Cellular strategies for accommodating replication-hindering adducts in DNA: control by the SOS response in Escherichia coli.

The replication of double-stranded plasmids containing a single adduct was analyzed in vivo by means of a sequence heterology that marks the two DNA strands. The single adduct was located within the sequence heterology, making it possible to distinguish trans-lesion synthesis (TLS) events from damage avoidance events in which replication did not proceed through the lesion. When the SOS system of the host bacteria is not induced, the C8-guanine adduct formed by the carcinogen N-2-acetylaminofluorene (AAF) yields less than 1% of TLS events, showing that replication does not readily proceed through the lesion. In contrast, the deacetylated adduct N-(deoxyguanosin-8-yl)-2-aminofluorene yields approximately 70% of TLS events under both SOS-induced and uninduced conditions. These results for TLS in vivo are in good agreement with the observation that AAF blocks DNA replication in vitro, whereas aminofluorene does so only weakly. Induction of the SOS response causes an increase in TLS events through the AAF adduct (approximately 13%). The increase in TLS is accompanied by a proportional increase in the frequency of AAF-induced frameshift mutations. However, the polymerase frameshift error rate per TLS event was essentially constant throughout the SOS response. In an SOS-induced delta umuD/C strain, both US events and mutagenesis are totally abolished even though there is no decrease in plasmid survival. Error-free replication evidently proceeds efficiently by means of the damage avoidance pathway. We conclude that SOS mutagenesis results from increased TLS rather than from an increased frameshift error rate of the polymerase.

Channel behavior in a gamma-aminobutyrate transporter.

Current produced by a gamma-aminobutyrate (GABA) transporter stably transfected into a mammalian cell line was observed in cell-attached and excised membrane patches. When GABA was absent, a fraction of the transporters produced cation-permeable channels. When GABA plus Na+ was on either side of the membrane, the majority of transporters produced a high-frequency current noise attributed to the movement of ions in an occluded pore.

Restricted expression of homeobox genes distinguishes fetal from adult human smooth muscle cells.

Smooth muscle cell plasticity is considered a prerequisite for atherosclerosis and restenosis following angioplasty and bypass surgery. Identification of transcription factors that specify one smooth muscle cell phenotype over another therefore may be of major importance in understanding the molecular basis of these vascular disorders. Homeobox genes exemplify one class of transcription factors that could govern smooth muscle cell phenotypic diversity. Accordingly, we screened adult and fetal human smooth muscle cell cDNA libraries with a degenerate oligonucleotide corresponding to a highly conserved region of the homeodomain with the idea that homeobox genes, if present, would display a smooth muscle cell phenotype-dependent pattern of expression. No homeobox genes were detected in the adult human smooth muscle cell library; however, five nonparalogous homeobox genes were uncovered from the fetal library (HoxA5, HoxA11, HoxB1, HoxB7, and HoxC9). Northern blotting of adult and fetal tissues revealed low and restricted expression of all five homeobox genes. No significant differences in transcripts of HoxA5, HoxA11, and HoxB1 were detected between adult or fetal human smooth muscle cells in culture. HoxB7 and HoxC9, however, showed preferential mRNA expression in fetal human smooth muscle cells that appeared to correlate with the age of the donor. This phenotype-dependent expression of homeobox genes was also noted in rat pup versus adult smooth muscle cells. While similar differences in gene expression have been reported between subsets of smooth muscle cells from rat vessels of different-aged animals or clones of rat smooth muscle, our findings represent a demonstration of a transcription factor distinguishing two human smooth muscle cell phenotypes.

Purification, cDNA cloning, functional expression, and characterization of a 26-kDa endogenous mammalian carboxypeptidase inhibitor.

The recent demonstration of the occurrence in rat brain and other nonpancreatic tissues of carboxypeptidase A (CPA) gene transcripts without associated catalytic activity could be ascribed to the presence of a soluble endogenous protein inhibitor. This tissue carboxypeptidase inhibitor (TCI), detected by the inhibition of added bovine pancreatic CPA, was purified from rat brain. Peptides were obtained by partial proteolysis of purified TCI, a protein of approximately 30 kDa, and starting from their sequences, a full-length cDNA encoding a 223-amino acid protein containing three potential phosphorylation sites was cloned from a cDNA library. Its identity with TCI was shown by expression in Escherichia coli of a recombinant protein recognized by antibodies raised against native TCI and display characteristic CPA-inhibiting activity. TCI appears as a hardly reversible, non-competitive, and potent inhibitor of CPA1 and CPA2 (Ki approximately 3 nM) and mast-cell CPA (Ki = 16 nM) and inactive on various other proteases. This pattern of selectivity might be attributable to a limited homology of a 11-amino acid sequence with sequences within the activation segments of CPA and CPB known to interact with residues within their active sites. The widespread expression of TCI in a number of tissues (e.g., brain, lung, or digestive tract) and its apparently cytosolic localization point to a rather general functional role, e.g., in the control of cytosolic protein degradation.

State of the art in continuous speech recognition.

In the past decade, tremendous advances in the state of the art of automatic speech recognition by machine have taken place. A reduction in the word error rate by more than a factor of 5 and an increase in recognition speeds by several orders of magnitude (brought about by a combination of faster recognition search algorithms and more powerful computers), have combined to make high-accuracy, speaker-independent, continuous speech recognition for large vocabularies possible in real time, on off-the-shelf workstations, without the aid of special hardware. These advances promise to make speech recognition technology readily available to the general public. This paper focuses on the speech recognition advances made through better speech modeling techniques, chiefly through more accurate mathematical modeling of speech sounds.

