MOB rules: Antibiotic Exposure Reprograms Metabolism to Mobilize Bacillus subtilis in Competitive Interactions

Antibiotics have dose-dependent effects on exposed bacteria. The medicinal use of antibiotics relies on their growth-inhibitory activities at sufficient concentrations. At subinhibitory concentrations, exposure effects vary widely among different antibiotics and bacteria. Bacillus subtilis responds to bacteriostatic translation inhibitors by mobilizing a population of cells (MOB-Mobilized Bacillus) to spread across a surface. How B. subtilis regulates the antibiotic-induced mobilization is not known. In this study, we used chloramphenicol to identify regulatory functions that B. subtilis requires to coordinate cell mobilization following subinhibitory exposure. We measured changes in gene expression and metabolism and mapped the results to a network of regulatory proteins that direct the mobile response. Our data reveal that several transcriptional regulators coordinately control the reprogramming of metabolism to support mobilization. The network regulates changes in glycolysis, nucleotide metabolism, and amino acid metabolism that are signature features of the mobilized population. Among the hundreds of genes with changing expression, we identified two, pdhA and pucA, where the magnitudes of their changes in expression, and in the abundance of associated metabolites, reveal hallmark metabolic features of the mobilized population. Using reporters of pdhA and pucA expression, we visualized the separation of major branches of metabolism in different regions of the mobilized population. Our results reveal a regulated response to chloramphenicol exposure that enables a population of bacteria in different metabolic states to mount a coordinated mobile response.

Mitochondrial Aconitase ACO2 Links Iron Homeostasis with Tumorigenicity in Non-Small Cell Lung Cancer

The ability of a patient tumor to engraft an immunodeficient mouse is the strongest known independent indicator of poor prognosis in early-stage non-small cell lung cancer (NSCLC). Analysis of primary NSCLC proteomes revealed low-level expression of mitochondrial aconitase (ACO2) in the more aggressive, engrafting tumors. Knockdown of ACO2 protein expression transformed immortalized lung epithelial cells, whereas upregulation of ACO2 in transformed NSCLC cells inhibited cell proliferation in vitro and tumor growth in vivo. High level ACO2 increased iron response element binding protein 1 (IRP1) and the intracellular labile iron pool. Impaired cellular proliferation associated with high level ACO2 was reversed by treatment of cells with an iron chelator, whereas increased cell proliferation associated with low level ACO2 was suppressed by treatment of cells with iron. Expression of CDGSH iron-sulfur (FeS) domain-containing protein 1 [CISD1; also known as mitoNEET (mNT)] was modulated by ACO2 expression level and inhibition of mNT by RNA interference or by treatment of cells with pioglitazone also increased iron and cell death. Hence, ACO2 is identified as a regulator of iron homeostasis and mNT is implicated as a target in aggressive NSCLC.

IMPLICATIONS

FeS cluster-associated proteins including ACO2, mNT (encoded by CISD1), and IRP1 (encoded by ACO1) are part of an "ACO2-Iron Axis" that regulates iron homeostasis and is a determinant of a particularly aggressive subset of NSCLC.

Rhodoquinone biosynthesis in C. elegans requires precursors generated by the kynurenine pathway

Parasitic helminths infect over a billion humans. To survive in the low oxygen environment of their hosts, these parasites use unusual anaerobic metabolism - this requires rhodoquinone (RQ), an electron carrier that is made by very few animal species. Crucially RQ is not made or used by any parasitic hosts and RQ synthesis is thus an ideal target for anthelmintics. However, little is known about how RQ is made and no drugs are known to block RQ synthesis. C. elegans makes RQ and can use RQ-dependent metabolic pathways - here, we use C. elegans genetics to show that tryptophan degradation via the kynurenine pathway is required to generate the key amine-containing precursors for RQ synthesis. We show that C. elegans requires RQ for survival in hypoxic conditions and, finally, we establish a high throughput assay for drugs that block RQ-dependent metabolism. This may drive the development of a new class of anthelmintic drugs. This study is a key first step in understanding how RQ is made in parasitic helminths.

Functional genetic discovery of enzymes using full-scan mass spectrometry metabolomics

Our understanding of metabolic networks is incomplete, and new enzymatic activities await discovery in well-studied organisms. Mass spectrometric measurement of cellular metabolites reveals compounds inside cells that are unexplained by current maps of metabolic reactions, and existing computational models are unable to account for all activities observed within cells. Additional large-scale genetic and biochemical approaches are required to elucidate metabolic gene function. We have used full-scan mass spectrometry metabolomics of polar small molecules to examine deletion mutants of candidate enzymes in the model yeast Saccharomyces cerevisiae. We report the identification of 25 genes whose deletion results in focal metabolic changes consistent with loss of enzymatic activity and describe the informatic approaches used to enrich for candidate enzymes from uncharacterized open reading frames. Triumphs and pitfalls of metabolic phenotyping screens are discussed, including estimates of the frequency of uncharacterized eukaryotic genes that affect metabolism and key issues to consider when searching for new enzymatic functions in other organisms.

Transposon insertional mutagenesis in Saccharomyces uvarum reveals trans-acting effects influencing species-dependent essential genes

To understand how complex genetic networks perform and regulate diverse cellular processes, the function of each individual component must be defined. Comprehensive phenotypic studies of mutant alleles have been successful in model organisms in determining what processes depend on the normal function of a gene. These results are often ported to newly sequenced genomes by using sequence homology. However, sequence similarity does not always mean identical function or phenotype, suggesting that new methods are required to functionally annotate newly sequenced species. We have implemented comparative analysis by high-throughput experimental testing of gene dispensability in Saccharomyces uvarum, a sister species of Saccharomyces cerevisiae. We created haploid and heterozygous diploid Tn7 insertional mutagenesis libraries in S. uvarum to identify species-dependent essential genes, with the goal of detecting genes with divergent functions and/or different genetic interactions. Comprehensive gene dispensability comparisons with S. cerevisiae predicted diverged dispensability at 12% of conserved orthologs, and validation experiments confirmed 22 differentially essential genes. Despite their differences in essentiality, these genes were capable of cross-species complementation, demonstrating that trans-acting factors that are background-dependent contribute to differential gene essentiality. This study shows that direct experimental testing of gene disruption phenotypes across species can inform comparative genomic analyses and improve gene annotations. Our method can be widely applied in microorganisms to further our understanding of genome evolution.

