The human DNA ligase III gene encodes nuclear and mitochondrial proteins

We provide evidence that the human DNA ligase III gene encodes a mitochondrial form of this enzyme. First, the DNA ligase III cDNA contains an in-frame ATG located upstream from the putative translation initiation start site. The DNA sequence between these two ATG sites encodes an amphipathic helix similar to previously identified mitochondrial targeting peptides. Second, recombinant green fluorescent protein harboring this sequence at its amino terminus was efficiently targeted to the mitochondria of Cos-1 monkey kidney cells. In contrast, native green fluorescent protein distributed to the cytosol. Third, a series of hemagglutinin-DNA ligase III minigene constructs were introduced into Cos-1 cells, and immunocytochemistry was used to determine subcellular localization of the epitope-tagged DNA ligase III protein. These experiments revealed that inactivation of the upstream ATG resulted in nuclear accumulation of the DNA ligase III protein, whereas inactivation of the downstream ATG abolished nuclear localization and led to accumulation within the mitochondrial compartment. Fourth, mitochondrial protein extracts prepared from human cells overexpressing antisense DNA ligase III mRNA possessed substantially less DNA ligase activity than did mitochondrial extracts prepared from control cells. DNA end-joining activity was also substantially reduced in extracts prepared from antisense mRNA-expressing cells. From these results, we conclude that the human DNA ligase III gene encodes both nuclear and mitochondrial enzymes. DNA ligase plays a central role in DNA replication, recombination, and DNA repair. Thus, identification of a mitochondrial form of this enzyme provides a tool with which to dissect mammalian mitochondrial genome dynamics.     

The activated human met gene encodes a protein tyrosine kinase

We have raised antibodies against a synthetic dodecapeptide corresponding to the carboxyl terminus of the predicted met gene product. Phosphorylation of 60 kDa and 65 kDa proteins on tyrosine residues was observed when immunoprecipitates of cells containing the activated human met gene were incubated with [gamma-32P]ATP. Phosphoproteins with the same molecular masses could be immunoprecipitated from cells metabolically labelled with [32P]orthophosphate. When considered together, these observations indicate that the activated human met gene encodes 60 kDa and 65 kDa proteins that can catalyse autophosphorylation on tyrosine residues.     

The P gene of human parainfluenza virus type 1 encodes P and C proteins but not a cysteine-rich V protein.

The nucleotide sequence of the P gene of human parainfluenza virus type 1 (PIV1) was determined from cloned cDNA copies of the mRNA. By analogy with the gene organization of Sendai virus, two open reading frames in the mRNA sense of the gene were identified as coding sequences for the P protein (568 amino acids with an estimated molecular weight of 64,655) and the C protein (204 amino acids with an estimated molecular weight of 24,108). Comparison of the deduced amino acid sequences of the P and C proteins of PIV1 with those of Sendai virus showed a high degree of homology. However, a sequence for the cysteine-rich V protein, which was considered a common feature of other paramyxoviruses, was interrupted by the presence of multiple stop codons. The sequence analysis of three P-gene-specific cDNA clones generated from genomic RNA by polymerase chain reaction and one additional clone generated from mRNA confirmed that the coding sequence for the cysteine-rich region is silent in the PIV1 gene and thus is not translated into protein. Two potential editing sites with the consensus sequence 3'UUYUCCC were found in the PIV1 P gene at positions 564 to 570 and 1430 to 1436. However, examination of the PIV1 mRNA population by a primer extension method indicated that neither of these sites is utilized. These results indicate that the PIV1 P gene has a coding strategy different from those of other paramyxovirus P genes.     

