On-line targeted metabolomics for real-time monitoring of relevant compounds in fermentation processes.

Fermentation monitoring is a powerful tool for bioprocess development and optimization. On-line metabolomics is a technology that is starting to gain attention as a bioprocess monitoring tool, allowing the direct measurement of many compounds in the fermentation broth at a very high time resolution. In this work, targeted on-line metabolomics was used to monitor 40 metabolites of interest during three Escherichia coli succinate production fermentation experiments every 5 min with a triple quadrupole mass spectrometer. This allowed capturing high-time resolution biological data that can provide critical information for process optimization. For nine of these metabolites, simple univariate regression models were used to model compound concentration from their on-line mass spectrometry peak area. These on-line metabolomics univariate models performed comparably to vibrational spectroscopy multivariate partial least squares regressions models reported in the literature, which typically are much more complex and time consuming to build. In conclusion, this work shows how on-line metabolomics can be used to directly monitor many bioprocess compounds of interest and obtain rich biological and bioprocess data.

Rapid HILIC-Z ion mobility mass spectrometry (RHIMMS) method for untargeted metabolomics of complex biological samples.

INTRODUCTION: Recent advances in high-throughput methodologies in the 'omics' and synthetic biology fields call for rapid and sensitive workflows in the metabolic phenotyping of complex biological samples. OBJECTIVE: The objective of this research was to evaluate a straightforward to implement LC-MS metabolomics method using a commercially available chromatography column that provides increased throughput. Reducing run time can potentially impact chromatography and therefore the effects of ion mobility spectrometry to expand peak capacity were also evaluated. Additional confidence provided via collision cross section measurements for detected features was also explored. METHODS: A rapid untargeted metabolomics workflow was developed with broad metabolome coverage, combining zwitterionic-phase hydrophilic interaction chromatography (HILIC-Z) with drift tube ion mobility-quadrupole time-of-flight (DTIM-qTOF) mass spectrometry. The analytical performance of our method was explored using extracts from complex biological samples, including a reproducibility study on chicken serum and a simple comparative study on a bacterial metabolome. RESULTS: The method is acronymised RHIMMS for rapid HILIC-Z ion mobility mass spectrometry. We present the RHIMMS workflow starting with data acquisition, followed by data processing and analysis. RHIMMS demonstrates improved chromatographic separation for a selection of metabolites with wide physicochemical properties while maintaining reproducibility at better than 20% over 200 injections at 3.5 min per sample for the selected metabolites, and a mean of 13.9% for the top 50 metabolites by intensity. Additionally, the combination of rapid chromatographic separation with ion mobility allows improved annotation and the ability to distinguish isobaric compounds. CONCLUSION: Our results demonstrate RHIMMS to be a rapid, reproducible, sensitive and high-resolution analytical platform that is highly applicable to the untargeted metabolomics analysis of complex samples.

On-line untargeted metabolomics monitoring of an Escherichia coli succinate fermentation process.

The real-time monitoring of metabolites (RTMet) is instrumental for the industrial production of biobased fermentation products. This study shows the first application of untargeted on-line metabolomics for the monitoring of undiluted fermentation broth samples taken automatically from a 5 L bioreactor every 5 min via flow injection mass spectrometry. The travel time from the bioreactor to the mass spectrometer was 30 s. Using mass spectrometry allows, on the one hand, the direct monitoring of targeted key process compounds of interest and, on the other hand, provides information on hundreds of additional untargeted compounds without requiring previous calibration data. In this study, this technology was applied in an Escherichia coli succinate fermentation process and 886 different m/z signals were monitored, including key process compounds (glucose, succinate, and pyruvate), potential biomarkers of biomass formation such as (R)-2,3-dihydroxy-isovalerate and (R)-2,3-dihydroxy-3-methylpentanoate and compounds from the pentose phosphate pathway and nucleotide metabolism, among others. The main advantage of the RTMet technology is that it allows the monitoring of hundreds of signals without the requirement of developing partial least squares regression models, making it a perfect tool for bioprocess monitoring and for testing many different strains and process conditions for bioprocess development.

Metabolomic Profiling of End-Stage Heart Failure Secondary to Chronic Chagas Cardiomyopathy.

