Structural organization, mapping, characterization and evolutionary relationships of CDKN2 gene family members in Xiphophorus fishes

Abstract

Xiphophorus fishes and their hybrids are used as models for the study of melanoma and other diseases. The cyclin-dependent kinase inhibitor gene family in humans is comprised of four members, including CDKN2A (P16), and dysregulation of this gene is implicated in numerous neoplasms including melanomas. We have investigated the status of the gene family in the southern platyfish X. maculatus. Xiphophorus harbors at least two such loci, which we now term CDKN2A/B and CDKN2D. Both loci map to Xiphophorus linkage group 5, a genomic area that has long been known to harbor the DIFF tumor suppressor locus. Within this report, we report on the complete cloning, genomic exon/intron boundary delineation, linkage mapping and expressional characteristics of Xiphophorus CDKN2D. We also compare and contrast this expression to that of the previously isolated CDKN2AB locus in normal and neoplastic tissues derived from non-hybrid and hybrid fishes. The hypothetical evolutionary relationships of gene family members and their involvement in melanoma is evaluated. In comparison to CDKN2A/B, the RNA expression of Xiphophorus CDKN2D differs in normal tissues and is not associated with melanotic/pathologic tissues, confirming functional divergence between obvious homologues.

Introduction

Xiphophorus is a freshwater fish genus currently comprised of 25 taxa. Hybrids between these distinct species can be produced by artificial insemination or by simple co-habitation of divergent species. Hybrid crosses between the southern platyfish, X. maculatus (Jp 163 B strain), and the green swordtail, X. helleri (Rio Sarabia strain), results in viable F₁ hybrids that are oftentimes used to create backcross hybrids (to X. helleri). The derived hybrid fishes often show abnormalities in sex determination, maturation and lack of gene regulation of critical gene targets. Some first generation backcross hybrids (BC₁) hybrids develop melanoma spontaneously, while others do not. This genetic predisposition is predominantly based on the genetic constitution at two key genomic regions within Xiphophorus linkage group (LG) 24 and LG 5 (Kazianis and Walter, 2002, Walter and Kazianis, 2001). LG 24 harbors an oncogene, XMRK that is constitutively expressed in BC₁ hybrid melanomas and also shows key mutations in its extracellular domain, rendering the derived protein independent of a need for a secretory ligand. While XMRK(Volff and Schartl, 2002) does not appear to be orthologous to any loci in mammals, the closest homologue is the epidermal growth factor receptor (EGFR). XMRK is derived from the X. maculatus progenitor and is properly regulated in a purebred background (Gomez et al., 2001, Gomez et al., 2002).

It has long been known that Xiphophorus LG 5 harbors a locus referred to as DIFF (Ahuja et al., 1980, Siciliano et al., 1976). This locus, discovered by classical genetics, theoretically equates to a tumor suppressor locus. BC₁ hybrids that are homozygous for a genomic region within LG 5 and inherit XMRK, are prone to develop melanoma. Sibling fishes that retain a DIFF allele from X. maculatus are less prone to develop spontaneous melanotic growths. If one accounts for the two genomic regions by genotyping studies in BC₁ hybrids, 80% of melanomas can be accounted for (Kazianis et al., 1998, Morizot et al., 1998, Nairn et al., 2001).

The CDKN2 gene family in mammals is comprised of 4 gene family members, CDKN2A, CDKN2B, CDKN2C and CDKN2D (Ortega et al., 2002, Ruas and Peters, 1998). These genes are alternatively known as the INK4 loci. All code for proteins with virtually indistinguishable biochemical characteristics, binding to cdk4 and cdk6 kinases and allosterically inhibiting interactions between these and the cyclin D family of proteins (cyclin D1, D2 and D3) (Ortega et al., 2002, Ruas and Peters, 1998). The CDKN2A locus of humans and several other mammals is distinguished from other CDKN2 loci, since it actually codes for two proteins, p14ARF and p16CDKN2A through use of an alternate exon and reading frame (Ruas and Peters, 1998, Satyamoorthy and Herlyn, 2002).

Germline and/or somatic mutations and deletions, and expressional inactivation through promoter methylation in CDKN2 loci, have all been implicated in carcinogenesis in a wide-variety of tumors (Ortega et al., 2002, Ruas and Peters, 1998). The CDKN2A locus, in particular, has been strongly implicated in familial and sporadic melanoma, showing all mentioned etiologic combinations (Ruas and Peters, 1998, Satyamoorthy and Herlyn, 2002). Since Xiphophorus fishes serve as a paradigm for the study of melanoma, we historically initiated our cloning efforts in hope of identifying a CDKN2A orthologue. A homologous sequence was subsequently cloned that showed obvious homology to the CDKN2A, but also to CDKN2B and CDKN2D gene family members. Therefore, the locus was termed CDKN2X (X for Xiphophorus) and this gene codes for a 13-kDa protein (Kazianis et al., 1999, Kazianis et al., 2000, Nairn et al., 1996). CDKN2X was mapped to LG 5, and the map position was consistent with a localization of the melanoma phenotype and the DIFF melanoma suppressor locus (Kazianis et al., 1998, Nairn et al., 1996, Nairn et al., 2001). Experiments have indicated that CDKN2X is highly expressed in melanomas, in contrast to discovered lack of expression seen for CDKN2A in human and rodent-derived melanomas (Kazianis et al., 1999, Kazianis et al., 2000). It is possible, however, in the fish model, that allelic differences in expression between species may be implicated in differential susceptibility in BC₁ hybrid pedigrees, since 80% of individuals with melanomas are homozygous for CDKN2X (with X. helleri alleles) (Kazianis et al., 1998, Nairn et al., 2001).

The discovery of CDKN2X does not preclude the existence of another locus on LG 5 that could be implicated in melanoma formation (Kazianis et al., 1998). This is especially possible since mammalian CDKN2A and CDKN2B loci are tightly linked loci in both rodents and humans (Ortega et al., 2002, Ruas and Peters, 1998). In addition, fundamental questions exist pertaining to the possible functional and expressional redundancy of gene family members, since their biochemical properties are nearly identical. Therefore, we undertook cloning efforts in pursuit of additional gene family members.