Consensus methods for finding and ranking DNA binding sites. Application to Escherichia coli promoters

A reexamination of information theory-based methods for DNA-binding site identification

Background

Searching for transcription factor binding sites in genome sequences is still an open problem in bioinformatics. Despite substantial progress, search methods based on information theory remain a standard in the field, even though the full validity of their underlying assumptions has only been tested in artificial settings. Here we use newly available data on transcription factors from different bacterial genomes to make a more thorough assessment of information theory-based search methods.

Results

Our results reveal that conventional benchmarking against artificial sequence data leads frequently to overestimation of search efficiency. In addition, we find that sequence information by itself is often inadequate and therefore must be complemented by other cues, such as curvature, in real genomes. Furthermore, results on skewed genomes show that methods integrating skew information, such as Relative Entropy, are not effective because their assumptions may not hold in real genomes. The evidence suggests that binding sites tend to evolve towards genomic skew, rather than against it, and to maintain their information content through increased conservation. Based on these results, we identify several misconceptions on information theory as applied to binding sites, such as negative entropy, and we propose a revised paradigm to explain the observed results.

Conclusion

We conclude that, among information theory-based methods, the most unassuming search methods perform, on average, better than any other alternatives, since heuristic corrections to these methods are prone to fail when working on real data. A reexamination of information content in binding sites reveals that information content is a compound measure of search and binding affinity requirements, a fact that has important repercussions for our understanding of binding site evolution.

DNA binding site characterization by means of Rényi entropy measures on nucleotide transitions

In this work, parametric information-theory measures for the characterization of binding sites in DNA are extended with the use of transitional probabilities on the sequence. We propose the use of parametric uncertainty measures such as Rényi entropies obtained from the transition probabilities for the study of the binding sites, in addition to nucleotide frequency-based Rényi measures. Results are reported in this work comparing transition frequencies (i.e., dinucleotides) and base frequencies for Shannon and parametric Rényi entropies for a number of binding sites found in E. Coli, lambda and T7 organisms. We observe that the information provided by both approaches is not redundant. Furthermore, under the presence of noise in the binding site matrix we observe overall improved robustness of nucleotide transition-based algorithms when compared with nucleotide frequency-based method.

Cohabitation of two different lexA regulons in Pseudomonas putida

In contrast to the vast majority of the members of the domain Bacteria, several Pseudomonas and Xanthomonas species have two lexA genes, whose products have been shown to recognize different LexA binding motifs, making them an interesting target for studying the interplay between cohabiting LexA regulons in a single species. Here we report an analysis of the genetic composition of the two LexA regulons of Pseudomonas putida KT2440 performed with a genomic microarray. The data obtained indicate that one of the two LexA proteins (LexA1) seems to be in control of the conventional Escherichia coli-like SOS response, while the other LexA protein (LexA2) regulates only its own transcriptional unit, which includes the imuA, imuB, and dnaE2 genes, and a gene (PP_3901) from a resident P. putida prophage. Furthermore, PP_3901 is also regulated by LexA1 and is required for DNA damage-mediated induction of several P. putida resident prophage genes. In silico searches suggested that this marked asymmetry in regulon contents also occurs in other Pseudomonas species with two lexA genes, and the implications of this asymmetry in the evolution of the SOS network are discussed.

Anatomy of Escherichia coli sigma70 promoters

Information theory was used to build a promoter model that accounts for the -10, the -35 and the uncertainty of the gap between them on a common scale. Helical face assignment indicated that base -7, rather than -11, of the -10 may be flipping to initiate transcription. We found that the sequence conservation of sigma70 binding sites is 6.5 +/- 0.1 bits. Some promoters lack a -35 region, but have a 6.7 +/- 0.2 bit extended -10, almost the same information as the bipartite promoter. These results and similarities between the contacts in the extended -10 binding and the -35 suggest that the flexible bipartite sigma factor evolved from a simpler polymerase. Binding predicted by the bipartite model is enriched around 35 bases upstream of the translational start. This distance is the smallest 5' mRNA leader necessary for ribosome binding, suggesting that selective pressure minimizes transcript length. The promoter model was combined with models of the transcription factors Fur and Lrp to locate new promoters, to quantify promoter strengths, and to predict activation and repression. Finally, the DNA-bending proteins Fis, H-NS and IHF frequently have sites within one DNA persistence length from the -35, so bending allows distal activators to reach the polymerase.

Extracting relations between promoter sequences and their strengths from microarray data

MOTIVATION

The relations between the promoter sequences and their strengths were extensively studied in the 1980s. Although these studies uncovered strong sequence-strength correlations, the cost of their elaborate experimental methods have been too high to be applied to a large number of promoters. On the contrary, a recent increase in the microarray data allows us to compare thousands of gene expressions with their DNA sequences.

