Krtap16, Characterization of a New Hair Keratin-associated Protein (KAP) Gene Complex on Mouse Chromosome 16 and Evidence for Regulation by Hoxc13*

Intermediate filament (IF) keratins and keratin-associated proteins (KAPs) are principal structural components of hair and encoded by members of multiple gene families. The severe hair growth defects observed upon aberrant expression of certain keratin and KAP genes in both mouse and man suggest that proper hair growth requires their spatio-temporally coordinated activation. An essential prerequisite for studying these cis-regulatory mechanisms is to define corresponding gene families, their genomic organization, and expression patterns. This work characterizes eight recently identified high glycine/tyrosine (HGT)-type KAP genes collectively designated Krtap16-n. These genes are shown to be integrated into a larger KAP gene domain on mouse chromosome 16 (MMU16) that is orthologous to a recently described HGT- and high sulfur (HS)-type KAP gene complex on human chromosome 21q22.11. All Krtap16 genes exhibit strong expression in a narrowly defined pattern restricted to the lower and middle cortical region of the hair shaft in both developing and cycling hair. During hair follicle regression (catagen), expression levels decrease until expression is no longer detectable in follicles at resting stage (telogen). Since isolation of the Krtap16 genes was based on their differential expression in transgenic mice overexpressing the Hoxc13 transcriptional regulator in hair, we examined whether bona fide Hoxc13 binding sites associated with these genes might be functionally relevant by performing electrophoretic mobility shift assays (EMSAs). The data provide evidence for sequence-specific interaction between Hoxc13 and Krtap16 genes, thus supporting the concept of a regulatory relationship between Hoxc13 and these KAP genes.

The distinct functional properties of diverse epithelial cell types are largely determined by their cytoskeletal architecture that includes keratins and keratin-associated proteins (KAPs) 1 as essential components. In vertebrates, there exist two major classes of keratins, acidic and basic, that are commonly known as type I and type II keratins, respectively, and which are encoded by more than 50 genes in mouse and man (1)(2)(3). Specific pairs of complementary type I and type II keratins are usually co-expressed and have an intrinsic capacity to form ␣-helical coiled-coil heterodimers; these heterodimer units are known to self-assemble into 10-nm intermediate filaments (IFs) through both end-to-end association and anti-parallel alignment (4). The cytoplasmic IFs build an intracellular fibrous network that extends from the cell membrane to the nucleus, and its properties are greatly influenced by interactions with keratin-associated proteins, also known as KAPs (1,5). The keratin and KAP gene expression profiles change with keratinocyte differentiation status, and the progression of hair follicle differentiation is characterized by the sequential activation of distinct sets of hair-specific keratin and KAP genes (5)(6)(7)(8)(9)(10)(11). Hair follicle development is initiated prenatally through mesenchymal-epithelial interactions that result in the formation of multilayered cylindrical structures (12). Two major outer layers of functionally distinct keratinocytes known as outer root sheath (ORS) and inner root sheath (IRS) surround the hair shaft which is composed of the cuticle, cortex, and medulla. Differentiation of the cells contributing to the formation of these compartments progresses along the proximaldistal axis of the follicle and originates in the lower bulbous portion that harbors largely proliferating and undifferentiated cells. In terminally differentiated cells of the hair shaft, the bulk of the structural molecules include hair-specific IF keratins and KAPs, which combined are estimated to include up to a 100 different proteins (5). In these cells, IFs are embedded into a dense matrix of KAPs that according to their biochemical composition have originally been classified into 3 major groups including high sulfur (HS), ultra-high sulfur (UHS), and high glycine-tyrosine (HGT). Due to the growing complexity among these KAPs, the three main groups have subsequently been divided into structurally distinct subgroups encoded by at least 23 hair KAP gene families (9 -11). The essential role of keratins in the structural organization of hair is underscored by the linkage of severe hair disorders to mutations in human hair keratins as exemplified in monilethrix (13). Furthermore, aberrant expression of keratins and KAPs in transgenic mice has been shown to affect hair structure and growth (14 -20). Combined, these data suggest that structural integrity and tight regulation of both groups of genes is essential for proper hair growth.
