Current methods for genome-wide analysis of gene expression require fragmentation of original transcripts into small fragments for short-read sequencing. In bacteria, the resulting fragmented information hides operon complexity. Additionally, in vivo processing of transcripts confounds the accurate identification of the 5′ and 3′ ends of operons. Here we develop a methodology called SMRT-Cappable-seq that combines the isolation of un-fragmented primary transcripts with single-molecule long read sequencing. Applied to E. coli, this technology results in an accurate definition of the transcriptome with 34% of known operons from RegulonDB being extended by at least one gene. Furthermore, 40% of transcription termination sites have read-through that alters the gene content of the operons. As a result, most of the bacterial genes are present in multiple operon variants reminiscent of eukaryotic splicing. By providing such granularity in the operon structure, this study represents an important resource for the study of prokaryotic gene network and regulation.

The high resolution refined structures of 23 enzymes were analyzed to determine the properties of amino acids involved in active site regions. These regions were found to be rich in G-X-Y or Y-X-G oligopeptides, where X and Y are polar and non-polar residues, respectively, that are small and with low polarity. Other regions of the enzyme molecules have significantly fewer of these sequences. These features suggest that glycine residues may provide flexibility necessary for enzyme active sites to change conformation, and the G-X-Y or Y-X-G oligopeptides may be a motif for the formation of enzyme active sites.

Computational evaluation of protein–DNA interaction is important for the identification of DNA-binding sites and genome annotation. It could validate the predicted binding motifs by sequence-based approaches through the calculation of the binding affinity between a protein and DNA. Such an evaluation should take into account structural information to deal with the complicated effects from DNA structural deformation, distance-dependent multi-body interactions and solvation contributions. In this paper, we present a knowledge-based potential built on interactions between protein residues and DNA tri-nucleotides. The potential, which explicitly considers the distance-dependent two-body, three-body and four-body interactions between protein residues and DNA nucleotides, has been optimized in terms of a Z-score. We have applied this knowledge-based potential to evaluate the binding affinities of zinc-finger protein–DNA complexes. The predicted binding affinities are in good agreement with the experimental data (with a correlation coefficient of 0.950). On a larger test set containing 48 protein–DNA complexes with known experimental binding free energies, our potential has achieved a high correlation coefficient of 0.800, when compared with the experimental data. We have also used this potential to identify binding motifs in DNA sequences of transcription factors (TF). The TFs in 79.4% of the known TF–DNA complexes have accurately found their native binding sequences from a large pool of DNA sequences. When tested in a genome-scale search for TF-binding motifs of the cyclic AMP regulatory protein (CRP) of Escherichia coli, this potential ranks all known binding motifs of CRP in the top 15% of all candidate sequences.

It was previously shown that a one-dimensional Ising model could successfully simulate the equilibrium binding of myosin S1 to regulated actin filaments (T. L. Hill, E. Eisenberg and L. Greene, Proc. Natl. Acad. Sci. U.S.A. 77:3186-3190, 1980). However, the time course of myosin S1 binding to regulated actin was thought to be incompatible with this model, and a three-state model was subsequently developed (D. F. McKillop and M. A. Geeves, Biophys. J. 65:693-701, 1993). A quantitative analysis of the predicted time course of myosin S1 binding to regulated actin, however, was never done for either model. Here we present the procedure for the theoretical evaluation of the time course of myosin S1 binding for both models and then show that 1) the Hill model can predict the "lag" in the binding of myosin S1 to regulated actin that is observed in the absence of Ca++ when S1 is in excess of actin, and 2) both models generate very similar families of binding curves when [S1]/[actin] is varied. This result shows that, just based on the equilibrium and pre-steady-state kinetic binding data alone, it is not possible to differentiate between the two models. Thus, the model of Hill et al. cannot be ruled out on the basis of existing pre-steady-state and equilibrium binding data. Physical mechanisms underlying the generation of the lag in the Hill model are discussed.