A Splice Site Mutation in the Gene of the Human Type I Hair Keratin hHa1 Results in the Expression of a Tailless Keratin Isoform*

In this study, we have elucidated the molecular mechanisms underlying the expression of an acidic 41-kDa protein inherited as an autosomal dominant trait of the hair keratin pattern of about 5% of the human population. We show that this protein is a size variant of the large type I hair cortex keratin hHa1 due to a genetic polymorphism in the hHa1 gene. We detected a G-A substitution in the 5′ splice site of intron 6 of the hHa1gene, which segregates with the 41-kDa protein phenotype in two pedigrees and is responsible for the formation of an abnormally spliced hHa1 mRNA species. The use of an alternative 5′ splice site leads to the retention of 41 nucleotides of the initial intron 6 sequences in the mature transcript. The open reading frame of the aberrant mRNA creates a premature stop codon immediately downstream of the mutation site. The resulting hHa1 protein variant, hHa1-t, is about 6-kDa smaller than the 47-kDa hHa1 hair keratin and lacks the complete nonhelical tail domain. We show that the tailless hHa1-t is functional, since both recombinant hHa1 and hHa1-t form identical keratin intermediate filaments when assembled in vitro with a type II hair keratin partner. This finding confirms the view of a noninvolvement of the keratin tail domain in filament assembly and explains the lack of a pathological hair phenotype in hHa1-t positive individuals.

In this study, we have elucidated the molecular mechanisms underlying the expression of an acidic 41-kDa protein inherited as an autosomal dominant trait of the hair keratin pattern of about 5% of the human population. We show that this protein is a size variant of the large type I hair cortex keratin hHa1 due to a genetic polymorphism in the hHa1 gene. We detected a G-A substitution in the 5 splice site of intron 6 of the hHa1 gene, which segregates with the 41-kDa protein phenotype in two pedigrees and is responsible for the formation of an abnormally spliced hHa1 mRNA species. The use of an alternative 5 splice site leads to the retention of 41 nucleotides of the initial intron 6 sequences in the mature transcript. The open reading frame of the aberrant mRNA creates a premature stop codon immediately downstream of the mutation site. The resulting hHa1 protein variant, hHa1-t, is about 6-kDa smaller than the 47-kDa hHa1 hair keratin and lacks the complete nonhelical tail domain. We show that the tailless hHa1-t is functional, since both recombinant hHa1 and hHa1-t form identical keratin intermediate filaments when assembled in vitro with a type II hair keratin partner. This finding confirms the view of a noninvolvement of the keratin tail domain in filament assembly and explains the lack of a pathological hair phenotype in hHa1-t positive individuals.
Besides the large number of epithelial or soft ␣-keratins, the complex keratin multigene family also comprises the numerically smaller group of hard ␣-keratins. These keratins are involved in the formation of various hard, keratinized structures such as hairs, nails, claws, beaks, etc., and according to their most intensely investigated site of expression, are commonly termed hair keratins (for a recent review see Ref. 1). One-and two-dimensional gel electrophoretic analyses of native hair keratins have shown that, independent of mammalian species, hair keratins are resolved into two subfamilies, each consisting of four distinct but closely spaced type I acidic  and type II basic to neutral (60 -58 kDa) members designated Ha1-Ha4 and Hb1-Hb4, respectively (2,3). Together with a minor hair keratin pair termed Hax/Hbx, the family of hair keratins is considered to be composed of 10 members (2,4). This notion seemed to be confirmed by the molecular cloning of four type I mouse hair keratins, mHa1-mHa4, and four type II sheep wool keratins, for which the designation K2.9-K2.12 has been proposed (5)(6)(7)(8). Expression studies showed that follicular hair keratin synthesis is compartmentalized into the cuticle, in which apparently only one keratin pair is expressed, and the cortex, which exhibits the sequential expression of several pairs of keratins (7,8).
Recent investigation in our laboratory aimed at characterizing human hair keratins has shown that the hair keratin family is distinctly more complex than previously assumed. At present, the human type I subfamily comprises seven members (hHa1 (9), hHa2 (10), hHa3-I, hHa3-II (11), hHa4, 1 hHa5 (12), hHRa1 1 ) characterized at the mRNA or genomic level. The type II subfamily consists of four keratins (hHb1 (11), hHb3, hHb5, hHb6 (13)), the type II partners for the cuticular keratin hHa2 and the hair-related keratin hHRa1, remaining to be elucidated. Moreover, contrary to previous evidence, hair keratin expression in the follicle is not restricted to the trichocytes of the cuticle and the cortex but also occurs much earlier in matrix cells of the hair bulb (12,13).