RESULTS

We studied the relations between the promoter sequences and their strengths using the Escherichia coli microarray data. We modeled those relations using a simple weight matrix, which was optimized with a novel support vector regression method. It was observed that several non-consensus bases in the '-35' and '-10' regions of promoter sequences act positively on the promoter strength and that certain consensus bases have a minor effect on the strength. We analyzed outliers for which the observed gene expressions deviate from the promoter strength predictions, and identified several genes with enhanced expressions due to multiple promoters and genes under strong regulation by transcription factors. Our method is applicable to other procaryotes for which both the promoter sequences and the microarray data are available.

Novel group of virulence activators within the AraC family that are not restricted to upstream binding sites

Several regulators within the AraC family control the expression of genes known or thought to be required for virulence of bacterial pathogens. One of these, Rns, activates transcription from an unprecedented variety of binding-site locations. Although nearly all prokaryotic activators bind within a small region upstream and adjacent to the promoter that they regulate, Rns does not bind within this region to activate its own promoter, Prns. Instead, to activate Prns, Rns requires one binding site 224.5 bp upstream and one downstream of the transcription start site. We show in this study that several other AraC family activators recognize the same binding sites as Rns and share with it the ability to utilize a downstream binding site. Like Rns, other members of this group of activators positively regulate the expression of virulence factors in pathogenic bacteria. These regulators are also able to activate transcription from promoter-proximal upstream binding sites since they are able to substitute for Rns at Pcoo, a promoter with only upstream binding sites. Thus, Rns is the prototype for a group of regulators, which include CfaR, VirF, AggR, and CsvR and which activate transcription from locations that are more diverse than those of any other known activator.

Promoter sequences and algorithmical methods for identifying them

This paper presents a survey of currently available mathematical models and algorithmical methods for trying to identify promoter sequences. The methods concern both searching in a genome for a previously defined consensus and extracting a consensus from a set of sequences. Such methods were often tailored for either eukaryotes or prokaryotes although this does not preclude use of the same method for both types of organisms. The survey therefore covers all methods; however, emphasis is placed on prokaryotic promoter sequence identification. Illustrative applications of the main extracting algorithms are given for three bacteria.

Recognition of regulatory sites by genomic comparison

Availability of complete bacterial genomes opens the way to the comparative approach to the recognition of transcription regulatory sites. Assumption of regulon conservation in conjunction with profile analysis provides two lines of independent evidence making it possible to make highly specific predictions. Recently this approach was used to analyze several regulons in eubacteria and archaebacteria. The present review covers recent advances in the comparative analysis of transcriptional regulation in prokaryotes and phylogenetic fingerprinting techniques in eukaryotes, and describes the emerging patterns of the evolution of regulatory systems.

New techniques for DNA sequence classification

DNA sequence classification is the activity of determining whether or not an unlabeled sequence S belongs to an existing class C. This paper proposes two new techniques for DNA sequence classification. The first technique works by comparing the unlabeled sequence S with a group of active motifs discovered from the elements of C and by distinction with elements outside of C. The second technique generates and matches gapped fingerprints of S with elements of C. Experimental results obtained by running these algorithms on long and well conserved Alu sequences demonstrate the good performance of the presented methods compared with FASTA. When applied to less conserved and relatively short functional sites such as splice-junctions, a variation of the second technique combining fingerprinting with consensus sequence analysis gives better results than the current classifiers employing text compression and machine learning algorithms.

A comprehensive library of DNA-binding site matrices for 55 proteins applied to the complete Escherichia coli K-12 genome

A major mode of gene regulation occurs via the binding of specific proteins to specific DNA sequences. The availability of complete bacterial genome sequences offers an unprecedented opportunity to describe networks of such interactions by correlating existing experimental data with computational predictions. Of the 240 candidate Escherichia coli DNA-binding proteins, about 55 have DNA-binding sites identified by DNA footprinting. We used these sites to construct recognition matrices, which we used to search for additional binding sites in the E. coli genomic sequence. Many of these matrices show a strong preference for non-coding DNA. Discrepancies are identified between matrices derived from natural sites and those derived from SELEX (Systematic Evolution of Ligands by Exponential enrichment) experiments. We have constructed a database of these proteins and binding sites, called DPInteract (available at http://arep.med.harvard.edu/dpinteract).

