Machine learning of spectra-property relationship for imperfect and small chemistry data.

Machine learning (ML) is causing profound changes to chemical research through its powerful statistical and mathematical methodological capabilities. However, the nature of chemistry experiments often sets very high hurdles to collect high-quality data that are deficiency free, contradicting the need of ML to learn from big data. Even worse, the black-box nature of most ML methods requires more abundant data to ensure good transferability. Herein, we combine physics-based spectral descriptors with a symbolic regression method to establish interpretable spectra-property relationship. Using the machine-learned mathematical formulas, we have predicted the adsorption energy and charge transfer of the CO-adsorbed Cu-based MOF systems from their infrared and Raman spectra. The explicit prediction models are robust, allowing them to be transferrable to small and low-quality dataset containing partial errors. Surprisingly, they can be used to identify and clean error data, which are common data scenarios in real experiments. Such robust learning protocol will significantly enhance the applicability of machine-learned spectroscopy for chemical science.

Divergent metabolome and proteome suggest functional independence of dual phloem transport systems in cucurbits.

Cucurbitaceous plants (cucurbits) have long been preferred models for studying phloem physiology. However, these species are unusual in that they possess two different phloem systems, one within the main vascular bundles [fascicular phloem (FP)] and another peripheral to the vascular bundles and scattered through stem and petiole cortex tissues [extrafascicular phloem (EFP)]. We have revisited the assumption that the sap released after shoot incision originates from the FP, and also investigated the long-standing question of why the sugar content of this sap is ~30-fold less than predicted for requirements of photosynthate delivery. Video microscopy and phloem labeling experiments unexpectedly reveal that FP very quickly becomes blocked upon cutting, whereas the extrafascicular phloem bleeds for extended periods. Thus, all cucurbit phloem sap studies to date have reported metabolite, protein, and RNA composition and transport in the relatively minor extrafascicular sieve tubes. Using tissue dissection and direct sampling of sieve tube contents, we show that FP in fact does contain up to 1 M sugars, in contrast to low-millimolar levels in the EFP. Moreover, major phloem proteins in sieve tubes of FP differ from those that predominate in the extrafascicular sap, and include several previously uncharacterized proteins with little or no homology to databases. The overall compositional differences of the two phloem systems strongly indicate functional isolation. On this basis, we propose that the fascicular phloem is largely responsible for sugar transport, whereas the extrafascicular phloem may function in signaling, defense, and transport of other metabolites.

Two Arabidopsis cytochrome P450 monooxygenases, CYP714A1 and CYP714A2, function redundantly in plant development through gibberellin deactivation.

The rice gene ELONGATED UPPERMOST INTERNODE1 (EUI1) encodes a P450 monooxygenase that epoxidizes gibberellins (GAs) in a deactivation reaction. The Arabidopsis genome contains a tandemly duplicated gene pair ELA1 (CYP714A1) and ELA2 (CYP714A2) that encode EUI homologs. In this work, we dissected the functions of the two proteins. ELA1 and ELA2 exhibited overlapping yet distinct gene expression patterns. We showed that while single mutants of ELA1 or ELA2 exhibited no obvious morphological phenotype, simultaneous elimination of ELA1 and ELA2 expression in ELA1-RNAi/ela2 resulted in increased biomass and enlarged organs. By contrast, transgenic plants constitutively expressing either ELA1 or ELA2 were dwarfed, similar to those overexpressing the rice EUI gene. We also discovered that overexpression of ELA1 resulted in a severe dwarf phenotype, while overexpression of ELA2 gave rise to a breeding-favored semi-dwarf phenotype in rice. Consistent with the phenotypes, we found that the ELA1-RNAi/ela2 plants increased amounts of biologically active GAs that were decreased in the internodes of transgenic rice with ELA1 and ELA2 overexpression. In contrast, the precursor GA(12) slightly accumulated in the transgenic rice, and GA(19) highly accumulated in the ELA2 overexpression rice. Taken together, our study strongly suggests that the two Arabidopsis EUI homologs subtly regulate plant growth most likely through catalyzing deactivation of bioactive GAs similar to rice EUI. The two P450s may also function in early stages of the GA biosynthetic pathway. Our results also suggest that ELA2 could be an excellent tool for molecular breeding for high yield potential in cereal crops.