Metabolite Extraction from Saccharomyces cerevisiae for Liquid Chromatography-Mass Spectrometry

Prior to mass spectrometric analysis, cellular small molecules must be extracted and separated from interfering components such as salts and culture medium. To ensure minimal perturbation of metabolism, yeast cells grown in liquid culture are rapidly harvested by filtration as described here. Simultaneous quenching of metabolism and extraction is afforded by immediate immersion in low-temperature organic solvent. Samples prepared using this method are suitable for a range of downstream liquid chromatography-mass spectrometry analyses and are stable in solvent for >1 yr at -80°C.

Metabolomics in Yeast

Budding yeast has from the beginning been a major eukaryotic model for the study of metabolic network structure and function. This is attributable to both its genetic and biochemical capacities and its role as a workhorse in food production and biotechnology. New inventions in analytical technologies allow accurate, simultaneous detection and quantification of metabolites, and a series of recent findings have placed the metabolic network at center stage in the physiology of the cell. For example, metabolism might have facilitated the origin of life, and in modern organisms it not only provides nutrients to the cell but also serves as a buffer to changes in the cellular environment, a regulator of cellular processes, and a requirement for cell growth. These findings have triggered a rapid and massive renaissance in this important field. Here, we provide an introduction to analysis of metabolomics in yeast.

Spectrophotometric Analysis of Ethanol and Glucose Concentrations in Yeast Culture Media

Fermentative growth on glucose is one of the most widely studied conditions of yeast growth in the laboratory. The production of ethanol from sugars is relevant to the wine, beer, and bread industries and to production of biofuels. Assaying the levels of glucose and ethanol in yeast growth medium allows the experimenter to determine the consumption of the carbon source glucose and the production of ethanol. This protocol describes enzyme-coupled assays for determination of glucose and ethanol concentrations in a sample of cell-free culture medium. Enzymes convert glucose or ethanol into other compounds through chemical reactions that reduce NAD(P)⁺ to NAD(P)H, and the production of NAD(P)H is measured using a spectrophotometer. The methods presented are highly sensitive, with a detection limit of ∼0.4 mg/L of glucose and 50 mg/L of ethanol, and also have the advantage of high specificity. For example, glucose and fructose have identical chemical formulas and thus cannot be distinguished by a mass spectrometer, but the enzyme assay presented here is specific for glucose. The glucose assay can be coupled to other assays to determine the quantity of additional carbohydrates such as fructose, trehalose, and glycogen.

A global genetic interaction network maps a wiring diagram of cellular function

We generated a global genetic interaction network for Saccharomyces cerevisiae, constructing more than 23 million double mutants, identifying about 550,000 negative and about 350,000 positive genetic interactions. This comprehensive network maps genetic interactions for essential gene pairs, highlighting essential genes as densely connected hubs. Genetic interaction profiles enabled assembly of a hierarchical model of cell function, including modules corresponding to protein complexes and pathways, biological processes, and cellular compartments. Negative interactions connected functionally related genes, mapped core bioprocesses, and identified pleiotropic genes, whereas positive interactions often mapped general regulatory connections among gene pairs, rather than shared functionality. The global network illustrates how coherent sets of genetic interactions connect protein complex and pathway modules to map a functional wiring diagram of the cell.

Proteomic analysis of the human KEOPS complex identifies C14ORF142 as a core subunit homologous to yeast Gon7

The KEOPS/EKC complex is a tRNA modification complex involved in the biosynthesis of N⁶-threonylcarbamoyladenosine (t⁶A), a universally conserved tRNA modification found on ANN-codon recognizing tRNAs. In archaea and eukaryotes, KEOPS is composed of OSGEP/Kae1, PRPK/Bud32, TPRKB/Cgi121 and LAGE3/Pcc1. In fungi, KEOPS contains an additional subunit, Gon7, whose orthologs outside of fungi, if existent, remain unidentified. In addition to displaying defective t⁶A biosynthesis, Saccharomyces cerevisiae strains harboring KEOPS mutations are compromised for telomere homeostasis, growth and transcriptional co-activation. To identify a Gon7 ortholog in multicellular eukaryotes as well as to uncover KEOPS-interacting proteins that may link t⁶A biosynthesis to the diverse set of KEOPS mutant phenotypes, we conducted a proteomic analysis of human KEOPS. This work identified 152 protein interactors, one of which, C14ORF142, interacted strongly with all four KEOPS subunits, suggesting that it may be a core component of human KEOPS. Further characterization of C14ORF142 revealed that it shared a number of biophysical and biochemical features with fungal Gon7, suggesting that C14ORF142 is the human ortholog of Gon7. In addition, our proteomic analysis identified specific interactors for different KEOPS subcomplexes, hinting that individual KEOPS subunits may have additional functions outside of t⁶A biosynthesis.

Targeted metabolomics in cultured cells and tissues by mass spectrometry: method development and validation

Metabolomics is the identification and quantitation of small bio-molecules (metabolites) in biological samples under various environmental and genetic conditions. Mass spectrometry provides the unique opportunity for targeted identification and quantification of known metabolites by selective reaction monitoring (SRM). However, reproducibility of this approach depends on careful consideration of sample preparation, chemical classes, and stability of metabolites to be evaluated. Herein, we introduce and validate a targeted metabolite profiling workflow for cultured cells and tissues by liquid chromatography-triple quadrupole tandem mass spectrometry. The method requires a one-step extraction of water-soluble metabolites and targeted analysis of central metabolites that include glycolysis, amino acids, nucleotides, citric acid cycle, and the hexosamine biosynthetic pathway. The sensitivity, reproducibility and molecular stability of each targeted metabolite were assessed under experimental conditions. Quantitation of metabolites by peak area ratio was linear with a dilution over a 4 fold dynamic range with minimal deviation R(2)=0.98. Inter- and intra-day precision with cells and tissues had an average coefficient of variation <15% for cultured cell lines, and somewhat higher for mouse liver tissues. The method applied in triplicate measurements readily distinguished immortalized cells from malignant cells, as well as mouse littermates based on their hepatic metabolic profiles.

