Patients with congenital ichthyosis and TGM1 mutations overexpress other ARCI genes in the skin: Part of a barrier repair response?

Abstract Autosomal recessive congenital ichthyosis (ARCI) is a group of monogenic skin disorders caused by mutations in any of at least 12 different genes, many of which are involved in the epidermal synthesis of ω‐O‐acylceramides (acylCer). AcylCer are essential precursors of the corneocyte lipid envelope crosslinked by transglutaminase‐1 (TGm‐1), or a yet unidentified enzyme, for normal skin barrier formation. We hypothesized that inactivating TGM1 mutations will lead to a compensatory overexpression of the transcripts involved in skin barrier repair, including many other ARCI‐causing genes. Using microarray, we examined the global mRNA expression profile in skin biopsies from five ARCI patients with TGM1 mutations and four healthy controls. There were a total of 599 significantly differentially expressed genes (adjusted P < 0.05), out of which 272 showed more than 1.5 log2fold‐change (FC) up‐ or down‐regulation. Functional classification of the latter group of transcripts showed enrichment of mRNA encoding proteins mainly associated with biological pathways involved in keratinocyte differentiation and immune response. Moreover, the expression of seven out of twelve ARCI‐causing genes was significantly increased (FC = 0.98‐2.05). Also, many of the genes involved in keratinocyte differentiation (cornified envelope formation) and immune response (antimicrobial peptides and proinflammatory cytokines) were upregulated. The results from the microarray analysis were also verified for selected genes at the mRNA level by qPCR and at the protein level by semi‐quantitative immunofluorescence. The upregulation of these genes might reflect a compensatory induction of acylCer biosynthesis as a part of a global barrier repair response in the patient′s epidermis.

transforms into either of four major clinical subtypes of ARCI, the most common of which is lamellar ichthyosis. [1][2][3][4][5] Severe forms of ARCI often cause a reduced quality of life and usually require chronic treatment with systemic retinoids.
Because TGm-1 is important in the final step of keratinization and many of the other mentioned gene products are preceding players in the formation of a skin barrier, a genetic deficiency of TGm-1 could hypothetically elicit compensatory gene regulations either via some pathway-specific feedback mechanisms or as a part of a more general epidermal repair mechanism.
To address this possibility, which has barely been studied in the past, we examined the global mRNA profile and expression of specific proteins in epidermis of patients with TGM1 mutations compared with healthy controls.

| Ethical statement
All studies were approved by the Regional Ethical Review Board (EPN) in Uppsala and conducted according to the Declaration of Helsinki Principles. Informed and written consent was obtained from the patients and healthy controls.

| Skin specimens
Skin punch biopsies were taken from the patients and controls (see Supporting Information) for RNA analysis and immunofluorescence staining as previously described. [16] A detailed description of the patients involved in this study can be found in Table 1 and in a previous report. [4] In brief, all patients were males, aged 29-80; four of them were born as collodion babies and later in life required maintenance treatment with oral acitretin, which for ethical reasons could not be stopped to obtain naive skin biopsies for the study. Topical therapy consisted of emollients twice a day which the patients were asked to abandon on gluteal skin for 12 hours before sampling.

| Microarray analysis
GeneChip ® ST Arrays (GeneChip ® Clariom D Human Array) was used for the microarray analysis of gene expression (see details in Supporting Information). Data were analysed as previously described. [17] The array data are deposited in Gene Expression Omnibus of NCBI (accession number GSE107462).

| Analysis of mRNA expression using qPCR
cDNA was synthesized from total RNA as previously described. [18,19] Semi-quantitative PCR was performed using TaqMan Gene Expression Assays (see Table S1 in Supporting Information) in an ABI7500Fast machine. Expression levels were measured in triplicate. The relative mRNA expression was determined by the 2ˆ(− ΔΔCt) method using RPL19 as a reference gene.

| Immunofluorescence (IF)
Patient and normal human skin biopsies were frozen and sectioned at 6 μm in a cryostat. Tissue sections were fixed, blocked, immunostained followed by microphotography and image analysis as described in Supporting Information and Table S2. The TGM1 mutations and degree of scaling (0-4) and erythema (0-4) have previously been reported. [4] P2 and P5 are unrelated but carry the same homozygous mutation p.
[Ser358Arg] which is common in Sweden and reduces the TGm-1 activity in vitro to <4%. [41] The splice site mutation c. [877-2A>G] in P4 reduces the activity to <7%, and so do the compound heterozygous point mutations in P3. [41] The point mutation p.
[Arg142His] in P1 abolishes the enzymatic activity in vitro, whereas the effect of the p.
[Gln463His] mutation has not previously been described, but is presumed to reduce the TGm-1 activity.

| Statistical analysis
Statistical methods are described in Supporting Information.

