Persistent Cutaneous Lesions of Darier Disease and Second-Hit Somatic Variants in ATP2A2 Gene

Key Points

Question

What is the molecular mechanism underlying the development of persistent skin lesions in Darier disease (DD)?

Findings

In this prospective case series of 9 patients with DD, somatic second-hit variants in the ATP2A2 gene were identified in 11 persistent skin lesions, whereas no somatic variants were found in transient lesions and normal skin.

Meaning

Somatic second-hit variants in the ATP2A2 gene may be associated with the development of persistent cutaneous lesions in DD.

This case series examines the molecular mechanisms underlying the development of persistent skin lesions in Darier disease.

Abstract

Importance

Darier disease (DD) is a rare genetic skin disorder caused by heterozygous variants in the ATP2A2 gene. Clinical manifestations include recurrent hyperkeratotic papules and plaques that occur mainly in seborrheic areas. Although some of the lesions wax and wane in response to environmental factors, others are severe and respond poorly to therapy.

Objective

To investigate the molecular mechanism underlying the persistency of skin lesions in DD.

Design, Setting, and Participants

In this case series, DNA was extracted from unaffected skin, transient and persistent lesional skin, and blood from 9 patients with DD. Genetic analysis was used using paired-whole exome sequencing of affected skin and blood or by deep sequencing of ATP2A2 of affected skin. Chromosomal microarray analysis was used to reveal copy number variants and loss of heterozygosity. All variants were validated by Sanger sequencing or restriction fragment length polymorphism.

Interventions or Exposures

Paired whole-exome sequencing and deep sequencing of ATP2A2 gene from blood and skin samples isolated from persistent, transient lesions and unaffected skin in patients with DD.

Main Outcomes and Measures

Germline and somatic genomic characteristics of persistent and transient cutaneous lesions in DD.

Results

Of 9 patients with DD, all had heterozygous pathogenic germline variants in the ATP2A2 gene, 6 were female. Participant age ranged from 40 to 69 years on enrollment. All 11 persistent skin lesions were associated with second-hit somatic variants in the ATP2A2 gene. The somatic variants were classified as highly deleterious via combined annotation-dependent depletion (CADD) scores or affect splicing, and 3 of them had been previously described in patients with DD and acrokeratosis verruciformis of Hopf. Second-hit variants in the ATP2A2 gene were not identified in the transient lesions (n = 2) or the normal skin (n = 2).

Conclusions and Relevance

In this study, persistent DD lesions were associated with the presence of second-hit somatic variants in the ATP2A2 gene. Identification of these second-hit variants offers valuable insight into the underlying mechanisms that contribute to the lasting nature of persistent DD lesions.

Introduction

Darier disease (DD) is a rare genetic skin disease caused by heterozygous loss of function (LOF) variants in the ATP2A2 gene, encoding the sarcoplasmic reticulum calcium ATPase pump, SERCA2.^1,2 DD has a wide phenotypic presentation involving skin, mucous membranes, and nails. The predominant clinical manifestation is widespread crusted or keratotic papules located over the scalp, face, neck, and upper trunk, which wax on environmental triggers such as UV, heat, sweating, and occlusion, and wane.³ In contrast to these transient lesions, some phenotypic features of DD are persistent and less likely to improve with therapy. Nearly 50% of patients with DD develop skin-colored flat-topped papules that are clinically and histologically indistinguishable from acrokeratosis verruciformis of Hopf (AKV), an allelic disorder of DD caused by variants in the ATP2A2 gene.^3,4,5 Persistent keratotic and comedonal papules and plaques were also reported,^6,7,8 raising the question of whether these are driven by additional genetic events that result in their more severe appearance and persistency.

The development of disease in a given organ/tissue, on the background of autosomal dominant variant is well-known from several genetic tumor/hamartoma syndromes such as neurofibromatosis, nevoid basal cell carcinoma syndrome, and PTEN-hamartoma syndrome. Although certain phenotypic features of these disorders such as macrocephaly and developmental delay are due to haploinsufficency, tumor development follows the Knudson 2-hit theory in which a somatic second-hit variant in the wild-type allele results in loss of heterozygosity (LOH)^{9,10,11,12,13} through copy-neutral LOH, copy-loss LOH or de-novo point variant in the wild-type allele.^14,15

Considering the difference in the biological behavior of transient and persistent DD skin lesions, we hypothesized that additional somatic events could explain DD lesion persistency. We therefore used next-generation sequencing (NGS) analysis of affected skin from 13 persistent or transient skin lesions in 9 patients.

