronger keratodermic effects inother mouse strains.

Surprisingly, we found that the healing process of incisionwounds in tail skin was shortened by 1 day in Cx31þ/F137L

mice when compared with wild-type mice. This result is remi-niscent to recent findings that induced downregulation of Cx43(10) or decreased expression of Cx43 due to localized appli-cation of antisense RNA to the skin of wild-type mice (18)resulted in a premature closure of the epidermal wound, com-pared to non-treated control mice. Our data suggest that theinteraction of Cx31 and Cx43 may lead to this effect. Kretzet al. (10) showed that downregulation of the Cx31 proteinoccurred at about the same time as the decreased Cx43protein levels, i.e. 1 day after wounding. The transdominantCx31F137L mutation may have the same inhibitory effecton wild-type Cx31 as downregulation of Cx43 by inducedablation (10). None of the other epidermal connexins wasaltered in its expression level in Cx31þF137L mice (Fig. 5).Dye transfer in the epidermis of Cx312/2 mice was hardlyaltered, in contrast to Cx43 ablated mouse epidermis whereit was strongly reduced (10). Which of the different steps ofwound healing is affected by transdominant interaction ofCx31 and Cx43 or by the decrease of Cx43 in mouse epider-mis needs to be clarified by further investigations.

MATERIALS AND METHODS

Construction of the Cx31F137L targeting vector

The blunted EcoRI–BamHI fragment (2202 bp) from themouse Cx31 gene locus (19, Fig. 1) was inserted into theNotI restriction site of the vector pBSK:lox-HPRT-lox(5693 bp), (20, modified in our laboratory) resulting invector A1 (7905 bp). Then, another BamHI-Hind III fragment(2369 bp) from the Cx31 gene locus (19, Fig. 1) was clonedinto pBluescript (2961 bps, Stratagene, La Jolla, USA) at the

identical restriction sites. This vector B1 was cut by SalI, fol-lowed by insertion of a SalI fragment from the vectorpBSK:frt-neo-frt (4796 bp; 13), containing the frt flanked neo-mycin resistance cassette, in order to generate vector B2(7214 bp). Cutting out the HPRT minigene of vector A withEcoRI and inserting a NotI-XhoI fragment of vector B2 ledto vector AB which now contained the 50-homologousregion, the wild-type Cx31 coding region and the frt-flankedNeomycin resistance cassette flanked by loxP sites. TheBamHI-XbaI F137L fragment (2677 bp), generated via PCRusing DNA of the Cx31 gene locus (19, Fig. 1), containedthe Cx31 coding region with the point mutation T to Cleading to the amino acid shift of F to L. For construction ofthe mutated coding region, we used the following PCRusing primers Eco_mCx31_F (50-G GAA TTC GGC ACCATG GAC TGG AAG AAG CTC-30) and mCx31F137LP2(50-GAT GAG CTT GAG GAT GAG GCT-30) orXba_mCx31_R (50-TGC TCT AGA AAT GGG GGT CAGGCT GGG CGC TGA-30) and mCx31F137LP3 (50-AGCCTC ATC CTC AAG CTC ATC-30). The two PCR fragmentswere fused together via a PCR using primers Eco_mCx31_F(50-G GAA TTC GGC ACC ATG GAC TGG AAG AAGCTC-30) and Xba_mCx31_R (50-TGC TCT AGA AAT GGGGGT CAG GCT GGG CGC TGA-30). Conditions for allthree PCRs were: 5 min at 948C, 30 cycles of 1 min at948C, 1 min at 558C, 1 min at 728C, followed by 10 min at728C. PCRs were carried out in 25 ml Taq buffer, containing1.5 mM MgCl2, 1.75 units of Taq polymerase (Promega,Germany), 0.2 mM dNTPs and 20 pmol of each primer.After the PCR reaction, the resulting fragment was digestedwith BamHI as well as XbaI and cloned into the pBSKvector (2961 bp, Stratagene) which was cut with the sameenzymes, resulting in vector C1 (5626 bp). DNA of the Gjb3locus (9, Fig. 1) was cut with XbaI and the resulting fragment(4075 bp) was used as 30-homologous region, inserted intoXbaI-cut C1 DNA, resulting in vector C2 (9701 bp). Thevector AB was digested with XhoI, the vector C2 with ClaIand NotI. Afterwards both fragments were blunted andligated to the final Cx31F137L targeting vector (16394 bp).

