Sequence, expression, and evolutionary conservation of a gene encoding a glycine/tyrosine-rich keratin-associated protein of hair.

In hair differentiation several families of keratin proteins with distinctive amino acid compositions are produced. To study the role and regulation of one of these families, the glycine/tyrosine-rich keratin-associated proteins encoded by the KAP6 gene family, a partial wool follicle cDNA clone encoding a sheep KAP6 protein was sequenced and the corresponding gene isolated from a sheep cosmid library. The KAP6.1 gene encodes a basic protein of 82 amino acids (M(r) = 8,296) with a combined glycine and tyrosine content of approximately 60 mol%. There are several KAP6 genes in the sheep genome, all located within a 1,050-kilobase SfiI fragment. Northern blot analysis demonstrated that at least one member of the KAP6 family is expressed in the wool follicle. A rabbit KAP6 gene was isolated and its sequence and expression patterns were compared with the sheep gene. The sheep and rabbit genes have a nucleotide sequence identity of 89%, suggesting that they are equivalent genes and indicating strong selection pressure during evolution. Both genes contain several conserved sequence motifs of 7-9 nucleotides in their 5'-flanking regions that may be involved in the regulation of their expression. Localization of KAP6 mRNAs in sheep wool and rabbit hair follicles by in situ hybridization suggests that the genes are expressed in the cells of the hair shaft cortex in varying expression patterns. KAP6 expression starts relatively late in hair follicle differentiation, and the proportion of hair cortical cells that express it may change from follicle to follicle.

of -60 mol %. There are several KAPG genes in the sheep genome, all located within a 1,050-kilobase SfiI fragment. Northern blot analysis demonstrated that at least one member of the KAP6 family is expressed in the wool follicle. A rabbit KAPG gene was isolated and its sequence and expression patterns were compared with the sheep gene. The sheep and rabbit genes have a nucleotide sequence identity of 89'70, suggesting that they are equivalent genes and indicating strong selection pressure during evolution. Both genes contain several conserved sequence motifs of 7-9 nucleotides in their 5"flanking regions that may be involved in the regulation of their expression. Localization of KAPG mRNAs in sheep wool and rabbit hair follicles by in situ hybridization suggests that the genes are expressed in the cells of the hair shaft cortex in varying expression patterns. KAPG expression starts relatively late in hair follicle differentiation, and the proportion of hair cortical cells that express it may change from follicle to follicle.
An intriguing feature of hair structure is the multiplicity of the hair keratin proteins. The wool fiber, for example, is composed of about 50-100 keratin proteins derived from several multigene families (1)(2)(3)(4)(5). These proteins are broadly classified into two groups: the intermediate filament (IF)' keratin proteins and the keratin IF-associated, or matrix, proteins (6). The keratin IF-associated proteins are characterized by high proportions of one or two amino acids and *This work was supported by a grant from the Wool Research Trust Fund on the recommendation of the Australian Wool Corporation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18  have been classified as ultra-high-sulfur, high-sulfur, or glycine/tyrosine-rich proteins (7). A unified nomenclature for these proteins has recently been proposed' and is used in this report. The hair matrix proteins are acknowledged as keratinassociated proteins and are referred to by the prefix KRTAP, abbreviated to KAP in general usage, with a number to denote each family. The KAPG family is the glycine/tyrosine-rich type I1 proteins, KAP7 is glycine/tyrosine-rich type I component C2, and KAP8 is glycine/tyrosine-rich type I component F.
Hairs have a common structure composed of a cuticle sheath, an inner cortex, and a central medulla, the relative proportions of which vary with fiber type. The cuticle and cortical cells synthesize the keratin proteins (8). Within the cortex of the wool fiber there are two major cell types, termed the paracortical and orthocortical cells (6), that contain different proportions of the keratin proteins (9). As terminal differentiation proceeds during fiber growth, the keratin IFs in the cortical cells are assembled into a filamentous scaffold that becomes embedded in a matrix of the KAPs (10). The orthocortical cells appear to have a higher 1F:matrix ratio than the paracortical cells and a different ultrastructural organization (11).