Methods

Patients and Lesion Definitions

Patients were recruited to the study based on the existence of both the clinical and histopathologic features of DD.¹ The participants provided their written informed consent prior to their inclusion in the study, according to the protocols approved by the Emek and Rabin Medical Centers (EMC-0086-15, RMC-0312-20) institutional review board, which approved the study.

Persistent lesions were defined as characteristic cutaneous lesions of DD (severe thick keratotic papules and plaques with/without erosions, AKV-like lesions, plaques seeded with comedos and isolated comedos), which had not waxed and waned in the previous 2 years and had not improved under topical antibiotic/steroid treatments. Persistence of lesions was confirmed in serial photographs taken in the clinic during follow-up before the study initiation. Transient lesions were less severe lesions defined as characteristic cutaneous lesions of DD that completely responded to therapy during follow-up, leaving normal skin or dyspigmentation following therapy. Those lesions were photographed before and after treatment and biopsied on recurrence.

We included persistent lesions from participants DD1 to DD4, DD6, and DD7 to DD9, transient lesions from participants DD3 and DD5, and normal skin samples from participants DD3 and DD9 (Figure 1). In participants DD1, DD4, DD6, and DD7 to DD9 we could not clearly identify transient lesions. Participant DD2 had transient lesions on their face and was reluctant to undergo biopsy from this area.

Figure 1. Persistent and Transient Skin Lesions in Participants With Darier Disease (DD).

Flat topped papules over dorsal hands and feet; verruciformis of Hopf–like lesions (A, F, and G). B, Hyperkeratotic papules and plaques. D, Comedonal Darier. Treatment responsive transient lesions over submammary areas (C), and trunk (E). Erosive thick papules and plaques (H). Black arrowhead and asterisks denote biopsy site of persistent and transient lesions, respectively.

Blood samples were obtained from all participants, followed by DNA extraction from leukocytes using a DNA isolation kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. Following a 4-mm biopsy collection from persistent/transient skin lesions or normal skin, epidermal sheaths were separated using dispase as previously described,¹⁶ and DNA was extracted as described.

Whole-Exome Sequencing

DNA of blood and affected tissue from patients DD1 to DD5 and of normal and affected tissue from patients DD8 and DD9 was sheared and barcoded, and whole-exome capture was performed (Twist Bioscience Human Core Exome kit +RefSeq panel). Illumina NovaSeq6000 instruments were used for high-throughput sequencing with 100 bp paired-end reads (Illumina) with an average variant depth of 150X.

We used the Genoox Data Analysis Platform (version 68) for data analysis; germline sequencing reads were aligned using the Burrows-Wheeler Aligner (BWA) against Human Reference Genome Build 38 (hg38).¹⁷ Duplicate reads were marked and removed using BamSorMaDup. Resulting BAM files were calibrated using GATK.¹⁸ Germline single nucleotide variants (SNVs) and indels were identified using HaplotypeCaller,¹⁹ somatic SNVs and indels were identified using MuTect2. All variants were annotated for functional effect with ANNOVAR.²⁰ The resultant SNVs and indels were filtered to exclude those with a somatic variant allele frequency of 3% or lower and a prevalence of more than 0.1% in gnomAD repository. Rare SNVs and INDELs were filtered based on pathogenicity using REVEL and scaled CADD scores.^21,22 Aligned reads were examined with the Broad Institute Integrative Genomics Viewer to exclude variants resulting from miscalls and alignment error.

Targeted ATP2A2 Sequencing

High depth amplicon-based sequencing of ATP2A2 was performed on DNA extracted from persistent skin lesions of DD6 (I, II [biopsies 1 and 2]) and DD7. An Illumina Miseq instrument was used with 250 bp paired-end reads with an average variant depth of 10 000X. Data analysis was performed on Genoox Data Analysis Platform with the same filtering criteria as used for whole-exome sequencing (WES) data.

Sanger Sequencing

Validation of ATP2A2 variants was obtained through Sanger sequencing using primer pairs (Figure 2, eFigure 2, eTables 2 and 3 in Supplement 1).

Figure 2. Somatic Second-Hit Variants in ATP2A2-Associated Persistent Cutaneous Lesions of Darier Disease.

Somatic ATP2A2 variants were confirmed via Sanger sequencing (A, B, D-F, and H-K), and polymerase chain reaction–restriction fragment length polymorphism analysis of DD3 and DD6-II (C and G). Agarose gel electrophoresis of digested amplicons of affected tissue and blood showing specific bands of the variant alleles that are absent in blood. The roman numerals indicate the first and second biopsies of the patient.