Screening of ES cell clones

ES cell culture, transfection and analyses were performed asdescribed (11,21). G418 resistant ES cell clones were screenedfor homologous recombination by PCR and were subsequentlyconfirmed by Southern blot hybridization.

The PCR with extracts from Cx31F137L ES cells indicatedhomologous recombination using the primers Cx31_hom_-For3 (50-CGA AGA TTC TTG CAG TAA TCC A-30) andCx31_hom_Rev (50-TAG TTC TAG AGC GGC CAATTC-30). PCR conditions were: 3 min at 948C, 35 cycles of1 min at 948C, 3 min at 668C, 1 min at 728C, followed by10 min at 728C. PCR was carried out in 50 ml RedTaq-Mix(Sigma, Germany), containing 50 pmol of each primer and25 ml water.

The Cx31F137L positive ES cell clones were further ana-lyzed by Southern blot analysis of DNA from lysates ofabout 15 million cells. Digestion of the PCR amplicon wasperformed by NdeI and AgeI. The probes were generated viaPCR using the vector Gx2 (19) for the internal and for the



external probe. The internal probe was generated using theprimers probe_int_for (50-CCA CGT GGT TGC TAG TATTCG-30) and probe_int_rev (50-GAG AAT CTG ATA ATCAAA CGA C-30), the external probe using primers pro-be_ext_for (50-GTT GTC TCT AGA TCT ATA ACC-30)and probe_ext_rev (50-GTG TTC TAG AAT GAA TGAAGA C-30). PCR conditions were: 5 min at 948C, 30 cyclesof 1 min at 948C, 1 min at 558C, 1 min at 728C, followed by10 min at 728C. PCR was carried out in 25 ml Taq buffer, con-taining 1.5 mM MgCl2, 1.75 units of Taq polymerase(Promega, Germany), 0.2 mM dNTPs and 20 pmol of eachprimer.

Characterization of Cx311/floxF137L ES cell clones

The homologously recombined ES cells were cultivated ongelatinized cell culture dishes (Falcon, USA) and the transienttransfection with the vector pCre-Pac was carried out by elec-troporation as described by Eckardt et al. (22).

Neurobiotin (N-2 [aminoethyl]-biotinamide hydrochloride;Vector Laboratories, Burlingame, USA) was iontophoreticallyapplied for 2–4 s in 0.1 M Tris buffer, pH 7.6, using positivecurrent of 20 nA (Ionophoresis Programmer model 160, Worldprecision instruments inc., New Haven, USA). During injec-tion, the cell culture dishes were kept on a heated block at378C. Five minutes after injection, cells were washed twicewith PBS, fixed for 10 min in paraformaldehyde (4%)/0.2%picrinic acid in sodium phosphate buffer (0.15 mM, pH 7.4),washed twice with PBS, incubated in 0.4% Triton X-100,washed with PBS, incubated with horseradish peroxidase-avidin D (Vector Laboratories) for 2 h, washed with PBS,incubated in 0.05% diaminobenzidine, 0.003% hydrogenperoxide solution for 15 min and examined using an invertedmicroscope (IM35, Zeiss, Oberkochem, Germany).

No difference in morphology was noticed during each seriesof microinjections with neurobiotin (up to about 1 h) in ES-cell transfectants.

STATISTICS

All data are represented as mean + standard deviation (SD).Each of the samples was tested for normality using theGauss grid method. Samples were subsequently comparedusing Student’s t-test. Samples were considered to be signifi-cantly different if P , 0.05.

