Identification of tumor suppressor candidate genes by physical and sequence mapping of the TSLC1 region of human chromosome 11q23.

Loss of heterozygosity for a locus on human chromosome 11q22-23 is observed at high frequency in non-small cell lung carcinoma (NSCLC). Introduction of a 1.1 Mb fragmented yeast artificial chromosome (YAC) mapping to this region completely suppresses the tumorigenic properties of a human NSCLC cell line, A549. Smaller fragmented YACs give partial but not complete suppression. To further localize the gene(s) responsible for this partial suppression, a bacterial artificial chromosome (BAC) and P1-based artificial chromosome (PAC) contig was constructed, completely spanning the candidate region. End sequence generated in the construction of the BAC/PAC contig identified a previously unmapped EST and served to order genomic sequence contigs from the publicly available Celera Genomics (CG) and Human Genome Project (HGP) efforts. Comparison showed that CG provided larger contigs, while HGP provided more coverage. Neither CG nor HGP provided complete sequence coverage, alone or in combination. The sequence was used to map 110 ESTs and to predict new genes, including two GenScan gene predictions that overlapped ESTs and were shown to be differentially expressed in tumorigenic and suppressed A549 cell lines.

Isolation of the TSLL1 and TSLL2 genes, members of the tumor suppressor TSLC1 gene family encoding transmembrane proteins.

We have recently identified the TSLC1 gene as a novel tumor suppressor in human non-small cell lung cancers. TSLC1 encodes a membrane glycoprotein with an extracellular domain homologous to those of immunoglobulin superfamily proteins. Truncation of TSLC1 in the cytoplasmic domain in a primary human tumor suggests that this domain is important for tumor suppressor activity. Here, we report the isolation of two TSLC1-like genes, TSLL1 and TSLL2, based on their structural homology with the sequences corresponding to the cytoplasmic domain of TSLC1. Significant similarity was also observed in the extracellular domain as well as in the overall gene structure, indicating that these three genes form a unique subfamily (the TSLC1-gene family) in the immunoglobulin superfamily genes. In contrast to the ubiquitous expression of TSLC1, TSLL1 is expressed exclusively in adult and fetal human brain, while TSLL2 is expressed in several specific tissues including prostate, brain, kidney and some other organs. Expression of TSLL1 and TSLL2 was lost or markedly reduced in many human glioma cell lines or some prostate cancer cell lines, suggesting that loss of expression of these genes might be involved in some human cancers.

Gene organization of bovine BCNT that contains a portion corresponding to an endonuclease domain derived from an RTE-1 (Bov-B LINE), non-LTR retrotransposable element: duplication of an intramolecular repeat unit downstream of the truncated RTE-1.

BCNT (a protein named after Bucentaur or craniofacial development protein 1) has a unique structure in Ruminantia. Bovine BCNT contains a region of the endonuclease domain derived from a truncated RTE-1 (previously called Bov-B LINE), a non-LTR retrotransposable repetitive element, and two repeat units (intramolecular repeat, IR) each with 40 amino acids in the C-terminal region. In contrast the human and mouse BCNT proteins contain one repeat unit and lack the RTE-1-derived portion. The 3' UTR of bovine bcnt cDNA also contains an approximately 300-bp portion homologous to the 3'-part of RTE-1. We examined the bovine bcnt genomic DNA sequence to understand how the bovine bcnt gene has been organized. The sequence of 3' UTR homologous portion was found to more closely resemble the Art2 element than the bovine RTE-1. By PCR screening a bovine/hamster hybrid somatic cell panel, the bovine bcnt gene was mapped to chromosome 18, syntenic human chromosome 16q on which human BCNT is located. The bcnt genomic DNA sequence corresponding to the cDNA downstream of a RTE-1 derived portion reveals that each IR unit is flanked by both 5'-side and 3'-side introns and that 3'-UTR consists of one exon. The alignment of the above sequence with a bovine RTE-1 did not show any significant homology downstream of the endonuclease domain. On the other hand, the alignment of the intron sequences with each other revealed that the six sequential homologous segments ranging in size from 40 to 453 bp existed over a 1 kb long sequence between both the 5'- and 3'-side introns flanking each bovine IR unit. In addition, both the 174-bp of 5'-side intron and 80-bp of 3'-side intron neighboring each 120-bp IR exon are significantly homologous among the two bovine IRs, human IR and mouse IR. These results suggest that a truncated bovine RTE-1 was inserted into the intron upstream of an IR unit of an ancestor bcnt gene and that a duplication of a relatively long region that includes IR occurred in the bovine genome.

