Abstract
The overexpression of the Xmrk oncogene (ONC-Xmrk) in pigment cells of certain Xiphophorus hybrids has been found to be the primary change that results in the formation of malignant melanoma. Spontaneous mutant stocks have been isolated that have lost the ability to induce tumor formation when crossed with Xiphophorus helleri. Two of these loss-of-function mutants were analyzed for genetic defects in ONC-Xmrk's. In the lof-1 mutant a novel transposable element, TX-1, has jumped into ONC-Xmrk, leading to a disruption of the gene and a truncated protein product lacking the carboxyterminal domain of the receptor tyrosine kinase. TX-1 is obviously an active LTR-containing retrotransposon in Xiphophorus that was not found in other fish species outside the family Poeciliidae. Surprisingly, it does not encode any protein, suggesting the existence of a helper function for this retroelement. In the lof-2 mutant the entire ONC-Xmrk gene was found to be deleted. These data show that ONC-Xmrk is indeed the tumor-inducing gene of Xiphophorus and thus the critical constituent of the tumor (Tu) locus.
MANY individuals of the teleost fish genus Xiphophorus harbor in their genome a tumor locus, Tu (Anders 1991), which contains an oncogenic version of the growth factor receptor tyrosine kinase (RTK) gene Xmrk (Wittbrodtet al. 1989). Generally, Tu does not exert its transforming function, because its activity is suppressed by another unlinked locus, R (Anders 1991). R appears to be involved in the transcriptional control of Xmrk, as overexpression of the RTK gene leads to unrestrained cell proliferation in vitro and tumor formation in transgenic animals (Wittbrodtet al. 1992; Winkleret al. 1994). Earlier concepts, however, had proposed that R acts via modification of pigment cell differentiation (see Anders 1991). The Xmrk oncogene (termed ONC-Xmrk) is always associated with a locus that determines a specific lineage of melanin-containing pigment cells, the macromelanophores (Weis and Schartl 1998). This locus is referred to as the macromelanophore locus and is designated Mdl for macromelanophore determining locus (Weis and Schartl 1998) or MACR (Morizotet al. 1998). Many different versions of Mdl exist, each of which specifies a certain black pigment pattern. Mdl-encoded pigmentation is a complex phenomenon, and the resulting pattern phenotype is modulated by a barely understood polygenic system of modifiers. In the platyfish, at least five different “pseudoalleles” of the macromelanophore locus are known. Mdl was proposed to consist of a number of closely linked genes, each of which can exist in several allelic states in the various populations from different river systems (see Kallman 1975).
The macromelanophore is also the cell type that gives rise to malignant melanoma. Spontaneous melanoma formation has been observed in very rare instances in older fish where obviously transcriptional repression of ONC-Xmrk becomes leaky (Schartlet al. 1995a). In the laboratory, melanoma can be produced at will by eliminating the R-mediated suppression through substitution of R-containing chromosomes (e.g., from Xiphophorus maculatus) by the R-free homologous chromosomes from other species (e.g., X. helleri) in introgressive breeding experiments. Genetic mapping and linkage analysis revealed that the Xiphophorus homologue of the tumor suppressor gene CDKN2 is a candidate gene encoded by the R locus (Nairnet al. 1996; Kazianiset al. 1998). There is, however, another genetic model that is also compatible with the data from the classical crossing experiments, namely that Tu activity is due to the presence of intensifying genes contributed by X. helleri chromosomes to the hybrid genome (see Schartl and Wellbrock 1998).
The Xmrk oncogene probably arose from its corresponding proto-oncogene (termed INV-Xmrk) through a nonhomologous recombination event. This led to a gene duplication of Xmrk. During this process the duplicated copy was fused in its 5′ region to another sequence and thereby became ONC-Xmrk. The gene duplication/recombination event generated a new promoter, which is responsible for the transcriptional activation of the additional Xmrk copy in the macromelanophore lineage of the hybrids (Adamet al. 1993).
In natural populations of Xiphophorus many fish do not contain ONC-Xmrk. Thus in wild fish the oncogene is an additional copy and in evolutionary terms can be regarded as a dispensable gene (Ohno 1985). All populations in which alleles of this oncogene have been found so far are polymorphic and often the majority of fish are oncogene free. It has been proposed that the gene has been frequently lost during evolution (Weis and Schartl 1998). We describe two cases, namely a deletion and an insertional disruption by a transposable element of ONC-Xmrk. In both cases the oncogenic function of the gene is impaired. This provides functional evidence that ONC-Xmrk is responsible for melanoma formation.
MATERIALS AND METHODS
Fish: All fish were bred and maintained in the aquarium facilities of the Biocenter under standard conditions (Kallman 1975).