Extracellular ATP measurements

The ES cells were cultured in gelatinized 35 mm plates to aconfluence of 40–50%, washed twice with PBS and incubatedwith 500 ml of Hank’s balanced salt solution (HBSS) with1 mM EGTA for 25 min at 378C and 5% CO2. After thisstimulation, 100 ml of the supernatant were collected and incu-bated for 20 min with 100 ml of the nucleotide releasingreagent from the ViaLight

TM

HS kit (Cambrex, East Rutherford,USA). Finally, the ATP concentration was measured for 10 swith the Berthold Microplate LB96V luminometer whichautomatically added 20 ml of the ATP monitoring reagent.The ATP release could be stimulated by the addition of

HBSS with 1 mM EGTA and could be decreased to a basiclevel by washing cells with PBS- and incubating them in EScell medium. All results of ATP release were normalizedafter determination of the total protein amount. Results wereexpressed as a mean + standard error of the mean (SEM),of relative luciferase activity.

Animals

Mice were kept in accordance with local governmental andinstitutional instructions for animal care and maintainedunder a 12/12 hours light/dark cycle. All transgenic micewere of mixed genetic background (at least 75% C57BL/6and at most 25% 129ola). In order to minimize the possibleinfluence of the genetic background, we performed all ana-lyses with littermates of heterozygote breedings, includingcontrol animals that lacked the Cre allele.

Genotyping of transgenic mice

Genotyping of Cre mediated deletion of the floxed region wasperformed as described previously (21). Cx31F137L, Cx31flox

and Cx31þ alleles were detected by the Cx31F137L PCR,using primers GenotypCreFor’ (50-CTC AAA GCT AGTCTG AGA TGC-30), Cx31hmrev (50-TAG TTC TAG AGCGGC CAA TTC-30) and GenotypCreRev (50-GCA TCACAA GGC TCC TAA GAA-30). PCR conditions were:5 min at 948C, 30 cycles of 1 min at 948C, 1 min at 608C,1 min at 728C, followed by 10 min at 728C. PCR wascarried out in 25 ml Taq buffer, containing 1.5 mM MgCl2,1.75 units Taq polymerase (Promega, Germany), 0.2 mM

dNTPs and 20 pmol of each primer. The F137L mutationwas detected by the Cx31CR-PCR, using primersCx31CR_For (50-GTT CCC TCA GGT GGG CAC AGC-30)and Cx31CR_Rev (50-CGG CTT CAC CCC TTC TCTAGC-30) followed by SmlI digestion. PCR conditions were:4 min at 948C, 35 cycles of 30 s at 948C, 1 min at 558C,1 min at 728C, followed by 10 min at 728C. PCR wascarried out in 50 ml RedTaq-Mix (Sigma, Germany), contain-ing 25 pmol of each primer and 25 ml water. After completingthe PCR 5 ml ‘buffer 4’, 1 ml of SmlI and 5 ml 10-fold concen-trated BSA (all from New England Biolabs, Ipswich, USA) areadded and the complete mix was incubated for 2 h at 558C.The PCR product was demonstrated as an 809 bp amplicon,the presence of the mutation was proven by two additionalbands at 408 bp and 401 bp.

The Cx31floxdelNEO allele was detected by theCx31F137L-NEO PCR, using primers Neo For neu (50-CGTATA TCC TTG ACA TAT CCA-30), Neo Rev 1 (50-GAAGTT TTT CCT GTC ATA CTT-30) and Neo Rev 2 (50-TTCGAG GTC GAC TCC ACC GCG-30). PCR conditions were:5 min at 948C, 30 cycles of 1 min at 948C, 1 min at 608C,1 min at 728C, followed by 10 min at 728C. PCR wascarried out in 25 ml Taq buffer, containing 1.5 mM MgCl2,1.75 units Taq polymerase (Promega, Germany), 0.2 mM

dNTPs and 20 pmol of each primer.Southern blot analysis of the Cx31þ/þ, Cx31þ/F137L, Cx31þ/flox,

Cx31flox/flox, Cx31þ/floxdelNeo, Cx31floxdelNeo/floxdelNeo geno-types were performed after NdeI and AgeI digestion of liver



DNA and hybridization with a probe spanning the completecoding region of the Cx31 gene.

