Implications of confounding from unmodeled interactions between explanatory variables when using latent variable regression models to make inferences.

With linear dependency between the explanatory variables, partial least squares (PLS) regression is commonly used for regression analysis. If the response variable correlates to a high degree with the explanatory variables, a model with excellent predictive ability can usually be obtained. Ranking of variable importance is commonly used to interpret the model and sometimes this interpretation guides further experimentation. For instance, when analyzing natural product extracts for bioactivity, an underlying assumption is that the highest ranked compounds represent the best candidates for isolation and further testing. A problem with this approach is that in most cases the number of compounds is larger than the number of samples (and usually much larger) and that the concentrations of the compounds correlate. Furthermore, compounds may interact as synergists or as antagonists. If the modelling process does not account for this possibility, the interpretation can be thoroughly wrong since unmodelled variables that strongly influence the response will give rise to confounding of a first order PLS model and send the experimenter on a wrong track. We show the consequences of this by a practical example from natural product research. Furthermore, we show that by including the possibility of interactions between explanatory variables, visualization using a selectivity ratio plot may provide model interpretation that can be used to make inferences.

Discovery of megapolipeptins by genome mining of a Burkholderiales bacteria collection.

Burkholderiales bacteria have emerged as a promising source of structurally diverse natural products that are expected to play important ecological and industrial roles. This order ranks in the top three in terms of predicted natural product diversity from available genomes, warranting further genome sequencing efforts. However, a major hurdle in obtaining the predicted products is that biosynthetic genes are often 'silent' or poorly expressed. Here we report complementary strain isolation, genomics, metabolomics, and synthetic biology approaches to enable natural product discovery. First, we built a collection of 316 rhizosphere-derived Burkholderiales strains over the course of five years. We then selected 115 strains for sequencing using the mass spectrometry pipeline IDBac to avoid strain redundancy. After predicting and comparing the biosynthetic potential of each strain, a biosynthetic gene cluster that was silent in the native Paraburkholderia megapolitana and Paraburkholderia acidicola producers was cloned and activated by heterologous expression in a Burkholderia sp. host, yielding megapolipeptins A and B. Megapolipeptins are unusual polyketide, nonribosomal peptide, and polyunsaturated fatty acid hybrids that show low structural similarity to known natural products, highlighting the advantage of our Burkholderiales genomics-driven and synthetic biology-enabled pipeline to discover novel natural products.

Discovery of a lagriamide polyketide by integrated genome mining, isotopic labeling, and untargeted metabolomics.

Microorganisms from the order Burkholderiales have been the source of a number of important classes of natural products in recent years. For example, study of the beetle-associated symbiont Burkholderia gladioli led to the discovery of the antifungal polyketide lagriamide; an important molecule from the perspectives of both biotechnology and chemical ecology. As part of a wider project to sequence Burkholderiales genomes from our in-house Burkholderiales library we identified a strain containing a biosynthetic gene cluster (BGC) similar to the original lagriamide BGC. Structure prediction failed to identify any candidate masses for the products of this BGC from untargeted metabolomics mass spectrometry data. However, genome mining from publicly available databases identified fragments of this BGC from a culture collection strain of Paraburkholderia. Whole genome sequencing of this strain revealed the presence of a homologue of this BGC with very high sequence identity. Stable isotope feeding of the two strains in parallel using our newly developed IsoAnalyst platform identified the product of this lagriamide-like BGC directly from the crude fermentation extracts, affording a culturable supply of this interesting compound class. Using a combination of bioinformatic, computational and spectroscopic methods we defined the absolute configurations for all 11 chiral centers in this new metabolite, which we named lagriamide B. Biological testing of lagriamide B against a panel of 21 bacterial and fungal pathogens revealed antifungal activity against the opportunistic human pathogen Aspergillus niger, while image-based Cell Painting analysis indicated that lagriamide B also causes actin filament disruption in U2-OS osteosarcoma cells.

Targeted sampling of natural product space to identify bioactive natural product-like polyketide macrolides.