Four single-nucleotide polymorphisms in the human BUB1 gene.

Four single-nucleotide polymorphisms have been found in the human BUB1 gene, which encodes a kinase involved in the mitotic spindle checkpoint. A cytosine-to-thymine change in exon 10, corresponding to codon 375 (c.1124C>T), causes an amino acid substitution of serine to phenylalanine. A guanine/cytosine polymorphism in exon 4 (c.279G>C) and a thymine/cytosine polymorphism in exon 12 (c.1293T>C) do not cause amino acid substitution. The other polymorphism, of thymine/cytosine (IVS9-8T>C), is found at 8bp upstream of exon 10. As mutations of the hBUB1 gene were reported in a subset of human cancers, these polymorphisms could provide useful tools for the genetic study of susceptibility to various human cancers.

TSLC1 is a tumor-suppressor gene in human non-small-cell lung cancer.

The existence of tumor-suppressor genes was originally demonstrated by functional complementation through whole-cell and microcell fusion. Transfer of chromosome 11 into a human non-small-cell lung cancer (NSCLC) cell line, A549, suppresses tumorigenicity. Loss of heterozygosity (LOH) on the long arm of chromosome 11 has been reported in NSCLC and other cancers. Several independent studies indicate that multiple tumor-suppressor genes are found in this region, including the gene PPP2R1B at 11q23-24 (ref. 7). Linkage studies of NSCLC are precluded because no hereditary forms are known. We previously identified a region of 700 kb on 11q23.2 that completely suppresses tumorigenicity of A549 human NSCLC cells. Most of this tumor-suppressor activity localizes to a 100-kb segment by functional complementation. Here we report that this region contains a single confirmed gene, TSLC1, whose expression is reduced or absent in A549 and several other NSCLC, hepatocellular carcinoma (HCC) and pancreatic cancer (PaC) cell lines. TSLC1 expression or suppression is correlated with promoter methylation state in these cell lines. Restoration of TSLC1 expression to normal or higher levels suppresses tumor formation by A549 cells in nude mice. Only 2 inactivating mutations of TSLC1 were discovered in 161 tumors and tumor cell lines, both among the 20 primary tumors with LOH for 11q23.2. Promoter methylation was observed in 15 of the other 18 primary NSCLC, HCC and PaC tumors with LOH for 11q23.2. Thus, attenuation of TSLC1 expression occurred in 85% of primary tumors with LOH. Hypermethylation of the TSLC1 promoter would seem to represent the 'second hit' in NSCLC with LOH.

Microsatellite instability and the PTEN1 gene mutation in a subset of early onset gliomas carrying germline mutation or promoter methylation of the hMLH1 gene.

High-frequent microsatellite instability (MSI-H) was detected in two of the 80 gliomas examined, whlie the other 78 gliomas showed microsatellite stable (MSS) phenotype. Both of the two MSI-H tumors were glioblastomas which developed in teenage patients. One of the patient was diagnosed as having Turcot's syndrome and had a germline mutation in the hMLH1 gene. Loss of expression due to promoter methylation was selectively observed in the wild type allele of the hMLH1 gene in the tumor of this patient. The other patient had neither a family history nor a past personal history of malignancy. Although no mutation in the mismatch repair genes was detected in the tumor of this patient, the level of expression of the hMLH1 gene was markedly decreased and the promoter sequence of the gene was highly methylated. In the tumor of this patient, the PTEN1 gene, one of the genes carrying microsatellite sequences in their coding regions, was altered by a slippage mutation within five adenine repeat sequences. These findings indicate that the genetic or epigenentic inactivation of the hMLH1 gene is involved in a subset of early-onset gliomas and the PTEN1 gene could be a downstream target for mutation as observed in glioblastoma without MSI.

