Advertisement

HOXC13 Is Involved in the Regulation of Human Hair Keratin Gene Expression*

      At present, HOXC13 is the only member of the HOX multigene family that produces a fragile hair phenotype when mutated or overexpressed in mice. To determine whether hair keratin genes are targets for this transcription factor, we analyzed the HOXC13 responsiveness of human hair keratin genes, whose expression matched that of nuclear HOXC13, immunologically revealed in cells of the lower hair-forming compartment of the human anagen hair follicle. We show that HOXC13, but not a homeobox-deleted HOXC13, strongly activated the promoters of the genes, with the respective proximal promoter regions being sufficient for optimal activation. The hair keratin promoters contained numerous putative Hox binding core motifs TAAT, TTAT, and TTAC. Electrophoretic mobility shift assays revealed that HOXC13 bound exclusively to distinct TAAT and TTAT core motifs that were clearly concentrated in the proximal promoter regions. A comparison of the sequences flanking HOXC13 binding and nonbinding core motifs, respectively, allowed the deduction of an extended 8-bp HOXC13 consensus binding sequence TT(A/T)ATNPuPu. Thus, the DNA binding conditions for HOXC13 were distinct from those of other members of the paralogous group 13, i.e. murine Hoxb13 and HOXd13, for which previous investigations yielded the consensus binding sequence TTTA(T/C)NPuPu. Collectively, our data speak for a direct involvement of HOXC13 in the control of hair keratin expression during early trichocyte differentiation.
      ISH
      in situ hybridization
      IIF
      indirect immunofluorescence
      IRS
      inner root sheath
      EMSA
      electrophoretic mobility shift assay
      DAPI
      4′,6-diamidino-2-phenylindole
      Hox genes encode evolutionarily conserved transcription factors that are important gene regulators involved in cell fate determination during embryonic development. In mammals the 39 Hox genes are organized into four separate chromosomal clusters, Hoxathrough Hoxd. Based upon sequence homology and location within a cluster, these genes have been divided into 13 paralogous groups (for reviews see Refs.
      • Gehring W.J.
      • Affolter M.
      • Bürglin T.
      ,
      • Maconochie M.
      • Nonchev S.
      • Morrison A.
      • Krumlauf R.
      ). Due to sequence homologies within the conserved homeobox, Hox genes are related to DrosophilaHOM-C genes with the paralogs 9–13 all being related to the Abd-B gene (
      • Izpisua-Belmonte J.
      • Falkenstein H.
      • Dollé P.
      • Renucci A.
      • Duboule D.
      ). During embryonic development, paralogous genes located at the 3′-end of each cluster are activated first, whereas the more 5′-genes are transcribed progressively later. In addition, there is spatial colinearity of position in a cluster such that members of successive paralogous groups have increasingly posterior-anterior limits of expression (
      • Peterson R.L.
      • Papenbrock T.
      • Davda M.M.
      • Awgulewitsch A.
      ,
      • Capecchi M.R.
      ). Thus, recent studies on the expression of Hoxc13 during mouse embryogenesis have shown that this 5′-most gene of the Hoxc gene cluster is first expressed at E10.5 in the tail bud (
      • Peterson R.L.
      • Papenbrock T.
      • Davda M.M.
      • Awgulewitsch A.
      ,
      • Godwin A.R.
      • Capecchi M.R.
      ), followed at E12.5 in the epithelia of the wrist and ankle regions of the limbs in which its expression becomes localized to the developing foot pads and nails at E13.5 (
      • Godwin A.R.
      • Capecchi M.R.
      ). However, the studies also revealed that at later embryonic stages, Hoxc13 expression obviously deviated from the code of colinearity, as it also occurred in both vibrissae and all body hair follicles, in the filiform papillae of tongue epithelium and in footpad epidermis (
      • Godwin A.R.
      • Capecchi M.R.
      ). At postnatal day 7, Hoxc13 transcripts in growing anagen hair follicles were seen mainly in the matrix of the hair bulb and the precortical region of the hair shaft and could also be demonstrated at the base of the posterior unit of the filiform tongue papilla (
      • Godwin A.R.
      • Capecchi M.R.
      ). Correspondingly, besides defects in caudal tail vertebrae, Hoxc13-null mice showed malformation of nails and filiform tongue papillae, and, notwithstanding morphologically normal looking hair follicles, totally lacked vibrissae and pelage hairs due to the premature fracture of hair shafts at the surface of the skin (Ref.
      • Godwin A.R.
      • Capecchi M.R.
      and for reviews see Refs.
      • Duboule D.
      ,
      • Godwin A.R.
      • Capecchi M.R.
      ). Since in the mouse, most of these Hoxc13 expressing anatomical regions are also known sites of hair keratin synthesis (9, 10 and references therein), it has been hypothesized that Hoxc13 might possess special functions in hair and filiform papillae development by being involved in the control of hair keratin gene expression (
      • Godwin A.R.
      • Capecchi M.R.
      ,
      • Duboule D.
      ,
      • Godwin A.R.
      • Capecchi M.R.
      ). In line with this assumption, recent investigations in Hoxc13-overexpressing mice have shown that a variety of hair follicle-specific genes are regulated by this transcription factor (
      • Tkatchenko A.V.
      • Visconti R.P.
      • Shang L.
      • Papenbrock T.
      • Pruett N.D.
      • Ito T.
      • Ogawa M.
      • Awgulewitsch A.
      ).
      In the past years, our laboratory has elucidated the organization of the human type I and type II hair keratin gene loci and characterized the sequential expression of their members during trichocyte differentiation (
      • Rogers M.A.
      • Winter H.
      • Wolf C.
      • Heck M.
      • Schweizer J.
      ,
      • Langbein L.
      • Rogers M.A.
      • Winter H.
      • Praetzel S.
      • Beckhaus U.
      • Rackwitz H.R.
      • Schweizer J.
      ,
      • Rogers M.A.
      • Winter H.
      • Langbein L.
      • Wolf C.
      • Schweizer J.
      ,
      • Langbein L.
      • Rogers M.A.
      • Winter H.
      • Praetzel S.
      • Schweizer J.
      ) Based on this knowledge, we selected three human type I hair keratin genes hHa5, hHa2, andhHa7 as well as the transcribed pseudogeneϕhHaA, whose mRNA expression patterns in the lower hair forming compartment of the human hair follicle (
      • Langbein L.
      • Rogers M.A.
      • Winter H.
      • Praetzel S.
      • Beckhaus U.
      • Rackwitz H.R.
      • Schweizer J.
      ,
      • Winter H.
      • Langbein L.
      • Krawczak M.
      • Cooper D.N.
      • Jave-Suarez L.
      • Rogers M.A.
      • Praetzel S.
      • Heidt P.J.
      • Schweizer J.
      ) corresponded to that described for Hoxc13 in mouse hair follicles (
      • Godwin A.R.
      • Capecchi M.R.
      ), and investigated whether they are target genes for HOXC13. In the present study, we provide strong evidence that these hair keratin genes are transcriptionally up-regulated by HOXC13 primarily via binding of the transcription factor to distinct core recognition motifs, TAAT and TTAT, in the respective proximal promoters.