Several earlier reports on one-and two-dimensional electrophoretic analyses of keratin extracts from human hairs repeatedly emphasized the occurrence of a rather prominent but unusual protein component that invariably migrated about 3-4 kDa below the collective of type I hair keratins (14 -19). Remarkably, this low molecular weight component was observed only sporadically and seemed to be restricted to hair keratin patterns present in about five percent of the human population (15,17). Systematic large scale family studies revealed that the protein component was inherited by autosomal dominant transmission, however, without the manifestation of a distinct hair phenotype (14 -19). Despite the intriguing properties and expression characteristics of this protein, up until now no efforts were made regarding its molecular characterization. Therefore, in the course of our investigations on the elucidation of new members of the human hair keratin family, we felt that this hair-associated protein could not be left unconsidered. In the present study we show that the low molecular weight protein belongs to the hair keratins. It is, however, not a novel hair keratin but represents a truncated isoform of the large type I hair keratin hHa1, resulting from a 5Ј donor splice site mutation in intron 6 of the hHa1 gene.

Extraction of Human Hair Keratins, SDS-Gel Electrophoresis and
Western Blot Procedure-Human hair clippings from male and female volunteers were washed with ethanol, dried, and minced to 2-3 mm in length. The material was homogenized in a glass-glass homogenizer in 8 M urea, 2% SDS, 100 mM DTT, 2 100 mM Tris-HCl, pH 8.0, in the presence of a small amount of sterile quartz sand. The homogenate was incubated overnight at 37°C, heated at 60°C for 3 min, and centrifuged at 10,000 rpm for 5 min. The supernatant was recovered and used 1:1 diluted with SDS sample buffer. Recombinant human hair keratins hHa1, hHa1-t, and hHb3 were diluted from 9.5 M urea with 3 ϫ SDS sample buffer. One-dimensional SDS-polyacrylamide gel electrophoresis of total hair keratin preparations and recombinant hair keratins was carried out according to Laemmli (20) in 10 or 12% gels. Parallel gels were stained with Coomassie Blue or were blotted electrophoretically onto nitrocellulose (BA85; Schleicher & Schuell). The membranes were blocked with 10% nonfat milk powder and 0.1% Tween in Trisbuffered saline for 1 h. Samples were incubated with primary keratin antibodies at a suitable dilution (see antibodies) for 1 h in Tris-buffered saline followed by three washes with Tris-buffered saline Tween for 20 min each. Peroxidase-coupled secondary goat antibodies to guinea pig or murine IgG (Dianova; Hamburg, Germany) were diluted 1:10,000 in 5% nonfat milk powder in Tris-buffered saline and applied for 1 h. Subsequently, specimens were washed three times for 20 min each. Secondary antibodies were detected by chemiluminescence (ECL; Amersham, Braunschweig, Germany).
In Situ Hybridization-The in situ hybridization procedure was performed essentially as described (13,21). Briefly, plucked anagen hairs were immediately snap-frozen in isopentane and cooled with liquid nitrogen, and cryostat sections (5 m) were mounted on sialinized glass slides. The hHa1 specific antisense cRNA probe was in vitro transcribed from a 1,040-bp cDNA subclone (9) containing the complete 3Ј-untranslated region using 35 S-CTP and T7-polymerase (Boehringer Mannheim). A 630-bp hHa1-t-specific probe of intron 6 of the hHa1 gene was generated by PCR with forward primer 5Ј-TGAGTACATGGGCAGA-CGTGTTTG-3Ј and reverse primer 5Ј-AAAATCATATGCCAGAGATA-TCCC-3Ј and cloned into pMOS Blue vector (Amersham). The antisense cRNA probe was synthesized by in vitro transcription with T7-polymerase and 35 S-GTP. All probes were used in their full length without alkaline hydrolyzation.
For microscopic recording of the in situ hybridization signals, a confocal laser scanning microscope (LSM 410 UV; Carl Zeiss, Oberkochen/Jena, Germany) was used. The instrument allowed simultaneous visualization in epi-illumination for the detection of the hybridization signals and in transmitted light in bright field for hematoxylin staining. The two signal channels were combined with an overlay in false colors (transmission image in green channel electronically changed by the computer to black/white; reflection image in red).
