Profile-based string kernels for remote homology detection and motif extraction

We introduce novel profile-based string kernels for use with support vector machines (SVMs) for the problems of protein classification and remote homology detection. These kernels use probabilistic profiles, such as those produced by the PSI-BLAST algorithm, to define position-dependent mutation neighborhoods along protein sequences for inexact matching of k-length subsequences ("k-mers") in the data. By use of an efficient data structure, the kernels are fast to compute once the profiles have been obtained. For example, the time needed to run PSI-BLAST in order to build the profiles is significantly longer than both the kernel computation time and the SVM training time. We present remote homology detection experiments based on the SCOP database where we show that profile-based string kernels used with SVM classifiers strongly outperform all recently presented supervised SVM methods. We further examine how to incorporate predicted secondary structure information into the profile kernel to obtain a small but significant performance improvement. We also show how we can use the learned SVM classifier to extract "discriminative sequence motifs"--short regions of the original profile that contribute almost all the weight of the SVM classification score--and show that these discriminative motifs correspond to meaningful structural features in the protein data. The use of PSI-BLAST profiles can be seen as a semi-supervised learning technique, since PSI-BLAST leverages unlabeled data from a large sequence database to build more informative profiles. Recently presented "cluster kernels" give general semi-supervised methods for improving SVM protein classification performance. We show that our profile kernel results also outperform cluster kernels while providing much better scalability to large datasets.     

Functional Proteomics Screen Enables Enrichment of Distinct Cell Types from Human Pancreatic Islets

The current world-wide epidemic of diabetes has prompted attempts to generate new sources of insulin-producing cells for cell replacement therapy. An inherent challenge in many of these strategies is the lack of cell-surface markers permitting isolation and characterization of specific cell types from differentiating stem cell populations. Here we introduce an iterative proteomics procedure allowing tag-free isolation of cell types based on their function. Our method detects and associates specific cell-surface markers with particular cell functionality by coupling cell capture on antibody arrays with immunofluorescent labeling. Using this approach in an iterative manner, we discovered marker combinations capable of enriching for discrete pancreatic cell subtypes from human islets of Langerhans: insulin-producing beta cells (CD9^high/CD56⁺), glucagon-producing alpha cells (CD9^- /CD56⁺) and trypsin-producing acinar cells (CD9^- /CD56^-). This strategy may assist future beta cell research and the development of diagnostic tools for diabetes. It can also be applied more generally for function-based purification of desired cell types from other limited and heterogeneous biological samples.     

A rapid and affordable screening platform for membrane protein trafficking

Background

Membrane proteins regulate a diversity of physiological processes and are the most successful class of targets in drug discovery. However, the number of targets adequately explored in chemical space and the limited resources available for screening are significant problems shared by drug-discovery centers and small laboratories. Therefore, a low-cost and universally applicable screen for membrane protein trafficking was developed.

Results

This high-throughput screen (HTS), termed IRFAP-HTS, utilizes the recently described MarsCy1-fluorogen activating protein and the near-infrared and membrane impermeant fluorogen SCi1. The cell surface expression of MarsCy1 epitope-tagged receptors can be visualized by simple addition of SCi1. User-friendly, rapid, and quantitative detection occurs on a standard infrared western-blotting scanner. The reliability and robustness of IRFAP-HTS was validated by confirming human vasopressin-2 receptor and dopamine receptor-2 trafficking in response to agonist or antagonist. The IRFAP-HTS screen was deployed against the leucine-rich G protein-coupled receptor-5 (Lgr5). Lgr5 is expressed in stem cells, modulates Wnt/ß-catenin signaling, and is therefore a promising drug target. However, small molecule modulators have yet to be reported. The constitutive internalization of Lgr5 appears to be one primary mode through which its function is regulated. Therefore, IRFAP-HTS was utilized to screen 11,258 FDA-approved and drug-like small molecules for those that antagonize Lgr5 internalization. Glucocorticoids were found to potently increase Lgr5 expression at the plasma membrane.

Conclusion

The IRFAP-HTS platform provides a versatile solution for screening more targets with fewer resources. Using only a standard western-blotting scanner, we were able to screen 5,000 compounds per hour in a robust and quantitative assay. Multi-purposing standardly available laboratory equipment eliminates the need for idiosyncratic and more expensive high-content imaging systems. The modular and user-friendly IRFAP-HTS is a significant departure from current screening platforms. Small laboratories will have unprecedented access to a robust and reliable screening platform and will no longer be limited by the esoteric nature of assay development, data acquisition, and post-screening analysis. The discovery of glucocorticoids as modulators for Lgr5 trafficking confirms that IRFAP-HTS can accelerate drug-discovery and drug-repurposing for even the most obscure targets.

