REPORTS

Allelic Variations of Human Keratins K4 and K5 provide Polymorphic Markers Within the Type II Keratin Gene Cluster on Chromosome 12

Reinhard Wanner, Hans-Heinrich Forster, lrmgard Tilmans,. ~nd Dietmar Misch~e . . .. . . t fur Experimentelle Onkologle und TransplantatlOnsmedlzm, Ul1Iversltatsklllukum RudolfVlrchow, Frele Ul1Iversltat Berlm,

lnstltu

Berlin, F.R.G.

-----To appreciate poi~t m~tations in keratin gen~s ~s causes for h editary epithehal diseases, the normal vanatlQn of these e~e sequences in the population must be known. Because

ge etic polymorphism of keratins at the protein level due to gil:lic variation has been described for the type II keratins 4 a d 5 we have analyzed their corresponding genes using a~ gle~strand conformation polymorphism gel electrophore­:~~ and sequence anAalysils hof pOhlymerase chain rea~ti?n al11pli­& d genomic DN . A t oug no sequence vanatlons were £ eund in the carboxyl-terminal and rod domains we were able t~ map the molecular ?ifferences ~mong the allele.s to t.heir

ino-terminal domams. In particular, we have Identified :{;ee alleles of keratin 4. Two a.lleles ?iffered.by ~ nucleot~de transition causing a neutral ammo aCid SUbStltutlOn (alamne

Keratins (K) are the tissue- and cell type - specific struc­tura l c?mponen~s of inter.mediate filaments ~n epith­elia. Like other 1l1termedlate filament prote1l1s, they share a central a-helical domain of approximately

. 310 amino acids that is flanked by non-helical amino­terminal head and carboxyl-terminal tail domains that vary consider­bly in size and sequence [1] .

a In humans there are more than 20 different keratins encoded by at I ast as many differentially expressed genes [2 - 5]. Based on their elatedness they can be subdivided into two classes: the acidic type I ~:ratins (K9 - K 19) and the neutral-basic type II keratins (K1- K8) . I addition, genetic polymorphism of keratins due to allelic varia­?on has been described for some human keratins at the protein level, ~mely, for the type II keratins K1, K4, and K5 and the type I keratin K10 [6-8]. In case of K1 and K10, the alleles differ by qariable numbers and sizes of glycine loops within the carboxyl­cerminal V2 subdomain [9,10). At least seven different alleles were thus detected for the single copy gene Kl0 [10].

The likely involvement of keratins in hereditary epithelial dis­eases (11 - 16] and t~e ~rospect of additio~al information on the role of keratin subdoma1l1s 111 filament formation [1,17,18] prompted us

Manuscript received December 10, 1992; accepted for publication

January 28, 1993. . . ' .. . Reprint requests to: Dr. DI~tmar Mls.chke, Instltut fur Expenmentelle

Onkologie und TransplantatlOnsmedlzll1, Umversltatskhmkum Rudolf Virchow, Freie Universitat Berlin, Spandauer Damm 130,0-1000 Berlin 19, FRG.

Abbreviations: EBS, epidermolysis bullosa simplex; K 1, keratin 1; PCR, polymerase chain reacti~n; PIC, polym~rphism information content; SSCP, single-strand conformatIOn polymorphism.

to valine) and one allele had a 42-bp in-frame deletion corre­sponding to 14 amino acids within the V1 subdomain. Three alleles were also recognized for the keratin 5 locus, all being elicited by single nucleotide substitutions. Of these, only one altered the amino acid sequence, replacing an uncharged (glycine) with a charged (glutamic acid) amino acid in the Hi subdomain. Pedigree analyses in three families showed the alleles to be inherited as autosomal Mendelian traits . Thus, these normal alleles of keratins 4 and 5 will provide favorable polymorphic markers for linkage analysis directly within the cluster of type II keratin genes located on chro­mosome 12q to elucidate the potential involvement of ~hese and other keratin genes in disorders of squamous ceJl differ­entiation. ] Invest Dermatol1 00:735 -741, 1993

to determine the basis of the observed polymorphism of human K4 and K5. We show here that the allelic variations map to the amino­terminal domains, i.e., the V1 subdomain of the K4 gene or the H1 subdomain of the K5 gene, respectively. The transmission of these alleles as Mendelian traits provides final proof for our previous hypothesis of a genetic polymorphism of these keratin genes and thus renders the already complex fam.ily of human keratins even more complex.

MATERIALS AND METHODS

Cytoskeletal Preparations and Gel Electrophoresis Epithe­lial tissue samples were homogenized and extracted in the presence of protease inhibitors as described [6]. The resulting Tritonjhigh­salt resistant pellet, highly enriched in keratins, was solubilized in sample buffer for gel electrophoresis. One-dimensional sodium do­de~yl sulphate polyacrylamide and two-dimensional gel electropho­reSIS were performed as described [71-

Nucleic Acids Preparation Genomic DNA and cytoplasmic RNA were isolated according to standard procedures [19]. Briefly, samples obtained from Iymphoepithelial tissue of palatine tonsll ~, squamous cell carcinomas of the head and neck, normal interfolh­cular epidermis, and cells from scrapings of buccal mucos~ were either ground to a fine powder under liquid nitrogen or dIrectly digested with proteinase K (200 .ugjml) and then phenoljchlor?­form extracted. Finally, the 'ethanol precipitated DNA was diS­solved in water.

