Clinical, genetic and experimental studies of the Brooke–Spiegler (CYLD) skin tumor syndrome

Abstract

Brooke–Spiegler syndrome (BSS; a.k.a. tuban tumor syndrome) is an autosomal dominant inherited skin disorder caused by germline mutations in the CYLD tumor suppressor gene. BSS is characterized by multiple skin adnexal tumors, mainly cylindromas and spiradenomas on the head and neck. The tumors are often severely disfiguring and require repeated surgical interventions. Here, we describe a four-generation BSS-family with a novel germline c.1613_1614delGC CYLD mutation that introduces a premature STOP codon predicted to result in a truncated, inactivated CYLD protein. In addition, we present a pilot study describing establishment of the first patient-derived xenografts (PDXs) from cutaneous CYLD-defective cylindromas. Fresh tumor tissues from cylindromas were transplanted into immunocompromised mice to generate PDXs. One xenograft showed progressive tumor growth after 3 months whereas the others remained unchanged in size during the 6 months study period. Histopathological and immunohistochemical analyses of the PDXs revealed that they recapitulate the histological and molecular features of their respective primary tumors, including expression of NTRK3 and the oncogenic driver MYB. In summary, we present the first preclinical BSS-model that morphologically and genetically recapitulates human CYLD-defective cylindromas. This model will be useful for preclinical therapeutic drug testing and for further studies of the molecular pathogenesis of inherited cylindromas.

Keywords:

• Tumor
• syndrome
• craniofacial

Introduction

Brooke–Spiegler syndrome (BSS) is an autosomal dominant inherited skin disorder characterized by the appearance of multiple skin adnexal tumors such as cylindroma, spiradenoma and trichoepithelioma [1]. The tumors typically occur in the head and neck region, particularly on the scalp, and increase in size and number during the lifetime of the patient. They are usually benign but may occasionally undergo malignant transformation. In addition to skin tumors, BSS patients may on rare occasions develop salivary gland tumors morphologically resembling their cutaneous counterparts [1,2]. Cylindromas may rarely also occur in the breast [1,3]. Despite the fact that BSS tumors are benign they can become quite large and adjacent lesions can coalesce (so called turban tumors). The treatment of BSS skin tumors is limited to repeated surgical excisions with or without concomitant skin transplantation. Thus, there is a need for novel, non-surgical treatment options for these patients.

BSS is associated with germline mutations in the CYLD tumor suppressor gene [4]. CYLD encodes an ubiquitin specific protease and functional inactivation of CYLD results in constitutively active NfκB-signaling and resistance to apoptosis, contributing to the development of skin adnexal tumors [5,6]. Recently, the MYB oncogene was also implicated in the pathogenesis of these tumors. Thus, MYB is overexpressed and drives proliferation of inherited, CYLD-defective cylindromas [7] and sporadic, non-BSS cylindromas have MYB-activation through MYB–NFIB gene fusion [8]. Collectively, these studies identify MYB as a potential therapeutic target in cylindromas.

The purpose of this pilot study was to explore the possibility of establishing patient-derived xenografts (PDXs) of CYLD-defective cylindromas for subsequent preclinical therapeutic drug testing and for further studies of the molecular pathogenesis of these neoplasms. In addition, we describe a new CYLD mutation in a BSS-family.

Patients and methods

Patient material

The index patient was a 66-year-old male presenting with multiple cutaneous tumors located predominantly on the scalp (Patient 1; Figure 1). He noticed the first lesions at the age of 25 years and has since then had numerous (>50) tumors removed surgically. The tumors have reached a largest size of approximately 20 mm in diameter. A 75-year-old brother have also developed multiple skin lesions mainly on the scalp since the age of 40 years (Patient 2). His lesions were, however, smaller with a maximum diameter of about 10 mm (Figure 1). Both patients also had scattered tumors on the trunk. Further investigation of the family history revealed that both the father of the two brothers and the daughter of Patient 2 were affected by the same disease (Figure 2).

Figure 1. Nucleotide sequence analysis showing a germline two base monoallelic deletion in CYLD exon 11 in two brothers presenting with multiple cutaneous cylindromas on the scalp (white arrowheads). The asterisk indicates a STOP codon predicted to terminate translation. Nucleotides deleted in the wild-type allele are underlined.

Figure 2. Pedigree of the four-generation family with the Brooke–Spiegler (CYLD) skin tumor syndrome.

Fresh tumor tissue was obtained from two surgically removed lesions from Patient 1 and one lesion from Patient 2. The remaining tumor tissues were fixed in formalin and embedded in paraffin and subjected to routine histopathological examination. Blood samples were collected from both patients and stored at −80 °C. The studies were approved by the Regional Human Ethics Board (Dno: 717–15), and written informed consent was obtained from the patients.