Uniform nomenclature for the mitochondrial contact site and cristae organizing system

The mitochondrial inner membrane contains a large protein complex that functions in inner membrane organization and formation of membrane contact sites. The complex was variably named the mitochondrial contact site complex, mitochondrial inner membrane organizing system, mitochondrial organizing structure, or Mitofilin/Fcj1 complex. To facilitate future studies, we propose to unify the nomenclature and term the complex “mitochondrial contact site and cristae organizing system” and its subunits Mic10 to Mic60.

Heritability and genetic basis of protein level variation in an outbred population

The genetic basis of heritable traits has been studied for decades. Although recent mapping efforts have elucidated genetic determinants of transcript levels, mapping of protein abundance has lagged. Here, we analyze levels of 4084 GFP-tagged yeast proteins in the progeny of a cross between a laboratory and a wild strain using flow cytometry and high-content microscopy. The genotype of trans variants contributed little to protein level variation between individual cells but explained >50% of the variance in the population’s average protein abundance for half of the GFP fusions tested. To map trans-acting factors responsible, we performed flow sorting and bulk segregant analysis of 25 proteins, finding a median of five protein quantitative trait loci (pQTLs) per GFP fusion. Further, we find that cis-acting variants predominate; the genotype of a gene and its surrounding region had a large effect on protein level six times more frequently than the rest of the genome combined. We present evidence for both shared and independent genetic control of transcript and protein abundance: More than half of the expression QTLs (eQTLs) contribute to changes in protein levels of regulated genes, but several pQTLs do not affect their cognate transcript levels. Allele replacements of genes known to underlie trans eQTL hotspots confirmed the correlation of effects on mRNA and protein levels. This study represents the first genome-scale measurement of genetic contribution to protein levels in single cells and populations, identifies more than a hundred trans pQTLs, and validates the propagation of effects associated with transcript variation to protein abundance.

Interspecies systems biology uncovers metabolites affecting C. elegans gene expression and life history traits

Diet greatly influences gene expression and physiology. In mammals, elucidating the effects and mechanisms of individual nutrients is challenging due to the complexity of both the animal and its diet. Here, we used an interspecies systems biology approach with Caenorhabditis elegans and two of its bacterial diets, Escherichia coli and Comamonas aquatica, to identify metabolites that affect the animal's gene expression and physiology. We identify vitamin B12 as the major dilutable metabolite provided by Comamonas aq. that regulates gene expression, accelerates development, and reduces fertility but does not affect lifespan. We find that vitamin B12 has a dual role in the animal: it affects development and fertility via the methionine/S-Adenosylmethionine (SAM) cycle and breaks down the short-chain fatty acid propionic acid, preventing its toxic buildup. Our interspecies systems biology approach provides a paradigm for understanding complex interactions between diet and physiology.

Broad metabolic sensitivity profiling of a prototrophic yeast deletion collection

Background

Genome-wide sensitivity screens in yeast have been immensely popular following the construction of a collection of deletion mutants of non-essential genes. However, the auxotrophic markers in this collection preclude experiments on minimal growth medium, one of the most informative metabolic environments. Here we present quantitative growth analysis for mutants in all 4,772 non-essential genes from our prototrophic deletion collection across a large set of metabolic conditions.

Results

The complete collection was grown in environments consisting of one of four possible carbon sources paired with one of seven nitrogen sources, for a total of 28 different well-defined metabolic environments. The relative contributions to mutants' fitness of each carbon and nitrogen source were determined using multivariate statistical methods. The mutant profiling recovered known and novel genes specific to the processing of nutrients and accurately predicted functional relationships, especially for metabolic functions. A benchmark of genome-scale metabolic network modeling is also given to demonstrate the level of agreement between current in silico predictions and hitherto unavailable experimental data.

Conclusions

These data address a fundamental deficiency in our understanding of the model eukaryote Saccharomyces cerevisiae and its response to the most basic of environments. While choice of carbon source has the greatest impact on cell growth, specific effects due to nitrogen source and interactions between the nutrients are frequent. We demonstrate utility in characterizing genes of unknown function and illustrate how these data can be integrated with other whole-genome screens to interpret similarities between seemingly diverse perturbation types.

Biochemical and structural studies of conserved Maf proteins revealed nucleotide pyrophosphatases with a preference for modified nucleotides

Maf (for multicopy associated filamentation) proteins represent a large family of conserved proteins implicated in cell division arrest but whose biochemical activity remains unknown. Here, we show that the prokaryotic and eukaryotic Maf proteins exhibit nucleotide pyrophosphatase activity against 5-methyl-UTP, pseudo-UTP, 5-methyl-CTP, and 7-methyl-GTP, which represent the most abundant modified bases in all organisms, as well as against canonical nucleotides dTTP, UTP, and CTP. Overexpression of the Maf protein YhdE in E. coli cells increased intracellular levels of dTMP and UMP, confirming that dTTP and UTP are the in vivo substrates of this protein. Crystal structures and site-directed mutagenesis of Maf proteins revealed the determinants of their activity and substrate specificity. Thus, pyrophosphatase activity of Maf proteins toward canonical and modified nucleotides might provide the molecular mechanism for a dual role of these proteins in cell division arrest and house cleaning.