Immunoblot analyses

Wounded or uninjured pieces of tail skin were dissected on iceand immediately frozen in liquid nitrogen. Homogenizedtissue was taken up in protein lysis buffer [60 mM Tris HCl,pH 7.4, and 3% sodium dodecyl sulfate (SDS)], supplementedwith proteinase inhibitor ‘Complete’ (Roche, Mannheim,Germany) and sonicated three times for 20 s on ice. Proteinconcentration was determined using the assay with bichincho-ninic acid (Sigma, Taufkirchen, Germany). For electrophoresis,50 mg of protein were separated on 10% SDS-polyacrylamidegels. Proteins were electroblotted on nitrocellulose membranes(Hybond ECL, Biosciences, Bucks, UK) for 2 h at 100 V.Membranes were blocked for 1 h with blocking solution[20 mM Tris, HCl, pH 7.4, 150 mM NaCl, 0.1% Tween 20(TBS-Tween)] and 5% skim milk powder (w/v) and after-wards incubated with primary antibodies overnight at 48C.The rabbit antibodies were diluted in blocking solution[anti-Cx43: 1:500 (23); anti-Cx31: 1:250 (24); anti-Cx30.3:1:500 (Zymed); anti-Cx30: 1:250 (Zytomed, Berlin,Germany); anti-Cx26: 1:500 (Zytomed)]. After washing for30 min in blocking solution, membranes were incubatedwith anti-rabbit horseradish peroxidase-conjugated secondaryantibodies at room temperature for 1 h (Dianova, Hamburg,Germany), diluted 1:20000 for anti-Cx43 and 1:5000 for theother antibodies in blocking solution. Afterwards, membraneswere washed for 1 h in TBS-Tween and incubated with anECL chemiluminescence detection system (Amersham Bio-sciences). In order to check equal loading in all lanes of theimmunoblot, we performed Ponceau staining of the blottingmembranes, followed by densitometric analyses of thestained protein. Lysates of HeLa cells stably transfected withthe corresponding connexin and lysates of untransfectedHeLa cells were used as positive and negative controls.

Immunofluorescence analyses

Cryosections (5 mm) of tail epidermis were stained with thefollowing rabbit antibody solutions: anti-Cx43 (1:2000; 23),anti-Cx31 (1:100; 24), anti-Cx30 (1:300; Zytomed),anti-Cx26 (1:500; Zytomed). Primary antibodies weredetected with Alexa594 conjugated goat anti-rabbit immuno-globulin (1:2000; MoBiTech, Goettingen, Germany). Forimmunofluorescence analyses, cryosections were fixed in100% ethanol at 2208C for 5 min, blocked with 4% bovineserum albumin (BSA, PAA Laboratories GMBH, Linz,Austria) in phosphate buffered saline (PBS) and incubatedwith antibodies diluted in 0.4% BSA in PBS overnight at48C. Afterwards, sections were washed with 0.4% BSA inPBS and incubated with Alexa conjugated antibodies for 1 hat room temperature. Nuclear staining was performed for15 min with propidium iodide (0.2 mg/ml) in PBS. Sliceswere mounted with fluorescent mounting medium (Dako,Glostrup, Denmark).

Antibody stained tissue slices were analyzed using thephotomicroscope Axiophot or the Laser Scan Microscope(both from Zeiss, Jena, Germany). Sections of tissues from

the corresponding connexin deficient mice were used as nega-tive controls. HeLa cells stably transfected with the corre-sponding connexin expression vector or sections of connexinexpressing tissues were used as positive controls.

Preparation of skin sections and histochemistry

Mice were killed by cervical dislocation. The tail skin was dis-sected, immediately frozen in liquid nitrogen and stored at2708C. Tail epidermis was cut lengthwise (10 mm sections)with a cryostat.