The glycine/tyrosine-rich KAPs, the smallest of the hair keratins (Mr = 6,000-9,000) were originally separated into two groups on the basis of amino acid content and solubility (lZ), type I (KAP7 and 8), and type I1 (KAP6 family). These proteins are rich in glycine, tyrosine, serine, and phenylalanine, accounting for -50 mol % of the amino acid content for KAP7 and 8 and -77 mol % for KAPG proteins. The glycine/ tyrosine-rich KAP group is heterogeneous and may contain 15-30 components (13), although there is some speculation that the apparent heterogeneity may, in part, have arisen from the conditions of protein preparation and fractionation (14). This uncertainty can be resolved by molecular cloning of members of this family.
The glycine/tyrosine-rich KAPs vary in abundance from less than 3% in human hair and the wool of Lincoln sheep to 13% in Merino wool, and up to 30-40% in echidna quill (15). The wide range in the content of these proteins in wool and hair raises intriguing questions concerning the regulation and function of these proteins in the matrix structure of the fiber. Further, they are subject to variation through dietary, physiological, and genetic factors (7).
The determination of the amino acid sequences of KAP7 (14) and KAP8 (16) enabled the synthesis of specific oligonucleotide probes to isolate the corresponding sheep cDNAs and genes (17,18). Recently, a partial amino acid sequence was derived from a KAPG protein isolated from wool (7).
Based on this sequence, an oligonucleotide probe was used to isolate a partial cDNA clone encoding a KAP6 protein from a sheep wool follicle cDNA l i b r a r~.~ We report in this study the use of this clone for the isolation and sequencing of sheep and rabbit KAPG genes, the investigation of the genomic organization of the sheep KAP6 family, and the expression patterns of the genes during hair follicle differentiation.

MATERIALS AND METHODS
Characterization of a Sheep KAP6.1 cDNA Clone-A sheep wool follicle poly(A)+ cDNA library (17) was screened with a 15-mer oligonucleotide mix (5'-RTARTANCCRAANCC-3'; Fig. 2A, n t 232-246) complementary to the pentapeptide (Gly-Phe-Gly-Tyr-Tyr) at the carboxyl terminus of a sheep KAPG protein (7). A strongly hybridizing cDNA clone, pKAP6.1, was isolated and pu~ified.~ The cDNA insert encoded 80% of the coding sequence for a KAPG protein, as identified by comparison with the protein sequence (7), a 3'noncoding region of 296 bp and a poly(A) tail of 26 bp (Fig. 2 A , n t 51-549). Both coding (Fig. 2 A , nt 51-270) and 3"noncoding probes ( Fig. 2 A , n t 295-503) were subcloned from a Bal-31-deletion clone of pKAP6.1. Unless stated otherwise, these probes were used in the library screenings, Southern and Northern blot analyses, and for the sheep in situ hybridizations.
Sheep Cosmid Library Screen-Approximately 1.5 genome equivalents of a sheep (Merino X Dorset Horn cross) cosmid library (Clontech) were screened by the colony hybridization method (19) with the sheep xAP6.Z 3"noncoding probe. From the four positive clones, one clone (KAPG Cosl) was further purified by two additional rounds of rescreening. Cosmid DNA was prepared from 5-10-ml cultures grown overnight at 30 "C by standard methods (20).
Rabbit X Library Screen-Approximately 5 genome equivalents of a rabbit genomic EMBL3 library (Clontech) were screened by the plaque hybridization method (21) using the sheep KAP6.I coding probe. Of the 91 hybridizing X clones, 12 were further purified with an additional three rounds of rescreening, and one was selected for further analysis. X DNA was prepared from plate lysates (20).