Based on differential gene expression analysis in postnatal skin of mice overexpressing Hoxc13, we have previously identified a novel subset of HGT-type KAP genes that we collectively termed Krtap16-n whose expression was uniformly down-regulated in the skin of this GC13 transgenic mouse; we mapped these genes to a distal region of mouse chromosome 16 (MMU16) that is of conserved linkage with human chromosome (HSA) 21q22.11 (19). Here we demonstrate that these genes are integrated into a larger KAP gene domain of about 0.82 Mb in size and that this region is homologous to a recently described domain of HGT-and HS-type KAP genes in humans (10). We examined the Krtap16 gene expression patterns in both developing and cycling hair follicles by in situ hybridization. Furthermore, we show that a HOXC13 consensus binding motif, 5Ј-TT(A/T)ATNPuPu-3Ј, implicated in the regulation of human IF hair keratin genes (21) matches HGT-2, a conserved motif that had previously been speculated to play a role in the transcriptional regulation of certain HGT-type KAP genes (15). Our in vitro DNA binding studies provide evidence for Hoxc13-dependent regulation of Krtap16 genes involving the HGT-2 motif. The Krtap16 gene complex described here might thus serve as a model for examining regulatory interactions of a Hox factor with clustered downstream target genes.

Genomic Map, Protein Sequence Comparison, and Phylogenetic Tree
Analysis-A genomic map of the distal region of MMU16 harboring the Krtap16 genes was derived from the NCBI mouse genome data base (www.ncbi.nlm.nih.gov/genome/guide/mouse). A region of ϳ0.82 Mb stretching roughly from position 90,094 to 90,912 kb of the MMU16 draft sequence was used for establishing this map. All KAP genes and KAP-related ESTs with continuous open reading frames (ORFs) present in this region were included in the map, and the corresponding protein sequences were derived from conceptual translation of the ORFs starting with an ATG start codon positioned in a sequence context that resembles a common consensus motif for eukaryotic translational initiation (22). Alignment of deduced KAP protein sequences was achieved by using the AlignX program included in the Vector NTI software package. Phylogenetic tree analysis was performed using the same software tool.
Expression Analysis-For determining Krtap16 gene expression patterns in skin of 5 day postnatal FVB mice, the animals were euthanized in an atmosphere of CO 2 . Skin from the scapular region was fixed in 4% paraformaldehyde in phosphate-buffered saline at 4°C overnight, then transferred to 30% sucrose solution in phosphate-buffered saline and incubated at 4°C overnight, finally frozen in OCT compound and stored at Ϫ80°C. In situ hybridization was performed with 10-m cryosections essentially as described (23,24). Briefly, plasmids containing Krtap16 cDNAs (19) were linearized with appropriate restriction endonucleases to generate templates for the synthesis of digoxigenin (Roche Applied Science)-labeled antisense and sense (control) probes using SP6 and T7 RNA polymerases. Prior to hybridization, sections were treated with proteinase K (1 g/ml) for 10 min and acetylated with acetic anhydride in 0.1 M triethanolamine. Hybridization with 5 l of probe (1/20 of in vitro transcription reaction product) in 150 l of hybridization solution containing 10 mM Tris, pH 7.5, 600 mM NaCl, 1 mM EDTA, 50% formamide, 10% dextran sulfate, and 200 g/ml yeast tRNA was carried out at 65°C overnight. To reduce background, sections were treated with RNaseA at 37°C for 30 min after the hybridization. Post-hybridization washes were performed in 50% formamide with descending concentrations of SSC starting at 2ϫ SSC until 0.2ϫ SSC at 65°C. Hybridization signals were visualized by using the standard NBT/BCIP detection system (Roche Applied Science).
For expression analysis in cycling hair, synchronized follicle growth in dorsal skin of FVB mice was induced by depilation as described (25). Fresh skin samples at 9 days, 17 days, and 23 days post-depilation were used for the preparation of RNA according to Chirgwin et al. (26). RNA samples were used for the synthesis of complex cDNA probes and employed in reverse Northern hybridizations to Krtap16 cDNAs arrays as described (27,19); parallel skin samples were embedded into an OCT compound for the preparation of frozen sections. In situ hybridization to these sections was performed with 35 S-labeled antisense and sense (control) RNA probes synthesized from linearized plasmids containing Krtap16 cDNAs (see Ref. 19 for a description of all Krtap16 cDNAs used here, except for Krtap16-10; primers used for synthesizing Krtap16-10 cDNA were 5Ј-TCACGGCAACTACTATGGT-3Ј[forward] and 5Ј-AGG-GACTGAAGTAGCCATAA-3Ј [reverse]). A detailed description of the in situ hybridization protocol utilizing 35 S-labeled probes including probe synthesis, probe adjustment, section pretreatment, hybridization, and washing conditions has been published (28). Hybridized sections were exposed to Kodak NTB2 autoradiographic emulsion for 24 h, and developed slides were counterstained with hematoxylin and eosin.