      MATERIALS AND METHODS

       Hair Keratin Reporter Plasmids and Deletion Constructs

      Promoter and promoter deletion constructs of type I hair keratin genes hHa5, hHa2, ϕhHaA, andhHa7 (GenBankTM accession numbers Y16791,X90761, Y16795, and Y16793) were generated by PCR using the previously described PAC3 clone (
      • Rogers M.A.
      • Winter H.
      • Wolf C.
      • Heck M.
      • Schweizer J.
      ) as template DNA as well as forward primers containing an EcoRI-site and reverse primers containing aXhoI-site. After EcoRI/XhoI digestion of the PCR products, the gel-purified fragments were cloned into theEcoRI/XhoI-digested β-galactosidase reporter vector pNassβ (CLONTECH, Heidelberg, Germany).
      Primer pairs used for the generation of hair keratin promoter fragments were as follows. 1) 0.9 kb, 0.4 kb, and 0.3 kbhHa5 promoter fragments; forward primers: 5′-atccaaagtagaattcattgatagggctgtgtgaaaaga-3′ (0.9a5); 5-acaaaatctagaattcgccaataaaagaagtcaaagttg-3′ (0.4a5); 5′-tttgaagaacgaattcgtgtcctgaagctctctcatggtga-3′ (0.3a5); common reverse primer: 5′-gtaagataagctcgagcttgagagacccagaagagaag-3′. 2) 0.9 kb, 0.6kb, and 0.3kb hHa2 promoter fragments; forward primers: 5′-acaaattctagaattcgcatgcggctgaccctttgaaga-3′ (0.9a2); 5′-acaaaatctagaattcagttcattctctccccactgata-3′ (0.6a2); 5′-acaaaatctagaattcgtccagagggggcaagacagag-3′ (0.3a2); common reverse primer: 5′-gtaagataagctcgagcctttctcctcagccacagctacct-3′. 3) 0.8kb, 0.6kb, 0.4kb, and 0.2kb ϕhHaA promoter fragments; forward primers: 5′-agatttaacagaattcgaaagtgggctgcttgaagaagacc-3′ (0.8aA); 5′-atagatgtgaattcgttccgcactgggcactctacaaat-3′ (0.6aA); 5′-cagtgagctgaattccacctcgaaatctttgcagctgtct-3′ (0.4aA); 5′-acaaattctagaattcgcatgcggctgaccctttgaaga-3′ (0.2aA); common reverse primer: 5′-tgcttgacactcgagtcttccttctcctgtagacctt-3′. 4) 0.2 kbhHa7 promoter fragment; forward primer: 5′-gagccaagaggaattcagatttgtcaacattgctttaat-3′; reverse primer: 5′-gctaaggctgctcgagtgcttcagatcagctgggaaggc-3′. The cloned PCR fragments and the promoter/vector junctions were verified by DNA sequencing.

       HOXC13 Expression Vectors

      RNA isolated from freshly plucked human hair follicles was reverse transcribed, and the resulting cDNA was used as template for PCR amplification of full-length human HOXC13 cDNA using the forward primer 5′-gcggccgctagctcgctgcctctggc-3′ and the reverse primer 5′-ggggaaagcagcggccgcgtggtcaggtggagtggagatg-3′. ΔHOXC13 cDNA was obtained by the same procedure using, however, the oligonucleotide 5′-tcagcgcttctctttggtgatgaac-3′ as reverse primer, which introduces a stop codon in the homeobox domain. Both cDNAs were cloned into theNotI site of the cytomegalovirus promoter-containing expression vector pcDNA3 (Invitrogen); their integrity was confirmed by DNA sequencing. The resulting expression vectors were designated pHOXC13 and pΔHOXC13, respectively.

       Cell Culture and Transient Transfections

      At present, both primary human or mouse trichocytes and trichocyte cell lines are not available. For transient transfection essays we therefore chose the epithelial rat kangaroo kidney cell line PtK2. Based on the Hoxc13 expression restrictions (
      • Godwin A.R.
      • Capecchi M.R.
      ,
      • Tkatchenko A.V.
      • Visconti R.P.
      • Shang L.
      • Papenbrock T.
      • Pruett N.D.
      • Ito T.
      • Ogawa M.
      • Awgulewitsch A.
      ) PtK2 cells should not express Hoxc13. The cells were grown at 37 °C in Dulbecco's modified Eagle's medium (pH 7.2), supplemented with l-glutamine, 10% bovine calf serum, and penicillin-streptomycin. Consistently, cells were plated at a density of 3 × 105 cells/35-mm dishes. One day after plating, cells were transfected with 0.5 μg of the indicated reporter plasmid and 0.5 μg of the respective expression vector or 0.5 μg pcDNA3, to bring the total amount of DNA to 1.0 μg. Each DNA mixture contained 2 μl of FuGENETM6 transfection reagent (Roche Molecular Biochemicals). Two days after transfection, cells were harvested, washed in phosphate-buffered saline, and incubated for 10 min in 200 μl of lysis buffer (60 mm Na2HPO4, 40 mmNaH2PO4, 10 mm KCl, 1 mm MgSO4, 2.5 mm EDTA, pH 8.3, 0.25% Nonidet P-40). Protein concentrations were determined using 10 μl of lysate. Twenty μg of each protein sample was incubated for 5 min in 200 μl of buffer A (60 mmNa2HPO4, 40 mmNaH2PO4, 10 mm KCl, 1 mm MgSO4, 2.5 mm EDTA, pH 8.3, 50 mm 2-mercaptoethanol) and supplemented with 50 μl of substrate solution (8 mm chlorophenol red-galactopyranoside (Roche Molecular Biochemicals) in buffer A). β-Galactosidase activity was determined at 570 nm. Transfections were performed in triplicate in at least three independent experiments.

       Preparation of Nuclear Extracts

      PtK2 cells were transiently transfected with 6 μg of pHOXC13 and pΔHOXC13 expression plasmid. After 48 h in Dulbecco's modified Eagle's medium, cells were washed twice in cold phosphate-buffered saline, collected by centrifugation, and resuspended in 1 ml of a hypotonic buffer (10 mm Hepes, pH 7.9, 1.5 mm MgCl2, 10 mm KCl, 0.5 mm dithiothreitol, and 1 tablet of the protein inhibitors mix (Complete mini EDTA-free; Roche Molecular Biochemicals) per 10 ml of buffer). After 15 min on ice, 100 μl of 10% Nonidet P-40 solution was added, and nuclei were precipitated by centrifugation. The nuclei were then resuspended in 200 μl of high salt buffer (20 mm Hepes, pH 7.9, 25% glycerol, 1.5 mm MgCl2, 420 mm KCl, 0.5 mm dithiothreitol, and 1 tablet Complete mini EDTA-free per 10 ml of buffer) and incubated on ice for 40 min. Extracts, cleared by centrifugation for 5 min at 14,000 rpm, were subjected to protein quantification and stored in aliquots at −80 °C.

       Electrophoretic Mobility Shift and Supershift Assays

      DNA fragments containing putative HOX binding sites of the hHa5,hHa2, ψhHaA, and hHa7 promoters were used as probes in DNA/protein binding assays. Oligonucleotides were synthesized as overhanging complementary strands, annealed, and end-labeled with [γ-32P]ATP and T4 polynucleotide kinase. After electrophoresis using 15% preparative polyacrylamide gels, specific bands of annealed oligonucleotides were excised and eluted by shaking in 300 μl of TEN buffer (10 mmTris-HCl, pH 7.5, 100 mm NaCl, and 1 mm EDTA) at 4 °C overnight. For 20 μl of binding assays, nuclear extracts (2 μg of protein) were pre-incubated at 4 °C for 5 min in binding buffer (100 mm KCl, 2 mm MgCl2, 4 mm spermidine, 0.1 mm EDTA, 0.25 mmdithiothreitol, 10 mm Tris-HCl, 10% glycerol, 100 μg/ml bovine serum albumin). Binding reaction was initiated by the addition of 2 μl of 32P-labeled probe (∼10,000 cpm). After 30 min of incubation at 4 °C, DNA/protein complexes were resolved in 6% polyacrylamide gels in low ionic strength buffer (0.5× TBE) at 200 V for 3 h. Gels were dried and autoradiographed. To ascertain binding specificity, a 100-fold excess of unlabeled specific or unspecific competitor oligonucleotides were incubated with the protein extract prior to the addition of the labeled oligonucleotide. For supershift assays, 1 μl of a specific HOXC13 antiserum (see “In Situ Hybridization and Indirect Immunofluorescence”) was preincubated with nuclear extracts in binding buffer for 30 min on ice prior to the addition of the labeled oligonucleotides to allow the formation of protein-antibody complexes. For control experiments, 1 μl of an antibody against the human androgen receptor, kindly provided by Frank Claessens, Leuven, Belgium, was used.