Total cytoplasmic RNA was isolated from the epithelial lining of palatine tonsils obtained by heating the tonsils in 0.9% NaCI, 2 mM ZnCI2 [20J at 60 °C for 5 min. Poly A+-RNA was prepared from guanidinium isothiocyanate extracted RNA using Hybond-mAB-

0022-202Xj93jS06.00 Copyright © 1993 by The Society for Investigative Dermatology, Inc.

735

736 WANNER ET AL THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Table I. List of Oligonucleotide Primers Used for PCR and Sequencing

Denotation

K4-10' K4-10b K4-16' K4_11J KS-I0' KS-I0f KS-l1g KS-l1 h

Sequence (5' to 3')

CAGACTCTTCAGGTATCCGTGG AGGAAGCACAGAGACACCAGTG CCCAGCTCGCTCTCTGCAGCC TGGATCTCAGGGTCAATCTCCA AGCAGTGTTTCCTCTGGATATG CACTTGGAAGGCAGTGACTTGC AGAGCCACCTTCTGCGTCCTGC CTGGATGCTGGGGTCGATTTGC

.,! Bases 1213-1234 and 2040-2061. respectively of [23] . ' Bases 1-21 of this paper (EMBL/Genbank X67683). J Bases 35-56 of[23] . ·JBases 1583 -1604 and 1880 - 1901. respectively of[24]. g, ! Bases 3-24 and 562-583. respectively of[24].

paper (Amersham). RNA integrity was confirmed by in vitro trans­lation in a commercial rabbit teticulocyte lysate (Amersham) supple­mented with 35S-methionine .

To complete the N-terminal cDNA sequence ofK4 and to allow for selection of an appropriate oligonucleotide primer (K 4-16a, Table I), RNA from individuals homozygous or heterozygous for the K4 alJeles was reverse transcribed using primer K4-11b. The cDNA was then tailed with polyG using terminal transferase and amplified with the same primer, a polyC-tailed anchoring primer, and an anchoring primer (5' GCGCGGCCGCGGAGGCC 3'). The complete N-terminal cDNA sequence of K4 will be available under EMBL/GenBank accession number X67683.

Polymerase Chain Reaction (PCR) PCR amplification of human K4 and K5 genes was performed according to the protocol supplied by Perkin-Elmer (Norwalk, CT) using, ill a 30-,u1 reaction, 30 ng of genomic DNA and 100 nM of the appropriate oligonu­cleotide primers (see Table I). The reactions were performed for 35 cycles, each cycle consisting of 1 min at 94 ° C, 1 min at the indicated primer pair specific annealing temperature (Table I), and 1 min at 72 ° C. An elongation step at 72 ° C for 7 min completed the amplifi­cation procedure. For labeling amplified DNA, [a-32P]dGTP or [a-32P]dCTP was included at a concentration of 1.5 ,uCi per 10,u1 reaction. For amplification of mRNA, the rTth DNA polymerase kit (Perkin-Elmer) was used.

Single-Strand Conformation Polymorphism (SSCP) Analy­sis To generate DNA fragments suitable for SSCP analysis, PCR products were trimmed to size by appropriate restriction enzyme digests. Restricted DNA corresponding to 1/100th of the original

Annealing Temperature

6S · C

60· C

6S·C

6S· C

Position

C-terminal

N-terminal

C-terminal

N-terminal

Length (bp)

849

455/413

319

581

PCR sample was denatured in 10 vol of 95% formamide, 10 roM NaOH, 10 mM ethylenediaminetetraacetic acid, 0.05% Brom­phenolblue, 0.05% Xylene Cyanole by heating at 90°C for 2 min, immediately transferred onto ice, and loaded on a 5% polyacrylam­ide gel (300 X 140 X 0.75 mm) containing 90 mM Tris-borate, pH 8.3, 4 mM EDTA for SSCP analysis as described [21]. Electro­phoresis was in the 2001 vertical electrophoresis chamber (LKB) at 40 W for 1 h with cooling to 20 °C. The same gel system was used to separate undenatured DNA samples. The gel was dried on filter paper and exposed to X-ray fi lm (4 - 24 h with intensifying screen) or silver stained.

Direct Sequencing PCR products were either separated on 4% Nusieve 3:1 agarose or polyacrylamide gels according to [19] and DNA from appropriate bands eluted. To minimize incorrect se­quence information due to the occasional infidelity ofTaq polymer­ase, the entire population of PCR products was analyzed by direct sequencing double-stranded DNA with Sequenase (U.S . Biochemi­cals) in the presence of single-strand binding protein (USB) as de­scribed [22].