The following strains were used.
X. maculatus, Rio Jamapa, Sd. This stock was derived from the Jp163A strain (ca. the 1960s) and is kept as a closed colony, randomly inbred. Females of this stock are homozygous for the X-chromosomal loci MdlSd-ONC-Xmrk.
X. helleri, Rio Lancetilla, Db-2−. These fish have neither ONC-Xmrk nor Mdl (Weis and Schartl 1998). They are used for introgressive breeding to separate the MdlSd-ONC-Xmrk from R. This stock is also known as the hIII strain.
Backcross hybrids of X. maculatus (see 1) with X. helleri (see 2) using X. helleri as the recurrent parent (third backcross generation). The fish carried the wild-type X chromosome of X. maculatus with MdlSd-ONC-Xmrk and develop, depending on the presence or absence of R, benign or malignant melanoma. They are known as Sd-melanoma fish from the classical Gordon-Kosswig cross (e.g., see Anders 1991; Kazianiset al. 1998).
Lof-1, backcross hybrids carrying the Sd loss-of-function X chromosome, derived from the wild-type MdlSd-ONC-Xmrk chromosome, in the genetic background of X. helleri. Lof-1 fish are hemizygous for the X. maculatus sex chromosome. The sex-determining locus on the X. maculatus X chromosome overrides the sex-determining mechanism of X. helleri. Therefore, the majority of fish with the lof-1 chromosome like the corresponding backcross hybrids with the wild-type MdlSd-ONC-Xmrk X chromosome are females (lof-1, 99%; wt, 90%). The stock is maintained by crossing lof-1 females with purebred X. helleri males from the Rio Lancetilla stock (see 2). Lof-1 is a spontaneous mutant that was isolated by A. Anders, Giessen, in the early eighties. This strain is also known as the ArDr strain.
Lof-2 is another spontaneous loss-of-function mutant from the Sd locus that occurred in our laboratory in a backcross brood of Sd melanoma fish with X. helleri (see 2) as the recurrent parent. Lof-1 and lof-2 are maintained in the same genetic background.
Lof-3 is also a spontaneous loss-of-function mutant of the MdlSd-ONC-Xmrk X chromosome, which was isolated by A. and F. Anders. The phenotype in backcross hybrids is identical with that of lof-1 and lof-2. Nothing is known about the molecular basis of the mutation. The mutant chromosome was reintroduced in the genetic background of X. maculatus from Rio Jamapa. This platyfish strain has been designated the Dr strain (Anderset al. 1973) or Δ-Sd strain (Woolcocket al. 1994).
For analysis of the phylogenetic distribution of the newly identified transposable element, TX-1, DNA from the following fish kept as randomly inbred lines at the Biocenter Würzburg was used (strain designation or geographic origin in parentheses): X. maculatus (Rio Jamapa, see 1, Rio Usumacinta), X. milleri (Laguna Catemaco), X. helleri (Rio Lancetilla, see 2), X. couchianus (Apodaca), X. meyeri (Musquiz), X. andersi (Rio Atoyac), X. nezahualcoyotl (Rio El Salto), X. montezumae (Cascadas de Tamasopo), X. variatus (Ciudad Mante), X. cortezi (Rio Axtla), X. malinche (Rio Calnali), Gambusia affinis affinis (Pena Blanca, Santa Cruz River system, north of Nogales), Poeciliopsis gracilis (Rio Jamapa), Heterandria bimaculata (Rio Tonto system), H. formosa (Fort Lauderdale), Phallichthyes amates (aquarium stock), Poecilia mexicana (Media Luna), P. latipinna (Key Largo), P. formosa (Tampico), Girardinus metallicus (aquarium stock), and G. falcatus (aquarium stock). All these species belong to the family Poeciliidae. As representatives of other families Fundulus spp. (Laguna de Labradores), Oryzias latipes (Medaka strains HB32C and Carbio, see Honget al. 1998), and Danio rerio (zebrafish strain m14) were used. Genomic DNA from Nile Tilapia (Oreochromis niloticus) and from Chinese perch (Siniperca chuatsi) were a gift from S. Chen (University of Würzburg). Genomic DNA from salmon (Salmo salar) was obtained from C. Cunningham (Sars Center for Molecular Marine Biology, Bergen). DNA and tissues from Battrachocottus baikalensis (Baikal lake) were kindly provided by S. Kirilchik and M. Grachev (Institute of Limnology, Irkutsk, Russia). Rainbow trout (Oncorhynchus mykiss), pike (Esox lucius), common carp (Cyprinus carpio), european eel (Anguilla anguilla), and sturgeon (Acipenser sturio) were obtained from a local fish farm.