DNA sequence analysis

Total DNA was extracted from blood samples using the DNeasy Blood & Tissue Kit (Qiagen, Venlo, The Netherlands). CYLD coding exons were amplified by PCR using gene-specific primers (Table 1). PCR products were gel-purified and sequenced using the ABI 3730xl DNA Analyzer (Applied Biosystems, Foster City, CA). The resulting sequences were analyzed using the Sequencher 4.10.1 software and compared with known genetic variants reported in dbSNP, ClinVar and COSMIC databases using Wannovar [9].

Table 1. CYLD primers used for PCR amplification and nucleotide sequencing.

Primer	CYLD target	Sequence 5′-3′
Exon 4A_F	Intron 3	TCAGTGTCTTGGAAGCAAACA
Exon 4A_R	Exon 4	AGGAATCCTTGAATGCCCCA
Exon 4B_F	Exon 4	CTACTGGGAAGAGCGGATTT
Exon 4B_R	Intron 4	ACACAGACACACATGAACACA
Exon 5_F	Intron 4	ACCCCTTTTCCTATGGATCGT
Exon 5_R	Intron 5	ACCCCTGTTCTATTTCAATGCTC
Exon 6_F	Intron 5	AGCAAAACCTTATTACTATGGCA
Exon 6_R	Intron 6	GGTGGCAAGGGGTAAAAGAC
Exon 7_F	Intron 6	TCTGTCCTCAAATCCACTGTG
Exon 7_R	Intron 7	CAGAAAGCCAGAGACACACA
Exon 8_F	Intron 7	GCTGAATTGTATTTTGAGAGGGT
Exon 8_R	Intron 8	AGCAAAACACACAAGACCCT
Exon 9_F	Intron 8	TGACTTTTGGGCTGAAAAATATT
Exon 9_F	Intron 9	ACAGCAGTACGGTTCAAACT
Exon 10A_F	Intron 9	AGCGAGAACACTGTTGGAGA
Exon 10B_F	Exon 10	GAACTCACTGACCACCGAGA
Exon 10A_R	Exon 10	AGACCATGTGAGTTCCCAGG
Exon 10B_R	Intron 10	CGGCAAAGGGAAGTCAGAAA
Exon 11_F^a	Intron 10	CGAAGAGGGAGTGGTGAGAA
Exon 11_R^a	Intron 11	ATAAGTCCCAAACGCACGTC
Exon 12_F	Intron 11	AGCAGTTAACCAAAGCCTACT
Exon 12_R	Intron 12	TCCCTTTCAATCCTGACCCA
Exon 13_F	Intron 12	GGGTATCTTCACAGGGGTCC
Exon 13_R	Intron 13	TGTTCTACACCTTGGCCACA
Exon 14_F	Intron 13	AAAATTAGCTGGATGTGGTGGT
Exon 14_R	Intron 14	AGAACCAGCCAGCTTCTTAAA
Exon 15_F	Intron 14	TCAGGAAAACTTGAAAACTGTGC
Exon 15_R	Intron 15	ATGGTACTGTGGCAAGGACA
Exon 16_F	Intron 15	GTTCTCAAATACTGCTGGGACA
Exon 16_R	Intron 16	TGGTGTTTTCTTTGGCAGCA
Exon 17_F	Intron 16	TGAGCCACTTGAAGACTTGT
Exon 17_R	Intron 17	AGAGTCTAACCCACAACTGA
Exon 18_F	Intron 17	TGAAGCCATGAACATTGTCCT
Exon 18_R	Intron 18	CTTTCCAAAACGAGCATTGTGT
Exon 19_F	Intron 18	AACTCCTGACCTCGTGATCC
Exon 19_R	Intron 19	CACATGCAGAAGAAAGGCGT
Exon 20A_F	Intron 19	GTTGGCCCCAGAATCCTTTC
Exon 20A_R	Exon 20	GTGAACAGAACTGCCAGCTC
Exon 20B_F	Exon 20	ACATTCCTCAAGTCACCCCA
Exon 20B_R	3′-UTR	ACTTCAGGGACACAAAAGCA

^a Primers detecting the c.1613_1614delGC mutation.

Mouse xenografts

Tumor tissue samples (∼4 × 4×2 mm) from two cylindromas from Patient 1 were transplanted subcutaneously into the dorsal flanks of four immunocompromised, non-obese severe combined immunodeficient interleukin-2 chain receptor γ knockout mice (NOG mice; Taconic, Lille Skensved, Denmark) sedated with isoflurane anesthesia. Tumors were measured using a steel caliper and tumor volumes were calculated with the formula V=(length × width²)/2. The animal experiments were approved by the Regional Animal Ethics Committee (Idno: 000036–2014, 000273–2017).