Southern Blot Analysis-DNA restriction fragments to be used as probes were labeled with either [ c Y -~* P ]~A T P and/or [ o -~* P ]~C T P (3,000 Ci/mmol, Bresatec, Adelaide, South Australia) by the oligolabeling method (22) using a Bresatec kit. Restriction enzyme-digested DNAs, as well as HindIII-digested X DNA and EcoRI-digested SPP-1 DNA molecular weight markers (Bresatec), were electrophoresed on 0.7-1.0% agarose gels and transferred to Zeta-Probe or Zeta-Probe G T (Bio-Rad) in 0.4 M NaOH (23) by a vacuum blotting apparatus (Pharmacia LKB Biotechnology Inc.). The membrane was then UV cross-linked (Stratagene). For hybridization, filter bound DNA (or RNA, see below) was prehybridized for at least 2 h at 41 "C in 50% formamide, 10% dextran sulfate, 5 X SSC, 0.5% SDS, 1 X Denhardt's reagent (20), and 100 pg/ml autoclaved salmon sperm DNA. Heatdenatured labeled probe was added to the same solution at 1 X lo6 cpm/ml and hybridization was carried out overnight. The filters were washed in two changes of 2 X SSC, 0.1% SDS at 25 "C and then twice a t 65 'C for 30 min. If required, filters were washed in 0.1 X SSC, 0.1% SDS at 65 "C for 30 min. Blots were stripped by two washes in boiling 0.1 x SSC, 0.1% SDS for 15 min, followed by a brief rinse in 2 X SSC a t 25 "C.
Northern Blot Analysis-Total RNA was isolated from wool follicles of Corriedale sheep by the acid guanidinium thiocyanate method (24). RNA (10 fig) and RNA molecular weight markers (GIBCO/ BRL) were electrophoresed on a 1.0% agarose gel containing 0.66 M formaldehyde (20) and transferred to Zeta-Probe GT using vacuum blotting and 10 X SSPE as the transfer buffer. The membrane was UV cross-linked.
DNA Subcloning and Sequencing-DNA fragments generated by digestion with restriction enzymes, progressive deletion using Bal-31 (25), or with Exonuclease 111 (26) using a kit (Pharmacia) were subcloned into appropriate M13 mp18/19 (27) or pGEM (Promega) vectors using established protocols (20). Single-stranded template DNA for dideoxy sequencing (28, 29) was prepared by the method of Winter and Fields (30). Double-stranded pGEM template DNA was prepared by the boiling-lysis method (20) and purified through a Sepharose CL-GB (Pharmacia) mini-column prior to sequencing. Template DNAs were sequenced using [ c Y -~~S I~A T P (1,000-1,500 Ci/ mmol, Bresatec) and either the Klenow sequencing kit (Bresatec) for E. Kuczek, unpublished experiments. single-stranded templates or a Sequenase Version 2.0 sequencing kit (U. S. Biochemical Corp.) for double-stranded templates. The DNAs were sequenced in both directions. Sequencing data were manipulated with programs from the IDEAS sequence analysis software package (31).
Pulsed Field Gel Electrophoresis-Sheep genomic DNA was isolated from white blood cells and embedded in agarose beads (32). Agarose beads containing -15 pg of DNA were digested with SacII, SalI, Sfii, and Not1 in 1 X digestion buffer (33 mM Tris-HAC, pH 7.8, 62.5 mM KAc, 10 mM MgAc, 4 mM Spermidine, 0.5 mM dithioeryth-rit01)~ and, with Saccharomyces cereuisiae chromosomes and X multimers as molecular weight markers (Bio-Rad), electrophoresed on 1.0% agarose gels using a CHEF-DR I1 system (Bio-Rad) with a ramp pulse time of 1-10 s, for resolving DNAs in the 23-200-kb range, or 50-90 s for resolving DNAs in the 200-2,200-kb range. The electrophoresis run time varied from 16 h at 200 V for the first ramp times, to 24 h at 175 V for the second ramp times. After electrophoresis, the gels were placed in 0.25 M HCI for 20 min, in 0.5 M NaOH, 1.5 M NaCl for 30 min and then in 1 M NH~Ac, 0.02 M NaOH for 20 min. The gels were transferred onto Zeta-Probe G T using capillary transfer (20) and 1 M NH~Ac, 0.02 M NaOH as the transfer buffer. The filters were given a 5-min rinse in 2 X SSPE and then baked in a vacuum oven a t 80 "C for 2 h.