Immunofluorescence Microscopy-The indirect immunofluorescence procedure was carried out as described previously (21). Human anagen hair follicles were plucked and immediately frozen in isopentane precooled with liquid nitrogen. Cryosections were fixed by immersion in methanol at Ϫ20°C for 10 min. Specimens were air-dried and incubated with the primary antibodies for 45 min at room temperature. After washing with phosphate-buffered saline, specimens were incubated with the respective secondary anti-mouse or antiguinea pig immunoglobulin antibodies coupled to Cy3 and diluted 1:300 (Dianova) for 45 min at room temperature. The specimens were washed thoroughly in phosphate-buffered saline, dipped in distilled water, and mounted in Elvanol. Micrographs were taken with a fluorescence microscope (Axiophot; Carl Zeiss) using identical processing parameters.
Antibodies-All antibodies used were diluted in phosphate-buffered saline, 5% normal goat serum. Primary antibodies were pan type I hair keratin guinea pig antiserum gp19 (Ref. 4 DNA Isolation, PCR Amplification, and Sequencing-Blood was drawn from consenting members of two pedigrees. Genomic DNA was prepared using the blood and tissue culture DNA extraction system (QIAGEN, Hilden, Germany). PCR was used to amplify segments of the type I hair keratin genes hHa1 and hHa5 (9,12) that spanned the coding sequences for the terminal part of the ␣-helix, the entire carboxyl terminus including intron 6, and in the case of hHa5, the additional intron 7 (12). Amplification primers were designed from the hHa1 cDNA sequence ((9) EMBL accession number X86570: forward primer, 5Ј-TCACCAACGTGGAGTCCCAG-3Ј, nucleotide position 1006 -1025; reverse primer, 5Ј-TGCATCCTTGCTCCTCTGGCA-3Ј, nucleotide position 1316 -1336) and the hHa5 cDNA sequence ((12); EMBL accession number X90763: forward primer, 5Ј-TGGAGGCCCAGCTGGCCG-AGAT-3Ј, nucleotide position 1128 -1149; reverse primer, 5Ј-GCCTTT-GTGACAGGCTCTGAGA-3Ј, nucleotide position 1460 -1481). PCR was carried out with the Expand Long Template-PCR system (Boehringer Mannheim). After initial denaturation at 94°C for 2 min, 30 cycles were performed consisting of 94°C for 10 s, 60°C for 30 s, and 72°C for 2 min. Final extension was at 72°C for 15 min. PCR-amplified products purified by agarose gel extraction were sequenced directly according to the Sequenase-PCR sequencing protocol (U. S. Biochemical) using either the PCR primers or gene-specific internal primers.
RNA Isolation and Reverse Transcription-PCR-RNA was extracted from the bulbs of about 10 -15 freshly plucked anagen hair follicles using the RNeasy system (QIAGEN). Reverse transcription was performed with an oligo(dT) adaptor-primer and Superscript Reverse Transcriptase according to the manufacturer's instructions (Life Technologies, Inc.). Amplification of cDNA fragments with the hHa1-specific primer pair was carried out as described for PCR amplification of genomic DNA. Alternatively, the respective forward primer was replaced by the primer 5Ј-CAGTTACCAGTCCTATTTTAAG-3Ј derived from exon 1 sequences of the hHa1 cDNA ((9), nucleotide position 356 -377).

Recombinant Expression of Hair Keratins in Escherichia coli and Protein
Purification-cDNA fragments containing the complete coding sequence of either the type I hair keratin hHa1 or the type II hair keratin hHb3 were generated by PCR using the available cDNA clones for both keratins as template (9,13). Each of the respective forward primers contained an EcoRI cloning site and a Shine-Dalgarno sequence upstream of the start codon (23). The reverse primer sequences were derived from the hHa1-and hHb3-specific 3Ј-untranslated regions following the stop codon, each containing a HindIII site for cloning. For the generation of the cDNA of the truncated hair keratin hHa1-t, total cDNA obtained by reverse transcription of hair follicle RNA of a hHa1-t positive individual was used as a template. Whereas the forward primer was the same as for hHa1 cDNA amplification, the reverse primer contained intron 6 sequences specifically retained in the hHa1t mRNA. The three cDNAs were cloned into the prokaryotic expression vector pDs5 (24), sequenced to verify the integrity of the reading frame and expressed in E. coli.
Bacterial clones producing hair keratins were identified by immunoblot analysis using the guinea pig antisera gp16, specific for type II hair keratins, or gp19, specific for type I hair keratins (4). Bound primary antibodies were detected by the alkaline phosphatase system of Promega, Madison, WI.