RESULTS

Polymorphic K4 and KS Polypeptides Analysis of keratin­enriched cytoskeletal residues obtained from human squamous non-keratinizing epithelia by high-resolution sodium dodecylsul­fate gel electrophoresis revealed an inter-individual variation for K4 and K5 within the characteristic polypeptide pattern consisting of the type II keratins K4, K5, and K6 and the type I keratins K13 and K14 (Fig 111) . This was brought about by protein bands showing a

2 3 4a/b

2~ "3 - 4 5

Figure 1. Polymorphic K4 and KS polypep­tides in vivo and in vitro. Coomassie-Blue stained sodium dodecylsulfate - polyacry­lamide gel electrophoresis (SDS-PAGE) (a) of keratin-enriched cytoskeletal preparations of non-keratinizing squamous epithelia from dif­ferent individuals with the phenorypes K4a + KSb (lane 1), [K4a + K4b] + KSb (lane 2), K4b + KSb (Iatle 3), K4a + KSa (lane 4), and K4a + [KSa + KSb] (lane 5). Autoradiographs of SDS-PAGE -separa~ed (b) and NEpHG/ SDS-PAGE-separated (e,d) ill vitro transla­tion products of RNA from individuals with the fo llowing phenotypes: K4b + KSb (b, latle 1), [K4a + K4b] + [KSa + KSb] (b, lalle 2), K4a + K5b (b, [aile 3), [K4a + K4b] + KSb (e), and K4a + [KSa + KSb] (d) .

4(~~ 5(a~ ab~

13 -14 - -

1~~

5b ..... \ -6 13-

• ". 4a

5b 5a W \ \ -

13-

VOL. 100. NO. 6 JUNE 1993

-- .. j(.- --

~ I> I>

* *-* *

b 2 3 4 6 7 8 9 10 " 12

figure 2. P~Rampli~cation ofche N·ccrminal do.main ofK4. To allow for 5SCP analysIs the ampltficatlon products of genomIc DNA from mdividuals with known K4 polypeptide phenotype were digested with Haem and DdeI into fragments of 9, 25, 26, 35, 44, 54, 90, and 172 (K4a) and/or 130 bp (!C4b). (a) Polyacrylamide gel electroph.oresis of undenatured HaeIII /Ode! fragments Showlllg the 172·bp (K 4a; solid arrowhead), 130-bp (K 4b; asterisk), ,nd 90-bp fragments. The opel! arrow indicates heteroduplexes with slower migration rates than perfectly matched strands that form only in samples from heterozygous individuals. (b) SSCP analysis for the same individuals detecting K4b (asterisks) and also the two alleles of K4a (K4a2, closed arrow­heads; K4al, open arrowheads). Because the PCR products in lalles 1-5 and /0-12 were labeled with [a-32PJdGTP and those in lalles 6-9 with 1(X.32P)d<;:~P, th~ two corresponding single strands show complcmclltary changes In IntenSIty.

slight difference in migration velocity and presenting either as a doublet 0: ~s a single ~~nd ,:it~ the lower .or the higher electropho­retic mobIlIty. In addition, 111 vitro translatlon of polyA+ -RNA from squamous non-keratinizing epithelia of individual donors yielded the same phenotype of keratin alleles as the itl vivo set, thus indicat­ing that each allelic polypeptide is encoded by its own mRNA (Fig lb-d). Collectively, these data suggested that both keratin genes are polymorphic and express co-dominant alleles. Individuals display­ing keratin doublets may then be interpreted as being heterozygous and those showingjust one keratin species as being homozygous for fhe respective variant allele a or b.

Amplification of the C-Terminal Domains of K4 and K5 Because the main features of intermediate fil ament chain structure suggest that the N - and C-terminal domains particularly

A C G T A C

5· A C G A \ -c -A --A -- -G .. G T

~ ~ G T~C

C T G / c T T T 3· -

a Ua, Kh!

N-terminU9

K4a2

N-term inu s

t

K4b

b

HUMAN POLYMORPHIC KERATINS 737

to T A C G 1 A C G T

- ",':Q ..

rod Y.S'1!V",r.;~6" .... d 0 m a i n

rod ~om a in

Figure 3. Sequence analysis and deduced amino acid sequence of the K4 alleles. (a) Segment of the sequencing gel showing the single nucleotide exchange ~hat characterizcs alleles K4a 1 and K4a2 and the 42.bp deletion in K4a21eadmg to allele K4b. (b) Rcpresentation of the N-terminal amino acid sequence showing the deletion and mutation in the V1 domain.

speci.fy the members of each class by their characteristic size, com­pOSition, an? sequence (1] , we analyzed these domains by means of the P<:R, slllgie-strand conformation gel electrophoresis, and se­quenClll.g. Geno~ic DNA from individuals homozygous for K4a or K4b at

the protell1 level was amplified with primers K4-10a and K4-10b ~Table 1.111 Materials and Methods) yielding fragments of 849 bp, Irrespectively of the genotype (not shown). Direct sequencing of these PCR rroducts confirmed the avai lable eDNA sequence infor­matIOn [23 and established the sequence of introns 7 and 8 of K4 (EMBL/GenBank accession number X61028). Similarly, genomic DNA from individuals homozygous for K5a or K5b at the protein level was amplified using primers K5-10a and K5-10b to give frag­ments of 319 bp, again irrespectively of the genotype (not shown). Sequencing confirmed the published K5 C-terminal sequence [24,25] leading us to conclude that, in contras t to the Kl and KlO alleles [9 ,10], the differences among the K4 and K5 alleles could not reside within their C-terminal domains.