Training and search methods for speech recognition.

Speech recognition involves three processes: extraction of acoustic indices from the speech signal, estimation of the probability that the observed index string was caused by a hypothesized utterance segment, and determination of the recognized utterance via a search among hypothesized alternatives. This paper is not concerned with the first process. Estimation of the probability of an index string involves a model of index production by any given utterance segment (e.g., a word). Hidden Markov models (HMMs) are used for this purpose [Makhoul, J. & Schwartz, R. (1995) Proc. Natl. Acad. Sci. USA 92, 9956-9963]. Their parameters are state transition probabilities and output probability distributions associated with the transitions. The Baum algorithm that obtains the values of these parameters from speech data via their successive reestimation will be described in this paper. The recognizer wishes to find the most probable utterance that could have caused the observed acoustic index string. That probability is the product of two factors: the probability that the utterance will produce the string and the probability that the speaker will wish to produce the utterance (the language model probability). Even if the vocabulary size is moderate, it is impossible to search for the utterance exhaustively. One practical algorithm is described [Viterbi, A. J. (1967) IEEE Trans. Inf. Theory IT-13, 260-267] that, given the index string, has a high likelihood of finding the most probable utterance.

Low density lipoprotein receptor-related protein mediates apolipoprotein E-dependent neurite outgrowth in a central nervous system-derived neuronal cell line.

The epsilon 4 allele of apolipoprotein E (apoE) is a major risk factor for Alzheimer disease, suggesting that apoE may directly influence neurons in the aging brain. Recent data suggest that apoE-containing lipoproteins can influence neurite outgrowth in an isoform-specific fashion. The neuronal mediators of apoE effects have not been clarified. We show here that in a central nervous system-derived neuronal cell line, apoE3 but not apoE4 increases neurite extension. The effect of apoE3 was blocked at low nanomolar concentrations by purified 39-kDa protein that regulates ligand binding to the low density lipoprotein receptor-related protein (LRP). Anti-LRP antibody also completely abolished the neurite-promoting effect of apoE3. Understanding isoform-specific cell biological processes mediated by apoE-LRP interactions in central nervous system neurons may provide insight into Alzheimer disease pathogenesis.

The rat suprachiasmatic nucleus is a clock for all seasons.

Seasonal changes of daylength (photoperiod) affect the expression of hormonal and behavioral circadian rhythms in a variety of organisms. In mammals, such effects might reflect photoperiodic changes in the circadian pace-making system [located in the suprachiasmatic nucleus (SCN) of the hypothalamus] that governs these rhythms, but to date no functionally relevant, intrinsic property of the SCN has been shown to be photoperiod dependent. We have analyzed the temporal regulation of light-induced c-fos gene expression in the SCN of rats maintained in long or short photoperiods. Both in situ hybridization and immunohistochemical assays show that the endogenous circadian rhythm of light responsiveness in the SCN is altered by photoperiod, with the duration of the photosensitive subjective night under the short photoperiod 5-6 h longer than under the long photoperiod. Our results provide evidence that a functional property of the SCN is altered by photoperiod and suggest that the nucleus is involved in photoperiodic time measurement.

Ordered restriction endonuclease maps of yeast artificial chromosomes created by optical mapping on surfaces.

We have developed a surface mounting technology for the rapid construction of ordered restriction maps from individual DNA molecules. Optical restriction maps constructed from yeast artificial chromosome DNA molecules mounted on specially derivatized glass surfaces are accurate and reproducible, and the technology is amenable to automation. The mounting procedures described here should also be useful for fluorescence in situ hybridization studies. We believe these improvements to optical mapping will further stimulate the development of nonelectrophoretic approaches to genome analysis.

DNA recombination is sufficient for retroviral transduction.

Oncogenic retroviruses carry coding sequences that are transduced from cellular protooncogenes. Natural transduction involves two nonhomologous recombinations and is thus extremely rare. Since transduction has never been reproduced experimentally, its mechanism has been studied in terms of two hypotheses: (i) the DNA model, which postulates two DNA recombinations, and (ii) the RNA model, which postulates a 5' DNA recombination and a 3' RNA recombination occurring during reverse transcription of viral and protooncogene RNA. Here we use two viral DNA constructs to test the prediction of the DNA model that the 3' DNA recombination is achieved by conventional integration of a retroviral DNA 3' of the chromosomal protooncogene coding region. For the DNA model to be viable, such recombinant viruses must be infectious without the purportedly essential polypurine tract (ppt) that precedes the 3' long terminal repeat (LTR) of all retroviruses. Our constructs consist of a ras coding region from Harvey sarcoma virus which is naturally linked at the 5' end to a retroviral LTR and artificially linked at the 3' end either directly (construct NdN) or by a cellular sequence (construct SU) to the 5' LTR of a retrovirus. Both constructs lack the ppt, and the LTR of NdN even lacks 30 nucleotides at the 5' end. Both constructs proved to be infectious, producing viruses at titers of 10(5) focus-forming units per ml. Sequence analysis proved that both viruses were colinear with input DNAs and that NdN virus lacked a ppt and the 5' 30 nucleotides of the LTR. The results indicate that DNA recombination is sufficient for retroviral transduction and that neither the ppt nor the complete LTR is essential for retrovirus replication. DNA recombination explains the following observations by others that cannot be reconciled with the RNA model: (i) experimental transduction is independent of the packaging efficiency of viral RNA, and (ii) experimental transduction may invert sequences with respect to others, as expected for DNA recombination during transfection.