iRhom2 Mutation Leads to Aberrant Hair Follicle Differentiation in Mice

iRhom1 and iRhom2 are inactive homologues of rhomboid intramembrane serine proteases lacking essential catalytic residues, which are necessary for the maturation of TNFα-converting enzyme (TACE). In addition, iRhoms regulate epidermal growth factor family secretion. The functional significance of iRhom2 during mammalian development is largely unclear. We have identified a spontaneous single gene deletion mutation of iRhom2 in Uncv mice. The iRhom2Uncv/Uncv mice exhibit hairless phenotype in a BALB/c genetic background. In this study, we observed dysplasia hair follicles in iRhom2Uncv/Uncv mice from postnatal day 3. Further examination found decreased hair matrix proliferation and aberrant hair shaft and inner root sheath differentiation in iRhom2Uncv/Uncv mutant hair follicles. iRhom2 is required for the maturation of TACE. Our data demonstrate that iRhom2Uncv cannot induce the maturation of TACE in vitro and the level of mature TACE is also significantly reduced in the skin of iRhom2Uncv/Uncv mice. The activation of Notch1, a substrate of TACE, is disturbed, associated with dramatically down-regulation of Lef1 in iRhom2Uncv/Uncv hair follicle matrix. This study identifies iRhom2 as a novel regulator of hair shaft and inner root sheath differentiation.

Introduction iRhom1 and iRhom2, which are inactive homologues of the rhomboid intramembrane serine proteases that lack the essential catalytic residues [1]. iRhom1 and iRhom2 regulate the secretion of the epidermal growth factor (EGF) family by endoplasmic reticulum-associated degradation [2]. iRhom2 is required for the release of tumor necrosis factor a (TNFa) in macrophages by controlling the maturation of TNFa-converting enzyme (TACE, also called ADAM17) [3][4][5]. Moreover, iRhom2 controls the activation and substrate selectivity of TACEdependent shedding events [6]. Dominant mutations of iRhom2 is the cause of human tylosis esophageal cancer [7,8]. In addition, iRhom2 plays an important role in inflammatory arthritis [9]. iRhom2 knockout mice could survive in a lethal lipopolysaccharide dose [5]. However, the functional significance of iRhom2 during mammalian skin development is unclear.
In the developing hair follicle, signals from adjacent mesenchymal dermal papilla cells instruct the overlying epithelium to form hair placodes [10,11]. The placode proliferates to form a larger bulb (matrix) and further differentiates into a central hair shaft consisting of the medulla, cortex and hair shaft cuticle surrounded by the inner root sheath (IRS), which consists of the inner root sheath cuticle and Huxley's and Henle's layers. The outer root sheath (ORS) is outside the IRS, is contiguous with the interfollicular epidermis and contains a reservoir of quiescent SCs that are known as the bulge. The bone morphogenetic protein (BMP), Notch and Wnt/b-catenin signaling pathways allow the normal differentiation of matrix cells into the hair shaft and the IRS envelope [12][13][14][15]. Gata3 is expressed in IRS, the Gata3 mutant mice generate primary IRS defects, which lead to alterations in the shaft [16,17]. Foxn1 and Hoxc13, which are important hair shaft gene regulators, both of which cause hair defects when mutated [18,19].
Here, we report a role for iRhom2 in mouse skin development. In a BALB/c genetic background, homozygous uncovered (Uncv, MGI: 1261908) mice have a hairless phenotype [20,21]. We identified a spontaneous non-frameshift deletion mutation in the N-terminal cytoplasmic domain of of iRhom2 (iRhom2 Uncv ) in Uncv mice by sequence capture array and sequencing platform. iRhom2 Uncv could not induce the maturation of TACE, and dcresease the level of NICD and Lef1. This study suggests that iRhom2 regulates hair shaft and IRS differentiation by specifically modulating Notch1 and Wnt signaling pathway which maybe mediated by TACE.