Chronic Chagas cardiomyopathy (CCC) is the most frequent and severe clinical form of chronic Chagas disease, representing one of the leading causes of morbidity and mortality in Latin America, and a growing global public health problem. There is currently no approved treatment for CCC; however, omics technologies have enabled significant progress to be made in the search for new therapeutic targets. The metabolic alterations associated with pathogenic mechanisms of CCC and their relationship to cellular and immunopathogenic processes in cardiac tissue remain largely unknown. This exploratory study aimed to evaluate the potential underlying pathogenic mechanisms in the failing myocardium of patients with end-stage heart failure (ESHF) secondary to CCC by applying an untargeted metabolomic profiling approach. Cardiac tissue samples from the left ventricle of patients with ESHF of CCC etiology (n = 7) and healthy donors (n = 7) were analyzed using liquid chromatography-mass spectrometry. Metabolite profiles showed altered branched-chain amino acid and acylcarnitine levels, decreased fatty acid uptake and oxidation, increased activity of the pentose phosphate pathway, dysregulation of the TCA cycle, and alterations in critical cellular antioxidant systems. These findings suggest processes of energy deficit, alterations in substrate availability, and enhanced production of reactive oxygen species in the affected myocardium. This profile potentially contributes to the development and maintenance of a chronic inflammatory state that leads to progression and severity of CCC. Further studies involving larger sample sizes and comparisons with heart failure patients without CCC are needed to validate these results, opening an avenue to investigate new therapeutic approaches for the treatment and prevention of progression of this unique and severe cardiomyopathy.

Functional colour genes and signals of selection in colour-polymorphic salamanders.

Coloration has been associated with multiple biologically relevant traits that drive adaptation and diversification in many taxa. However, despite the great diversity of colour patterns present in amphibians the underlying molecular basis is largely unknown. Here, we use insight from a highly colour-variable lineage of the European fire salamander (Salamandra salamandra bernardezi) to identify functional associations with striking variation in colour morph and pattern. The three focal colour morphs-ancestral black-yellow striped, fully yellow and fully brown-differed in pattern, visible coloration and cellular composition. From population genomic analyses of up to 4,702 loci, we found no correlations of neutral population genetic structure with colour morph. However, we identified 21 loci with genotype-phenotype associations, several of which relate to known colour genes. Furthermore, we inferred response to selection at up to 142 loci between the colour morphs, again including several that relate to coloration genes. By transcriptomic analysis across all different combinations, we found 196 differentially expressed genes between yellow, brown and black skin, 63 of which are candidate genes involved in animal coloration. The concordance across different statistical approaches and 'omic data sets provide several lines of evidence for loci linked to functional differences between colour morphs, including TYR, CAMK1 and PMEL. We found little association between colour morph and the metabolomic profile of its toxic compounds from the skin secretions. Our research suggests that current ecological and evolutionary hypotheses for the origins and maintenance of these striking colour morphs may need to be revisited.

Changing environments and genetic variation: natural variation in inbreeding does not compromise short-term physiological responses.

Selfing plant lineages are surprisingly widespread and successful in a broad range of environments, despite showing reduced genetic diversity, which is predicted to reduce their long-term evolutionary potential. However, appropriate short-term plastic responses to new environmental conditions might not require high levels of standing genetic variation. In this study, we tested whether mating system variation among populations, and associated changes in genetic variability, affected short-term responses to environmental challenges. We compared relative fitness and metabolome profiles of naturally outbreeding (genetically diverse) and inbreeding (genetically depauperate) populations of a perennial plant, Arabidopsis lyrata, under constant growth chamber conditions and an outdoor common garden environment outside its native range. We found no effect of inbreeding on survival, flowering phenology or short-term physiological responses. Specifically, naturally occurring inbreeding had no significant effects on the plasticity of metabolome profiles, using either multivariate approaches or analysis of variation in individual metabolites, with inbreeding populations showing similar physiological responses to outbreeding populations over time in both growing environments. We conclude that low genetic diversity in naturally inbred populations may not always compromise fitness or short-term physiological capacity to respond to environmental change, which could help to explain the global success of selfing mating strategies.

MetaboCraft: building a Minecraft plugin for metabolomics.

Motivation: The rapid advances in metabolomics pose a significant challenge in presentation and interpretation of results. Development of new, engaging visual aids is crucial to advancing our understanding of new findings. Results: We have developed MetaboCraft, a Minecraft plugin which creates immersive visualizations of metabolic networks and pathways in a 3D environment and allows the results of user experiments to be viewed in this context, presenting a novel approach to exploring the metabolome. Availability and implementation: https://github.com/argymeg/MetaboCraft/; https://hub.docker.com/r/ronandaly/metabocraft/. Supplementary information: Supplementary data are available at Bioinformatics online.