Tissue in Situ Hybridizations-In situ hybridizations on paraformaldehyde-fixed and sectioned sheep wool and rabbit hair follicle biopsies were performed as previously described (33). Final wash conditions were 0.1 X SSPE at 55-60 "C for 30 min. The cRNA and RNA probes were labeled with [ o -~~S J U T P (1,402 Ci/mmol, Bresatec) using either T7 or SP6 RNA polymerase (34) with a kit (Bresatec). For the sheep in situ hybridizations, a pGEM-SZf(+) clone containing the sheep KAP6.1 coding insert was linearized with Hind111 and transcribed with T 7 RNA polymerase to produce antisense RNA. The same clone was linearized with EcoRI and transcribed with SP6 RNA polymerase to produce sense RNA. For the rabbit in situ hybridizations, a pGEM-7Zf(+) clone containing the 1,169-bp EcoRI fragment encoding the rabbit KAP6.Z gene was linearized with Hinff ( Fig. 3, nt -61 to -56) and transcribed with T 7 RNA polymerase to produce antisense RNA. For the sense RNA, a clone containing the same insert but in the opposite orientation was linearized with SmaI (in vector polylinker), transcribed with T7 RNA polymerase, and then subjected to partial alkaline hydrolysis (35).

RESULTS
Characterization and Sequence of a Sheep KAP6.1 Gene-A sheep cosmid clone was isolated with a 3"noncoding probe from the cDNA clone pKAP6.1 (see "Materials and Methods"). Southern blot analysis of KAPG Cosl indicated that three EcoRI fragments of 9.9, 7.3, and 1.4 kb hybridized with the KAP6.1 coding probe, and only the 9.9-kb EcoRI fragment hybridized with the KAP6.1 3"noncoding probe (Fig. 1). The 9.9-kb EcoRI fragment was subcloned and Southern blot analysis of this clone identified a 1.6-kb PstI fragment that hybridized with both the KAP6.1 coding and 3'-noncoding probes (data not shown). The nucleotide and deduced amino acid sequence of the KAP6.1 gene encoded by the 1.6-kb PstI fragment is shown in Fig. 2 4 . The nucleotide sequence of pKAP6.1 and the genomic clone were identical, indicating that the cosmid contained the genomic equivalent of the cDNA clone. The KAP6.1 gene, like the other IF-associated keratin genes expressed in the hair follicle (6), does not contain introns. A number of general regulatory elements were identified in the noncoding and flanking regions of the gene (Fig. 2 A ) . In addition, there were several conserved sequence motifs of 7-9 nucleotides that had been initially identified in the 5"flanking regions of the KAP7 and KAPB genes, namely the HGT-1 and HGT-2 motifs (39). From the nucleotide sequence, an mRNA of 610 nt is N o t e the faint hvlxidizat ion signal seen nt 7.3 kh in lnnc 2 is most likrly nonsprrilir sinre its intensity is ~~pproximatrlv the same as the signal seen at 8.8 kh in Inn(' 2 whirh is the cosmid vrctor, p\VEl5 (20). predicted (exclusive of the polV(A) tail) encoding a hasic protein of 82 amino acids ( M , = 8,296) with a combined glycine (37.8 mol '7)) and tyrosine (22 mol %) content of -60 mol %. The protein had a serine and cysteine content of 14.6 and 11 mol 5 , respectively, with leucine, arginine, phenylalanine, and asparagine accounting for the rest of the compo-&ion. Featured in the coding seqr~ence were $-and 12-amino acid repeats (Fig. 2 H ) .
