LIPH Expression in Skin and Hair Follicles of Normal Coat and Rex Rabbits

Natural mutations in the LIPH gene were shown to be responsible for hair growth defects in humans and for the rex short hair phenotype in rabbits. In this species, we identified a single nucleotide deletion in LIPH (1362delA) introducing a stop codon in the C-terminal region of the protein. We investigated the expression of LIPH between normal coat and rex rabbits during critical fetal stages of hair follicle genesis, in adults and during hair follicle cycles. Transcripts were three times less expressed in both fetal and adult stages of the rex rabbits than in normal rabbits. In addition, the hair growth cycle phases affected the regulation of the transcription level in the normal and mutant phenotypes differently. LIPH mRNA and protein levels were higher in the outer root sheath (ORS) than in the inner root sheath (IRS), with a very weak signal in the IRS of rex rabbits. In vitro transfection shows that the mutant protein has a reduced lipase activity compared to the wild type form. Our results contribute to the characterization of the LIPH mode of action and confirm the crucial role of LIPH in hair production.


Introduction
In rabbits, hair follicles are structured into groups, usually constituted of one central primary hair follicle surrounded by 2-4 lateral primary hair follicles and by 20-50 secondary down hair follicles ( Figure 1). These three types of hair follicles appear sequentially during fetal development and early after birth. At day 19 of gestation (the average gestation in rabbits lasts 30 days), the central primary hair follicles rise followed at day 25 by the primary lateral hair follicles. At day 29 of gestation, a secondary hair follicle for each of the 2 to 4 lateral hair follicles appears. Finally, secondary derived hair follicles emerging from the skin by the same hair channel, appears during the early childhood of the animals [1]. Normal rabbit fur is composed of three different types of hairs: guard hairs produced by central primary hair follicles (3-4 cm long for a diameter of 50-60 mm), awn hairs produced by lateral primary hair follicles (3-3.5 cm/25-30 mm) which both constitute the physical ''outer coat'' protection, and down hairs produced by secondary hair follicles (2.5-3 cm/15 mm), and the inner coat for thermal protection. Down hairs are the most abundant and represent about 90-95% of all hairs. In 1919, a mutant phenotype (rex) with soft short hairs was observed by a French breeder in a litter of European rabbits (Oryctolagus cuniculus) [2]. In the early nineteen-eighties, the rex trait (downs .95%) was improved by selection on an INRA experimental farm to further reduce the number of guard and awn hairs and is now commercialized as orylagH (downs .98% with a residual variability of downs percentage). The ''r1'' mutation was confirmed to segregate within INRA families as a monogenic, autosomal and recessive trait. Furthermore, a deletion of a single nucleotide in exon 9 of the LIPH gene (1362delA) was identified in rex rabbits [3]. This mutation results in a frameshift and introduces a premature stop codon shortening the predicted protein by 19 aminoacids. LIPH is a membrane-bound member of the mammalian triglyceride lipase family, the phosphatidic acidselective phospholipase A1 (PLA1). It specifically hydrolyzes phosphatidic acid (PA) to produce 2-acyl lysophosphatidic acid (LPA), which is a lipid mediator with diverse biological properties including stimulation of cell proliferation and motility [4]. Disruption of the LIPH gene in the mouse results in various phenotypes including retarded hair growth and postnatal lethality [5]. In 2006, a deletion in the LIPH gene was identified as being responsible for Hypotrichosis Simplex (HS) in 50 families of Russian people [6]. Hairs of these patients are abnormally short, dystrophic and fragile due to retarded or arrested hair growth. Mutations have also been reported in other exons for HS and for Autosomal Recessive Woolly Hair syndrome (ARWH) partly associated to HS [7][8][9][10][11][12][13][14]. It is known that the LIPH gene is expressed abundantly in human HF as detected by RT-PCR without precise localization [6]. Later, in situ hybridization studies of skin sections showed that LIPH mRNA is strongly expressed in the human precortex, hair shaft cuticle and Huxley layer of the inner root sheath (IRS) of the bulb portion and in a more prominent manner in the outer root sheath (ORS) in the upper portion of the HF [12]. The mPA-PLA1 alpha (LIPH) protein was found to be associated with the detergent-resistant membrane fraction of insect transfected cells [15] confirmed by confocal fluorescence microscopy [16]. In spite of its important role in the hair growth process, little is known about the way LIPH is implicated in the hair follicle function and its expression, localization and exact role are still a subject of speculations. In this study we analyzed the spaciotemporal expression of the LIPH gene in fetal and adult rabbit skin by quantitative PCR as well as in hair follicles by in situ hybridization and immunochemistry. The expression of LIPH mRNA and protein was compared for normal and rex phenotypes. In addition, the activity of both normal and mutant proteins was estimated in vitro using transfected mammalian cell cultures.