Pattern of iRhom2 expression in the mouse
In a BALB/c genetic background, homozygous Uncv mice are hairless; heterozygous Uncv mice have a sparse hair coat ( Figure A in S1 Fig.). A previous study demonstrated that the Uncv mouse hair abnormalities were linked to a single autosomal gene mutation and incomplete dominant inheritance [20]. Uncv mice with heterozygous parents exhibited the expected Mendelian ratio (wildtype:heterozygous:homozygous 5105:191:103). These results show that Uncv/ Uncv mice only have a single gene mutation. Using a genetic linkage analysis of Uncv mice, the mutated gene was mapped to a region between markers D11mit338 and D11mit337 on mouse chromosome 11 [20,22]. We identified a 309 bp spontaneous non-frameshift deletion mutation in the N-terminal cytoplasmic domain of iRhom2 (iRhom2 Uncv ) in Uncv mice by sequence capture array and sequencing platform (Figure B in S1 Fig.). PCR analysis showed that each hairless mouse's genotype is iRhom2 Uncv/Uncv , each sparse mouse's genotype is iRhom2 Uncv/+ . The hair phenotype of mouse is depended on the dose of iRhom2. All the results indicated that iRhom2 mutation leads to the hairless in mice.
The pattern of iRhom2 mRNA expression was analyzed in 8-week-old mice. iRhom2 was expressed at high levels in the lung and spleen and at moderate levels in the skin (Fig. 1A). Next, we examined the expression of iRhom2 mRNA during skin development. iRhom2 was found to be specifically expressed at high levels on postnatal days (P) 2-15 and P28-35 (Fig. 1B), which corresponds well with the hair follicle growth phase (anagen). The expression of iRhom2 was detected at much lower levels in the hair follicle morphogenesis stage [embryonic day (E) 13.5-P0] and telogen stage (P21) (Fig. 1B). The cyclic expression pattern of iRhom2 implies that it maybe play a role in the progression of anagen in hair follicles. The strong ubiquitous cellular expression of iRhom2 is detected in the hair follicles (Fig. 1C-E and Figure B in S2 Fig.). Double immunofluorescence staining demonstrated that iRhom2 and K14 (marker for the ORS), AE13 (marker for the hair shaft cuticle and cortex keratins) or AE15 (marker for the the IRS and medulla of the hair shaft) were colocalized in the hair follicles ( Fig. 1C-E). iRhom2 was not expressed in dermal papilla cells (Fig. 1E). Moreover, iRhom2 was also expressed in basal layer of epidermis ( Figure C in S2 Fig.). There is a non-frameshift deletion mutation of iRhom2 in Uncv mice, the antibody we used could not discriminate iRhom2 and iRhom2 Uncv , so iRhom2 expression could still be detected in iRhom2 Uncv/Uncv mice ( Figure E in S2 Fig.). However, we performed real-time PCR analysis of iRhom2 and iRhom2 # (The PCR primers were located in the deletion region of iRhom2 Uncv ) in wild type and iRhom2 Uncv/Uncv mice, the results showed that iRhom2 # could not be detected in iRhom2 Uncv/Uncv mice ( Figure A in S2 Fig.). To confirm whether iRhom2 was expressed in the dermis, the expression of iRhom2 was separately assessed in the epidermis and dermis at P5 using real-time PCR. The results indicated that the dermis expresses very little iRhom2 (Fig. 1F).
Aberrant hair shaft and inner root sheath differentiation in iRhom2 Uncv/Uncv mice In mice, the development of the primary hair follicles is initiated at approximately E13 and extends to P16 [10]. To determine whether the deletion mutation in iRhom2 affects hair follicle morphogenesis, we examined the dorsal skin histology of iRhom2 Uncv/Uncv and wild-type mice at E15.5, E17.5, P0, P3 and P9. At hair follicle morphogenesis stage (E15.5, E17.5 and P0), the follicles of iRhom2 Uncv/Uncv were histologically similar to wild-type mice ( Fig. 2A-F). Skin follicle density was not significantly different in iRhom2 Uncv/Uncv mice compared to the wild-type at E17.5 (12¡2 compared to 12¡1 follicles per mm, respectively; P50.543, n58, Fig. 2S). At P3, the differentiating stage, iRhom2 Uncv/Uncv mice showed slightly shorter hair follicles ( Fig. 2G and H). However, the skin of iRhom2 Uncv/Uncv mice iRhom2 in Follicle Differentiation displayed striking defects in hair follicle morphology at later postnatal stages. By P9, the hair follicles of wild-type mice were in mid-anagen become fully differentiated with large hair bulbs that had descended deep into the fat layer of the skin. In contrast, the hair follicles of iRhom2 Uncv/Uncv mice were smaller and misshapen, and the majority of follicles failed to produce hair shafts ( Fig. 2I and J). Few apoptotic cells were detected by TUNEL staining in wildtype and iRhom2 Uncv/Uncv hair follicle matrix of P9 (2.94¡1.12 vs. 3.49¡1.49, P50.42, n58, Fig. 2K and L). At P18, hair follicles of both wildtype and mutant mice entered catagen [23] ( Figure A and B in S3 Fig.), and subsequently entered  follicle cycled normally and the marked shrinkage of the hair follicle matrix in mutant mice is not due to premature catagen development.
