Moonlighting glutamate formiminotransferases can functionally replace 5-formyltetrahydrofolate cycloligase

Metabolite Damage and Damage Control in a Minimal Genome

Analysis of the genes retained in the minimized Mycoplasma JCVI-Syn3A genome established that systems that repair or preempt metabolite damage are essential to life. Several genes known to have such functions were identified and experimentally validated, including 5-formyltetrahydrofolate cycloligase, coenzyme A (CoA) disulfide reductase, and certain hydrolases. Furthermore, we discovered that an enigmatic YqeK hydrolase domain fused to NadD has a novel proofreading function in NAD synthesis and could double as a MutT-like sanitizing enzyme for the nucleotide pool. Finally, we combined metabolomics and cheminformatics approaches to extend the core metabolic map of JCVI-Syn3A to include promiscuous enzymatic reactions and spontaneous side reactions. This extension revealed that several key metabolite damage control systems remain to be identified in JCVI-Syn3A, such as that for methylglyoxal.

Approaches for completing metabolic networks through metabolite damage and repair discovery

Metabolites are prone to damage, either via enzymatic side reactions, which collectively form the underground metabolism, or via spontaneous chemical reactions. The resulting non-canonical metabolites that can be toxic, are mended by dedicated "metabolite repair enzymes." Deficiencies in the latter can cause severe disease in humans, whereas inclusion of repair enzymes in metabolically engineered systems can improve the production yield of value-added chemicals. The metabolite damage and repair loops are typically not yet included in metabolic reconstructions and it is likely that many remain to be discovered. Here, we review strategies and associated challenges for unveiling non-canonical metabolites and metabolite repair enzymes, including systematic approaches based on high-resolution mass spectrometry, metabolome-wide side-activity prediction, as well as high-throughput substrate and phenotypic screens.

Evolution of folate biosynthesis and metabolism across algae and land plant lineages

Tetrahydrofolate and its derivatives, commonly known as folates, are essential for almost all living organisms. Besides acting as one-carbon donors and acceptors in reactions producing various important biomolecules such as nucleic and amino acids, as well as pantothenate, they also supply one-carbon units for methylation reactions. Plants along with bacteria, yeast and fungi synthesize folates de novo and therefore constitute a very important dietary source of folates for animals. All the major steps of folate biosynthesis and metabolism have been identified but only few have been genetically characterized in a handful of model plant species. The possible differences in the folate pathway between various plant and algal species have never been explored. In this study we present a comprehensive comparative study of folate biosynthesis and metabolism of all major land plant lineages as well as green and red algae. The study identifies new features of plant folate metabolism that might open new directions to folate research in plants.

A missense variant in FTCD is associated with arsenic metabolism and toxicity phenotypes in Bangladesh

Inorganic arsenic (iAs) is a carcinogen, and exposure to iAs via food and water is a global public health problem. iAs-contaminated drinking water alone affects >100 million people worldwide, including ~50 million in Bangladesh. Once absorbed into the blood stream, most iAs is converted to mono-methylated (MMA) and then di-methylated (DMA) forms, facilitating excretion in urine. Arsenic metabolism efficiency varies among individuals, in part due to genetic variation near AS3MT (arsenite methyltransferase; 10q24.32). To identify additional arsenic metabolism loci, we measured protein-coding variants across the human exome for 1,660 Bangladeshi individuals participating in the Health Effects of Arsenic Longitudinal Study (HEALS). Among the 19,992 coding variants analyzed exome-wide, the minor allele (A) of rs61735836 (p.Val101Met) in exon 3 of FTCD (formiminotransferase cyclodeaminase) was associated with increased urinary iAs% (P = 8x10-13), increased MMA% (P = 2x10-16) and decreased DMA% (P = 6x10-23). Among 2,401 individuals with arsenic-induced skin lesions (an indicator of arsenic toxicity and cancer risk) and 2,472 controls, carrying the low-efficiency A allele (frequency = 7%) was associated with increased skin lesion risk (odds ratio = 1.35; P = 1x10-5). rs61735836 is in weak linkage disequilibrium with all nearby variants. The high-efficiency/major allele (G/Valine) is human-specific and eliminates a start codon at the first 5´-proximal Kozak sequence in FTCD, suggesting selection against an alternative translation start site. FTCD is critical for catabolism of histidine, a process that generates one-carbon units that can enter the one-carbon/folate cycle, which provides methyl groups for arsenic metabolism. In our study population, FTCD and AS3MT SNPs together explain ~10% of the variation in DMA% and support a causal effect of arsenic metabolism efficiency on arsenic toxicity (i.e., skin lesions). In summary, this work identifies a coding variant in FTCD associated with arsenic metabolism efficiency, providing new evidence supporting the established link between one-carbon/folate metabolism and arsenic toxicity.

