A harlequin ichthyosis pig model with a novel ABCA12 mutation can be rescued by acitretin treatment.

Harlequin ichthyosis (HI) is a severe genetic skin disorder and caused by mutation in the ATP-binding cassette A12 (ABCA12) gene. The retinoid administration has dramatically improved long-term survival of HI, but improvements are still needed. However, the ABCA12 null mice failed to respond to retinoid treatment, which impedes the development of novel cure strategies for HI. Here we generated an ethylnitrosourea mutagenic HI pig model (named Z9), which carries a novel deep intronic mutation IVS49-727 A>G in the ABCA12 gene, resulting in abnormal mRNA splicing and truncated protein production. Z9 pigs exhibit significant clinical symptom as human patients with HI. Most importantly, systemic retinoid treatment significantly prolonged the life span of the mutant pigs via improving epidermal maturation, decreasing epidermal apoptosis, and triggering the expression of ABCA6. Taken together, this pig model perfectly resembles the clinical symptom and molecular pathology of patients with HI and will be useful for understanding mechanistic insight and developing therapeutic strategies.

Introduction

Harlequin ichthyosis (HI) is an inherited disease, which mainly affects the skin. HI is the most severe form of autosomal recessive congenital ichthyosis (ARCI), which includes HI, congenital ichthyosiform erythroderma, and lamellar ichthyosis (Akiyama and Shimizu, 2008). Patients with HI are born with a thick covering of armor-like scales over the entire body. Affected infants also have abnormal facial features, including ectropion, eclabium, and flattening of the ears and nose (Farhadi and Kazemi, 2013; Parikh et al., 2016). According to clinical data of 45 patients with HI, the ages of the survivors ranged from 10 months to 25 years, with an overall survival rate of 56% (Rajpopat et al., 2011). Most infants died from microbial infections, feeding problems or respiratory distress. The survival rate of patients with HI increased to ~80% after being treated with systemic retinoids, which is an anchor therapy in disorders of keratinization (Rajpopat et al., 2011). Therefore, a more comprehensive therapeutic protocol for this disease is still under development, which certainly relies on obtaining more clear understanding of pathological mechanism with the aid of optimal animal models.

Two independent research groups have reported that the ABCA12 gene is the causative gene of HI (Akiyama et al., 2005; Kelsell et al., 2005). The ABCA12 gene belongs to ATP-binding cassette (ABC) transporter superfamily (Annilo et al., 2002; Lefévre et al., 2003); the ABC genes encode a large number of membrane proteins involved in energy-dependent transport of various substrates across cell membranes, and mutations in these genes cause several human genetic disorders (Allikmets et al., 1996; Dean and Allikmets, 2001). The ABCA12 protein is located within the lamellar granules (LGs) in the epidermal keratinocytes and is supposed to be involved in lipid transport into the extracellular space. LGs, also called lamellar bodies, are specialized vesicular structures in the cytoplasm, which function to transport lipids to extracellular space (Sakai et al., 2007). ABCA12 thus plays a critical role in transporting lipids within the epidermal cells and formation of the intercellular lipid layer, ultimately influencing the epidermal barrier function.

Three independent research groups have generated ABCA12 null mice by either ethylnitrosourea (ENU) mutagenesis or homologous recombination (Smyth et al., 2008; Yanagi et al., 2008; Zuo et al., 2008). These models partially reproduce the human HI skin phenotype and aid the investigation of pathological mechanisms of HI due to ABCA12 deficiency. However, even treated with fetal retinoid, all ABCA12 null mice die within a few hours of birth, showing neither improved skin phenotype nor extended survival period (Yanagi et al., 2008). This situation severely limits the use of mouse model for guiding the development of innovative treatments, including new drugs and nursing methods, for ichthyosis with ABCA12 deficiency.

Pigs and humans share numerous similarities in genetics, anatomy, physiology, pathology, and immunology (Swindle et al., 2012; Lin et al., 2017), which make pigs very suitable for modeling many human diseases. The structure and physiology of porcine and human skin are very similar, which make pigs a useful model to study human skin related diseases (Debeer et al., 2013; Summerfield et al., 2015). Till now, many pig models have been developed and used in human skin-related studies, including wound repair (Ansell et al., 2012), burn research (Abdullahi et al., 2014), infectious diseases (Meurens et al., 2012), as well as chemical penetration studies (Barbero and Frasch, 2009).

In the current study, we created a novel ENU-induced deep intronic mutation IVS49-727 A>G in the ABCA12 gene that results in abnormal mRNA splicing and truncated protein production, causing hyperkeratotic skin in Bama miniature pigs. The affected (named Z9 later) pigs show pathological features as human patients with HI. Most importantly, therapeutic effects were observed in pigs treated with systemic retinoids therapy. These pigs provide a remarkable model of human HI to better understand the pathological mechanisms of this disease and develop novel prevention and treatment strategies.