Atnplification of the K4 N-Terminal Domain However, amplification of genomic DNA using the N-terlninal specific primers K4-16a and K4-11b yielded two amplification products of 455 bp (K4a) and 413 bp (K4b) in case of heterozygous subjects, or either one or the other in homozygous individuals (Fig 2a). In all instances, the number of the bands and their sizes were in complete accordance with the corresponding protein phenotype thus estah-

738 WANNER ET AL

]314 nts

* * * A ~ . J> ~ V ]267 nts

~

* ~ ~

* J> V * J>

2 3 4 5 6

Figure 4. SSCP ana lysis of the N-terminal domain of KS. The S81-bp amplification products of genomic DNA from selected individuals gener­ated with the primer pair KS-11a and KS-11b were digested with EcoRV into fragments of 314 and 267 bp and analyzed for SSCP. The gel was subjected to silver sta ining. All six possible combinations of three alleles in homozygous and heterozygous,individuals are displayed in iaucs 1 to 6 [a/a; b1/b1; b2/b2; a/b1; a/b2; bl/b2]. The single strands corresponding to the 267-bp restriction fragment are indicated as follows: KSa, asterisks; KSb1, open arrowheads; KSb2, closed arrowheads. Apparently, one of the single strands ofKSb1 and KSb2 as well as one ofKSa and KSb1 shows the same migration velocity on SSCP gels.

lishing the two size alleles K4a and K4b. Yet, SSCP analysis of HaeIII and DdeI double-digested DNA showed the resulting 172-bp restriction fragment of allele K4a to further separate into two pairs of single strands that were named K4a1 and K4a2, accordingly (Fig 2b). For example, the individuals shown in lanes 1, 8, and 10 are heterozygous for K4a1 and K4a2.

Sequenc;:ing indicated that the difference between alleles K4a2 and K4b is due to a 42-bp deletion and between K4a 1 and K4a2 to a nucleotide transition from T to C (Fig 3a). The deduced amino acid sequence (Fig 3b) then showed the deletion to be in frame and equivalent to the loss of 14 amino acids in the V1 subdomain. The point mutation caused a neutral amino acid substitution from valine to alanine, also located in the V1 subdomain.

Amplification of the KS N-Terminal Domain Genomic DNA from individuals heterozygous or homozygous for KSa and/ or KSb was amplified with primers KS-11 a and KS-11 b and yielded fragments of the same size (518 bp), irrespective of genotype (not shown). Differences among the alleles became apparent only after SSCP analysis of EcoRV digested amplification products as shown in Fig 4. A total of three pairs of single strands could be discerned. Two of those belonged to the protein phenotype KSb and were therefore labeled KSb1 and KSb2. For these, direct sequencing re­vealed a transition from C to T in the third base of a leucine codon, accordingly having no effect on this amino acid in position 116 of the V1 subdomain at al l.

The alleles KSa and KSb1 were distinguished by a transversion from A to G for the respective homozygous individuals; in the heterozygous individual the corresponding bands appeared with equal intensities in both lanes of the sequencing gel (Fig Sa). This sequence variation is located in the H 1 subdomain leading to a replacement of a glycine by a glutamic acid (Fig Sb). Because an uncharged amino acid is substituted for a charged one, the overall charge of the polypeptides is affected yielding 1soelectric points of 7.80 and 8.1 6 for KSa and KSb, respectively confirming the ob­served separation of the corresponding polypeptide variants by two­dimensional gel electrophoresis (cf. Fig 1d and [6,7]).

Segregation ofK4 and KS Alleles To evaluate the distribution of the alleles in the three families that had already been analyzed at the protein level [8] , buccal mucosa DNA was amplified with the appropriate primer pairs. In all cases, the previous polypeptide­based pedigree analyses were confirmed but could also be refined with respect to the newly detected sequence variations in K4a (Fig

T G

...

-- ~ ..

a KSa

KSa

KSb

b

THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

T G A

~

--KSbl KSa + KSbl

5' C

/ c C T C C T G

A<>G A G G T A T C

'" C A A 3'

rod -domain

Figure 5. Sequence analysis and deduced amino acid sequence of the K5 alleles. (a) The position of the A to G transversion is indicated in rhe homozygous and heterozygous individuals in this segment of the sequenc­ing ge l. (b) Representation of the N-terminal amino acid sequence showing the amino acid exchange in H 1 and the position of the silent mutation in the V1 subdomain.

