Mechanism of activation of the human trk oncogene.

The human trk oncogene was generated by a genetic rearrangement that replaced the extracellular domain of the normal trk tyrosine kinase receptor by sequences coding for the 221 amino-terminal residues of a nonmuscle tropomyosin. Molecular dissection of a cDNA clone of the trk oncogene indicated that both the tropomyosin and tyrosine kinase domains were required for proper transforming activity. Replacement of nonmuscle tropomyosin sequences with those of other tropomyosin isoforms had no deleterious effect. However, when tropomyosin sequences were replaced with those of another cytoskeletal gene, such as beta-actin or beta-globin, transforming activity was completely abolished. These results illustrate the important role of tropomyosin sequences in endowing the trk kinase with transforming properties. Functionally unrelated subdomains of the tropomyosin molecule were equally efficient in activating the trk gene. Moreover, the transforming activity of the trk oncogene was not affected when its subcellular localization was drastically altered. Therefore, tropomyosin sequences are likely to contribute to the malignant activation of the trk oncogene not by facilitating its interaction with defined cytoskeletal structures as initially suspected, but by allowing its kinase domain to fold into a constitutively active configuration.

Human trk oncogenes activated by point mutation, in-frame deletion, and duplication of the tyrosine kinase domain.

Malignant activation of the human trk proto-oncogene, a member of the tyrosine protein kinase receptor family, has been implicated in the development of certain human cancers, including colon and thyroid papillary carcinomas. trk oncogenes have also been identified in cultured cells transfected with various DNAs. In this study, we report the characterization of three in vitro-generated trk oncogenes, trk2, trk4, and trk5 (R. Oskam, F. Coulier, M. Ernst, D. Martin-Zanca, and M. Barbacid, Proc. Natl. Acad. Sci. USA 85:2964-2968, 1988), in an effort to understand the spectrum of mutational events that can activate the human trk gene. Nucleotide sequence analysis of cDNA clones of trk2 and trk4 revealed that these oncogenes were generated by a head-to-tail arrangement of two trk tyrosine protein kinase domains connected by a purine-rich region. These oncogenes code for cytoplasmic molecules of 67,000 (p67trk2) and 69,000 (p69trk4) daltons. In contrast, the product of the trk5 oncogene, gp95trk5, is a cell surface glycoprotein of 95,000 daltons. This oncogene was generated by a 153-base-pair in-frame deletion within sequences coding for the extracellular domain of the trk receptor. This activating deletion encompasses a triplet coding for one of the nine cysteine residues that the trk receptor shares with the product of the highly related trkB tyrosine protein kinase gene. Introduction of a single point mutation (TGT----AGT) in this codon resulted in a novel trk oncogene whose product, gp140S345, differs from the nontransforming trk proto-oncogene receptor in a single amino acid residue, Ser-345 instead of Cys-345. These results illustrate that multiple molecular mechanisms, including point mutation, internal deletion, and kinase domain duplication, can result in the malignant activation of the human trk proto-oncogene.

Genome-wide search for loss of heterozygosity shows extensive genetic diversity of human breast carcinomas.

Deletions of genomic regions involving tumor suppressor genes are thought to be important in the initiation and progression of breast cancer. We conducted a genome-wide search for deleted regions in a series of 75 human breast carcinomas by studying the allelic patterns of 184 microsatellite markers distributed over all chromosomes and looking for loss of heterozygosity (LOH). We identified 56 regions of consistent LOH. Strikingly, every tumor had a different set of deletions. To study this complexity, we applied a phylogenetic-like type of analysis. Each region was involved in a certain proportion of tumors, ranging from 20 to 62%; the most frequently involved regions were on chromosome arms 8p, 11q, 16q, and 17p. There was a correlation (P = 0.005) between the level of LOH and the size of the tumors. Tumors with a high level of LOH were also highly proliferative and had a high mitotic index.

Putative structure of the FGF6 gene product and role of the signal peptide.