Results

Pig mutants presented with sclerotic, dry, and chapped skin, which mimic symptoms of human HI

The Z9 mutants were identified in a three-generation breeding scheme based on a large-scale screen of ENU-mutagenized porcine populations (Hai et al., 2017a). Affected G3 offspring were generated by backcrossing G2 sows with a G1 founder, and Z9 pigs were identified with sclerotic, dry, and chapped skin (Figure 1A). Ectropion, eclabium, flattening ears, decreased activities, and neonatal death were also observed in Z9 pigs. Furthermore, the birth weight of Z9 pigs was significantly lower than that of wild type (WT) littermates (WT, 0.51 ± 0.13 kg, n = 51; Z9, 0.39 ± 0.10 kg, n = 18; P < 0.001; Supplementary Figure S1).

Figure 1

Skin phenotype of an ENU-induced pig disease model. (A) Newborn Z9 pigs displayed a phenotype of sclerosis, dry and chapped skin, and neonatal death. (B) Histological sections of epidermis of neonatal Z9 pigs exhibited hyperkeratotic SC, lack of normal skin folds, and disordered SS. The lower panels show the magnified images of the selected region (red frames) of upper panels. Scale bar, 100 μm. (C) Ultrastructural defects in the epidermis of Z9 pigs were observed by TEM, including massive hyperkeratotic structure in the SC (top) and numerous multivesicular bodies in the SG (middle and bottom). Scale bar, 1 μm (top and middle) and 200 nm (bottom). The bottom panels show the magnified images of the selected region (red frames) of middle panels. (D) Cornified envelopes isolated from newborn Z9 pigs were irregular and fragile compared with WT controls. Scale bar, 100 μm. (E) Toluidine blue staining of euthanized neonatal Z9 pigs shows defects in epidermal barrier function. (F) The TEWL assay shows a defect in barrier formation of the dorsal and abdominal skin from Z9 pigs (n = 4; P < 0.05 between WT and Z9 pigs).

To investigate the pathological changes of Z9 skin, skin samples were collected for histological analysis. Histological sections from the skin of neonatal Z9 pigs revealed markedly hyperkeratotic stratum corneum (SC), as well as a lack of normal skin folds (Figure 1B). The spinous cells and basale cells of the skin from Z9 pigs also exhibited abnormal arrangement. These characteristics of skin from Z9 pigs are consistent with those from patients with HI.

Transmission electron microscopy (TEM) was performed to investigate the ultrastructure defects in the skin tissue of neonatal Z9 pigs. Clear lipid lamellar structures in the SC of WT skin were observed. In contrast, Z9 pigs were observed to have massive hyperkeratotic SC, losing normal lamellar structures in the interstitial spaces (Figure 1C). The WT stratum granulosum (SG) showed numerous LGs with typical structure and some LGs are close to interstitial lamellar structures, while the Z9 cytoplasm of the SG showed plenty of multivesicular bodies instead of normal LGs.

To examine the SC, cornified envelopes were isolated from epidermis of WT and newborn Z9 pigs and then photographed under a light microscope. Cornified envelopes were prepared as previously described (Koch et al., 2000). Z9-cornified envelopes exhibit several defects in comparison with those of WT pigs (Figure 1D). In detail, cornified envelopes from Z9 pigs showed more irregular and fragile appearance, while those from WT pigs showed mature and rigid appearance. In addition, Z9-cornified envelopes tended to stack up, while cornified envelopes from WT pigs had smooth surface and were neatly arranged in liquid.

Since severe dehydration was observed in Z9 carcasses, the barrier function of the epidermis was investigated, and skin permeability assay was carried out. Euthanized neonatal pigs were stained with toluidine blue and washed with phosphate buffered saline (PBS). Results showed that the skin of neonatal Z9 pigs had increased inward permeability to toluidine blue (Figure 1E). A gravimetric transepidermal water loss (TEWL) assay, which is able to measure water evaporation from the skin surface, was performed on samples harvested from neonatal pig dorsal skin and abdominal skin. Water loss of the Z9 skin was significantly higher than that of WT skin at any time point (Figure 1F). These results indicate that the epidermal permeability barrier was disrupted in Z9 pigs.

Mapping and identification of the causative mutation in ABCA12 gene

The mutant phenotype was not observed in G1 and G2 pigs but was discovered in both male and female G3 pigs (Figure 2A). Among 50 offspring from 7 litters, 37 WT pigs and 13 mutant pigs were identified, with a ratio of ~3:1 (Figure 2B). These data implied that the mutation in the Z9 mutant pedigree is inherited in an autosomal recessive manner.

Figure 2

and identification of the novel ABCA12 mutation. (A) Pedigree map shows the autosomal recessive inheritance of the Z9 mutation. (B) Distribution of the number of WT and mutant pigs in the G3 offsprings. (C) Through genetic linkage analysis, only one linkage signal (LOD, 4.13) among the whole genome was identified in a 19 Mb region on chromosome 15. (D) Reverse-transcription PCR indicated abnormal pre-mRNA splicing. (E) Sequencing of ABCA12 mRNA revealed a 132-nt insertion in Z9 pigs, which introduced a premature stop codon. The red asterisk indicates stop codon (UAA). (F) ABCA12 with the Z9 mutation expresses a truncated protein, which lacked part of its ABC domain. (G) A deep intronic mutation of IVS49-727 A>G in ABCA12 results in splicing alteration. (H) Brief summary of gene mapping: a deep intronic mutation in ABCA12 was located. (I) Minigene experiment was carried out to verify the effect of the ABCA12 IVS49-727 A>G mutation on pre-mRNA splicing in vitro.