6a) and KSb (Fig 6b). Again, all alleles segregated truely as autoso­mal Mendelian traits.

Allele Frequencies of the K4 and KS Alleles A total of 142 alleles of K4 and 110 alleles of KS were analyzed in our survey. Of these, 98 K4 alleles and 90 KS alleles were from unrelated individu­als. Although our samples were initially non-randomly chosen, i.e., with respect to the phenotype of polypeptide variants expressed, the alleles K4al and K4a2 as well as the alleles K5b1 and KSb2 could be considered as randomly detected among the 69 K4a alleles and the 68 KSb alleles of non-related individuals, respectively. Together with the previously determined allele frequencies for K4 and K5 polypeptides [8] the allele frequencies can therefore be calculated as presented in Table II. This would result in calculated PIC values of 0.44 and 0.33 and heterozygosity indices of 0.48 and 0.36 for K4 and KS, respectively.

VOL. 100, NO.6 JUNE 1993

.2.2

I 81.2 82 b

-- - ---- -* - - *-

* * a

a b1 II.

b figure 6. Segregation of the K4 and KS alleles. DNA from scrapings of buccal mUCosa was PCR amplified with the appropriate primer pairs. The edigree has been aligned to the lanes on the autoradiog.ram of the SSCP ge l.

~a) The amplificatIOn products of K4 were dIgested WIth HaeIII/Odd and separated by SSCP electrophoresis (only the 172- and 130-bp fragments arc shown). Allele K4al is denoted by opell arrow/,eads, allele K4a2 by closed arrowheads, and allele K4b by asterisks. (b) The products of KS amplification were digested with HaeIII/EcoRV into restriction fragments of 314,166, and 101 bp, and separated by SSCP electrophoresis (only the 101 bp frag­ment is shown). Allele KSa is denoted by an OpW arrowhead, allele KSbl by a closed arrowhead. This pattern is best explained by assuming that one single strand takes up two conformations, one that comigrates with the respective complementary strand and the other having the same migration velociry for both alleles. The icons used in the pedigree are appropriately labeled for some individuals.

DISCUSSION

We have identified, by SSCP electrophoresis and sequence analysis of PeR-amplified genomic DNA, the differences that distinguish the alleles of keratins K4 and KS, respectively. Unlike the Kl and KI0 alleles, which differ in the sizes of their V2 subdomains near the C-terminus [9,10]' the C-terminal domains of the K4 and KS genes did not reveal any variation among the different genotypes analyzed, neither in size nor in sequence. Similarly, sequence analy­sis of the rod domain as amplified from mRNA (not shown) did not reveal any systematic differences that would account for the ob­served polypeptide variants.

HUMAN POLYMORPHIC KERATINS 739

Table II. Allele Frequencies of the K4 and KS Alleles

Allele Frequency

K4al 0.lS8 K4a2 0.682 K4b 0.160 KSa 0.082 KSb1 0.783 KSb2 0.13S

However, such systematic variations were found within the gene-specific sequences encoding the N-terminal domains. Thus, not only the cause for the observed migration behavior of the kera­tin polypeptides could be identified either as a 42-bp size difference in the V1 subdomain of K4 or as a sequence variation in the H1 subdomain affecting the overall charge of KS, but also one addi­tional allele for each gene rendered both loci even more polymor­phic than anticipated.

Since point mutations in keratin genes have been described to cause autosomal dominant skin blistering diseases such as epidermo­lysis bullosa simplex (EBS) and epidermolytic hyperkeratosis. [11-16,26], the availability of polymorphic markers directly wlthl11 the cluster of type II keratins localized on chromosome 12ql1-q13 ([27] and references therein) will be of immediate significance for future linkage analysis of suspected keratin abnormalities and epithelial diseases. Even when compared to anonymous markers with higher PIC values, like microsatellite DNA polymorphisms, these poly­morphic keratin genes can be expected to be more informative be­cause of the lower likelihood of recombination due to close phYSical association or direct involvement. And indeed, the polypeptide po­lymorphism ofKS has already been instrumental in unravelling the case of one EBS family [13] as has the Kl polymorphism 111 a fami ly with epidermolytic hyperkeratosis [16]. Thus, the knowledge ofK4 and KS sequence variations within the population will furtherfoster the Identification of alterations in type II keratin genes leadmg to disease phenotypes.

Although none of the described alleles and their underlying se­quence differences seem to be associated with apparent, namely clinical, impairment of epithelial development and differentiation, it is interesting to note that a Glu to Gly change like the one mani­festing in the HI subdomain of alleles KSb and KSa is linked with EBS when the very same change occurs in the conserved helix termination peptide at the end of the rod domain of KS [1 3]. In addition, a leucine to proline mutation in the HI subdomain ofKl, located 11 amino acids downstream towards the a-helical rod domain, has also been shown to cause defective keratin filament formation in one family with epidermolytic hyperkeratosis [16]. Therefore, the physiologic and hitherto possibly undetected patho­phYSIOlogiC consequences of allelic variations will ment further investigations.