The LGI1/epitempin gene encodes two protein isoforms differentially expressed in human brain

The leucine-rich, glioma inactivated 1 (LGI1)/Epitempin gene has been linked to two phenotypes as different as gliomagenesis and autosomal dominant lateral temporal epilepsy. Its function and the biochemical features of the encoded protein are unknown. We characterized the LGI1/Epitempin protein product by western blot analysis of mouse and human brain tissues. Two proteins of about 60 and 65 kDa were detected by an anti-LGI1 antibody within the expected molecular mass range. The two proteins appeared to reside in different subcellular compartments, as they were fractionated by differential centrifugation. The specificity of both polypeptides was validated by cell transfection assay and mass spectrometry analysis. Immunoblot analysis of protein distribution in various zones of the human brain revealed variable amounts of both proteins. Notably, these proteins were more abundant in the temporal neocortex than in the hippocampus, the difference in abundance of the 65-kDa product being particularly pronounced. These results suggest that the two protein isoforms encoded by LGI1/Epitempin are differentially expressed in the human brain, and that higher expression levels of these proteins in the lateral temporal cortex may underlie the susceptibility of this brain region to the epileptogenic effects of LGI1/Epitempin mutations.     

The neurofibromatosis type 1 gene encodes a protein related to GAP

cDNA walking and sequencing have extended the open reading frame for the neurofibromatosis type 1 gene (NF1). The new sequence now predicts 2485 amino acids of the NF1 peptide. A 360 residue region of the new peptide shows significant similarity to the known catalytic domains of both human and bovine GAP (GTPase activating protein). A much broader region, centered around this same 360 amino acid sequence, is strikingly similar to the yeast IRA1 product, which has a similar amino acid sequence and functional homology to mammalian GAP. This evidence suggests that NF1 encodes a cytoplasmic GAP-like protein that may be involved in the control of cell growth by interacting with proteins such as the RAS gene product. Mapping of the cDNA clones has confirmed that NF1 spans a t(1;17) translocation mutation and that three active genes lie within an intron of NF1, but in opposite orientation.     

The platelet-derived growth factor-inducible KC gene encodes a secretory protein related to platelet alpha-granule proteins

A full length cDNA clone of the platelet-derived growth factor-inducible KC gene has been isolated, sequenced, and expressed in COS cells. Both sequence analysis and expression studies indicate that KC encodes a secretory protein. Sequence analysis shows that, furthermore, the protein encoded by KC belongs to a growing superfamily of inducible proteins with a common ancestral linkage to the platelet alpha-granule proteins, platelet factor 4, and connective tissue-activating peptide III. A computer-generated phylogenetic tree documents interrelationships between KC and six additional members of this peptide superfamily. The KC gene is, in all probability, the murine homologue of a human gene termed "gro." By extension, the KC protein is the murine counterpart of the protein encoded by the gro gene. The gro protein corresponds to a factor described as "melanoma growth-stimulating activity."     

The human homolog of the JE gene encodes a monocyte secretory protein.

The mouse fibroblast gene, JE, was one of the first platelet-derived growth factor-inducible genes to be described as such. The protein encoded by JE (mJE) is the prototype of a large family of secreted, cytokinelike glycoproteins, all of whose members are induced by a mitogenic or activation signal in monocytes macrophages, and T lymphocytes; JE is the only member to have been identified in fibroblasts. We report the identification of a human homolog for murine JE, cloned from human fibroblasts. The protein predicted by the coding sequence of human JE (hJE) is 55 amino acids shorter than mJE, and its sequence is identical to that of a recently purified monocyte chemoattractant. When expressed in COS cells, the human JE cDNA directed the secretion of N-glycosylated proteins of Mr 16,000 to 18,000 as well as proteins of Mr 15,500, 15,000, and 13,000. Antibodies raised against mJE recognized these hJE species, all of which were secreted by human fibroblasts. hJE expression was stimulated in HL60 cells during phorbol myristate acetate-induced monocytoid differentiation. However, resting human monocytes constitutively secreted hJE; treatment with gamma interferon did not enhance hJE expression in monocytes, and treatment with phorbol myristate acetate or lipopolysaccharide inhibited its expression. Thus, human JE encodes yet another member of the large family of JE-related cytokinelike proteins, in this case a novel human monocyte and fibroblast secretory protein.     