| Transcriptome analysis in patients with TGM1 mutations versus healthy controls-initial results
We performed a microarray analysis to investigate the global mRNA expression profile in skin biopsies from five male patients (P1-5) with ARCI in different severity and various types of inactivating TGM1 genes with UniGene ID, 599 of which were differentially expressed genes (DEGs) ( Figure 1A and Table S3). The controls and patients were well separated by two principal components generated from the global expression data ( Figure 1B). Out of these DEGs, 205 increased >1.5 log2fold-change (FC) and 67 decreased <−1.5 FC (which equals a 2.8-fold change after back-transformation), making a total of 272 markedly altered transcripts (Table S4) which were functionally enriched in different gene ontology (GO) terms ( Figure 1C).
A number of DEGs were identified as ichthyosis-causing and lipid biosynthesis genes (Table S3). For instance, the expression of seven With reference to lipid biosynthesis, two genes encoding key enzymes in the early steps of ceramide generation, SPTLC2 and SMPD1, were increased 1.59 and 1.09 FC, respectively (adjusted P < 0.054).

| Gene ontology analysis
We used the Database for Annotation, Visualization and Integrated Discovery (DAVID) to examine clusters among the 272 DEGs with adjusted P < 0.05 for FC>1.5 or <−1.5 (Table S4). The results show additional clustering of DEGs coding for proteins associated with biological pathways associated with keratinocyte differentiation (GO:0030216) and immune response (GO:0006955) in Table S5. Among the DEGs annotated to keratinocyte differentiation (also involving the formation of CE), eight small proline-rich proteins (SPRRs) were increased 3.18-9.31 FC (Table S6). Furthermore, involucrin (IVL) and two late cornified envelope proteins (LCE3E and LCE3D) were induced, as well as PPARD, a nuclear ligand-activated transcription factor known to promote keratinocyte differentiation and lipid metabolism in keratinocytes. In contrast, we did not observe any significant difference in the expression of the keratin genes.
Furthermore, two serine (or cysteine) protease inhibitors, SERPINB3 (Clade B [Ovalbumin]) and SERPINB4, known to be affected in early inflammation and barrier dysfunction, [22] were among the most strongly induced (8.09 and 7.70 FC, respectively) (

| Verification by qPCR
The array data, which indicated changes in the expression of many ichthyosis-causing and inflammatory-associated genes, prompted a reinvestigation of the transcripts by qPCR for verification. In total agreement with the array results, mRNA levels of ABCA12, SLC27A4, ELOVL4, CYP4F22, CERS3, SDR9C7, ALOX12B and LIPN were all increased in patients versus controls, but not so for ALOXE3, NIPAL4 and PNPLA1 ( Figure 3A). In addition, five strongly induced transcripts, but unrelated to any known ARCI aetiology, viz. CCL20, S100A7, FABP5, CD36 and IL36G, were markedly increased by qPCR analysis ( Figure 3B).

| Verification by immunofluorescence staining
Some genes showing pertinent changes at the transcriptional level were also analysed at the protein level by immunofluorescence (IF) staining of skin sections ( Figure 4A and B). By applying image analysis, protein expression-represented by median fluorescence intensitywas also analysed in different epidermal layers using recently developed pipelines for the CellProfiler software [23] ( Figure 4C). It can be seen that the expression of CYP4F22, FATP4 (encoded by SLC27A4) and CerS3 in patient skin is markedly increased in the upper spinous and granular layers and also extends to additional epidermal layers compared to control skin ( Figure 4A-C). S100-A7, which is almost undetectable in control skin, is highly expressed in the patients, especially in the differentiated layers of epidermis ( Figure 4). Likewise, the fatty acid transporter CD36 was strikingly increased in the upper spinous layer and granular layer of patient skin (Figure 4), whereas fatty acid binding protein 5 (FABP5) was more variably increased in stratum corneum of some patients compared to controls (Figure 4), thus corroborating our microarray and qPCR results.

| D ISCUSS I ON
This study is one of the first addressing global changes in the cutaneous gene expression in ARCI patients with mutations in TGM1.

| Effects on genes involved in innate immunity and inflammation
Induced AMP expression has been previously reported in ARCI patients with TGM1 mutations, as well as in TGm1-deficient mice. [24] Our study revealed a number of highly expressed transcripts encoding proteins with antimicrobial activity, viz. S100-A7, S100-A7A, the S100-A8/S100-A9 complex (calprotectin), SKALP/ELAFIN and beta defensin-4A and 4B. Several of the induced AMP genes may also alter keratinocyte differentiation and proliferation. [25] Our findings are reminiscent of the situation in lesional psoriatic skin, [25] except that the expression of Rnase7, LL-37 and dermcidin was not induced in the ARCI patients with TGM1 mutations. Typically, neither of these conditions is prone to skin infections despite the barrier abnormalities. However, other subtypes of ichthyoses, that is, harlequin ichthyosis caused by ABCA12 mutations, Netherton syndrome caused by SPINK5 mutations and epidermolytic ichthyosis caused by KRT1 or KRT10 mutations show increased prevalences of secondary skin infections, which can be attributed both to a more severe barrier failure and to distinctive abnormalities in the lamellar bodies secretory system that collectively decrease the bioavailability of AMPs. [26] Speculatively, a stimulation of innate immune responses in patient skin might be a consequence of the observed reduction of IL37, which is a known suppressor of innate immunity. [27] Furthermore, an enhanced adaptive immunity, reflected in increased IL36G transcription in the array and qPCR analyses, may in turn induce various chemokines and psoriasin (S100-A7), [28][29][30] which was also observed in our study.
It is well established that the Th17/IL-23 pathway is induced in psoriasis [31] and a previous study of ARCI patients with TGM1 mutations could link this pathway to ichthyosis severity and inflammation. [32] Furthermore, an increased IL-17/IL-22 activation in peripheral blood in patients with lamellar ichthyosis was shown to correlate with clinical variables in a recent study. [33] Indeed, we could confirm increased expression of several of the previously reported IL-17/TNFα-regulated genes, namely IL36G/IL1F9, PI3, CCL20, LCN2, DEFB4 and S100A9, whereas the expression of the cytokines IL17A, IL17C and TNFA was virtually unchanged.

