Introduction

Multiple mouse mutations lead to hair loss by disrupting the hair cycle (Sundberg, 1994). Of these, the best characterized are allelic mutations at the hairless (Hr) locus (Brooke, 1926; Howard, 1940). Hairless acts as a transcriptional corepressor by heterodimerizing with the thyroid hormone receptor (Thompson and Bottcher, 1997), vitamin D receptor (Hsieh et al., 2003), and retinoic acid-like orphan-receptor (Moraitis et al., 2002), possibly mediated through association with histone deacetylases (Potter et al., 2002; Hsieh et al., 2003). A series of allelic mutations with various degrees of severity have arisen in murine Hr, all of which manifest in mice as an inability to regrow a normal coat of hair after the initial catagen stage, resulting in a progressive loss of hair from head-to-tail beginning around 14 days of age (Brooke, 1926; Howard, 1940). Although the molecular bases differ for Hr mutations characterized to date, all display recessive inheritance.

The only mouse mutation purported to be allelic with Hr but not displaying recessive inheritance was first described by Stelzner in 1983 (Stelzner, 1983) and designated near-naked (Hr^N). The mutation arose spontaneously in the mouse colony at Oak Ridge National Laboratory. Hr^N was defined as an allele of Hr based on the absence of wild-type mice among 106 offspring produced from the backcross mating of mice doubly heterozygous for Hr^N and Hr^hr Hr^hr/Hr^hr mice. However, these data were only suggestive of allelism, as they rule out a crossover frequency of 5.6% or higher at the 5% significance level (Stelzner, 1983). Hr^N differs from classical Hr mutations in that it exerts its effects in a semi-dominant rather than recessive manner. Heterozygous mice display a very sparse coat that undergoes some level of cyclic loss and regrowth, whereas homozygotes are virtually hairless. The etiology of the phenotype in Hr^N mutants also differs markedly from that of other Hr mutants in that Hr^N/Hr^N mice never grow a normal coat of hair, rather than failing to regrow the coat after the initial catagen (Stelzner, 1983).

In this study, we sought to identify the molecular basis for the Hr^N mutation. We also used histological and microarray analyses to define the morphological and molecular alterations that underlie hair loss in Hr^N mutant mice. On the basis of the results described herein, the Hr^N mutation causes hair loss by disrupting expression of genes necessary for normal hair structure. We also suggest that this phenotype results not from a coding region mutation in Hr but rather from either a closely linked gene or from a regulatory mutation that alters Hr expression.

Results

Phenotype

Hr^N mutant mice were distinguished from wild-type littermates at approximately 5 days of age by reduced numbers (Hr^N/+) or lack of (Hr^N/Hr^N) emerging hairs. At 13 days of age, Hr^N/+ mice displayed a sparse but uniform coat, whereas Hr^N/Hr^N mouse skin was virtually devoid of hair (Figure 1a). With age, Hr^N/Hr^N mice developed prominent wrinkles, similar to those of Hr/Hr or some of the milder forms of the rhino allelic mutations of Hr (Figure 1b). Vibrissae of Hr^N/Hr^N were sparse, short, and wavy, whereas those of Hr^N/+ mice were intermediate between Hr^N/Hr^N and +/+ in both length and texture (Figure 1c and d). Both body weight (Figure 1e) and body length (Figure 1f) were significantly reduced in Hr^N/Hr^N mice beginning at 7 days of age (P<0.05). Hr^N/+ mice displayed a similar reduction in size that was significant at some but not all ages measured (P<0.05). Gross necropsy did not reveal significant lesions in any organs other than skin (data not shown).

Figure 1.

The phenotype of Hr^N mutant mice. (a) Hair loss and gross appearance of 13-day-old +/+, Hr^N/+, and Hr^N/Hr^N mice (left-to-right); (b) skin wrinkling in an adult Hr^N/Hr^N mouse (5-month-old); vibrissae of +/+, Hr^N/+, and Hr^N/Hr^N mice (left-to-right) at (c) 7 days and (d) 5 weeks of age; (e, f) effect of the Hr^N mutation on body growth from 7 days to 5 weeks of age; (e) body weight and (f) nose-to-tail length were measured in independent sets of Hr^N/Hr^N, Hr^N/+, and +/+ mice at each time point (n=3–6 per time point). Data were analyzed for significant differences between genotypes using Student's t-test; ^*P<0.05.

Full figure and legend (140K)

Plucked hairs from wild-type mice had normal septation or septulation patterns based on hair fiber type. By contrast, hair fibers from Hr^N/+ mice, although relatively straight, lacked the septation pattern of the hair fiber medulla. The few fibers that emerged from hair follicles of Hr^N/Hr^N mice had short, disproportionate hair fibers with a wide base and a banding pattern, but no medullary septations (Figure 2). Scanning electron microscopy (SEM) confirmed the sparse (Hr^N/+) and virtually absent (Hr^N/Hr^N) coats compared with wild-type mice (Figure 3a–c). Fibers from heterozygous and homozygous mutant mice had regular but deformed cuticular scales relative to wild-type (Figure 3d–f).

Figure 2.

Light microscopy of hairs from Hr^N mice. Individual hairs were plucked and examined for gross morphology by light microscopy; (a–d) wild-type; (e–h) Hr^N/+; (i–l) Hr^N/Hr^N; a, b, e, f, i, and j 4 magnification; c, d, g, h, k, and l 25 magnification.

Full figure and legend (51K)

Figure 3.

SEM. Sections of dorsal skin were prepared for SEM as described; (a and d) wild-type; (b and e) Hr^N/+; (c and f) Hr^N/Hr^N; bar, a–c=200 m; d–f=20 m.