Southern blot, PCR, and RNA expression analysis: DNA from individual fish was isolated as described (Schartlet al. 1995b). RNA was extracted from pooled organs of several fish using the TRIZOL reagent (GIBCO-BRL, Grand Island, NY) according to the supplier's recommendations. Reverse transcription was done with 2 μg total RNA using Superscript II reverse transcriptase (GIBCO-BRL) and random primers (GIBCO-BRL). cDNA from 10 ng (actin) to 300 ng of total RNA was used for PCR. We used 50 pmol of the following gene-specific primers: for Xmrk Ins4 5′ GCC TCC TGG GAG GAC AGC GAC 3′ and Ins5 5′ AGC GAG CCC TGC ATC CCG CCG 3′ at an annealing temperature of 72°, Ana1 5′ CCC GTC CAG CAG ACC CAG GTT CA 3′ in combination with Ins4 at an annealing temperature of 72°, Lof1A 5′ CCG TTA GAT GGA GAG CTG GTA GC 3′ in combination with Ana1 at an annealing temperature of 72°, and for actin Act1 5′ GTA GGT GAT GAA GCC CAG AGC 3′ and Act2 5′ AGG GAG CTC GTA GCT CTT CTC 3′ at an annealing temperature of 66°. PCR bands were quantified using the software from the Easy plus Rev3.16 image analysis system (Herolab).
For Southern analysis 4.5 μg of genomic DNA was digested with restriction enzymes, separated on a 0.8% agarose gel and blotted onto nylon membrane (Hybond N+, Amersham Buchler, Braunschweig, Germany). For analysis of the ONC-Xmrk locus of the lof-2 mutant, filters were hybridized with several different probes: a 0.7-kb BamHI fragment (probe II) and a 5.3-kb EcoRI fragment of the first intron (probe III), the total Xmrk cDNA (probe I), a PCR product amplified with the primers Ins2/Hg80 (Ins2 5′ AGG GAA TGA ACT ACC TGG AAG AGC 3′, Hg80 5′ AGG GTG AAG GCA GGT AGG TG 3′) specific for the Xmrk kinase domain (probe IV), a PCR product amplified with the primers Sw2/Sw1 (Sw2 5′ GCT GAC GGG ATG AAC GC 3′, Sw1 5′ TGA GAA TCC AGT TTC AAC C 3′), specific for the Xmrk carboxyterminal domain (probe V), and the 3′-end probe of the cosmid 008-G01 (probe VI) containing X. maculatus sequences 3′ of the Xmrk coding region (J.-N. Volff, K. Sweeney, K. Wichert, S. Weis and M. Schartl, unpublished results). For species distribution of the transposable element TX-1, the 1.5- and 1.3-kb BamHI fragments from p338 were used.
Hybridization probes were radiolabeled by random priming according to Feinberg and Vogelstein (1983). High-stringency hybridizations were performed at 42° in buffer containing 50% formamide, 5× SSC. Filters were washed at 68° with 0.1× SSC, 1% SDS. Under conditions of low stringency, hybridization was done in 35% formamide and washing was at 50° with 2× SSC, 1% SDS.
A total of 100 ng of lof-2 genomic DNA was used for PCR experiments with the primers Prom3 (5′ AAT GAC TGG GCA GTG CTA AGG 3′) and Prom2 (5′ CCG CTC CTC CGC GCA GAA AC 3′). These primers flank the chromosomal breakpoint of ONC-Xmrk and generate a product only from the oncogenic copy of Xmrk, but not from INV-Xmrk. Amplification was carried out as described (Gutbrod and Schartl 1999) with an annealing temperature of 64° using 1 unit of Taq polymerase (GIBCO-BRL) in a reaction volume of 50 μl.
Cloning and sequencing: DNA from a single fish of the lof-1 mutant was digested to completion with EcoRI and separated on a 0.8% low-melting-point agarose gel. The 10–15 kb range was cut out of the gel, eluted, and cloned into the EcoRI site of dephosphorylated λEMBL4 arms. Recombinant phage DNA was packaged in vitro and the resulting library was screened with the kinase domain probe p17-2 (Wittbrodtet al. 1989) of Xmrk. The entire 15-kb EcoRI insert of one of the plaque-purified positive phages was cloned into pBluescript KS+ to generate plasmid p338. Generation of overlapping subclones and sequencing using dideoxy chain termination was essentially done according to standard protocols. The sequence of TX-1 is deposited in GenBank under accession no. AF130854. The RT-PCR product with primers, Ana1/Ins4 was cloned into pUC 18 using the SureClone ligation kit (Amersham Pharmacia Biotech, Freiburg, Germany) and sequenced. Sequence analysis and similarity searches were done using the program package of NIX at the HGMP (http://menu.hgmp.mrc.ac.uk/menu-bin/Nix/Nix.pl).