Annealed oligonucleotides were end-labeled with [␥-32 P]ATP and T4 polynucleotide kinase, and purified oligonucleotides (Ϸ10,000 cpm) were incubated with nuclear extracts (Ϸ2 g of protein) from rat kangaroo kidney cells (PtK2 cells) transiently transfected with a human HOXC13 expression vector; nuclear extracts from non-transfected cells were used as a control. Protein/DNA complexes were resolved by electrophoresis in 6% polyacrylamide gels in 0.5ϫ TBE. Binding specificity was ascertained by adding 100-fold excess of unlabeled oligonucleotides (see Fig. 7) in parallel reactions. For supershift assays, 1 l of HOXC13specific antiserum (21) was incubated with nuclear extracts prior to the incubation with labeled oligonucleotides.
For immunohistochemical detection of Hoxc13 in 10-m frozen skin sections, HOXC13-specific antiserum (21) was used at a 1:1000 dilution; bound antibody was visualized by incubating the slides with biotinylated goat anti-guinea pig IgG and the Vector Elite ABC kit was used for substrate reaction and color development.

Genomic and Structural Characterization of Krtap16
Genes-Based on differential gene expression analysis in 5 day skin of normal versus Hoxc13 overexpressing transgenic mice, we have recently identified a subset of new HGT-type KAP genes designated originally Krtap16-1-10, and by using a mouse-hamster radiation hybrid panel we mapped them to the distal portion of MMU16 (Ref. 19; the sequence for Krtap16-10 was submitted separately to the GenBank TM data base under accession number AF477980). The subsequent publication of a draft sequence of MMU16 and the entire mouse genome (29,30) facilitated a closer analysis of the corresponding genomic region (see mouse genome data base, www.ncbi.nlm.nih.gov/ genome/guide/mouse). This analysis revealed redundancy in the original assignment of these Krtap16 cDNAs to separate genes, thus reducing the actual number of Krtap16 genes isolated by us to 8 (see Table I). A map of the region harboring the Krtap16 and neighboring genes is presented in Fig. 1A. Whereas all 8 Krtap16 genes are located on the minus strand within a subregion of about 0.54 Mb of DNA, 6 of them are clustered within less than 100 kb; the seventh (Krtap16-8) and eighth (Krtap16-7) gene are located at a distance of ϳ90 and 450 kb upstream of the 6-member subcluster, respectively. The Krtap16 genes are integrated into a larger domain of 0.82 Mb of genomic DNA harboring predominantly KAP genes (Fig. 1A). Within this domain, the Krtap16 genes are found among 16 annotated KAP genes in addition to 12 ESTs and cDNAs with similarities to known KAP genes. A conceptual translation of the ORFs of these 28 KAP genes and KAP gene-related cDNA sequences revealed considerable heterogeneity with regard to glycine/tyrosine and cysteine content, as well as the presence of conserved amino acid sequence motifs ( Fig. 2 and Table I). The glycine/tyrosine content of the predicted Krtap16 proteins ranges from 40.7 mol% (Krtap16-3) to 65.4 mol% (Krtap16-8) (Table I), thus placing them into the HGT-type class (5) that is known to include several structurally distinct subfamilies in both mouse and human (5,10). Remarkably, there is a more than 4-fold variation in cysteine content among the Krtap16 proteins ranging from 1.76 mol% for Krtap16-10 to 18 mol% for Krtap16-7 (Table I). Interestingly, the two presumptive proteins with the highest cysteine content of 13 mol%, i.e. Krtap16-7 and Krtap16-8, are also those with the highest glycine/tyrosine content of 61.8 and 65.4 mol%, respectively. These differences in cysteine and, perhaps to a lesser degree, in glycine/tyrosine content between individual Krtap16 proteins are associated with distinct amino acid repeats and terminal peptide motifs (Fig. 2), as well as distinct patterns of spatial segregation and genomic clustering of the corresponding genes ( Fig. 1A).
To gain a better understanding of the structural and phylogenetic relationships between Krtap16 and surrounding KAP genes, we performed a phylogenetic tree analysis based on the protein sequences predicted for these genes. This analysis revealed the existence of distinct clusters of phylogenetically related genes, thus permitting a subdivision of the 28 KAP genes listed in Table I into 8 different groups and subgroups ( Fig. 1B). Overall, these groups correlate well with the distribution of specific amino acid motifs and peptide repeats (Fig. 2), as well as the genomic clustering of the corresponding genes (Fig. 1A). The three members of group I all encode proteins with a cysteine content of about 12 mol% that consequently fall into the class of HS-type KAPs. These proteins are structurally highly conserved and share an N-terminal M(A/V)YSCCSGN-FSS motif ( Fig. 2A). The genes of this group are located at the proximal end of the KAP domain ( Fig. 1A) and are followed by two members of group II, Krtap14 (pmg-1; Refs. 31 and 32) and Krtap15 (pmg-2; Ref. 33). We assigned two additional KAP genes including Krtap11-1 (originally known as Hacl-1, Ref. 34) and a Krtap11-related cDNA (GenBank TM accession number: AK017387) that mark the distal end of this domain to group II ( Fig. 1, A and B). A characteristic of all four genes is their overall heterogeneity and lack of extended conserved sequence motifs (Fig. 2B). In addition, the relatively low glycine/tyrosine and cysteine content of their products (Table I) averaging around 18 and 8 mol%, respectively, makes it difficult to assign them either to the HGT or HS class, although Krtap11-1 might be an exception with a cysteine content of close to 12 mol% (Table I).