Full figure and legend (126K)

Histologic evaluation of the skin revealed onset of subtle lesions as early as 3 days of age for Hr^N/Hr^N mutant mice (Figure 4a). Wild-type follicles were complicated, well-ordered, and differentiated structures in late anagen by 7 days of age (Figure 4b). Follicles of 7-day-old heterozygous mice were very similar to wild-type with formation of hair fibers containing well-differentiated septations (Figure 4c). By contrast, homozygous 7-day-old mutant mice had a precortical region that became brightly eosinophilic, suggesting premature maturation or abnormal keratinization/cornification that resulted in a deformed hair fiber-lacking septulations (Figure 4d–f). This abnormality became prominent between 2 and 4 weeks of age (Figure 4g–j), resulting in cornified, fibrotic, and mineralized cysts by 6 months of age (Figure 4k and q). As hair follicles underwent involution on entry into catagen, with termination of hair fiber production, precortical changes were less obvious depending upon the orientation of sections (Figure 4l). There was marked apoptosis at the base of the follicle with either cornified remnants filling the hair follicle or traces of a thin hair fiber (Figure 4l). Heterozygous mice formed hair fibers that appeared normal from the precortex to the lower follicle but abruptly transitioned to weak and deformed fibers (Figure 4n). Six-month-old wild-type mouse skin stained with Masson's trichrome for collagen revealed little except for around blood vessels and small nerves in the dermis (Figure 4m). Similar amounts were present in Hr^N/+ mice (Figure 4o), but with occasional small fibrous tracks running from the base of involuted follicles into the hypodermal fat (Figure 4p), suggesting some degree of dermal scarring. In Hr^N/Hr^N mice (Figure 4q) there was marked epidermal thickening, a nonspecific finding in many mutant mice with alopecia (Sundberg, 1994). Follicles were distended with laminated cornified material that was mineralized (black in von Kossa-stained sections (Figure 4k) or blue with Masson's stains (Figure 4q)). These follicular cysts ruptured causing a foreign body granuloma and fibrosis (Figure 4q). Although proliferation rates appeared to be similar in late anagen-stage hair follicles in 7-day-old mice when labeled with Ki67, the lengths of the hair follicles were progressively shorter suggesting that miniaturization of the follicles is a sequella of increased copy number of the mutated allele (Figure 4r–t).

Figure 4.

Progressive histologic lesions in Hr^N mice; scale bar in each panel represents 100 m. (a) Changes in Hr^N/Hr^N mutant mice were visible as bright eosinophilic areas in the precortex in 3-day-old mice. (b) Normal 7-day-old mice had well-formed late anagen follicles. (c) Heterozygous mice appeared to have normal follicles at 7 days of age although by 9 days of age the normal septated hair fibers underwent abrupt changes and dystrophy (n, double-headed arrow). (d–f) Mutant mice had prominent eosinophilic cells with elongated nuclei in the precortical region at 7 days of age. (e) This change resulted in formation of a thin, deformed fiber with no septation pattern in its medulla. (f, arrows) When the follicle was placed out of focus to focus on the fiber these changes were evident. (g–j; g and h, 2 weeks of age; i, 3 weeks of age; j, 4 weeks of age) The precortical change was associated with formation of a highly deformed fiber to no fiber. (k) In older mice at 5 months of age, follicular cysts with amphophilic regions stained positive for mineralization using a von Kossa stain. (l) Involuting Hr^N/Hr^N follicles upon entry into catagen; arrow, apoptosis. (m, double arrows) Masson's trichrome stain revealed normal collagen around small blood vessels and nerves in wild-type and (o, arrow) heterozygous mice. (p, arrow) Occasionally, there were fibrous tracks into the hypodermal fat in heterozygous mice suggesting a follicular scar. (n) Heterozygous hair fibers abruptly became deformed (arrows) above a normal-appearing precortex and follicle region. (q) Homozygous mutant mice formed follicular cysts of uniform size filled with laminated cornified material that when ruptured resulted in a foreign body granuloma with fibrosis. (r–t) Ki67 expression was similar in +/+, Hr^N/+, and Hr^N/Hr^N at 7 days of age although miniaturization of follicles was evident.

Full figure and legend (410K)

Sequencing the Hr locus

Northern blots demonstrated that the Hr transcript was of similar size between Hr^N and wild-type mice, indicating that the mutation was not because of a large deletion or insertion in the mRNA, and did not cause production of a truncated transcript (data not shown). The Hr-coding region (coordinates 62,332,793–62,348,704, UCSC Genome Browser, May 2004 build) was sequenced from skin cDNA templates prepared from Hr^N/Hr^N mice and from mice of both the C3H/HeJ and 101/Rl wild-type inbred strains, the two parental strains in which the original mutation potentially occurred. The Hr^N cDNA sequence matched perfectly to cDNA sequences from both C3H/HeJ and 101/R1 (and to the published Hr sequence, NM_021877), demonstrating that the Hr^N mutation does not lie in the Hr-coding region. No sequence alterations specific to Hr^N/Hr^N mice were identified from genomic DNA sequencing of intron–exon boundaries, an upstream CpG island (62,330,506–62,331,530),1591 bp immediately downstream of exon 20 (62,349,943–62,351,533), and two regions upstream of exon 1 that are highly conserved between mouse and human (62,328,470–62,328,735 and 62,329,920–62,330,140). The cDNA for bone morphogenetic protein 1 (Bmp1), which lies within 36 kb of Hr, was also sequenced because of its proximity to Hr and because Bmp-receptor type 1A mutations cause hair loss (Kobielak et al., 2003; Andl et al., 2004). However, we did not identify a polymorphism that was specific to Hr^N/Hr^N mice.

