Gene expression profile of the skin in the 'hairpoor' (HrHp) mice by microarray analysis

Background The transcriptional cofactor, Hairless (HR), acts as one of the key regulators of hair follicle cycling; the loss of function mutations is the cause of the expression of the hairless phenotype in humans and mice. Recently, we reported a new Hr mutant mouse called 'Hairpoor' (HrHp). These mutants harbor a gain of the function mutation, T403A, in the Hr gene. This confers the overexpression of HR and HrHp is an animal model of Marie Unna hereditary hypotrichosis in humans. In the present study, the expression profile of HrHp/HrHp skin was investigated using microarray analysis to identify genes whose expression was affected by the overexpression of HR. Results From 45,282 mouse probes, differential expressions in 43 (>2-fold), 306 (>1.5-fold), and 1861 genes (>1.2-fold) in skin from HrHp/HrHp mice were discovered and compared with skin from wild-type mice. Among the 1861 genes with a > 1.2-fold increase in expression, further analysis showed that the expression of eight genes known to have a close relationship with hair follicle development, ascertained by conducting real-time PCR on skin RNA produced during hair follicle morphogenesis (P0-P14), indicated that four genes, Wif1, Casp14, Krt71, and Sfrp1, showed a consistent expression pattern with respect to HR overexpression in vivo. Conclusion Wif1 and Casp14 were found to be upregulated, whereas Krt71 and Sfrp1 were downregulated in cells overexpressing HR in transient transfection experiments on keratinocytes, suggesting that HR may transcriptionally regulate these genes. Further studies are required to understand the mechanism of this regulation by the HR cofactor.


Background
With a complex and dynamic structure, hair is generated by hair-producing follicles and has a patterned cycle of growth and remodeling, which consists of growth (anagen), regression (catagen), and rest (telogen) stages. There are many genes involved in mature hair follicle (HF) regulation [1].
One of these genes, hairless (Hr), is expressed in skin, specifically in the suprabasal cell layer of the interfollicular epidermis and in the lower portion of the HF epithelium; its expression is dependent on the hair cycle. Hr encodes a 130 kDa protein (HR), which contains a zinc finger domain and is localized in the nucleus [2], and acts as a transcriptional corepressor that regulates transcription through directly binding to the thyroid hormone receptor [3,4], vitamin D receptor [5], and retinoic acid-like orphan receptor α [6].
Various Hr mutant mice have been studied to understand the function of HR, and most Hr mutant mice are created by causing the loss of HR function in their cells, giving them a typical phenotype with a recessive inheritance mode [7][8][9][10][11][12][13][14]. Microarray analysis of the skin of Hr tm1Cct /Hr tm1Cct mice has revealed that loss of HR function results in specific changes in the expression of epidermal differentiation-associated genes such as kera-tin10, loricrin, filaggrin, keratinocyte differentiationassociated protein (Kdap), and calmodulin 4, among other such mouse genes [7]. These results suggest that HR also plays a role in keratinocyte terminal differentiation through regulation of gene transcription.
The new Hr mutant mouse called 'hairpoor' (Hr Hp ) that we reported recently has a phenotype that is inherited in an autosomal semidominant manner [15]. Therefore, the heterozygote shows poor hair distribution, whereas the homozygote displays total alopecia. The Hr Hp mouse genome harbors a T-to-A substitution at position 403 in the noncoding exon 2 of Hr, and this mutation confers overexpression of HR in the mutant mouse skin [15]. This clearly distinguishes Hr Hp mice from other Hr mutant mice with loss of function of Hr.
Marie Unna hereditary hypotrichosis (MUHH; OMIM-146550) is an autosomal dominant disorder that displays coarse and twisted hair development in early age in humans and progresses to alopecia as the patients grow older. Recently, the genetic cause of MUHH was found to be similar to the mutation found in Hr Hp , namely a mutation in the 5' UTR of the HR gene [16,17]. This makes the Hr Hp mouse one of the valuable animal models for MUHH, and studying Hr Hp mice will facilitate the understanding of MUHH pathogenesis. Hr N mutant is another model for MUHH [15,18]. We recently reported that overexpression of HR is associated with alterations in the morphology and expression of a number of genes in the skin of the Hr Hp mouse [16]. In the present study, we performed microarray analysis and compared Hr Hp /Hr Hp skin with that of age-matched wild-type mice to identify the genes whose expression was affected by the overexpression of HR. We catalogued the genes showing differential expression in the mutant skin and found some of them to be tightly regulated by HR, which we confirmed using a reporter expression system.