Production of cDNA-engineered hair keratins, their enrichment in the inclusion body fraction, and purification by column chromatography was performed according to Hofmann and Franke (23) with the following modifications. All buffers used for purification contained 2 mM DTT. After purification by the Q-Sepharose column (Pharmacia Biotech Inc.), keratin-containing fractions were pooled and dialyzed against 9.5 M urea, 30 mM sodium formiate (pH 4.0), 1 mM EDTA, 2 mM DTT for 1 h at room temperature and applied to a 15-ml S-Sepharose column (Pharmacia). Bound protein was eluted in a 100-ml gradient of guanidine HCl (0 -0.6 M) in column buffer. Keratin-containing fractions were checked for purity by SDS-polyacrylamide gel electrophoresis and pooled.
Assembly Experiments-Equimolar amounts of purified type I and type II hair keratins were mixed in 9.5 M urea, 10 mM Tris-HCl (pH 7.5), 2.5 mM EDTA, 2 mM DTT. The urea concentration was gradually lowered by step-wise dialysis against 5 M urea and 2.5 M urea, buffered with 10 mM Tris-HCl (pH 7.5), 2.5 mM EDTA, 2 mM DTT at room temperature. Assembly was induced by a final dialysis step against 160 mM NaCl, 25 mM Tris-HCl (pH 7.5), 2 mM DTT for 2 h at room temperature. The assembled protein material was analyzed by negative staining as described (25,26).

Screening for the Low Molecular Weight 41-kDa Hair Protein: Evidence for a Type I Hair Keratin
Variant-In a large scale screening study, keratins were extracted from clipped hairs of 30 unrelated healthy male and female individuals. Upon one-dimensional electrophoresis, keratins from mature hairs were generally separated into two broad bands, one migrating between 59 and 54 kDa and comprising the members of the type II hair keratin subfamily, the other occurring between 48 and 44 kDa, harboring the type I subfamily (2). However, the keratin patterns of two unrelated male individuals clearly exhibited the additional low molecular weight protein (see arrowheads in Fig. 1A, individual 2, and Fig. 1B, individual 2). The protein possessed an estimated size of 41 kDa in our gel system, which is in good agreement with previous reports (14 -19). Subsequent two-generation family hair keratin analyses of the 41-kDa-positive individuals confirmed the autosomal dominant transmission of the protein (Fig. 1, A and B). Light microscopic examination of plucked hairs did not reveal macroscopically visible differences between 41-kDa positive and negative individuals of both pedigrees.
In several previous electrophoretic hair keratin analyses, a diminished protein staining in the high molecular weight part of the type I hair keratins was consistently noticed in hair keratin patterns containing the 41-kDa component (16,18,19). This observation could be confirmed in pedigree A (Fig. 1), in particular by comparing the type I hair keratin pattern of the 41-kDa protein-negative individual 1 with those of the 41-kDa protein-positive individuals 2 and 3 (Fig. 1A). Moreover, Yu et al. (18) recently described a family in which both parents were positive for the 41-kDa protein. These individuals as well as two of their offsprings showed type I hair keratin patterns similar to those of individuals 2 and 3 of pedigree A (Fig. 1), whereas a third member of the F1 generation clearly lacked a high molecular weight type I hair keratin (18). Collectively, these data led us to hypothesize that the 41-kDa protein might be a size variant of a large type I hair keratin. Out of the seven human type I hair keratins recently cloned in our laboratory, the cortex keratin hHa1 and the matrix keratin hHa5 are the largest proteins, having calculated molecular weights of 47,378 Da and 47,727 Da, respectively (9,12). Assuming alterations in the genes of these two keratins, they were subjected to mutation analyses.
Mutation Analyses of the Type I hHa1 and hHa5 Hair Keratin Genes: Detection of a Splice Site Mutation in the hHa1 Gene-At the gene level, a possible relationship between the hHa1 or hHa5 hair keratin and the 41-kDa protein can most easily be explained by a mutational event leading to a premature stop codon in one of the candidate genes and entailing the loss of about 50 amino acid residues. Since the non-␣-helical tail domains of hHa1 and hHa5 are almost equally long (49 and 47 amino acid residues, respectively (9,12)), such a mutation should in both cases be located in the gene region encoding the transition between the ␣-helical rod domain and the carboxyl terminus. Therefore, genomic DNA fragments comprising the coding sequences of this region for both genes were amplified from blood DNA of 41-kDa protein-negative and -positive individuals of pedigrees A and B (Fig. 1) by means of hHa1-and hHa5-specific PCR primers. Since the genes of both keratins are known (partial hHa1 and hHa5 gene sequences (12,27) and own data for the complete hHa1 and hHa5 genes), 1 the size of the amplified genomic fragments could be predicted.