Charactcrizntion nnd Sequence of n Rnhhit KAI'6.1 Gene-To examine the evolutionary conservation ofthe KAP6 genes, a rabbit X genomic clone was isolated using the sheep pKAP6.1 coding probe (see "Materials and Methods"). A Southern blot filter of the X clone (KAP6 X I ) digested with I.:coRI was hyhridized with the sheep KAP6.1 coding prohe, and a unique hybridizing restriction fragment of 1.2 kt, was suhcloned. The nucleotide and deduced amino acid sequence of the KAI'6.1 gene encoded hv the 1.2-kh E h R I fragment. is shown in Fig. 3. The gene does not contain introns. Many of the motifs noted in the 5"noncoding and -flanking region of the sheep gene ( Fig.  2A ) were also identified in the same regions of the rabhit gene (Fig. 3).

~A T A G A T A T G C -C A G~A~T c A c M T~M G~c A T~~c A T A G A f l A G M A c T c A -7.1 A T T A~T C A C T A C C T G M T G G T~~G~A G~T G G~~T A G A~~~~C G T G T -T A G M T A C A T A~
-712

M A A T M T T . M T T T A T T T C M T A G G G~T C C T T G T G C G T T~G~A T A T . T T A C A~T T A C A A A~~A G T T T~ -m 6
Mlf -1'1 "CI-?

G A T T M G T G M A G T C T A G~~T G A G G T A G G A~~C~~~[ . A~~G T G A T T G T C T C C T C T C~G A G C C A T T G A T T G G A T A~G C
-111  ing probe showed that only the 9.9-kb F h R I fragment hybridized to the prohe, corresponding to the EcoRI fragment, in the cosmid clone (data not. shown). When a Southern blot, of EcoRI-digested rabhit genomic DNA was hyhridized with the sheep KAP6.1 coding prohe, six hybridizing restriction fragments were detected (dat.a not shown).

TGA G G A T A C C~G G A G C C T C T C A C C C T C M C M C A T T G A A C A G T T C T G M C C C C A G~
I'ulscd Field Blot Ann/.vsis-To investigate the genomic organization of the sheep U P 6 genes, a pulsed field hlot containing genomic DNAs resolved in the 200-2,200-kh range was hyhridized with R coding probe from the sheep KAP6.1 gene. The results showed that the U P 6 genes were contained wit.hin single S/iI (1,050 kh) and SOCII, S d I , and Not1 (1,800 kb) fragments (Fig. 5). No additional hybridizing restriction fragments were detected on a hlot containing identically digested genomic DNAs resolved in the 32-200-kh range (data not shown). Northern Blot Analysis of Sheep KA P6 (knp IJxpr(vsion in Wool Folliclps-Sheep KAP6. I coding and :j'-noncoding prohes hybridized to a follicle RNA t,and(s) of -650 hases (Fig. 6) in agreement with the predicted mRNA size derived from the gene. Under identical hybridization and washing conditions, and exposure times, a stronger hyhridization signal was obtained with the coding prohe than with the 3 'noncoding prohe. Both prohes were approximatelv the same size and laheled to approximately the same specific activity. This implied t.hat the coding prohe was detecting multiple KAPG mRNAs with highly conserved coding regions h u t different 3"noncoding regions.