Quantitative reverse transcription PCR (QPCR) assay revealed that Hr expression in 7-day-old mice was significantly increased in both Hr^N/+ and Hr^N/Hr^N (1.8- and 2.3-fold, respectively; P<0.05) compared with wild-type mice. At 5 weeks of age, however, Hr expression was significantly reduced in mice carrying the Hr^N mutation (40 and 20% of +/+ levels in Hr^N/+ and Hr^N/Hr^N, respectively). This change in expression could suggest a regulatory mutation in Hr as the basis for the Hr^N phenotype.

Microarray expression profiling

A focused cDNA array designed to represent genes important in skin and hair biology was used as an initial screen for expression changes in skin from 5-week-old Hr^N/+ and +/+ mice coisogenic on the BALC/cRl background. These hybridizations identified a total of 20 genes that were significantly differentially expressed across a panel of three pairs of 5-week-old wild-type and Hr^N/+ mice, using a false discovery rate of 10% (Table 1). Several genes involved in keratinocyte differentiation and hair structure were downregulated in Hr^N/+ skin, including hair keratins 2 and 3 (Krt1-2 and Krt1-3), keratin 1-c29 (Krt1-c29), four keratin-associated proteins (krtaps), S100a3 (a Ca⁺²-binding protein expressed specifically in postmitotic differentiated cells of the hair follicle) (Kizawa and Ito, 2005), and chaperonin subunit 7, a protein also downregulated in wound healing (Darden et al., 2000; Kizawa and Ito, 2005). Genes with increased expression were enriched in members of lipid metabolism pathways. In the absence of marked changes in sebaceous gland size or the thickness of the dermal adipose layer, upregulation of lipid metabolism in Hr^N/+ skin suggests an increase in sebum production, which may be in response to the alopecia.

Table 1 - Genes differentially expressed ( twofold change) in skin of 5-week-old Hr^N/+ mice compared with +/+ littermates.

Full table

To collect a broader profile of gene expression changes underlying the phenotype induced by Hr^N and to identify the early changes that coincide with manifestation of the phenotype, we next hybridized cutaneous RNA from 7-day-old Hr^N/Hr^N and +/+ littermates to a long oligo (65-mer) microarray representing 21,000 mouse transcripts. A total of 522 genes were differentially expressed in Hr^N/Hr^N skin compared with +/+, using a reasonably low projected level of false discovery (0.6%). Genes of interest, a subset of the 522, are listed in Table 2, whereas all of 522 genes are shown in Table S1. Strikingly, 99% (517) of these genes were downregulated, with only five genes showing significantly increased expression compared with wild-type. Those expressed at higher levels in Hr^N/Hr^N skin include GLI-Kruppel family member Gli3 (Gli3), adipsin (Adn), mast cell protease 2 (Mcpt2), Ig heavy chain (Ighg), and an uncharacterized cDNA (AK017710) that lies within paired box gene 5 (Pax5). QPCR was used to validate increased expression for four of these five genes (Gli3, Adn, Mcpt2, and AK017710) and confirmed significantly increased expression for Adn (5.2-fold), Mcpt2 (1.8-fold), and AK017710, expression of which appeared to be turned on in Hr^N/Hr^N skin. According to QPCR, Gli3 levels varied considerably among individuals within each genotype but was not significantly increased by the Hr^N mutation.

Table 2 - Partial list of genes differentially expressed ( twofold change) in skin of 7-day-old Hr^N/Hr^N mice compared with +/+ littermates.

Full table

Several genes with known or suspected roles in keratinocyte differentiation and hair formation showed reduced expression in Hr^N/Hr^N skin, including psoriasis susceptibility 1 candidate 2 (Psors1c2), small proline-rich protein 1B (Sprr1B), and mu-crystallin (Crym) (Mischke et al., 1996; Aoki et al., 2000; International Psoriasis Genetics Consortium, 2003). Consistent with the focused array data from 5-week-old mice, a total of seven keratins (including four hair keratins and three cytokeratins) and five krtaps also displayed decreased expression. QPCR assay of krtap genes not present on the arrays identified six additional krtaps with reduced expression in 7-day-old Hr^N/Hr^N compared with +/+, for a total of 11 when combined with those identified from microarrays. Therefore, the Hr^N mutation directly or indirectly disrupts the expression levels of a large group of genes critical for normal hair structure.

A variety of Gene Ontology (GO) categories showed statistically significant overrepresentation among the set of differentially expressed genes, highlighting functional pathways that were altered by the Hr^N mutation (Table S2). Ten GO terms with high specificity (level 5) within the Biological Process category included at least 10 genes each and collectively these 10 categories represented 21% of all differentially expressed genes (Table S3). The most highly populated GO term was transcription, DNA-dependent, containing a total of 31 genes (7.0%); all were significantly downregulated in Hr^N skin (Table S4). Eleven of these genes encode proteins containing a homeobox domain(s) (InterPro ID IPR001356), which was also significantly overrepresented among differentially expressed genes (P<0.00612).

Discussion

We characterized a novel spontaneous mouse mutation that was originally mapped to the Hr locus (Stelzner, 1983). On the basis of the initial allelism testing with Hr, we began by sequencing the Hr cDNA. Despite Hr as an obvious candidate for hair loss in this region of mouse Chr 14, we did not identify a mutation in the Hr-coding region, nor in the intron–exon boundaries or 5'- or 3'-untranslated region. We are therefore left to consider two alternate possibilities for the Hr^N phenotype. The first is that a mutation in another gene near Hr on Chr 14 is responsible for the Hr^N mutant phenotype. Several other genes known to be expressed in skin reside near Hr on m. Chr 14, including Bmp1, glial cell line-derived neurotrophic factor family-receptor alpha 2 (Gfra2), and protein phosphatase 3, catalytic subunit, gamma isoform (Ppp3cc). We did not identify a coding-region mutation in Bmp1, but other appealing candidates such as Gfra2 remain to be examined (Botchkareva et al., 2000). Further analysis and positional cloning are required to determine if one of these or another closely linked gene contains the mutation causing the Hr^N phenotype.