     

Coping with Space Neophobia in Drosophila melanogaster: The Asymmetric Dynamics of Crossing a Doorway to the Untrodden

We discover and examine within a wide phylogenetic perspective spatial neophobia, avoidance of untrodden terrain, in fruit flies, in an experimental setup that reduces the gap between the field and the laboratory. In our setup, fruit flies use a natal fruit as their origin, freely exploring for days their surroundings, which consists of a mixture of trodden and untrodden terrain. The interface between trodden and untrodden is, however, reduced in our setup to a wide doorway, opened within a surrounding wall. Crossing this doorway, characterized by a sharp contrast interface between trodden and untrodden, generates a behavior whose dynamics betrays the flies'' space neophobia. The moment-by-moment dynamics of crossing is remarkably similar to that reported in mouse models of anxiety. This means that neophobic behavior is either homologous across arthropods and vertebrates or, not less interesting, convergent, whereby the same behavior is mediated in the two phyla by two completely different schemata.     

Improved Strains and Plasmid Vectors for Conditional Overexpression of His-Tagged Proteins in Haloferax volcanii

Research into archaea will not achieve its full potential until systems are in place to carry out genetics and biochemistry in the same species. Haloferax volcanii is widely regarded as the best-equipped organism for archaeal genetics, but the development of tools for the expression and purification of H. volcanii proteins has been neglected. We have developed a series of plasmid vectors and host strains for conditional overexpression of halophilic proteins in H. volcanii. The plasmids feature the tryptophan-inducible p.tnaA promoter and a 6×His tag for protein purification by metal affinity chromatography. Purification is facilitated by host strains, where pitA is replaced by the ortholog from Natronomonas pharaonis. The latter lacks the histidine-rich linker region found in H. volcanii PitA and does not copurify with His-tagged recombinant proteins. We also deleted the mrr restriction endonuclease gene, thereby allowing direct transformation without the need to passage DNA through an Escherichia coli dam mutant.Over the past century, our understanding of fundamental biological processes has grown exponentially, and this would have been impossible without the use of organisms that are amenable to experimental manipulation. Model species, such as Escherichia coli, Saccharomyces cerevisiae, Caenorhabditis elegans, Drosophila melanogaster, Mus musculus, and Arabidopsis thaliana, have become a byword for scientific progress (15). The rational choice of a model organism is critically important, and certain features are taken for granted, such as ease of cultivation, a short generation time, and systems for genetic manipulation. This list has now grown to include a genome sequence and methods for biochemical analysis of purified proteins in vitro.Research into archaea has lagged behind work on bacteria and eukaryotes but has nonetheless yielded profound insights (2). One hurdle has been the paucity of archaeal organisms suitable for both biochemistry and genetics. For example, Methanothermobacter thermautotrophicus is a stalwart of archaeal biochemistry but has proved resistant to even the most rudimentary genetic manipulation (2). Progress has recently been made with another biochemical workhorse, Sulfolobus spp., and a few genetic tools are now available (6, 13, 37). Methanosarcina spp. and Thermococcus kodakaraensis offer alternative systems with an increasing array of techniques (16, 35, 36), but sophisticated genetics has traditionally been the preserve of haloarchaea, of which Haloferax volcanii is the organism of choice (39). It is easy to culture, the genome has been sequenced (19), and there are several selectable markers and plasmids for transformation and gene knockout (3, 7, 31), including a Gateway system (14), as well as reporter genes (20, 33) and a tightly controlled inducible promoter (26).The genetic prowess of H. volcanii is not yet fully matched by corresponding systems for protein overexpression and purification. Like other haloarchaea, H. volcanii grows in high salt concentrations (2 to 5 M NaCl), and to cope with the osmotic potential of such environments, it accumulates high intracellular concentrations of potassium ions (12). Consequently, halophilic proteins are adapted to function at high salt concentrations and commonly feature a large excess of acidic amino acids; the negative surface charge is thought to be critical to solubility (28). This can pose problems for expression in heterologous hosts, such as E. coli, since halophilic proteins can misfold and aggregate under conditions of low ionic strength. The purification of misfolded halophilic enzymes from E. coli has relied on the recovery of insoluble protein from inclusion bodies, followed by denaturation and refolding in hypersaline solutions (8, 11). This approach is feasible only where the protein is well characterized and reconstitution of the active form can be monitored (for example, by an enzymatic assay). Furthermore, archaeal proteins expressed in heterologous bacterial hosts lack posttranslational modifications, such as acetylation or ubiquitination (4, 22), which are critical to understanding their biological function.Systems for expression of halophilic proteins in a native haloarchaeal host are therefore required. A number of studies have successfully purified recombinant proteins with a variety of affinity tags after overexpression in H. volcanii. For example, Humbard et al. employed tandem affinity tagging to purify 20S proteasomal core particles from the native host (23). However, the protein expression constructs used in these studies were custom made and somewhat tailored to the application in question. We report here the development of “generic” plasmid vectors and host strains for conditional overexpression of halophilic proteins in H. volcanii. The plasmids feature a tryptophan-inducible promoter derived from the tnaA gene of H. volcanii (26). We demonstrate the utility of these vectors by overexpressing a hexahistidine-tagged recombinant version of the H. volcanii RadA protein. Purification was greatly facilitated by a host strain in which the endogenous pitA gene was replaced by an ortholog from Natronomonas pharaonis. The latter protein lacks the histidine-rich linker region found in H. volcanii PitA (5) and therefore does not copurify with His-tagged recombinant proteins. Finally, we deleted the mrr gene of H. volcanii, which encodes a restriction enzyme that cleaves foreign DNA methylated at GATC residues. The mrr deletion strain allows direct transformation of H. volcanii without the need to passage plasmid DNA through an E. coli dam mutant (21).     