A second target of the antimalarial and antibacterial agent fosmidomycin revealed by cellular metabolic profiling.

Antimicrobial drug resistance is an urgent problem in the control and treatment of many of the world's most serious infections, including Plasmodium falciparum malaria, tuberculosis, and healthcare-associated infections with Gram-negative bacteria. Because the non-mevalonate pathway of isoprenoid biosynthesis is essential in eubacteria and P. falciparum and this pathway is not present in humans, there is great interest in targeting the enzymes of non-mevalonate metabolism for antibacterial and antiparasitic drug development. Fosmidomycin is a broad-spectrum antimicrobial agent currently in clinical trials of combination therapies for the treatment of malaria. In vitro, fosmidomycin is known to inhibit the deoxyxylulose phosphate reductoisomerase (DXR) enzyme of isoprenoid biosynthesis from multiple pathogenic organisms. To define the in vivo metabolic response to fosmidomycin, we developed a novel mass spectrometry method to quantitate six metabolites of non-mevalonate isoprenoid metabolism from complex biological samples. Using this technique, we validate that the biological effects of fosmidomycin are mediated through blockade of de novo isoprenoid biosynthesis in both P. falciparum malaria parasites and Escherichia coli bacteria: in both organisms, metabolic profiling demonstrated a block of isoprenoid metabolism following fosmidomycin treatment, and growth inhibition due to fosmidomycin was rescued by media supplemented with isoprenoid metabolites. Isoprenoid metabolism proceeded through DXR even in the presence of fosmidomycin but was inhibited at the level of the downstream enzyme, methylerythritol phosphate cytidyltransferase (IspD). Overexpression of IspD in E. coli conferred fosmidomycin resistance, and fosmidomycin was found to inhibit IspD in vitro. This work has validated fosmidomycin as a biological reagent for blocking non-mevalonate isoprenoid metabolism and suggests a second in vivo target for fosmidomycin within isoprenoid biosynthesis, in two evolutionarily diverse pathogens.

Alternative isoleucine synthesis pathway in cyanobacterial species.

Cyanothece sp. ATCC 51142 is an aerobic N(2)-fixing and hydrogen-producing cyanobacterium. Isotopomer analysis of its amino acids revealed an identical labelling profile for leucine and isoleucine when Cyanothece 51142 was grown mixotrophically using 2-(13)C-labelled glycerol as the main carbon source. This indicated that Cyanothece 51142 employs the atypical alternative citramalate pathway for isoleucine synthesis, with pyruvate and acetyl-CoA as precursors. Utilization of the citramalate pathway was confirmed by an enzyme assay and LC-MS/MS analysis. Furthermore, the genome sequence of Cyanothece 51142 shows that the gene encoding the key enzyme (threonine ammonia-lyase) in the normal isoleucine pathway is missing. Instead, the cce_0248 gene in Cyanothece 51142 exhibits 53 % identity to the gene encoding citramalate synthase (CimA, GSU1798) from Geobacter sulfurreducens. Reverse-transcription PCR indicated that the cce_0248 gene is expressed and its transcriptional level is lower in medium with isoleucine than in isoleucine-free medium. Additionally, a blast search for citramalate synthase and threonine ammonia-lyase implies that this alternative isoleucine synthesis pathway may be present in other cyanobacteria, such as Cyanothece and Synechococcus. This suggests that the pathway is more widespread than originally thought, as previous identifications of the citramalate pathway are limited to mostly anaerobic bacteria or archaea. Furthermore, this discovery opens the possibility that such autrotrophic micro-organisms may be engineered for robust butanol and propanol production from 2-ketobutyrate, which is an intermediate in the isoleucine biosynthesis pathway.

Characterization of the central metabolic pathways in Thermoanaerobacter sp. strain X514 via isotopomer-assisted metabolite analysis.