The human FGF6 gene is an oncogene related by sequence similarities to the fibroblast growth factor (FGF) gene family, which encodes mitogenic peptides implicated in various physiological processes including angiogenesis, morphogenesis, tissue regeneration and survival and oncogenesis. Nucleotide sequence analysis of the FGF6 gene and of cDNA clones revealed an open reading frame able to code for a protein of 208 residues. The FGF6 protein shares 32-70% residues with the other members of the family within the C-terminal two-thirds of the molecule. In vitro, three in-frame ATG codons are able to initiate the translation of three peptides of 175, 198 and 208 residues. These three peptides differ at their amino termini with respect to the relative position of a hydrophobic leader peptide, which extends from residues 16 to 40, and is therefore absent from the shorter (175 amino acids) form. In-vitro analysis indicates that this signal peptide is able to drive the FGF6 protein through the endoplasmic reticulum, where it becomes glycosylated. The presence of this signal peptide sequence appears essential for the in vivo transforming capacity of the FGF6 gene.

Cytotoxic activity of a diptheria toxin/FGF6 mitotoxin on human tumour cell lines.

The FGFs constitute a family of, at least, 12 polypeptides (FGF1 to FGF12) implicated in a number of physiological and pathological processes throughout embryogenesis and adult life. They bind to at least three types of cell surface molecules, including four high affinity transmembrane tyrosine kinase receptors (FGFR1 to FGFR4). In addition to important roles during development, FGF involvement in pathological conditions, including tumour formation, has been suspected, and overexpression of FGFR in tumour specimens is well documented. Diphtheria Toxin/FGF6 (DT/FGF6) mitotoxin has been shown to selectively and effectively target FGFR1-expressing cells. We show here that DT/FGF6 targets myoblasts engineered to express either one of the four FGFR, as well as FGFR-expressing tumour cells.

Characterization of the HST-related FGF.6 gene, a new member of the fibroblast growth factor gene family.

By screening a mouse cosmid library with a human HST probe under reduced conditions of stringency, we isolated several positive clones. One of them was identified as a new member of the fibroblast growth factor gene family, and called FGF.6. The human FGF.6 gene was subsequently isolated and sequenced. The deduced amino-acid sequence exhibited 70% identity with the HST gene product over the C-terminal two-thirds of the putative protein. FGF.6 was mapped to chromosome 12 at band p13 by in situ hybridization. The cloned normal human gene was able to transform mouse NIH3T3 fibroblasts using both focus- and tumorigenicity-assays.

let-756, a C. elegans fgf essential for worm development.

In vertebrates, Fibroblast Growth Factors (FGFs) and their receptors are involved in various developmental and pathological processes, including neoplasia. The number of FGFs and their large range of activities have made the understanding of their precise functions difficult. Investigating their biology in other species might be enlightening. A sequence encoding a putative protein presenting 30-40% identity with the conserved core of vertebrate FGFs has been identified by the C. elegans sequencing consortium. We show here that this gene is transcribed and encodes a putative protein of 425 amino acids (aa). The gene is expressed at all stages of development beyond late embryogenesis, peaking at the larval stages. Loss-of-function mutants of the let-756 gene are rescued by the wild type fgf gene in germline transformation experiments. Two partial loss-of-function alleles, s2613 and s2809, have a mutation that replaces aa 317 by a stop. The truncated protein retains the FGF core but lacks a C-termins portion. These worms are small and develop slowly into clear and scrawny, yet viable and fertile adults. A third allele, s2887, is inactivated by an inversion that disrupts the first exon. It causes a developmental arrest early in the larval stages. Thus, in contrast to the other nematode fgf gene egl-17, let-756/fgf is essential for worm development.

Murine FGF-12 and FGF-13: expression in embryonic nervous system, connective tissue and heart.