LIPH expression during fetal hair follicle development
The organization of the cell structures observed after Roan staining of the skin sections validated the expected timing of the formation of the different types of hair follicles (HF) in normal rabbits ( Figure 1, row B2). At day 19, the bulb of the central primary hair follicles appeared in both phenotypes with a higher meiotic activity of the external cells revealed by the red coloration. At day 25, the epidermis was well organized and looked very much the same in normal and orylagH skins. The central primary HF was well visible as well as primary lateral HF in formation from the epidermis layers by invagination ( Figure 1B1). It seemed that the meiotic activity was more intense in the parts that could be identified as the bulge and the bulb. At day 29, the group structure of the HF was visible with the apparition of the lateral primary HF in both phenotypes but with observable secondary HF in the normal rabbit not seen in the orylagH. Hair shafts were clearly identified in some of the central primary HF in both phenotypes ( Figure 1, row B1). There was no apparent noticeable hair follicle group density difference between normal and orylagH rabbits. The mRNA expression level estimated by quantitative RT-PCR was significantly different at the three different stages of gestation between the orylagH and the normal rabbit skins (H = 9.9; p = 0.0017). There was on average three times less expression in the orylagH fetal skin than in normal fetal rabbit skin ( Figure 1A). The presence of the mPA-PLA1 was detected by immunochemistry at all fetal stages. At day 19, it was present in the epidermis and in the center of the growing central primary HF in both phenotypes. At day 25 and day 29, it seemed absent from the lower part of the follicle bulb in normal and in orylagH skin sections.

LIPH expression in adult rabbits
In adults, we compared by qPCR the three genotypes at the LIPH locus as well as extreme phenotypes within the orylagH population homozygous for the LIPH deletion ( Figure 2). LIPH mRNA was detected by qPCR in homozygous and heterozygous wild type adults, rex and orylagH skin. The expression level of LIPH mRNA was found to be closely related to the genotype. The skin of rex rabbits expressed about three times less LIPH mRNA compared to that in the skin of normal animals (H = 8.2; p = 0.004). The skin of average orylagH rabbits expressed about seven times less LIPH mRNA compared to that of the skin of normal animals (H = 16.8; p = 4.10 25 ). In addition, within the orylagH phenotype, the rabbits with less guard hairs (orylagH+) had significantly less LIPH mRNA than those with more guard hairs (orylagH2) (H = 5.1; p = 0.02) (Figure 2A). The level in the heterozygous genotype is intermediary and significantly different to that of orylagH+ (H = 6.4; p = 0.01) and normal rabbits (H = 4.3; p = 0.03) with whom they were mated. The LIPH mRNA ( Figure 2C) and protein ( Figure 2C) were detected by in situ hybridization and immunohistochemistry respectively in skin sections of all genotypes and phenotypes. We observed a gradient of the LIPH mRNA and protein expression, which approximately agreed with those of the qPCR results, with the highest expression in normal rabbits and the lowest expression in the finest coats of orylagH, while the heterozygous and the rex were intermediate. It was not possible to differentiate between orylagH+ and orylagH2 since the signal was very weak in both cases. In normal coat rabbits, the LIPH mRNA and protein were detected in the IRS (inner root sheath) and ORS (outer root sheath) of the three hair follicle classes. For all genotypes and phenotypes, the mPA-PLA1 is less expressed in the IRS compared to the ORS. In orylagH, LIPH was only clearly detected in the ORS. LIPH expression during hair follicle cycle There was no difference for LIPH between the two cycle stages in the orylagH (H = 0.05; p = 0.8), on the contrary to heterozygous rabbits that present a significant higher expression level (more than twice) of LIPH mRNA at the anagen stage compared to the catagen/telogen stage (H = 5.2; p = 0.02). Whatever the stage, mRNA in orylagH was much less expressed than in heterozygous animals ( Figure 3A). There was no significant difference between LIPH levels at the catagen/telogen stage in normal rabbits and both values for orylagH (H = 0.02; p = 0.9). In situ hybridization indicated that the LIPH mRNA was weakly expressed in the orylagH and at the same low level in both stages. In heterozygous animals, the level of expression was much higher in the anagen compared to the catagen/telogen stage ( Figure 3B). Similarly, mPA-PLA1a was weakly expressed in the orylagH and there was no difference of expression between the two stages. On the contrary, the mPA-PLA1a was clearly more expressed in the anagen stage compared to the catagen/telogen stage in heterozygous animals. In these rabbits, the expression was of the same order of magnitude at the catagen/telogen stage than that of orylagH at either stages ( Figure 3C). These results show that LIPH mRNA and protein levels approximately agree with those of qPCR results.