Results
Hr overexpression preceded histological changes in the skin of Hr Hp /Hr Hp mice Although we have shown that the overexpression of HR causes a number of morphological alterations in the skin, we know little about the molecular basis of these changes. Because the HR protein functions as a transcriptional co-repressor [4,5], we set out to identify the genes whose expression was specifically affected by HR overexpression using microarray analysis. To investigate the initial events underlying the morphological changes, we determined the time point at which morphological changes and Hr expression in the skin of Hr Hp /Hr Hp mice first occurred.
The fur of +/+ and Hr Hp /Hr Hp mice showed noticeable differences by P7, so we examined the histology of the skin at the earlier time points by hematoxylin and eosin staining. At E18.5 and P0, we did not find any differences in the skins of +/+ and Hr Hp /Hr Hp mice ( Figure  1A-D) However we observed that HFs were much shorter and did not grow deep into the subcutis layer in Hr Hp /Hr Hp skin than in the +/+ skin at P3. In addition, hair bulbs in the Hr Hp /Hr Hp mice displayed short and round shape compared to the extended orval shape of the hair bulbs in the wiltype mice ( Figure 1G). This observation clearly showed the morphological changes occurred in the mutant skin at P3.
Next, we compared HR protein expression in the skin of Hr Hp /Hr Hp mice with that of +/+ mice by western blot analysis. The expression of HR was increased in Hr Hp /Hr Hp skin compared with +/+ skin at P0 and P3, as shown in Figure 1H, indicating that HR was overexpressed in Hr Hp /Hr Hp skin.
Based on these results, we decided to perform microarray analysis on the skin of mice at P0 as HR was overexpressed without prominent morphological changes in the skin of Hr Hp /Hr Hp compared to that of +/+ mice at this stage.
In addition, we also analyzed microarray data using q-value. We found differential expression of 23 (>2-fold, q < 0.05), 90 (>1.5-fold, q < 0.05), and 283 genes (>1.2fold, q < 0.05) in the skin of Hr Hp /Hr Hp mice compared with the skin of +/+ mice at P0. Fewer genes were found to be differentially expressed based on the q-values than on the p-values. While some of the genes with higher fold change in expression were found to be not differently expressed (Cox7a1 and Hbb-b1), many genes such as Cidea, Cyp2g1, Krt71 Sepina3h, Hmgcs2 and Odc1 showed significant differential expression with significant q-values ( Table 2). All of 283 genes are listed in Additional file 1 and 2.
Using the database of Kyoto Encyclopedia of genes and genomes (KEGG), 306 genes (>1.5-fold, p < 0.05), which were identified to be affected by overexpression of HR in Hr Hp /Hr Hp mice, were found to be involved in the biological pathways related to cell-cell signaling and communication, various cancers, metabolism, and regulation of the actin cytoskeleton (Table 3). These results suggested that pathways involved in the communication and proliferation of cells were affected by HR overexpression.