RESULTS
Lof-1: insertion of a transposable element: The Xmrk lof-1 mutant is a spontaneous mutant from a brood of backcross hybrids carrying the wild-type X. maculatus X chromosome of the Rio Jamapa strain with the ONC-Xmrk allele from the Tu-Sd locus (= MdlSd-ONC-Xmrk) in the genetic background of X. helleri (Rio Lancetilla stock). The mutant has lost the ability to develop spontaneously malignant melanomas (Figure 1). Gross structural analysis of ONC-Xmrk revealed that the lof-1 mutant has a rearranged fragment for the kinase domain of 12 kb instead of 5 kb in EcoRI digests due to an ~7-kb insertion in exon 24 (earlier denominated exon W) of the Xmrk oncogene (Wittbrodtet al. 1989). The rearranged EcoRI fragment was isolated from a partial genomic library. Sequence analysis of the borders of the insertion revealed the site of insertion to be between nucleotides (nt) 1652 and 1653 of the X-chromosomal Sd allele of ONC-Xmrk (numbering according to Adamet al. 1991). The inserted DNA thus interrupts the reading frame of the wild-type sequence at the 3′ border of exon 24.
Backcross hybrids carrying (a) the wild-type MdlSd-ONC-Xmrk chromosome, (b) the mutant lof-1 chromosome, and (c) the mutant lof-2 chromosome.
When the wild-type and lof-1 mutant ONC-Xmrk sequences were compared, an imperfect 6-bp duplication at the site of insertion was found (5′ border ACA GGT, 3′ border CAA GGT). As this is a known feature of a mutational event generated by the insertion of a transposable element, the structure of the entire interrupting sequence (denominated TX-1) was determined (Figure 2). The insert has a total length of 7157 nt. The 5′ and 3′ ends are composed of 2377- and 2313-nt-long terminal direct repeats (96% identity) each of which has almost perfect internal repeats of a 68-nt and a 91-nt sequence. The internal unique sequence lacks an apparent continuous long open reading frame. The longest open reading frame that would translate into a 181-aa protein was found at the 3′ border of the 5′ LTR. When exon-based gene prediction programs were used, no apparent spliced gene structure was predicted. Searching the publicly available databases revealed no significant similarity to known sequences of any organism.
Southern blot analysis of X. maculatus DNA using an internal fragment of the insertion as a probe in digests with enzymes that cut within the inserted sequence revealed only the expected fragments of the insertion (data not shown). However, the hybridization signal was very strong. In digests with enzymes that do not cut the insertion sequence but only cut the flanking genomic DNA, a multiplicity of fragments ranging in size from 1 kb to >20 kb was obtained. This indicates multiple copies of TX-1, certainly >50–100 per genome. Shorter fragments might represent truncated copies. A similar result was obtained for other X. maculatus strains as well as for X. milleri and X. helleri males and females (Figure 3A), X. couchianus, X meyeri, X. andersi, X. nezahualcoyotl, X. montezumae, X. variatus, X. cortezi, X. malinche, and the X. maculatus/X. helleri hybrid melanoma cell line PSM (data not shown). TX-1-cross-hybridizing sequences were also detected under high-stringency conditions in all members of the family Poeciliidae tested, including Poecilia formosa, Gambusia affinis, Girardinus falcatus, P. gracilis, Heterandria bimaculata, Phallichthyes amates (Figure 3B), P. mexicana, P. latipinna, G. metallicus, and H. formosa (data not shown). In contrast, even under low-stringency conditions, no signals were found in other non-poeciliid species. Even in the closely related fishes Fundulus spp. (Fundulidae) and medakafish (Oryzias latipes, Oryziatidae), which both belong to the same order Cyprinidontiformes, no TX-1-related sequences appeared to be present. The same is true for the more distantly related representatives of the main branches of the phylogenetic tree (Nelson 1994) of the bony fishes (Osteichthyes), Nile Tilapia (Oreochromis nilotica, Cichlidae), rainbow trout and salmon (Oncorhynchus mykiss and Salmo salar, Salmonidae), pike (Esox lucius, Esocidae), zebrafish (Danio rerio, Cyprinidae), common carp (Cyprinus carpio, Cyprinidae), european eel (Anguilla anguilla, Anguillidae), sturgeon (Acipenser sturio, Acipenseridae), human (Figure 3C), Chinese perch (Siniperca chuatsi, Serranidae), and Battrachocottus baikalensis (Cottidae; not shown). This strongly suggests that TX-1 is restricted to the family of Poeciliidae.