•

Maf proteins represent a family of nucleoside triphosphate pyrophosphatases

•

Maf proteins hydrolyze the canonical nucleotides dTTP, UTP, and CTP

•

Maf proteins are also active against m⁵UTP, m⁵CTP, pseudo-UTP, and m⁷GTP

•

Maf structures reveal the molecular mechanisms of their substrate selectivity

Nucleotide degradation and ribose salvage in yeast

Nucleotide degradation is a universal metabolic capability. Here we combine metabolomics, genetics and biochemistry to characterize the yeast pathway. Nutrient starvation, via PKA, AMPK/SNF1, and TOR, triggers autophagic breakdown of ribosomes into nucleotides. A protein not previously associated with nucleotide degradation, Phm8, converts nucleotide monophosphates into nucleosides. Downstream steps, which involve the purine nucleoside phosphorylase, Pnp1, and pyrimidine nucleoside hydrolase, Urh1, funnel ribose into the nonoxidative pentose phosphate pathway. During carbon starvation, the ribose-derived carbon accumulates as sedoheptulose-7-phosphate, whose consumption by transaldolase is impaired due to depletion of transaldolase's other substrate, glyceraldehyde-3-phosphate. Oxidative stress increases glyceraldehyde-3-phosphate, resulting in rapid consumption of sedoheptulose-7-phosphate to make NADPH for antioxidant defense. Ablation of Phm8 or double deletion of Pnp1 and Urh1 prevent effective nucleotide salvage, resulting in metabolite depletion and impaired survival of starving yeast. Thus, ribose salvage provides means of surviving nutrient starvation and oxidative stress.

Reconstitution and characterization of eukaryotic N6-threonylcarbamoylation of tRNA using a minimal enzyme system

The universally conserved Kae1/Qri7/YgjD and Sua5/YrdC protein families have been implicated in growth, telomere homeostasis, transcription and the N6-threonylcarbamoylation (t⁶A) of tRNA, an essential modification required for translational fidelity by the ribosome. In bacteria, YgjD orthologues operate in concert with the bacterial-specific proteins YeaZ and YjeE, whereas in archaeal and eukaryotic systems, Kae1 operates as part of a larger macromolecular assembly called KEOPS with Bud32, Cgi121, Gon7 and Pcc1 subunits. Qri7 orthologues function in the mitochondria and may represent the most primitive member of the Kae1/Qri7/YgjD protein family. In accordance with previous findings, we confirm that Qri7 complements Kae1 function and uncover that Qri7 complements the function of all KEOPS subunits in growth, t⁶A biosynthesis and, to a partial degree, telomere maintenance. These observations suggest that Kae1 provides a core essential function that other subunits within KEOPS have evolved to support. Consistent with this inference, Qri7 alone is sufficient for t⁶A biosynthesis with Sua5 in vitro. In addition, the 2.9 Å crystal structure of Qri7 reveals a simple homodimer arrangement that is supplanted by the heterodimerization of YgjD with YeaZ in bacteria and heterodimerization of Kae1 with Pcc1 in KEOPS. The partial complementation of telomere maintenance by Qri7 hints that KEOPS has evolved novel functions in higher organisms.

The future of deciphering personal genomes? The flies (and yeast and worms) still have it

A report on the 13th International Conference on Systems Biology, held in Toronto, Ontario, Canada, 19-23 August 2012.

Quantitative measurement of allele-specific protein expression in a diploid yeast hybrid by LC-MS

Understanding the genetic basis of gene regulatory variation is a key goal of evolutionary and medical genetics. Regulatory variation can act in an allele-specific manner (cis-acting) or it can affect both alleles of a gene (trans-acting). Differential allele-specific expression (ASE), in which the expression of one allele differs from another in a diploid, implies the presence of cis-acting regulatory variation. While microarrays and high-throughput sequencing have enabled genome-wide measurements of transcriptional ASE, methods for measurement of protein ASE (pASE) have lagged far behind. We describe a flexible, accurate, and scalable strategy for measurement of pASE by liquid chromatography-coupled mass spectrometry (LC-MS). We apply this approach to a hybrid between the yeast species Saccharomyces cerevisiae and Saccharomyces bayanus. Our results provide the first analysis of the relative contribution of cis-acting and trans-acting regulatory differences to protein expression divergence between yeast species.

Accurate proteome-wide protein quantification from high-resolution 15N mass spectra

In quantitative mass spectrometry-based proteomics, the metabolic incorporation of a single source of ¹⁵N-labeled nitrogen has many advantages over using stable isotope-labeled amino acids. However, the lack of a robust computational framework for analyzing the resulting spectra has impeded wide use of this approach. We have addressed this challenge by introducing a new computational methodology for analyzing ¹⁵N spectra in which quantification is integrated with identification. Application of this method to an Escherichia coli growth transition reveals significant improvement in quantification accuracy over previous methods.

Design of custom oligonucleotide microarrays for single species or interspecies hybrids using Array Oligo Selector

New technologies for DNA sequencing have made it feasible to determine the genome sequence of any organism of interest. This sequence is the resource required to create tools for downstream studies, including DNA microarrays. A number of vendors can produce DNA microarrays containing customer-specified sequences, allowing investigators to design and order arrays customized for any species of interest. Freely available, user-friendly computer programs are available for designing microarray probes. These design programs can be used to create probes that distinguish between two related genomes, allowing investigation of gene expression or gene representation in intra- or interspecies hybrids or in samples containing DNA from multiple species.

Dissection of genetically complex traits with extremely large pools of yeast segregants

Most heritable traits, including many human diseases 1, are caused by multiple loci. Studies in both humans and model organisms, such as yeast, have failed to detect a large fraction of the loci that underlie such complex traits 2,3. A lack of statistical power to identify multiple loci with small effects is undoubtedly one of the primary reasons for this problem. We have developed a method in yeast that allows the use of dramatically larger sample sizes than previously possible and hence permits the detection of multiple loci with small effects. The method involves generating very large numbers of progeny from a cross between two strains and then phenotyping and genotyping pools of these offspring. We applied the method to 17 chemical resistance traits and mitochondrial function, and identified loci for each of these phenotypes. We show that the range of genetic complexity underlying these quantitative traits is highly variable, with some traits influenced by one major locus and others due to at least 20 loci. Our results provide an empirical demonstration of the genetic complexity of many traits and show that it is possible to identify many of the underlying factors using straightforward techniques. Our method should have broad applications in yeast and can be extended to other organisms.