Expression pattern of HF associated genes during HF morphogenesis
As the first step to delineating the function of the genes whose expression was affected by HR overexpression, we assessed genes directly associated with HF development. Because few genes were directly associated with HF morphogenesis and/or development in the 43 (>2fold) or 306 (>1.5-fold) genes showing differential expression, we broadened our search to include the 1861 genes (>1.2-fold). Among those genes, we focused on the Wnt signaling pathway-associated genes (Sfrp1, Wif1, Wnt7b), caspase-14 (Casp14), Janus kinase-2 (Jak2), keratins (Krt71, Krt15) and fibroblast growth factor 10 (Fgf10) ( Table 4). Wnt signaling is not only involved in HF development [19,20] but also in HF cycling [21]. HR is reported to play a role in these processes [15,21]. HF undergoes vast apoptosis during catagen; Casp14 [22] and Jak2 [23] play a role in apoptosis, and in addition, Casp14 is directly associated with epidermal cell differentiation [24]. Krt71 and Krt15 are some of the main constituents of structures that grow from the skin. Krt71 is expressed in the inner root sheath (IRS), specifically in Henle's and Huxley's layers [25], and Krt15 is expressed in the basal layer of the outer root sheath [26]. Fgf10 is expressed in dermal papilla, the outer root sheath, and keratinocytes [27].
We validated the microarray analysis data using realtime PCR. This was carried out using gene-specific primers and the same RNA sources used for the microarray analysis. Results found by real-time PCR corroborated those from the microarray analysis, as shown in Table 4. Six upregulated genes, Wif1, Wnt7b, Casp14, Jak2, Krt15, and Fgf10, showed similar or higher upregulation by real-time PCR than microarray analysis. Two downregulated genes, Sfrp1 and Krt71, showed a similar fold reduction in expression with both measurement techniques.
To analyze the potential role of these genes in HF morphogenesis, we investigated their expression during early HF morphogenesis (P0-P14) by comparing their expression levels in the skin of mutant mice with those of wild-type mice. Further real-time PCR analysis revealed that the downregulated genes, Krt71 and Sfrp1, Figure 2 Gene expression profile in Hr Hp /Hr Hp skin using microarray. (A) Hierarchical clustering represents differential expression of the genes between +/+ and Hr Hp /Hr Hp skins. (B) Using DEG finding criteria, we found differential expression in 43 (>2-fold) and 306 genes (>1.5-fold) in the skin from Hr Hp /Hr Hp mice compared with the skin from wild type mice.  On the other hand, there were two subclasses of the upregulated genes; one group displayed a consistent expression pattern, whereas the other group showed an inconsistent pattern with respect to the HR expression. The relative expression levels of Wnt7b, Krt15, Jak2, and Fgf10 did not show a consistent pattern with respect to the HR expression in Hr Hp /Hr Hp mice. In contrast, the levels of Wif1 and Casp14 mRNA gradually increased over time. Thus, by P14, the relative expression levels of Wif1 and Casp14 in Hr Hp /Hr Hp skin were 5.77-and 5.35-fold higher than those of +/+ skin, respectively. This continuous increase in expression was consistent with the HR overexpression pattern in Hr Hp / Hr Hp mice (Figure 3) [16].
Expression of Wif1, Casp14, Sfrp1, and Krt71 in keratinocytes To further analyze whether the expressions of Wif1, Sfrp1, Casp14, and Krt71 were directly regulated by HR, we investigated changes in expression of these genes in the presence of overexpressed HR in a transient expression system using the mouse keratinocyte cell line, PAM212. RT-PCR revealed all the genes normally expressed in PAM212 cells ( Figure 4A); transfection of PAM212 cells with Hr cDNA construct resulted in expression of HR ( Figure 4B). The expression of all four genes was affected by the presence of HR: expression of Wif1 and Casp14 was increased 1.85-and 1.57-fold in HR-overexpressed cells compared to the mocktransfected PAM212 cells, respectively. In contrast, the relative expression of Sfrp1 and Krt71 was decreased to 0.61-and 0.52-fold ( Figure 4C) in HR-expressing cells compared with control cells. These results were consistent with their expression pattern in vivo and strongly suggested that HR may directly regulate expression of these genes.