Detection of amplification of a chromosomal fragment at 6p21 including the cyclin D3 gene in a glioblastoma cell line by arbitrarily primed polymerase chain reaction.

DNA from 10 human glioma cell lines was analyzed by arbitrarily primed polymerase chain reaction. By fingerprinting of the DNA fragments obtained, the presence of fragment Qx with an abnormal signal was detected in one of the glioblastoma cell lines, CCF-STTG1. The nucleotide sequence of this fragment of 387 base pairs showed no homology with any known sequences. Southern-blot analysis using Qx as a probe revealed that the abnormal signal was caused by amplification of DNA by about 50-fold. By analysis of radiation hybrid panels, the fragment was shown to be derived from a chromosomal region on 6p21. The cyclin D3 (ccnd3) gene and an EST locus, H40682, both of which were located in this region, were amplified by about 50-fold in this cell line. Two other loci, R75654 and M78872, flanking the Qx, CCND3 and H40682 loci, were not amplified, suggesting that the size of the amplicon was less than 62 cR. Since over-expression of the ccnd3 gene, but not the H40682 locus, was detected in the cell line CCF-STTG1, the increased amounts of cyclin D3 caused by gene amplification could be involved in the development and/or progression of this glioblastoma.

Partial nuclear localization of a bovine phosphoprotein, BCNT, that includes a region derived from a LINE repetitive sequence in Ruminantia.

BCNT, named after Bucentaur, is a protein that contains a 324-amino-acid region derived from part of a long interspersed DNA sequence element (LINE) in Ruminantia. However, the unique portion is completely missing in human and mouse BCNTs. Since no significant information on their function has been obtained by homology search, we at first examined cellular localization and biochemical characteristics of bovine BCNT to get a hint on its function. Subcellular fractionation and immunohistochemical analyses using a normal bovine epithelial cell line and bovine brain revealed that a significant amount of bovine BCNT is localized in the nuclei, while the major portion is present in the cytosol. Furthermore, it was shown that bovine BCNT is a phosphoprotein and that both bovine and human BCNTs are phosphorylated by casein kinase II in vitro. These results show that BCNTs consist of a unique family, probably a substrate of casein kinase II, which may contribute further to the understanding of gene evolution.

Localization of tumor suppressor activity important in nonsmall cell lung carcinoma on chromosome 11q.

Loss of heterozygosity on chromosome 11q23 is observed at high frequency in human nonsmall cell lung carcinomas (NSCLCs), suggesting the presence of a tumor suppressor gene. Previous analysis of DNA from 79 patients identified a commonly deleted segment of 5 centimorgans. Complementation analysis was used to further localize a putative tumor suppressor gene. Three yeast artificial chromosome (YAC) clones spanning the minimal loss of heterozygosity region were modified, and spheroplast fusion was used to transfer them into human A549 NSCLC or murine Lewis lung carcinoma (LLC) cell lines. The resulting yeast x human hybrid cell lines containing an intact copy of a 1.6-Mb YAC, 939b12, showed reduced growth in vitro. Injection of parental A549 cells into athymic (nu/nu) mice resulted in tumor formation at 27 of 28 injection sites. In contrast, two independent 939b12-containing cell lines formed tumors at only 3 of 20 injection sites. 939b12 also suppressed tumor formation by LLC NSCLC cells in nude mice, but YACs 785e12 and 911f2, which flank 939b12, had no suppressor activity. Further localization of tumor suppression activity on 939b12 was accomplished by introduction of defined fragmentation derivatives into A549 cells and by analysis of YACs that were broken on transfer into LLC cells. This complementation approach localized tumor suppression activity to the central 700 kb of 939b12 and provides a functional assay for positional cloning of this tumor suppressor gene.

Existence of a bovine LINE repetitive insert that appears in the cDNA of bovine protein BCNT in ruminant, but not in human, genomes.