These heterogeneous pmg-related genes flank larger clusters of HGT-type KAPs that were assigned to groups III, IV, and V. Group III is the largest group including 12 genes that are organized into 3 structurally and spatially distinct sub-clusters, III-A, -B, and -C. In addition to Krtap8-2, group III-A includes 5 of the 8 Krtap16 genes, i.e. Krtap16-1, Ϫ16-3, Ϫ16-4, Ϫ16-5, and Ϫ16-9; except for Krtap16-3, all of these KAPs encode a consensus MS(Y/H)YX(S/G)Y(Y/S)GGLG N terminus and, Krtap16-3 included, a conserved YGFSXFY C terminus (Fig. 2C). Embedded within the group III-A cluster are the two members of group III-C, Krtap16-10, and a Krtap16-10-related gene (GenBank TM accession number: AY026312) that are lo- The first column lists known KAP genes under their originally published name; corresponding GenBank TM accession numbers are listed in the second column. ESTs or cDNAs that have previously not been reported in the literature are only identified by their GenBank TM accession numbers. The third column indicates gene location on the centromeric (ϩ) or telomeric (Ϫ) strand of the chromosome. The remaining columns specify characteristics of the conceptual translation product including peptide size, molecular weight, glycine-tyrosine content and cysteine content.   Table I. Optimal alignment was achieved by grouping the sequences according to the clustering observed upon phylogenetic tree analysis (Fig. 1B)  cated just downstream of the III-A distal-most member, Krtap16-3. The group III-C genes encode proteins that share a distinct MSYY(H/Y)GNYYGG N-terminal domain (Fig. 2E). Members of subgroup III-B form a loose cluster stretched over 100 kb near the middle of this 0.83 Mb domain (Fig. 1A) and contain previously unknown KAP gene sequences. The presumptive proteins encoded by these sequences share a characteristic MCYY(R/G)(G/S)YYGGLG N terminus (Fig. 2D). Interspersed between groups III-A and III-B is a tight cluster of 5 highly conserved Krtap6-1-related KAPs assigned to group IV-A, which includes in addition to Krtap6-1, Krtap16-8, as well as 3 additional KAP genes (GenBank TM accession numbers: D86419, D86421, and AK003924). Conceptual translation of the ORFs of all 5 genes indicates that their protein sequences are nearly identical throughout and start with a distinct MC-GYYGNYYGGRGYG N terminus (Fig. 2F). The two members of subgroup IV-B, Krtap16-7 and Krtap6-2 are located more than 180 kb distal to the III-B subcluster. Their conceptual protein structures show strong similarities between the Nterminal halves and both have extensive -CGYGSGYG-repeats; however, their C termini are unique (Fig. 2G). Finally, two cDNAs (GenBank TM access numbers: AK004025 and D86423) were found to encode structurally heterogeneous HGT-type KAPs; since they showed little similarity to members of the other groups, they were assigned to a separate group V (Figs. 1  and 2H). These two genes are located between groups III-A and IV-A (AK004025), as well as group IV-B and the two distal members of group II (D86423).
We have pointed out earlier that the distal region of MMU16 where we originally mapped the Krtap16 genes is of conserved linkage with HSA 21q22.11 (19), and recently a corresponding human KAP gene domain has been defined in this region (10). A comparison between the organization of the genes and subclusters included in the KAP gene domain on MMU16 and those found in the corresponding human KAP gene domain is schematically depicted in Fig. 3 and discussed below. Overall, the results reveal a great degree of similarity with respect to the order and arrangement of subclusters and the orientation of these domains in mouse and human.