Localization of S17eep and Rabbit KA 1'6 Gene Transcripts in Actiw Wool and Hnir Follicles-To investigate the expression of the KAPG family during follicle differentiation, in situ hybridizations were performed with coding region prohes.  (lane 2). In the case of each hyhridization. the filters were given a final wash in 2 X SSC, 0.15 SDS at 65 "C and exposed for the same length of time. The sizes of the hvhridizing fragments are given in These prohes should detect most of the KAI% mRXAs hecause the stringency conditions o f the in situ hyhridization were lower than those of the genomic Southern analysis (0.1 X SSPE at 60 and 65 "C, respectively) that detected multiple components (Fig. 4 ) . KAI'fi expression w*as lirst tletected in differentiating hair shaft keratinocytes in sheep wool follicles and rahhit hair folliclrs at the same relative locations, at)out 200 gm ahove the proliferative zone of the follicle hit) (Figs.   7 , A and H , and 8, A and N). In rahhit follicles t w o patterns of KAP6 gene expression were seen in longitudinal sect inns (Fig. 8, A-c'). Either expression seemed to occur in ;I11 cortical cells, or it was restricted to cells in one part o f the cortrxx. Amino Acid Scyucvw ('omparisons and Scc.ondrrr\. Structure Analwis of tl7c (;l~~cinc/7~\~rosinc-ric.l~ KAl's-The aminn acid sequence derived from the sheep K,4 I'6.1 gene (Fig. 'LA j was compared with the partial sheep K A P 6 amino acid sequence deduced hv protein sequencing ( 7 ) . hereafter referred t o as KAI'6.P. antl with the sheep KA1'7 and hill% genederived amino acid sequences (18). There were a t least eight differences hetween the KAP6.1 and tiAP6.2 s e q u e n c e s (Pig.  ( 7 ) . rrfrrrrd t o ns KAl'G.2. Identical amino acids are hoxrd. Note: the KAP6.2 sequence has two tvrosine residues at the carlxngl trrminus. wherras the tiAl'6.l srqurncr contains three. This difference was most likelv due to limitations in the peptidr sequenrinc terhnicpe used to rrsnlve thr numhrr o f tyrosine residues in the peptide fragment at the carhoxyl terminus (M. Gillespie, personal communirntion). In addition, the numher and srqurnrr o f the amino acids at the amino terminus (hrnclwt) was uncertain (7). I<, the predicted amino x i t l srquenrr from the shrep Kl\/'G./ grne is compared with the derived amino acid sequences from the KA1'7 and KAI'R genes (18). 'I'he g'nps in t he seql~rnces have h e n introtlucrrl to facilitate maximum alignment. Identical amino acids are boxed. C', comparison of the gene-drrived sherp antl rahhit KAl'6.l r~rnlnr) nrid sequences. Insertions of 7 and 10 amino acid residues in the rahhit and sheep KAl'6.l srquencrs. rrsprctivrlg. arr horcrf and thrir pcjsitions are indicated IIV srnoll nrrou3.s. Identical amino acids are b/J.wd. quences ( Fig. 9 R ) . All three proteins contained a tyrosine residue a t t h e carboxyl terminus. Despite the high content of glycine the sequence Glv-Gly is relatively uncommon, occurring twice in the KAP7 and KAPG.1 sequences (three times in the rabhit KAPG.1 sequence, see Fig. 9C') and not found in the KAPR sequence. This is in contrast. to the glycine-rich kerat,ins of birds (47), reptiles (48), and human (49) and of mouse (50) epidermis.
The sheep and rabhit. KAP6.1 proteins are highly conserved, exhibiting 11-amino acid changes and two insertions/ delet,ions of 7 and 10 amino acids (Fig. 9C). I f the 7-and 10amino acid insertions/deletions in the rahhit and sheep proteins are excepted, the two genes had an 85 and 8grE. identity at the amino acid and nucleotide sequence level, respectively. An analysis of the sheep KAPG.1, KAl'7, and KA1'8 amino acid sequences using a suite of 10 protein secondary structure prediction programs (51) indicated that all thrw proteins contained a high degree of random turns. some ,j-sheets, antl no n-helices (data not shown). The glycine/t?frosine-rich KAPs may contain the structural motif' termed the glycine loop, a structure proposed to exist in glycine-rich regions of proteins (52).