Topic modeling for untargeted substructure exploration in metabolomics.

The potential of untargeted metabolomics to answer important questions across the life sciences is hindered because of a paucity of computational tools that enable extraction of key biochemically relevant information. Available tools focus on using mass spectrometry fragmentation spectra to identify molecules whose behavior suggests they are relevant to the system under study. Unfortunately, fragmentation spectra cannot identify molecules in isolation but require authentic standards or databases of known fragmented molecules. Fragmentation spectra are, however, replete with information pertaining to the biochemical processes present, much of which is currently neglected. Here, we present an analytical workflow that exploits all fragmentation data from a given experiment to extract biochemically relevant features in an unsupervised manner. We demonstrate that an algorithm originally used for text mining, latent Dirichlet allocation, can be adapted to handle metabolomics datasets. Our approach extracts biochemically relevant molecular substructures ("Mass2Motifs") from spectra as sets of co-occurring molecular fragments and neutral losses. The analysis allows us to isolate molecular substructures, whose presence allows molecules to be grouped based on shared substructures regardless of classical spectral similarity. These substructures, in turn, support putative de novo structural annotation of molecules. Combining this spectral connectivity to orthogonal correlations (e.g., common abundance changes under system perturbation) significantly enhances our ability to provide mechanistic explanations for biological behavior.

Novel compounds targeting the enterohemorrhagic Escherichia coli type three secretion system reveal insights into mechanisms of secretion inhibition.

Anti-virulence (AV) compounds are a promising alternative to traditional antibiotics for fighting bacterial infections. The Type Three Secretion System (T3SS) is a well-studied and attractive AV target, given that it is widespread in more than 25 species of Gram-negative bacteria, including enterohemorrhagic E. coli (EHEC), and as it is essential for host colonization by many pathogens. In this work, we designed, synthesized and tested a new series of compounds that block the functionality of the T3SS of EHEC. Affinity chromatography experiments identified the primary target of the compounds as the T3SS needle pore protein EspD, which is essential for effector protein translocation into host cells. These data were supported by mechanistic studies that determined the coiled-coil domain 1 of EspD as a key compound-binding site, thereby preventing correct assembly of the T3SS complex on the cell surface. However, binding of inhibitors to EspD or deletion of EspD itself did not result in transcriptional down-regulation of effector proteins. Instead, we found the compounds to exhibit dual-functionality by also down-regulating transcription of the entire chromosomal locus encoding the T3SS, further demonstrating their desirability and effectiveness.

PiMP my metabolome: an integrated, web-based tool for LC-MS metabolomics data.

SUMMARY: The Polyomics integrated Metabolomics Pipeline (PiMP) fulfils an unmet need in metabolomics data analysis. PiMP offers automated and user-friendly analysis from mass spectrometry data acquisition to biological interpretation. Our key innovations are the Summary Page, which provides a simple overview of the experiment in the format of a scientific paper, containing the key findings of the experiment along with associated metadata; and the Metabolite Page, which provides a list of each metabolite accompanied by 'evidence cards', which provide a variety of criteria behind metabolite annotation including peak shapes, intensities in different sample groups and database information. AVAILABILITY AND IMPLEMENTATION: PiMP is available at http://polyomics.mvls.gla.ac.uk, and access is freely available on request. 50 GB of space is allocated for data storage, with unrestricted number of samples and analyses per user. Source code is available at https://github.com/RonanDaly/pimp and licensed under the GPL. CONTACT: karl.burgess@glasgow.ac.uk. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Unexpected differential metabolic responses of Campylobacter jejuni to the abundant presence of glutamate and fucose.

INTRODUCTION: Campylobacter jejuni is the leading cause of foodborne bacterial enteritis in humans, and yet little is known in regard to how genetic diversity and metabolic capabilities among isolates affect their metabolic phenotype and pathogenicity. OBJECTIVES: For instance, the C. jejuni 11168 strain can utilize both L-fucose and L-glutamate as a carbon source, which provides the strain with a competitive advantage in some environments and in this study we set out to assess the metabolic response of C. jejuni 11168 to the presence of L-fucose and L-glutamate in the growth medium. METHODS: To achieve this, untargeted hydrophilic liquid chromatography coupled to mass spectrometry was used to obtain metabolite profiles of supernatant extracts obtained at three different time points up to 24 h. RESULTS: This study identified both the depletion and the production and subsequent release of a multitude of expected and unexpected metabolites during the growth of C. jejuni 11168 under three different conditions. A large set of standards allowed identification of a number of metabolites. Further mass spectrometry fragmentation analysis allowed the additional annotation of substrate-specific metabolites. The results show that C. jejuni 11168 upon L-fucose addition indeed produces degradation products of the fucose pathway. Furthermore, methionine was faster depleted from the medium, consistent with previously-observed methionine auxotrophy. CONCLUSIONS: Moreover, a multitude of not previously annotated metabolites in C. jejuni were found to be increased specifically upon L-fucose addition. These metabolites may well play a role in the pathogenicity of this C. jejuni strain.