An alternative explanation for the Hr^N phenotype is that it results from a regulatory mutation in Hr, a transcriptional regulator known to play important roles in keratinocytes and the hair follicle (Potter et al., 2001, 2002; Hsieh et al., 2003). Hr expression was significantly increased in skin of 7-day-old Hr^N/Hr^N mice and to a lesser extent in mice carrying one mutant Hr^N allele. However, if the Hr^N mutation regulates Hr it does so in a hair cycle-dependent manner, given that the Hr expression profile was reversed across genotypes in 5-week-old mice. Elevated levels of Hr have been shown to disrupt the normal expression profile of keratinocytes both in vitro and in vivo (Beaudoin et al., 2005; Xie et al., 2006). Transgenic mice overexpressing Hr in mice under control of a K14-promoter (K14–rHr) mice displayed shorter hairs, a phenotype qualitatively similar to that of Hr^N/Hr^N mice, but no hair loss (Beaudoin et al., 2005). The observation that the vast majority (99%) of genes differentially expressed in 7-day-old Hr^N/Hr^N skin were downregulated also would be expected from increased expression of a transcriptional corepressor. In fact, the corollary expectation was recently verified in skin of Hr null mice (Hr^tm1Cct/Hr^tm1Cct), in which 93% of differentially expressed genes displayed increased expression relative to wild-type controls (Zarach et al., 2004). There was, however, no overlap between the two sets of genes differentially expressed in Hr null mice and Hr^N mutants. If Hr^N is because of increased HR, it might be predicted that the heterozygous phenotype would be more severe when Hr^N was balanced by either the rhino (Hr^rh) or hairless Hr^hr allele instead of the wild-type allele because of a net reduction in HR. However, Stelzner (1983) reported that the phenotype of double mutants between Hr^N and Hr^hr was indistinguishable from that of Hr^N/Hr^N, arguing against gain of function as the cause of Hr^N. Histologically, there were no significant similarities between the effects of Hr^N and hairless or rhino mutations. Hr^N mice have dystrophic hairs but normal hair cycles. The late anagen precortex undergoes premature cornification with changes more similar to those seen in desmoglein 4 (Dsg4) mutant mice (Kljuic et al., 2003) than in homozygous Hr^rh or Hr^hr mutants. By contrast, rhino and hairless mice initially develop a normal juvenile hair coat but lose it because of disrupted initial postnatal catagen by 2 weeks of age, when embryonic hair follicle development ends and the adult hair cycle begins. The dermal papilla does not reassociate normally with the bulge to reinitiate the hair cycle. Subsequently, the infundibulum and the dermal papilla with its associated keratinocytes develop separately to form utricles and deep dermal cysts, respectively (reviewed in Panteleyev et al., 1998a, 1998b; Sundberg et al., 1999). These lesions are histologically distinct from those that occur with age in Hr^N/Hr^N mice. Therefore, Hr^N mice display some of the phenotypic consequences (downregulated gene expression in skin, especially keratins and krtaps; shorter hairs) that would be predicted to result from overexpression of Hr, but lack features such as disrupted catagen and utricle formation that are hallmarks of known Hr mutations.

Although the exact molecular defect remains unknown, hair loss in Hr^N mice appears because of reduced expression of hair keratins and krtaps, resulting in dystrophic hairs that are unable to emerge consistently from the piliary canal and vibrissae that are shortened and crimped. This may be owing to direct effects of Hr^N on transcription of keratins and krtaps, similar to the mechanisms of alopecia consequent to mutations in the transcription factors Hoxc13 and Foxn1 (Flanagan, 1966; Rigdon and Packchanian, 1974; Kopf-Maier et al., 1990; Godwin and Capecchi, 1998; Meier et al., 1999; Schlake et al., 2000; Mecklenburg et al., 2001; Tkatchenko et al., 2001; Jave-Suarez et al., 2002). Alternatively, disrupted keratin and krtap expression could be secondary to reduced proliferation/accelerated terminal differentiation of matrix keratinocytes, similar to what has been described in mice lacking Dsg4 expression because of the lanceolate hair mutation (Dsg4^lah) (Sundberg et al., 2000; Kljuic et al., 2003). Further study will be necessary to investigate these possibilities.

Given the rapidly advancing nature of the field, it is intriguing to speculate that the Hr^N mutation lies not in a protein-coding gene but in a small non-coding RNA near the Hr locus. Such a mutation could explain the overwhelmingly disproportionate number of downregulated genes, for example a mutation that increased microRNA abundance. Two microRNAs (mmu-mir-124a and mmu-mir-320) are predicted to reside within 5 Mb of Hr (UCSC Genome Browser, mouse August 2005 assembly). We sequenced the precursors of each of these microRNAs but did not find a mutation (data not shown). At this point, however, we cannot eliminate the possibilities that the Hr^N mutation alters the expression of one of the nearby microRNAs or that it might be contained within an as yet unidentified, small non-coding RNA near Hr on Chr 14.