Discussion
Recently, we reported the Hr Hp mouse generated by N-ethyl-N-nitrosourea mutagenesis as an animal model of human MUHH [15]. As an initial step to delineate the molecular basis of the underlying mechanism for the Hr Hp phenotype, we investigated the differential expression of genes in the skin immediately before morphological changes occurred in the Hr Hp /Hr Hp mouse.  Microarray analysis revealed that various biological pathways and the expression of many genes were affected by overexpression of HR.
Other studies have reported systematic screening of differentially expressed genes in the skin of mice with the Hr mutation, including analyses of gene expression in the skin of Hr N mice, another Hr-overexpressing mutant, and Hr tm1Cct , an Hr-loss-of-function mutant [7,18]. Hr N mutants harbor the mutation A402G, which abolishes the same uATG as in Hr Hp mutants, indicating that they have an identical defect [16]. A comparison of our microarray analysis results with those reported for the Hr N mutant did not show that the same genes displayed differential expression. This may be due to the difference in the developmental stage of the HF and epidermis (P0 vs. P7 or 5 weeks after birth) used in these analyses. However, many keratin-associated protein genes were detected in both mutants. At P0 in Hr Hp / Hr Hp mice, the expression of Krtap6-2, Krtap16-7, and Krtap16-3 increased, whereas the expression of Krtap5-1 and Krt71 decreased. Similarly, Krtap6-3, Krtap8-2, Krtap14, Krt1-1, and Krt1-3 were downregulated in Hr N /Hr N mice at P7 [18]. These results suggest that HR-regulated genes are associated with keratinocyte differentiation and/or hair-shaft structure. Furthermore, changes in the expression of keratin10, Kdap, and many epidermal differentiation-associated genes were also detected in Hr tm1Cct /Hr tm1Cct mice at P12 [7]. Results suggest that mutation of Hr causes the abnormal expression of many keratin-associated genes during HF morphogenesis and result in disruption of normal hair formation [7].
Of four genes with consistent expression patterns with respect to HR overexpression, two genes, Sfrp1 and Wif1, belong to Wnt inhibitor families. Both inhibitors interfere with Wnt signaling transduction by binding directly to the WNT protein [28,29], but though they function in a similar fashion in the Wnt signaling pathway, they displayed completely different expression patterns in Hr Hp / Hr Hp skin compared with age-matched +/+ skin. Surprisingly, Sfrp1 was downregulated by overexpression of the HR protein, whereas Wif1 was upregulated. Although we cannot rule out the possibility that this result may be caused by different locations of Sfrp1 and Wif1 expression in the skin, this difference is more likely to result from the differential transcriptional regulation of these genes, as seen in keratinocyte cells with HR overexpression ( Figure 4). The Hr overexpression in keratinocyte cells results in suppression of Sfrp1 by 39% and activation of Wif1 by 85%, suggesting that these promoters respond to HR differently. It is not known whether HR functions as an activator for Wif1 transcription, and further study is required to understand its mechanism of action. Regulation of the Wnt pathway by HR through transcriptional regulation of Wise, Soggy and Sfrp2 is reported to be important for proper HF cycling. While expression of Wise and Soggy mRNAs were upregulated in the Hr tm1Cct /Hr tm1Cct skin. Their expression levels were reduced in the skin of HR over-expressing transgenic mouse (2-fold). Sfrp2 was also shown to be downregulated in the Hr Hp /Hr Hp skin. [16,21,30]. We may include two more Wnt inhibitors, Sfrp1 and Wif1, for being regulated by HR based on this work. Further study is required to understand their function(s) in HF development and/ or cycling.
Casp14 is another gene upregulated by overexpression of HR, and is expressed in IRS and corneous cells of the outer root sheath in HFs [31]. It plays a role in the formation of the stratum corneum and terminal differentiation of keratinocytes [22,31]. The higher levels of Casp14mRNA suggest that an increase in terminal differentiation of keratinocytes occurs in the mutant skin. This result is comparable with our observation that the epidermis of the mutant skin shows increased differentiation in three-week-old mutant mice compared with age-matched wild-type mice [16]. Interestingly, Casp14 is also highly expressed in Hr tm1Cct /Hr tm1Cct mice, which display the hairless phenotype and show an increase in terminal differentiation of the epidermis with loss of function of Hr [7]. Furthermore its striking up-regualtion is age specific, which closely resembles its expression pattern in the Hr Hp /Hr Hp mice. Thus, despite having the opposite molecular defect, Hr Hp /Hr Hp and Hr tm1Cct /Hr tm1Cct mice exhibit increased expression of Casp14and displayed a similar increase in the terminal differentiation of keratinocytes, suggesting that, Casp14 may play an important role in proper differentiation of ketatinocyte. Thus, although it is not clear how HR regulates Casp14expression, both loss of Hr expression and Hr overexpression lead to alopecia through expression of Casp14 expression. Krt71 is also expressed in the IRS, specifically in Henle's and Huxley's layers; gene knockout mice show irregularly formed Henle's and Huxley's layers compared with the wild-type mice, indicating that Krt71 plays an important role in the formation of linear IRS intermediate filaments [25]. Results show that the downregulation of Krt71expression in the skin of Hr Hp /Hr Hp mice is comparable with that in Krt71 knockout mice in that IRSs in Hr Hp /Hr Hp mice were also abnormal.

Conclusion
Wif1 and Casp14 were found to be upregulated, whereas Krt71 and Sfrp1 were downregulated by overexpression of HR. These results suggest that HR may transcriptionally regulate these genes. Clearly, further studies are required to delineate the molecular mechanisms underlying how HR regulates its target genes and to elucidate the role of Hr in HF development, leading to a better understanding of MUHH.

Methods
Histological study of the skins of wild-type and Hr Hp /Hr Hp mice The skin was gathered from the buttocks of mice at E (Embryonic day)18.5, P0(Postnatal day 0) and P3 of wild-type (+/+) and Hr Hp /Hr Hp as previously described [15]. Paraffin sections were prepared, cut 5-μm in thickness and stained with hematoxylin and eosin (H&E), following a standard method. The histological morphology was observed with an optical microscope (Olympus).