Restriction map and schematic structure of the lof-1 insertion. Numbering of nucleotides is according to the inserted transposable element TX-1; numbers in parentheses indicate the position with respect to the ONC-Xmrk sequence. Large arrows indicate the LTRs; small arrows show the placement of small internal direct repeats.
Lof-2: deletion of ONC-Xmrk: The lof-2 mutant was detected in a brood of backcross hybrids of the same type as described above for lof-1. No melanoma development was observed among the mutant and its progeny after further backcrossing with X. helleri (Figure 1c). The phenotype of lof-2 fish is exactly the same as that of lof-1 fish except that all lof-1 animals are females (like fish with the wild-type chromosome), while lof-2 fish are males. Crossing of lof-1 to lof-2 fish resulted in offspring that all had the mutant phenotype (not shown). This demonstrates that the two mutations do not complement each other. Southern blot analysis with different hybridization probes spanning the whole genomic region of the Xmrk oncogene from the 5′ flanking sequences downward up to sequences 3′ of the last exon revealed that the whole region is deleted (Figure 4). PCR amplification with the primers prom3/prom2 extending from nt −584 to nt +25 across the promoter region and transcriptional start site failed to produce the expected 609-bp fragment that is obtained from the wild-type X-chromosomal allele of ONC-Xmrk. In summary these data indicate deletion of the entire oncogene locus.
Expression analysis: To analyze the expression of the Xmrk proto-oncogene (INV-Xmrk) the wild-type and mutant ONC-Xmrk alleles (wt ONC-Xmrk, lof-1 Xmrk, lof-2 Xmrk), an allele-specific RT-PCR was set up (Figure 5a). The wt ONC-Xmrk allele from the X chromosome of X. maculatus differs from INV-Xmrk by a 1344-bp deletion that includes exon 26 [numbering of exons according to Gutbrod and Schartl (1999); this exon has been denoted Y in earlier articles (Adamet al. 1991)].
Primers that flank the deletion (Ins4/5) give an RT-PCR product of 226 bp from the INV-Xmrk transcript, while the ONC-Xmrk transcript is amplified as a 128-bp product. In a competitive PCR, a semiquantitative determination of expression levels can be achieved. INV-Xmrk was found to be expressed at low levels in all healthy organs including skin (number of independent RNA preparations tested: n = 6), muscle (n = 3), liver (n = 9), kidney (head nephros; n = 3), testes (n = 2), brain (n = 3), eye (n = 14), and gills (n = 18) from parental platyfish (with and without ONC-Xmrk) and melanoma-bearing as well as healthy backcross hybrids. A similar low expression was seen in benign (n = 4) and malignant melanoma (n = 4). The wild-type ONC-Xmrk was highly expressed in malignant melanoma (~10-fold higher than INV-Xmrk), but was expressed only at a low level in benign melanoma (at similar levels as INV-Xmrk). The wild-type ONC-Xmrk was found to be expressed also in all healthy organs analyzed (except liver) of ONC-Xmrk-bearing platyfish (n = 5) as well as malignant and benign melanoma-bearing backcross hybrid fish (n = 9; Figure 5b), but at much lower levels than INV-Xmrk.
Southern blot analysis of TX-1 distribution in (A) the genus Xiphophorus, (B) the family Poeciliidae, and (C) other bony fishes. See materials and methods and results for description of the fishes analyzed. Human DNA was used as a control for unspecific hybridization. Genomic DNAs were cut with HindIII. Probes are the direct repeats of TX-1. Filters A and B were washed at high stringency (0.1× SSC, 68°), filter C at low stringency (2× SSC, 50°). Extreme overexposure of filter C did not reveal any specific signal in fishes outside of the family Poeciliidae.
It has been proposed that ONC-Xmrk expression plays a role in macromelanophore pigment cell pattern formation (Woolcocket al. 1994). To investigate this, we prepared RNA of X. maculatus (MdlSd-ONC-Xmrk) from dorsal fins with and without macromelanophore spots and caudal fins without macromelanophore spots; we also prepared RNA of X. maculatus carrying the lof-3 mutant X chromosome from dorsal fins without macromelanophore spots. In all four samples a similar expression level of ONC-Xmrk was detected (Figure 5c), indicating that expression of the oncogene is not functionally related to formation of macromelanophores.