Metabolite damage and repair in metabolic engineering design

The necessarily sharp focus of metabolic engineering and metabolic synthetic biology on pathways and their fluxes has tended to divert attention from the damaging enzymatic and chemical side-reactions that pathway metabolites can undergo. Although historically overlooked and underappreciated, such metabolite damage reactions are now known to occur throughout metabolism and to generate (formerly enigmatic) peaks detected in metabolomics datasets. It is also now known that metabolite damage is often countered by dedicated repair enzymes that undo or prevent it. Metabolite damage and repair are highly relevant to engineered pathway design: metabolite damage reactions can reduce flux rates and product yields, and repair enzymes can provide robust, host-independent solutions. Herein, after introducing the core principles of metabolite damage and repair, we use case histories to document how damage and repair processes affect efficient operation of engineered pathways - particularly those that are heterologous, non-natural, or cell-free. We then review how metabolite damage reactions can be predicted, how repair reactions can be prospected, and how metabolite damage and repair can be built into genome-scale metabolic models. Lastly, we propose a versatile 'plug and play' set of well-characterized metabolite repair enzymes to solve metabolite damage problems known or likely to occur in metabolic engineering and synthetic biology projects.

Longevity of major coenzymes allows minimal de novo synthesis in microorganisms

Coenzymes are vital for cellular metabolism and act on the full spectrum of enzymatic reactions. Intrinsic chemical reactivity, enzyme promiscuity and high flux through their catalytic cycles make coenzymes prone to damage. To counteract such compromising factors and ensure stable levels of functional coenzymes, cells use a complex interplay between de novo synthesis, salvage, repair and degradation. However, the relative contribution of these factors is currently unknown, as is the overall stability of coenzymes in the cell. Here, we use dynamic ¹³C-labelling experiments to determine the half-life of major coenzymes of Escherichia coli. We find that coenzymes such as pyridoxal 5-phosphate, flavins, nicotinamide adenine dinucleotide (phosphate) and coenzyme A are remarkably stable in vivo and allow biosynthesis close to the minimal necessary rate. In consequence, they are essentially produced to compensate for dilution by growth and passed on over generations of cells. Exceptions are antioxidants, which are short-lived, suggesting an inherent requirement for increased renewal. Although the growth-driven turnover of stable coenzymes is apparently subject to highly efficient end-product homeostasis, we exemplify that coenzyme pools are propagated in excess in relation to actual growth requirements. Additional testing of Bacillus subtilis and Saccharomyces cerevisiae suggests that coenzyme longevity is a conserved feature in biology.

Purine biosynthesis is the bottleneck in trimethoprim-treated Bacillus subtilis

PURPOSE

Trimethoprim is a folate biosynthesis inhibitor. Tetrahydrofolates are essential for the transfer of C₁ units in several biochemical pathways including purine, thymine, methionine, and glycine biosynthesis. This study addressed the effects of folate biosynthesis inhibition on bacterial physiology.

EXPERIMENTAL DESIGN

Two complementary proteomic approaches were employed to analyze the response of Bacillus subtilis to trimethoprim. Acute changes in protein synthesis rates were monitored by radioactive pulse labeling of newly synthesized proteins and subsequent 2DE analysis. Changes in protein levels were detected using gel-free quantitative MS.

RESULTS

Proteins involved in purine and histidine biosynthesis, the σ^B -dependent general stress response, and sporulation were upregulated. Most prominently, the PurR-regulon required for de novo purine biosynthesis was derepressed indicating purine depletion. The general stress response was activated energy dependently and in a subpopulation of treated cultures an early onset of sporulation was observed, most likely triggered by low guanosine triphosphate levels. Supplementation of adenosine triphosphate, adenosine, and guanosine to the medium substantially decreased antibacterial activity, showing that purine depletion becomes the bottleneck in trimethoprim-treated B. subtilis.