A novel protein, BCNT, originally isolated from bovine brain and named after Bucentaur, contains an internal portion that is translated from part of bovine LINE repetitive sequence (Bov-B LINE). Human cDNA highly homologous to the bovine bcnt (bbcnt) cDNA has been isolated but does not contain a sequence similar to the Bov-B LINE insert (Nobukuni, T., Kobayashi, M., Omori, A., Ichinose, S., Iwanaga, T., Takahashi, I., Hashimoto, K., Hattori, S., Kaibuchi, K., Miyata, Y., Masui, T., Iwashita, S., 1997. An Alu-linked repetitive sequence corresponding to 280 amino acids is expressed in a novel bovine protein, but not in its human homologue. J. Biol. Chem. 272, 2801-2807). In this study, we conducted a polymerase chain reaction analysis to investigate whether such a Bov-B LINE insert is present in bcnt orthologs in other animals and in the genomic sequence of the human BCNT (hBCNT) gene. The results indicate that the Bov-B LINE insert is present in the genomic sequences of bcnt orthologs from sheep, goats, axis deer, and mouse deer (chevrotain), that is in Ruminantia, but not in pigs or human. Analysis of the bbcnt genomic sequence around the Bov-B LINE insert revealed a large part of the inserted Bov-B LINE sequence to be included in an exon; this is followed by a 54-nucleotide sequence that is highly homologous to Bov-B LINE in the 3'-side intron. The hBCNT gene was isolated and found to consist of seven exons and six introns, among which the intron corresponding to the Bov-B LINE insertion site in the bbcnt genome is 16.5kb in length with no sequence similar to Bov-B LINE. Based on these results, it seems likely that the Bov-B LINE insert is derived from a long Bov-B LINE repetitive sequence transposed to an ancestral bcnt gene in Ruminantia and reformed as a new exon through new splicing sites in the transposed sequence.

An Alu-linked repetitive sequence corresponding to 280 amino acids is expressed in a novel bovine protein, but not in its human homologue.

A novel protein harboring a 280-amino acid region from an Alu-linked repetitive sequence (bovine Alu-like dimer-driven family) was isolated from a bovine brain S-100 fraction using monoclonal antibodies against a rat GTPase-activating protein that shares the same epitope. The protein has an apparent molecular mass of 97 kDa (p97). Western blot analysis using extracts prepared from various tissues showed p97 to be predominantly detected in brain and moderately in liver and lung. From sequence analysis of the cDNA encoding p97, it was found that the 840-base pair sequence homologous to a part of the bovine Alu-like dimer-driven family, which has never been shown to be expressed, occurs in the middle of the protein coding region. The protein also contains a pair of intramolecular repeats composed of 40 highly hydrophilic amino acids at the C terminus. Human cDNA homologous to p97 was cloned, and its nucleotide sequence demonstrates that the 840-base pair repetitive sequence and one of the intramolecular repeats are missing. We named p97 bovine BCNT after Bucentaur. These results show that bovine BCNT is a unique molecule and suggest that an analysis of the relationship between bovine bcnt and its human homologue may help further the understanding of gene organization and evolution.

Identification of the prfC gene, which encodes peptide-chain-release factor 3 of Escherichia coli.

The termination of protein synthesis in bacteria requires two codon-specific polypeptide release factors, RF-1 and RF-2. A third factor, RF-3, which stimulates the RF-1 and RF-2 activities, was originally identified in Escherichia coli, but it has received little attention since the 1970s. To search for the gene encoding RF-3, we selected nonsense-suppressor mutations by random insertion mutagenesis on the assumption that a loss of function of RF-3 would lead to misreading of stop signals. One of these mutations, named tos-1 (for transposon-induced opal suppressor), mapped to the 99.2 min region on the E. coli chromosome and suppressed all three stop codons. Complementation studies and analyses of the DNA and protein sequences revealed that the tos gene encodes a 59,442-Da protein, with sequence homology to elongation factor EF-G, including G-domain motifs, and that the tos-1 insertion eliminated the C-terminal one-fifth of the protein. Extracts containing the overproduced Tos protein markedly increased the formation of ribosomal termination complexes and stimulated the RF-1 or RF-2 activity in the codon-dependent in vitro termination assay. The stimulation was significantly reduced by GTP, GDP, and the beta,gamma-methylene analog of GTP, but not by GMP. These results fit perfectly with those described in the original publications on RF-3, and the tos gene has therefore been designated prfC. A completely null prfC mutation made by reverse genetics affected the cell growth under the limited set of physiological and strain conditions.