Unsupervised Discovery and Comparison of Structural Families Across Multiple Samples in Untargeted Metabolomics.

In untargeted metabolomics approaches, the inability to structurally annotate relevant features and map them to biochemical pathways is hampering the full exploitation of many metabolomics experiments. Furthermore, variable metabolic content across samples result in sparse feature matrices that are statistically hard to handle. Here, we introduce MS2LDA+ that tackles both above-mentioned problems. Previously, we presented MS2LDA, which extracts biochemically relevant molecular substructures ("Mass2Motifs") from a collection of fragmentation spectra as sets of co-occurring molecular fragments and neutral losses, thereby recognizing building blocks of metabolomics. Here, we extend MS2LDA to handle multiple metabolomics experiments in one analysis, resulting in MS2LDA+. By linking Mass2Motifs across samples, we expose the variability in prevalence of structurally related metabolite families. We validate the differential prevalence of substructures between two distinct samples groups and apply it to fecal samples. Subsequently, within one sample group of urines, we rank the Mass2Motifs based on their variance to assess whether xenobiotic-derived substructures are among the most-variant Mass2Motifs. Indeed, we could ascribe 22 out of the 30 most-variant Mass2Motifs to xenobiotic-derived substructures including paracetamol/acetaminophen mercapturate and dimethylpyrogallol. In total, we structurally characterized 101 Mass2Motifs with biochemically or chemically relevant substructures. Finally, we combined the discovered metabolite families with full scan feature intensity information to obtain insight into core metabolites present in most samples and rare metabolites present in small subsets now linked through their common substructures. We conclude that by biochemical grouping of metabolites across samples MS2LDA+ aids in structural annotation of metabolites and guides prioritization of analysis by using Mass2Motif prevalence.

Probing the metabolic network in bloodstream-form Trypanosoma brucei using untargeted metabolomics with stable isotope labelled glucose.

Metabolomics coupled with heavy-atom isotope-labelled glucose has been used to probe the metabolic pathways active in cultured bloodstream form trypomastigotes of Trypanosoma brucei, a parasite responsible for human African trypanosomiasis. Glucose enters many branches of metabolism beyond glycolysis, which has been widely held to be the sole route of glucose metabolism. Whilst pyruvate is the major end-product of glucose catabolism, its transamination product, alanine, is also produced in significant quantities. The oxidative branch of the pentose phosphate pathway is operative, although the non-oxidative branch is not. Ribose 5-phosphate generated through this pathway distributes widely into nucleotide synthesis and other branches of metabolism. Acetate, derived from glucose, is found associated with a range of acetylated amino acids and, to a lesser extent, fatty acids; while labelled glycerol is found in many glycerophospholipids. Glucose also enters inositol and several sugar nucleotides that serve as precursors to macromolecule biosynthesis. Although a Krebs cycle is not operative, malate, fumarate and succinate, primarily labelled in three carbons, were present, indicating an origin from phosphoenolpyruvate via oxaloacetate. Interestingly, the enzyme responsible for conversion of phosphoenolpyruvate to oxaloacetate, phosphoenolpyruvate carboxykinase, was shown to be essential to the bloodstream form trypanosomes, as demonstrated by the lethal phenotype induced by RNAi-mediated downregulation of its expression. In addition, glucose derivatives enter pyrimidine biosynthesis via oxaloacetate as a precursor to aspartate and orotate.

Glycolysis and pyrimidine biosynthesis are required for replication of adherent-invasive Escherichia coli in macrophages.