Wounding and isolation of mouse tail skin

Incision wounds into mouse tail were cut with a scalpel, aspreviously described for rat tail (25). Six to eight transversesections through the tail skin with a length of 1 cm were per-formed per mouse. Two of them were processed for immuno-fluorescence, two for histological and four were used forimmunoblot analyses. At each time point after wounding,mutated and control mice were analyzed.

SUPPLEMENTARY MATERIAL

Supplementary Material is available at HMG Online.

ACKNOWLEDGEMENTS

We gratefully acknowledge the technical assistance ofChristine Siegmund. Furthermore, we like to acknowledgediscussions with Dr Gabriele Richard (Philadelphia) to selectthe F137L EKV mutation for our studies. Work in the Bonnlaboratory was supported by a grant of the research group onkeratinocytes from the German Research Association to K.W.

Conflict of Interest statement. None declared.

REFERENCES

1. Evans, W.H. and Martin, P.E. (2002) Gap junctions: structure andfunction (Review). Mol. Membr. Biol., 19, 121–136.

2. Bedner, P., Niessen, H., Odermatt, B., Kretz, M., Willecke, K. andHarz, H. (2006) Selective permeability of different connexin channelsto the second messenger cyclic AMP. J. Biol. Chem., 281, 6673–6681.

3. Richard, G. (2005) Human connexins in skin development and skindisorders. In Winterhager, (Eds), Gap Junctions in Development andDisease. Springer Verlag, Berlin, pp. 135–171.

4. Arita, K., Akiyama, M., Tsuji, Y., Onozuka, T. and Shimizu, H. (2003)Erythrokeratoderma variabilis without connexin 31 or connexin 30.3 genemutation: immunohistological, ultrastructural and genetic studies.Acta Derm. Venereol., 83, 266–270.

5. Plantard, L., Huber, M., Macari, F., Meda, P. and Hohl, D. (2003)Molecular interaction of connexin 30.3 and connexin 31 suggests adominant-negative mechanism associated with erythrokeratodermiavariabilis. Hum. Mol. Genet., 12, 3287–3294.

6. Richard, G., Brown, N., Rouan, F., Van der Schroeff, J.G., Bijlsma, E.,Eichenfield, L.F., Sybert, V.P., Greer, K.E., Hogan, P., Campanelli, C.et al. (2003) Genetic heterogeneity in erythrokeratodermia variabilis:novel mutations in the connexin gene GJB4 (Cx30.3) andgenotype-phenotype correlations. J. Invest. Dermatol., 120, 601–609.

7. Gottfried, I., Landau, M., Glaser, F., Di, W.L., Ophir, J., Mevorah, B., BenTal, N., Kelsell, D.P. and Avraham, K.B. (2002) A mutation in GJB3 isassociated with recessive erythrokeratodermia variabilis (EKV) and leadsto defective trafficking of the connexin 31 protein. Hum. Mol. Genet., 11,1311–1316.



8. Rouan, F., Lo, C.W., Fertala, A., Wahl, M., Jost, M., Rodeck, U.,Uitto, J. and Richard, G. (2003) Divergent effects of two sequencevariants of GJB3 (G12D and R32W) on the function of connexin 31in vitro. Exp. Dermatol., 12, 191–197.

9. Plum, A., Winterhager, E., Pesch, J., Lautermann, J., Hallas, G.,Rosentreter, B., Traub, O., Herberhold, C. and Willecke, K. (2001)Connexin31-null mutation in mice causes transient placentaldysmorphogenesis but does not impair hearing and skin differentiation.Dev. Biol., 231, 334–347.

10. Kretz, M., Euwens, C., Hombach, S., Eckardt, D., Teubner, B., Traub, O.,Willecke, K. and Ott, T. (2003) Altered connexin expression and woundhealing in the epidermis of connexin deficient mice. J. Cell Sci., 116,3443–3452.

11. Theis, M., Magin, T.M., Plum, A. and Willecke, K. (2000) General or celltype-specific deletion and replacement of connexin-coding DNA in themouse. Methods, 20, 205–218.