DISCI:SSIOK
We have described the molecular characterization and expression patterns o ! sheep and rahhit KAl'6./ genes during hair follicle differentiation. The sheep and rabbit genes show a high degree of sequence identity, both at the nucleotide and amino acid levels, which indicates a strong selection pressure during evolution, estimated to be -90 million years for the sheep and rabbit genomes (53). It further suggests that the conserved regions in the two proteins have specific functions, possibly involving interactions with the hair keratin IFs of the cell cortex and the other keratin IF-associated proteins.
K A P G Multigene Family-Both the sheep and rabbit KAPG.l genes are derived from multigene families (Fig. 4, and data not shown). Since all the hair IF-associated genes sequenced thus far lack introns, including the two KAPG. 1 genes described in this report, and the coding regions are usually less than 1 kb in size (6), it is possible that each EcoRI fragment detected in the sheep genomic Southern blot (Fig.  4) represents a KAPG gene. All the sheep KAPG genes appear to be contained within a 1,050-kb SfiI fragment (Fig. 5). The sheep wool keratin genes from the various families tend to occur in gene clusters (1)(2)(3)(4)(5), and experiments are now in progress to determine the genomic organization of these gene families by chromosomal in situ hybridization and pulsed field blot analysis.
Northern blot analysis suggests that more than one KAPG gene is expressed in the follicle (Fig. 6). The KAPG protein sequenced by Gillespie (7) exhibited several amino acid differences to the sheep KAP6.1 protein (Fig. 9A) and most likely represents a different member of the KAPG family which we suggest be referred to as KAP6.2. Autoradiographs of two-dimensional polyacrylamide gel patterns of S-carboxymethylated wool keratin proteins (6) reveal multiple protein spots in the glycine/tyrosine-rich KAP region. One estimate of the complexity by protein chemical methods is 15-30 components (13), although there is some speculation of artefactual heterogeneity produced by the preparative procedures (14). Since the KAP7 and KAP8 proteins are encoded by unique genes (18) and there may be several KAPG genes, a molecular genetic estimate of the number of sheep glycine/ tyrosine-rich KAPs could be about 10.
Glycine/tyrosine-rich KAPs constitute about 19% of mouse hair protein (7). The mouse genome contains unique KAP7 and KAP8 genes (54) and up to 20 KAPG-hybridizing EcoRI restriction fragments (Fig. 4) possibly representing an equivalent number of genes (see above). In contrast, the glycine/ tyrosine-rich KAP content of human hair appears to be less than 3% (15), yet the human genome contains unique KAP7 and KAP8 genes (54) and probably several KAPG genes (Fig.  4). A number of possible explanations may account for this wide variation in the glycine/tyrosine protein content between the hairs of different species. It is well known that proteins are difficult to extract from human hair, extractions sometimes resulting in yields of less than 5% of total protein (55), and thus the low glycine/tyrosine protein content may be an underestimate. Alternatively, human hair may simply contain a greater proportion of other keratin proteins, and the glycine/ tyrosine-rich KAP content is indeed low. Some of the genes may differ in transcription rates between species. The human glycine/tyrosine-rich KAP genes may be expressed at a low level, but in the mouse genome, which could contain up to twice the number of KAPG genes, some KAPG genes may be expressed at high levels. Another interesting finding concerns a sheep wool fiber mutant known as the felting luster mutant. It produces a wool that is virtually devoid of glycine/tyrosinerich KAPs (56), yet the KAP7 and KAP8 genes that are abundantly expressed in Merino breed crosses (18) are present in the felting luster genome (54). If the KAPG gene family is also present that could have interesting implications for the control of expression of those genes, particularly as the genes appear to be clustered. These two issues can now be addressed by comparative analyses of glycine/tyrosine-rich KAP gene numbers and expression by Northern blot analysis and in situ hybridization with conserved KAPG, 7, and 8 gene probes.