Thermoanaerobacter sp. strain X514 has great potential in biotechnology due to its capacity to ferment a range of C(5) and C(6) sugars to ethanol and other metabolites under thermophilic conditions. This study investigated the central metabolism of strain X514 via (13)C-labeled tracer experiments using either glucose or pyruvate as both carbon and energy sources. X514 grew on minimal medium and thus contains complete biosynthesis pathways for all macromolecule building blocks. Based on genome annotation and isotopic analysis of amino acids, three observations can be obtained about the central metabolic pathways in X514. First, the oxidative pentose phosphate pathway in X514 is not functional, and the tricarboxylic acid cycle is incomplete under fermentative growth conditions. Second, X514 contains (Re)-type citrate synthase activity, although no gene homologous to the recently characterized (Re)-type citrate synthase of Clostridium kluyveri was found. Third, the isoleucine in X514 is derived from acetyl coenzyme A and pyruvate via the citramalate pathway rather than being synthesized from threonine via threonine ammonia-lyase. The functionality of the citramalate synthase gene (cimA [Teth514_1204]) has been confirmed by enzymatic activity assays, while the presence of intracellular citramalate has been detected by mass spectrometry. This study demonstrates the merits of combining (13)C-assisted metabolite analysis, enzyme assays, and metabolite detection not only to examine genome sequence annotations but also to discover novel enzyme activities.

A liquid chromatography-tandem mass spectrometry-based assay for indole-3-acetic acid-amido synthetase.

Indole-3-acetic acid (IAA) amide conjugates play an important role in balancing levels of free IAA in plant cells. The GH3 family of proteins conjugates free IAA with various amino acids. For example, auxin levels modulate expression of the Oryza sativa (rice) GH3-8 protein, which acts to prevent IAA accumulation by coupling the hormone to aspartate. To examine the kinetic properties of the enzyme, we developed a liquid chromatography-tandem mass spectrometry (LC-MS/MS) assay system. Bacterially expressed OsGH3-8 was purified to homogeneity and used to establish the assay system. Monitoring of the reaction confirms the reaction product as IAA-Asp and demonstrates that production of the conjugate increases proportionally with both time and enzyme amount. Steady-state kinetic analysis using the LC-MS/MS-based assay yields the following parameters: V/E(t)(IAA)=20.3 min(-1), K(m)(IAA)=123 microM, V/E(t)(ATP)=14.1 min(-1), K(m)(ATP)=50 microM, V/E(t)(Asp)=28.8 min(-1), K(m)(Asp)=1580 microM. This is the first assignment of kinetic values for any IAA-amido synthetase from plants. Compared with previously described LC- and thin-layer chromatography (TLC)-based assays, this LC-MS/MS method provides a robust and sensitive means for performing direct kinetic studies on a range of IAA-conjugating enzymes.

Application of stable isotope-assisted metabolomics for cell metabolism studies.

The applications of stable isotopes in metabolomics have facilitated the study of cell metabolisms. Stable isotope-assisted metabolomics requires: (1) properly designed tracer experiments; (2) stringent sampling and quenching protocols to minimize isotopic alternations; (3) efficient metabolite separations; (4) high resolution mass spectrometry to resolve overlapping peaks and background noises; and (5) data analysis methods and databases to decipher isotopic clusters over a broad m/z range (mass-to-charge ratio). This paper overviews mass spectrometry based techniques for precise determination of metabolites and their isotopologues. It also discusses applications of isotopic approaches to track substrate utilization, identify unknown metabolites and their chemical formulas, measure metabolite concentrations, determine putative metabolic pathways, and investigate microbial community populations and their carbon assimilation patterns. In addition, 13C-metabolite fingerprinting and metabolic models can be integrated to quantify carbon fluxes (enzyme reaction rates). The fluxome, in combination with other "omics" analyses, may give systems-level insights into regulatory mechanisms underlying gene functions. More importantly, 13C-tracer experiments significantly improve the potential of low-resolution gas chromatography-mass spectrometry (GC-MS) for broad-scope metabolism studies. We foresee the isotope-assisted metabolomics to be an indispensable tool in industrial biotechnology, environmental microbiology, and medical research.

A nucleotide metabolite controls stress-responsive gene expression and plant development.