Adherent-invasive Escherichia coli (AIEC) have been implicated in the aetiology of Crohn's disease (CD), a chronic inflammatory bowel condition. It has been proposed that AIEC-infected macrophages produce high levels of pro-inflammatory cytokines thus contributing to the inflammation observed in CD. AIEC can replicate in macrophages and we wanted to determine if bacterial replication was linked to the high level of cytokine production associated with AIEC-infected macrophages. Therefore, we undertook a genetic analysis of the metabolic requirements for AIEC replication in the macrophage and we show that AIEC replication in this niche is dependent on bacterial glycolysis. In addition, our analyses indicate that AIEC have access to a wide range of nutrients in the macrophage, although the levels of purines and pyrimidines do appear to be limiting. Finally, we show that the macrophage response to AIEC infection is indistinguishable from the response to the non-replicating glycolysis mutant (ΔpfkAB) and a non-pathogenic strain of E. coli, MG1655. Therefore, AIEC does not appear to subvert the normal macrophage response to E. coli during infection.

Potent trypanocidal curcumin analogs bearing a monoenone linker motif act on trypanosoma brucei by forming an adduct with trypanothione.

We have previously reported that curcumin analogs with a C7 linker bearing a C4-C5 olefinic linker with a single keto group at C3 (enone linker) display midnanomolar activity against the bloodstream form of Trypanosoma brucei. However, no clear indication of their mechanism of action or superior antiparasitic activity relative to analogs with the original di-ketone curcumin linker was apparent. To further investigate their utility as antiparasitic agents, we compare the cellular effects of curcumin and the enone linker lead compound 1,7-bis(4-hydroxy-3-methoxyphenyl)hept-4-en-3-one (AS-HK014) here. An AS-HK014-resitant line, trypanosomes adapted to AS-HK014 (TA014), was developed by in vitro exposure to the drug. Metabolomic analysis revealed that exposure to AS-HK014, but not curcumin, rapidly depleted glutathione and trypanothione in the wild-type line, although almost all other metabolites were unchanged relative to control. In TA014 cells, thiol levels were similar to untreated wild-type cells and not significantly depleted by AS-HK014. Adducts of AS-HK014 with both glutathione and trypanothione were identified in AS-HK014-exposed wild-type cells and reproduced by chemical reaction. However, adduct accumulation in sensitive cells was much lower than in resistant cells. TA014 cells did not exhibit any changes in sequence or protein levels of glutathione synthetase and γ-glutamylcysteine synthetase relative to wild-type cells. We conclude that monoenone curcuminoids have a different mode of action than curcumin, rapidly and specifically depleting thiol levels in trypanosomes by forming an adduct. This adduct may ultimately be responsible for the highly potent trypanocidal and antiparasitic activity of the monoenone curcuminoids.

Recovery from rapamycin: drug-insensitive activity of yeast target of rapamycin complex 1 (TORC1) supports residual proliferation that dilutes rapamycin among progeny cells.

The target of rapamycin complex 1 (TORC1) is a key conserved regulator of eukaryotic cell growth. The xenobiotic rapamycin is a potent inhibitor of the yeast complex. Surprisingly, the EGO complex, a nonessential in vivo activator of TORC1, is somehow required for yeast cells to recover efficiently from a period of treatment with rapamycin. Why? Here, we found that rapamycin is only a partial inhibitor of TORC1. We confirmed that saturating amounts of rapamycin do not fully inhibit proliferation of wild-type cells, and we found that the residual proliferation in the presence of the drug is dependent on the EGO complex and on the activity of TORC1. We found that this residual TORC1-dependent proliferation is key to recovery from rapamycin treatment. First, the residual proliferation rate correlates with the ability of cells to recover from treatment. Second, the residual proliferation rate persists long after washout of the drug and until cells recover. Third, the total observable pool of cell-associated rapamycin is extremely stable and decreases only with increasing cell number after washout of the drug. Finally, consideration of the residual proliferation rate alone accurately and quantitatively accounts for the kinetics of recovery of wild-type cells and for the nature and severity of the ego- mutant defect. Overall, our results revealed that rapamycin is a partial inhibitor of yeast TORC1, that persistence of the drug limits recovery, and that rapamycin is not detoxified by yeast but is passively diluted among progeny cells because of residual proliferation.

The metabolic enzyme AdhE controls the virulence of Escherichia coli†O157:H7.