Expression of the ONC-Xmrk gene in healthy organs allowed for expression analysis of the mutant alleles from lof-1 and lof-2 (Figure 5d). Consistent with the genomic structure, no expression of ONC-Xmrk in lof-2 was detected with primers Ins4/5. From the lof-1 mutant allele two transcripts were detected. Amplification with primers Ana1/lof-1A indicated presence of a transcript, which reads through from exon 24 into the 5′ LTR of the insertion sequence. From this mRNA premature termination of the Xmrk protein will occur due to a stop codon located at 107 bp in the inserted sequence (Figure 6a). A second mRNA that was amplified with primers Ana1/Ins4 is produced. Cloning and sequencing of the PCR product revealed that a cryptic splice donor site (CAG/GTACCT) in exon 24 that is located 16 bp 5′ of the wild-type splice site (CAG/GTACGC) is used. In this transcript the entire inserted DNA is spliced out together with intron 24. However, a deletion of 16 bases is created due to the use of the new splice donor, which leads to a frameshift and termination after 15 codons (Figure 6b). Both transcripts are present at similar levels and both translate into a truncated Xmrk receptor protein that terminates shortly after the end of the kinase domain and lacks the carboxyterminal domain with the autophosphorylation and substrate binding sites.
DISCUSSION
We have analyzed the molecular basis of two melanoma loss-of-function mutants in Xiphophorus and found that both are due to impairment of the oncogenic copy of Xmrk. In the lof-2 mutant the whole ONC-Xmrk gene is missing. ONC-Xmrk is flanked by the RY locus (red and yellow pigment patterns) and Mdl, the macromelanophore locus (Gutbrod and Schartl 1999). The deletion of lof-2 obviously does not extend into the RY locus, because the phenotype of the pigment pattern Dr is unchanged compared to wild-type fish. Whether the deletion extends into Mdl is unclear. No macromelanophores are produced. Thus Mdl could be partly or totally missing, but an explanation similar to that for the lof-1 mutant (see below) may be given as well.
In the lof-1 mutant an insertion in ONC-Xmrk results in a disruption of the gene leading to aberrant transcripts from which only a truncated protein can be produced. The mutant protein would lack the whole carboxyterminal domain, which contains the binding sites for several important substrate proteins (Wellbrock and Schartl 1999; Wellbrocket al. 1999). Thus the truncated receptor should be unable to exert its signaling function, which is required to initiate and maintain the proliferation of the transformed pigment cells. This would explain the phenotype of the lof-1 mutant, which, contrary to the corresponding wild type, does not develop melanoma anymore. The fact that disruption of the Xmrk oncogene results in a loss of the ability to form melanoma demonstrates that ONC-Xmrk is indeed the melanoma-inducing gene from the Tu locus. As ONC-Xmrk was isolated by a positional cloning strategy, this proof of its identity appears especially important.
Analysis of the genomic structure of the ONC-Xmrk region of the lof-2 mutant. (a) Genomic organization of the X allele of ONC-Xmrk of wild-type X. maculatus. Black boxes represent exons; numbers give the size of the EcoRI fragments. Fragments used as hybridization probes on filters with genomic DNA of the lof-2 mutant and wt fish and sequences amplified by PCR are indicated by roman numbers. The PCR products VII, amplified with primers Prom3 and Prom2; VIII, amplified with the primers Xfor/Xrev (Gutbrod and Schartl 1999); and IX, amplified with Ins5/Ins4, are missing in lof-2. (b) Southern blot analysis with probes II, III, IV, and VI. X. mac (II–IV), male X. maculatus with X- and Y-ONC-Xmrk; lof-2, lof-2 mutant fish; BC3, fish of the same pedigree as lof-2, but carrying the wild-type X chromosome with ONC-Xmrk; X. mac (VI), female X. maculatus with X-ONC-Xmrk; X. mac*, female X. maculatus without Xmrk oncogene. (II) A 0.7-kb BamHI fragment was used as hybridization probe. The 10-kb band of X. maculatus represents INV-Xmrk, whereas the 10-kb band in lof-2 and BC3; is the INV-Xmrk allele inherited from the X. helleri parent. In these two individuals the X. maculatus-derived INV allele on the X chromosome is 7.4 kb in size (J. N. Volff, unpublished results). (III) The 5.3-kb EcoRI probe specific for the X allele of ONC-Xmrk detects its genomic counterpart, a 4.4-kb fragment derived from the X. helleri INV-Xmrk gene, a 4.1-kb X. maculatus INV allele, and a 3.4-kb Y-ONC-specific fragment. (IV) The probe covering a part of the Xmrk kinase domain-encoding sequence detects a 7-kb band representing both X. maculatus and X. helleri INV-Xmrk alleles, a 6.5-kb band that is Y-ONC-Xmrk specific, and a 5-kb fragment that is the X-ONC allele. (VI) A probe from the 3′ end of Xmrk detects a 7.8-kb band common to X. maculatus and X. helleri INV-Xmrk and a 5-kb X-ONC-specific fragment.