mPA-PLA1a enzymatic activity in cell cultures
The activity of the 1362delA LIPH mutation on lipid metabolism was assessed by an in vitro assay as described by Pasternack [17] using PED-A1 as a substrate. A highly significant difference between wild-type mPA-PLA1a and 1362delA mPA-PLA1a activities was observed (H = 8.5; p = 0.003) ( Table 1). At 10 minutes, the 1362delA mPA-PLA1a activity was that of the background while the wild-type mPA-PLA1a already started converting the substrate by its phospholipase activity. After 30 minutes, the wild-type mPA-PLA1a activity was 1.5 higher than that of 1362delA mPA-PLA1a, a difference conserved after one hour of reaction.

Discussion
We first provide evidence showing that the 1362delA LIPH mutation [3] does not alter the histological structure of rabbit skin. Indeed, no gross abnormalities were observed after staining of skin cross sections, neither at the three fetal stages, nor in adults. Hair follicle structures were similar between orylagH and normal rabbits at all development stages. Likewise, no differences of hair follicle group density could be evidenced. These data suggest that LIPH does not play a major role in hair follicle formation and development. We then show that LIPH is expressed in three hair  follicle types, mainly in the ORS and to a lesser extent in the IRS. These findings raise the question of how this LIPH mutation affects mainly primary hair follicles producing guard hair. In orylagH rabbits, LIPH was only clearly detected in the ORS with no detectable signal in the IRS. This could mean that the orylagH selection suppressed the LIPH transcription in the IRS. However, due to the weaker LIPH signal in rex and orylagH follicles, we cannot rule out that we are at the detection limit of the method. Expression of the LIPH gene was also monitored by qPCR on depilated skin to compare the different genotypes at different development stages. We provide evidence that LIPH is regulated during hair follicle cycles. Expression levels decrease by a factor 3 in normal rabbits between anagen and catagen/telogen stages. Interestingly, the LIPH mRNA expression does not seem to be regulated during development in the orylagH strain and the expression level appears similar to that of normal rabbits at the catagen/telogen stage. This suggests that LIPH may be required for the migration and/or differentiation of keratinocytes into distinct epithelial hair lineages during the anagen phase, when hairs are growing [15].
OrylagH and rex rabbits exhibited a marked decrease of LIPH mRNA level compared to control rabbits at both fetal and adult stages, in agreement with in situ hybridization. Thus, the LIPH mutation induces a reduced expression of both mRNA and protein. One possible explanation could be the nonsense mediated decay (NMD) mechanism [18] but the LIPH 1362delA characteristics are not in agreement with NMD or STAU1-mediated decay [19] criteria. We also explored the possible involvement of regulatory elements. Interestingly, the 1362delA mutation leads to a perfect match with the seed sequence of hsa-miR-518c*, also described in mice (mmu-miR-1904) and primates (ppt-miR1031a and b). This miRNA was shown to be down-regulated in psoriatic skin [20]. This miRNA could play a major role in the decrease of LIPH mRNA expression in rex rabbits if expressed in rabbit skin. We cannot exclude that the 1362delA mutation could also affect the secondary mRNA structure leading to its degradation.
Interestingly, orylagH individuals conserve a variability of LIPH RNA levels, discriminating between best orylagH with a perfect fur (down exclusively, orylagH+) and low quality orylagH furs(orylagH2). These findings suggest that some particular features of the rex allele and/or some other regulatory genes may have been selected in orylagH animals. We thus investigated two regions located 500 bp and 3 kb downstream of the transcription start site predicted to contain binding sites for transcription factors involved in hair follicle development and maintenance [21][22][23]. No polymorphisms provoking changes in these regions could be identified among normal, rex and orylagH (data not shown). Other genes with more tenuous action on the hair growth process may also have been selected during the improvement of the rex trait to obtain the orylagH phenotype.