CONCLUSIONS AND CLINICAL RELEVANCE

The frequently prescribed antibiotic trimethoprim causes purine depletion in B. subtilis, which can be complemented by supplementing purines to the medium.

Simultaneous presence of fhs and purT genes is disadvantageous for the fitness of Escherichia coli growth

In bacteria, alternate mechanisms are known to synthesize N(10)-formyltetrahydrofolate (N(10)-formyl-THF) and formyl glycinamide ribotide (fGAR), which are important in purine biosynthesis. In one of the mechanisms, a direct transfer of one carbon unit from formate allows Fhs to convert tetrahydrofolate to N(10)-formyl-THF, and PurT to convert glycinamide ribotide (GAR) to fGAR. Our bioinformatics analysis of fhs and purT genes (encoding Fhs and PurT) showed that in a majority of bacteria (∼94%), their presence was mutually exclusive. A large number of organisms possessing fhs lacked purT and vice versa. The phenomenon is so penetrating that even within a genus (Bacillus) if a species possessed fhs it lacked purT and vice versa. To investigate physiological importance of this phenomenon, we used Escherichia coli, which naturally lacks fhs (and possesses purT) as model. We generated strains, which possessed fhs and purT genes in singles or together. Deletion of purT from E. coli in the presence or absence of fhs did not confer a detectable growth disadvantage in pure cultures. However, growth competition assays revealed that the strains possessing either of the single genes outcompeted those possessing both the genes suggesting that mutual exclusion of purT and fhs in organisms confers fitness advantage in mixed cultures.

Functional analysis by protein biochemistry

To date a number of cereal genomes are fully sequenced and more are near completion. The information within these genomes will be of most use to scientists when every gene has been functionally characterized leading to the complete annotation of these genomes. This chapter describes how functional characterization of plant proteins can be achieved via in vitro or in vivo methods. The first section of this chapter describes the use of Escherichia coli as a host for expression of plant genes, followed by purification and in vitro characterization of the resultant enzyme. The second section of this chapter details the methods involved in transient gene expression in Zea mays leaf protoplasts for in vivo functional characterization of protein localization.

Regulation of transcription by 6S RNAs: insights from the Escherichia coli and Bacillus subtilis model systems

Whereas, the majority of bacterial non-coding RNAs and functional RNA elements regulate post-transcriptional processes, either by interacting with other RNAs via base-pairing or through binding of small ligands (riboswitches), 6S RNAs affect transcription itself by binding to the housekeeping holoenzyme of RNA polymerase (RNAP). Remarkably, 6S RNAs serve as RNA templates for bacterial RNAP, giving rise to the de novo synthesis of short transcripts, termed pRNAs (product RNAs). Hence, 6S RNAs prompt the enzyme to act as an RNA-dependent RNA polymerase (RdRP). Synthesis of pRNAs exceeding a certain length limit (~13 nt) persistently rearrange the 6S RNA structure, which in turn, disrupts the 6S RNA:RNAP complex. This pRNA synthesis-mediated "reanimation" of sequestered RNAP molecules represents the conceivably fastest mechanism for resuming transcription in cells that enter a new exponential growth phase. The many different 6S RNAs found in a wide variety of bacteria do not share strong sequence homology but have in common a conserved rod-shaped structure with a large internal loop, termed the central bulge; this architecture mediates specific binding to the active site of RNAP. In this article, we summarize the overall state of knowledge as well as very recent findings on the structure, function, and physiological effects of 6S RNA examples from the two model organisms, Escherichia coli and Bacillus subtilis. Comparison of the presently known properties of 6S RNAs in the two organisms highlights common principles as well as diverse features.

6S RNA: recent answers--future questions

6S RNA is a non-coding RNA, found in almost all phylogenetic branches of bacteria. Through its conserved secondary structure, resembling open DNA promoters, it binds to RNA polymerase and interferes with transcription at many promoters. That way, it functions as transcriptional regulator facilitating adaptation to stationary phase conditions. Strikingly, 6S RNA acts as template for the synthesis of small RNAs (pRNA), which trigger the disintegration of the inhibitory RNA polymerase-6S RNA complex releasing 6S RNA-dependent repression. The regulatory implications of 6S RNAs vary among different bacterial species depending on the lifestyle and specific growth conditions that they have to face. The influence of 6S RNA can be seen on many different processes including stationary growth, sporulation, light adaptation or intracellular growth of pathogenic bacteria. Recent structural and functional studies have yielded details of the interaction between E. coli 6S RNA and RNA polymerase. Genome-wide transcriptome analyses provided insight into the functional diversity of 6S RNAs. Moreover, the mechanism and physiological consequences of pRNA synthesis have been explored in several systems. A major function of 6S RNA as a guardian regulating the economic use of cellular resources under limiting conditions and stress emerges as a common perception from numerous recent studies.