During follicle maturation, hair matrix keratinocytes rapidly propagate and differentiate, forming columns of cells that become the hair shaft and IRS [17]. To investigate hair follicular proliferation in iRhom2 Uncv/Uncv mice, we performed Ki67 immunofluorescence staining. In normal hair follicles, a large number of Ki67-positive cells was concentrated in the hair matrix at P0 and P3. A similar number of Ki67-positive cells was observed in iRhom2 Uncv/Uncv mouse follicles at P0 (21.13¡2.38 in wild-type follicles compared to 20.75¡3.19 in mutant follicles; P50.82, n58, Next, we examined the expression of several markers of differentiation in the hair follicle at P3 and P9. The ORS expresses keratin (K) 14. In iRhom2 Uncv/Uncv mouse follicles, K14 was expressed at normal levels ( Fig. 3A, B, I and J). K6 was expressed in the companion layer of the hair follicle [24]. The expression of K6 was increased in iRhom2 Uncv/Uncv mouse hair follicles (Fig. 3C, D, K and L). In contrast, AE15, which is normally expressed in the IRS and medulla of the hair shaft [25], was markedly reduced in the majority of iRhom2 Uncv/Uncv mouse follicles and was absent from the most distorted follicles, indicating defective IRS differentiation and an absence of the hair shaft medulla (Fig. 3E, F, M and N). AE13, a specific marker for the hair shaft cuticle and cortex keratins [26], was absent in the majority of iRhom2 Uncv/Uncv mouse follicles (Fig. 3G, H, O and P). We confirmed these results by real-time PCR, and the results showed that the expression of IRS markers (K71 and K72) and the hair shaft marker (K85) was significantly decreased in iRhom2 Uncv/Uncv mouse follicles and that the expression of K6 was increased (Fig. 3Q). These results indicate that the hair matrix cells failed to differentiate toward the hair shaft and the IRS in iRhom2 Uncv/Uncv mouse follicles, suggesting a crucial role for iRhom2 in hair follicle differentiation.
The iRhom2 Uncv mutant protein cannot induce the maturation of TACE iRhoms inhibits secretion of EGF family ligands by inducing degradation of epidermal growth factor receptor (EGFR) ligands in mammalian cells [2]. Therefore, we analyzed the expression of phospho-EGFR in P3 and P5 mouse skin. The expression of phospho-EGFR was similar in the iRhom2 Uncv/Uncv and the wild-type hair follicles (Fig. 4A-E), so was the expression of phospho-ERK (Fig. 4E). These data indicated that during hair follicle maturation the expression of phospho-EGFR and phospho-ERK were not altered in iRhom2 Uncv/Uncv mice. Furthermore, an in vitro EGF degradation assay showed that both wild-type iRhom2 and iRhom2 Uncv could promote EGF degradation (Fig. 4F). This iRhom2 in Follicle Differentiation degradation is dependent on the proteasome; the proteasome inhibitor MG132 could partly restore the expression of EGF (Fig. 4F). Therefore, the deleted region in the iRhom2 Uncv is not necessary for the iRhom2-mediated degradation of EGF.
