Go with the Flow

Although two-dimensional (2-D) gel electrophoresis continues to be a robust, informative approach to proteomic analysis, much of the recent attention in the field has centered on multidimensional liquid chromatographic separation of peptides followed by tandem mass spectrometry. Despite the popularity of strong cation-exchange chromatography combined with reversed-phase liquid chromatography (RPLC), isoelectric focusing (the first dimension of 2-D gel electrophoresis in which proteins are separated by p1) looks to be enjoying a revival. Recent work has demonstrated that isoelectric focusing via commercial device, gel, or silica capillary can replace the cation-exchange step in traditional liquid chromatographic approaches. Although these published reports confirm that high-resolution separation is possible, they suffer from troubling limitations on sample loading capacity. To address this issue, some proteomics researchers have sought inspiration from the past. Continuous free flow electrophoresis made its debut in the early 1970s; it is a liquid-phase isoelectric focusing technique in which carrier ampholyte solution passes through a flat chamber in a direction perpendicular to an applied electric field. Because samples can be continuously injected, large amounts of sample can be accommodated. Moreover, the ongoing flow means that downtime for system regeneration is avoided, with obvious benefits to sample throughput. What has been missing from isoelectric focusing for shotgun proteomics is, however, an easy way to link the first separation with RPLC-MS. Malmström et al., now describe a protocol allowing just that. Given that some of the authors are behind PeptideProphet, a popular tool for evaluating the reliability of MS-based peptide identification, the team also describes a modification of that statistical program to incorporate pI data. Working with a cytoplasmic extract of Drosophila cells, the authors show that the combination of an optimized pI gradient, separation media, and data processing workflow led to accurate, efficient edification of more than 10,000 unique peptides. Hence, this pre-MS separation method represents an attractive option for shotgun proteomics.

- Malmström et al. 2006. Optimized peptide separation and identification for mass spectrometry based proteomics via free-flow electrophoresis. Journal of Proteome Research [Epub ahead of print, July 29, 2006].

The Long and Short of It

Methods development is seen as an intensely practical, empirical effort. However, theory also contributes significantly to improved practices, particularly when procedures have become so second nature that they are no longer questioned. In that spirit, M.C. Wendl, who is based at Washington University's Genome Sequencing Center, revisits the theory behind shotgun sequencing in a recent paper. Although the underlying chemistry was not substantially modified, the pressure of the human genome project forced rapid evolution of sequencing strategies in order to maximize throughput. The optimized process that resulted typically involves concentrating on paired reads of short inserts, selective use of large-insert clones to derive long-range linking information, and significant redundancy (sequencing depth). These time-tested strategies have clearly been enormously successful; however, Wendl, points out that an increasing number of projects focus on smaller-scale targets, such as small macro-nuclear chromosomes or genomic islands derived by filtering libraries of repetitive elements. In addition, current theory may not apply to technologies such as pyrosequencing. Wendl, provides a detailed mathematical treatment of read pairing and edge effects in order to develop a revised model that analyzes how sequence coverage might be optimized for genomic targets of small to moderate size. Although his paper discusses a large number of theoretical implications, the most significant findings suggest that better coverage may result from reduced read length; that single-stranded reads are more likely to yield better results than the now-standard double-stranded approach; and that sequencing should be limited to 4- or 5-fold redundancy. Wendl, observes that these theoretical specifications match well with the properties of pyrosequencing and suggests that the theory implies that a single run cycle of a pyrosequencing instrument could be divided among two or more projects without compromising sequencing quality.

- Wendl. 2006. A general coverage theory for shotgun DNA sequencing. Journal of Computational Biology 13:1177-1196.

Transcription Proscription

Despite the ongoing enthusiasm for RNA interference, successful gene expression manipulation by gene therapy is likely to require a variety of tools. One means to block transcription before it even begins is by introducing DNA molecules encoding transcription factor binding sites. These exogenous molecules could act as decoys and sequester particular transcription factors from interacting with their genuine targets. One important challenge to using this approach is that oligonucleotides do not generally cross cell membranes unassisted. However, nanoparticles decorated with appropriate targeting peptides can enter the nucleus. Thus, an important question is whether decoy DNA-studded nanoparticles can efficiently interfere with transcription. A team at North Carolina State University, having extensive experience with biofunctionalized gold nanoparticles, investigated this matter in a model system involving in vitro T7 RNA polymerase transcription assays. They conjugated putative decoy oligonucleotides to gold nanoparticles using standard chemistries and controlled surface density by using a passivating agent that competed with the thiol-based DNA-nanoparticle linkages. Through a fluorescence assay, the researchers determined that the surface density of the singlestranded DNA needed to be limited in order to ensure efficient hybridization of the complementary DNA strand. While surface density also plays a role in the efficacy of the decoy molecules as assessed by IC₅₀ assays, the more important parameter appears to be particle size. Although smaller nanoparticles might be expected to pose fewer steric hindrances to the approach of a protein, the authors instead observed that 15-nm particles outperformed 10-nm particles. This effect exceeded the level of advantage provided by the increased number of binding sites on the larger particle. This study shows the value of empirically examining parameters for optimal nanoparticle-protein binding and suggests that bioconjugated nanoparticles can act as decoys for transcription interference.

- Agbasi-Porter et al. 2006. Transcription inhibition using oligonucleotide-modified gold nanoparticles. Bioconjugate Chemistry [Epub ahead of print, July 12, 2006].

Just Add Water

In addition to wet-lab approaches, protein-protein interactions can also be mapped by computationally modeling the binding interface through docking. Although water molecules are important in real-life interactions, existing protein-protein docking methods have not, to date, incorporated solvent contributions. This situation has become increasingly ripe for change as additional information regarding the role of water molecules in structure-resolved protein complexes becomes available. van Dijk and Bonvin, draw upon this emerging trend and describe a modification to their docking program HADDOCK that explicitly incorporates water molecules during prediction of the protein-protein interface. In the first step of the approach, the individual proteins are solvated with one layer of water molecules. Each hydrated polypeptide is then considered as a rigid body, and the docking procedure is initiated. In these early steps and before the subsequent energy minimization, extraneous water molecules (not present at the interface) are removed. However, for further refinement of the structure, all bystander waters must be deleted. In order to identify those water molecules that should be removed, the authors examined a set of high-resolution protein structures and derived statistical descriptions of the observed patterns of amino acid-water contacts. After using the resulting probabilistic model to filter the remaining water molecules to 25% of the input, the docking protocol performs another rigid body energy minimization step. As a final step, the structures are docked under semi-flexible conditions. To evaluate this computational protocol using known protein complexes, van Dijk and Bonvin, docked components of 10 previously characterized complexes. By comparing docking results using unmodified HADDOCK and the new solvation-dependent procedure, the authors confirm that including water molecules produces structures that more closely match the experimentally determined results. In addition to improving prediction of protein-protein binding interfaces, the authors suggest that solvated HADDOCK may be of particular interest for modeling DNA-protein complexes, which often make extensive use of water in intermolecular contacts.

- van Dijk and Bonvin. 2006. Solvated docking: introducing water into the modeling of biomolecular complexes. Bioinformatics [Epub ahead of print, August 9, 2006].