To identify the causative mutation in the Z9 mutant pedigree, a family-based genome-wide association study (GWAS) and candidate gene sequencing were performed. GWAS results showed a strong signal at 119 Mb to 133 Mb on chromosome 15 (Supplementary Figure S2). Similarly, the linkage analysis also identified the significant linkage signal of a 19 Mb region on chromosome 15 between 120 Mb and 139 Mb (Figure 2C). There are 38 annotated genes (Supplementary Table S1) in the overlapped genomic interval (120 Mb–133 Mb). Among these genes, ABCA12, whose mutations are known to result in ARCI, was selected for mutation screening.

Sanger sequencing of all 53 exons of ABCA12 revealed that no mutation co-segregated with Z9 phenotype in this pedigree. In order to determine if there was abnormal splicing of the ABCA12 gene in Z9 pigs, the mRNA was amplified and sequenced using a series of primer sets (Supplementary Table S2) that allows complete analysis of 7788 bp coding sequences (CDSs) of ABCA12 gene in overlapping fragments. An unexpected length of polymerase chain reaction (PCR) products amplified by CDS-11 primers was observed in Z9 pigs (Figure 2D), and sequencing results showed a splicing alteration of mRNA with a 132-nt insertion between exon 49 and exon 50 in Z9 pigs (Figure 2E). The 132-nt insertion in ABCA12 mRNA of Z9 pigs was mapped to intron 49 of the ABCA12 gene by basic local alignment search tool (BLAST). The insertion introduced a premature stop codon, resulting in the expression of a truncated ABCA12 protein that lacked the second ABC domain (Akiyama, 2010; Figure 2F). In order to identify the causative mutation that results in the splicing alteration, the whole intron 49 was sequenced in Z9 pigs. As expected, a deep intronic mutation of A to G (IVS49-727 A>G) was identified next to the 3′ end of the insertion sequence in Z9 pigs (Figure 2G). According to the GT–AG rule, this mutation leads to a new splicing site and produces a 132-nt pseudoexon (named Ψ49b) (Figure 2H). The IVS49-727 A>G mutation in ABCA12 was only found in Z9 pedigree rather than in the SNP database or commercial pig breeds (Supplementary Table S3), indicating that this mutation might be the causative mutation and induced by ENU mutagenesis.

To assess if the IVS49-727 A>G mutation affects ABCA12 pre-mRNA splicing directly, we constructed WT and Z9 minigenes containing the genomic sequence from exon 49 to exon 50 of ABCA12 and transfected minigenes into 293T cells. The results showed that the IVS49-727 A>G mutation from Z9 pigs caused alterations in pre-mRNA splicing, similar to that seen in vivo (Figure 2I). Taken together, these data indicate that the IVS49-727 A>G mutation in ABCA12, which resulted in a 132-nt insertion in mature mRNA, was the causative mutation of Z9 pigs.

Interestingly, we have observed that the PCR product of Z9 mRNA contained minor bands that are shorter than the main band, which indicates that more splicing isoforms were produced by the mutation (Figure 2D). Sequencing results of the minor bands showed that ABCA12 gene with IVS49-727 A>G mutation also contains WT transcript and two other abnormally spliced transcripts (mutant transcripts 2 and 3) containing different exon skipping (Supplementary Figure S3A). Real-time quantitative PCR (RT-qPCR) results showed that the WT transcript expression level in Z9 skin is only 1/36 of that in WT skin and most of the ABCA12 expressions are alternative splicing isoforms (Supplementary Figure S3B).

IVS49-727 A>G mutation in ABCA12 causes abnormal skin lipid composition and increased lipid accumulation

As ABCA12 is a lipid transporter involved in lipid transport from cytoplasm to the extracellular space, lipidomics analysis was performed to investigate the effects of the ABCA12 IVS49-727 A>G mutation on lipid composition in porcine skin (Supplementary Table S4). Significantly different lipidomic profile of WT and Z9 skins was observed, suggesting that ABCA12 IVS49-727 A>G mutation should disrupt lipid transport process in ABCA12 skins (Figure 3A). Heatmap revealed that the mass levels of lipid species were significantly different between WT and Z9 skin (P < 0.05; Figure 3B). The significantly altered lipid subclasses account for 49.9% (190/381) of total subclasses. Besides the significantly decreased diacylglycerides, triacylglycerides, significantly increased levels of ceramide (Cer) were detected in skin tissues of Z9 pigs, which were consistent with reports in mice (Smyth et al., 2008). Glucosylceramides (GluCer) rather than galactosylceramides were markedly increased. Sphingosines (Sph), a breakdown product of Cer, and free cholesterols were also significantly increased (Figure 3C). To confirm the impaired lipid transport of Z9 pigs in vitro, ABCA12, ABCA12, and ABCA12 fibroblasts were cultured in the presence of acetylated low-density lipoprotein (AcLDL) as the donor of lipid and the liver X receptor (LXR) agonist TO-901317 as the activator of the ABCA12 gene. The oil red O staining revealed that the lipid accumulation in ABCA12 fibroblasts was notably higher than that observed in ABCA12 and ABCA12 fibroblasts (Figure 3D and E). Taken together, these data indicated that the lipid homeostasis is disrupted in the skin of Z9 pigs, suggesting that ABCA12 is not only involved in the transport of lipids in skin but also plays a wider role in lipid metabolism.