A general procedure for locating and analyzing protein-binding sequence motifs in nucleic acids

In the last decade, two tools, one drawn from information theory and the other from artificial neural networks, have proven particularly useful in many different areas of sequence analysis. The work presented herein indicates that these two approaches can be joined in a general fashion to produce a very powerful search engine that is capable of locating members of a given nucleic acid sequence family in either local or global sequence searches. This program can, in turn, be queried for its definition of the motif under investigation, ranking each base in context for its contribution to membership in the motif family. In principle, the method used can be applied to any binding motif, including both DNA and RNA sequence families, given sufficient family size.

Promoter region of the Escherichia coli O7-specific lipopolysaccharide gene cluster: structural and functional characterization of an upstream untranslated mRNA sequence

We report the identification of the promoter region of the Escherichia coli O7-specific lipopolysaccharide (LPS) gene cluster (wbEcO7). Typical -10 and -35 sequences were found to be located in the intervening region between galF and rlmB, the first gene of the wbEcO7 cluster. Data from RNase protection experiments revealed the existence of an untranslated leader mRNA segment of 173 bp, including the JUMPStart and two ops sequences. We characterized the structure of this leader mRNA by using the program Mfold and a combination of nested and internal deletions transcriptionally fused to a promoterless lac operon. Our results indicated that the leader mRNA may fold into a series of complex stem-loop structures, one of which includes the JUMPStart element. We have also found that one of the ops sequences resides on the predicted stem and the other resides on the loop region, and we confirmed that these sequences are essential for the RfaH-mediated regulation of the O polysaccharide cluster. A very similar stem-loop structure could be predicted in the promoter region of the LPS core operon encoding the waaQGPSBIJYZK genes. We observed another predicted stem-loop, located immediately downstream from the wbEcO7 transcription initiation site, which appeared to be involved in premature termination of transcription. This putative stem-loop is common to many other O polysaccharide gene clusters but is not present in core oligosaccharide genes. wbEcO7-lac transcriptional fusions in single copy numbers were also used to determine the effects of various environmental cues in the transcriptional regulation of O polysaccharide synthesis. No effects were detected with temperature, osmolarity, Mg2+ concentration, and drugs inducing changes in DNA supercoiling. We therefore conclude that the wbEcO7 promoter activity may be constitutive and that regulation takes place at the level of elongation of the mRNA in a RfaH-mediated manner.

Expression, purification, and functional analysis of the TyrR protein of Haemophilus influenzae

The gene that was inferred to encode the TyrR protein of Haemophilus influenzae Rd was synthesized by polymerase chain reaction and inserted into a T7-based expression vector. Methods were developed to overexpress the TyrR protein of H. influenzae in Escherichia coli and to purify the protein on a large scale. Both in vitro and in vivo functional comparisons of the H. influenzae and E. coli TyrR proteins were carried out. The TyrR protein of H. influenzae was able to bind in vitro to an operator target upstream of the aroF-tyrA gene of E. coli. In the presence of [gamma-S]ATP, the DNA binding ability of the H. influenzae TyrR protein was drastically reduced. Despite the much shorter peptide chain length (318 amino acid residues vs 513), the TyrR protein of H. influenzae was as active in repressing the aroF promoter as the TyrR protein of E. coli. Repression by both proteins was enhanced in the presence of tyrosine; however, the transcriptional activation function associated with the TyrR protein of E. coli could not be detected when the H. influenzae TyrR protein was expressed in E. coli. By computer analysis, at least five operator targets for TyrR were identified within the genomic DNA of H. influenzae. These observations show that the assignment of function to the tyrR gene of H. influenzae was correctly made. Further studies of the H. influenzae TyrR protein in comparison to its E. coli counterpart should provide valuable mechanistic information on transcriptional regulation in this system.

Genetic analysis of the dTDP-rhamnose biosynthesis region of the Escherichia coli VW187 (O7:K1) rfb gene cluster: identification of functional homologs of rfbB and rfbA in the rff cluster and correct location of the rffE gene

The O-repeating unit of the Escherichia coli O7-specific lipopolysaccharide is made of galactose, mannose, rhamnose, 4-acetamido-4,6-dideoxyglucose, and N-acetyglucosamine. We have recently characterized the genes involved in the biosynthesis of the sugar precursor GDP-mannose occurring in the E. coli O7:K1 strain VW187 (C. L. Marolda and M. A. Valvano, J. Bacteriol. 175:148-158, 1993). In the present study, we identified and sequenced the rfbBDAC genes encoding the enzymes for the biosynthesis of another precursor, dTDP-rhamnose. These genes are localized on the upstream end of the rfbEcO7 region, and they are strongly conserved compared with similar genes found in various enteric and nonenteric bacteria. Upstream of rfbB we identified a DNA segment containing the rfb promoter and a highly conserved untranslated leader sequence also present in the promoter regions of other surface polysaccharide gene clusters. Also, we have determined that rfbB and rfbA have homologs, rffG (o355) and rffH (o292), respectively, located on the rff cluster, which is involved in the synthesis of enterobacterial common antigen. We provide biochemical evidence that rffG and rffH encode dTDP-glucose dehydratase and glucose-1-phosphate thymidylyltransferase activities, respectively, and we also show that rffG complemented the rfbB defect in the O7+ cosmid pJHCV32. We also demonstrate that rffG is distinct from rffE and map the rffE gene to the second gene of the rff cluster.