The only human hair syndrome (other than those owing to known mutations in HR) that maps to the same genomic region as Hr^N is Marie Unna hereditary hypotrichosis (OMIM %146550) (Cichon et al., 2000; Green et al., 2003). Marie Unna hereditary hypotrichosis is a rare, autosomal-dominant disorder characterized by coarsely textured hair that is progressively lost with age, beginning in early adulthood, with no other obvious phenotypic consequences (Argenziano et al., 1999; Cichon et al., 2000; Green et al., 2003). SEM images of hair fibers from Marie Unna hereditary hypotrichosis patients reveal a longitudinal groove and an irregular form, similar to that seen in Hr^N mice (Kim et al., 2001). For both the Hr^N mouse and Marie Unna hereditary hypotrichosis patients, mutations within the Hr-coding region were excluded as the cause of the hair phenotype (van Steensel et al., 1999; Cichon et al., 2000; Lefevre et al., 2000). It is possible that mutations in a homologous gene underlie both disorders or that both are because of a regulatory mutation in Hr.

In summary, the Hr^N mutation disrupts normal hair formation, paralleled by reduced expression of genes critical for hair structure and integrity. The mutation maps near the Hr locus but is not because of a mutation in the Hr cDNA. Further study is necessary to determine if the hair loss phenotype of Hr^N mice results from a mutation that drives increased Hr expression or to a mutation in a nearby gene on mouse Chr 14.

Materials and Methods

Animals and tissue collection

All mice were bred at Oak Ridge National Laboratory and experiments were conducted under approved Institutional Animal Care and Use Committee protocols. Hr^N mice were maintained on a congenic BALB/cRl background by crossing Hr^N/+ males (identified phenotypically) with wild-type BALB/cRl females (Stelzner, 1983). Hr^N/Hr^N mice used for sequencing were produced from matings of Hr^N/+ pairs of mice. Because intercross matings of Hr^N/+ mice on the BALB/cRl genetic background failed to reliably yield homozygous offspring, Hr^N/Hr^N mice used to produce samples for histology and RNA were produced by first outcrossing Hr^N/+ males to FVB/NTac females and then intercrossing F1 Hr^N/+ mice to produce the three genotypes (+/+, Hr^N/+, and Hr^N/Hr^N). Dorsal skin collected for RNA extraction was harvested in RNALater (Ambion, Austin, Texas). Tissues harvested for histologic analysis were fixed in 10% neutral buffered formalin for 24–48 hours and then stored at room temperature in 70% ethanol. Liver and spleen samples used for genomic DNA isolation were collected in liquid nitrogen and stored at -80°C.

PCR reactions and DNA sequencing

The full-length Hr cDNA (NM_021877) and Bmp1 cDNA (NM_009755) were sequenced using cDNA templates prepared from reverse transcription of skin RNA extracted from Hr^N/Hr^N mice on the BALB/cRl background and from C3H/HeJ and 101/Rl strains, the two potential strains on which the original Hr^N mutation arose. Portions of the Hr gene (Chr14: 62,324,459–62,349,944 on the UCSC Genome Browser (mouse May 2004 assembly), including intron/exon boundaries, sequence upstream of the coding region and downstream of exon 20 (the last exon) were sequenced from genomic DNA of Hr^N/Hr^N and +/+ littermate mice. Genomic DNA was extracted from liver and spleen using a standard protocol (Bultman et al., 1992) and cDNA was synthesized from DNAseI-treated total cutaneous RNA according to the manufacturer's protocol (Invitrogen, Carlsbad, CA). Primers to produce overlapping amplicons for both genomic DNA and cDNA templates were designed using the Primer3 database (Rozen and Skaletsky, 2000). After PCR optimization, purified amplicons were sequenced bidirectionally using BigDye Version 3.1 dye-terminator kit (ABI, Foster City, CA) and analyzed on an ABI PRISM 3100 Genetic Analyzer according to standard protocols. The sequences of all primers used are available on request.

RNA isolation

RNA was isolated from dorsal skin using the RNeasy Mini Kit (Qiagen, Valencia, CA). RNA quality was assessed by visualization in denaturing agarose gel electrophoresis and spectrophotometrically by the 260/280 nm ratio of absorbance. Samples were quantified spectrophotometrically based on the absorbance at 260 nm. Only RNA samples of high quality were used for further analysis.

Microarray construction, labeling, hybridization, and data analysis

Focused cDNA microarrays were designed and fabricated to represent genes of interest in skin and hair biology. Plasmid clones for selected cDNAs were purchased from Research Genetics (Huntsville, AL) and amplified and purified according to standard protocols (Hegde et al., 2000). cDNA inserts contained in plasmid DNA were verified by bidirectional sequencing using M13 universal primers. Sequences were analyzed using basic local alignment search tool (Altschul et al., 1990). Clones with confirmed identity were amplified by PCR, lyophilized, resuspended in 3 standard saline citrate, and spotted on triplicate on Corning UltraGAPS slides using an SDDC-2 arrayer (Virtek, Waterloo, Ontario, Canada). Microarrays representing 21,547 unique mouse genes were printed by the Center for Applied Genomics (PHRI, Newark, NJ) using the Compugen Mouse OligoLibrary™ (2.0). After printing, slides were air-dried and the cDNAs irreversibly immobilized by UV-crosslinking. Spot quality was assessed by hybridization with fluorescently labeled panomers according to manufacturer's protocols (Molecular Probes, Carlsbad, CA).