Krtap16 Expression Patterns-The circumstance that all Krtap16 genes described here were isolated as differentially expressed cDNAs in 5 day postnatal skin of our GC13 hair mutant (19) implies that they are transcriptionally active in skin at that stage of development. Since the Krtap16 genes bear structural similarities to previously characterized hairspecific HGT-type KAP genes, we focused on examining expression in hair. To define spatial patterns of Krtap16 expression, we performed in situ hybridization analysis in 5-day postnatal skin that is known to harbor fully differentiated hair follicles (35). Furthermore, to determine whether Krtap16 genes might continue to be active also in cycling hair, we examined expression in adult skin containing depilation-induced anagen phase hair follicles.
Expression analyses in 5 day postnatal skin was performed by using digoxigenin-labeled RNA probes specific for each of the Krtap16 genes, except for Krtap16-10, whose expression was only examined in cycling hair (see below). Hybridization in longitudinal follicle sections yielded similar hair-specific patterns for all genes examined (Fig. 4). In most cases, expression was restricted to a relatively narrow region of the hair shaft with a proximal boundary near the neck of the bulb, a region known to harbor cells undergoing terminal differentiation (36); for Krtap16-7, however, the proximal expression boundary was shifted slightly more distal compared with the remaining genes (Fig. 4F). Hybridization signals were observed in all differentiated hair follicles found in scapular skin samples, and the patterns suggested expression to be restricted to the cortical region of the hair shaft (Fig. 4, A-D and F-I). Further evidence for the cortically restricted expression was obtained by hybridization to follicular cross sections. In all cases, expression was found to be restricted to a circular layer of uniformly shaped cells surrounding the medulla, which itself lacked detectable hybridization signals (Fig. 5).
For five of the six members of group III, we determined expression in skin at different stages of the hair cycle including the progressive growth phase (anagen), the phase of follicle regression (catagen), and the resting phase (telogen). This was done by either reverse Northern (Krtap16-1, Ϫ16-3, Ϫ16-5, Ϫ16-9) or in situ hybridization (Krtap16-10). The results show a drastic decline in expression levels during follicle regression (catagen), and in skin containing telogen follicles expression was no longer detectable (Fig. 6). In addition, we examined expression of the remaining Krtap16 genes in cycling anagen hair follicles by in situ hybridization with radioactive probes; this revealed expression patterns that were very similar to those observed upon hybridization with digoxigenin-labeled probes in 5-day postnatal hair (data not shown). Combined, these findings are consistent with the hair cycle-dependent expression of other Krtap16-related HGT-type KAP genes in mouse skin previously reported (37).
Combined, the expression data show that all Krtap16 genes described here are transcriptionally active in both developing and cycling hair follicles and that their expression is restricted to cortical keratinocytes. The reduction and lack of expression demonstrated for a subset of Krtap16 genes during catagen and telogen, respectively, suggests the existence of hair cycle-dependent control mechanisms for the regulation of transcrip-tional activity. The majority of Krtap16 genes, including Krtap16-1, Ϫ16-3, Ϫ16-4, Ϫ16-5, and Ϫ16-9 have, together with the previously isolated Krtap8-2 gene (GenBank TM accession no. D86422, Ref. 37), been assigned to group III-A as based on our phylogenetic tree alignment (Fig. 1B), and consequently the great similarities in expression (Fig. 4) to the cortical pattern reported for Krtap8-2 (37) are not surprising.
Furthermore, the cortical patterns of group III-A genes seem to be remarkably similar to the spatially restricted patterns of expression of certain members of the KAP19 subfamily of human HGT-type KAP genes in human beard hair, including KAP19. 2, -19.3, and -19.7 (10). Group III-A genes are structurally most closely related to KAP19 genes located in a domain of HGT-type KAP genes on human chromosome 21q22.11 ( Fig. 3;  Ref. 10). However, in contrast to the relative uniformity in Krtap16 expression patterns, some members of the human KAP19 family (i. e. KAP19.1, -19.4, and -19.6) exhibit rather heterogeneous patterns that may either include or be restricted to cuticular cells of beard follicles (10).
DNA Binding Studies-The fact that all Krtap16 genes were found to be down-regulated in the skin of Hoxc13-overexpressing transgenic mice (19) might suggest a regulatory relationship between Hoxc13 and these KAP genes. As a first step toward examining potential direct interactions with Hoxc13, we searched for the presence of presumptive Hoxc13 binding sites in Krtap16 flanking regions. The consensus binding sequence (5Ј-TT(A/T)ATNPuPu-3Ј) for human HOXC13 has previously been defined (21), and given the high degree of overall similarity of 98% between mouse Hoxc13 and human HOXC13 proteins, as well as the structural identity of the DNA binding homeodomains in both proteins, it is to be anticipated that murine Hoxc13 will interact specifically with the same consensus sequence. A systematic search for bona fide Hoxc13 binding sites within a genomic interval of 23 kb harboring a subcluster of four tandemly arranged Krtap16 genes, i.e. Krtap16-9, -16-1, -  1A and 7A), identified 34 perfect matches for the HOXC13 consensus binding sequence (5Ј-TT(A/T)AT-NPuPu-3Ј) in this region alone (Fig. 7A). In most cases one or two presumptive binding sites were closely associated with a TATA box (Fig. 7A). To obtain evidence for interaction of HOXC13 with these presumptive binding sites, we selected the two most proximal sites upstream of Krtap16.5 for EMSAs. The experiments were performed with nuclear extracts from HOXC13-transfected and non-transfected PtK2 cells and 32 Plabeled double-stranded oligonucleotides containing a bona fide HOXC13 binding sequence and a mutated version of this sequence.