Independent of the source of the DNA, amplification with the specific hHa5 primers yielded a fragment that corresponded to the calculated size of 780 bp and that contained sequences for intron 6 as well as for an additional intron 7 (12). It should be emphasized that except for hHa5, type I hair keratin genes do not possess a 7th intron (Ref. 12 1 ; similar to the epithelial type I keratin K9 (28), the hHa5 7th intron is located in the 3Ј noncoding region of the gene). Sequencing of the fragments did not reveal sequence variations between 41-kDa protein negative and positive individuals (results not shown).
Amplification with the specific hHa1 primers yielded a fragment corresponding to the calculated size of 1007 bp that contained 677 bp of intron 6 (partial hHa1 gene sequence (27) and own complete hHa1 gene sequences 1 ; in Ref. 27, this intron has erroneously been designated intron 7). The sequences of the DNA fragments obtained from 41-kDa protein-negative individuals did not deviate from the corresponding region of the published hHa1 cDNA and genomic sequence (9). 1 In contrast, sequencing of the 5Ј end of the DNA fragment from 41-kDa protein-positive individuals led to the detection of a heterozygous G-A mutation. Fig. 2 illustrates that this point mutation affects the initial G of the 5Ј donor splice site, GTGAGT, of intron 6 of the hHa1 gene. In both pedigrees, the heterozygous G-A mutation within the hHa1 gene segregated with the 41-kDa protein phenotype, whereas 30 unrelated 41-kDa proteinnegative individuals were homozygous for the wild type donor splice site sequence, GTGAGT (results not shown).
To investigate the consequences of the donor splice site mutation in the hHa1 gene at the mRNA level, RNA isolated from anagen hair follicles of a 41-kDa protein-negative and a 41-kDa protein-positive individual was reverse transcribed and amplified by means of the same hHa1-specific primer pair in exons 6 and 7 used for the amplification of genomic DNA. In the 41-kDa protein-negative individual, this amplification yielded a single cDNA fragment a that exhibited the expected size of 329 bp (Fig. 3a, lane 2), whereas the amplification of the corresponding cDNA region of the 41-kDa protein-positive individual led to an additional, slightly larger 370-bp fragment b (Fig. 3a,  lane 3). Sequencing of fragment a from both individuals revealed complete sequence identity with the corresponding region of the published hHa1 cDNA clone (9). This was also true for fragment b from the 41-kDa protein-positive individual, except for the intercalation of a 41-nucleotide stretch (Fig. 3c). These additional nucleotides corresponded to the 5Ј end of intron 6 but contained the G-A-mutated 5Ј splice site already detected in one hHa1 allele of 41-kDa-positive individuals (Fig.  2). The retention of only 41 out of the 677 nucleotides of intron 6 (27) in this aberrant mRNA species could be attributed to a cryptic donor splice site motif, GTGAGT, following the 41th Total hair keratins from mature hairs of each family member were resolved by either 10% (family A) or 12% SDS-polyacrylamide gel electrophoresis (family B). Collective type I and type II hair keratins are indicated at the right-hand side of each gel panel. The 41-kDa protein is marked by an arrowhead. Note that in 10% gels, the type I hair keratins tend to be resolved into two bands (i.e. family A, lane 1), whereas in 12% gels the same phenomenon is observed for type II hair keratins. The two bars at the left-hand side of each panel represent molecular mass markers vimentin (57 kDa) and actin (43 kDa).
intron nucleotide, which is sequentially identical with the authentic 5Ј splice site of intron 6 ( Fig. 3d) and is obviously used as an alternative splice site for the elimination of the remaining intron 6 sequences in the resulting transcript.
To investigate whether, besides normal hHa1 transcripts and aberrant hHa1 transcripts containing intron 6 sequences, additional hHa1 transcripts were present in 41-kDa proteinpositive hair follicles, reverse transcription-PCRs with RNA from the 41-kDa protein-negative and -positive individual were performed using a forward primer in exon 1 instead of exon 6 of the hHa1 gene. Except for a 980-bp fragment, virtually no other products were seen in both amplification assays (Fig. 3b, lanes  1 and 2). Subcloning of the fragment from the 41-kDa proteinpositive individual and sequencing of several clones revealed a nearly 1:1 ratio of clones exhibiting either the normal or the aberrant hHa1 cDNA sequence (results not shown).