Polyketide or polyketide-like macrolides (pMLs) continue to serve as a source of inspiration for drug discovery. However, their inherent structural and stereochemical complexity challenges efforts to explore related regions of chemical space more broadly. Here, we report a strategy termed the Targeted Sampling of Natural Product space (TSNaP) that is designed to identify and assess regions of chemical space bounded by this important class of molecules. Using TSNaP, a family of tetrahydrofuran-containing pMLs are computationally assembled from pML inspired building blocks to provide a large collection of natural product-like virtual pMLs. By scoring functional group and volumetric overlap against their natural counterparts, a collection of compounds are prioritized for targeted synthesis. Using a modular and stereoselective synthetic approach, a library of polyketide-like macrolides are prepared to sample these unpopulated regions of pML chemical space. Validation of this TSNaP approach by screening this library against a panel of whole-cell biological assays, reveals hit rates exceeding those typically encountered in small molecule libraries. This study suggests that the TSNaP approach may be more broadly useful for the design of improved chemical libraries for drug discovery.

Antibacterial Marinopyrroles and Pseudilins Act as Protonophores.

Elucidating the mechanism of action (MoA) of antibacterial natural products is crucial to evaluating their potential as novel antibiotics. Marinopyrroles, pentachloropseudilin, and pentabromopseudilin are densely halogenated, hybrid pyrrole-phenol natural products with potent activity against Gram-positive bacterial pathogens like Staphylococcus aureus. However, the exact way they exert this antibacterial activity has not been established. In this study, we explore their structure-activity relationship, determine their spatial location in bacterial cells, and investigate their MoA. We show that the natural products share a common MoA based on membrane depolarization and dissipation of the proton motive force (PMF) that is essential for cell viability. The compounds show potent protonophore activity but do not appear to destroy the integrity of the cytoplasmic membrane via the formation of larger pores or interfere with the stability of the peptidoglycan sacculus. Thus, our current model for the antibacterial MoA of marinopyrrole, pentachloropseudilin, and pentabromopseudilin stipulates that the acidic compounds insert into the membrane and transport protons inside the cell. This MoA may explain many of the deleterious biological effects in mammalian cells, plants, phytoplankton, viruses, and protozoans that have been reported for these compounds.

Multivariate pattern analysis: a method and software to reveal, quantify, and visualize predictive association patterns in multicollinear data.

BACKGROUND: Strongly multicollinear covariates, such as those typically represented in metabolomics applications, represent a challenge for multivariate regression analysis. These challenges are commonly circumvented by reducing the number of covariates to a subset of linearly independent variables, but this strategy may lead to loss of resolution and thus produce models with poorer interpretative potential. The aim of this work was to implement and illustrate a method, multivariate pattern analysis (MVPA), which can handle multivariate covariates without compromising resolution or model quality. RESULTS: MVPA has been implemented in an open-source R package of the same name, mvpa. To facilitate the usage and interpretation of complex association patterns, mvpa has also been integrated into an R shiny app, mvpaShiny, which can be accessed on www.mvpashiny.org . MVPA utilizes a general projection algorithm that embraces a diversity of possible models. The method handles multicollinear and even linear dependent covariates. MVPA separates the variance in the data into orthogonal parts within the frame of a single joint model: one part describing the relations between covariates, outcome, and explanatory variables and another part describing the "net" predictive association pattern between outcome and explanatory variables. These patterns are visualized and interpreted in variance plots and plots for pattern analysis and ranking according to variable importance. Adjustment for a linear dependent covariate is performed in three steps. First, partial least squares regression with repeated Monte Carlo resampling is used to determine the number of predictive PLS components for a model relating the covariate to the outcome. Second, postprocessing of this PLS model by target projection provided a single component expressing the predictive association pattern between the outcome and the covariate. Third, the outcome and the explanatory variables were adjusted for the covariate by using the target score in the projection algorithm to obtain "net" data. We illustrate the main features of MVPA by investigating the partial mediation of a linearly dependent metabolomics descriptor on the association pattern between a measure of insulin resistance and lifestyle-related factors. CONCLUSIONS: Our method and implementation in R extend the range of possible analyses and visualizations that can be performed for complex multivariate data structures. The R packages are available on github.com/liningtonlab/mvpa and github.com/liningtonlab/mvpaShiny.