The results show band shifts only for reactions performed with nuclear extracts from HOXC13 transformed cells in the presence of oligonucleotides that contain an intact 5Ј-TTAAT-GAG-3Ј sequence matching the HOXC13 consensus binding site 5Ј-TT(A/T)ATNPuPu-3Ј in both cases, thus indicating the formation of sequence-specific HOXC13/DNA complexes ( Fig.  7B and C; compare lanes 1, 2, and 3). In the presence of excess cold competitor oligonucleotide containing either the authentic sequence or an alteration from 5Ј-TTAATGAG-3Ј to 5Ј-TGCCG-GAG-3Ј, the shifted bands were either abolished or maintained, respectively (Fig. 7, B and C; compare lanes 4 and 5). Supershift assays performed by adding HOXC13-specific antiserum (21)  typical pattern of Hoxc13 expression (Fig. 7D) previously reported in both mouse and human hair follicles at RNA (19,38), as well as both RNA and protein levels (21), respectively. Furthermore, this apparent Hoxc13 expression is shifted toward the keratogenous zone of the cortex in GC13 mutant hair follicles (Fig. 7E). DISCUSSION Our data show that the Krtap16 genes are embedded within a larger KAP gene domain that occupies ϳ0.82 Mb of DNA and includes 28 KAP genes (Fig. 1); 20 of these genes, including the Krtap16 genes, have previously been reported in the literature, on occasion under alternative names attributed by different authors. The remaining 8 KAP genes were listed in the mouse genome data base as ESTs and cDNAs with KAP gene similarities (www.ncbi.nlm.nih.gov/genome/guide/mouse). In addition, 3 ESTs for unknown genes without similarities to KAP genes have been mapped to this region (Fig. 1A). Outside of this 0.82 Mb region, no further KAP genes or KAP-related ESTs were identified, thus suggesting that the map presented in Fig. 1 defines the limits of a murine KAP gene complex in the distal portion of MMU16.
Like all KAP genes reported to date (see Refs. 5,[9][10][11], the Krtap16 genes have a simple structure consisting of a single exon with a short ORF, in this case ranging from 165 to 423 bp. Although all Krtap16 genes may be classified as HGT-type KAPs as based on the high glycine/tyrosine content of their presumptive products ranging from 40.7 to 65.4%, they do  -1, -16-3, -16-4, -16-5, -16-7, -16-8,  and -16-9, (A, B and D-H) and a sense (SE) control probe specific for Krtap16-3 (C) were hybridized to 10-m frozen sections of scapular skin derived from 5-day postnatal FVB mice. Hybridization signals were visualized by the NBT/BCIP color reaction that resulted in a dark precipitate; no signal was obtained with the control probe. A schematic representation of the sections depicting the individual follicle layers is shown in the scheme at the lower right (I). Expression of all genes is restricted to cortex (Ctx), while no hybridization signal was detected in the medulla (M) and surrounding cuticle (Cu). Space bar, 20 m.