A consensus sequence for binding of Lrp to DNA

Lrp (leucine-responsive regulatory protein) is a major regulatory protein involved in the expression of numerous operons in Escherichia coli. For ilvIH, one of the operons positively regulated by Lrp, Lrp binds to multiple sites upstream of the transcriptional start site and activates transcription. An alignment of 12 Lrp binding sites within ilvIH DNA from two different organisms revealed a tentative consensus sequence AGAAT TTTATTCT (Q. Wang, M. Sacco, E. Ricca, C.T. Lago, M. DeFelice, and J.M. Calvo, Mol. Microbiol. 7:883-891, 1993). To further characterize the binding specificity of Lrp, we used a variation of the Selex procedure of C. Tuerk and L. Gold (Science 249:505-510, 1990) to identify sequences that bound Lrp out of a pool of 10(12) different DNA molecules. We identified 63 related DNA sequences that bound Lrp and estimated their relative binding affinities for Lrp. A consensus sequence derived from analysis of these sequences, YAGHAWATTWT DCTR, where Y = C or T, H = not G, W = A or T, D = not C, and R = A or G, contains clear dyad symmetry and is very similar to the one defined earlier. To test the idea that Lrp in the presence of leucine might bind to a different subset of DNA sequences, we carried out a second selection experiment with leucine present during the binding reactions. DNA sequences selected in the presence or absence of leucine were similar, and leucine did not stimulate binding to any of the sequences that were selected in the presence of leucine. Therefore, it is unlikely that leucine changes the specificity of Lrp binding.

Analysis of consensus sequence patterns in Giardia cytoskeleton gene promoters

Protein-coding genes in the ancient eukaryote Giardia lamblia lack typical promoter consensus elements. We have analysed the immediate 5' flanking sequences of seven genes of related function (structural cytoskeleton proteins) to identify shared DNA motifs that might have a role in transcription initiation. Transcription start sites for five genes have been determined previously. Genomic mapping and mRNA primer extension experiments demonstrate additionally that the genes for beta-giardin and median body protein are (i) present as single copies in the genome, (ii) transcribed with very short 5' leader sequences. Two search algorithms designed to extract conserved motifs from either aligned or non-aligned sequences independently discovered three sites constituting a common pattern in all seven promoters. Sites were optimally aligned using weight matrix building trials to achieve the maximum 'information content'. Profiling the information content of best alignments defines the extent of the homologies as: a 9 bp box (initiator) at the start site and upstream 18 and 6 bp boxes. The initiator is the most highly conserved element and contains a universal Py-A-Pu motif at which transcription starts. We show that the best matrices can be combined in a search pattern that correctly locates transcription start sites in genomic DNA sequences.

Molecular structure of the Frankia spp. nifD-K intergenic spacer and design of Frankia genus compatible primer

The nifD-K intergenic spacer (IGS) of ArI3 and ACoN24d were found to have a length 265 and 199 nucleotides, respectively. They are markedly less conserved than the two neighbouring genes and have, in some instances, a repeated structure reminiscent of an insertion event. The repeated sequence and the IGSs have no detectable homology with sequences in DNA databanks. The IGS has a stem-loop structure with a low folding energy, lower than that between nifH and nifD. No convincing alignment of IGS sequences could be obtained among Frankia strains. Only between ACoN24d and ArI3, which belong to the same genomic species, was the alignment good enough to permit detection of a doubly repeated structure. No promoter could be detected in the IGSs. The putative nifK open reading frame (ORF) in Frankia strain ArI3 has a length of 1587 nucleotides, starting with a GTG codon, preceded by a ribosome binding site of a structure similar to that of nifH (GGAGGN7). The codon usage was similar to that of previously sequenced Frankia genes with a strong bias toward G- and C-ending codons except in the case of glycine where GGT is frequent. Alignment of the three Frankia nifK sequences (EUN1f; ArI3 and ACoN24d) with those of other nitrogen-fixing bacteria permitted detection of a sequence conserved among the three Frankia strains but absent in the other sequences. A primer targeted to that region in combination with FGPD807-85 amplified the nifD-KIGS sequences of all Frankia strains (except the non-nitrogen-fixing Frankia strains CN3 and AgB1-9) and yet failed to amplify DNA of all other nitrogen-fixing bacteria.(ABSTRACT TRUNCATED AT 250 WORDS)