Metabolomic analysis via reversed-phase ion-pairing liquid chromatography coupled to a stand alone orbitrap mass spectrometer

We present a liquid chromatography-mass spectrometry (LC-MS) method that capitalizes on the mass-resolving power of the orbitrap to enable sensitive and specific measurement of known and unanticipated metabolites in parallel, with a focus on water-soluble species involved in core metabolism. The reversed phase LC method, with a cycle time 25 min, involves a water-methanol gradient on a C18 column with tributylamine as the ion pairing agent. The MS portion involves full scans from 85 to 1000 m/z at 1 Hz and 100,000 resolution in negative ion mode on a stand alone orbitrap ("Exactive"). The median limit of detection, across 80 metabolite standards, was 5 ng/mL with the linear range typically >or=100-fold. For both standards and a cellular extract from Saccharomyces cerevisiae (Baker's yeast), the median inter-run relative standard deviation in peak intensity was 8%. In yeast exact, we detected 137 known compounds, whose (13)C-labeling patterns could also be tracked to probe metabolic flux. In yeast engineered to lack a gene of unknown function (YKL215C), we observed accumulation of an ion of m/z 128.0351, which we subsequently confirmed to be oxoproline, resulting in annotation of YKL215C as an oxoprolinase. These examples demonstrate the suitability of the present method for quantitative metabolomics, fluxomics, and discovery metabolite profiling.

The impact of incomplete knowledge on evaluation: an experimental benchmark for protein function prediction

MOTIVATION

Rapidly expanding repositories of highly informative genomic data have generated increasing interest in methods for protein function prediction and inference of biological networks. The successful application of supervised machine learning to these tasks requires a gold standard for protein function: a trusted set of correct examples, which can be used to assess performance through cross-validation or other statistical approaches. Since gene annotation is incomplete for even the best studied model organisms, the biological reliability of such evaluations may be called into question.

RESULTS

We address this concern by constructing and analyzing an experimentally based gold standard through comprehensive validation of protein function predictions for mitochondrion biogenesis in Saccharomyces cerevisiae. Specifically, we determine that (i) current machine learning approaches are able to generalize and predict novel biology from an incomplete gold standard and (ii) incomplete functional annotations adversely affect the evaluation of machine learning performance. While computational approaches performed better than predicted in the face of incomplete data, relative comparison of competing approaches-even those employing the same training data-is problematic with a sparse gold standard. Incomplete knowledge causes individual methods' performances to be differentially underestimated, resulting in misleading performance evaluations. We provide a benchmark gold standard for yeast mitochondria to complement current databases and an analysis of our experimental results in the hopes of mitigating these effects in future comparative evaluations.

AVAILABILITY

The mitochondrial benchmark gold standard, as well as experimental results and additional data, is available at http://function.princeton.edu/mitochondria.

Measuring differential gene expression by short read sequencing: quantitative comparison to 2-channel gene expression microarrays

Background

High-throughput cDNA synthesis and sequencing of poly(A)-enriched RNA is rapidly emerging as a technology competing to replace microarrays as a quantitative platform for measuring gene expression.

Results

Consequently, we compared full length cDNA sequencing to 2-channel gene expression microarrays in the context of measuring differential gene expression. Because of its comparable cost to a gene expression microarray, our study focused on the data obtainable from a single lane of an Illumina 1 G sequencer. We compared sequencing data to a highly replicated microarray experiment profiling two divergent strains of S. cerevisiae.

Conclusion

Using a large number of quantitative PCR (qPCR) assays, more than previous studies, we found that neither technology is decisively better at measuring differential gene expression. Further, we report sequencing results from a diploid hybrid of two strains of S. cerevisiae that indicate full length cDNA sequencing can discover heterozygosity and measure quantitative allele-specific expression simultaneously.

Computationally driven, quantitative experiments discover genes required for mitochondrial biogenesis

Mitochondria are central to many cellular processes including respiration, ion homeostasis, and apoptosis. Using computational predictions combined with traditional quantitative experiments, we have identified 100 proteins whose deficiency alters mitochondrial biogenesis and inheritance in Saccharomyces cerevisiae. In addition, we used computational predictions to perform targeted double-mutant analysis detecting another nine genes with synthetic defects in mitochondrial biogenesis. This represents an increase of about 25% over previously known participants. Nearly half of these newly characterized proteins are conserved in mammals, including several orthologs known to be involved in human disease. Mutations in many of these genes demonstrate statistically significant mitochondrial transmission phenotypes more subtle than could be detected by traditional genetic screens or high-throughput techniques, and 47 have not been previously localized to mitochondria. We further characterized a subset of these genes using growth profiling and dual immunofluorescence, which identified genes specifically required for aerobic respiration and an uncharacterized cytoplasmic protein required for normal mitochondrial motility. Our results demonstrate that by leveraging computational analysis to direct quantitative experimental assays, we have characterized mutants with subtle mitochondrial defects whose phenotypes were undetected by high-throughput methods.

Mitochondria are the proverbial powerhouses of the cell, running the fundamental biochemical processes that produce energy from nutrients using oxygen. These processes are conserved in all eukaryotes, from humans to model organisms such as baker's yeast. In humans, mitochondrial dysfunction plays a role in a variety of diseases, including diabetes, neuromuscular disorders, and aging. In order to better understand fundamental mitochondrial biology, we studied genes involved in mitochondrial biogenesis in the yeast S. cerevisiae, discovering over 100 proteins with novel roles in this process. These experiments assigned function to 5% of the genes whose function was not known. In order to achieve this rapid rate of discovery, we developed a system incorporating highly quantitative experimental assays and an integrated, iterative process of computational protein function prediction. Beginning from relatively little prior knowledge, we found that computational predictions achieved about 60% accuracy and rapidly guided our laboratory work towards hundreds of promising candidate genes. Thus, in addition to providing a more thorough understanding of mitochondrial biology, this study establishes a framework for successfully integrating computation and experimentation to drive biological discovery. A companion manuscript, published in PLoS Computational Biology (doi:10.1371/journal.pcbi.1000322), discusses observations and conclusions important for the computational community.