In the context of the understanding that intermediate filament proteins and-even more-the members of the keratin families, are closely re lated and have evolved through common, presumably lamin-like ancestors [1,28], our finding of genetic polymor~hisms may not be surprising but may rather reflect ongoing evolutionary processes leading to diversification and specialization. In particular, tandem repetition of sequence motifs followed by n:utat~onal changes has been suggested to contribute to such a diverSity wlth1l1 the variable N- and C-terminal domains of keratin protellls [3,29). Furthermore, because the different type II keratin genes are clus­tered, they may also be particularly prone to recombination events like unequal crossing-over or gene conversion.

In fact, the N-tenninal sequences of both genes reveal several motifs, i.e., rather long (up to 19 bp in case of K4) parallel and antiparallel repeats that are reminiscent of tandem dupltcatlOn and insertion. However, the opposite, i.e., a deletion, may also ha~e been acquired considering the sequence of allele K4a2. As shown 111

740 WANNER ET AL

Gl

K4a

.. 0 D:J .[ .[ <) 0

<) oD ~ oD

U

t!l

t!l

o

S'TTTGGGGGTGCTGGAGGC TTGGCACTGGTGGCTTTGGTG 0

'" X C' "~J;.AC.I®.Illll.C.Ill®.Hl GTGG AT T TGG G G G 3'

K4b 5' TTTGGGGGTGCTGGAGGCTT T GGCACTQGTGGCTT TGGTG GTGGATT TGGGGG 3'

Figure 7. Evolutionary origin of the K4 alleles. Intra-chromatidal align­ment of the two 19 bp direct repeats in allele K4a2 followed by illegal crossing-over would result in a 42-bp interstitial deletion generating allele K4b.

Fig 7, an intra-chromatidal alignment, e.g., by slipped-strand mis­pairing of the two 19-bp direct repeats found in the N -terminal domain of this allele followed by illegal crossing-over would ac­count for the 42-bp interstitial deletion present in allele K4b.

With respect to keratins, sequence variability has its constraints in function as many mutations will be lost during evolution, most likely due to deleterious effects on filament formation. In addition, the requirements for hetero-polymerization of type I and type II keratins are likely to be less tolerant to mutations in the proteins than the requirements of homo-polymerization as in most other intermediate filament proteins [30] . It has also been shown that the end domains have subtle roles in IF network formation and/or 10-nm fi lament assembly and stability, with recent findings sug­gesting that specific sequence motifs in the head domain and their spatial arrangement are particularly important [18,31- 35]. This may be reflected by our finding that, except for the 14 amino acid deletion in the VI subdomain of K4 corresponding to the loss of three glycine loops, all other allelic differences are point mutations. Of these, two involve C to T transitions at CpG dinucleotides suggesting methylation of cytosine followed by deamination and mutation [36].

In contrast, the Kl and, in particular, the KI0 alleles differ by rather large deletions in their C-terminal domains [9,10] as if this region tolerated more variability without impairment of function. Yet, the deletions in KI0 seem also not to be randomly distributed but restricted to three major sites as defined by glycine loops [37], which are thought to be involved in the structural organization of keratin filaments in epithelial cells and to contribute to the flexibil­ity of the entire epithelium. All point mutations found to be asso­ciated with the EBS phenotype as well as most point mutations associated with epidermolytic hyperkeratosis, however, map to the rod domain, either to the highly conserved helix initiation or helix termination motifs or to coil 2 of the rod [11-13,15], indicating that these regions are highly sensitive to sequence variations as they fulfill crucial roles in filament formation.

The question as to when-in evolution-the alleles arose re­mains open. However, the apparent lack of linkage disequilibrium among the K4 and K5 alleles l8] despite that both genes have been mapped to the type II keratin gene cluster on ch'romosome 12ql1-q13 [38 ,39] may argue in favor of very long histories for these alleles. The analysis of polymorphisms of keratin genes in other species will provide further insight into the mechanisms involved.

In conclusion, polymorphism of the keratin genes is rendering this already complex multigene family even more complex. How­ever, the genetic polymorphism of K4 and K5 as described in this paper together with allelic variations of the Kl gene [9] and the K2 and K3 genes (D. Mischke, unpublished observation) should allow for improved haplotype analysis and thus will help to study genetic

THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

linkage within the type II keratin gene cluster on chromosome 12 to

elucidate the potential involvement of keratins in disorders of squa­mous cell differentiation and neoplasia.

Fllllditig from tire Deutsclre Forscllllllgsgemeillschaft (Mi 210/6-1) is gratifrdly ackllowledged. We tlrallk Dr. Michael Hummel, IlISlifut fiir PatllOlogie, Utliver­sildtsklitlikum Steglitz, Free Utliversity of Berlin, for syllthesizillg some of the primers; Dr. Alldre Reis, IlISlitul fur HllI1lal1gelletik, Ulliversitdtsklillik'll1l Rudolf VircilOw, Free Utliversify of Berlill, for com ments Oil the mmluscript; alld Dr. A. Ziegler (tlris illstitute) for stimulatillg discussiolls alld etlcouragement.