Identification of the 41-kDa Protein as hHa1-t, a Truncated Isoform of hHa1-Amino acid translation of the aberrant hHa1 mRNA species led to a hHa1 protein variant that exhibited an asparagine-lysine substitution in the second last position of the ␣-helical rod domain but which prematurely terminated immediately thereafter due to an in-frame chain termination codon, TGA, originating from the intron 6-derived sequence of the mRNA (Fig. 3c). Thus, the resulting truncated hHa1 protein isoform lacked the last ␣-helical amino acid leucine (Leu-367) as well as the complete nonhelical tail domain of hHa1 and exhibited a calculated molecular mass of 42,290 Da. Considering that the aberrant hHa1 mRNA species could only be detected in 41-kDa-positive individuals, there is every reason to believe that its translation product is identical with the 41-kDa protein and justifies its designation as truncated hHa1 isokeratin hHa1-t.
To further prove this, cDNA reverse-transcribed from follicular RNA of a 41-kDa protein-positive individual was used as a template for the amplification of the complete coding region of hHa1-t mRNA species. The amplification product was cloned into a prokaryotic expression vector as was the coding portion of the hHa1 cDNA (9). After expression in E. coli, the recombinant hHa1 and hHa1-t proteins were purified and subjected to one-dimensional gel electrophoresis using the hair keratin preparation from individual 2 of pedigree A (Fig. 1) as a 41-kDa protein-positive reference pattern. This analysis clearly showed that the recombinant hHa1-t comigrated with the 41-kDa protein (Fig. 4a, lane 3, and arrowhead in lane 1). In Western blots, both proteins were recognized by a pan type I hair keratin antiserum (Fig. 4b, lanes 1 and 3), thus demonstrating the type I keratin nature of the 41-kDa protein. Moreover, a monoclonal antibody raised against the last 15 amino acid residues of the hHa1 carboxyl terminus (22), which selectively recognized its natural and recombinant antigens (Fig. 4c,  lanes 1 and 2), did not react with the recombinant tailless hHa1-t keratin and also left the 41-kDa protein undetected (Fig. 4c, lanes 1 and 3).
Gene Expression Studies of hHa1 and hHa1-t-In situ hybridization experiments on sections of human anagen hair follicles were first performed using a specific riboprobe comprising common 3Ј noncoding sequences of the hHa1 and hHa1-t mRNAs. As expected, this probe produced equally strong hybridization signals in hair follicles of both hHa1-t-negative individuals and individuals heterozygous for hHa1-t (Fig. 5 a  and aЈ). Hybridization signals decorated the entire hair cortex from an area just above the apex of the dermal papillae up to the keratogenous zone (Fig. 5, a and aЈ). In contrast, a riboprobe specifically tailored to contain only intron 6 sequences of the hHa1 gene did not react with hHa1-t-negative follicles (Fig.  5b), whereas follicles from individuals heterozygous for hHa1-t were clearly labeled over the same area as with the 3Ј-specific probe (Fig. 5bЈ). Apparently, the labeling intensity observed in Fig. 5bЈ does not appropriately reflect the 1:1 ratio of the hHa1 and hHa1-t transcripts relative to Fig. 5aЈ. Most probably, this is due to the fact that the hybridizing portion of the 630-bp probe comprised solely the initial 41 nucleotides of intron 6, the non-hybridizing remainder of the probe being removed during a RNase treatment step routinely performed after in situ hybridization. Expression studies on the protein level with the specific hHa1 carboxyl-terminal antibody (22) revealed a strong response in the cortex of hHa1-t-negative follicles, whereas under identical conditions, follicles containing hHa1 as well as hHa1-t exhibited the expected quantitative reduction in reaction intensity (Fig. 5, c and cЈ).
In Vitro Filament Assembly of hHa1 and hHa1-t with a Type II Hair Keratin-To compare the assembly capacity of the tailless hair keratin hHa1-t with that of its normal counterpart hHa1, we chose a recombinant type II hair keratin as an in vitro association partner. Considering that hHa1 and hHa1-t are expressed in the hair cortex, we selected keratin hHb3, a medium-sized type II hair keratin (calculated molecular mass, 54,200 Da), which exhibits a similar expression pattern in the hair cortex as hHa1 (13). As shown in Fig. 4a, lane 4, the recombinant hHb3 migrated as expected when compared with the total type II hair keratin pattern (Fig. 4a, lane 1) and was detected by a pan type II hair keratin specific antiserum (Ref. 4; results not shown).