Abiotic stress, such as drought and high salinity, activates a network of signaling cascades that lead to the expression of many stress-responsive genes in plants. The Arabidopsis FIERY1 (FRY1) protein is a negative regulator of stress and abscisic acid (ABA) signaling and exhibits both an inositol polyphosphatase and a 3',5'-bisphosphate nucleotidase activity in vitro. The FRY1 nucleotidase degrades the sulfation byproduct 3'-phosphoadenosine-5'-phosphate (PAP), yet its in vivo functions and particularly its roles in stress gene regulation remain unclear. Here we developed a LC-MS/MS method to quantitatively measure PAP levels in plants and investigated the roles of this nucleotidase activity in stress response and plant development. It was found that PAP level was tightly controlled in plants and did not accumulate to any significant level either under normal conditions or under NaCl, LiCl, cold, or ABA treatments. In contrast, high levels of PAP were detected in multiple mutant alleles of FRY1 but not in mutants of other FRY1 family members, indicating that FRY1 is the major enzyme that hydrolyzes PAP in vivo. By genetically reducing PAP levels in fry1 mutants either through overexpression of a yeast PAP nucleotidase or by generating a triple mutant of fry1 apk1 apk2 that is defective in the biosynthesis of the PAP precursor 3'-phosphoadenosine-5'-phosphosulfate (PAPS), we demonstrated that the developmental defects and superinduction of stress-responsive genes in fry1 mutants correlate with PAP accumulation in planta. We also found that the hypersensitive stress gene regulation in fry1 requires ABH1 but not ABI1, two other negative regulators in ABA signaling pathways. Unlike in yeast, however, FRY1 overexpression in Arabidopsis could not enhance salt tolerance. Taken together, our results demonstrate that PAP is critical for stress gene regulation and plant development, yet the FRY1 nucleotidase that catabolizes PAP may not be an in vivo salt toxicity target in Arabidopsis.

Selective chemical labeling reveals the genome-wide distribution of 5-hydroxymethylcytosine.

In contrast to 5-methylcytosine (5-mC), which has been studied extensively, little is known about 5-hydroxymethylcytosine (5-hmC), a recently identified epigenetic modification present in substantial amounts in certain mammalian cell types. Here we present a method for determining the genome-wide distribution of 5-hmC. We use the T4 bacteriophage ß-glucosyltransferase to transfer an engineered glucose moiety containing an azide group onto the hydroxyl group of 5-hmC. The azide group can be chemically modified with biotin for detection, affinity enrichment and sequencing of 5-hmC-containing DNA fragments in mammalian genomes. Using this method, we demonstrate that 5-hmC is present in human cell lines beyond those previously recognized. We also find a gene expression level-dependent enrichment of intragenic 5-hmC in mouse cerebellum and an age-dependent acquisition of this modification in specific gene bodies linked to neurodegenerative disorders.

Mammalian cytochrome P450 enzymes catalyze the phenol-coupling step in endogenous morphine biosynthesis.

A cytochrome P450 (P450) enzyme in porcine liver that catalyzed the phenol-coupling reaction of the substrate (R)-reticuline to salutaridine was previously purified to homogeneity (Amann, T., Roos, P. H., Huh, H., and Zenk, M. H. (1995) Heterocycles 40, 425-440). This reaction was found to be catalyzed by human P450s 2D6 and 3A4 in the presence of (R)-reticuline and NADPH to yield not a single product, but rather (-)-isoboldine, (-)-corytuberine, (+)-pallidine, and salutaridine, the para-ortho coupled established precursor of morphine in the poppy plant and most likely also in mammals. (S)-Reticuline, a substrate of both P450 enzymes, yielded the phenol-coupled alkaloids (+)-isoboldine, (+)-corytuberine, (-)-pallidine, and sinoacutine; none of these serve as a morphine precursor. Catalytic efficiencies were similar for P450 2D6 and P450 3A4 in the presence of cytochrome b(5) with (R)-reticuline as substrate. The mechanism of phenol coupling is not yet established; however, we favor a single cycle of iron oxidation to yield salutaridine and the three other alkaloids from (R)-reticuline. The total yield of salutaridine formed can supply the 10 nm concentration of morphine found in human neuroblastoma cell cultures and in brain tissues of mice.