Classical studies have focused on the role that individual regulators play in controlling virulence gene expression. An emerging theme, however, is that bacterial metabolism also plays a key role in this process. Our previous work identified a series of proteins that were implicated in the regulation of virulence. One of these proteins was AdhE, a bi-functional acetaldehyde-CoA dehydrogenase and alcohol dehydrogenase. Deletion of its gene (adhE) resulted in elevated levels of extracellular acetate and a stark pleiotropic phenotype: strong suppression of the Type Three Secretion System (T3SS) and overexpression of non-functional flagella. Correspondingly, the adhE mutant bound poorly to host cells and was unable to swim. Furthermore, the mutant was significantly less virulent than its parent when tested in vivo, which supports the hypothesis that attachment and motility are central to the colonization process. The molecular basis by which AdhE affects virulence gene regulation was found to be multifactorial, involving acetate-stimulated transcription of flagella expression and post-transcriptional regulation of the T3SS through Hfq. Our study reveals fascinating insights into the links between bacterial physiology, the expression of virulence genes, and the underlying molecular mechanism mechanisms by which these processes are regulated.

MetAssign: probabilistic annotation of metabolites from LC-MS data using a Bayesian clustering approach.

MOTIVATION: The use of liquid chromatography coupled to mass spectrometry has enabled the high-throughput profiling of the metabolite composition of biological samples. However, the large amount of data obtained can be difficult to analyse and often requires computational processing to understand which metabolites are present in a sample. This article looks at the dual problem of annotating peaks in a sample with a metabolite, together with putatively annotating whether a metabolite is present in the sample. The starting point of the approach is a Bayesian clustering of peaks into groups, each corresponding to putative adducts and isotopes of a single metabolite. RESULTS: The Bayesian modelling introduced here combines information from the mass-to-charge ratio, retention time and intensity of each peak, together with a model of the inter-peak dependency structure, to increase the accuracy of peak annotation. The results inherently contain a quantitative estimate of confidence in the peak annotations and allow an accurate trade-off between precision and recall. Extensive validation experiments using authentic chemical standards show that this system is able to produce more accurate putative identifications than other state-of-the-art systems, while at the same time giving a probabilistic measure of confidence in the annotations. AVAILABILITY AND IMPLEMENTATION: The software has been implemented as part of the mzMatch metabolomics analysis pipeline, which is available for download at http://mzmatch.sourceforge.net/.

Nanotopographical effects on mesenchymal stem cell morphology and phenotype.

There is a rapidly growing body of literature on the effects of topography and critically, nanotopography on cell adhesion, apoptosis and differentiation. Understanding the effects of nanotopography on cell adhesion and morphology and the consequences of cell shape changes in the nucleus, and consequently, gene expression offers new approaches to the elucidation and potential control of stem cell differentiation. In the current study we have used molecular approaches in combination with immunohistology and transcript analysis to understand the role of nanotopography on mesenchymal stem cell morphology and phenotype. Results demonstrate large changes in cell adhesion, nucleus and lamin morphologies in response to the different nanotopographies. Furthermore, these changes relate to alterations in packing of chromosome territories within the interphase nucleus. This, in turn, leads to changes in transcription factor activity and functional (phenotypical) signalling including cell metabolism. Nanotopography provides a useful, non-invasive tool for studying cellular mechanotransduction, gene and protein expression patterns, through effects on cell morphology. The different nanotopographies examined, result in different morphological changes in the cyto- and nucleo-skeleton. We propose that both indirect (biochemical) and direct (mechanical) signalling are important in these early stages of regulating stem cell fate as a consequence of altered metabolic changes and altered phenotype. The current studies provide new insight on cell-surface interactions and enhance our understanding of the modulation of stem cell differentiation with significant potential application in regenerative medicine.

Optimisation of surfactin yield in Bacillus using data-efficient active learning and high-throughput mass spectrometry.

Integration of machine learning and high throughput measurements are essential to drive the next generation of the design-build-test-learn (DBTL) cycle in synthetic biology. Here, we report the use of active learning in combination with metabolomics for optimising production of surfactin, a complex lipopeptide resulting from a non-ribosomal assembly pathway. We designed a media optimisation algorithm that iteratively learns the yield landscape and steers the media composition toward maximal production. The algorithm led to a 160 % yield increase after three DBTL runs as compared to an M9 baseline. Metabolomics data helped to elucidate the underpinning biochemistry for yield improvement and revealed Pareto-like trade-offs in production of other lipopeptides from related pathways. We found positive associations between organic acids and surfactin, suggesting a key role of central carbon metabolism, as well as system-wide anisotropies in how metabolism reacts to shifts in carbon and nitrogen levels. Our framework offers a novel data-driven approach to improve yield of biological products with complex synthesis pathways that are not amenable to traditional yield optimisation strategies.