It was reported that iRhom2 interacts with TACE and promotes its maturation in macrophages [3]. Immunoprecipitation of iRhom2-overexpressing 293T cells followed by immunoblotting revealed that iRhom2 Uncv did not affect it's interaction with TACE (Fig. 5D). We expressed wild-type iRhom2 and iRhom2 Uncv in 293T cells and assayed its effect on the maturation of TACE. Consistent with previous reports [3], the overexpression of wild-type iRhom2 caused the excessive maturation of TACE; however, exogenous iRhom2 Uncv could not induce the maturation of TACE compared with the empty vector (Fig. 5A). Moreover, exogenous iRhom2 Uncv does not significantly affect the maturation of TACE that is induced by exogenous wild-type iRhom2 (Fig. 5A). In western blots of iRhom2 Uncv/Uncv mouse skin, mature TACE was significantly reduced at P3, P5 and P9 (Fig. 5B). The mRNA of TACE was not significantly different between the iRhom2 Uncv/Uncv and wild-type mice at P5 (Fig. 5C). Double immunofluorescence staining revealed that TACE and iRhom2 were colocalized in the hair follicles of wild-type and iRhom2 Uncv/Uncv mice ( Figure D and E in S2 Fig.). These results indicate that iRhom2 Uncv cannot induce the maturation of TACE in vitro or in vivo.
Expression levels of regulators of inner root sheath and hair shaft differentiation in iRhom2 Uncv/Uncv follicles Notch1 and BMP signaling play critical roles in the differentiation of the hair shaft and the IRS [12,13,15,27]. Wnt/b-catenin signaling controls the differentiation of matrix cells along the hair shaft lineage [14]. TACE participates in the activation of the Notch pathway [28][29][30][31]. The expression of Notch intracellular  iRhom2 in Follicle Differentiation domain (NICD) was dramatically reduced in iRhom2 Uncv/Uncv mouse hair follicles from P3-P9 (Fig. 6A, B and K); however, the expression of Notch1 mRNA was not significantly different from the wild-type (Fig. 6N), indicating that the posttranslational maturation of the Notch1 protein was inhibited in iRhom2 Uncv/Uncv mouse skin. The Notch1 target genes Hes1, Hey1 and Hey2 [27] were also downregulated in iRhom2 Uncv/Uncv mouse skin (Fig. 6N). To determine whether iRhom2 affects Notch signaling transcriptional activity, HeLa cells were cotransfected with Notch-dependent CSL luciferase reporter (CSL-Luc) and increasing amounts of Flag-iRhom2 or Flag-iRhom2 Uncv mutation. As shown in Fig. 6L, Notch transcriptional activation activated by iRhom2 in a dosedependent manner, but iRhom2 Uncv was not able to activate Notch transcriptional activation. It was reported that expression of Lef1, the Wnt/b-catenin pathway transcriptional effector, could be regulated by Notch1 [32,33]. The expression of Lef1 was similar at E17.5 (Fig. 6C, D and M), dramatically reduced at P3 and P5 (Fig. 6E-H and M), and absent at P9 in iRhom2 Uncv/Uncv mouse follicles compared to wild-type follicles (Fig. 6I, J and M). The expression of Foxn1, Gata3 and Hoxc13, which are implicated in hair follicle differetionation [34], were dramatically reduced in iRhom2 Uncv/Uncv mouse follicles (Fig. 6N, Figure E and F in S4 Fig.). The nuclear localization of b-catenin, which complexes with Lef1 to activate the transcription of Wnt target genes, was normal in iRhom2 Uncv/Uncv mice ( Figure A and B in S4 Fig.). Other genes involved in the Wnt pathway, including Ctnnb1 (b-catenin), Tcf3, Dact1, Wif1 and Daam2 mRNA, were normally expressed in the dorsal skin of wild-type and iRhom2 Uncv/Uncv mice at P5 (Fig. 6N). The levels of phospho-Smad1/5/8 were comparable between iRhom2 Uncv/Uncv mouse follicles and the wild-type ( Figure C and D in S4 Fig.), as were the expression levels of Bmp2 and Bmp4 mRNA (Fig. 6N). These results indicated that iRhom2 is involved in hair follicle differentiation by controlling the Notch and Wnt/b-catenin signaling pathways.

Discussion
In this study, we discovered that iRhom2 regulate murine hair follicle differentiation by specifically modulating Notch1 and Wnt signaling pathway which maybe mediated by TACE.