Figure 3

ABCA12 mutation impacts lipid homeostasis. (A) Principal component analysis score plot of lipidomic profile of four paired WT and Z9 skin. (B) Heatmap revealed the mass level of lipid species, which showed significantly different between WT and Z9 skin (P < 0.05). CE, cholesteryl esters; Cer, ceramide; Cho, free cholesterol; CS, cholesteryl sulfate; DAG, diacylglycerides; GalCer, galactosylceramides; GluCer, glucosylceramides; GM, monosialo-dihexosyl gangliosides; LBPA, lyso-bisphosphatidic acids; LPC, lyso-PG; PA, phosphatidic acids; PC, phosphatidylcholines; PE, phosphatidylethanolamines; PG, phosphatidylglyerols; PI, phosphatidylinositols; PS, phosphatidylserines; Sph, sphingosines; TAG, triacylglycerides. (C) In the skin of Z9 pigs, the levels of ceramide, GluCer, Sph, and Cho were significantly increased compared with WT controls (*P < 0.05, **P < 0.01). (D) The lipid accumulation in ABCA12 fibroblasts was greater than that in ABCA12 and ABCA12 fibroblasts. Scale bar, 100 μm. (E) Quantitative analysis for D indicated that gray values per cell in ABCA12 fibroblasts were higher than those in ABCA12 and ABCA12 fibroblasts (n = 3; *P < 0.05, **P < 0.01).

Oral administration of acitretin improves the skin condition and survival rate of Z9 pigs

Acitretin, a synthetic retinoid, has been reported to be effective for the survival of patients with HI (Singh et al., 2001); however, the therapy neither improved skin manifestations nor extended the survival of mice used as an HI disease model (Farhadi and Kazemi, 2013). In order to investigate whether the porcine HI model responds to acitretin treatment, oral administration of acitretin to the pregnant sow (10 mg/kg daily; 30 consecutive days) and neonatal Z9 pigs (1 mg/kg daily; from the first day of birth) were carried out. As a result of acitretin treatment to the sow, neonatal Z9 pigs achieved significant improvement in the skin phenotype at birth (Figure 4A), and no fissures were observed over the whole body. Most importantly, acitretin treatment to the Z9 pigs resulted in extension of the survival period from 2 days to at most 23 days (mean ± SD: survival time, 7 ± 9 days, n = 5). In contrast, Z9 pigs without treatment survived no more than 48 h even with intensive care (Figure 4B). Histology results also showed improvement of skin condition in Z9 pigs with acitretin treatment (Figure 4C). The SC of Z9 pigs with treatment was slightly thinner than that without treatment, while the stratum basale (SB) of Z9 pigs with treatment was more like that of normal epidermis.

Figure 4

Oral administration of acitretin improves the survival rate of Z9 pigs. (A) Z9 pigs achieved significant improvement in the skin phenotype at birth with oral administration of acitretin to the pregnant sow (10 mg/kg daily; 30 consecutive days) and neonatal Z9 pigs (1 mg/kg daily; from the first day of birth). (B) Acitretin treatment to the Z9 pigs extended the survival period from 3 days to 23 days (ABCA12: n = 20; ABCA12: n = 6; ABCA12 with treatment: n = 5). (C) The histologic sections showed an improvement of skin SC and SB. The SB of Z9 pigs with treatment was more like that of normal epidermis. Scale bar, 100 μm.

Acitretin treatment improves the terminal differentiation of epidermis in Z9 pigs

To investigate the lipid barrier function of the epidermis, double immunofluorescent labeling of GluCer and Cer was performed on the skin from ABCA12 pigs, ABCA12 pigs, and treated ABCA12 pigs. Cer was evenly distributed in the entire epidermis of ABCA12 pigs, including the SC, but hardly distributed in the SC of ABCA12 pigs. GluCer showed intensive distribution in the SC of ABCA12 pigs but showed slight staining in that of ABCA12 pigs. It suggests that lipid barrier was disrupted in the Z9 epidermis. The ABCA12 pigs achieved partial improvement of GluCer distribution after treatment, which was more like the staining pattern of ABCA12 pigs. It suggests that the lipid barrier function in the epidermis of ABCA12 pigs was partially restored after acitretin treatment (Figure 5A).