Polymerase Chain Reaction–Restriction Fragment Length Polymorphism Analysis

In cases DD3-I and DD6-II, where Sanger sequencing did not detect clear somatic variants, variant allele amplification was used. Briefly, to maximize the detection of variant alleles at low percentage mosaicism, we selected specific restriction enzymes which digest the normal alleles. Primer sequences and restriction enzymes for each variant are given in eTable 4 in Supplement 1. Touchdown polymerase chain reaction program with 35 cycles and 25 cycles was used throughout, and products electrophoresed on a 3% agarose gel.

Copy Number Variation and LOH Analysis

For WES data we used a sliding window to detect runs of homozygosity (ROH) with a minimal size of 1 mega byte where there are at least 18 homozygous variants and no more than 2 candidate heterozygous variants. For DD6-I, DD6-II and DD7, which underwent targeted sequencing, we used SNV genotyping to plot b-allele frequency across chromosome 12 (eFigure 3 in Supplement 1).

Results

We recruited 9 patients with DD. All had classic clinical and histopathologic characteristics of DD. Their demographics and clinical presentations are detailed in eTable 1 in Supplement 1.

Of the 9 participants, 6 were female. Participant age ranged from 40 to 69 years on enrollment. Eight patients had a family history of DD. Eight patients had classic disease distribution with hyperkeratotic papules and plaques over seborrheic areas and v-shaped notching of the nails (Figure 1). Seven patients had acral flat papules reminiscent of AKV lesions (Figure 1, A and G). One patient had comedonal DD, manifested as keratin plugged brown-black papules and plaques over forehead, cheeks and trunk (Figure 1D). Three patients had persistent thick brown skin-colored papules and plaques over the torso (Figure 1, B and C), whereas some of their other cutaneous lesions were thinner and responded to topical corticosteroids or acitretin through follow-up (Figure 1, C and E).

Persistent Cutaneous Lesions of DD May Be Associated With Second-Hit Somatic Variants in the ATP2A2 Gene

To investigate whether somatic variants have developed in persistent and transient DD lesions, we performed WES of affected skin and blood or normal skin (Table). In 6 of 6 patients with 8 persistent lesions (DD1-4, DD8-9), we identified germline and second-hit somatic variants in ATP2A2 (Table, Figure 2). All were rare variants in gnomAD. Six variants predicted to have a pathogenic effect on the encoded protein according to scaled CADD score and REVEL; one of them had been previously reported in patients with DD.^16,17 Two variants were predicted to affect splicing (Table). Two distinct lesions were tested in DD3 and DD8 finding different somatic ATP2A2 variants in different lesions of the same patient.

Table. Second-Hit Somatic Variants of ATP2A2 Result in Persistent Darier Disease Skin Lesions.

Sample ID	Germline variant	CADD score	REVEL	Type of lesion	Natural history of lesion	Somatic variant	CADD score	REVEL	VAF in skin, %	VAF in blood, %	Genetic analysis approach
DD1	c.1000C>T, p.Arg334X	36	NA	AKV-like lesion	Persistent	c.1912A>T, p.Ile638Phe	25.8	Pathogenic	13	0	WES
DD2	c.2256_2256Dup, p.(Tyr753Leufs*60)	NA	NA	Keratotic papules in seborrheic areas	Persistent	c.2747C>T, p.Ser916Phe	32	Pathogenic	7	0	WES
DD3-I	C.530A>C, p.Gln177Pro	27.8	Pathogenic	AKV-like lesion	Persistent	c.816G>A, p.Arg272X	36	NA	3.5	0	WES
DD3-II				Thin papule on torso	Wax and wane	Not identified	NA	NA	NA	NA	WES
DD3-III				Erosive papule on torso	Persistent	c.2859 + 1G>T	NAa	NAa	9	0	WES
DD3-IV				Normal skin		Not identified					WES
DD4	c.1406_1415del, p.(Asn469Thrfs*7)	NA	NA	Comedo	Persistent	c.1484C>T, p.Ser495Leub	32	Pathogenic	3.5	0	WES
DD5	c.395A>C, p.Gln132Pro c.392G>T, p. Arg131Leu	22.9/ 29.9	Pathogenic	Thin papule on torso	Wax and wane	Not identified	NA	NA	NA	NA	WES
DD6-I	C.1582C>T, p.Arg528X	37	NA	AKV-like lesion	Persistent	c.2759C>T, p.Ser920Pheb	29	Pathogenic	20	0	Targeted sequencing
DD6-II				AKV-like lesion	Persistent	c.2759C>A, p.Ser920Tyrb	27.4	Pathogenic	7	0	Targeted sequencing
DD7	c.1327 A>C, p.Thr443Pro	27.4	Pathogenic	AKV-like lesion	Persistent	c.348C>G, p.Ile116Met	22.9	Uncertain	16	0	Targeted sequencing
DD8	C.530A>C, p.Gln177Pro	27.8	Pathogenic	Erosive papule on torso	Persistent	c.775C>T, p.Gln259Ter	37	NA	6	NA	WES
				Erosive papule on torso	Persistent	c.1542 + 6T>A	NAa	NAa	6	NA	WES
DD9	C.530A>C, p.Gln177Pro	27.8	Pathogenic	Erosive papule on torso	Persistent	c.815G>A, p.Trp272Ter	37	NA	5	0 in Normal skin	WES
				Normal skin		Not identified	NA	NA	NA	NA	WES