Immunohistochemistry

Formalin-fixed, paraffin-embedded tissue sections were deparaffinized, and antigen epitopes were retrieved with EnVision FLEX Target Retrieval Solution pH 9 (Agilent Technologies, Palo Alto, CA). Slides were rinsed and endogenous peroxidase activity was blocked with the EnVision Flex Mini Kit (Agilent Technologies). Slides were incubated for 1 h at room temperature with antibodies to MYB (SPM175, Santa Cruz Biotechnology, Dallas, TX) or NTRK3 (C44H5, Cell Signaling Technology, Danvers, MA). Bound antibodies were detected with HRP-conjugated secondary antibodies and visualized with the EnVision FLEX DAB + Chromogen substrate. As positive control for MYB and NTRK3 staining we used an adenoid cystic carcinoma (Supplementary Figure S1A) [8]. Control sections were treated identically with the exception that the primary antibodies were replaced by isotype specific IgGs (mouse IgG_2A MAB003 and rabbit IgG AB-105-C purchased from R&D Systems, Minneapolis, MN) (Supplementary Figure S1B). Images were captured with an Eclipse E1000 microscope (Nikon Imaging Inc., Tokyo, Japan) equipped with a ProgRes C7 digital camera (Jenoptik, Jena, Germany).

Results

Clinical and histopathological characteristics

The index patient and his brother presented with multiple cutaneous lesions located predominantly on the scalp and an anamnesis supportive of BSS (Figure 1). The four-generation pedigree revealed that two additional family members were affected, the father of the two brothers and the daughter of Patient 2 (Figure 2). Histologically, tumors from both Patients 1 and 2 showed the typical picture of cylindromas with islands of tumor cells forming a jigsaw pattern. The islands were composed of basaloid tumor cells and were surrounded by a rim of basement membrane material. One tumor from Patient 2 showed a picture consistent with a spiradenocylindroma. None of the tumors showed any signs of atypia or malignant transformation.

A novel germline CYLD mutation

To confirm the clinical suspicion of BSS, we performed targeted Sanger sequencing of the CYLD gene on DNA isolated from peripheral lymphocytes collected from Patient 1. Indeed, nucleotide sequence analysis revealed a novel, previously not reported germline c.1613_1614delGC CYLD mutation (Figure 1). The mutation introduces a premature STOP codon predicted to result in a truncated and inactivated CYLD protein. Sequence analysis of DNA isolated from peripheral lymphocytes obtained from Patient 2 confirmed the presence of an identical CYLD mutation (Figure 1), thus confirming the diagnosis of BSS in this family.

Patient-derived xenografts

To assess the growth potential of CYLD-defective cylindromas in immunocompromised mice, we transplanted fresh cylindroma tissues to the flank of NOG mice from two different cylindromas from Patient 1. Each tumor was transplanted into two mice. One mouse was sacrificed three weeks post-transplantation and the transplanted tumor tissue was removed and subjected to histopathological analysis. The tumor had a large central necrosis surrounded by islands of seemingly intact tumor tissue. The tumor had preserved its original cylindromatous morphology (data not shown). There were signs of nascent angiogenesis in the periphery of the tumor but no mitosis in the tumor islands. Immunohistochemical analysis revealed that MYB protein expression was lower in the xenograft compared to the primary tumor with scattered MYB-positive cells mainly in the periphery of the tumor nodules.

Three months after transplantation, increased tumor growth was detected in one of the three remaining mice (Figure 3(A)). The growth continued for a month after which it ceased. The other two xenografts showed no signs of increased tumor volume during the study period (Figure 3(A)). The mice were sacrificed 6 months after transplantation, after which the remaining tumor tissues were removed for histopathological and immunohistochemical analyses. The morphology of the xenografts was almost identical to the corresponding primary tumors (Figure 3(B)). The xenografted tumors were composed of multiple nodules of uniform, basaloid tumor cells. The nodules were arranged in a jigsaw pattern, typical of dermal cylindroma (Figure 3(B)), and were surrounded by a rim of hyalinized basement membrane material of variable thickness. There were no signs of atypia or of malignant transformation in any of the xenografts.

Figure 3. Growth characteristics and morphology of patient-derived primary cylindroma xenografts (PDXs). (A) Growth curves of three cylindroma PDXs. (B) Hematoxylin and Eosin (HE), MYB and NTRK3 immunostainings in primary cylindromas and PDX tumors. PT 1: primary tumor 1; PDX 1: PDX from primary tumor 1; PT 2: primary tumor 2; PDX 2: PDX from primary tumor 2; PDX 3 (panel A): PDX from primary tumor 2. Scale bars indicate 50 µm.