       Isolation of RNA and Reverse Transcription-PCR

      Total RNA was extracted from either the bulbs of freshly plucked human hairs or from ∼20 cryostat sections (10 μm each) of either human dorsal tongue or human footsole, obtained as surgical material, using the RNeasy System (Qiagen, Hilden, Germany). Reverse transcription was performed with oligo(dT)-15-primer and Superscript II Reverse Transcriptase (Invitrogen). The GC-rich PCR System (Roche Molecular Biochemicals) was used for amplification of cDNA fragments. The cycle program for PCR was: 95 °C for 3 min, followed by 35 cycles of 95 °C, 30 s; 58 °C, 30 s; 68 °C, 2 min; and a final extension for 7 min at 68 °C. Amplification primers used for HOXC13 were 5′-cgctgcctctggcaagtggagt-3′ (forward primer), 5′-tcggttatggtacaaagcggagac-3′ (reverse primer). Reverse transcription-PCR products were resolved in ethidium bromide-containing 1% agarose gels and visualized under UV light.

       In Situ Hybridization and Indirect Immunofluorescence

      ISH1was carried out on cryostat sections of human scalp (kindly provided by Dr. Bernard Cribier, Strasbourg, France) as previously described in detail (
      • Langbein L.
      • Rogers M.A.
      • Winter H.
      • Praetzel S.
      • Beckhaus U.
      • Rackwitz H.R.
      • Schweizer J.
      ,
      • Langbein L.
      • Rogers M.A.
      • Winter H.
      • Praetzel S.
      • Schweizer J.
      ). HOXC13 transcripts were detected using a specific PCR fragment that encompassed ∼300 bp of the 3′-untranslated regions of the human HOXC13 mRNA (
      • Stanchina E.
      • Gabelline D.
      • Norio P.
      • Giacca M.
      • Peverali F.A.
      • Riva S.
      • Falaschi A.
      • Biamonti G.
      ). The fragment was cloned into pCR2.1 vector (Invitrogen). Antisense RNA was generated by in vitrotranscription using 35Sr-CTP. The preparation of the antisense RNA probe used for the detection of the mRNA of hair keratin hHa5 has been described previously (
      • Langbein L.
      • Rogers M.A.
      • Winter H.
      • Praetzel S.
      • Beckhaus U.
      • Rackwitz H.R.
      • Schweizer J.
      ). For the recording of the ISH signals by reflection microscopy, a confocal laser scanning microscope (LSM 510 UV, Carl Zeiss, Oberkochen, Germany) was used. The instrument allows simultaneous visualization of ISH in epi-illumination for the detection of reflection signals and transmitted light in bright field for hematoxilin staining. The two signal channels were combined by an overlay in pseudocolor (transmission image in green, electronically changed into black/white using the LSib software (Carl Zeiss); reflection image, i.e. ISH signals in red).
      IIF on cryostat sections of plucked beard hair follicles and human scalp sections was carried out as described previously (
      • Langbein L.
      • Rogers M.A.
      • Winter H.
      • Praetzel S.
      • Beckhaus U.
      • Rackwitz H.R.
      • Schweizer J.
      ,
      • Langbein L.
      • Rogers M.A.
      • Winter H.
      • Praetzel S.
      • Schweizer J.
      ). The HOXC13 antiserum was generated using the synthetic oligopeptide PEPSGALPDGDDLS-C as antigen. This oligopeptide, was derived from the published human HOXC13 sequence (position 154–166 in Fig. 1C of Ref.
      • Stanchina E.
      • Gabelline D.
      • Norio P.
      • Giacca M.
      • Peverali F.A.
      • Riva S.
      • Falaschi A.
      • Biamonti G.
      ) and is unique to this member of the paralogous group 13. Moreover, comparison of the peptide sequence with the SWISSPROT Data base did not reveal similarities with any other homeodomain protein. Prior to immunization, a cysteine residue was added to the carboxyl-terminal end of the peptide for coupling to Keyhole limpet protein (Peptide Specialty Laboratory, Heidelberg, Germany). After the third booster injection, antiserum was obtained and used at a dilution of 1:700. For immunofluorescence, Cy-coupled goat anti-guinea pig IgGs (Dianova, Hamburg, Germany) were used at a dilution of 1:50. IIF results were documented with a photomicroscope (Axiophot 2, Carl Zeiss).

      RESULTS

       HOXC13 Is Expressed in the Human Anagen Hair Follicle

      To demonstrate HOXC13 expression in the adult human anagen hair follicle, we first performed reverse transcription-PCR with follicular RNA using specific primer pairs for the amplification of HOXC13 cDNA. As shown in Fig. 1A, lane a, HOXC13 mRNA was present in anagen hair follicles. Moreover, in line with previous expression data in mice epithelia (
      • Godwin A.R.
      • Capecchi M.R.
      ), HOXC13 mRNA could also be detected in both human dorsal tongue epithelium (Fig. 1A, lane b) and in plantar epidermis (Fig.1A, lane c), for which we recently gained evidence for a spatially ordered expression of a restricted number of hair keratins.
      H. Winter, L. Langbein, M. A. Rogers, and J. Schweizer, unpublished results.
      Figure thumbnail gr1
      Figure 1Demonstration of HOXC13 mRNA expression in human hair follicles and various tissues. A, PCR-based amplification of the complete HOXC13 cDNA (1098 bp) of human anagen hair follicles (lane a), human dorsal tongue (lane b), and human footsole epidermis (lane c). M, molecular mass markers in kb.B, in situ hybridization on longitudinal scalp sections with a specific HOXC13 cRNA probe. Transcripts are present in matrix and early to mid-cortex cells (a, a′). mRNA expression begins in cells lining the dermal papilla (green arrowheadsin a, a′) and is also seen in the lower hair cuticle (blue arrowheads in a′′). A cRNA probe specific for hair keratin hHa5 detects the respective transcripts in the HOXC13-expressing area except for cells lining the dermal papilla (white arrowheads inb), some of which were, however, hHa5-positive (green arrowheads in b). Red arrows demarcate regions of RNA synthesis. co, cortex; cu, cuticle; dp, dermal papilla; gp, germinative cell compartment. Bars = 150 μm.
      ISH of human scalp sections revealed, that HOXC13 transcripts occurred in follicular trichocytes. HOXC13 mRNA expression began in cells that were directly apposed to the dermal papilla, extended into the matrix, the cuticle and the lower cortex region, and declined in the mid-cortex region (Fig. 1B,panels a, a′, a“). The observed HOXC13 mRNA pattern was nearly identical with that obtained for the mRNAs of the cuticular keratin hHa2 (
      • Langbein L.
      • Rogers M.A.
      • Winter H.
      • Praetzel S.
      • Beckhaus U.
      • Rackwitz H.R.
      • Schweizer J.
      ) and the matrix and cuticle hair keratin hHa5 (Ref.
      • Langbein L.
      • Rogers M.A.
      • Winter H.
      • Praetzel S.
      • Beckhaus U.
      • Rackwitz H.R.
      • Schweizer J.
      , and Fig. 1B, panel b), except that the onset of hHa5 mRNA expression occurred essentially above the HOXC13-expressing cell row lining the dermal papilla in which only few cells were labeled (green arrowheads in Fig. 1B, panel b).
      IIF studies using an antiserum against a specific HOXC13 oligopeptide on longitudinal sections of plucked beard hairs confirmed the presence of HOXC13 protein in all trichocytes of the entire matrix, in the lower to mid-cortex region as well as in the lower medulla and hair cuticle (Fig. 2a). As expected for active transcription factors, HOXC13 was restricted to the nuclei of the respective cells. In contrast to the ISH data, HOXC13 could also be detected in the IRS cuticle at a height at which HOXC13 expression in the hair cuticle gradually ceased (Fig. 2, a,c). In accordance with the Hoxc13 expression profile in mouse hair follicles (
      • Godwin A.R.
      • Capecchi M.R.
      ), nuclear HOXC13 expression was also demonstrable in cells of the companion layer (Fig. 2, a,b). The expression pattern of HOXC13 in the human anagen hair follicle is illustrated schematically at the right hand side of Fig. 2.
      Figure thumbnail gr2
      Figure 2Localization of HOXC13 protein in human hair follicles. a, IIF on cryostat sections of plucked human beard hairs with a HOXC13 antiserum. Labeled nuclei occur in cells of the hair matrix, the lower to mid-cortex, the lower medulla and hair cuticle, the cuticle of the IRS, and the companion layer.a′, DAPI-staining of the section in a. b, higher magnification of the boxed area b in a. b′, DAPI-staining ofb. c, higher magnification of the boxed area c ina. c′, DAPI-staining of c. Note the slightly oblique angle of the longitudinal section. A schematic illustration of the HOC13 expression pattern in the hair follicle is given at the right hand side. ORS, outer root sheath; IRS, inner root sheath; ma, matrix; co, cortex;med, medulla; cu, cuticle of the hair;cu-IRS, cuticle of the IRS; cl, companion layer;dp, dermal papilla. Bars = 100 μm.