Abbreviations: AKV, acrokeratosis verruciformis; CADD, combined annotation-dependent depletion; DD, Darier disease; NA, not applicable for nonmissense variants; VAF, variant allele frequency; WES, whole-exome sequencing.

^a Predicted to alter splicing by SpliceAI.

^b Previously reported as a germline variant in DD.

Next, we looked for somatic ATP2A2 variants in transient DD lesions from DD3 and DD5. DD5 had 2 germline pathogenic ATP2A2 variants in cis (eFigure 3 in Supplement 1). No somatic ATP2A2 variants were identified (Figure 1, C and F, Table). There were no somatic ATP2A2 variants in normal skin from DD3 and DD9.

The chronically sun-exposed location of the persistent AKV-like and comedonal DD lesions used for analysis and the finding that 5 of 8 SNVs were C>T or G>A transitions raised the possibility that second hits resulted from UV light-induced mutagenesis. To test this, we compared variant burden, extent of UV signature SNVs, and the rare damaging coding SNVs in persistent (n = 8), transient (n = 2) lesions and normal skin (n = 2), and found no difference between sample types (eFigure 1 in Supplement 1).

Deep Sequencing of ATP2A2 Identified Somatic Variants in Persistent Cutaneous Lesions of DD

Considering our WES data supporting second-hit variants in the ATP2A2 gene in persistent cutaneous lesions of DD, we sought to expand our cohort and test our hypothesis through deep sequencing of the ATP2A2 gene in additional persistent DD lesions. First, we identified heterozygous pathogenic germline ATP2A2 variants in blood and skin samples of DD6 and DD7 with Sanger sequencing (eFigure 2 in Supplement 1). In participant DD6, deep sequencing revealed 2 different somatic ATP2A2 variants in 2 distinct AKV-like lesions biopsied from a patient’s dorsal hand (Figure 1F). Both variants involve cytosine in position 2759, exchanging amino acid serine, located between transmembrane loops 8 and 9 (c.2759C>T, p.Ser920Phe, and c.2759C>A. p.Ser920Tyr, respectively) (Figure 2, F and G). Both variants had been previously reported in patients with DD.^23,24,25 Biochemical analysis of c.2759C>A revealed major reduction of calcium pump activity and enhanced coprecipitation with the wild-type pump.²⁵ ATP2A2 deep sequencing of DNA isolated from an AKV-like lesion from participant DD7 (Figure 1G) identified a somatic c.348C>G, p.Ile116Met variant (Table, Figure 2H). ROH and BAF analyses did not reveal somatic LOH in any of the persistent and transient samples (eFigure 3 in Supplement 1).

Discussion

In this prospective case series study we aimed to identify the molecular mechanism that differentiates persistent from transient DD skin lesions. Whereas persistent DD lesions tend to be thicker, with minimal response to therapeutic interventions, transient DD lesions are thinner and respond to therapy. We identified somatic second-hit variants in the ATP2A2 gene in all persistent DD lesions (n = 7) from either seborrheic papules/plaques or AKV-like lesions, whereas transient DD lesions harbored the germline ATP2A2 variant only. Six of 11 variants involved the substitution of C to T or G to A, which could result from UV-light induced mutagenesis. Considering lesion locations (dorsal hands and upper trunk), it is plausible that UV was involved. Nevertheless, the other 3 lesions had other variants, including 1 sample from close proximity of C>T lesion (DD6-I and II. Table, Figure 2G), suggesting that other mechanisms are involved.