A Data Deposition Platform for Sharing Nuclear Magnetic Resonance Data.

Nuclear magnetic resonance (NMR) data are rarely deposited in open databases, leading to loss of critical scientific knowledge. Existing data reporting methods (images, tables, lists of values) contain less information than raw data and are poorly standardized. Together, these issues limit FAIR (findable, accessible, interoperable, reusable) access to these data, which in turn creates barriers for compound dereplication and the development of new data-driven discovery tools. Existing NMR databases either are not designed for natural products data or employ complex deposition interfaces that disincentivize deposition. Journals, including the Journal of Natural Products (JNP), are now requiring data submission as part of the publication process, creating the need for a streamlined, user-friendly mechanism to deposit and distribute NMR data.

Identification of the lipodepsipeptide selethramide encoded in a giant nonribosomal peptide synthetase from a Burkholderia bacterium.

Bacterial natural products have found many important industrial applications. Yet traditional discovery pipelines often prioritize individual natural product families despite the presence of multiple natural product biosynthetic gene clusters in each bacterial genome. Systematic characterization of talented strains is a means to expand the known natural product space. Here, we report genomics, epigenomics, and metabolomics studies of Burkholderia sp. FERM BP-3421, a soil isolate and known producer of antitumor spliceostatins. Its genome is composed of two chromosomes and two plasmids encoding at least 29 natural product families. Metabolomics studies showed that FERM BP-3421 also produces antifungal aminopyrrolnitrin and approved anticancer romidepsin. From the orphan metabolome features, we connected a lipopeptide of 1,928 Da to an 18-module nonribosomal peptide synthetase encoded as a single gene in chromosome 1. Isolation and structure elucidation led to the identification of selethramide which contains a repeating pattern of serine and leucine and is cyclized at the side chain oxygen of the one threonine residue at position 13. A (R)-3-hydroxybutyric acid moiety decorates the N-terminal serine. Initial attempts to obtain deletion mutants to probe the role of selethramide failed. After acquiring epigenome (methylome) data for FERM BP-3421, we employed a mimicry by methylation strategy that improved DNA transfer efficiency. Mutants defective in selethramide biosynthesis showed reduced surfactant activity and impaired swarming motility that could be chemically complemented with selethramide. This work unveils a lipopeptide that promotes surface motility, establishes improved DNA transfer efficiency, and sets the stage for continued natural product identification from a prolific strain.

MultiplexMS: A Mass Spectrometry-Based Multiplexing Strategy for Ultra-High-Throughput Analysis of Complex Mixtures.

High-throughput chemical analysis of natural product mixtures lags behind developments in genome sequencing technologies and laboratory automation, leading to a disconnect between library-scale chemical and biological profiling that limits new molecule discovery. Here, we report a new orthogonal sample multiplexing strategy that can increase mass spectrometry-based profiling up to 30-fold over traditional methods. Profiled pooled samples undergo subsequent computational deconvolution to reconstruct peak lists for each sample in the set. We validated this approach using in silico experiments and demonstrated a high assignment precision (>97%) for large, pooled samples (r = 30), particularly for infrequently occurring metabolites of relevance in drug discovery applications. Requiring only 5% of the previously required MS acquisition time, this approach was repeated in a recent biological activity profiling study on 925 natural product extracts, leading to the rediscovery of all previously reported bioactive metabolites. This new method is compatible with MS data from any instrument vendor and is supported by an open-source software package: https://github.com/liningtonlab/MultiplexMS.

Tolypocaibols: Antibacterial Lipopeptaibols from a Tolypocladium sp. Endophyte of the Marine Macroalga Spongomorpha arcta.

Two new lipopeptaibols, tolypocaibols A (1) and B (2), and the mixed NRPS-polyketide-shikimate natural product maximiscin [(P/M)-3)] were isolated from a Tolypocladium sp. fungal endophyte of the marine alga Spongomorpha arcta. Analysis of NMR and mass spectrometry data revealed the amino acid sequences of the lipopeptaibols, which both comprise 11 residues with a valinol C-terminus and a decanoyl acyl chain at the N-terminus. The configuration of the amino acids was determined by Marfey's analysis. Tolypocaibols A (1) and B (2) showed moderate, selective inhibition against Gram-positive and acid-fast bacterial strains, while maximiscin [(P/M)-3)] showed moderate, broad-spectrum antibiotic activity.