| Effects on other genes related to keratinocyte differentiation
Earlier studies have revealed five genes essential for psoriasis development: PPARD, GATA3, TIMP3, WNT5A and PTTG1. [34] In the present study, we also found an altered expression of PPARD, GATA3 and WNT5A. As a corollary, we note that a PPARβ/δ agonist has recently been shown to induce CERS3 and ELOVL4 in cultured keratinocytes. [35]

| Proposed mechanisms for transcriptional changes associated with TGM1 mutations
The observed upregulation of a multitude of genes in the skin of patients with TGM1 mutations probably represents a compensatory mechanism for the severe barrier dysfunction elicited by a marked reduction or total lack of TGm-1 activity. The eliciting factors involved in such a hypothesized compensatory mechanism are still unknown, but increased water flux in stratum corneum, [36] abnormal signalling by epidermal lipids and a disrupted [Ca 2+ ]-gradient might all be involved. Actually, the latter proposal is supported by the marked induction of GJB2 observed in the ARCI patients. In the GJB2-deficient mice model for Keratitis-ichthyosis-deafness (KID) syndrome, an altered calcium distribution in epidermis results in abnormal lipid processing, [37] thus implying GJB2 as an essential factor in barrier homoeostasis. Further studies are required to clarify whether the changed expressions of many ARCI genes in patients with TGM1 mutations are unique for this type of ichthyosis or represents ingredients of a general barrier repair mechanism.

| Study limitations
Although this study involved only 5 patients (and 4 controls), the results were in many cases pronounced and principally the same patterns were observed in four of the patients. However, patient P4 (age 80, homozygous for a splice site mutation in TGM1 and acitretin-treated for 20 years) showed more equivocal results despite overt lamellar ichthyosis. Incidentally, all patients except P1 were dependent on systemic retinoid therapy that could not be interrupted for a minimum of 1 month required to obtain treatment-naive skin biopsies. Therefore, we cannot entirely rule out the possibility that some of the observed changes in ARCIcausing and other genes were affected by retinoids. In an indirect attempt to address this possibility, we exposed human epidermal equivalents (HEE) generated from wild-type keratinocytes to alltrans retinoic acid (atRA) in vitro (see Supporting Information) and found reduced expression of almost all ichthyosis-causing genes (see Figure S1). This response is similar to previous results in monolayer keratinocytes exposed to atRA [17] and contradicts retinoid therapy as a cause of the abnormal gene expression changes in the patients. Primary keratinocytes from three of these patients with TGM1 mutations were isolated for functional studies, but the phenotypic characteristics were lost and problems in generating HEEs using these cells were encountered (see Supporting Information).
Furthermore, we previously analysed several retinoid-regulated genes in healthy volunteers exposed to topical atRA [38] and lamellar ichthyosis patients treated systemically with the retinoic acidmetabolism blocking agent, liarozole, [39] and among all genes found to be upregulated in these reports, only CRABP2 was upregulated in the ARCI patients of this study. It is worth noting that in a previous study of CRABP protein levels in epidermis, increased expressions were observed in untreated patients with lamellar ichthyosis and psoriasis. [40] Based on all these evidences, we conclude that acitretin therapy is an unlikely explanation to the transcriptional changes observed in patients P2-P5.

| CON CLUS ION
The present study shows that disruptive mutations in TGM1 are associated with marked upregulations of several other ichthyosisrelated and lipid biosynthesis genes, as well as modulations of gene expression important for CE formation and immune or inflammatory regulations, together probably constituting a repair mechanism for the epidermal barrier function.

ACK N OWLED G EM ENTS
We acknowledge the work of the Array and Analysis Facility,

CO N FLI C T O F I NTE R E S T
The authors have no conflict of interest to declare.  Table S1. TaqMan qPCR detection primers and probes used. Table S2. Primary antibodies used. Table S3. Differentially expressed genes (DEGs) with UniGene ID and adjusted P < 0.05 in ARCI patients (n = 5) with TGM1. Table S4. DEGs with more than 1.5 log2fold-change used in functional annotation cluster analysis with DAVID. Table S5. Functional annotation clusters of DEGs with more than 1.5 log2fold-change from gene ontology analysis with DAVID.