Directing experimental biology: a case study in mitochondrial biogenesis

Computational approaches have promised to organize collections of functional genomics data into testable predictions of gene and protein involvement in biological processes and pathways. However, few such predictions have been experimentally validated on a large scale, leaving many bioinformatic methods unproven and underutilized in the biology community. Further, it remains unclear what biological concerns should be taken into account when using computational methods to drive real-world experimental efforts. To investigate these concerns and to establish the utility of computational predictions of gene function, we experimentally tested hundreds of predictions generated from an ensemble of three complementary methods for the process of mitochondrial organization and biogenesis in Saccharomyces cerevisiae. The biological data with respect to the mitochondria are presented in a companion manuscript published in PLoS Genetics (doi:10.1371/journal.pgen.1000407). Here we analyze and explore the results of this study that are broadly applicable for computationalists applying gene function prediction techniques, including a new experimental comparison with 48 genes representing the genomic background. Our study leads to several conclusions that are important to consider when driving laboratory investigations using computational prediction approaches. While most genes in yeast are already known to participate in at least one biological process, we confirm that genes with known functions can still be strong candidates for annotation of additional gene functions. We find that different analysis techniques and different underlying data can both greatly affect the types of functional predictions produced by computational methods. This diversity allows an ensemble of techniques to substantially broaden the biological scope and breadth of predictions. We also find that performing prediction and validation steps iteratively allows us to more completely characterize a biological area of interest. While this study focused on a specific functional area in yeast, many of these observations may be useful in the contexts of other processes and organisms.

Genome sequencing has provided us with “parts lists” of genes for many organisms, but many of the biological roles these genes are still unknown. While a great deal of functional genomic data exists, providing information about these genes and their roles, the rate at which these data are leveraged into concrete biological knowledge lags far behind the rate of data generation. Many computational approaches have been developed to generate accurate predictions of gene functions, with the goal of bridging this divide. However, as no large-scale experimental efforts have been based on such approaches, their validity and utility remains unproven. We have performed a study that experimentally evaluates predictions from a combination of three computational function prediction approaches, focusing on mitochondrion-related processes in brewer's yeast as a model system. By using computational predictions to guide our laboratory investigation, we have greatly accelerated the rate at which proteins can be assigned to biological processes. Further, our results demonstrate that in order to achieve the best results, it is important for computational biologists to consider both the underlying data and the algorithmic foundations of the methods used to predict function. Lastly, we demonstrate that iterating through phases of prediction and validation has quickly and extensively expanded our knowledge of mitochondrial biology.

Predicting cellular growth from gene expression signatures

Maintaining balanced growth in a changing environment is a fundamental systems-level challenge for cellular physiology, particularly in microorganisms. While the complete set of regulatory and functional pathways supporting growth and cellular proliferation are not yet known, portions of them are well understood. In particular, cellular proliferation is governed by mechanisms that are highly conserved from unicellular to multicellular organisms, and the disruption of these processes in metazoans is a major factor in the development of cancer. In this paper, we develop statistical methodology to identify quantitative aspects of the regulatory mechanisms underlying cellular proliferation in Saccharomyces cerevisiae. We find that the expression levels of a small set of genes can be exploited to predict the instantaneous growth rate of any cellular culture with high accuracy. The predictions obtained in this fashion are robust to changing biological conditions, experimental methods, and technological platforms. The proposed model is also effective in predicting growth rates for the related yeast Saccharomyces bayanus and the highly diverged yeast Schizosaccharomyces pombe, suggesting that the underlying regulatory signature is conserved across a wide range of unicellular evolution. We investigate the biological significance of the gene expression signature that the predictions are based upon from multiple perspectives: by perturbing the regulatory network through the Ras/PKA pathway, observing strong upregulation of growth rate even in the absence of appropriate nutrients, and discovering putative transcription factor binding sites, observing enrichment in growth-correlated genes. More broadly, the proposed methodology enables biological insights about growth at an instantaneous time scale, inaccessible by direct experimental methods. Data and tools enabling others to apply our methods are available at http://function.princeton.edu/growthrate.

Predicting gene function in a hierarchical context with an ensemble of classifiers

Background:

The wide availability of genome-scale data for several organisms has stimulated interest in computational approaches to gene function prediction. Diverse machine learning methods have been applied to unicellular organisms with some success, but few have been extensively tested on higher level, multicellular organisms. A recent mouse function prediction project (MouseFunc) brought together nine bioinformatics teams applying a diverse array of methodologies to mount the first large-scale effort to predict gene function in the laboratory mouse.

Results:

In this paper, we describe our contribution to this project, an ensemble framework based on the support vector machine that integrates diverse datasets in the context of the Gene Ontology hierarchy. We carry out a detailed analysis of the performance of our ensemble and provide insights into which methods work best under a variety of prediction scenarios. In addition, we applied our method to Saccharomyces cerevisiae and have experimentally confirmed functions for a novel mitochondrial protein.

Conclusion:

Our method consistently performs among the top methods in the MouseFunc evaluation. Furthermore, it exhibits good classification performance across a variety of cellular processes and functions in both a multicellular organism and a unicellular organism, indicating its ability to discover novel biology in diverse settings.

Element 1360 and RNAi components contribute to HP1-dependent silencing of a pericentric reporter

In eukaryotes, distinct regions of the genome are packaged as euchromatin (less condensed, more active) or heterochromatin (condensed, silenced). Studies in yeast, plants, and flies suggest that RNA interference (RNAi) is linked to heterochromatin formation and transcriptional silencing of transposable element (TE) sequences. We previously reported that insertion of a mobile hsp70-white reporter within 10 kb of a 1360 element on chromosome four of Drosophila melanogaster correlates with variegation (silencing). Here, we report small RNAs (approximately 23 nt) corresponding to 1360, indicating processing by the RNAi machinery. To directly test the ability of 1360 to silence a nearby gene in vivo, we introduced a P element construct carrying a single copy of 1360 upstream of the hsp70-white reporter into flies. This 1360 element contributes to HP1-dependent variegation at a pericentric insertion site, as demonstrated by a decrease in silencing after FLP-mediated removal of 1360. In euchromatin, 1360 is not sufficient to induce silencing, suggesting that proximity to pericentric heterochromatin and/or a high local TE density contributes to heterochromatin formation. Silencing of the 1360, hsp70-white reporter is sensitive to mutations in RNAi components. Our results implicate 1360 as a target for sequence-specific heterochromatic silencing through an RNAi-dependent mechanism.