DD is caused by heterozygous germline variants in the ATP2A2 gene. The presumed genetic pathomechanism of DD is haploinsufficiency, wherein 1 copy of the ATP2A2 gene is inactivated due to loss of expression or loss of function effect,^26,27,28,29 and the remaining allele ATP2A2 product is not sufficient to preserve a normal function, especially in the presence of endoplasmic reticulum (ER) stressors such as UV exposure and infections that stimulate cytokine secretion downregulating ATP2A2 expression from the intact allele.^30,31,32 Decreased ATP2A2-encoded protein, SERCA2 protein level, leads to depletion of the ER Ca²⁺ storage, which further results in disassembly of desmosomes, induction of apoptosis and differentiation impairment.³² This hypothesis might explain the recurrent flares of patients with DD and the effect of sun exposure, sweating, and other physical triggers on disease severity. Although transient stimulation triggers the development of transient DD lesions via decreased ATP2A2 expression, it does not provide an explanation for the development of fixed, proliferative lesions that are treatment-resistant on seborrheic areas, dorsal hands, feet, palms, and soles.

The concept of somatic second hit is well described in oncologic diseases. However, during recent years several nonneoplastic, cutaneous proliferative lesions were also found to be driven by second-hit mechanism. Porokeratosis is a keratinization disorder caused by variation in genes of the mevalonate pathway. Patients with disseminated superficial actinic porokeratosis had mono-allelic congenital germline variants in MVK and MVD genes, whereas their cutaneous lesions presented additional somatic variants in the affected genes either by mitotic recombination or through C>T transition variants of the wild-type allele.³³ Another second-hit mechanism is the monoallelic 2-hit mechanism in PLCD1 in patients with frequent hair-bearing trichilemmal cysts, where a specific high-risk germline variant in PLCD1 gene predisposes to second-hit somatic variant on the same allele, abrogating downstream signaling.³⁴ While somatic second-hit is required for the development of clinical manifestations in porokeratosis/trichilemmal cysts, ATP2A2 germline variant is sufficient to cause DD lesions.

The clinical differentiation between transient and persistent DD lesions bears resemblance to the distinction observed in type 1 and type 2 mosaic DD. Type 1 mosaic DD involves a dominant, heterozygous postzygotic somatic variant occurring in a subset of precursor cells in an otherwise wild-type individual, and leading to a patch or stripe of affected skin.³⁵ Typically, these skin lesions present during the second decade of life and exhibit a favorable response to therapeutic interventions. On the other hand, type 2 mosaic DD entails an independent postzygotic ATP2A2 variant arising in a subpopulation of precursor cells in an individual who already carries a germline heterozygous ATP2A2 variant.^36,37 This results in the development of more severe lesions earlier in life that display reduced responsiveness to treatment.³⁶ Similarly, transient acantholytic dermatosis (Grover disease), sharing clinical and histopathologic features with DD, manifests skin lesions during the sixth decade of life or later and, as the name suggests, they are transient in nature. Recently, somatic ATP2A2 variants have been identified in Grover disease skin lesions.³⁸ Our findings are consistent with existing knowledge from mosaic DD and Grover disease, bolstering the concept that second hits may lead to more severe and persistent lesions, less responsive to therapy, whereas somatic heterozygosity in DD contributes to lesion transiency.

Limitations

Limitations of this study include the examination of a greater number of persistent than transient lesions. Although second hits were demonstrated we did not study whether they were mono- or bi-allelic. The contribution of other somatic variants to disease pathogenesis has not been studied but the recurrence of the ATP2A2 somatic variant strengthens their contribution to lesion severity.

Conclusions

In this prospective case series study, we found that second-hit somatic variants in the ATP2A2 gene may be associated with persistent DD lesions. These results suggest that second hits contribute to lesion persistency, whereas environmental factors leading to decreased expression of the wild-type allele result in the formation of less severe/transient DD lesions.

Supplement 1.

eTable 1. Demographic and Clinical characteristics of study cohort

eTable 2. Primer sets used for validation of germline variants

eTable 3. Primer sets used for validation of somatic variants

eTable 4. PCR- Restriction fragment length polymorphism (RFLP) analysis

eFigure 1. Somatic SNV burden in Darier disease skin lesions

eFigure 2. Germline pathogenic variants in ATP2A2 gene

eFigure 3. SNP genotyping across chromosome 12 in persistent skin lesions

eFigure 4. Sanger sequencing did not verify the somatic ATP2A2 variants in DD3 and DD6-II

Supplement 2.

Data Sharing Statement