The molecular cloning of cDNAs encoding murine fibroblast growth factor-13 (FGF-13/FHF-2) and three isoforms of murine FGF-12 (FHF-1) is described. Like their highly conserved human counterparts, murine FGF-12 and FGF-13 are part of a distinct subfamily of FGF-like proteins characterized by a greater degree of amino acid sequence cross-homology and by conserved N-terminal domains which do not include secretion signal sequences. In addition to their expression in several adult tissues, both of these FGF genes are prominently and regionally expressed in midgestation mouse embryos, as revealed by in situ hybridization. Fgf12 and fgf13. RNAs were detected in developing central nervous system in cells outside the proliferating ependymal layer, and fgf13 RNA was also found throughout the peripheral nervous system. Fgf12 is expressed in developing soft connective tissue of the limb skeleton and in presumptive connective tissue linking vertebrae and ribs. Both FGF genes are also expressed in the myocardium of the heart, with fgf12 RNA found only in the atrial chamber and fgf13 RNA detected in both atrium and ventricle. On the basis of their novel structure and patterns of expression, FGF-12 and FGF-13 are anticipated to perform embryonic functions distinct from other known FGF molecules.

Homeobox gene clusters and the human paralogy map.

Homeobox genes encode important developmental control proteins. In vertebrates, those encoding the proteins of the HOX class and their most closely related families, including paraHOX and metaHOX classes, are clustered in paralogous regions (or paralogons). We show that the majority of the other homeobox genes (we called contraHOX) can also be clustered and belong to paralogons in humans. This suggests that they duplicated during vertebrate evolution along the same processes as the HOX genes. We tentatively assembled several paralogons in superparalogons. One of the superparalogons contains the contraHOX genes. These observations were extended to hundreds of genes, and allowed to describe a primary human genome paralogy map.

Ancestrally-duplicated paraHOX gene clusters in humans.

A paraHox gene cluster has been described recently in Amphioxus. We show here using bioinformatics and cytogenetics that, as the probable result of the duplication of an ancestral paraHox gene cluster, human paraHOX genes are located in four paralogous regions of the genome, on chromosomes 4, 5, 13 and X. By analogy with the four HOX gene clusters, we propose to designate the four paraHOX gene clusters as paraHOX-A to D clusters. We also propose a scenario for the evolution of HOX and paraHOX genes. Several chromosomal translocation breakpoints of hemopathies are located in the paralogous regions that contain the paraHOX genes. Two of the paraHOX genes are involved in these rearrangements.

Expression of variant forms of fibroblast growth factor receptor 2 mRNA in human breast carcinomas.

FGF (fibroblast growth factors) and FGF receptors may play a role in stroma-epithelium relationships in the mammary gland. Dysregulation of their interactions may be important in mammary carcinogenesis. Isoforms of the FGFR2 receptor, differing in the structure of the extracellular region, are expressed in the mammary gland and may diversely affect stroma-epithelium relationships. We determined the mRNA variants encoding these isoforms in human cell lines and breast carcinomas. All possible combinations of variants were found. No correlation was observed between the presence of a particular variant and the expression of any FGF gene tested, or the status of any histoclinical parameter.

Oncogenesis and genome duplication maps (commentary).

Examples of activation of putative cancer genes belonging to multigene families are given. Their structure and chromosomal location helps in identifying genome duplications. Reciprocally, genome duplication maps may help in predicting the location and behaviour of putative cancer genes.

Sequence analysis of full length cDNA for enhancing factor/phospholipase A2.

Enhancing factor (EF) protein was initially purified as a modulator of epidermal growth factor from small intestines of mouse. The cDNA sequence, obtained by RT-PCR, revealed that EF belonged to the non-pancreatic, phospholipase A2 (PLA2) family. This was the first report of the mouse PLA2. In the present paper we report the complete cDNA sequence of EF gene, in which the 5' sequence has been obtained by RAcE-PCR. The predicted amino acid sequence was computer analysed and the putative sites for enzyme action, calcium binding and heparin binding have been identified. The complete protein sequence of EF along with 16 aligned sequences were used to infer a phylogenetic tree. From this data the mouse EF was grouped with other membrane associated PLA2 with a bootstrap value of 98% indicating that it belonged to this class.

Overexpression of human TRK proto-oncogene into mouse cells using an inducible vector system.