RNA samples were labeled using indirect dye incorporation, labeling reactions were purified and hybridized, and microarray slides were washed and scanned according to standard protocols (Hegde et al., 2000). Each hybridization consisted of a pair of RNA samples from littermate +/+ and Hr^N mutants (+/+ vs Hr^N/+, 5-week-old; +/+ vs Hr^N/Hr^N, 7-day-old). For each pair of mice, a dye swap was performed to control for dye-specific bias in labeling and to provide a replicate hybridization for each pair of samples. Slides were scanned for Cy3 and Cy5 fluorescence using a ScanArray 4000 confocal laser scanner (Perkin Elmer, Wellesley, MA), and resulting TIFF images were analyzed using Imagene (Biodiscovery, El Segundo, CA). Data were normalized using Lowess to adjust for intensity-dependent dye bias after removing spots of poor quality or low expression and subtracting local background (Yang et al., 2002). Data from each channel were transformed to log₂ values and combined into ratios for each hybridization using GeneSight (Biodiscovery, El Segundo, CA). Data were analyzed for statistically significant differences in expression using significance analysis of microarrays (Tusher et al., 2001). To estimate the numbers of genes likely to be identified as significantly different by chance alone, 500 permutations of the measurements were implemented and the false discovery rate was adjusted to produce an acceptably low-level expected false positives (FDR of 10.1% for data from 5–week-old mice and 0.6% for data from 7-day-old mice). Functional annotation of gene sets was performed using the Database for Annotation, Visualization and Integrated Discovery 2.1 (DAVID 2.1, http://apps1.niaid.nih.gov/David) (Dennis et al., 2003; Hosack et al., 2003). A one-tailed Fisher's exact probability value was calculated to identify GO terms for which overrepresentation among differentially expressed genes was statistically significant, using a P-value <0.05. The detailed protocols, details of each array platform, and primary data from this study are available at the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo), under accession numbers GSE4052 (data from 7-day-old mice) and GSE4053 (data from 5-week-old mice).

Real-time quantitative RT-PCR

QPCR was used to quantify differences in Hr expression between +/+, Hr^N/+, and Hr^N/Hr^N mice and also to verify differential expression of genes identified from microarray analyses. For array validation, RNA samples from the same mice were used for both microarray analyses and QPCR verification. All QPCR reactions were performed using a SmartCycler thermalcycler (Cepheid, Sunnyvale, CA), and software supplied by the manufacturer was used to identify threshold cycle (C_t) values. Expression levels of Hr and chaperonin subunit 7 were assayed using Assays-on-Demand™ predesigned gene-specific primer/TaqMan probe sets and TaqMan Universal Master Mix (Applied Biosystems, Foster City, CA), following the manufacturer's protocol. All other QCPR assays were performed using SYBR green I (Invitrogen, Carlsbad, CA) detection chemistry and custom primers designed using the Primer3 website (Rozen and Skaletsky, 2000) after reaction optimization. Primers were designed to span at least two exons and to produce amplicons 75–150 bp in size. Genes of interest were normalized against relative expression of 18S rRNA and data were analyzed using Excel.

Histology

Hematoxylin and eosin staining was done according to standard protocols (Prophet, 1992). Von Kossa staining was used to assess mineralization of dermal cysts according to a standard protocol (Klement et al., 2005). Actively replicating cells were selectively highlighted within tissue samples by immunolabeling with a rabbit anti-human antibody to the Ki67 antigen (NCL-Ki67p, Novocastra, Newcastle upon Tyne, UK) using standard protocols.

SEM of the hair fibers and skin

SEM was used to evaluate the skin surface and quality of hair fibers in areas of alopecia. Dorsal skin was fixed overnight in 2.5% glutaraldehyde in 0.1 M phosphate-buffered saline at 4°C. Samples were processed, dried, sputter coated with gold, and examined at 20 kV in a JOEL model S-3000N SEM (Hitachi, Japan) as described previously (Bechtold, 2000).

Conflict of Interest

The authors declare no conflict of interest.