Figure 5

Acitretin treatment improved the terminal differentiation of Z9 skin. (A) Immunofluorescence staining demonstrated the deficiency of both Cer and GluCer in the SC of ABCA12 pigs. (B) The ABCA12 protein was primarily detected in the SS of WT pigs but was concentrated in the upper SS in the ABCA12 epidermis. After acitretin treatment, ABCA12 protein in ABCA12 pigs was distributed in a pattern similar to that in WT pigs. (C) The disordered outline of K14 in the SB was greatly improved in the epidermis of ABCA12 pigs after treatment. (D–F) Staining for K10 as well as the cornified envelope marker loricrin and filaggrin showed diffuse expression in the SC ABCA12 pigs, revealing abnormal differentiation of keratinocytes. (G) ABCA12 epidermis exhibited premature apoptosis in the spinosum and granular layers. The cell nuclei were counterstained with Hoechst 33342 (blue). Scale bar, 50 μm.

The lipid barrier function of epidermis relies on ABCA12 protein. In WT pigs, the ABCA12 protein was primarily detected in the stratum spinosum (SS) of the epidermis, while the protein was concentrated in the upper SS in the ABCA12 epidermis, suggesting perturbed location of the ABCA12 protein. After treatment with acitretin, the distribution of ABCA12 protein in ABCA12 pigs shows a pattern similar to that in WT pigs (Figure 5B).

The epidermis of ABCA12 and ABCA12 pigs show different differentiation features. Keratin 14 (K14), a marker of SB cells, was detected in the lowest cellular layer of the epidermis with a regular outline of the SB in ABCA12+/+ pigs. In the ABCA12 epidermis, K14 expression was also in the lowest cellular layer but show a disordered outline of the SB. The disordered outline of the SB was greatly improved in the epidermis of ABCA12 pigs after treatment, and the K14 expression was more expansive (Figure 5C). Keratin 10, a marker of epidermal differentiation, was expressed sparsely in the SS and intensely in the SC. However, a defect of epidermal differentiation with hyperkeratinization was observed in ABCA12 pigs and this condition was improved by acitretin treatment (Figure 5D). Loricrin, a terminal differentiation marker and a major protein component of the cornified cell envelope (Yoneda and Steinert, 1993), was intensely expressed in the SG of the epidermis from ABCA12+/+ pigs, while loricrin was diffusely expressed in the whole SC and SG of the epidermis from ABCA12 pigs. However, in ABCA12 pigs treated with acitretin, loricrin showed a similar expression pattern as WT pigs (Figure 5E). Filaggrin, which is cross-linked to the cornified envelope, participates in coordinating the structure of the cornified cells (Steinert and Marekov, 1995). We also characterized filaggrin to further evaluate the cornified cell envelope. In the epidermis, filaggrin was expressed diffusely in the SS of ABCA12 pigs, whereas treatment with acitretin led to the intense expression of filaggrin in the SC and SG, resembling expression levels seen in ABCA12+/+ pigs (Figure 5F). We assessed cellular proliferation by staining the proliferation marker Ki67 (Supplementary Figure S4). Ki67 was observed in the basal layers of the porcine epidermis. The ki67 positive cells of ABCA12+/+, ABCA12, and treated ABCA12 epidermis were about an equal number.

Terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) assays were performed to investigate cell apoptosis in the skin (Figure 5G). It is detected that ABCA12 epidermis exhibited premature apoptosis in the spinosum and granular layers compared with that of ABCA12, while ABCA12 with treatment epidermis exhibited less expansive TUNEL staining than that of ABCA12. It was reported that the activation of the AKT signal pathway have an antiapoptotic function, and the level of Ser-473 phosphorylated AKT in the Abca12−/− mouse was higher than those of control mice (Yanagi et al., 2011). In the current study, we assessed the degree of AKT activation in Z9 skin before and after acitretin treatment by immunoblot analysis. Oppositely, the skin of Z9 pigs showed lower expression levels of phosphorylated AKT than those of the WT pigs (Supplementary Figure S5A and B).

ABCA6 expression is upregulated in acitretin-treated Z9 pigs

To better understand the pathogenesis underlying HI caused by ABCA12 mutations and the therapeutic mechanism of acitretin on HI, RNA sequencing (RNA-seq) analysis was performed to identify the transcriptional profiles in the skin from three groups of pigs. The differentially expressed genes (DEGs) were screened based on the criteria of false discovery rate (FDR)-value <0.05 and absolute fold change ≥2. The volcano plots showed a broad overview of the DEGs in the skin of WT vs. Z9 pigs (Figure 6A) as well as Z9 pigs vs. treated Z9 pigs (Figure 6B). Our results indicated that 1272 upregulated genes and 1887 downregulated genes were found in Z9 pigs, and 639 upregulated genes and 402 downregulated genes were found in Z9 pigs with acitretin treatment. Moreover, 15 top DEGs were selected and verified by RT-qPCR (Supplementary Figure S6A and B).