Identification of ribosome binding sites in Escherichia coli using neural network models

This study investigated the use of neural networks in the identification of Escherichia coli ribosome binding sites. The recognition of these sites based on primary sequence data is difficult due to the multiple determinants that define them. Additionally, secondary structure plays a significant role in the determination of the site and this information is difficult to include in the models. Efforts to solve this problem have so far yielded poor results. A new compilation of E. coli ribosome binding sites was generated for this study. Feedforward backpropagation networks were applied to their identification. Perceptrons were also applied, since they have been the previous best method since 1982. Evaluation of performance for all the neural networks and perceptrons was determined by ROC analysis. The neural network provided significant improvement in the recognition of these sites when compared with the previous best method, finding less than half the number of false positives when both models were adjusted to find an equal number of actual sites. The best neural network used an input window of 101 nucleotides and a single hidden layer of 9 units. Both the neural network and the perceptron trained on the new compilation performed better than the original perceptron published by Stormo et al. in 1982.

Prediction of function in DNA sequence analysis

Recognition of function of newly sequenced DNA fragments is an important area of computational molecular biology. Here we present an extensive review of methods for prediction of functional sites, tRNA, and protein-coding genes and discuss possible further directions of research in this area.

Nucleotide sequence of the region between crr and cysM in Salmonella typhimurium: five novel ORFs including one encoding a putative transcriptional regulator of the phosphotransferase system

A 4471 bp region between crr and cysM on the Salmonella typhimurium chromosome (49.5 min) has been sequenced. Five ORFs were found within this region, one of which is likely to be the putative regulatory gene, ptsJ, that corresponds in map position to a gene which when mutated allows expression of a cryptic Enzyme I of the phosphotransferase system. The deduced amino acid sequence of the encoded protein is similar to those of several open reading frames (ORFs) including ORFT2 of Rhodobacter spheroides with which it is 28% identical throughout most of its length (comparison score of 21 S.D.). PtsJ exhibits a putative, N-terminal, helix-turn-helix, DNA binding domain that is similar in sequence to those in members of the GntR family of transcriptional regulators. Analyses of the sequences of the ORFs encoded within this region are presented.

Functional organization of the bacteriophage PRD1 genome

PRD1 is a broad-host-range virus that infects Escherichia coli cells. It has a linear double-stranded DNA genome that replicates by a protein-primed mechanism. The virus particle is composed of a protein coat enclosing a lipid membrane. On the basis of this structure, PRD1 is being used as a membrane biosynthesis and structure model. In this investigation, we constructed the transcription map of the 15-kb-long phage genome. This was achieved by a computer search of putative promoters, which were then tested for activity by primer extension and for the capability to promote the synthesis of chloramphenicol acetyltransferase.

Posttranscriptional accumulation of chloroplast tufA (elongation factor gene) mRNA during chloroplast development in Chlamydomonas reinhardtii

Light induces chloroplast (Cp) differentiation in dark-grown y-1 strains of Chlamydomonas reinhardtii. Slot blot analysis was used to quantitate tufA, psbA, psbK, rbcL, and 16S rRNA transcript accumulation and transcription during Cp differentiation. When etiolated cc-125 y-1 cells were illuminated for 5 h, a 1710 bp tufA mRNA accumulated up to 5-fold while the psbA, rbcL, and 16S rRNA transcripts accumulated less than 1.5-fold. The tufA gene encodes translational elongation factor EF-Tu. The light-induced accumulation of tufA mRNA did not occur in cc-1931, a strain that does not become etiolated in darkness. Pulse labelling was used to measure the transcription of Cp transcripts during tufA mRNA accumulation, and no detectable change in tufA transcription was observed. These results imply that the half life of the tufA transcript increases during the greening process.

Compilation of E. coli mRNA promoter sequences

An updated compilation of 300 E. coli mRNA promoter sequences is presented. For each sequence the most recent relevant paper was checked, to verify the location of the transcriptional start position as identified experimentally. We comment on the reliability of the sequence databanks and analyze the conservation of known promoter features in the current compilation. This database is available by E-mail.