References

1. Andl T, Ahn K, Kairo A, Chu EY, Wine-Lee L, Reddy ST et al. (2004) Epithelial Bmpr1a regulates differentiation and proliferation in postnatal hair follicles and is essential for tooth development. Development 131:2257–2268 | Article | PubMed | ISI | ChemPort |
2. Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ (1990) Basic local alignment search tool. J Mol Biol 215:403–410 | Article | PubMed | ISI | ChemPort |
3. Aoki N, Ito K, Ito M (2000) mu-Crystallin, thyroid hormone-binding protein, is expressed abundantly in the murine inner root sheath cells. J Invest Dermatol 115:402–405 | Article | PubMed | ChemPort |
4. Argenziano G, Sammarco E, Rossi A, Delfino M, Calvieri S (1999) Marie Unna hereditary hypotrichosis. Eur J Dermatol 9:278–280 | PubMed | ChemPort |
5. Beaudoin GM III, Sisk JM, Coulombe PA, Thompson CC (2005) Hairless triggers reactivation of hair growth by promoting Wnt signaling. Proc Natl Acad Sci USA 102:14653–14658 | Article | PubMed | ChemPort |
6. Bechtold LS (2000) Ultrastructural evaluation of mouse mutations. In: Systematic Characterization of Mouse Mutations. (Boggess D, Sundberg JP, ed). Boca Raton, FL: CRC Press, pp 121–129
7. Botchkareva NV, Botchkarev VA, Welker P, Airaksinen M, Roth W, Suvanto P et al. (2000) New roles for glial cell line-derived neurotrophic factor and neurturin: involvement in hair cycle control. Am J Pathol 156:1041–1053 | PubMed | ISI | ChemPort |
8. Brooke HC (1926) Hairless mice. J Hered 17:173–174
9. Bultman SJ, Michaud EJ, Woychik RP (1992) Molecular characterization of the mouse agouti locus. Cell 71:1195–1204 | Article | PubMed | ISI | ChemPort |
10. Cichon S, Kruse R, Hillmer AM, Kukuk G, Anker M, Altland K et al. (2000) A distinct gene close to the hairless locus on chromosome 8p underlies hereditary Marie Unna type hypotrichosis in a German family. Br J Dermatol 143:811–814 | Article | PubMed | ChemPort |
11. Darden DL, Hu FZ, Ehrlich MD, Gorry MC, Dressman D, Li HS et al. (2000) RNA differential display of scarless wound healing in fetal rabbit indicates downregulation of a CCT chaperonin subunit and upregulation of a glycophorin-like gene transcript. J Pediatr Surg 35:406–419 | Article | PubMed | ChemPort |
12. Dennis G Jr, Sherman BT, Hosack DA, Yang J, Gao W, Lane HC et al. (2003) DAVID: database for annotation, visualization, and integrated discovery. Genome Biol 4:3 | Article |
13. Flanagan SP (1966) "Nude", a new hairless gene with pleiotropic effects in the mouse. Genet Res 8:295–309 | PubMed | ISI | ChemPort |
14. Godwin AR, Capecchi MR (1998) Hoxc13 mutant mice lack external hair. Genes Dev 12:11–20 | PubMed | ISI | ChemPort |
15. Green J, Fitzpatrick E, de Berker D, Forrest SM, Sinclair RD (2003) Progressive patterned scalp hypotrichosis, with wiry hair, onycholysis, and intermittently associated cleft lip and palate: clinical and genetic distinction from Marie Unna. J Investig Dermatol Symp Proc 8:121–125 | Article | PubMed |
16. Hegde P, Qi R, Abernathy K, Gay C, Dharap S, Gaspard R et al. (2000) A concise guide to cDNA microarray analysis. Biotechniques 29:548–550, 552–4, 556 | PubMed | ISI | ChemPort |
17. Hosack DA, Dennis G Jr, Sherman BT, Lane HC, Lempicki RA (2003) Identifying biological themes within lists of genes with EASE. Genome Biol 4:R70 | Article | PubMed |
18. Howard A (1940) "Rhino", an allele of hairless in the house mouse. J Hered 31:467–470
19. Hsieh JC, Sisk JM, Jurutka PW, Haussler CA, Slater SA, Haussler MR et al. (2003) Physical and functional interaction between the vitamin D receptor and hairless corepressor, two proteins required for hair cycling. J Biol Chem 278:38665–38674 | Article | PubMed | ISI | ChemPort |
20. International Psoriasis Genetics Consortium (2003) The International Psoriasis Genetics Study: assessing linkage to 14 candidate susceptibility loci in a cohort of 942 affected sib pairs. Am J Hum Genet 73:430–437 | Article |
21. Jave-Suarez LF, Winter H, Langbein L, Rogers MA, Schweizer J (2002) HOXC13 is involved in the regulation of human hair keratin gene expression. J Biol Chem 277:3718–3726 | Article | PubMed | ISI | ChemPort |
22. Kim HS, Kim SC, Lee WS (2001) Marie Unna hypotrichosis in an Asian family. J Dermatol 28:149–152 | PubMed | ChemPort |
23. Kizawa K, Ito M (2005) Characterization of epithelial cells in the hair follicle with S100 proteins. Methods Mol Biol 289:209–222 | PubMed |
24. Klement JF, Matsuzaki Y, Jiang QJ, Terlizzi J, Choi HY, Fujimoto N et al. (2005) Targeted ablation of the abcc6 gene results in ectopic mineralization of connective tissues. Mol Cell Biol 25:8299–8310 | Article | PubMed | ISI | ChemPort |
25. Kljuic A, Bazzi H, Sundberg JP, Martinez-Mir A, O'Shaughnessy R, Mahoney MG et al. (2003) Desmoglein 4 in hair follicle differentiation and epidermal adhesion: evidence from inherited hypotrichosis and acquired pemphigus vulgaris. Cell 113:249–260 | Article | PubMed | ISI | ChemPort |
26. Kobielak K, Pasolli HA, Alonso L, Polak L, Fuchs E (2003) Defining BMP functions in the hair follicle by conditional ablation of BMP receptor IA. J Cell Biol 163:609–623 | Article | PubMed | ISI | ChemPort |
27. Kopf-Maier P, Mboneko VF, Merker HJ (1990) Nude mice are not hairless. A morphological study. Acta Anat (Basel) 139:178–190 | PubMed | ChemPort |
28. Lefevre P, Rochat A, Bodemer C, Vabres P, Barrandon Y, de Prost Y et al. (2000) Linkage of Marie-Unna hypotrichosis locus to chromosome 8p21 and exclusion of 10 genes including the hairless gene by mutation analysis. Eur J Hum Genet 8:273–279 | Article | PubMed | ISI | ChemPort |
29. Mecklenburg L, Nakamura M, Sundberg JP, Paus R (2001) The nude mouse skin phenotype: the role of Foxn1 in hair follicle development and cycling. Exp Mol Pathol 71:171–178 | Article | PubMed | ISI | ChemPort |