Our study demonstrated that the iRhom2 Uncv/Uncv mouse has hair differentiation abnormalities and displays a hairless phenotype in a BALB/c genetic background. The iRhom2 Uncv/Uncv mouse showed decreased hair matrix proliferation and could not differentiated into IRS and hair shaft. In P9, the hair follicles of wild-type and mutant mice were in mid-anagen and had descended deep into the subcutis. However, the hair matrix of mutant follicles were less proliferative than that of wildtype. Comparable and few apoptotic cells were detected in both wildtype and iRhom2 Uncv/Uncv hair matrix of P9, implying that mutant mice did not entered into catagen stage prematurely. Further examination showed that hair follicle cycling of mutant mice was normal. So, the marked iRhom2 in Follicle Differentiation shrinkage of the hair follicle matrix in mutant mice is not due to premature catagen development. The iRhom2 Uncv/Uncv hair matrix failed to differentiate into IRS and hair shaft which resulted in the hairless phenotype of iRhom2 Uncv/Uncv mouse. However, the hair follicle phenotype was not reported in iRhom2 knockout C57BL/6 mouse [3,5]. We noticed that cub/cub mice display a hairless or a wavy-coated phenotype depending on the modifier gene mcub [35]. When homozygous for the recessive mcub allele, cub/cub mice appear hairless. A single copy of the dominant Mcub allele confers a full, curly coat to cub/cub mice. The hairless phenotype in cub/cub, mcub/mcub mice resemble the Uncv/Uncv mice. The mapping regions of cub and Uncv cover a large overlapping region including the iRhom2 gene. We speculate that cub and Uncv are the same gene, iRhom2. The modifier gene, Mcub or mcub, maybe affect hair follicle phenotype of iRhom2 knockout mouse.
iRhom2 is necessary for the maturation of TACE. The Adam17 knockout mice displayed a disorganized distribution and structure of hair follicles [36]. Consistent with phenotype in hair follicle of Adam17 knockout mice, the iRhom2 Uncv/Uncv mice showed irregularly positioned and oriented hair follicles with abnormal structure. Adam17 conditional knockout mice induced by K14-Cre showed delayed hair outgrowth, shortened and disorganized hair follicles, abnormal epidermal proliferation and there was a dramatically increased infiltration of inflammatory macrophages [37]. iRhom2 Uncv/Uncv mice showed defects in hair development and macrophages infiltration ( Figure A, B and C in S5 Fig.) resembling those in Adam17 conditional knockout mice. There are some disparity between iRhom2 Uncv/Uncv mice and Adam17 knockout mice. The majority of Adam17 knockout mice died between E17.5 and the first day after birth, those few mice that survived for several weeks had 20 to 40% body weights loss than those of littermates [36], while iRhom2 Uncv/Uncv mice could survive up to 12 months and had 20% body weights loss at 4 weeks after birth. It was reported that there is some redundancy between them iRhom1 and iRhom2 [38], this provides some explanation iRhom2 Uncv/Uncv mice do not show the several defects seen in Adam17 knockouts. We noticed that Adam17 knockouts have perturbed hair coats and curly vibrissae [36], however, iRhom2 Uncv/Uncv mice have hairless phenotype. This discrimination implies that iRhom2 has additional physiologically substrates other than Adam17.
iRhom2 is required for controlling the activation of TACE-dependent shedding events [6]. Consistently, iRhom2 Uncv cannot induce the maturation of TACE in vitro, and mature TACE was also significantly reduced in iRhom2 Uncv/Uncv mouse skin. Furthermore, iRhom2 Uncv does not affect the maturation of TACE induced by wild-type iRhom2. These findings indicated that iRhom2 Uncv is a loss of function with respect to the maturation of TACE. However, the deleted region in the iRhom2 Uncv mutant is not necessary for the degradation of EGF. So, the iRhom2 Uncv mutant is not a simple loss of function mutation, which selectivity affects the client protein.