       HOXC13 Is Able to Activate Human Hair Keratin Promoters

      To investigate whether hair keratin genes are target genes of HOXC13, we first selected the two functional human hair keratin genes, hHa5 and hHa2, whose expression patterns matched that of HOXC13 in the lower hair-forming compartment of the anagen hair follicle (Ref.
      • Langbein L.
      • Rogers M.A.
      • Winter H.
      • Praetzel S.
      • Beckhaus U.
      • Rackwitz H.R.
      • Schweizer J.
      and Fig. 1B). Moreover, we included the transcribed pseudogene ϕhHaA, whose defective and untranslated mRNA species could previously be demonstrated in pre- to mid-cortex cells of the hair shaft (
      • Winter H.
      • Langbein L.
      • Krawczak M.
      • Cooper D.N.
      • Jave-Suarez L.
      • Rogers M.A.
      • Praetzel S.
      • Heidt P.J.
      • Schweizer J.
      ).
      In initial experiments, various promoter deletion fragments (0.9–0.2 kb) of the three hair keratin genes, cloned into the β-galactosidase expression vector pNassβ, were transiently transfected into PtK2 cells either without or together with the expression vector pHOXC13 (Fig. 3). In each case, cotransfection of the largest promoter constructs (i.e. 0.9a5-β-gal, 0.9a2-β-gal, 0.8aA-β-gal) with pHOXC13 led to a strong 7.5- to 13-fold increase in β-galactosidase activity (Fig. 3,A–C). While slight variations in β-galactosidase activity were observed for the respective medium sized promoter fragments (constructs 0.4a5-β-gal, 0.6a2-β-gal,0.6aA-β-gal, 0.4aA-β-gal), consistently, cotransfections of the smallest promoter fragments (constructs 0.3a5-β-gal, 0.3a2-β-gal, 0.2aA-β-gal) led to β-galactosidase activities that were comparable with those obtained with the largest promoter versions (Fig. 3, A–C).
      Figure thumbnail gr3
      Figure 3hHa5, hHa2, and ϕhHaA hair keratin reporter genes are activated by HOXC13. Schematic presentation of β-galactosidase reporter constructs containing promoter fragments of hair keratin genes/pseudogenes hHa5(A), hHa2 (B), and ϕhHaA(C). The numbering of the three promoters refers to the respective gene sequences. Deletion variants of the individual promoters are indicated according to their length in kb.Arrowheads denote the position of the TATA box in each promoter. Reporter constructs were transiently cotransfected with a HOXC13 expression vector into PtK2 cells, x-fold β-galactosidase activity (hatched columns) refers to values obtained upon transfection of the respective reporter constructs alone. Each value represents the mean of at least four independent experiments; calculated standard deviations are shown.

       Multiple Putative HOXC13 Binding Core Motifs Are Present in the Promoters of Hair Keratin Genes