Since the in vitro assembly conditions for isolated hard akeratins have not yet been investigated, we first dialyzed mixtures of near equimolar amounts of either hHa1/hHb3 or hHa1t/hHb3 in 9.5 M urea under conditions usually employed to assemble epithelial keratins in vitro (i.e. Tris-HCl concentrations between 5 and 50 mM (29 -31)). By these means, we were unable to demonstrate the formation of hair keratin IFs. Surprisingly, however, buffer conditions generally used to induce the in vitro assembly of vimentin, i.e. NaCl concentrations in the range of 100 -160 mM (25,26,32), clearly led to the formation of smooth-surfaced IFs of 9 -10-nm width and 500-nm to 1-m length, independent of whether hHa1 or hHa1-t were associated with hHb3 (Fig. 6, a and b). It should be mentioned that similar looking filaments were also obtained when the recombinant type II epithelial keratin K8 was used as a partner for hHa1 or hHa1-t, although in this case, supplementation of the assembly buffer with salt could be omitted (results not shown). Collectively, our in vitro assembly data indicate that, in comparison to its normal counterpart, the capacity of tailless hHa1-t to form IFs with a type II keratin is not impaired.

DISCUSSION
In the present study we have shown that a point mutation in the 5Ј donor splice site of intron 6 in one allele of the human type I hair keratin gene hHa1 is responsible for the heterozygous expression of a truncated hHa1 protein variant that completely lacks the entire nonhelical tail domain. The tailless hair keratin hHa1-t is identical with a low molecular weight, 41-kDa protein previously described as a sporadically occurring, but genetically transmitted component of human hair keratin patterns (14 -19).
The tailless hHa1-t represents the first example of a keratin variant originating from a point mutation in a keratin gene intron splice site. The generation of the hHa1-t mRNA in response to the mutation in the 5Ј splice site of intron 6 that involves the activation of a cryptic 5Ј splice site 41 nucleotides downstream of the G-A mutation site deserves some comments.
It is well documented that mutations in either the first or the second base of the obligate 5Ј donor splice site consensus sequence GT of introns can induce different forms of missplicing. Among the observed aberrations are the skipping of the exon immediately upstream of the mutated intron by the alternative use of the 5Ј splice site of the preceding intron, the activation of normally quiescent cryptic 5Ј splice sites either downstream or, less frequently, upstream of the mutated wild type splice site, or the intron read through in the absence of suitable cryptic splice sites (for review, see Refs. 33 and 34). Generally, exon skipping is more frequently observed than the use of a cryptic splice site, but there are also numerous examples where a 5Ј splice site mutation generates multiple aberrant transcripts with relatively widely differing proportions that result from combinations of either all or two of the above cited missplicing mechanisms (33-37).  2 and 3). c, Western blot of a using the hHa1 tail-specific monoclonal antibody LHTric-1 (22). Note staining of natural (lane 1) and recombinant hHa1 (lane 2). Labels at the right-hand site indicate molecular mass markers: carbonic anhydrase (29 kDa), glyceraldehyde-3-phosphate dehydrogenase (36.9 kDa), ovalbumin (45 kDa), bovine serum albumin (66 kDa) In the case of the 5Ј splice site mutation in intron 6 of the hHa1 gene, we did not observe multiple misspliced transcripts. For example, a mRNA species larger than that encoding hHa1-t, resulting from intron 6 read through, could not be detected. Similarly, the formation of a mRNA species smaller than the hHa1-t transcript and originating from skipping of exon 6 could be excluded by PCR amplification of reversetranscribed follicular RNA from a hHa1-t-positive individual by means of an exon 1 primer in combination with a primer in the 3Ј noncoding region of hHa1 cDNA (see Fig. 3b). The apparently exclusive utilization of the cryptic 5Ј splice site of intron 6 in the processing of the mutated hHa1 pre-mRNA is in line with the notion that this type of alternative splicing is preferred over exon skipping, the closer the cryptic splice site is to the normal splice site and the more their extended sequences are homologous (33,34). Both of these features are particularly pronounced in the present case. In addition, there is strong evidence that the splicing machinery recognizes the 5Ј donor and 3Ј acceptor site of a given intron as a unit (38). The 5Ј splice site sequences of introns 5 (GTGTGT) and 4 (GTGGGC) of the hHa1 gene both deviate over the the first six bases from the corresponding site of intron 6. Therefore, once the latter is mutationally altered, the selection of the sequentially identical cryptic splice site in intron 6 becomes the most suitable and efficient option for the processing of the hHa1-t pre-mRNA.