REFERENCES

1. Steinert PM, Roop DR: Molecular and cellular biology of intermediate filaments. Annu Rev Biochem 57:593-625,1988

2. Moll R, Franke WW, Schiller DL, Geiger B, Krepler R: The catalog of human cytokeratins: patterns of expression in normal epithelia, tumors and cultured cells. Cell 31: 1-24, 1982

3. Sun T-T, Eichner R, Schermer A, Cooper 0, Nelson WG, Weiss RA: Classification, expression, and possible mechanisms of evolution of mammalian epithelial keratins: a unifying model. In: Levine A, Topp W, Vande Woude G, WatsonJD (cds.). Cancer Cells 1: The Transformed Phenotype. Cold Spring Harbor Laboratory, New York, pp 169-176, 1984

4. Fuchs E: Keratins as biochemical markers of epithelial differentiation. Trends Genet 4:277 - 281, 1988

5. Oshima RG: Intermediate filament molecular biology. Curr Opin Cell Biol4:110-116, 1992

6. Wild G-A, Mischke 0: Variation and frequency of cytokeratin poly­peptide patterns in human squamous non-keratinizing epithelium. Exp Cell Res 162:114-126, 1986

7. Mischke 0, Wild G: Polymorphic keratins in human epidermis. J Invest Dermatol 88:191-197, 1987

8. Mischke 0, Wille G, Wild AG: Allele frequencies and segregation of human polymorphic keratins K4 and K5. Am J Hum Genet 46:548 - 552, 1990

9. Korge BP, ComptonJG, Steinert PM, Mischke 0: The two size alleles of human keratin 1 are due to a deletion in the glycine-rich car­boxyl-terminal V2-subdomain. J Invest Dermatol 99:697 - 702, 1992

10. Korge BP, Gan S-Q, McBride OW, Mischke 0, Steinert PM: Exten­sive size polymorphism of the human keratin 10 chain resides in the C-terminal V2 subdomain due to variable numbers and sizes of glycine loops. Proe Nat! Acad Sci USA 89:910-914, 1992

11. Bonifas JM, Rothman AL. Epstein EH: Epidermolysis bullosa simplex: evidence in two fami lies for keratin gene abnormalities. Science 254:1202-1205, 1991

12. Coulombe PA, Hutton ME, Letai A, Hebert A, Paller AS, Fuchs E: Point mutations in human keratin 14 genes of Epidermolysis bullosa simplex patients: genetic and functional analyses. Cell 66:1301-1311,1991

13. Lane EB, Rugg EL, Navsaria H, Leigh 1M, Heagerty AHM, Ishida­Yamamoto A, Eady RAJ: A mutation in the conserved helix termina­tion peptide of keratin 5 in hereditary skin blistering. Nature 356:244-346,1992

14. Compton JG, DiGiovanna JJ, Santucci SK, Kearns KS, Amos CI, Abangan DL, Korge BP, McBride OW, Steinert PM, Bale SJ: Link­age of epidermolytic hyperkeratosis to the type II keratin gene clus­ter on chromosome 12q. Nature Genetics 1:301- 305, 1992

15. Rothnagel JA, Dominey AM, Dempsey LD, Longley MA, Green­halgh DA, Gagne TA, Huber M, Frenk E, Hohl D, Roop DR: Mutations in the rod domains of keratins 1 and 10 in epidermolytie hyperkeratosis. Science 257:1128-1139, 1992

16. Chipev CC, Korge BP, Markova N, Bale SJ, DiGiovanna]], Compton JG, Steinert PM: A leucine > proline mutation in the HI subdo­main of keratin 1 causes epidermolytic hyperkeratosis. Cell 70:821-828,1992

17. Hatzfeld M, Weber K: A synthetic peptide representing the consensus

VOL. 100, NO.6 JUNE 1993 HUMAN POLYMORPHIC KERATINS 741

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

sequence motif at the carboxy-terminal end of the rod domain inhib­its intermediate filament assembly and disassembles preformed fila­ments.] Cell BioI 110:1199-1210, 1990

Herrmann H, Hofmann I, Franke WW: Identification of a nonapep­tide motif in the vimentin head domain involved in intermediate filament assembly.] Mol BioI 223:637 -650, 1992

Maniatis T, Fritsch EF, SambrookJ: Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory, New York, 1982

Schvveizer], Goerttler K: Synthesis in vitro of keratin polypeptides by mRNA isolated from newborn and adult mouse epidermis. Eur] Biochem 112:243-249,1980

Orita M, Suzuki TS, Hayashi K: Rapid and sensitive detection of point mutations and DNA polymorphisms using the polymerase chain reaction. Genomics 5:874-879, 1989