Figure 6

Transcriptome analysis of the skin tissue of WT, Z9, and treated Z9 pigs. (A) The volcano plot revealed a broad overview of the difference in gene expression between WT and Z9 pigs. (B) The volcano plot revealed a broad overview of the difference in gene expression between Z9 pigs and Z9 pigs treated with acitretin. (C) Functional enrichment analysis of upregulated and downregulated DEGs in Z9 pigs compared with WT pigs identified the significantly overrepresented GO terms (corrected P < 0.05). (D) Functional enrichment analysis of upregulated and downregulated DEGs in acitretin-treated Z9 pigs compared with Z9 pigs identified significantly enriched GO terms (corrected P < 0.05). (E) Heatmap showed the upregulated expression of genes in GO term of epidermis development (GO:0008544) in Z9 pigs. (F) Heatmap showing the downregulated expression of genes in GO term of cornification (GO:0070268) after acitretin treatment of Z9 pigs. (G) mRNA abundance determined by RNA-seq showed that STRA6, RBP1, and RARB expressions were notably increased (FPKM) after acitretin treatment (n = 3; P < 0.05). (H) mRNA abundance determined by RNA-seq showed that ABCA6, ABCA8, and ABCA9 expression levels were notably increased after acitretin treatment (n = 3; P < 0.05).

The functional enrichment analysis identified the upregulated epidermis development (GO:0008544) as the most significantly enriched gene ontology (GO) term, indicating epidermis development of Z9 pigs were disrupted (Figure 6C and E). Interestingly, we also found that genes in GO term of cornification (GO:0070268) were significantly downregulated after acitretin treatment of Z9 pigs, suggesting that acitretin treatment partially restore the epidermis development in Z9 pigs (Figure 6D and F).

Among these top DEGs, three genes related to retinol metabolism were identified, including STRA6 (FDR, 9.04 × 10−32; log2 fold change, 6.1), which encodes a membrane protein involved in the metabolism of retinol; RBP1 (FDR, 1.06 × 10−13; log2 fold change, 2.97), which encodes the carrier protein involved in the transport of retinol; and RARB (FDR, 3.23 × 10−8; log2 fold change = 2.75), which is a nuclear transcriptional regulators in cytoplasm (Figure 6G). This result implied that the oral administration of acitretin induced the response of retinol metabolism in skin and thus played its therapeutic effects.

Importantly, ABCA6, a gene that encodes a cholesterol transporter and belongs to the superfamily of ABC transporters, was significantly upregulated in treated Z9 pigs (Figure 6H), and it was confirmed by RT-qPCR. RT-qPCR results showed no change in the expression level of ABCA8 and ABCA9 after acitretin treatment (Supplementary Figure S7A). Furthermore, acitretin was demonstrated to trigger ABCA6 expression in vitro using HaCat cells cultured in the present of acitretin for 60 h (Supplementary Figure S7B). In brief, we inferred that the triggered expression of ABCA6 may partially compensate the loss of function of ABCA12, as a mechanism that how acitretin work in treating HI in human beings and pigs.

Discussion

We have successfully established a systematic three-generation ENU mutagenesis porcine program and screened a large scale of mutants with a broad range of phenotypes that could be potentially developed into models for human diseases (Hai et al., 2017a, 2017b; Zhang et al., 2017). Forward genetic screening is an important method that can be used to establish human disease models and to facilitate studies of gene function (Hrabé de Angelis and Balling, 1998) due to the advantages like high throughput and identification of novel molecular signal pathways without prior hypothesis (de Bruijn et al., 2009). ENU mutagenesis generates a series of point mutations that frequently mimic the subtlety and heterogeneity of human genetic lesions (Funato et al., 2016; Oliver and Davies, 2012). Pigs are increasingly being used as human disease models due to their anatomical and physiological similarities with human and many breeding and handling advantages (Swindle et al., 2012). Most dominantly, porcine skin more closely resembles human skin than rodent skin, making pigs a better model to study skin-related disease.

In this study, we report the discovery of a novel mutation (ABCA12 IVS49-727 A>G) in Bama miniature pigs induced by ENU mutagenesis, which led to a remarkable model of human HI. The IVS49-727 A>G mutation mainly results in a splicing alteration with a 132-nt insertion between exon 49 and exon 50, introducing a premature stop codon in the ABCA12 gene. It is consistent with human HI findings that only truncation or conserved region deletion mutations could seriously affect ABCA12 protein function and lead to the HI phenotype (Akiyama et al., 2005). Moreover, up to 13% (5 in 38) of the reported ABCA12 mutations in human HI are splice site mutations (Rajpopat et al., 2011). Therefore, future gene therapy aiming at splice site mutations should be feasible for human. The IVS49-727 A>G mutation happen to generate a typical restriction of 5′ consensus sequence of intron (from ATAAGT to GTAAGT; Qian et al., 2014); as a result, ABCA12 pre-mRNA of Z9 pig shows abnormal splicing. Therefore, targeting the sequence including this 6-nt sequence by genome editing tool should be an effective treatment method to restore the abnormal alternative splicing. Interestingly, extremely low level of ABCA12 WT transcript, generated by alternative splicing, was observed in Z9 skin. However, despite of the expression of WT transcript, the phenotype cannot be recovered in Z9 pigs.