Immunostaining of the MYB oncoprotein revealed that the majority of the tumor cells in both the primary and xenografted tumors showed positive nuclear staining for MYB (Figure 3(B)). Similarly, the majority of tumor cells both in the primary and xenografted tumors showed positive membranous and/or cytoplasmic staining of NTRK3 (Figure 3(B)), thus confirming that this tyrosine kinase receptor is active also in cylindromas after long-term growth as xenografts in immundeficient mice [10].

Discussion

BSS is a disease in which the affected individuals are genetically predisposed to develop numerous cutaneous adnexal tumors predominantly in the head and neck region [1]. Although the tumors are typically benign, they can nonetheless have significant impact on the quality of life of the affected patients. The tumors may be both painful and disfiguring and usually require repeated surgical interventions. Here, we present a pilot study describing the first PDX-model of this disease for preclinical testing of novel treatment strategies. In addition, we describe a novel CYLD mutation in a BSS-family.

Previous studies of Cyld-defective mouse models have shown that they do not recapitulate the histological and clinical characteristics of BSS particularly well [11,12]. Thus, these mice do not spontaneously develop skin tumors mimicking cylindromas or spiradenomas. Instead, they need to be challenged with topical administration of chemicals before skin tumors appear, suggesting that Cyld-inactivation itself is not sufficient for tumorigenesis. Interestingly, a recent study indicate that specific inactivation of Cyld in K14-positive hair follicle and basal epidermal cells followed by topical challenge with DMBA and TPA results in development of sebaceous and basaloid tumors with some resemblance to BSS-associated tumors [12]. However, since there are no prior reports of established cell lines or xenografts of tumors from BSS patients, the present PDX model is thus the first preclinical model that truly recapitulates human inherited CYLD-defective cylindromas.

Histopathological and immunohistochemical analyses of the PDX tumors revealed that they recapitulate the histological and molecular features of the respective primary tumors. Thus, they showed the typical jigsaw pattern with islands and nests of basaloid tumor cells surrounded by a rim of hyalinized basement membrane material. The xenografted tumors also stained positive for MYB, a driver of growth of both sporadic and CYLD-mutated inherited cylindromas [7,8]. In addition, the PDXs expressed NTRK3 (TRKC), a tyrosine kinase receptor regulating cell survival and differentiation. NTRK3 signaling has previously been shown to promote colony formation and proliferation of cylindromas and the receptor is overexpressed in primary, CYLD-defective tumors [10]. Both MYB and NTRK3 are potential therapeutic targets in CYLD-defective cylindromas [7,10] that can be evaluated using the present preclinical PDX-model. Our studies, although preliminary, demonstrate that the cylindroma xenografts survive and retain their morphological and molecular characteristics for up to at least 6 months in vivo, which is ample time to allow for preclinical treatment studies. The reason why the PDX-tumors did not grow progressively may be due to the lack of specific factors in the mouse subcutaneous tissue microenvironment necessary for continous growth of cylindromas. This may, however, potentially be resolved by co-transplantation of human cylindroma tissues containing both epidermis, tumor, and dermal connective tissue.

Genetic testing of the two affected brothers revealed a novel germline c.1613_1614delGC CYLD mutation in both patients, thus confirming the BSS diagnosis. The mutation is located in CYLD exon 11 and introduces a premature STOP codon predicted to result in a truncated and inactivated CYLD protein (Figure 1). Germline mutations in CYLD are found in 80–85% of patients with a classical BSS phenotype [1,13]. Approximately 100 different CYLD mutations have been identified so far [1,13]. Frameshift and nonsense mutations are most common and are found in approximately 75% of the cases. Most mutations cluster in the 3′-part of CYLD (exons 9–20), which encodes the catalytic residues of the ubiquitin protease, and are predicted to result in truncated CYLD proteins. The novel mutation presented here belongs to this large subgroup of mutations.

In summary, we present the first preclinical BSS-model that truly recapitulates human CYLD-defective cylindromas. Our model will be useful for preclinical therapeutic drug testing and for further studies of the molecular pathogenesis of inherited cylindromas.

Acknowledgements

We thank Ywonne Andrén, Therese Carlsson and Sofia Stenqvist for technical assistance.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This study was supported by the Swedish Cancer Society, BioCARE – a National Strategic Cancer Research Program at University of Gothenburg, and Assar Gabrielsson Cancer Research Foundation.