However, another aspect of the lof-1 phenotype is not so easily explained by the molecular data. It was shown that Mdl, which specifies the phenotype of the macromelanophore (larger size, lack of distance regulation in pattern formation, etc.), is a separate genetic entity and not identical to the ONC-Xmrk (Weis and Schartl 1998). Thus it would have been expected that lof-1 fish would still develop some macromelanophores like the parental platyfish. These, however, should not proliferate and grow out to melanoma because of the lack of ONC-Xmrk function. Surprisingly, not a single macromelanophore has been detected in hundreds of lof-1 fish. A possible explanation would be that the large insertion in ONC-Xmrk has a remote effect on neighboring genes, e.g., the very closely linked Mdl, through changes in chromatin structure. Also, genes associated with transposable elements could be mistakenly identified as genome invaders and become methylated and silenced as well (Henikoff and Matzke 1997). Thus insertion of the TX-1 sequence may have a negative regulatory effect on the expression of Mdl, which has been mapped close to ONC-Xmrk (Gutbrod and Schartl 1999).
Also in the lof-2 mutant, another locus, which is obviously not affected in its primary structure by the mutation, is affected in its phenotype. Lof-2 fish show a normal expression of the RY locus located 3′ of the deleted ONC-Xmrk. But in these fish the sex-determining locus that resides behind the RY locus further away from ONC-Xmrk has obviously changed its function from female to male.
Expression analysis of the wild-type ONC-Xmrk gene and the lof-1 mutant. (a) Location of primers and sizes of the corresponding RT-PCR products from the Xmrk proto-oncogene (INV-Xmrk), the wild-type X-chromosomal oncogenic copy (wt ONC-Xmrk) with the deletion including exon 26, and the derived mutant ONC-Xmrk allele from lof-1 fish (Lof1 ONC-Xmrk) with insertion of TX-1. (b) RT-PCR analysis of normal organs from tumor-bearing backcross hybrids and of benign pigment lesions and malignant melanoma. Primers Ins5 and Ins4 were used, which amplify a 227-bp fragment from the INV-Xmrk and a 122-bp fragment from the X-chromosomal ONC-Xmrk transcript. For quantification a PCR for actin with primers Act1 and Act2 (product size 565 bp) was performed from each RT reaction. (c) RT-PCR analysis with primers Ins5 and Ins4 for ONC-Xmrk expression in macromelanophore spot-containing dorsal fins (1); macromelanophore spot-free dorsal fin areas (2); tail fins (3), which contain no macromelanophores, of wild-type X. maculatus from Rio Jamapa with the pattern “spotted dorsal”; and dorsal fins without macromelanophores (4) from the lof-3 mutant. (d) RT-PCR for analysis of expression of the ONC-Xmrk locus in normal organs from lof-1 fish. With primers Ana1 and Lof1A the transcript that extends into the inserted TX-1 sequence is specifically amplified (fragment size 188 bp) from the mutant ONC-Xmrk locus in lof-1 fish but not from wild-type (wt) fish (melanoma-bearing backcross hybrids with benign or malignant lesions). With primers Ana1/Ins4 a transcript of 308 bp is amplified in which the TX-1 insertion sequence is spliced out; this transcript has a sequence similar to the wild-type transcript (324 bp). The 429-bp product is amplified by this pair of primers from the INV-Xmrk transcripts.
Alternatively, ONC-Xmrk, although not part of Mdl, could have some function in X. maculatus for macromelanophore pattern formation. In this context the expression of the oncogenic copy of Xmrk in all normal tissues except liver is somehow enigmatic, because, contrary not show any sign of hyperplasia or other proliferative disorders in any of their organs nor do hybrids develop tumors of any other cell type except pigment cells. An explanation may be the low level of expression in normal organs, which is below that of the proto-oncogene and is considerably lower than in the benign pigment lesions. The reasoning that a certain threshold level of ONC-Xmrk transcripts is necessary to elicit neoplastic transformation is in accordance with studies on the transforming capacity of ONC-Xmrk in transgenic fish and in in vitro cell systems where high expression of the oncogenic receptor was required (Wittbrodtet al. 1992; Winkleret al. 1994). Moreover, the malignancy of the tumors, which is determined by the amount of undifferentiated, rapidly dividing cells, correlates positively with the amount of ONC-Xmrk mRNA (Adamet al. 1991; Maüeleret al. 1993).