      To be able to interpret the results of the transfection experiments described above, we set out to systematically screen the largest promoter fragments of the functional hHa5 and hHa2 genes for putative HOXC13 binding sites. Although recent studies have provided evidence that members of Hox paralogs 1–8 preferentially bind to oligonucleotides containing conical TAAT core motifs, while members of the Abd-B-like Hox paralogs 9–13 exhibited a binding preference for either TTAT or TTAC core motifs (18–21; see also “Discussion”), we analyzed the hHa5 and hHa2 promoters for all three core motifs. Fig. 4Ashows that the 0.9-kb hHa5 promoter fragment harbored 13 putative HOXC13 binding sites. Considering that motif 7 contained overlapping TAAT and TTAC sequences, all in all, six motifs were TAAT and four were either TTAT or TTAC. To determine which of them are binding sites for HOXC13, we generated 12 35–40 mer oligonucleotides (underlined in Fig. 4A), containing the numbered core motifs in a central position. Except for the two oligonucleotides that collectively comprised putative binding motifs 3, 4, 5, and 4, 5, respectively, the remaining oligonucleotides contained only one core motif (Fig. 4A). The HOXC13 binding capacity of these oligonucleotides was explored by means of electrophoretic mobility shift assays, using nuclear extracts of both untransfected PtK2 cells and PtK2 cells that had been transfected with a pHOXC13 expression vector, encoding the entire HOXC13 protein. Fig. 4B reveals that of the 12 oligonucleotides investigated, only six contained HOXC13 binding sequences (Oa5–2, Oa5–3/4/5, Oa5–4/5 (not shown), Oa5–6, Oa5–9, and Oa5–13). Consistently, the DNA/protein complexes showed up as a major, fast migrating and a minor, slower migrating shifted band (arrowheads in Fig. 4B). To determine which of the three core motifs in oligonucleotide 0a5–3/4/5 were involved in the particularly strong HOXC13 binding, we first mutated motif 3. As shown in Fig. 4B, oligonucleotide Oa5-m3/4/5 formed a much weaker DNA/protein complex compared with its normal counterpart. An essentially comparable result was obtained upon mutation of motif 4 (oligonucleotide Oa5–3/m4/5; Fig. 4B). Since mutation of motif 4 in the HOXC13 binding oligonucleotide Oa5–4/5 (i.e.oligonucleotide Oa5-m4/5) did not lead to a binding complex (Fig.4B), these data imply that in oligonucleotide Oa5–3/4/5, only core motifs 3 and 4 were binding sites for HOXC13.
      Figure thumbnail gr4
      Figure 4HOXC13 does not bind to all T(T/A)A(T/C) core motifs of the hHa5 promoter (A) hHa5promoter sequence. The TATA box is indicated by a gray box. Putative Hox binding core motifs are boxed in dark gray and numbered 1–13 in 3′ to 5′ direction. Oligonucleotides used for EMSA inB, C, and D are underlined. HOXC13 binding core motifs are indicated by circled numbers. Bent arrows denote the 5′-ends of promoter constructs used for transfection studies (see Fig. ). B, oligonucleotides derived from the hHa5 promoter are generally designated 0a5, followed by numbers corresponding to the individual core motifs boxed in A. Mutation of distinct core motifs (generally from T(T/A)A(T/C) to GCCG) are indicated by m in front of the oligonucleotide designation. Double stranded, end-labeled oligonucleotides were incubated with nuclear protein extracts of either untransfected (lanes 1) or pHOXC13 transfected PtK2 cells (lanes 2). Oligonucleotides exhibiting HOXC13 binding areunderlined. DNA/protein binding complexes are marked byarrowheads. C, HOXC13 binding oligonucleotides 0a5–2, 0a5–3/4/5, 0a5–6, and 0a5–9 incubated with nuclear protein extracts of untransfected (lanes 1) or pHOXC13-transfected PtK2 cells (lanes 2) were assayed for competition with a 100-fold excess of the corresponding unlabeled oligonucleotide (lanes 3), a random oligonucleotide (lanes 4), or following incubation with nuclear protein extracts of PtK2 cells transfected with ΔpHOXC13, i.e. a homeobox-truncated HOXC13 version (lanes 5). D, lanes 1and 2 correspond to the respective lanes in B andC. The indicated oligonucleotides were further incubated with nuclear protein extracts of pHOXC13-transfected PtK2 cells, to which an antibody against either HOXC13 (lanes 3) or the human androgen receptor (lanes 4) was added. HOXC13 binding complexes are marked by arrowheads, supershifted complexes are denoted by asterisks.
      A more detailed EMSA analysis of the most proximal binding core motifs 2, 3/4, 6, and 9 is given in Fig. 4C. In each case, the strong HOXC13 binding (Fig. 4C, lanes 2) was completely abolished by a 100-fold excess of the respective unlabeled oligonucleotide (Fig. 4C, lanes 3), but remained undisturbed upon addition of a heterologous probe of a random sequence (Fig. 4C, lanes 4). As expected, no binding complexes were observed with ΔHOXC13, coding for a homeobox-truncated HOXC13 protein (Fig. 4C, lanes 5). Moreover, supplementation of the binding assays with the specific HOXC13 antiserum led to a strong supershift of both the major and minor complexes (Fig. 4D, lanes 3; asterisksdenote the supershifted DNA/protein complexes), which was not observed upon addition of an unrelated antibody (Fig. 4D, lanes 4).
      An identical study was performed for the 12 putative HOXC13 binding motifs present in the hHa2 promoter (Fig.5A). Considering that motif 1 contained three overlapping TAAT motifs and that motifs 2 and 8 harbored overlapping TAAT/TTAT and TAAT/TTAC motifs, respectively, nine motifs were TAAT, six were TTAT, and one was TTAC (Fig. 5A). Of the 10 oligonucleotides (underlined in Fig. 5A) generated for the analysis of the HOXC13 binding capacities of the various core motifs, only oligonucleotides Oa2–1/2 and Oa2–9/10 contained two closely spaced core motifs. EMSA revealed that HOXC13 bound only to four oligonucleotides, Oa2–1/2, Oa2–3, Oa2–6, and Oa2–11 (Fig.5B). Since in Oa2–1/2, mutation of the complex motif 1 (oligonucleotide Oa2-m1/2), but not of motif 2 (oligonucleotide Oa2–1/m2), completely abolished the binding capacity (Fig.5B), the observed HOXC13 binding to the original oligonucleotide could be ascribed to motif 1. Once again, complex formation involving motifs 1, 3, 6, and 11 was not affected by unspecific competition (Fig. 5C, lanes 4), but inhibited either by specific competition (Fig. 5C,lanes 3) or by exposure to truncated ΔHOXC13 protein (Fig.5C, lanes 5). In keeping with these data, ΔHOXC13 also failed to activate the reporter gene of β-galactosidase constructs 0.3a5, 0.3a2, and also 0.2aA (see Fig. 1,A–C) in cotransfection assays (Fig.6). DNA/protein binding complexes formed by HOXC13 with the proximal core motifs 1 and 3 were supershifted in the presence of HOXC13 antiserum (Fig. 5C, lanes 3; asterisks denote the supershifted DNA/protein complexes), but not in the presence of an unrelated antibody (Fig.5C, lanes 4).
      Figure thumbnail gr5
      Figure 5HOXC13 does not bind to all T(T/A)A(T/C) core motifs of the hHa5 promoter (A) hHa2 promoter sequence.B, oligonucleotides derived from the hHa2promoter are generally designated 0a2, followed by numbers corresponding to the individual core motifs boxed in A. Cand D, control experiments for HOXC13 binding oligonucleotides 0a2–1/2, 0a2–3, 0a2–6, and 0a2–11. For details see legend to Fig. .
      Figure thumbnail gr6
      Figure 6Hair keratin reporter genes are not transactivated by a homeobox-truncated HOXC13. PtK2 cells were transiently transfected with reporter constructs 0.3a5, 0.3a2, 0.2aA, or 0.2a7 alone (black columns) or together with expression plasmids pHOXC13 (shaded columns), and pΔHOXC13 (white columns). Results are given as × fold β-galactosidase activity relative to control (average of four experiments ± S.D).

       HOXC13 Binding to Hair Keratin Promoters Involves Core Motifs TAAT and TTAT, but Not TTAC

      The analysis of HOXC13 binding core motifs in the hHa5 and hHa2 promoters revealed that only TAAT and TTAT sequences, but in no case TTAC sequences, were involved in HOXC13 binding (see Figs. 4, A, B and 5,A, B). Since, however, not all of the TAAT and TTAT motifs bound HOXC13, we systematically compared the flanking sequences of all binding and non-binding core motifs present in thehHa5 and hHa2 promoters. As shown in Fig.7A, all binding TAAT and TTAT core motifs were preceded by a T. Moreover, while the position immediately 3′ to binding core motifs was variable, the next two positions were, with one exception, purine residues (Fig.7A). From this it appears that HOXC13 binding sites in thehHa5 and hHa2 hair keratin promoters exhibit an 8-bp consensus sequence 5′-TT(A/T)ATNPuPu-3′ (boxed in Fig.7A). This implies that of the three overlapping TAAT core motifs present in the complex motif 1 of the hHa2 promoter (Fig. 5A), the one being flanked by T and TAG, respectively, most probably represents the active HOXC13 binding site (Fig.7A).
      Figure thumbnail gr7
      Figure 7Elucidation of a DNA consensus sequence for HOXC13 binding to hair keratin promoters. A, HOXC13 binding sequences. Aligned are the core sequences as well as 5′- and 3′-flanking sequences of the individual HOXC13 binding sites of thehHa5, hHa2, and hHa7 promoters (see Figs. A, A, and A). The deduced HOXC13 consensus binding sequence is boxed. The respective core motifs are given in blue, conserved flanking nucleotides are indicated in red. B, HOXC13 nonbinding sequences. Deviations from the consensus binding sequence are given in green. 1, denotes overlapping core motifs.
      The majority of the 20 non-binding core motifs of the hHa5and hHa2 promoters exhibited two or three deviations from the consensus sequence (Fig. 7B). However, also single base deviation, such as T→G and T→C in position 1 (motifs 7a of thehHa5 promoter and 5 of the hHa2 promoter as well as motif 1a of the hHa2 promoter and T→C in position 5 (motifs 1 and 11 of the hHa5 promoter) (Fig. 7B), were sufficient to completely abolish the HOXC13 binding capacity of the respective motifs. In this context it is, however, noteworthy that two non-binding motifs, i.e. 1b of the hHa2promoter and 8 of the hHa5 promoter, exhibited a single base variation, Pu→T and Pu→C, in position 7 of the consensus sequence (Fig. 7B), while a similar deviation (i.e.Pu→T) did not affect the HOXC13 binding capacity of motif 11 of thehHa2 promoter (Fig. 7A).