Comprehensive Metabolite Profiling of Two Edible Garcinia Species Based on UPLC-ESI-QTOF-MSE Coupled with Bioactivity Assays.

In China, the endemic species Garcinia yunnanensis and native Garcinia xanthochymus are known as edible and medicinal plants. However, a systematic metabolomic and bioactivity evaluation of different plant parts from both species is lacking. In this study, comprehensive investigations of 11 plant parts of G. yunnanensis and 10 of G. xanthochymus employing UPLC-ESI-QTOF-MSE-based metabolomic analysis in conjunction with three bioactivity assays were undertaken. A customized chemotaxonomic-based in-house library containing 6456 compounds was constructed and coupled to the Progenesis QI informatic platform for metabolite annotations. From these two species, a total of 235 constituents were characterized using multiple criteria. Differences in metabolite profiles between the plant parts within each species were uncovered using multivariate analysis. Based on orthogonal partial least-squares discriminant analysis (OPLS-DA), 23 markers were identified as highly differential metabolites from G. xanthochymus and 20 from G. yunnanensis. Comparative assessment of the biological assays revealed the activity variations among different plant parts. The seeds of both species and G. yunnanensis latex exhibited excellent cytotoxic and antibacterial activities, while G. xanthochymus roots and G. yunnanensis arils showed strong anti-inflammatory effects. S-plot analysis identified 26 potential biomarkers for the observed activities, including the known cytotoxic agent cycloxanthochymol and the anti-inflammatory compound garcimultiflorone B, which likely explains some of the potent observed bioactivity.

Interaction Metabolomics to Discover Synergists in Natural Product Mixtures.

Mass spectrometry metabolomics has become increasingly popular as an integral aspect of studies to identify active compounds from natural product mixtures. Classical metabolomics data analysis approaches do not consider the possibility that interactions (such as synergy) could occur between mixture components. With this study, we developed "interaction metabolomics" to overcome this limitation. The innovation of interaction metabolomics is the inclusion of compound interaction terms (CITs), which are calculated as the product of the intensities of each pair of features (detected ions) in the data matrix. Herein, we tested the utility of interaction metabolomics by spiking known concentrations of an antimicrobial compound (berberine) and a synergist (piperine) into a set of inactive matrices. We measured the antimicrobial activity for each of the resulting mixtures against Staphylococcus aureus and analyzed the mixtures with liquid chromatography coupled to high-resolution mass spectrometry. When the data set was processed without CITs (classical metabolomics), statistical analysis yielded a pattern of false positives. However, interaction metabolomics correctly identified berberine and piperine as the compounds responsible for the synergistic activity. To further validate the interaction metabolomics approach, we prepared mixtures from extracts of goldenseal (Hydrastis canadensis) and habañero pepper (Capsicum chinense) and correctly correlated synergistic activity of these mixtures to the combined action of berberine and several capsaicinoids. Our results demonstrate the utility of a conceptually new approach for identifying synergists in mixtures that may be useful for applications in natural products research and other research areas that require comprehensive mixture analysis.

Modulation of in Vitro SARS-CoV-2 Infection by Stephania tetrandra and Its Alkaloid Constituents.

Botanical natural products have been widely consumed for their purported usefulness against COVID-19. Here, six botanical species from multiple sources and 173 isolated natural product compounds were screened for blockade of wild-type (WT) SARS-CoV-2 infection in human 293T epithelial cells overexpressing ACE-2 and TMPRSS2 protease (293TAT). Antiviral activity was demonstrated by an extract from Stephania tetrandra. Extract fractionation, liquid chromatography-mass spectrometry (LC-MS), antiviral assays, and computational analyses revealed that the alkaloid fraction and purified alkaloids tetrandrine, fangchinoline, and cepharanthine inhibited WT SARS-CoV-2 infection. The alkaloids and alkaloid fraction also inhibited the delta variant of concern but not WT SARS-CoV-2 in VeroAT cells. Membrane permeability assays demonstrate that the alkaloids are biologically available, although fangchinoline showed lower permeability than tetrandrine. At high concentrations, the extract, alkaloid fractions, and pure alkaloids induced phospholipidosis in 293TAT cells and less so in VeroAT cells. Gene expression profiling during virus infection suggested that alkaloid fraction and tetrandrine displayed similar effects on cellular gene expression and pathways, while fangchinoline showed distinct effects on cells. Our study demonstrates a multifaceted approach to systematically investigate the diverse activities conferred by complex botanical mixtures, their cell-context specificity, and their pleiotropic effects on biological systems.