FIG. 6. Krtap16 gene expression during the hair cycle. A, reverse Northern hybridization to arrayed cDNAs specific for Krtap16-1, - 16-3, -16-5, and -16-9 as indicated with complex cDNA probes derived from adult mice whose hair growth was at different stages of the hair cycle. The hair cycle had been synchronized by depilation (25) and skin samples were taken at days 9, 17, and 21 post-depilation (p.d.) that corresponded to anagen, catagen, and telogen phases, respectively. Please note the dramatic reduction in signal and loss of signal for all four genes in catagen and telogen, respectively. S, total mouse genomic DNA used as a standard. B/BЈ, D/DЈ, E/EЈ, in situ hybridization with 35 S-labeled Krtap16-10-specific antisense probe to frozen sections of adult skin that was depilated for synchronized entry into anagen, catagen, and telogen phase of the hair cycle as described above; C/CЈ, hybridization with Krtap16-10-specific sense control probe. Hybridized sections were counterstained with hematoxylin/ eosin as shown in brightfield (B-E), and silver grains corresponding to hybridization signal were visualized by light scattering (bright grains) using darkfield microscopy (BЈ Ϫ EЈ). Please note Krtap16-10 expression in the lower cortex (Ctx) of the growing hair shaft during anagen (B/BЈ) and the reduced expression in catagen hair (D/DЈ); no specific hybridization signal was detectable in the shaft of telogen hair (E/EЈ), and no signal was detectable in anagen follicles upon hybridization with the sense probe (C/CЈ) used as control. DP, dermal papilla; HS, hair shaft; M, medulla. Space bar, 50 m. belong to diverse groups as revealed by phylogenetic tree analysis (Fig. 1B), and they are found scattered throughout much of the 0.82 Mb KAP gene complex. Interestingly, putative proteins encoded by Krtap16-7 and Ϫ16-8, the two Krtap16 proteins with the highest glycine/tyrosine content of 61.8 and 65.4%, respectively, have also the highest cysteine content of 18 and 13% (Table I). The same applies also to other members of the Krtap16-8 and Ϫ16-7 subgroups IV-A and -B, respectively, as well as members of subgroup III-B (Table I, Fig. 1). This circumstance might illustrate that the traditional dichotomy of KAP classification into HS (and UHS) and HGT-type KAPs is insufficient for adequate categorization, as it has been recognized previously (5).
By and large, the clustering of the described KAP genes into distinct groups and subgroups as demonstrated by phylogenetic tree alignment corresponds well with their spatial clustering in the genome (Fig. 1, A and B). This overall genomic organization is conserved between mouse and human where a corresponding KAP gene domain has recently been defined on HSA 21q22.11 (10). In both cases, the domains start at the centromeric end with a group of Krtap13-related (HS-type) genes that is followed by Pmg-related KAPs, which precede a subcluster of HGT-type KAP genes related to murine Krtap8-2, i.e. groups III-A and KAP19 in mouse and human, respectively (Fig. 3). This is followed by groups IV-A and KAP6 in mouse and human, respectively, which in both cases include genes with similarities to Krtap6-1. Subsequent to this, the density of KAP genes decreases and includes group III-B and IV-B genes in mouse and their respective human KAP20 and KAP21 equivalents. Finally, the complexes terminate distally with heterogeneous Pmg-like genes in both cases. This remarkable conservation in gene order on a relatively small scale reflects on a much larger scale the conserved synteny between distal MMU16 and HSA21 corresponding to segments of 22.37 and 28.42 Mb of the mouse and human chromosomes, respectively (29). The order of 111 genes in this MMU16 segment has been reported to follow exactly the order of corresponding human genes in this region (29), and consequently, the overall conservation of the murine KAP gene domain is of no surprise. Although the mouse and human genomes have been extensively shuffled by chromosomal rearrangements since their phylogenetic separation from a common ancestor ϳ65-75 million years ago, the gene order in syntenic segments has been found to be largely intact (30). However, this does not exclude isolated local rearrangements and gene family expansions that are likely to have occurred during the evolution of the murine KAP gene domain on chromosome 16 and its human counterpart and resulted in paralogous genes.
The expression patterns of the Krtap16 genes in the keratogenous zone of the hair shaft most closely resemble the patterns of two previously isolated members of their domain including Krtap8-2 (originally isolated under the name HGTptype Ia, see Ref. 37) and Krtap11-1 (originally known as Hacl-1, see Ref. 34). Although these two genes are structurally heterogeneous (Fig. 2) and belong to different groups according to our phylogenetic alignment analysis (Fig. 1B), their reported expression patterns are very similar. This relative uniformity contrasts with a considerable diversity in expression patterns among members of the human equivalent of the MMU16 KAP gene domain. In that case, several genes exhibited expression in both cortical and cuticular cells with remarkable variation in proximal-distal expression domains, which for certain Pmg-like KAP genes, including hKAP13.1 and -15.1, initiated in the matrix of beard follicles (10). Furthermore, several of the human KAP genes showed striking cylindrical asymmetries in expression, a feature that previously had been observed also for Krtap6-1 expression in hair follicles of sheep and rabbit (39). These asymmetries are likely to reflect differential expression  (21). EMSAs were performed with two separate double-stranded oligonucleotides, K16-5 (1) (B) and K16-5 (2) (C) that contained the presumptive binding sites, as well as two corresponding control oligonucleotides, K16-5 (1) M (B) and K16-5 (2) M (C) that contained mutated versions of these sites (changes in HOXC13 consensus binding sequence from 5Ј-TTAATGAG-3Ј 3 5Ј-TGCCGGAG-3Ј). For a description of gel lanes, see key to the right. Specific HOXC13-DNA complexes are only seen in lanes 2 and 5 for either oligonucleotide as indicated by the arrows at the left. Lane 1, reaction with nuclear extracts from non-transfected PtK2 cells; lane 6, supershift assay with HOXC13-specific antibody (21) as marked by the asterisk. D and E, immunohistochemical detection of Hoxc13 protein expression pattern (brown staining) with anti-HOXC13 antiserum in normal (D) and GC13 mutant (E) hair follicle of 5-day postnatal skin counterstained with hematoxylin. Follicles depicted in D and E have been aligned and divided into 5 horizontal zones of 75 m comparable to the zones seen in Fig. 4. Hoxc13 expression originates in the matrix (Mtx) of the bulb and reaches into precortical/cortical regions above the apex of the dermal papilla (P). In GC13 hair (E), this expression reaches into the 4th zone (red arrow); in the same zone no cortical staining is visible in the normal follicle (D) as marked by the red arrow, although specific staining is seen in the medulla (M). Mtx, matrix. Space bar, 50 m.