CD25 deficiency causes an immune dysregulation, polyendocrinopathy, enteropathy, X-linked-like syndrome, and defective IL-10 expression from CD4 lymphocytes

BACKGROUND

Immune dysregulation, polyendocrinopathy, enteropathy, X-linked (IPEX) results in systemic autoimmunity from birth and can be caused by mutations in the transcription factor forkhead box P3 (FOXP3).

OBJECTIVE

To determine if Foxp3 is required for the generation of IL-10-expressing T regulatory cells.

METHODS

CD4 lymphocytes were isolated from patients with IPEX-like syndromes and activated with antibodies to CD3 and CD46 to generate IL-10-expressing T regulatory cells.

RESULTS

We describe a patient with clinical manifestations of IPEX that had a normal Foxp3 gene, but who had CD25 deficiency due to autosomal recessive mutations in this gene. This patient exhibited defective IL-10 expression from CD4 lymphocytes, whereas a Foxp3-deficient patient expressed normal levels of IL-10.

CONCLUSION

These data show that CD25 deficiency results in an IPEX-like syndrome and suggests that although Foxp3 is not required for normal IL-10 expression by human CD4 lymphocytes, CD25 expression is important.

CLINICAL IMPLICATIONS

Any patient with features of IPEX but with a normal Foxp3 gene should be screened for mutations in the IL-2 receptor subunit CD25.

Induction and biochemical purification of RNA-induced silencing complex from Drosophila S2 cells

The discovery of RNA interference (RNAi) has greatly simplified the process of suppressing genes in many experimental systems, including Caenorhabditis elegans, Drosophila, and mammalian cells. A sequence-specific nuclease complex, called the RNA-induced silencing complex (RISC), can be purified from cells undergoing RNAi. RISC shows RNase activity when exposed to RNAs homologous to the input double-stranded RNA (dsRNAs) but lacks activity in the presence of nontargeted RNAs. We describe the induction of RNAi by dsRNA in cultured Drosophila Schneider-2 (S2) cells and detail procedures for RISC purification from these cells. This purification approach has allowed us to identify several RISC components, including siRNAs, Argo naute 2 (Ago-2), Drosophila Fragile X related protein (dFXR), Vasa intronic gene (VIG), and the micrococcal nuclease family member Tudor-SN (Drosophila CG7008). RNAi is carried out by an endogenous pathway important for normal development in many organisms. In fact, organisms express hundreds of different microRNAs (miRNAs), small hairpin RNAs that function through the RNAi pathway to suppress expression of endogenous genes. The function of miRNAs is poorly understood, and most of their targets are unknown. Purified RISC complexes contain short interfering RNAs and endogenously expressed miRNAs and will be useful for studying many aspects of the RNAi machinery.

Short hairpin activated gene silencing in mammalian cells

RNA interference (RNAi) is now a popular method for silencing gene expression in a variety of systems. RNAi methods use double-stranded RNAs (dsRNAs) to target complementary RNAs for destruction. In mammalian systems, very short dsRNAs (22-25 bp) such as short interfering RNAs (siRNAs) or short hairpin RNAs (shRNAs) are used to avoid endogenous nonspecific antiviral responses that target longer dsRNAs. siRNAs elicit a transient silencing response, while shRNAs can be expressed continuously to establish stable gene silencing. shRNAs can be introduced into cells and animals using a variety of standard vectors as well as retroviral or lentiviral expression systems. This chapter describes the design, construction, validation, and use of shRNAs for silencing genes. We report our results from testing a variety of shRNA design features and shRNA expression vectors. We also provide methods that use shRNAs to permit different levels of gene expression. Additionally, we discuss some aspects important for constructing an information pipeline to support development of a large shRNA library.

A micrococcal nuclease homologue in RNAi effector complexes

RNA interference (RNAi) regulates gene expression by the cleavage of messenger RNA, by mRNA degradation and by preventing protein synthesis. These effects are mediated by a ribonucleoprotein complex known as RISC (RNA-induced silencing complex). We have previously identified four Drosophila components (short interfering RNAs, Argonaute 2 (ref. 2), VIG and FXR) of a RISC enzyme that degrades specific mRNAs in response to a double-stranded-RNA trigger. Here we show that Tudor-SN (tudor staphylococcal nuclease)--a protein containing five staphylococcal/micrococcal nuclease domains and a tudor domain--is a component of the RISC enzyme in Caenorhabditis elegans, Drosophila and mammals. Although Tudor-SN contains non-canonical active-site sequences, we show that purified Tudor-SN exhibits nuclease activity similar to that of other staphylococcal nucleases. Notably, both purified Tudor-SN and RISC are inhibited by a specific competitive inhibitor of micrococcal nuclease. Tudor-SN is the first RISC subunit to be identified that contains a recognizable nuclease domain, and could therefore contribute to the RNA degradation observed in RNAi.

Fragile X-related protein and VIG associate with the RNA interference machinery

RNA interference (RNAi) is a flexible gene silencing mechanism that responds to double-stranded RNA by suppressing homologous genes. Here, we report the characterization of RNAi effector complexes (RISCs) that contain small interfering RNAs and microRNAs (miRNAs). We identify two putative RNA-binding proteins, the Drosophila homolog of the fragile X mental retardation protein (FMRP), dFXR, and VIG (Vasa intronic gene), through their association with RISC. FMRP, the product of the human fragile X locus, regulates the expression of numerous mRNAs via an unknown mechanism. The possibility that dFXR, and potentially FMRP, use, at least in part, an RNAi-related mechanism for target recognition suggests a potentially important link between RNAi and human disease.