Sequence and conceptual translation of the mutant ONC-Xmrk transcripts of lof-1. (a) “Read through” transcript: cDNA sequence and conceptual translation of the transcript from the wild-type ONC-Xmrk gene (top) and from the lof-1 mutant locus (bottom). In the mutant, transcription continues into the inserted sequence and translation stops at a TAA codon in TX-1. (b) “Splice-out” transcript: genomic structure, cDNA sequence, and conceptual translation of the transcript from the wild-type ONC-Xmrk (top) and from the lof-1 mutant locus (bottom). Because TX-1 is inserted into the wild-type splice donor site of the exon 24/intron 24 border, a cryptic splice donor site 16 bases upstream that has a similar sequence is used (CAG/GTACC vs. CAG/GTACG). The transcript continues correctly with exon 25, but a frameshift created through the position of the cryptic spice donor site leads to a different amino acid sequence and terminates after 15 residues.
The low-level expression of ONC-Xmrk in normal organs could point to a normal function of this gene. However, all Xiphophorus populations are polymorphic for ONC-Xmrk, and many individuals do not have this gene. Using an experimental strategy similar to the one in this study Woolcock et al. (1994) found expression of ONC-Xmrk in dorsal fins with macromelanophore spots, but not in fins of the loss-of-function mutant lof-3, which do not develop macromelanophore spots. This result was interpreted as reflection of a function of ONC-Xmrk for formation of the “spotted dorsal” pigment pattern. However, this appears unlikely, because macromelanophore patterns are found in fish of several Xiphophorus species in the absence of ONC-Xmrk, indicating that it is not necessary for development of this cell type (Weis and Schartl 1998). In addition, we have shown that expression of ONC-Xmrk in the fin tissue is not connected to formation of macromelanophores, as its transcript was detected at similar levels also in fins without spots.
The insertion sequence in lof-1 has several features, which characterize it as a mobile genetic element. We thus propose to name it TX-1 (for transposable element of Xiphophorus). First, TX-1 has been observed to move into the Xmrk oncogene by identification of the first mutant fish. This is a so far unique situation because fish hemizygous for ONC-Xmrk can be easily bred and because of the dominant phenotype of the Xmrk oncogene. In the other examples in fish where mutants are due to insertion of sequences with structural similarities to transposable elements, the “no-tail” mutant of the zebrafish (Schulte-Merkeret al. 1994) and the albino mutants i1 and i4 of the medaka (Koga and Hori 1997), the mutations are recessive. As spontaneous mutants they might have been around in the natural fish population or laboratory strain for quite some time before becoming apparent. Thus the creation of the mutant allele and the invasion of the element could not be observed. It should be noted, however, that frequent active excision of the transposable element of the i4 mutant has been observed (Kogaet al. 1996). Second, TX-1 created a 6-bp target sequence duplication at its margins. This is typically seen for all mobile elements but evidently does not occur during insertion events by other types of sequences. Third, TX-1 has long direct terminal repeats. This would assign TX-1 to the class of LTR-containing retrotransposons of mobile elements. Fourth, TX-1 is present in multiple copies in the genome of all poeciliid species tested, but absent from other fish species.
TX-1 has specific features, which distinguish it from most transposable elements analyzed so far. Its terminal repeat is unusually long and may be one of the largest known. The TX-1 sequence shows no similarity to other retrotransposons, has no long open reading frames that would give rise to proteins known from other mobile DNA elements, and in particular gives no indication of encoding a functional polymerase that would make it capable of transposition. As TX-1 only recently moved into the Xmrk gene, either it has lost such sequences during the process of integration or it is dependent on a helper element that contributes the reverse transcriptase function. Elements of the SINE family also do not encode proteins for reverse transcription. TX-1, however, has no structural features that are characteristics of SINEs. The overall structure of TX-1 resembles that of the murine Etn elements and human THE-1. THE-1, however, is 2.3-kb long and has only a 350-bp LTR (Paulsonet al. 1985). Etn elements are almost 6-kb long, have no long open reading frames, and lack sequence similarity to any known retroelements, but have small LTRs (Sonigoet al. 1987).
Acknowledgments
We thank A. Telling, S. Hölter, and R. Kolb for technical assistance. Founder fish for some of our stocks were kindly supplied by A. and F. Anders, Giessen. We are especially grateful to them for making the lof-1 mutant available to us. The zebrafish strain m14 was obtained from W. Driever, Freiburg. We are grateful to S. Chen (University of Würzburg), C. Cunningham (Sars Center for Marine Molecular Biology, Bergen, Norway), and S. Kirilchik and M. Grachev (Institute of Limnology, Irkutsk, Russia) for the gift of fish organs and DNA and to Y. Hong for help in dissecting fish for DNA isolation. This work was supported by grants to M.S. supplied by the Deutsche Forschungsgemeinschaft through Sonderforschungsbereich 465, the Commission of the European Community (FAIR PL97-3796) and Fonds der Chemischen Industrie.
Footnotes
-
Communicating editor: C. Kozak
- Received March 25, 1999.
- Accepted July 16, 1999.
- Copyright © 1999 by the Genetics Society of America