       The Consensus Motif TT(A/T)ATNPuPu in the Proximal Promoters of Hair Keratin Genes Is Involved in HOXC13-mediated Gene Activation

      Our transfection studies described in Fig. 1 clearly showed that the HOXC13-induced β-galactosidase gene activation is essentially mediated by an about 200-bp proximal promoter region upstream of the TATA box. In the hHa5 gene, this promoter region contained three HOXC13 binding TAAT core motifs besides one binding TTAT core motif (Fig. 4A), while the correspondinghHa2 promoter region harbored only one HOXC13 binding core motif of each type (Fig. 5A). Considering that murine Hoxb13 and Hoxd13 preferred TTAT and TTAC over TAAT as core binding motifs (
      • Shen W.F.
      • Montgomery J.C.
      • Rozenfeld S.
      • Moskow J.J.
      • Lawrence H.J.
      • Buchberg A.M.
      • Largman C.
      ,
      • Shen W.F.
      • Rozenfeld S.
      • Lawrence H.J.
      • Largman C.
      ), the possible participation of TAAT core sequences in HOXC13-mediated reporter gene activation seemed surprising. Interestingly, however, the screening of further hair keratin promoters revealed that an ∼200-bp region upstream of the TATA box of hair keratin gene hHa7, expressed in the lower medulla,
      L. F. Jave-Suarez, H. Winter, L. Langbein, M. A. Rogers, and J. Schweizer, manuscript in preparation.
      contained two TAAT motifs 1 and 2, with motif 1 overlapping with a TTAT motif (Fig. 8A). Of these two potential HOXC13 binding sites, only motif 2 obeyed to the 8-bp consensus sequence TT(A/T)ATNPuPu (Fig.7A), while the corresponding sequences of the complex motif 1 deviated in either one or two positions (Fig. 7B). Accordingly, EMSA with oligonucleotides harboring either motif 1 (oligonucleotide 0a7–1) or motif 2 (oligonucleotide 0a7–2, both underlined in Fig. 8A) clearly demonstrated that only TAAT motif 2 bound HOXC13 (Fig. 8B, lane 2). Control experiments, such as specific (Fig. 8C, lane 3) and unspecific competition (Fig. 8C, lane 4) or the use of the homeobox-truncated ΔHOXC13 protein (Fig.8C, lane 5), yielded results as expected for a functional core motif. Therefore, the weak, but significant ∼1.5-fold β-galactosidase activation observed upon cotransfection of the 0.2a7 reporter gene construct with pHOXC13, but not with pΔHOXC13 (Fig. 6), was clearly mediated by HOXC13 binding to the single TTAATGAG consensus sequence in the proximal hHa7promoter.
      Figure thumbnail gr8
      Figure 8Elucidation of HOXC13 binding core motifs in the proximal hHa7 promoter (A) hHa7 promoter sequence. The TATA box is indicated by a gray box. Putative HOXC13 binding core motifs are boxed in dark gray and numbered 1–2 in 3′ to 5′ direction. Oligonucleotides 0a7–1 and 0a7–2 used for EMSA inB and C are underlined. B, EMSA with oligonucleotides 0a7–1 and 0a7–2. For details see legend to Fig. B. C, control experiments for HOXC13 binding oligonucleotide 0a7–2. For details see legend to Fig.C.