Furthermore, skipping of exon 6 as a possible consequence of the 5Ј splice site mutation in intron 6 of the hHa1 gene can also be conceptually excluded. It is well documented that mutations in epithelial keratin genes can be causally related to the manifestation of a large number of inherited epidermal fragility syndromes (for recent reviews, see Refs. 39 and 40). We have recently also shown that mutated hair keratin genes may un-(right-hand side panels). a and aЈ, in situ hybridization with a cRNA probe recognizing both hHa1 and hHa1-t mRNAs. Note the equally intense hybridization signals in the entire hair cortex of both types of follicles. b and bЈ, in situ hybridization with a hHa1-t mRNA-specific cRNA probe. Note the complete absence of hybridization signals in the follicle in b and weak signals over the cortex of the follicle in bЈ (for further details, see "Results"). c and cЈ, indirect immunofluorescence with monoclonal antibody LHTric-1 specifically directed against the penultimate carboxyl-terminal hHa1 sequence (22). The micrographs were taken and processed under identical photographic conditions. Note the strong reaction in the follicle homozygous for hHa1 (c) and the appropriately weaker reaction in the follicle with heterozygous hHa1-t expression (cЈ). derlie the etiology of certain hereditary hair diseases (41,50). Most frequently, these pathological mutations occur in the gene regions that encode the highly conserved initiation and termination motifs of the central ␣-helix of keratins. The structural integrity of these ␣-helical segments has been shown to be indispensable for the formation of both intact keratin heterodimers and higher order intermediate filaments (39,40).
In the case of an intron 6 mutated hHa1 gene, skipping of exon 6 would imply the removal of the complete terminal region of the hHa1 rod domain including its termination motif. Predictably, this would lead to the expression of a highly pathogenic hHa1 keratin variant, which would clearly be incompatible with both the unremarkable hair phenotype and the normal appearance of hair follicles in individuals who carry the splice site mutation in the hHa1 gene.
These features can, however, be reconciled with the expression of the truncated hHa1-t keratin in the hair cortex. There is a large body of evidence that the nonhelical tail domain of keratins is dispensable for normal IF formation. This has first been concluded from the occurrence of the type I keratin K19, which exhibits a rudimentary, 13-amino acid residues-long carboxyl terminus but is fully competent in filament assembly (42). The formation of normal sized and normal looking IFs has also been shown by in vitro association studies with tailless K8/K18 or K5/K14 keratins, although IF association and stability were reported to be slightly altered in tail-clipped keratin studies (30,31). Moreover, transfection experiments with a mutated keratin K14 cDNA encoding a protein variant that contained only the two carboxyl-terminal amino acid residues adjacent to the rod domain showed this tailless keratin to be incorporated into the keratin network of the host cell without any alteration in morphology (45,46). Similarly, transfections of cDNA clones coding for tailless K8 and K18 into cells that by their own did not express keratins led to the de novo formation of regular IFs (47). Consistent with these data, our own in vitro assembly studies of either hHa1 or hHa1-t with the type II hair keratin hHb3 revealed no differences in the appearance of the formed IFs and thus justify the conclusion that the in vivo expression of the tailless hHa1-t does not compromise the formation of a normal IF network in hair cortex cells. In this context it should be emphasized that besides the complete tail portion, hHa1-t also lacks the last two regular amino acid residues of the ␣-helical rod, which in addition terminates with a lysine residue instead of an asparagine residue. It is, however, known that the last few residues of the helix termination motif are generally less stringently conserved among keratins (48). Moreover, synthetic helix termination motif peptides, nonconservatively mutated in these positions, have been shown to retain full disassembling capacity on an existing keratin intermediate filament network (49). These data indicate that alterations in these ␣-helical positions are less critical for proper filament formation and that mutations therein are not likely to lead to pathology.
Although the tail domain of hair keratins seems to be as dispensable for undisturbed IF formation as that of epidermal keratins, this does not generally imply a lack of function. Unlike epithelial keratins, hair keratins, in particular those expressed in the cortex, possess a high content of positionally largely conserved cysteine residues in their head and tail domains (7,8,13) that are thought to undergo covalent S-Sbridging with hair-specific high sulfur and ultra high sulfur matrix proteins to stabilize protofibrillar interactions at late stages of hair differentiation (43,44). Thus subtle alterations in the organization and stability of IF-matrix protein complexes in hairs of hHa1t-positive individuals may not be excluded. Considering that in the hair cortex at least four different pairs of keratins are expressed (7,8,13), it is, however, conceivable that the heterozygous expression of a single tailless cortex keratin occurs without a morphologically discernable hair phenotype.