Metabolite damage and its repair or pre-emption

It is increasingly evident that metabolites suffer various kinds of damage, that such damage happens in all organisms and that cells have dedicated systems for damage repair and containment. First, chemical biology is demonstrating that diverse metabolites are damaged by side reactions of 'promiscuous' enzymes or by spontaneous chemical reactions, that the products are useless or toxic and that the unchecked buildup of these products can be devastating. Second, genetic and genomic evidence from prokaryotes and eukaryotes is implicating a network of new, conserved enzymes that repair damaged metabolites or somehow pre-empt damage. Metabolite (that is, small-molecule) repair is analogous to macromolecule (DNA and protein) repair and seems from comparative genomic evidence to be equally widespread. Comparative genomics also implies that metabolite repair could be the function of many conserved protein families lacking known activities. How--and how well--cells deal with metabolite damage affects fields ranging from medical genetics to metabolic engineering.

Multifunctional enzymes in archaea: promiscuity and moonlight

Enzymes from many archaea colonizing extreme environments are of great interest because of their potential for various biotechnological processes and scientific value of evolution. Many enzymes from archaea have been reported to catalyze promiscuous reactions or moonlight in different functions. Here, we summarize known archaeal enzymes of both groups that include different kinds of proteins. Knowledge of their biochemical properties and three-dimensional structures has proved invaluable in understanding mechanism, application, and evolutionary implications of this manifestation. In addition, the review also summarizes the methods to unravel the extra function which almost was discovered serendipitously. The study of these amazing enzymes will provide clues to optimize protein engineering applications and how enzymes might have evolved on Earth.

6S RNA - an old issue became blue-green

6S RNA from Escherichia coli acts as a versatile transcriptional regulator by binding to the RNA polymerase and changing promoter selectivity. Although homologous 6S RNA structures exist in a wide range of bacteria, including cyanobacteria, our knowledge of 6S RNA function results almost exclusively from studies with E. coli. To test for potential structural and functional conservation, we selected four predicted cyanobacterial 6S RNAs (Synechocystis, Synechococcus, Prochlorococcus and Nostoc), which we compared with their E. coli counterpart. Temperature-gradient gel electrophoresis revealed similar thermodynamic transition profiles for all 6S RNAs, indicating basically similar secondary structures. Subtle differences in melting behaviour of the different RNAs point to minor structural variations possibly linked to differences in optimal growth temperature. Secondary structural analysis of three cyanobacterial 6S RNAs employing limited enzymic hydrolysis and in-line probing supported the predicted high degree of secondary structure conservation. Testing for functional homology we found that all cyanobacterial 6S RNAs were active in binding E. coli RNA polymerase and transcriptional inhibition, and had the ability to act as template for transcription of product RNAs (pRNAs). Deletion of the 6S RNA gene in Synechocystis did not significantly affect cell growth in liquid media but reduced fitness during growth on solid agar. While our study shows that basic 6S RNA functions are conserved in species as distantly related as E. coli and cyanobacteria, we also noted a subtle degree of divergence, which might reflect fundamental differences in transcriptional regulation and lifestyle, thus providing the first evidence for a possible physiological role in cyanobacteria.

Frontiers in metabolic reconstruction and modeling of plant genomes

A major goal of post-genomic biology is to reconstruct and model in silico the metabolic networks of entire organisms. Work on bacteria is well advanced, and is now under way for plants and other eukaryotes. Genome-scale modelling in plants is much more challenging than in bacteria. The challenges come from features characteristic of higher organisms (subcellular compartmentation, tissue differentiation) and also from the particular severity in plants of a general problem: genome content whose functions remain undiscovered. This problem results in thousands of genes for which no function is known ('undiscovered genome content') and hundreds of enzymatic and transport functions for which no gene is yet identified. The severity of the undiscovered genome content problem in plants reflects their genome size and complexity. To bring the challenges of plant genome-scale modelling into focus, we first summarize the current status of plant genome-scale models. We then highlight the challenges - and ways to address them - in three areas: identifying genes for missing processes, modelling tissues as opposed to single cells, and finding metabolic functions encoded by undiscovered genome content. We also discuss the emerging view that a significant fraction of undiscovered genome content encodes functions that counter damage to metabolites inflicted by spontaneous chemical reactions or enzymatic mistakes.