The murine Pes1 gene encodes a nuclear protein containing a BRCT domain

Pescadillo was originally identified in the zebrafish Danio rerio as a site of a retrovirus-insertion mutation that caused severe defects during embryogenesis. In particular, growth of the fetal zebrafish liver was significantly affected by loss of pescadillo function. To begin to understand the role of pescadillo during mammalian hepatogenesis we identified the murine homologue of pescadillo and named it Pes1. A single gene localized to chromosome 11 on the mouse genome encodes Pes1. Although Pes1 mRNA was detected in all tissues examined it was present at the highest levels in both adult and fetal liver. Analysis of the predicted amino acid sequence of Pes1 found it to contain a BRCT domain, which has previously been found in several proteins involved in cell-cycle checkpoints and DNA repair. Consistent with a putative role in these processes we found that when recombinant Pes1 protein was expressed in HepG2 cells it localized to the nucleus.     

The Saccharomyces cerevisiae checkpoint gene BUB1 encodes a novel protein kinase.

Normal cell multiplication requires that the events of mitosis occur in a carefully ordered fashion. Cells employ checkpoints to prevent cycle progression until some prerequisite step has been completed. To explore the mechanisms of checkpoint enforcement, we previously screened for mutants of Saccharomyces cerevisiae which are unable to recover from a transient treatment with a benzimidazole-related microtubule inhibitor because they fail to inhibit subsequent cell cycle steps. Two of the identified genes, BUB2 and BUB3, have been cloned and described (M. A. Hoyt, L. Totis, and B. T. Roberts, Cell 66:507-517, 1991). Here we present the characterization of the BUB1 gene and its product. Genetic evidence was obtained suggesting that Bub1 and Bub3 are mutually dependent for function, and immunoprecipitation experiments demonstrated a physical association between the two. Sequence analysis of BUB1 revealed a domain with similarity to protein kinases. In vitro experiments confirmed that Bub1 possesses kinase activity; Bub1 was able to autophosphorylate and to catalyze phosphorylation of Bub3. In addition, overproduced Bub1 was found to localize to the cell nucleus.     

The Bradyrhizobium japonicum nolA gene encodes three functionally distinct proteins

Examination of nolA revealed that NolA can be uniquely translated from three ATG start codons. Translation from the first ATG (ATG1) predicts a protein (NolA1) having an N-terminal, helix-turn-helix DNA-binding motif similar to the DNA-binding domains of the MerR-type regulatory proteins. Translation from ATG2 and ATG3 would give the N-terminally truncated proteins NolA2 and NolA3, respectively, lacking the DNA-binding domain. Consistent with this, immunoblot analyses of Bradyrhizobium japonicum extracts with a polyclonal antiserum to NolA revealed three distinct polypeptides whose molecular weights were consistent with translation of nolA from the three ATG initiation sites. Site-directed mutagenesis was used to produce derivatives of nolA in which ATG start sites were sequentially deleted. Immunoblots revealed a corresponding absence of the polypeptide whose ATG start site was removed. Translational fusions of the nolA mutants to a promoterless lacZ yielded functional fusion proteins in both Escherichia coli and B. japonicum. Expression of NolA is inducible upon addition of extracts from 5-day-old etiolated soybean seedlings but is not inducible by genistein, a known inducer of the B. japonicum nod genes. The expression of both NolA2 and NolA3 requires the presence of NolA1. NolA1 or NolA3 is required for the genotype-specific nodulation of soybean genotype PI 377578.     

The mouse ortholog of the human SMARCB1 gene encodes two splice forms

The human SMARCB1 gene (SWI/SNF-related, matrix-associated, actin-dependent regulator of chromatin, subfamily b, member 1, previously named the INI1/hSNF5 gene) is a tumor suppressor gene located on chromosome 22q11.2 and is inactivated in malignant rhabdoid tumors. By using an EST-based approach, we cloned two splice forms of the Smarcb1 gene in mouse and a longer splice form of the human ortholog. Proteins corresponding to the longer (385 aa) and the shorter (376 aa) forms are 100% conserved between human and mouse. Meningiomas and schwannomas are tumors frequently deleting various regions on chromosome 22, including the SMARCB1 locus. We therefore directly sequenced seven SMARCB1 exons (90% of the open reading frame) in search for mutations in 41 meningiomas and 23 schwannomas. No inactivating mutations were observed, which suggests that the SMARCB1 gene is not involved in the pathogenesis of these tumors.