Molecular analysis of the 3,6-dideoxyhexose pathway genes of Yersinia pseudotuberculosis serogroup IIA

Salmonella enterica and Yersinia pseudotuberculosis are the only examples in nature known to use a variety of 3,6-dideoxyhexose derivatives as O antigen constituents. To allow a comparison of the responsible biosynthetic genes of the two organisms, we have sequenced a section of the Y. pseudotuberculosis serogroup IIA rfb region that contained the genes for the abequose biosynthetic pathway. Comparison of the identified genes with the rfb region of S. enterica LT2 showed that the two dideoxyhexose pathway gene clusters are related. The arrangement of the genes was largely conserved, and the G + C compositions of the two DNA regions were strikingly similar; however, the degree of conservation of nucleotide and protein sequences suggested that the two gene clusters have been evolving independently for considerable time. Hybridization experiments showed that the dideoxyhexose pathway genes are widespread throughout the various serogroups of Y. pseudotuberculosis.

An assessment of neural network and statistical approaches for prediction of E. coli promoter sites

We have constructed a perceptron type neural network for E. coli promoter prediction and improved its ability to generalize with a new technique for selecting the sequence features shown during training. We have also reconstructed five previous prediction methods and compared the effectiveness of those methods and our neural network. Surprisingly, the simple statistical method of Mulligan et al. performed the best amongst the previous methods. Our neural network was comparable to Mulligan's method when false positives were kept low and better than Mulligan's method when false negatives were kept low. We also showed the correlation between the prediction rates of neural networks achieved by previous researchers and the information content of their data sets.

Escherichia coli promoters: neural networks develop distinct descriptions in learning to search for promoters of different spacing classes

Back-propagation neural networks were trained to recognize promoter sequences of each of the three major spacing classes found in E. coli. These networks were trained with the object of maximizing their ability to generalize while maintaining the level of false positive identifications at a fraction of 1 percent. These objectives were generally met. Networks for the 16 base spacing class captured between 78 and 100% of previously unseen promoters in different tests; networks for the 17 base class identified 97% of the test promoters; networks for the 18 base class identified 79% of the test promoters. A tandem poll of networks for all three spacing classes produced a cumulative false positive level of less than 0.5%. In each case, the weight matrices used by the networks in their classification were analyzed to determine the relative weight assigned to the occurrence of a given base at a given position within the promoter. In this fashion, an approximate description of the network's definition of the promoter can be obtained.

Isolation of DNA damage-inducible promoters in Escherichia coli: regulation of polB (dinA), dinG, and dinH by LexA repressor

A new genetic screening method has been developed to isolate Escherichia coli promoters which are components of the SOS regulon. Plasmids containing the regulatory regions of polB (dinA) and two new loci, dinG and dinH, were characterized. Galactokinase gene fusion experiments indicated that transcription of these genes is inducible by treatment with mitomycin and conforms to a classical model of SOS regulation involving simple LexA repression. Mapping studies using the E. coli DNA library of Kohara et al. (Y. Kohara, K. Akiyama, and K. Isono, Cell 50:495-508, 1987) revealed that dinG and dinH are located at 17.8 and 19.8 min on the chromosome, respectively. The nucleotide sequence of the dinH regulatory region contains a segment which is very similar to previously characterized binding sites for LexA protein. An asymmetric, noncanonical 20-bp LexA operator in the cloned dinG promoter region was identified. Additional experiments have revealed that the nucleotide sequence of the gene immediately downstream of the DNA damage-inducible polB locus encodes a polypeptide which has extensive sequence homology to several known and putative DNA and RNA helicase proteins. This gene, which is not regulated by the LexA repressor, has been designated hepA. The predicted amino acid sequence of the product of hepA contains several highly conserved sequence motifs that are also found in enzymes such as the RecQ and UvrB proteins of E. coli and the Rad3 protein of Saccharomyces cerevisiae.

Expectation maximization algorithm for identifying protein-binding sites with variable lengths from unaligned DNA fragments

An Expectation Maximization algorithm for identification of DNA binding sites is presented. The approach predicts the location of binding regions while allowing variable length spacers within the sites. In addition to predicting the most likely spacer length for a set of DNA fragments, the method identifies individual sites that differ in spacer size. No alignment of DNA sequences is necessary. The method is illustrated by application to 231 Escherichia coli DNA fragments known to contain promoters with variable spacings between their consensus regions. Maximum-likelihood tests of the differences between the spacing classes indicate that the consensus regions of the spacing classes are not distinct. Further tests suggest that several positions within the spacing region may contribute to promoter specificity.

Molecular characterization of mutations affecting expression level and growth rate-dependent regulation of the Escherichia coli zwf gene

We characterized three cis dominant mutations which elevate glucose 6-phosphate dehydrogenase level. Growth rate-dependent regulation and oxidative stress control of enzyme level were altered by the mutations. DNA sequencing and transcript mapping showed that the "up" mutations created new promoters whose hyperactive expression overrides the normal regulation of the native promoter.