      DISCUSSION

      Hox genes encode transcription factors that act as master switches during embryonic development by controlling the activities of a plethora of downstream target genes. The knowledge about direct Hox target genes is presently poor and most of them have been identified in the fruit fly Drosophila (
      • Graba Y.
      • Aragnol D.
      • Pradel J.
      ). In mammals both the Hoxc8-mediated repression of the osteopontin gene and the HOXA5-controlled activation of the progesterone receptor gene have recently been investigated at the molecular level (
      • Shi X.
      • Yang X.
      • Chen D.
      • Chang Z.
      • Cao X.
      ,
      • Raman V.
      • Tamori A.
      • Vali M.
      • Zeller K.
      • Korz D.
      • Sukumar S.
      ). Based on strong evidence indicating a regulatory function of Hoxc13 for hair follicle-specific genes in mice, which either under- or overexpressed this transcription factor (
      • Godwin A.R.
      • Capecchi M.R.
      ,
      • Tkatchenko A.V.
      • Visconti R.P.
      • Shang L.
      • Papenbrock T.
      • Pruett N.D.
      • Ito T.
      • Ogawa M.
      • Awgulewitsch A.
      ), the present study was aimed at investigating whether human hair keratin genes are targets for HOXC13. A prerequisite thereof was the demonstration of HOXC13 expression in the adult human anagen hair follicle. Using a HOXC13-specific antiserum, we extended the existing murine Hoxc13 mRNA expression data (
      • Godwin A.R.
      • Capecchi M.R.
      ) by showing that within the hair-forming compartment of the hair follicle, HOXC13 was strongly present in the nuclei of cells of the entire matrix, including cells lining the dermal papilla, the lower hair cuticle and cortex, in which, however, HOXC13 expression gradually vanished. In beard hairs, cells in the lowermost portion of the central medulla also contained HOXC13 protein in their nuclei. Accordingly, we generated β-galactosidase reporter gene constructs with promoter fragments of type I hair keratin genes hHa2,hHa5, and ϕhHaA, which are expressed in these regions (
      • Langbein L.
      • Rogers M.A.
      • Winter H.
      • Praetzel S.
      • Beckhaus U.
      • Rackwitz H.R.
      • Schweizer J.
      ,
      • Langbein L.
      • Rogers M.A.
      • Winter H.
      • Praetzel S.
      • Schweizer J.
      ,
      • Winter H.
      • Langbein L.
      • Krawczak M.
      • Cooper D.N.
      • Jave-Suarez L.
      • Rogers M.A.
      • Praetzel S.
      • Heidt P.J.
      • Schweizer J.
      ), for cotransfection studies with a HOXC13 expression vector. These studies clearly showed that hair keratin promoters strongly activated the reporter gene in the presence of functional but not homeobox deleted HOXC13.
      Numerous investigations, relying essentially on both DNA site selection protocols and gelshift assays with appropriate oligonucleotides, have provided evidence that Hox proteins may activate or repress their target genes (
      • Shi X.
      • Yang X.
      • Chen D.
      • Chang Z.
      • Cao X.
      ,
      • Raman V.
      • Tamori A.
      • Vali M.
      • Zeller K.
      • Korz D.
      • Sukumar S.
      ) primarily via binding of their highly conserved homeobox domain to specific DNA binding motifs. It could be demonstrated that Hox proteins of paralogous groups 1–8 bound preferentially to TAAT core motifs, while both DrosophilaAbd-B protein and Abd-B-like Hox9–13 paralogs exhibited a binding preference for both TTAT and TTAC core motifs (
      • Ekker S.C.
      • Jackson D.G.
      • von Kessler D.P.
      • Sun B.I.
      • Young K.E.
      • Beachy P.A.
      ,
      • Shen W.F.
      • Montgomery J.C.
      • Rozenfeld S.
      • Moskow J.J.
      • Lawrence H.J.
      • Buchberg A.M.
      • Largman C.
      ,
      • Shen W.F.
      • Rozenfeld S.
      • Lawrence H.J.
      • Largman C.
      ,
      • Shen W.F.
      • Rozenfeld S.
      • Kwong A.
      • Kömüves L.G.
      • Lawrence H.J.
      • Largman C.
      ). Notwithstanding these apparent binding site preferences, we have screened the endogenous promoters of the functional hair keratin geneshHa5 and hHa2 for all three versions of the core motif. EMSA showed that of the 30 putative core binding motifs collectively present in the two hair keratin promoters, only 10 bound to HOXC13, with six of them being concentrated in the respective proximal promoter regions. In accordance with the existing rules for the site selection preference of Abd-B protein and Hox9–13 paralogs (
      • Ekker S.C.
      • Jackson D.G.
      • von Kessler D.P.
      • Sun B.I.
      • Young K.E.
      • Beachy P.A.
      ,
      • Shen W.F.
      • Montgomery J.C.
      • Rozenfeld S.
      • Moskow J.J.
      • Lawrence H.J.
      • Buchberg A.M.
      • Largman C.
      ,
      • Shen W.F.
      • Rozenfeld S.
      • Lawrence H.J.
      • Largman C.
      ), we confirmed a generally strong binding of HOXC13 to three oligonucleotides containing TTAT motifs. However, while virtually no binding was observed for TTAC motifs, surprisingly, HOXC13 clearly bound to seven oligonucleotides containing TAAT motifs. Collectively, the intensities of the TTAT or TAAT HOXC13 binding complexes were comparably strong. This finding clearly contrasted to previous binding studies with the murine Hoxb13 and d13 paralogs, which did not efficiently select for TAAT motifs, but rather exhibited a strong preference for TTAC motifs (
      • Shen W.F.
      • Montgomery J.C.
      • Rozenfeld S.
      • Moskow J.J.
      • Lawrence H.J.
      • Buchberg A.M.
      • Largman C.
      ,
      • Shen W.F.
      • Rozenfeld S.
      • Lawrence H.J.
      • Largman C.
      ). Given that HOXC13 binding to both TTAT and TAAT motifs in the hair keratin promoters was not observed with a truncated HOXC13 protein and that in addition, the binding complexes could be specifically supershifted with the HOXC13 antiserum, these data clearly demonstrated a direct binding of HOXC13 to these core motifs in the hair keratin promoters.
      Previous analyses of the binding sequences for Abd-B-like Hox proteins also revealed distinct base constraints in the regions flanking the core motifs (
      • Ekker S.C.
      • Jackson D.G.
      • von Kessler D.P.
      • Sun B.I.
      • Young K.E.
      • Beachy P.A.
      ,
      • Shen W.F.
      • Montgomery J.C.
      • Rozenfeld S.
      • Moskow J.J.
      • Lawrence H.J.
      • Buchberg A.M.
      • Largman C.
      ,
      • Shen W.F.
      • Rozenfeld S.
      • Lawrence H.J.
      • Largman C.
      ,
      • Shen W.F.
      • Rozenfeld S.
      • Kwong A.
      • Kömüves L.G.
      • Lawrence H.J.
      • Largman C.
      ). An inspection of all 10 HOXC13 binding core motifs of the endogenous hHa5 and hHa2 hair keratin promoters showed that these were invariably preceded 5′ by a T and followed 3′ by a variable nucleotide and, as a rule, by two purine residues, thus giving rise to an extended 8-bp consensus-binding sequence TT(A/T)ATNPuPu. Obviously, the T in position 1 is generally mandatory for binding since it was invariably present in oligonucleotides that were site-selected with bothDrosophila Abd-B protein (
      • Ekker S.C.
      • Jackson D.G.
      • von Kessler D.P.
      • Sun B.I.
      • Young K.E.
      • Beachy P.A.
      ) and a large number of Abd-B-like Hox proteins (Hoxb9, Hoxa10, Hoxa11, Hoxd12, Hoxd13, Hoxb13 (
      • Shen W.F.
      • Montgomery J.C.
      • Rozenfeld S.
      • Moskow J.J.
      • Lawrence H.J.
      • Buchberg A.M.
      • Largman C.
      ,
      • Shen W.F.
      • Rozenfeld S.
      • Lawrence H.J.
      • Largman C.
      )). Moreover, its selective substitution in the extended HOXC13 consensus binding sequence entailed a complete abolishment of HOXC13 binding. Whether the T upstream of position 1 of the consensus sequence, which preceded 8 of 11 HOXC13 binding sequences (Fig. 7), is involved in binding remains to be seen. Regarding base constraints 3′ to the core motifs, Drosophila Abd-B protein seems to prefer a G in position six of the consensus sequence (
      • Ekker S.C.
      • Jackson D.G.
      • von Kessler D.P.
      • Sun B.I.
      • Young K.E.
      • Beachy P.A.
      ), while the above mentioned Hox9–13 proteins and HOXC13 obviously tolerate any base at this position (
      • Shen W.F.
      • Montgomery J.C.
      • Rozenfeld S.
      • Moskow J.J.
      • Lawrence H.J.
      • Buchberg A.M.
      • Largman C.
      ,
      • Shen W.F.
      • Rozenfeld S.
      • Lawrence H.J.
      • Largman C.
      ). In contrast, efficient binding of bothDrosophila Abd-B protein and Abd-B-like Hox members requires a purine base in position 7 of the consensus sequence (18–20 and this study). Interestingly, while Drosophila Abd-B protein and Hox9–12 paralogs appear to have a strong preference for a C in position 8 of the consensus sequence (
      • Ekker S.C.
      • Jackson D.G.
      • von Kessler D.P.
      • Sun B.I.
      • Young K.E.
      • Beachy P.A.
      ,
      • Shen W.F.
      • Montgomery J.C.
      • Rozenfeld S.
      • Moskow J.J.
      • Lawrence H.J.
      • Buchberg A.M.
      • Largman C.
      ,
      • Shen W.F.
      • Rozenfeld S.
      • Lawrence H.J.
      • Largman C.
      ), murine Hoxd13 and Hoxb13 (
      • Shen W.F.
      • Montgomery J.C.
      • Rozenfeld S.
      • Moskow J.J.
      • Lawrence H.J.
      • Buchberg A.M.
      • Largman C.
      ,
      • Shen W.F.
      • Rozenfeld S.
      • Lawrence H.J.
      • Largman C.
      ) as well as human HOXC13, clearly exhibit a higher specificity for a G in this position. Thus the present data on the optimal sequences necessary for binding of the three Hox13 paralogs investigated so far suggest that all require a purine base in position 8 of the consensus sequence, which distinguishes them from the remaining Abd-B-like Hox members. Within the paralogous group 13, however, HOXC13 differs from Hoxb13 and Hoxd13 by the absolute need of a T in position 5, while the others also accept a C. Conversely, while there seems to be an absolute constraint for a T in position 3 for efficient Hoxb13 and Hoxd13 binding, HOXC13 seems to prefer an A, which is normally discriminatory for the binding of Hox1–8 proteins (
      • Ekker S.C.
      • Jackson D.G.
      • von Kessler D.P.
      • Sun B.I.
      • Young K.E.
      • Beachy P.A.
      ,
      • Shen W.F.
      • Montgomery J.C.
      • Rozenfeld S.
      • Moskow J.J.
      • Lawrence H.J.
      • Buchberg A.M.
      • Largman C.
      ,
      • Shen W.F.
      • Rozenfeld S.
      • Lawrence H.J.
      • Largman C.
      ).
      Our transfection studies with reporter gene constructs ofhHa5 and hHa2 promoter fragments have clearly shown that the proximal 200 bp containing varying numbers of HOXC13 binding TAAT and TTAT motifs are sufficient for optimal reporter gene activation by HOXC13. Remarkably, HOXC13-dependent transactivation was also observed for a β-galactosidase construct containing a proximal promoter region of the medullar hair keratin genehHa7,3 which, however, exhibited only one HOXC13 binding TTAATGAG consensus binding sequence. This strongly suggests that both TAAT and TTAT core motifs are involved in HOXC13-mediated reporter gene activation. Whether the relatively low β-galactosidase gene activation observed after cotransfection of pHOXC13 with the 0.2 bp hHa7 promoter construct as compared with that obtained with the corresponding hHa5 andhHa2 promoter constructs is indicative of a hierarchy within TAAT and TTAT core motifs regarding their individual contribution to gene activation can only be resolved by detailed mutation/deletion studies of the critical core motifs of the hair keratin promoters.
      It is evident that hair keratin promoters would be well suited for studies aimed at investigating the role of cofactors that influence Hox binding. It has been shown that for efficient binding to their consensus sequences, Hox9–10 paralogs either associate with members of Exd/Pbx (
      • Mann R.S.
      • Chan S.K.
      ,
      • Phelan M.L.
      • Rambaldi I.
      • Featherstone M.S.
      ) or Meis1 (
      • Moskow J.J.
      • Bullrich F.
      • Huebner K.
      • Daar I.O.
      • Buchberg A.M.
      ) family of homeodomain proteins or form triple complexes with these proteins (
      • Shen W.F.
      • Montgomery J.C.
      • Rozenfeld S.
      • Moskow J.J.
      • Lawrence H.J.
      • Buchberg A.M.
      • Largman C.
      ,
      • Shen W.F.
      • Rozenfeld S.
      • Lawrence H.J.
      • Largman C.
      ,
      • Shen W.F.
      • Rozenfeld S.
      • Kwong A.
      • Kömüves L.G.
      • Lawrence H.J.
      • Largman C.
      ). In contrast, Hox11–13 paralogs are only able to interact with Meis1, but have been shown to bind equally strong to their consensus sequence in the absence of Meis1(19–21). These studies were performed using oligonucleotides containing either the Edx/Pbx binding consensus sequence ATGAT or the Meis1 binding sequence TGACAG contiguous with a Hox binding sequence (
      • Shen W.F.
      • Montgomery J.C.
      • Rozenfeld S.
      • Moskow J.J.
      • Lawrence H.J.
      • Buchberg A.M.
      • Largman C.
      ,
      • Shen W.F.
      • Rozenfeld S.
      • Lawrence H.J.
      • Largman C.
      ,
      • Shen W.F.
      • Rozenfeld S.
      • Kwong A.
      • Kömüves L.G.
      • Lawrence H.J.
      • Largman C.
      ). Since all HOXC13 binding oligonucleotides derived from thehHa5 and hHa2 promoters lacked a Meis1 binding sequence adjacent to the extended HOXC13 binding sequence, none of the two HOXC13 binding complexes consistently observed in our bandshift assays can be attributed to a HOXC13-Meis1 interaction. Meis1 binding consensus sequences are, however, present in both hair keratin promoters. The hHa5 promoter contains two TGACAG motifs, one immediately upstream of the 0.3a5 fragment, the other being located 15 nucleotides upstream of the first motif (see Fig. 4), while in thehHa2 promoter one TGACAG motif lies within the proximal 0.3a2 fragment, 27 nucleotides upstream of the nonbinding TAAT motif 4 (see Fig. 5). Both Meis1 binding studies, and in particular cotransfection studies of appropriately tailored and mutated/deletedhHa2 and hHa5 promoter fragments with a Meis1 expression vector into cells with constitutive HOXC13 expression, will help to better define the relevance of this cofactor for HOXC13 mediated gene activation.
      Our evidence in vitro for HOXC13-mediated hair keratin gene expression needs to be considered in the context of a recent study that showed that Hoxc13 overexpression in mice led to fragile hairs concomitant with a drastic down-regulation of late, hair-specific genes encoding members of the large multigene family of hair keratin-associated proteins, KAPs (
      • Powell B.C.
      • Rogers G.E.
      ,
      • Rogers M.A.
      • Langbein L.
      • Winter H.
      • Ehmann C.
      • Korn B.
      • Schweizer J.
      ), while changes in hair keratin gene expression were not reported (
      • Tkatchenko A.V.
      • Visconti R.P.
      • Shang L.
      • Papenbrock T.
      • Pruett N.D.
      • Ito T.
      • Ogawa M.
      • Awgulewitsch A.
      ). We believe that this observation is due to a selective shift of overexpressed Hoxc13 into the mid- to upper cortical compartment (
      • Tkatchenko A.V.
      • Visconti R.P.
      • Shang L.
      • Papenbrock T.
      • Pruett N.D.
      • Ito T.
      • Ogawa M.
      • Awgulewitsch A.
      ), where it may aberrantly affect KAP gene expression. In contrast, our finding that in normal anagen hair follicles, the nuclear HOXC13 expression pattern in the lower hair-forming compartment coincides with that of the hair keratins investigated her, strongly speaks for a gene-activating function of HOXC13 during early trichocyte differentiation.