Expression of the Sheep and Rabbit KAPG Genes in Active Hair Follicles-Hair shaft keratinocytes begin to express differentiation-specific hair keratins as soon as they leave the proliferative region of the follicle bulb (5,6). The KAPG genes are expressed in the hair shaft cortical keratinocytes (Figs. 7 and 8) a considerable distance (200 pm) above the proliferative zone of the follicle bulb and after the transcription of the hair keratin IF genes has commenced (5). This is a relatively late stage in the differentiation of the hair shaft keratinocytes (6). It is not possible to conclude whether the KAPG genes are also expressed in the cuticle cells because the cuticle layer of these hairs is only one cell thick and, at the position in the follicle where cortical KAPG expression occurs, the cuticle cells are becoming compressed and are only a few microns wide. However, if expression does occur in the cuticle cells it must be restricted to those adjacent to the cortical cells that show expression. This would suggest local variation in the differentiation of cuticle cells within a follicle, but in view of the uniform ultrastructure of cuticle cells (57) this seems unlikely. As the KAPG family has the same expression characteristics in two evolutionarily well separated species, it is likely that these data are typical of KAPG expression in mammals in general.
An intriguing feature of KAPG gene expression in hair follicle differentiation is the variation in spatial expression. In some follicles expression appears to be restricted to a few cortical keratinocytes whereas in adjacent follicles most cortical cells express the genes (Figs. 7, D and F, and 8, E and G). Furthermore, the observation of asymmetric expression along the hair shaft of some rabbit hair follicles (Fig. 8C) suggests that the proportion of cortical cells that express KAPG is fixed within a follicle but variable between follicles. To establish any underlying theme it will be necessary to examine the expression of the KAPG gene family in a series of transverse sections covering the length of many follicles. In sheep follicles there are two major types of cortical keratinocytes, orthocortical and paracortical cells, whose proportions and spatial arrangement are believed to be responsible for the crimp or curliness of wool fibers (6). KAPG expression might be confined to one of these cell types.
Conserved Motifs in the 5'-Noncoding and -Flanking Regions-The 5"noncoding and -flanking regions of the sheep and rabbit KAPG.l genes are moderately conserved with -70% identity over 394 nucleotides (compare Figs. 2A and 3). The position and sequence of several general regulatory and glycine/tyrosine-rich KAP-specific motifs are conserved in the 5"noncoding and -flanking regions of both genes, suggesting that some of these motifs may have functional roles in the regulation of transcription or translation.
The putative TATA motif of the sheep gene (TACAAA; Fig. 2 A ) is identical to that present in the hamster desmin gene (58). Other putative TATA motifs were noted in the promoter region of the sheep and rabbit genes about 50-100 bp further upstream in both cases ( Figs. 2A and 3). However, from an examination of the promoter regions of several hair keratin genes (1)(2)(3)(4)(5) these putative TATA motifs are further from the initiation codon than expected and furthermore, no CAAT motifs were identified further upstream from them. RNase protection analyses would be required to determine the functional TATA motifs.
Fisher et al. (59) have shown by immunohistochemical means that Fos protein is present in the differentiating cells of the rat hair follicle and they suggest that Fos plays a role in keratinization. The proteins Fos and Jun form a common transcriptional activator complex that binds to AP-1 motifs (60), and since the sheep and rabbit KAP6.1 genes and a hair keratin type I1 IF gene (5) contain AP-1 motifs, a role for AP-1 in hair keratin gene transcription is implicated.
Among the hair keratin genes several conserved DNA sequences have been noted, and the HGT-1 and HGT-2 motifs appear to be unique to the glycine/tyrosine-rich KAP genes (39). These motifs may be involved in determining the timing and spatial pattern of KAP6 gene expression during hair follicle differentiation and suitable experimental strategies, such as deletion and footprint analyses and oligonucleotide binding studies (61), can now be initiated to test them.