A 5-formyltetrahydrofolate cycloligase paralog from all domains of life: comparative genomic and experimental evidence for a cryptic role in thiamin metabolism

A paralog (here termed COG0212) of the ATP-dependent folate salvage enzyme 5-formyltetrahydrofolate cycloligase (5-FCL) occurs in all domains of life and, although typically annotated as 5-FCL in pro- and eukaryotic genomes, is of unknown function. COG0212 is similar in overall structure to 5-FCL, particularly in the substrate binding region, and has distant similarity to other kinases. The Arabidopsis thaliana COG0212 protein was shown to be targeted to chloroplasts and to be required for embryo viability. Comparative genomic analysis revealed that a high proportion (19%) of archaeal and bacterial COG0212 genes are clustered on the chromosome with various genes implicated in thiamin metabolism or transport but showed no such association between COG0212 and folate metabolism. Consistent with the bioinformatic evidence for a role in thiamin metabolism, ablating COG0212 in the archaeon Haloferax volcanii caused accumulation of thiamin monophosphate. Biochemical and functional complementation tests of several known and hypothetical thiamin-related activities (involving thiamin, its breakdown products, and their phosphates) were, however, negative. Also consistent with the bioinformatic evidence, the COG0212 proteins from A. thaliana and prokaryote sources lacked 5-FCL activity in vitro and did not complement the growth defect or the characteristic 5-formyltetrahydrofolate accumulation of a 5-FCL-deficient (ΔygfA) Escherichia coli strain. We therefore propose (a) that COG0212 has an unrecognized yet sometimes crucial role in thiamin metabolism, most probably in salvage or detoxification, and (b) that is not a 5-FCL and should no longer be so annotated.

Mthfs is an Essential Gene in Mice and a Component of the Purinosome

Tetrahydrofolates (THF) are a family of cofactors that function as one-carbon donors in folate-dependent one-carbon metabolism, a metabolic network required for the de novo synthesis of purines, thymidylate, and for the remethylation of homocysteine to methionine in the cytoplasm. 5-FormylTHF is not a cofactor in one-carbon metabolism, but serves as a storage form of THF cofactors. 5-formylTHF is mobilized back into the THF cofactor pool by methenyltetrahydrofolate synthetase (MTHFS), which catalyzes the irreversible and ATP-dependent conversion 5-formyltetrahydrofolate to 5,10-methenyltetrahydrofolate. Mthfs is not an essential gene in Arabidopsis, but MTHFS expression is elevated in animal tumors, enhances de novo purine synthesis, confers partial resistance to antifolate purine synthesis inhibitors and increases rates of folate catabolism in mammalian cell cultures. However, the mechanisms underlying these effects of MTHFS expression have yet to be established. The purpose of this study was to investigate the role and essentiality of MTHFS in mice. Mthfs was disrupted through the insertion of a gene trap vector between exons 1 and 2. Mthfs^gt/+ mice were fertile and viable. No Mthfs^gt/gt embryos were recovered from Mthfs^gt/+ intercrosses, indicating Mthfs is an essential gene in mice. Tissue MTHFS protein levels are decreased in both Mthfs^gt/+ and Mthfs^+/+ mice placed on a folate and choline deficient diet, and mouse embryonic fibroblasts from Mthfs^gt/+ embryos exhibit decreased capacity for de novo purine synthesis without impairment in de novo thymidylate synthesis. MTHFS was shown to co-localize with two enzymes of the de novo purine synthesis pathway in HeLa cells in a cell cycle-dependent manner, and to be modified by the small ubiquitin-like modifier (SUMO) protein. Mutation of the consensus SUMO modification sites on MTHFS eliminated co-localization of MTHFS with the de novo purine biosynthesis pathway under purine-deficient conditions. The results from this study indicate that MTHFS enhances purine biosynthesis by delivering 10-formylTHF to the purinosome in a SUMO-dependent fashion.