The Z9 pigs perfectly mimic the clinical manifestations of human patients with HI, including hyperkeratosis, ectropion, eclabium, developmental delay, abnormal LGs, anomalous skin barrier function, lower birth weight, and defects in lipid metabolism, which provide us a valuable model to understand the pathological mechanism of HI. A disordered arrangement of spinous cells and basale cells was observed in Z9 pig epidermis but has not been reported in ABCA12 null mouse epidermis. However, these findings are similar to those seen in skin biopsy of human patients with HI, which exhibited disordered basale cells (Rathore et al., 2015). Therefore, we infer that epidermal disordered arrangement is one of the common features between human patients with HI and Z9 pigs. Ki67 staining showed that imbalanced cell proliferation in the SB might result in disordered basale cells of the skin from Z9 pigs.

Previous lipidomics studies in the skin of ABCA12 null mice led to conflicting results (Smyth et al., 2008; Zuo et al., 2008). Smyth et al. showed that Cer was significantly increased in the skin of ABCA12 null mice, while the study from Zou et al. showed that Cer was decreased. Our results of lipidomics analysis in pigs were consistent with the view that Cer was significantly increased in ABCA12 mutant skin. These data certainly demonstrate that the ABCA12 mutation impairs its transporter activity and the accumulated Cer results in detrimental effects to skin keratinization.

We observed that Z9 epidermis exhibited premature apoptosis by TUNEL assay. It was reported that mutations of ABCA12 gene would result in Cer accumulation in the cytoplasm (Smyth et al., 2008), which may inhibit the activation of AKT (Bourbon et al., 2002). Akt pathway is an essential pathway for cell survival and growth during development, and it is well known that inactivation of AKT promotes apoptosis (Brunet et al., 1999). In the current study, the inactivation AKT in Z9 skin was observed and may be resulted from the Cer accumulation. These results indicated that AKT plays an antiapoptotic role in ABCA12-deficient keratinocytes.

The ABCA12 null mice die shortly after birth, and retinoid therapy to the pregnant mice failed to improve the pups’ skin phenotype or extend the survival period of the ABCA12 defective newborns (Yanagi et al., 2008). In contrast, Z9 pigs showed remarkable therapeutic effects with the acitretin treatment, a drug commonly used clinically to increase the survival rate of patients with HI. The drug enters the cells by nonreceptor-mediated endocytosis, and once in the nucleus, acitretin activates two classes of nucleic acid receptors—retinoic acid receptors (RAR) and retinoid X receptors (RXR)—which subsequently activate expression of their downstream genes (Dogra and Yadav, 2014). Acitretin has multiple effects on epidermal cell growth and differentiation through induction of epidermal growth factor and transforming growth factor β (Tong et al., 1990). However, the exact mechanism of the acitretin effects on HI remains unknown. Z9 pigs offer the potential of studying the mechanisms of acitretin treatment and possibility of improving therapies of human HI.

Among the top DEGs identified by RNA-seq, STRA6, RBP1, and RARB play roles in retinoid metabolism, and upregulation of these genes indicated that administration of acitretin to pregnant sow and neonatal pigs has induced a therapeutic response in skin cells of Z9 pigs. The analysis of genes clustered in epidermis development and cornification shows that acitretin treatment in Z9 pigs gained improvement in the skin phenotype through alleviation of skin cornification. These results confirmed that the clinical improvement of acitretin patients might be obtained via the activation of RAR and RXR. The upregulated epidermis development of Z9 pigs was identified by the functional enrichment analysis, while genes in GO term of cornification were significantly downregulated after acitretin treatment in Z9 pigs, which might explain the therapeutic mechanism of acitretin in treating HI. Besides, ABCA6, a transporter of cholesterol that belongs to ABCA family, was significantly upregulated by acitretin treatment, suggesting that ABCA6 might play a role through the compensation of function loss of ABCA12 gene. These results provide new insights for explaining the therapeutic mechanism of acitretin.

In summary, we have generated a novel HI pig model based on a deep intronic mutation in ABCA12. This pig model faithfully replicates the abnormalities and pathologies seen in human patients and shows positive responses to acitretin treatment. This model provides a useful tool for understanding the disease development and a new material for developing treatments strategies.

Materials and methods

Animals

Bama miniature pigs were raised at the Beijing Farm Animal Research Center (attached to Institute of Zoology, Chinese Academy of Sciences), where ad libitum access to feed and water is supplied during the experimental period. All experiments involving animals were performed according to the guidelines for the Care and Use of Laboratory Animals established by the Beijing Association for Laboratory Animal Science and were approved by the Animal Ethics Committee of the Institute of Zoology, Chinese Academy of Sciences.