      Acknowledgments

      We thank Silke Praetzel for excellent technical assistance, Herbert Spring (both from this Center) for help with LSM microscopy, and Frank Claessens, University of Leuven, Belgium, for the gift of an antibody against the human androgen receptor.

      REFERENCES

        • Gehring W.J.
        • Affolter M.
        • Bürglin T.
        Annu. Rev. Biochem. 1994; 63: 487-526
        • Maconochie M.
        • Nonchev S.
        • Morrison A.
        • Krumlauf R.
        Annu. Rev. Genet. 1996; 30: 529-556
        • Izpisua-Belmonte J.
        • Falkenstein H.
        • Dollé P.
        • Renucci A.
        • Duboule D.
        EMBO J. 1991; 10: 2279-2289
        • Peterson R.L.
        • Papenbrock T.
        • Davda M.M.
        • Awgulewitsch A.
        Mech. Dev. 1994; 47: 253-260
        • Capecchi M.R.
        Cold Spring Harbor Symp. Quant. Biol. 1997; LXII: 273-281
        • Godwin A.R.
        • Capecchi M.R.
        Genes Dev. 1998; 12: 11-20
        • Duboule D.
        Genes Dev. 1998; 12: 1-4
        • Godwin A.R.
        • Capecchi M.R.
        J. Invest. Dermatol. 1999; 4: 244-247
        • Tkatchenko A.V.
        • Visconti R.P.
        • Shang L.
        • Papenbrock T.
        • Pruett N.D.
        • Ito T.
        • Ogawa M.
        • Awgulewitsch A.
        Development. 2001; 128: 1547-1558
        • Rogers M.A.
        • Winter H.
        • Wolf C.
        • Heck M.
        • Schweizer J.
        J. Biol. Chem. 1998; 273: 26683-26691
        • Langbein L.
        • Rogers M.A.
        • Winter H.
        • Praetzel S.
        • Beckhaus U.
        • Rackwitz H.R.
        • Schweizer J.
        J. Biol. Chem. 1999; 274: 19874-19884
        • Rogers M.A.
        • Winter H.
        • Langbein L.
        • Wolf C.
        • Schweizer J.
        J. Invest. Dermatol. 2000; 114: 464-472
        • Langbein L.
        • Rogers M.A.
        • Winter H.
        • Praetzel S.
        • Schweizer J.
        J. Biol. Chem. 2001; 276: 35123-35132
        • Winter H.
        • Langbein L.
        • Krawczak M.
        • Cooper D.N.
        • Jave-Suarez L.
        • Rogers M.A.
        • Praetzel S.
        • Heidt P.J.
        • Schweizer J.
        Hum. Genet. 2001; 108: 37-42
        • Stanchina E.
        • Gabelline D.
        • Norio P.
        • Giacca M.
        • Peverali F.A.
        • Riva S.
        • Falaschi A.
        • Biamonti G.
        J. Mol. Biol. 2000; 299: 667-680
        • Ekker S.C.
        • Jackson D.G.
        • von Kessler D.P.
        • Sun B.I.
        • Young K.E.
        • Beachy P.A.
        EMBO J. 1994; 13: 3551-3560
        • Shen W.F.
        • Montgomery J.C.
        • Rozenfeld S.
        • Moskow J.J.
        • Lawrence H.J.
        • Buchberg A.M.
        • Largman C.
        Mol. Cell. Biol. 1997; 17: 6448-6458
        • Shen W.F.
        • Rozenfeld S.
        • Lawrence H.J.
        • Largman C.
        J. Biol. Chem. 1997; 272: 8198-8206
        • Shen W.F.
        • Rozenfeld S.
        • Kwong A.
        • Kömüves L.G.
        • Lawrence H.J.
        • Largman C.
        Mol. Cell. Biol. 1999; 19: 3051-3061
        • Graba Y.
        • Aragnol D.
        • Pradel J.
        BioEssays. 1997; 19: 379-388
        • Shi X.
        • Yang X.
        • Chen D.
        • Chang Z.
        • Cao X.
        J. Biol. Chem. 1999; 274: 13711-13717
        • Raman V.
        • Tamori A.
        • Vali M.
        • Zeller K.
        • Korz D.
        • Sukumar S.
        J. Biol. Chem. 2000; 275: 26551-26555
        • Mann R.S.
        • Chan S.K.
        Trends Genet. 1996; 12: 258-262
        • Phelan M.L.
        • Rambaldi I.
        • Featherstone M.S.
        Mol. Cell. Biol. 1995; 15: 3989-3997
        • Moskow J.J.
        • Bullrich F.
        • Huebner K.
        • Daar I.O.
        • Buchberg A.M.
        Mol. Cell. Biol. 1995; 15: 5434-5443
        • Powell B.C.
        • Rogers G.E.
        Jollès P. Zahn H. Hoecker H. Formation and Structure of Human Hair. Birkhäuser Verlag, Basel, Boston, MA1997: 59-148
        • Rogers M.A.
        • Langbein L.
        • Winter H.
        • Ehmann C.
        • Korn B.
        • Schweizer J.
        J. Biol. Chem. 2001; 276: 19440-19451