30. Meier N, Dear TN, Boehm T (1999) Whn and mHa3 are components of the genetic hierarchy controlling hair follicle differentiation. Mech Dev 89:215–221 | Article | PubMed | ISI | ChemPort |
31. Mischke D, Korge BP, Marenholz I, Volz A, Ziegler A (1996) Genes encoding structural proteins of epidermal cornification and S100 calcium-binding proteins form a gene complex ("epidermal differentiation complex") on human chromosome 1q21. J Invest Dermatol 106:989–992 | Article | PubMed | ISI | ChemPort |
32. Moraitis AN, Giguere V, Thompson CC (2002) Novel mechanism of nuclear receptor corepressor interaction dictated by activation function 2 helix determinants. Mol Cell Biol 22:6831–6841 | Article | PubMed | ISI | ChemPort |
33. Panteleyev AA, Paus R, Ahmad W, Sundberg JP, Christiano AM (1998a) Molecular and functional aspects of the hairless (hr) gene in laboratory rodents and humans. Exp Dermatol 7:249–267 | Article | PubMed | ISI | ChemPort |
34. Panteleyev AA, van der Veen C, Rosenbach T, Muller-Rover S, Sokolov VE, Paus R (1998b) Towards defining the pathogenesis of the hairless phenotype. J Invest Dermatol 110:902–907 | Article | PubMed | ISI | ChemPort |
35. Potter GB, Beaudoin GM III, DeRenzo CL, Zarach JM, Chen SH, Thompson CC (2001) The hairless gene mutated in congenital hair loss disorders encodes a novel nuclear receptor corepressor. Genes Dev 15:2687–2701 | Article | PubMed | ISI | ChemPort |
36. Potter GB, Zarach JM, Sisk JM, Thompson CC (2002) The thyroid hormone-regulated corepressor hairless associates with histone deacetylases in neonatal rat brain. Mol Endocrinol 16:2547–2560 | Article | PubMed | ChemPort |
37. Prophet E (1992) AFIP Laboratory Methods in Histotechnology. Washinton, DC: American Registry of Pathology
38. Rigdon RH, Packchanian AA (1974) Histologic study of the skin of congenitally athymic "nude" mice. Tex Rep Biol Med 32:711–723 | PubMed | ChemPort |
39. Rozen S, Skaletsky H (2000) Primer3 on the WWW for general users and for biologist programmers. Methods Mol Biol 132:365–386 | PubMed | ChemPort |
40. Schlake T, Schorpp M, Maul-Pavicic A, Malashenko AM, Boehm T (2000) Forkhead/winged-helix transcription factor Whn regulates hair keratin gene expression: molecular analysis of the nude skin phenotype. Dev Dyn 217:368–376 | Article | PubMed | ISI | ChemPort |
41. Stelzner KF (1983) Four dominant autosomal mutations affecting skin and hair development in the mouse. J Hered 74:193–196 | PubMed | ChemPort |
42. Sundberg J (1994) In: Handbook of Mouse Mutations with Skin and Hair Abnormalities: Animal Models and Biomedical Tools. (Sundberg J, ed). Boca Raton, FL: CRC Press, pp 291–312
43. Sundberg JP, Boggess D, Bascom C, Limberg BJ, Shultz LD, Sundberg BA et al. (2000) Lanceolate hair-J (lahJ): a mouse model for human hair disorders. Exp Dermatol 9:206–218 | Article | PubMed | ISI | ChemPort |
44. Sundberg JP, Price VH, King LE Jr (1999) The "hairless" gene in mouse and man. Arch Dermatol 135:718–720 | Article | PubMed | ISI | ChemPort |
45. Thompson CC, Bottcher MC (1997) The product of a thyroid hormone-responsive gene interacts with thyroid hormone receptors. Proc Natl Acad Sci USA 94:8527–8532 | Article | PubMed | ChemPort |
46. Tkatchenko AV, Visconti RP, Shang L, Papenbrock T, Pruett ND, Ito T et al. (2001) Overexpression of Hoxc13 in differentiating keratinocytes results in downregulation of a novel hair keratin gene cluster and alopecia. Development 128:1547–1558 | PubMed | ISI | ChemPort |
47. Tusher VG, Tibshirani R, Chu G (2001) Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci USA 98:5116–5121 | Article | PubMed | ChemPort |
48. van Steensel M, Smith FJ, Steijlen PM, Kluijt I, Stevens HP, Messenger A et al. (1999) The gene for hypotrichosis of Marie Unna maps between D8S258 and D8S298: exclusion of the hr gene by cDNA and genomic sequencing. Am J Hum Genet 65:413–419 | Article | PubMed | ISI | ChemPort |
49. Xie Z, Chang S, Oda Y, Bikle DD (2006) Hairless suppresses vitamin D receptor transactivation in human keratinocytes. Endocrinology 147:314–323 | PubMed | ChemPort |
50. Yang YH, Dudoit S, Luu P, Lin DM, Peng V, Ngai J et al. (2002) Normalization for cDNA microarray data: a robust composite method addressing single and multiple slide systematic variation. Nucleic Acids Res 30:e15 | Article | PubMed |
51. Zarach JM, Beaudoin GM III, Coulombe PA, Thompson CC (2004) The co-repressor hairless has a role in epithelial cell differentiation in the skin. Development 131:4189–4200 | Article | PubMed | ISI | ChemPort |

Acknowledgments

We thank Dr JH Kim for critical review of this manuscript. We also thank Dr P Hoyt for his help in fabricating cDNA microarrays, Dr LE King (Vanderbilt University) for insightful comments on the pathogenesis of hair loss in Hrⁿ mice, and Dr J Dunlap (University of Tennessee, Knoxville) for his gracious help with scanning electron microscopy. This work was supported by the Office of Biological and Environmental Research, US Department of Energy, under contract DE-AC05-00OR22725 with UT-Battelle, and by grants from the National Institutes of Health (RR00173).

SUPPLEMENTARY MATERIAL

Table S1. Complete list of genes differentially expressed in 7-day-old Hr^N/Hr^N mice compared with +/+ littermates.

Table S2. Complete list of significantly enriched GO categories at all levels of the ontologies biological process (BP), cellular component (CC), and molecular function (MF) across genes differentially expressed in skin of 7-day-old Hr^N/Hr^N mice.

Table S3. Highly represented GO categories (BP level 5) among genes differentially expressed in skin of 7-day-old Hr^N/Hr^N mice.

Table S4. Genes annotated with the GO term regulation of transcription, DNA-dependent that were differentially expressed in 7-day-old Hr^N/Hr^N mice.