In vivo acetylation of CheY, a response regulator in chemotaxis of Escherichia coli

CheY, the excitatory response regulator in the chemotaxis system of Escherichia coli, can be modulated by two covalent modifications: phosphorylation and acetylation. Both modifications have been detected in vitro only. The role of CheY acetylation is still obscure, although it is known to be involved in chemotaxis and to occur in vitro by two mechanisms—acetyl-CoA synthetase-catalyzed transfer of acetyl groups from acetate to CheY and autocatalyzed transfer from AcCoA. Here, we succeeded in detecting CheY acetylation in vivo by three means—Western blotting with a specific anti-acetyl-lysine antibody, mass spectrometry, and radiolabeling with [¹⁴C]acetate in the presence of protein-synthesis inhibitor. Unexpectedly, the level and rate of CheY acetylation in vivo were much higher than that in vitro. Thus, before any treatment, 9-13% of the lysine residues were found acetylated, depending on the growth phase, meaning that, on average, essentially every CheY molecule was acetylated in vivo. This high level was mainly the outcome of autoacetylation. Addition of acetate caused an incremental increase in the acetylation level, in which acetyl-CoA synthetase was involved too. These findings may have far-reaching implications for the structure-function relationship of CheY.     

Particle-associated and free-living bacterial communities in an oligotrophic sea are affected by different environmental factors

In the oceans and seas, environmental conditions change over multiple temporal and spatial scales. Here, we ask what factors affect the bacterial community structure across time, depth and size fraction during six seasonal cruises (2 years) in the ultra-oligotrophic Eastern Mediterranean Sea. The bacterial community varied most between size fractions (free-living (FL) vs. particle-associated), followed by depth and finally season. The FL community was taxonomically richer and more stable than the particle-associated (PA) one, which was characterized by recurrent ‘blooms’ of heterotrophic bacteria such as Alteromonas and Ralstonia. The heterotrophic FL and PA communities were also correlated with different environmental parameters: the FL population correlated with depth and phytoplankton, whereas PA bacteria were correlated primarily with the time of sampling. A significant part of the variability in community structure could, however, not be explained by the measured parameters. The metabolic potential of the PA community, predicted from 16S rRNA amplicon data using PICRUSt, was enriched in pathways associated with the degradation and utilization of biological macromolecules, as well as plastics, other petroleum products and herbicides. The FL community was enriched in predicted pathways for the metabolism of inositol phosphate, a potential phosphorus source, and of polycyclic aromatic hydrocarbons.     

Divergence and Selectivity of Expression-Coupled Histone Modifications in Budding Yeasts

Various histone modifications are widely associated with gene expression, but their functional selectivity at individual genes remains to be characterized. Here, we identify widespread differences between genome-wide patterns of two prominent marks, H3K9ac and H3K4me3, in budding yeasts. As well as characteristic gene profiles, relative modification levels vary significantly amongst genes, irrespective of expression. Interestingly, we show that these differences couple to contrasting features: higher methylation to essential, periodically expressed, ‘DPN’ (Depleted Proximal Nucleosome) genes, and higher acetylation to non-essential, responsive, ‘OPN’ (Occupied Proximal Nucleosome) genes. Thus, H3K4me3 may generally associate with expression stability, and H3K9ac, with variability. To evaluate this notion, we examine their association with expression divergence between the closely related species, S. cerevisiae and S. paradoxus. Although individually well conserved at orthologous genes, changes between modifications are mostly uncorrelated, indicating largely non-overlapping regulatory mechanisms. Notably, we find that inter-species differences in methylation, but not acetylation, are well correlated with expression changes, thereby proposing H3K4me3 as a candidate regulator of expression divergence. Taken together, our results suggest distinct evolutionary roles for expression-linked modifications, wherein H3K4me3 may contribute to stabilize average expression, whilst H3K9ac associates with more indirect aspects such as responsiveness.