SQUIRREL: Sequence QUery, Information Retrieval and REporting Library. A program package for analyzing signals in nucleic acid sequences for the VAX

A computer tool is described for comparison, analysis and search of genetic signals. The method is based on sequence consensus matrices. It assumes that a genetic signal (such as a promoter, enhancer or whatever) is composed of several signal blocks separated from each other by variable distances. A set of programs is presented to perform the analysis. The result of such an analysis is a description of the investigated signal including matrices for each signal block, distances between each block and distribution of the values. Programs are provided to search for a signal using results from previous analysis. The method is able to align large sets of sequences within a few minutes and to check the quality of the alignment. An analysis of E.coli promoters is provided as an example.

Sequence and evolution of the FruR protein of Salmonella typhimurium: a pleiotropic transcriptional regulatory protein possessing both activator and repressor functions which is homologous to the periplasmic ribose-binding protein

The repressor of the fructose (fru) operon of Salmonella typhimurium (FruR) has been implicated in the transcriptional regulation of dozens of genes concerned with central metabolic pathways of carbon utilization. We here report the nucleotide sequence of the gene encoding FruR and analyse both its operator-promoter region and its deduced amino acyl sequence. The FruR protein was overexpressed and was shown to have a molecular weight of about 36 kDa in agreement with the molecular weight deduced from the gene sequence. Sequence analyses revealed that FruR is homologous to 9 distinct bacterial DNA-binding proteins, most of which recognize sugar inducers and all of which possess helix-turn-helix motifs within their N-terminal regions and exhibit sequence identity throughout most of their lengths. FruR is also homologous to the periplasmic ribose-binding protein which serves as a constituent of the ribose transport/chemoreception system. The ribose-binding protein is in turn homologous to binding proteins specific for arabinose and galactose. The periplasmic binding proteins, the structures of some of which have been elucidated in three dimensions, lack the N-terminal helix-turn-helix region, but instead possess N-terminal hydrophobic signal sequences which target them to the periplasm. A phylogenetic tree for the more closely related proteins of this superfamily was constructed, and a signature motif was identified which should facilitate future detection of additional transcriptional regulatory proteins belonging to this family.

Nucleotide sequence of the Rhodobacter capsulatus fruK gene, which encodes fructose-1-phosphate kinase: evidence for a kinase superfamily including both phosphofructokinases of Escherichia coli

The fruK gene encoding fructose-1-phosphate kinase (FruK), located within the fructose (fru)-catabolic operon of Rhodobacter capsulatus, was sequenced. FruK of R. capsulatus (316 amino acids; molecular weight = 31,232) is the same size as and is homologous to FruK of Escherichia coli, phosphofructokinase B (PfkB) of E. coli, phosphotagatokinase of Staphylococcus aureus, and ribokinase of E. coli. These proteins therefore make up a family of homologous proteins, termed the PfkB family. A phylogenetic tree for this new family was constructed. Sequence comparisons plus chemical inactivation studies suggested the lack of involvement of specific residues in catalysis. Although the Rhodobacter FruK differed markedly from the other enzymes within the PfkB family with respect to amino acid composition, these enzymes exhibited similar predicted secondary structural features. A large internal segment of the Rhodobacter FruK was found to be similar in sequence to the domain bearing the sugar bisphosphate-binding region of the large subunit of ribulose 1,5-bisphosphate carboxylase/oxygenase of plants and bacteria. Proteins of the PfkB family did not exhibit statistically significant sequence identity with PfkA of E. coli. PfkA, however, is homologous to other prokaryotic and eukaryotic ATP- and PPi-dependent Pfks (the PfkA family). These eukaryotic, ATP-dependent enzymes each consist of a homotetramer (mammalian) or a heterooctamer (yeasts), with each subunit containing an internal duplication of the size of the entire PfkA protein of E. coli. In some of these enzymes, additional domains are present. A phylogenetic tree was constructed for the PfkA family and revealed that the bacterial enzymes closely resemble the N-terminal domains of the eukaryotic enzyme subunits whereas the C-terminal domains have diverged more extensively. The PPi-dependent Pfk of potato is only distantly related to the ATP-dependent enzymes. On the basis of their similar functions, sizes, predicted secondary structures, and sequences, we suggest that the PfkA and PfkB families share a common evolutionary origin.