Collateral sensitivity profiling in drug-resistant Escherichia coli identifies natural products suppressing cephalosporin resistance.

The rapid emergence of antimicrobial resistance presents serious health challenges to the management of infectious diseases, a problem that is further exacerbated by slowing rates of antimicrobial drug discovery in recent years. The phenomenon of collateral sensitivity (CS), whereby resistance to one drug is accompanied by increased sensitivity to another, provides new opportunities to address both these challenges. Here, we present a high-throughput screening platform termed Collateral Sensitivity Profiling (CSP) to map the difference in bioactivity of large chemical libraries across 29 drug-resistant strains of E. coli. CSP screening of 80 commercial antimicrobials demonstrated multiple CS interactions. Further screening of a 6195-member natural product library revealed extensive CS relationships in nature. In particular, we report the isolation of known and new analogues of borrelidin A with potent CS activities against cephalosporin-resistant strains. Co-dosing ceftazidime with borrelidin A slows broader cephalosporin resistance with no recognizable resistance to borrelidin A itself.

Annotation of natural product compound families using molecular networking topology and structural similarity fingerprinting.

Spectral matching of MS2 fragmentation spectra has become a popular method for characterizing natural products libraries but identification remains challenging due to differences in MS2 fragmentation properties between instruments and the low coverage of current spectral reference libraries. To address this bottleneck we present Structural similarity Network Annotation Platform for Mass Spectrometry (SNAP-MS) which matches chemical similarity grouping in the Natural Products Atlas to grouping of mass spectrometry features from molecular networking. This approach assigns compound families to molecular networking subnetworks without the need for experimental or calculated reference spectra. We demonstrate SNAP-MS can accurately annotate subnetworks built from both reference spectra and an in-house microbial extract library, and correctly predict compound families from published molecular networks acquired on a range of MS instrumentation. Compound family annotations for the microbial extract library are validated by co-injection of standards or isolation and spectroscopic analysis. SNAP-MS is freely available at www.npatlas.org/discover/snapms .

Metabolomics Reveals a 26-Membered Macrolactone Produced by Endophytic Colletotrichum spp. from Alcatrazes Island, Brazil.

Comparative metabolomics analysis of nonphytotoxic endophytic Colletotrichum spp. isolated from Anthurium alcatrazense endemic to Alcatrazes island (Brazil) and phytopathogenic Colletotrichum spp. isolated from the mainland of Brazil revealed significant differences in chemical composition. Examination of endophytic Colletotrichum spp. from Alcatrazes island led to the discovery of a 26-member macrolactone, colletotrichumolide (1), containing a phosphatidyl choline side chain. Further examination of the phytopathogenic strains from the mainland identified a family of phytopathogenic metabolites not present in the nonpathogenic island-derived strains, suggesting that geographical isolation could influence the secondary metabolism of fungal strains.

High-throughput functional annotation of natural products by integrated activity profiling.