12. He, L.Q., Liu, Y., Cai, F., Tan, Z.P., Pan, Q., Liang, D.S., Long, Z.G.,Wu, L.Q., Huang, L.Q., Dai, H.P. et al. (2005) Intracellular distribution,assembly and effect of disease-associated connexin 31 mutants in HeLacells. Acta Biochem. Biophys. Sin. (Shanghai), 37, 547–554.

13. Rodriguez, C.I., Buchholz, F., Galloway, J., Sequerra, R., Kasper, J.,Ayala, R., Stewart, A.F. and Dymecki, S.M. (2000) High-efficiencydeleter mice show that FLPe is an alternative to Cre-loxP. Nat. Genet.,25, 139–140.

14. Diestel, S., Richard, G., Doring, B. and Traub, O. (2002) Expression of aconnexin31 mutation causing erythrokeratodermia variabilis is lethal forHeLa cells. Biochem. Biophys. Res. Commun., 296, 721–728.

15. Elfgang, C., Eckert, R., Lichtenberg-Frate, H., Butterweck, A., Traub, O.,Klein, R.A., Hulser, D.F. and Willecke, K. (1995) Specific permeabilityand selective formation of gap junction channels in connexin-transfectedHeLa cells. J. Cell Biol., 129, 805–817.

16. Common, J.E., Di, W.L., Davis, D., Galvin, H., Leigh, I.M.,O’ Toole, E.A. and kelsell, D.P. et al. (2003) Cellular mechanismsof mutant Connexins is skin disease and hearing loss. Cell Comm. Adhes.,10, 347–359.

17. Bakirtzis, G., Choudry, R., Aasen, T., Shore, L., Brown, K., Bryson, S.,Forrow, S., Tetley, L., Finbow, M., Greenhalgh, D. and Hodgins, M.(2003) Targeted epidermal expression of mutant Connexin26(D66H)mimics true Vohwinkel syndrome and provides a model for thepathogenesis of dominant connexin disorders. Human Mol. Gen., 12,1737–1744.

18. Qiu, C., Coutinho, P., Frank, S., Franke, S., Law, L.Y., Martin, P.,Green, C.R. and Becker, D.L. (2003) Targeting connexin43 expressionaccelerates the rate of wound repair. Curr. Biol., 13, 1697–1703.

19. Plum, A., Hallas, G. and Willecke, K. (2002) Expression of the mouse gapjunction gene Gjb3 is regulated by distinct mechanisms in embryonic stemcells and keratinocytes. Genomics, 79, 24–30.

20. Magin, T.M., McWhin, J. and Melton, D.W. (1992) A new mouseembryonic stem cell line with good germ line contribution and genetargeting frequency. Nucl. Acids Res., 20, 3795–3796.

21. Theis, M., de Wit, C., Schlaeger, T.M., Eckardt, D., Kruger, O.,Doring, B., Risau, W., Deutsch, U., Pohl, U. and Willecke, K. (2001)Endothelium-specific replacement of the connexin43 coding region by alacZ reporter gene. Genesis, 29, 1–13.

22. Eckardt, D., Theis, M., Doring, B., Speidel, D., Willecke, K. andOtt, T. (2004) Spontaneous ectopic recombination in cell-type-specificCre mice removes loxP-flanked marker cassettes in vivo. Genesis, 38,159–165.

23. Traub, O., Eckert, R., Lichtenberg-Frate, H., Elfgang, C., Bastide, B.,Scheidtmann, K.H., Hulser, D.F. and Willecke, K. (1994)Immunochemical and electrophysiological characterization of murineconnexin40 and -43 in mouse tissues and transfected human cells.Eur. J. Cell Biol., 64, 101–112.

24. Butterweck, A., Elfgang, C., Willecke, K. and Traub, O. (1994)Differential expression of the gap junction proteins connexin45, -43, -40,-31 and -26 in mouse skin. Eur. J. Cell Biol., 65, 152–163.

25. Goliger, J.A. and Paul, D.L. (1995) Wounding alters epidermal connexinexpression and gap junction-mediated intercellular communication. Mol.

Biol. Cell., 6, 1491–1501.



gap junctions are intercellular conduits through which small

Documents