Skin permeability assay and gravimetric TEWL assay

In situ skin permeability assays were performed with toluidine blue as previously described (Hardman et al., 1998). WT and Z9 pigs were euthanized after birth and the entire bodies were washed with PBS, incubated in 0.1% toluidine blue for 5 min, washed with PBS three times, and then photographed.

TEWL assay was performed on samples harvested from the neonatal dorsal skin, which is able to measure the water evaporation from the skin surface. Skin samples sealed on laboratory film (Bemis) with Vaseline were weighed on a microbalance every 30 min for 5 h at ambient temperature. Each skin sample was photographed and the areas were calculated with imageJ software. TEWL was calculated as milligrams of water loss per square millimeter of epidermis per hour.

Gene mapping analysis of Z9 pigs

Twenty-four genomic DNA samples (including the G1 founder, three G2 female pigs, 12 G3 WT pigs, and eight G3 affected pigs) were isolated from ear skin tissues using a phenol–chloroform method. DNA samples were subjected to whole-genome single-nucleotide polymorphism (SNP) genotyping using a porcine SNP60 BeadChips (Illumina), which contains 62163 SNP markers. Raw genotype data were processed using standard quality control procedures. A family-based GWAS (transmission disequilibrium test analysis) was performed to detect SNP loci showing a significant association with the Z9 trait using PLINK software (Purcell et al., 2007). Additionally, Merlin software (Abecasis et al., 2002) was also used to conduct a parametric linkage analysis based on an autosomal recessive model, and the logarithm of the odds (LOD) score was calculated to assess the evidence for linkage.

Identification of causative mutation

Sanger sequencing of all CDSs of 53 exons and intronic sequence of intron 49 in ABCA12 gene was performed. All primers used in this paper are available in Supplementary Table S2.

Minigene assay

To construct minigenes, genomic sequence of ABCA12 ranging from exon 49 to 50 were amplified by PCR with DNA template from WT and Z9 pigs. Forward primer (HindIII): 5′-ACTAAGCTTACAGGGTGTCAACTTCAGTGAG-3′. Reverse primer (XhoI): 5′-ACTCTCGAGAGGCGAGTATGGTACTGTGGG-3′. WT and Z9 PCR products were digested with HindIII and XholI and then cloned into pcDNA3.1 vector. For each minigene experiment, 2.5 μg minigene vectors were transfected to 293T cells. After 24 h, the cells were collected for RNA isolation and reverse-transcription PCR using M-MLV reverse transcriptase (Promega). Transcription products of the minigenes were amplified with universal primers T7 and BGH and then subjected to Sanger sequencing.

Lipid accumulation assay

Fibroblasts were cultured from ear skin tissues of the ABCA12+/+, ABCA12+/, and ABCA12 pigs. Assays of lipid accumulation were performed according to instruction previously described (Smyth et al., 2008). Fibroblasts were incubated in serum-containing medium with TO-901317 (4 mM; Cayman Chemical) and AcLDL (10 mg/ml; Yeasen Biotech) for 18 hrs. Fibroblasts were washed with PBS, fixed with 4% formaldehyde, and then stained with Oil Red O (Sigma-Aldrich) working solution.

Acitretin therapeutic trials

Pregnant pigs were administered orally with acitretin (Huapont Pharm; 1 or 10 mg/kg daily) 30 days before parturition, and newborn piglets were dosed orally with acitretin (1 mg/kg daily) soon after birth.

Deep sequencing and reads mapping

RNA-seq analysis was conducted at Annoroad Gene Technology. Paired-end libraries for sequencing were prepared according to the Illumina PE library preparation protocol (Illumina), and the qualified libraries were sequenced on an Illumina Hiseq 2500 sequencing platform. The qualified reads (clean reads) were aligned to the pig built 11.1 reference sequence using TopHat software with default parameters. The BAM files (generated from TopHat) that contained the read alignments were then used to evaluate gene expression levels by fragments per kilobaseof exon per million fragments mapped (FPKM) values using Cufflinks software.

DEG identification

To identify the gene expression level changes, the number of the uniquely mapped reads assigned to each gene in the pig genome (using the gene annotation file for pigs ‘Sus_scrofa.Sscrofa11.1.94.gtf’) was first analyzed using the featureCounts package. The read counts of each gene were used as input and the differentially expressed genes in different groups with a threshold value of FDR  <0.05 and a fold change ≥2 were identified by edgeR software.

Clustering analysis and GO analysis

The hierarchical clustering of DEGs was performed using the heatmap.2 function in gplots, and a heatmap was generated to represent the gene clusters showing similar expression patterns. GO enrichment analysis of upregulated and downregulated genes was performed using the Metascape toolkits. The GO terms with P-values (enrichment score) <0.05 were regarded as ‘statistically significant’.

Statistics

All data are expressed as mean ± standard deviation (SD). Student’s t-test was used for all analyses, and statistical significance was defined as P-value <0.05. The replicates in each analysis can be found in the figure legends.

Click here for additional data file.