Xenopus ribosomal RNA gene intergenic spacer elements conferring transcriptional enhancement and nucleolar dominance-like competition in oocytes

Repeated within the intergenic spacers that separate adjacent ribosomal RNA (rRNA) genes in Xenopus laevis are several distinct sequence elements. These include transcription terminators, "region 0" repeats, "region 1" repeats, duplicated spacer promoters, and 42-bp enhancer elements that are embedded within 60 or 81-bp repeats. All have been reported to stimulate RNA polymerase I transcription from an adjacent gene promoter. A greater number of 42-bp enhancers/gene have been suggested to explain the preferential transcription of X. laevis rRNA genes in X. laevis x Xenopus borealis hybrids, an epigenetic phenomenon known as nucleolar dominance. However, the possible contribution of regions 0/1 and/or spacer promoters to the preferential transcription of X. laevis (over X. borealis) rRNA genes has never been tested directly. In this study, we systematically tested the various intergenic spacer elements for their contributions to promoter strength and nucleolar dominance-like competition in oocytes. In disagreement with a previous report, region 0 and region 1 repeats do not have significant enhancer activity, nor do they play a discernible role in X. laevis-X. borealis rRNA gene competition. Minigenes containing X. laevis spacer sequences are only dominant over minigenes having complete X. borealis spacers if a spacer promoter is located upstream of the 42-bp enhancers; X. laevis enhancers alone are not sufficient. These results provide additional evidence that spacer promoters together with adjacent enhancers form a functional activating unit in Xenopus oocytes.

Short hairpin RNAs (shRNAs) induce sequence-specific silencing in mammalian cells

RNA interference (RNAi) was first recognized in Caenorhabditis elegans as a biological response to exogenous double-stranded RNA (dsRNA), which induces sequence-specific gene silencing. RNAi represents a conserved regulatory motif, which is present in a wide range of eukaryotic organisms. Recently, we and others have shown that endogenously encoded triggers of gene silencing act through elements of the RNAi machinery to regulate the expression of protein-coding genes. These small temporal RNAs (stRNAs) are transcribed as short hairpin precursors (approximately 70 nt), processed into active, 21-nt RNAs by Dicer, and recognize target mRNAs via base-pairing interactions. Here, we show that short hairpin RNAs (shRNAs) can be engineered to suppress the expression of desired genes in cultured Drosophila and mammalian cells. shRNAs can be synthesized exogenously or can be transcribed from RNA polymerase III promoters in vivo, thus permitting the construction of continuous cell lines or transgenic animals in which RNAi enforces stable and heritable gene silencing.

Stable suppression of gene expression by RNAi in mammalian cells

In a diverse group of organisms including plants, Caenorhabditis elegans, Drosophila, and trypanosomes, double-stranded RNA (dsRNA) is a potent trigger of gene silencing. In several model systems, this natural response has been developed into a powerful tool for the investigation of gene function. Use of RNA interference (RNAi) as a genetic tool has recently been extended to mammalian cells, being inducible by treatment with small, approximately 22-nt RNAs that mimic those produced in the first step of the silencing process. Here, we show that some cultured murine cells specifically silence gene expression upon treatment with long dsRNAs (approximately 500 nt). This response shows hallmarks of conventional RNAi including silencing at the posttranscriptional level and the endogenous production of approximately 22-nt small RNAs. Furthermore, enforced expression of long, hairpin dsRNAs induced stable gene silencing. The ability to create stable "knock-down" cell lines expands the utility of RNAi in mammalian cells by enabling examination of phenotypes that develop over long time periods and lays the groundwork for by using RNAi in phenotype-based, forward genetic selections.

Regulation of transcriptional activation domain function by ubiquitin

The ability of transcriptional activation domains (TADs) to signal ubiquitin-mediated proteolysis suggests an involvement of the ubiquitin-proteasome pathway in transcription. To probe this involvement, we asked how ubiquitylation regulates the activity of a transcription factor containing the VP16 TAD. We show that the VP16 TAD signals ubiquitylation through the Met30 ubiquitin-ligase and that Met30 is also required for the VP16 TAD to activate transcription. The requirement for Met30 in transcription is circumvented by fusion of ubiquitin to the VP16 activator, demonstrating that activator ubiquitylation is essential for transcriptional activation. We propose that ubiquitylation regulates TAD function by serving as a dual signal for activation and activator destruction.

Argonaute2, a link between genetic and biochemical analyses of RNAi

Double-stranded RNA induces potent and specific gene silencing through a process referred to as RNA interference (RNAi) or posttranscriptional gene silencing (PTGS). RNAi is mediated by RNA-induced silencing complex (RISC), a sequence-specific, multicomponent nuclease that destroys messenger RNAs homologous to the silencing trigger. RISC is known to contain short RNAs ( approximately 22 nucleotides) derived from the double-stranded RNA trigger, but the protein components of this activity are unknown. Here, we report the biochemical purification of the RNAi effector nuclease from cultured Drosophila cells. The active fraction contains a ribonucleoprotein complex of approximately 500 kilodaltons. Protein microsequencing reveals that one constituent of this complex is a member of the Argonaute family of proteins, which are essential for gene silencing in Caenorhabditis elegans, Neurospora, and Arabidopsis. This observation begins the process of forging links between genetic analysis of RNAi from diverse organisms and the biochemical model of RNAi that is emerging from Drosophila in vitro systems.

Post-transcriptional gene silencing by double-stranded RNA

Imagine being able to knock out your favourite gene with only a day's work. Not just in one model system, but in virtually any organism: plants, flies, mice or cultured cells. This sort of experimental dream might one day become reality as we learn to harness the power of RNA interference, the process by which double-stranded RNA induces the silencing of homologous endogenous genes. How this phenomenon works is slowly becoming clear, and might help us to develop an effortless tool to probe gene function in cells and animals.

Role for a bidentate ribonuclease in the initiation step of RNA interference

RNA interference (RNAi) is the mechanism through which double-stranded RNAs silence cognate genes. In plants, this can occur at both the transcriptional and the post-transcriptional levels; however, in animals, only post-transcriptional RNAi has been reported to date. In both plants and animals, RNAi is characterized by the presence of RNAs of about 22 nucleotides in length that are homologous to the gene that is being suppressed. These 22-nucleotide sequences serve as guide sequences that instruct a multicomponent nuclease, RISC, to destroy specific messenger RNAs. Here we identify an enzyme, Dicer, which can produce putative guide RNAs. Dicer is a member of the RNase III family of nucleases that specifically cleave double-stranded RNAs, and is evolutionarily conserved in worms, flies, plants, fungi and mammals. The enzyme has a distinctive structure, which includes a helicase domain and dual RNase III motifs. Dicer also contains a region of homology to the RDE1/QDE2/ARGONAUTE family that has been genetically linked to RNAi.