Western blot analysis
Protein extracts were prepared from mouse skin at P0, and P3 in RIPA buffer (150 mM sodium chloride, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM Tris-HCl pH8.0) following a standard method. Bradford assay was performed to quantify protein amount. Two hundred micrograms of protein were used for western blot analysis as described earlier [16]. Rabbit polyclonal HR antibody [16] and αtubulin antibody (Santa Cruz) were used at a dilution of 1: 5000 and 1:2500, respectively. Antigen-antibody complexes were detected using ECL system (Amersham Bioscience) and exposure to x-ray film (Kodak).

Microarray hybridization and data analysis
Total RNA was extracted from the skins of wild type and Hr Hp /Hr Hp mice at P0 (N = 6) using TRIZOL following the manufacturer's instructions (Invitrogen). RNA that passed the quality check using an Agilent Bioanalyzer were used for microarray analysis using Mouse WG-6 v2.0 Expression BeadChip Kits (Illumina). Total RNA was reverse transcribed to cDNAs which were used to synthesize biotin labeled cRNA in an in vitro transcription reaction (Ambion). Two micrograms of biotin-labeled cRNA were loaded on to BeadChip and hybridization and washing experiments were carried out following the protocol from the manufacturer (Illumina). The slides were scanned using Illumina BeadArray Reader.

Gene pathway analysis
To analyze biological pathways associated with differentially expressed genes in the mutant skin, we input a total of 306 genes (>1.5 fold, p < 0.05) to the KEGG database http://www.genome.jp/kegg/pathway.html

Statistical analysis
The 'affy' and 'gcrma' packages of Bioconductor were used to preprocess and normalize the data following import of CEL files into the R statistical package (Affymetrix, Inc, Santa Clara, CA, USA). The GC Robust Multiarray Analysis (GC-RMA) was used to adjust perfect match (PM) probe data for background noise. Normalization was performed on adjusted perfect match (PM) data with an algorithm based on rank invariant probes [32].
After normalization, differential gene expression between groups was assessed by Significance Analysis of Microarrays (SAM) [33]. The t-test was calculated for statistical comparisons, and p-values were obtained with 100 permutations. The q-value, which is a Bayesian equivalent to the false discovery rate (FDR)-adjusted p-value, is estimated [34]. The q-value is a well suited measure of significance for the genomewide tests of significance.

RT-PCR and Realtime PCR
Total RNA was extracted from the skins of wild type and Hr Hp /Hr Hp mice at P0 (N = 6), P3 (N = 3), P7 (N = 3), P10 (N = 3) and P14 (N = 3) as described above. Single stranded cDNAs were synthesized using the PrimeScript 1 st strand cDNA Synthesis kit (Takara). PCR was performed using Peltier Thermal Cycler-100 (MJ Research). PCR conditions were 2 min at 95°C followed by 28 cycles of 15 sec at 94°C, 15 sec at 62°C, 15 sec at 72°C. The final extension was for 10 min at 72°C. Forward and reverse primer sequences of each gene are listed in Additional file 3. Realtime PCR was performed with SYBR Premix Ex Taq (Takara) using an Mx3000P (Stratagene). The cycling condition was initial denaturation for 2 min at 95°C followed by 45 cycles of 15 sec at 94°C, 15 sec at 62°C and 15 sec at 72°C. The gene expression levels were measured by the comparative ΔΔCt method [35], and the relative mRNA expression levels were determined based on the realtime PCR performed in duplicate using three independent samples. Statistical significance was determined by a student ttest using Sigma plot.

Cell culture and transient transfection experiment
Mouse keratinocyte cells (PAM212 cell line) were cultured in DMEM (Invirogen) containing 10% FBS with 5% CO 2 at 37°C incubator. An Hr full-length cDNA clone (BC049182) and pcDNA 3.1(+) vector were purchased from Invitrogen. Transfection was carried out using polyethyleneimine (Sigma-Aldrich) following the manufacturer's instruction. 8 x 10 5 cells were seeded in 60 mm dishes in triplicate. Twenty-four hours later, 3 μg of Hr cDNA construct and 0.2 μg of pCMV3.1/βgal were introduced into cells, which were harvested 48 hr post transfection and protein and total RNAs were extracted for western blot and realtime PCR analyses, respectively. Plasmid pcDNA 3.1 DNA was used as a control, and the relative expression level was normalized against transfection efficiency determined by β-Galactosidase activity.

Additional material
Additional file 1: Up-regulated genes in the skin of Hr Hp /Hr Hp at P0 compared with that of age matched wild type (>1.2-fold, p and q < 0.05).
Additional file 2: Down-regulated genes in the skin of Hr Hp /Hr Hp at P0 compared with that of age matched wild type (>1.2-fold, p and q < 0.05).
Additional file 3: List of gene-specific primers.