Synergistic use of plant-prokaryote comparative genomics for functional annotations

BACKGROUND

Identifying functions for all gene products in all sequenced organisms is a central challenge of the post-genomic era. However, at least 30-50% of the proteins encoded by any given genome are of unknown or vaguely known function, and a large number are wrongly annotated. Many of these 'unknown' proteins are common to prokaryotes and plants. We set out to predict and experimentally test the functions of such proteins. Our approach to functional prediction integrates comparative genomics based mainly on microbial genomes with functional genomic data from model microorganisms and post-genomic data from plants. This approach bridges the gap between automated homology-based annotations and the classical gene discovery efforts of experimentalists, and is more powerful than purely computational approaches to identifying gene-function associations.

RESULTS

Among Arabidopsis genes, we focused on those (2,325 in total) that (i) are unique or belong to families with no more than three members, (ii) occur in prokaryotes, and (iii) have unknown or poorly known functions. Computer-assisted selection of promising targets for deeper analysis was based on homology-independent characteristics associated in the SEED database with the prokaryotic members of each family. In-depth comparative genomic analysis was performed for 360 top candidate families. From this pool, 78 families were connected to general areas of metabolism and, of these families, specific functional predictions were made for 41. Twenty-one predicted functions have been experimentally tested or are currently under investigation by our group in at least one prokaryotic organism (nine of them have been validated, four invalidated, and eight are in progress). Ten additional predictions have been independently validated by other groups. Discovering the function of very widespread but hitherto enigmatic proteins such as the YrdC or YgfZ families illustrates the power of our approach.

CONCLUSIONS

Our approach correctly predicted functions for 19 uncharacterized protein families from plants and prokaryotes; none of these functions had previously been correctly predicted by computational methods. The resulting annotations could be propagated with confidence to over six thousand homologous proteins encoded in over 900 bacterial, archaeal, and eukaryotic genomes currently available in public databases.

Bacterial conversion of folinic acid is required for antifolate resistance

Antifolates, which are among the first antimicrobial agents invented, inhibit cell growth by creating an intracellular state of folate deficiency. Clinical resistance to antifolates has been mainly attributed to mutations that alter structure or expression of enzymes involved in de novo folate synthesis. We identified a Mycobacterium smegmatis mutant, named FUEL (which stands for folate utilization enzyme for leucovorin), that is hypersusceptible to antifolates. Chemical complementation indicated that FUEL is unable to metabolize folinic acid (also known as leucovorin or 5-formyltetrahydrofolate), whose metabolic function remains unknown. Targeted mutagenesis, genetic complementation, and biochemical studies showed that FUEL lacks 5,10-methenyltetrahydrofolate synthase (MTHFS; also called 5-formyltetrahydrofolate cyclo-ligase; EC 6.3.3.2) activity responsible for the only ATP-dependent, irreversible conversion of folinic acid to 5,10-methenyltetrahydrofolate. In trans expression of active MTHFS proteins from bacteria or human restored both antifolate resistance and folinic acid utilization to FUEL. Absence of MTHFS resulted in marked cellular accumulation of polyglutamylated species of folinic acid. Importantly, MTHFS also affected M. smegmatis utilization of monoglutamylated 5-methyltetrahydrofolate exogenously added to the medium. Likewise, Escherichia coli mutants lacking MTHFS became susceptible to antifolates. These results indicate that folinic acid conversion by MTHFS is required for bacterial intrinsic antifolate resistance and folate homeostatic control. This novel mechanism of antimicrobial antifolate resistance might be targeted to sensitize bacterial pathogens to classical antifolates.

Phenotypic landscape of a bacterial cell

The explosion of sequence information in bacteria makes developing high-throughput, cost-effective approaches to matching genes with phenotypes imperative. Using E. coli as proof of principle, we show that combining large-scale chemical genomics with quantitative fitness measurements provides a high-quality data set rich in discovery. Probing growth profiles of a mutant library in hundreds of conditions in parallel yielded > 10,000 phenotypes that allowed us to study gene essentiality, discover leads for gene function and drug action, and understand higher-order organization of the bacterial chromosome. We highlight new information derived from the study, including insights into a gene involved in multiple antibiotic resistance and the synergy between a broadly used combinatory antibiotic therapy, trimethoprim and sulfonamides. This data set, publicly available at http://ecoliwiki.net/tools/chemgen/, is a valuable resource for both the microbiological and bioinformatic communities, as it provides high-confidence associations between hundreds of annotated and uncharacterized genes as well as inferences about the mode of action of several poorly understood drugs.