The decrease of the mutant mPA-PLA1a activity could be the result of the deletion, directly reducing the intrinsic activity of the protein or that the normally active protein does not localize at the right place at the cell membrane. Confocal microscopy should be able to clarify this point.
Here, we show that the rex mutation, not located in the lipase domain of mPA-PLA1a, does not totally disrupt the lipase activity. It thus seems that the remaining activity is sufficient to produce down hair and nanified coarse hairs.
Another noticeable point is that mPA-PLA1a, belonging to the pancreatic lipase family, needs a colipase to open and activate its catalytic domain. The C-terminal region contains a motif which, when recognized by the colipase, allows it to match the lid domain. In silico protein structure prediction of both normal and mutant mPA-PLA1a rabbit protein suggests that the mutant protein is unable to match with the colipase. This could explain the decrease in the lipase activity.
The 1362delA mutation could also affect the tertiary structure of the mPA-PLA1a and interfere with its cell membrane expected localization. The modeling indicates that the rex protein presents a destabilization of the C-terminal domain in comparison with the normal protein. The mutant mPA-PLA1a could thus present a default of localization, hampering normal activity.
In contrast to the pathological defects reported in humans, the situation of rex and orylagH rabbits is considered more as an advantage than as a defect. This is due to the fact that the deletion reduces but does not abolish the hair growth. Our rabbit mutant and normal phenotypes represent a most suitable model to better understand the function of LIPH in the hair growth and development process and should stimulate new investigative areas for human applications. Further investigations are needed to explore the mode of action of this gene in such a complex, but so instructive, mechanism of hair growth cycling.

Animals and tissue collection
Mating plans were achieved in order to obtain rabbit fetuses at 19, 25 and 29 days postcoïtum. One normal and one orylagH female were used for each stage leading to the collection of 5 to 6 fetuses per female. We pooled the fetus tissues of the same litter. Adult rabbits 11-14 weeks old were selected to ensure obtaining fully developed hair follicle groups. Skin biopsies were made on 7 orylagH+, 7 orylagH2, 10 rex, 10 normal (INRA2066 strain) and 10 heterozygous for 1362delA LIPH rabbits. We produced adult rabbits with hair cycles in phase for the growing anagen and the resting catagen/telogen by plucking skin areas of 20 cm 2 . Skin specimens were immediately snap frozen in liquid nitrogen and stored at 280uC.
In France, no mandatory ethical committee approval is yet necessary to conduct experiments on the animals of this study.

RNA preparation and RT-qPCR analysis
Skin samples were processed for total RNA extraction using RNA Now reagent (Ozyme) as described by the manufacturer. Reverse transcriptions (RT) were performed on 5 mg of total RNA using the Superscript First Strand Synthesis System (Invitrogen) according to the manufacturer's instructions. Quantitative PCR were performed on RT products using the Mastercycler Realplex (Eppendorf). Reaction conditions consisted of 15 min at 95uC (1 cycle), 15 s at 95uC, 30 s at 60uC and 30 s at 72uC(40 cycles) and final 5 min at 72uC with primers (5 mM) using Absolute QPCR Sybr-Green (Thermo Scientific). Primers are presented in Table 2. The relative abundance of each gene was determined by the formula 2 DCt [24]. The DCt value corresponds to the Ct of each gene after normalization with the GAPDH housekeeping gene.

