ALPK1 hotspot mutation as a driver of human spiradenoma and spiradenocarcinoma

Abstract

Spiradenoma and cylindroma are distinctive skin adnexal tumors with sweat gland differentiation and potential for malignant transformation and aggressive behaviour. We present the genomic analysis of 75 samples from 57 representative patients including 15 cylindromas, 17 spiradenomas, 2 cylindroma-spiradenoma hybrid tumors, and 24 low- and high-grade spiradenocarcinoma cases, together with morphologically benign precursor regions of these cancers. We reveal somatic or germline alterations of the CYLD gene in 15/15 cylindromas and 5/17 spiradenomas, yet only 2/24 spiradenocarcinomas. Notably, we find a recurrent missense mutation in the kinase domain of the ALPK1 gene in spiradenomas and spiradenocarcinomas, which is mutually exclusive from mutation of CYLD and can activate the NF-κB pathway in reporter assays. In addition, we show that high-grade spiradenocarcinomas carry loss-of-function TP53 mutations, while cylindromas may have disruptive mutations in DNMT3A. Thus, we reveal the genomic landscape of adnexal tumors and therapeutic targets.

Fig. 1

The driver gene landscape of skin adnexal tumors. Genomic data for the 52 cases where matched tumor/normal DNA-sequencing data were available. Additional cases are shown in Supplementary Fig. 5. The germline and somatic mutations in this plot were validated by high-depth targeted exome sequencing. Only mutations in coding regions are shown, except for TERT promoter variants and the splice region mutation in CYLD. Note that PD30271 and PD29730 are the same patient who had multiple tumors analyzed

Fig. 2

Mutations identified in CYLD and ALPK1. Variants in CYLD (a) and ALPK1 (b) against the translation of the longest transcript of these genes (ENST00000458497.5 and ENST00000311559.13). Protein domains are from UniProt. All of the variants shown were validated by high-depth targeted exome sequencing. Adjacent normal represents morphologically normal tissue from the same block as the tumor which was used as a germline sample for somatic variant calling. Variants in red were called somatically. Variants in green were called from the adjacent normal tissue. The color of the circles indicates tumor/tissue type. The somatic splice region mutation in PD29703a in CLYD is not shown. c Protein alignment of ALPK1 across vertebrates. The conservation score represents constrained elements in multiple alignments by quantifying substitution deficits. The arrow indicates the position of the p.V1092 residue in humans

Fig. 3

The somatic genetic landscape of adnexal tumors. a The contribution of published mutational signatures in adnexal tumors was computed using deconstructSigs. Total contribution per sample adds up to one. For this analysis, we used all variants, including those in noncoding regions, such as 5′ and 3′ UTRs. b The copy-number landscape of adnexal tumors. This analysis was performed using Sequenza to define the absolute copy number for chromosomal segments. These analyses were performed using the tumors shown in Fig. 1

Fig. 4

Assessment of the MYB-NFIB fusion and p65 expression in adnexal tumors. a Fluorescence in situ hybridization (FISH) imaging of the MYB-NFIB fusion in an adenoid cystic carcinoma and assessment in cylindroma samples. Previous reports have suggested that adnexal tumors, such as cylindomas, carry MYB-NFIB fusions which have been associated with MYB overexpression. The left panel shows an adenoid cystic carcinoma, which is a positive control for the fusion event. Yellow signal results from the overlap of the green NFIB probe and red MYB probe. Right panel: a representative cylindroma which was fusion negative. b Representative histopathological images of a cylindroma at ×100 magnification, spiradenoma at ×100 magnification, high-grade spiradenocarcinoma at ×200 magnification, and a low-grade spiradenocarcinoma at ×200 magnification. c p65 immunohistochemistry of a CYLD mutant cylindroma (left) and an ALPK1 p.V1092A mutant spiradenoma (right) at ×20 magnification