in local subpopulations of cortical keratinocytes and might be a contributing factor in determining the texture and crimp of hair (39). Accordingly, the greater diversity in human KAP gene expression patterns compared with their counterparts on MMU16 is likely part of the molecular signature determining textural differences between human beard and murine coat hair.
The down-regulation of the Krtap16 genes in Hoxc13 overexpressing (GC13) mice led us to speculate about a regulatory relationship between Hoxc13 and these KAP genes. In support of this idea we found multiple putative Hoxc13 binding sites associated with members of the MMU16 KAP gene domain (Fig. 7A); EMSAs carried out for two of these sites located 5Ј of Krtap16-5 indicated sequence-specific interaction with human HOXC13, whose DNA binding homeodomain is identical to the murine Hoxc13 homeodomain. Furthermore, both Hoxc13 and the Krtap16 genes are expressed in the same lineage of cortical keratinocytes, although at different stages of differentiation. While the Krtap16 genes are expressed in the keratogenous zone of the cortex, Hoxc13 is expressed in the matrix and the precortical region, which means that there is little overlap between the Krtap16 and Hoxc13 expression domains in normal hair follicles. Taking this circumstance into account, one may consider at least two alternative mechanisms for explaining the down-regulation of KAP genes in skin of GC13 mice. The first is based on indirect regulation mediated by another transcription factor, whose expression might directly be controlled by Hoxc13. A good candidate for this is Hoxc12, whose expression domain in the cortex and precortex of differentiated hair follicles conspicuously overlaps with both the Hoxc13 domain (38) and the Krtap16 expression domains (Fig. 4). During axial patterning, Hox genes frequently restrict the activity of their next downstream neighbor within the same cluster that usually occupies a more anterior expression domain along the longitudinal embryonic axis, a phenomenon known as posterior prevalence (see Refs. 40 and 41). A similar principle might apply to the regulation of Hox gene activities along the longitudinal axis of the differentiating hair follicle (42). In our case, the concept of a regulatory relationship between Hoxc13 and Hoxc12 is supported by the presence of multiple bona fide Hoxc13 binding sites upstream of Hoxc12, (as well as Hoxc13; data not shown), which itself might positively regulate Krtap16 expression. In this context it is worth mentioning that Hoxc13 might control also its own expression via negative autoregulation as it has been postulated based on the reduced expression of a Hoxc13-lacZ reporter gene construct in Hoxc13 overexpressing mice (19). The presence of putative Hoxc13 binding sites found in genomic regions flanking Hoxc13 is consistent with this idea.
The second mechanism for explaining the down-regulation of Krtap16 genes in hair follicles overexpressing Hoxc13 is based on an expansion of the Hoxc13 expression domain toward the keratogenous zone, thus resulting in Hoxc13 expression in cells that normally express Krtap16 genes but not Hoxc13. Under the premise that Hoxc13 normally acts as a negative regulator of Krtap16 genes to prevent their premature activity, this would result in their reduced expression. Our comparative immunohistochemical analysis of Hoxc13 expression in normal and GC13 hair follicles indeed suggests a distal expansion of the Hoxc13 domain into a zone where Hoxc13 and Krtap16 gene expression overlap (compare Figs. 7 and 4). In summary, the phylogenetically conserved KAP gene complex on mouse chro-mosome 16 might provide a useful paradigm for studying mechanisms that regulate the coordinated expression of clustered hair-specific genes in a cyclical manner.