Histology
Rabbit skins were fixed 48 h in formol 10% at room temperature, dehydrated through successive baths of ethanol (70%, 265 h; 80%, 265 h; 95%, 5 h and 100%, 6 h) and xylene substitute (UltraclearH, 266 h). Then, they were embedded in three successive baths of paraplast (60uC, 365 h). Paraffin sections (5 mm) were cut at room temperature and mounted on superfrostH plus slides, air dried overnight at 45uC and stored at room temperature until use for the IHC and ISH treatment. Skin sections on slides were stained with ROAN solution and then dipped into nuclear red for 7 min and rinced before being dipped into Orange G for 4 min and in Aniline blue for 3 min, dehydrated and mounted.

In situ hybridization
A 213 pb normal rabbit LIPH cDNA was amplified with ''ISH primers'' ( Table 2) for in vitro RNA probe synthesis and cloned into the pGEM-T vector system (Promega A3600). ISH riboprobes were synthetized with sp6 and T7 promoters. The transcription mixture included 1 mg of linearized plasmid (from M13 amplification), dNTP mix with DIG-UTP (Roche, DIG RNA labelling mix), RNase Inhibitor (Roche), and sp6 or T7 RNA polymerase (Roche). Tissues on slides were deparaffined and rehydrated through successive baths of xylene and ethanol and then in PBS-Tween 0.1% before being fixed with PFA 4% for 20 minutes and equilibrated in SSC5X for 15 min. Prehybridizations were performed with hybridization buffer pre-warmed at 85uC (50% formamide, 16 salt buffer, 10% dextran sulfate, 1 mg/mL yeast RNA, 16 Denhardt) during 1 hour, at 65uC in a dry oven. Hybridization was performed overnight at 65uC in a dry oven, with LIPH riboprobes diluted at 1/200 with hybridization buffer. Slides were rinsed through successive baths of SSC and blocked 1 h30 in MABT-Normal Goat Serum 20% at room temperature. Then, hybridization with the alkaline phosphatase-coupled antidigoxigenin antibody (Roche), 1/2000 diluted on MABT-Normal Goat Serum 2%, was performed overnight at +4uC. Revelation was performed with the substrate of the alkaline phosphatase BMPurple (Roche) for 72 h (4-5 days for fetuses, data not shown). Slides were washed in PBS and PFA 4%, and mounted with an aqueous mounting media (abcys, H-5501).

Immunohistochemistry
Tissues on slides were deparaffined and rehydrated through successive baths. The protocol involved a polyclonal antibody produced in the chicken against the rabbit LIPH peptide YHQVSLLARFNQDLDKVAE (ProteogenixH, 1/200), a horse-radish peroxidase secondary antibody (HRP goat polyclonal to chicken IgY, Abcam) and Histogreen (Abcys) as chromogens for the peroxidase activity. Fetal slides were led to saturation in order to get a sufficient signal.

Enzymatic activity
Each plasmid was transfected into CHO cells as described previously [25]. LIPH wild-type and LIPH 1362delA were amplified with the ''cDNA LIPH'' primers ( Table 2). After maxiprep (NucleobondH PC100, Macherey-Nagel), PCR products and empty pcDNA3.1+ vector (Invitrogen) were double digested by KpnI and XhoI (Biolabs). The two LIPH constructs were transformed into pcDNA3.1 plasmid and amplified by maxiprep. CHO cells (Invitrogen) from passages 2-3 were then transiently transfected with the two LIPH constructs or with the empty vector. Cells were seeded into a 96-well plate (BD Biosciences, Franklin Lakes, NJ). At day 2 after transfection, the medium was replaced by HBSS (Hank balanced salt solution) containing 0.5 mM PED-A1 (Invitrogen). mPA-PLA1a enzyme activity was monitored using the PED-A1 substrate quenching system (Invitrogen). The phospholipase activity was monitored every minute for 1 hour at 37uC using an epi-fluorescence microscope (excitation filter 485 nm, emission filter 540 nm).