Wanner R, Tilmans I, Mischke D: Avoiding strand reassociation in direct sequencing of double-stranded PCR products with thermola­bile polymerases. PCR Meth Appl 1:193-194, 1992

Leube R, Bader BL, Bosch FX, Zimbclmann R, AchtstaetterT, Franke WW: Molecular characterization and expression of the stratifica­tion-related cytokeratins 4 and 15. ] Cell BioI 106:1249-1261, 1988

Eckert RL, Rorke EA: The sequence of the human epidermal 58-kD (# 5) rype II keratin reveals an absence of 5' upstream sequence conservation between co expressed epidermal keratins. DNA 7:337-345,1988

Lersch R, Fuchs E: Sequence and expression of a rype II keratin, K5, in human epidermal cells. Mol Cell Bioi 8:486-493,1988

Ryynanen M, Knowlton RG, Uitto J: Mapping of epidermolysis bul­losa simplex mutation to chromosome 12. Am ] Hum Genet 49:978 - 984, 1991

Craig LW, McBride OW: Report of the committee on the genetic constitution of chromosome 12. Cytogenet Cel l Genet 58:555-579,1991

Dodemont H, Riemer D, Weber K: Structure of an invertebrate gene encoding cytoplasmic intermediate filament (IF) proteins: implica­tions for the origin and the diversification of IF proteins. EMBO ] 9:4083-4094,1990

29. Klinge EM, Sylvestre YR, Freedberg 1M, Blumenberg M: Evolution of keratin genes: different domains evolve by different pathways.] Mol EvoI24:319-329,1987

30. Myers MW, Lazzarini RA, Lee VM-L, Schlaepfer WW, Nelson DL: The human mid-size neurofilament subunit: a repeated protein se­quence and the relationship of its gene to the intermediate filament gene family. EMBO ] 6: 1617 - 1626, 1987

31. Geisler N , Weber K: Phosphorylation of desmin in vitro inhibits formation of intermediate filaments: identification of three kinase A sites in the aminoterminal head domain. EMBO] 7:15-20,1988

32. Raats ]MH, Pieper FR, Vree Egberts WTM, Verrijp KN, Ramaekers FCS, Bloemendahl H: Assembly of amino-terminally deleted des­min in vimentin-free cells. ] Cell BioI 111: 1971-1985, 1990

33. Lu X, Lane EB: Retrovirus-mediated transgenic keratin expression in cultured fibroblasts: specific domain functions in keratin stabiliza­tion and filament formation. Cell 62:681-696, 1990

34. Traub P, Scherbarth W , Wiegers W, Shoeman RL: Salt-stable interac­tion of the amino-terminal head region of vimentin with the a-heli­cal rod domain of cytoplasmic intermediate filament proteins and its relevance to protofilament structure and filament formation and stabiliry.] Cell Sci 101 :363 - 381, 1992

35. Parry DAD, Steinert PM: Intermediate filament structure. Curr Opin Cell BioI 4:94 - 98, 1992

36. Cooper DN, Youssoufian H: The CpG dinucleotide and human ge­netic disease. Hum Genet 78:151 - 155, 1988

37. Steinert PM, Mack]W, Korge BP, Gan S-Q, Haynes SR, Steven AC: Glycine loops in proteins: their occurrence in certain intermediate filament chains, loricrins and single-stranded RNA binding pro­teins. Int] Bioi Macromol 13:130-139, 1991

38. Barletta C, Batticane N, Ragusa RM, Leube R, Peschle C, Romano V: Subchromosomal localization of two human cytokeratin genes (KRT4 and KRTI5) by in situ hybridization. Cytogenet Cell Genet 54:148-150,1990

39. Rosenberg M, Fuchs E, Le Beau MM, Eddy RL, Shows TB: Three epidermal and one simple epithelial rype II keratin genes map to human chromosome 12. Cytogenet Cell Genet 57:33-38, 1991

ANNOUNCEMENT

The Jackson Laboratory presents a workshop, "Mouse Mutations as Animal Models and Bio­medical Tools: Skin and Hair Mutations, September 29 - October 2, 1993 at Bar Harbor, Maine.

Focus This new workshop will present essential information on the valuable source of mouse mutations as animal models of human diseases of the skin and hair. We will focus on recent advances it: the use of animal models for understanding skin and hair in normal and pathologic states and mtroduce new mUrine models for dermatological research.

Applications Placement is limited to 125, including faculty. Applicants need to possess a doctoral degree or equivalent or an advanced graduate degree.

Applicati~n for adm.is~ion is made by submitting 1) the intent form; 2) your curriculum vitae; 3) a letter bnefly descnbll1g your work and stating why you want to attend this workshop; and, for graduate students, 4) a letter of recommendation from your major advisor. Applications will be reviewed on a competitive selection basis starting in February and will continue until the workshop is full. For further information, contact Suzanne Serreze at 207/ 288-3371, ext. 1378, FAX 207/288-8254, or e-mail sbs@aretha.jax.org.