The TRK proto-oncogene encodes a tyrosine kinase receptor for an, as yet, unidentified ligand. In order to help the identification of this ligand, we have constructed an expression vector capable of overexpressing the TRK protein in an inducible fashion. We report here the characterization of the TRK proto-oncogene products obtained from this expression vector.

MetaHox gene clusters.

Homeobox genes encode important developmental control proteins. The Drosophila fruit fly HOM complex genes are clustered in region 84-89 of chromosome 3. Probably due to large-scale genome duplication events, their human HOX orthologs belong to four paralogous regions. A series of 13 other homeobox genes are also clustered in region 88-94, on the same chromosome of Drosophila. We suggest that they also duplicated during vertebrate evolution and belong to paralogous regions in humans. These regions are on chromosome arms 4p, 5q, 10q, and 2p or 8p. We coined the term "paralogon" to designate paralogous regions in general. We propose to call these genes "meta Hox" genes. Like Hox genes, metaHox genes are present in one cluster in Drosophila and four clusters (metaHox A-D) in humans on the 4p/5q/10q paralogon.

Coparalogy: physical and functional clusterings in the human genome.

Two rounds of large-scale duplications are thought to have occurred in early vertebrate ancestry; this is now known as the "2R hypothesis." They have led to the constitution of subfamilies of paralogous genes. Chromosomal regions that contain present-day paralogs (paralogous regions or paralogons) have been identified in mammals. We show that sets of paralogons (PGs) can be assembled in a tentative "human genome paralogy map" that includes all autosomes and X. A total of 14 PGs, containing more than 1600 genes, were assembled in this paralogy map. Genes that belong to the same PG are coparalogs. We show that identification of coparalogy can be used (i) to broaden data on gene mapping, (ii) to identify physical gene clusters that derive from early cis-duplications, and (iii) to speculate on coevolution and coregulation of genes sharing a common structure or function (functional clusters). Thus, coparalogy analyses should parallel phylogenetic analyses and can help draw hypotheses on gene and genome evolution.

trkB, a novel tyrosine protein kinase receptor expressed during mouse neural development.

We have isolated a novel member of the tyrosine protein kinase family of cell surface receptors. This gene, designated trkB, is highly related to the human trk proto-oncogene. At the amino acid level, their respective products share a 57% homology in their extracellular regions including 9 of the 11 cysteines present in the trk proto-oncogene. This homology increases to 88% within their respective tyrosine kinase catalytic domains. Both trk and trkB are equally distantly related to the other members of this gene family of receptors. A biologically active cDNA clone of trkB can direct the synthesis of gp145trkB, a glycoprotein of 145 kd of which only 93 kd correspond to its polypeptide backbone. In adult mice, trkB is preferentially expressed in brain tissue, although significant levels of trkB RNA have also been observed in lung, muscle and ovaries. In addition, trkB transcripts can be detected in mid and late gestation embryos. The trkB locus exhibits a complex pattern of transcription. At least seven RNA species ranging in size from approximately 9 kb to 2 kb have been identified in brain. However, only a subset of these transcripts appears to be expressed in the other tissues. In situ hybridization analysis of 14 and 18 day old mouse embryos indicates that trkB transcripts are localized in the central (CNS) and peripheral (PNS) nervous systems, including brain, spinal cord, spinal and cranial ganglia, paravertebral trunk of the sympathetic nervous system and various innervation pathways. These results suggest that trkB may code for a novel cell surface receptor involved in neurogenesis.

"Paleogenomics": looking in the past to the future.

The complete sequence of the human and other vertebrate and nonvertebrate genomes provide a wealth of information on the organization, relationships and evolution of the metazoans. Soon the fine structure of our innermost biological identity will be unveiled and what has so far remained deep and secret will shine like an unearthed treasure and shape and fuel our future quests. A key treasure, for many molecular scientists interested in molecular evolution and development would be the knowledge of the genome of the ancestral precursor of all metazoans. In the absence of fossil DNA, this knowledge will forever remain a yearning for dreamy molecular biologists. And yet, will not the power of deduction and reconstitution of information gained through man's sophisticated technologies one day recreate a "virtual" metazoan ancestor?