Determining mechanism of action (MOA) is one of the biggest challenges in natural products discovery. Here, we report a comprehensive platform that uses Similarity Network Fusion (SNF) to improve MOA predictions by integrating data from the cytological profiling high-content imaging platform and the gene expression platform Functional Signature Ontology, and pairs these data with untargeted metabolomics analysis for de novo bioactive compound discovery. The predictive value of the integrative approach was assessed using a library of target-annotated small molecules as benchmarks. Using Kolmogorov-Smirnov (KS) tests to compare in-class to out-of-class similarity, we found that SNF retains the ability to identify significant in-class similarity across a diverse set of target classes, and could find target classes not detectable in either platform alone. This confirmed that integration of expression-based and image-based phenotypes can accurately report on MOA. Furthermore, we integrated untargeted metabolomics of complex natural product fractions with the SNF network to map biological signatures to specific metabolites. Three examples are presented where SNF coupled with metabolomics was used to directly functionally characterize natural products and accelerate identification of bioactive metabolites, including the discovery of the azoxy-containing biaryl compounds parkamycins A and B. Our results support SNF integration of multiple phenotypic screening approaches along with untargeted metabolomics as a powerful approach for advancing natural products drug discovery.

Ferrocene-appended anthraquinone and coumarin as redox-active cytotoxins.

Appending of ferrocene (Fc) to biologically-active organic backbones can generate novel multi-functional species for targeting bacteria and cancer. In this work Fc was linked to coumarin and anthraquinone with the goal of harnessing the redox-active Fc centre to generate new compounds that exhibit cytoxicity through the generation of toxic reactive oxygen species (ROS). A Cu(I)-catalyzed azide-alkyne cycloaddition "click" reaction was used to connect the organic and Fc components via a triazole linker. Cyclic voltammetry shows that the Fc potentials are suitable for oxidation by biological hydrogen peroxide to give reactive ferrocenium (Fc+) species, which can then generate hydroxyl radicals. The ability of the compounds to generate hydroxyl radicals in the presence of hydrogen peroxide was shown directly using EPR spin-trapping experiments. Furthermore, in vitro studies in MCF-7 breast cancer cells show significant increases in ROS following incubation with the Fc-functionalized compounds. Screening for antibacterial activity produced negative results for all of the Fc compounds, consitent with low levels of hydrogen peroxide typically found in bacteria. By contrast, Fc-coumarin showed cytotoxicity against A549 lung cancer and SKOV3 ovarian cancer cell lines, whereas the parent compound was inactive. This is consistent both with the cytoxic potential of the Fc group and the elevated hydrogen peroxide levels found in many cancers. Interestingly, the anthraquinone compounds showed the opposite effect with the parent compounds showing modest activity against A549 cells, but the Fc compounds being inactive. This demonstrates other potential negative impacts of including Fc, such as significantly increased lipophilicity.

Potent Bactericidal Antimycobacterials Targeting the Chaperone ClpC1 Based on the Depsipeptide Natural Products Ecumicin and Ohmyungsamycin A.

Ohmyungsamycin A and ecumicin are structurally related cyclic depsipeptide natural products that possess activity against Mycobacterium tuberculosis (Mtb), the causative agent of tuberculosis (TB). Herein, we describe the design and synthesis of a library of analogues of these two natural products using an efficient solid-phase synthesis and late-stage macrolactamization strategy. Lead analogues possessed potent activity against Mtb in vitro (minimum inhibitory concentration 125-500 nM) and were shown to inhibit protein degradation by the mycobacterial ClpC1-ClpP1P2 protease with an associated enhancement of ClpC1 ATPase activity. The most promising analogue from the series exhibited rapid bactericidal killing activity against Mtb, capable of sterilizing cultures after 7 days, and retained bactericidal activity against hypoxic non-replicating Mtb. This natural product analogue was also active in an in vivo zebrafish model of infection.

NP Analyst: An Open Online Platform for Compound Activity Mapping.

Few tools exist in natural products discovery to integrate biological screening and untargeted mass spectrometry data at the library scale. Previously, we reported Compound Activity Mapping as a strategy for predicting compound bioactivity profiles directly from primary screening results on extract libraries. We now present NP Analyst, an open online platform for Compound Activity Mapping that accepts bioassay data of almost any type, and is compatible with mass spectrometry data from major instrument manufacturers via the mzML format. In addition, NP Analyst will accept processed mass spectrometry data from the MZmine 2 and GNPS open-source platforms, making it a versatile tool for integration with existing discovery workflows. We demonstrate the utility of this new tool for both the dereplication of known compounds and the discovery of novel bioactive natural products using a challenging low-resolution antimicrobial bioassay data set. This new platform is available at www.npanalyst.org.