Structure and sequence of the rfb (O antigen) gene cluster of Salmonella serovar typhimurium (strain LT2)

The rfb gene cluster of Salmonella LT2 has been cloned and sequenced. The genes rfbA, rfbB, rfbD, rfbF, rfbG, rfbK, rfbM and rfbP were located individually and the gene rfbL was located outside the cluster. Approximately 16 open reading frames were found in the region which is essential for the expression of O antigen. The gene products of rfbB and rfbG were found to have homology with the group of dehydrogenase and related enzymes described previously. Analysis of the G + C ratio of the rfb cluster extended the area of low-G + C composition previously found in the sequence of rfbJ to the whole rfb gene cluster. Three to five segments with discrete G + C contents and codon adaptation indices are present in the rfb region, indicating a heterogeneous origin of these segments. Potential promoters were found near the start of the rfb region, supporting the possibility that the rfb gene cluster is an operon.

Molecular characterization of the Escherichia coli K-12 zwf gene encoding glucose 6-phosphate dehydrogenase

In Escherichia coli K-12, expression of zwf, the gene for glucose 6-phosphate dehydrogenase, is coordinated with the cellular growth rate and induced by superoxide-generating agents. To initiate the study of the molecular mechanisms regulating its expression, the gene was cloned and its DNA sequence was determined. The 5' ends of zwf mRNA isolated from cells growing in glucose and acetate minimal media were mapped. The map was complex in that transcripts mapped to -45, -52, and -62, with respect to the beginning of the coding sequence. Three analytical methods were used to search the DNA sequence for putative promoters. Only one sequence for a promoter recognized by the sigma 70 form of RNA polymerase was found by all three search routines that could be aligned with a mapped transcript, indicating that the other transcripts arise by processing of the mRNA. A computer-assisted search did not reveal a thermodynamically stable long-range mRNA secondary structure that is capable of sequestering the translation initiation region, which suggests that growth-rate-dependent regulation of glucose 6-phosphate dehydrogenase level may not be carried out by a mechanism similar to the one for the gene (gnd) for 6-phosphogluconate dehydrogenase. The DNA segment between the -10 hexamer and the start point of transcription resembles the discriminator sequence of stable RNA genes, which has been implicated in stringent control and growth-rate-dependent regulation.

Training back-propagation neural networks to define and detect DNA-binding sites

A three layered back-propagation neural network was trained to recognize E. coli promoters of the 17 base spacing class. To this end, the network was presented with 39 promoter sequences and derivatives of those sequences as positive inputs; 60% A + T random sequences and sequences containing 2 promoter-down point mutations were used as negative inputs. The entire promoter sequence of 58 bases, approximately -50 to +8, was entered as input. The network was asked to associate an output of 1.0 with promoter sequence input and 0.0 with non-promoter input. Generally, after 100,000 input cycles, the network was virtually perfect in classifying the training set. A trained network was about 80% effective in recognizing 'new' promoters which were not in the training set, with a false positive rate below 0.1%. Network searches on pBR322 and on the lambda genome were also performed. Overall the results were somewhat better than the best rule-based procedures. The trained network can be analyzed both for its choice of base and relative weighting, positive and negative, at each position of the sequence. This method, which requires only appropriate input/output training pairs, can be used to define and search for any DNA regulatory sequence for which there are sufficient exemplars.

Overproduction and characterization of the iclR gene product of Escherichia coli K-12 and comparison with that of Salmonella typhimurium LT2

The iclR gene of Escherichia coli K-12, which encodes a regulatory protein (repressor) for the aceBAK operon, is located between that operon and metH in the 91-min region of the chromosome. The iclR gene was cloned and expressed in a coupled T7 RNA polymerase/promoter system and the gene product was identified by specific binding to a fragment containing the aceBAK operator region. The iclR gene product is a polypeptide of 274 amino acids (aa) with a calculated Mr of 29,741. Comparison of the deduced IclR aa sequence to that of Salmonella typhimurium revealed that the two IclR repressors exhibit 89% identity. A possible helix-turn-helix motif characteristic of DNA-binding proteins was found within the IclR sequence. A search in protein data banks revealed that IclR has a score of similarity of 43.7% with GylR, a transcriptional regulator of the glycerol operon of Streptomyces coelicolor.

An expectation maximization (EM) algorithm for the identification and characterization of common sites in unaligned biopolymer sequences

Statistical methodology for the identification and characterization of protein binding sites in a set of unaligned DNA fragments is presented. Each sequence must contain at least one common site. No alignment of the sites is required. Instead, the uncertainty in the location of the sites is handled by employing the missing information principle to develop an "expectation maximization" (EM) algorithm. This approach allows for the simultaneous identification of the sites and characterization of the binding motifs. The reliability of the algorithm increases with the number of fragments, but the computations increase only linearly. The method is illustrated with an example, using known cyclic adenosine monophosphate receptor protein (CRP) binding sites. The final motif is utilized in a search for undiscovered CRP binding sites.