Our results indicated that the deleted region in the iRhom2 Uncv is necessary for the maturation of TACE, which is required for the normal processing of Notch [28,29,39]. The Notch precursor protein is cleaved by furin to produce a bipartite heterodimeric molecule. Notch ligand-receptor interactions induce S2 and S3 proteolytic cleavage. S2 cleavage within the extracellular domain is mediated by TACE. Subsequently, S3 cleavage by the c-secretase releases NICD, which translocates to the nucleus [39,40]. We demonstrated that the iRhom2 Uncv mutation affects TACE activity and inhibits the activation of Notch1. It was reported that Notch and WNT signaling exists cross-talk [32,33,41]. Moreover, the Notch intracellular domain can function as a coactivator for Lef1 and regulator for expression of Lef1 [32,41]. We observed that the expression of NICD and Lef1 were dramatically reduced from P3, which is the initial stage of hair follicle differentiation and the numbers of hair matrix cells were comparable between the wild-type and the iRhom2 Uncv/Uncv mice, indicating that the loss of NICD and Lef1 is not due to loss of the matrix cells. Therefore, the reduction in NICD and Lef1 expression in the iRhom2 Uncv/Uncv mouse hair matrix appeared before the change in hair bulb morphology, which resulted in disorders of follicle differentiation. This study identifies that iRhom2 regulates hair shaft and IRS differentiation by specifically modulating Notch1 and Wnt signaling pathway which maybe mediated by TACE.

Mouse strains and genotyping
Uncv mice were maintained in a BALB/c background in a specific pathogen-free environment. Uncv mice were bred in a heterozygous mating heterozygous format. The genotyping primers were iRhom2 fwd, 59-CACAGCCCAGTGGTTTGGGGTCA -39, and iRhom2 rev, 59-GAGGGCGGCGGCTGCCTGAAAGCT -39. In wild-type mice, iRhom2 was amplified by standard PCR to yield a 469 bp fragment. In Uncv/Uncv mutant mice, iRhom2 was amplified to yield a 160 bp fragment. Wild-type mice littermates were used as controls. All animal studies were approved by the Review Board of the Institute of Radiation Medicine, Beijing, China.

Cell culture, transfection and immunoprecipitation
293T cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum and 2 mM L-glutamine at 37˚C in a humidified incubator with 5% CO 2 . iRhom2 plasmids and the pcDNA3.0-Flag iRhom2 in Follicle Differentiation vector were transfected into 293T cells using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA). The immunoprecipitation was performed as described previously [42].

Histological analysis, immunohistochemistry and immunofluorescence
Dorsal skin tissue samples were fixed in 4% paraformaldehyde at 4˚C overnight, embedded in paraffin, sectioned at 4 mm and stained with hematoxylin and eosin for histological analysis. Immunohistochemistry and immunofluorescence was performed as described previously [43,44]. For the iRhom2 antibody-blocking experiment, iRhom2 antibody was pre-mixed with iRhom2 antibody (Nterminus) blocking peptides (Abgent) before incubation with the sections.
Terminal deoxyribonucleotidyl transferase-mediated dUTP nickend labeling assay The terminal deoxynucleotidyl transferase dUTP nick-end labeling (TUNEL) assay was performed on 4-mm-thick sections of dorsal skin using the ApopTag Peroxidase In Situ Apoptosis Detection Kit (Roche) following the manufacturer's directions.

RNA isolation and real-time PCR
Total RNA was isolated from the dorsal skin using TRIzol reagent (Life Technologies) following the manufacturer's protocol, and cDNA was then prepared using the Prime Script RT reagent kit (TaKaRa) according to the manufacturer's instructions with oligo-dT primers. SYBR Premix Ex Taq (TaKaRa) was used for real-time quantification, and gene expression was normalized to GAPDH using the DDcycle threshold method. Primer sequences are available upon request. To assess the iRhom2 expression separately in the epidermis and dermis, dorsal skin was incubated in 0.25% trypsin at 4˚C for 24 to 48 hours; the epidermis was removed as soon as it could be separated from the dermis as an intact sheet.

Luciferase assays
HeLa cells were transfected with Notch-dependent CSL luciferase reporter containing CSL binding sites, Renilla luciferase reporter, iRhom2 and iRhom2 Uncv plasmids. Renilla luciferase was used as a transfection control and signals are given as fold Firefly/Renilla corrected for background. 48 h after transfection, luciferasereporters activity was measured. Datas are representative of at least three independent experiments.

Statistical analysis
All results are presented as the means ¡ SE. All statistical analyses were performed using the SPSS software. The significance of the differences between groups was determined using Student's t-test; P,0.05 was considered significant.