Due to excellent separation capacity for complex mixtures of chemicals, comprehensive two-dimensional gas chromatography (GC × GC) is being utilized with increasing frequency for metabolomics analyses. This review describes recent advances in GC × GC method development for metabolomics, organismal sampling techniques compatible with GC × GC, metabolomic discoveries made using GC × GC, and recommendations and best practices for collecting and reporting GC × GC metabolomics data.

The study of chemical bioactivity in the rhizosphere has recently broadened to include microbial metabolites, and their roles in niche construction and competition via growth promotion, growth inhibition, and toxicity. Several prior studies have identified bacteria that produce volatile organic compounds (VOCs) with antifungal activities, indicating their potential use as biocontrol organisms to suppress phytopathogenic fungi and reduce agricultural losses. We sought to expand the roster of soil bacteria with known antifungal VOCs by testing bacterial isolates from wild and cultivated cranberry bog soils for VOCs that inhibit the growth of four common fungal and oomycete plant pathogens, and Trichoderma sp. Twenty one of the screened isolates inhibited the growth of at least one fungus by the production of VOCs, and isolates of Chromobacterium vaccinii had broad antifungal VOC activity, with growth inhibition over 90% for some fungi. Fungi exposed to C. vaccinii VOCs had extensive morphological abnormalities such as swollen hyphal cells, vacuolar depositions, and cell wall alterations. Quorum-insensitive cviR − mutants of C. vaccinii were significantly less fungistatic, indicating a role for quorum regulation in the production of antifungal VOCs. We collected and characterized VOCs from co-cultivation assays of Phoma sp. exposed to wild-type C. vaccinii MWU328, and its cviR − mutant using stir bar sorptive extraction and comprehensive two-dimensional gas chromatography-time-of-flight mass spectrometry (SBSE-GC×GC-TOFMS). We detected 53 VOCs that differ significantly in abundance between microbial cultures and media controls, including four candidate quorum-regulated fungistatic VOCs produced by C. vaccinii. Importantly, the metabolomes of the bacterial-fungal co-cultures were not the sum of the monoculture VOCs, an emergent property of their VOC-mediated interactions. These data suggest semiochemical feedback loops between microbes that have co-evolved for sensing and responding to exogenous VOCs.

Coccidioides immitis and C. posadasii are soil dwelling dimorphic fungi found in North and South America. Inhalation of aerosolized asexual conidia can result in asymptomatic, acute, or chronic respiratory infection. In the United States there are approximately 350,000 new infections per year. The Coccidioides genus is the only known fungal pathogen to make specialized parasitic spherules, which contain endospores that are released into the host upon spherule rupture. The molecular determinants involved in this key step of infection remain largely elusive as 49% of genes are hypothetical with unknown function. An attenuated mutant strain C. posadasii cts2/ ard1/ cts3 in which chitinase genes 2 and 3 were deleted was previously created for vaccine development. This strain does not complete endospore development, which prevents completion of the parasitic lifecycle. We sought to identify pathways active in the wild-type strain during spherule remodeling and endospore formation that have been affected by gene deletion in the mutant. We compared the transcriptome and volatile metabolome of the mutant cts2/ ard1/ cts3 to the wild-type C735. First, the global transcriptome was compared for both isolates using RNA sequencing. The raw reads were aligned to the reference genome using TOPHAT2 and analyzed using the Cufflinks package. Genes of interest were screened in an in vivo model using NanoString technology. Using solid phase microextraction (SPME) and comprehensive two-dimensional gas chromatography -time-of-flight mass spectrometry (GC × GC-TOFMS) volatile organic compounds (VOCs) were collected and analyzed. Our RNA-Seq analyses reveal approximately 280 significantly differentially regulated transcripts that are either absent or show opposite expression patterns in the mutant compared to the parent strain. This suggests that these genes are tied to networks impacted by deletion and may be critical for endospore development and/or spherule rupture in the wild-type strain. Of these genes, 14 were specific to the Coccidioides genus. We also found that the wild-type and mutant strains differed

Valley fever (coccidioidomycosis) is an endemic fungal pneumonia of the North and South American deserts. The causative agents of Valley fever are the dimorphic fungi Coccidioides immitis and C. posadasii, which grow as mycelia in the environment and as spherules within the lungs of vulnerable hosts. Current diagnostics for Valley fever are severely lacking due to poor sensitivity and invasiveness, contributing to a 23-day median time to diagnosis, and therefore, new diagnostic tools are needed. We are working toward the development of a breath-based diagnostic for coccidioidomycosis, and in this initial study, we characterized the volatile metabolomes (or volatilomes) of in vitro cultures of Coccidioides. Using solid-phase microextraction (SPME) and comprehensive two-dimensional gas chromatography coupled to time of flight mass spectrometry (GC×GC-TOFMS), we characterized the volatile organic compounds (VOCs) produced by six strains of each species during mycelial or spherule growth. We detected a total of 353 VOCs that were at least 2-fold more abundant in a Coccidioides culture than in medium controls and found that the volatile metabolome of Coccidioides is more dependent on the growth phase (spherules versus mycelia) than on the species. The volatile profiles of C. immitis and C. posadasii have strong similarities, indicating that a single suite of Valley fever breath biomarkers can be developed to detect both species. IMPORTANCE Coccidioidomycosis, or Valley fever, causes up to 30% of community-acquired pneumonias in highly populated areas of the U.S. desert southwest where the disease is endemic. The infection is difficult to diagnose by standard serological and histopathological methods, which delays appropriate treatment. Therefore, we are working toward the development of breath-based diagnostics for Valley fever. In this study, we characterized the volatile metabolomes (or volatilomes) of six strains each of Coccidioides immitis and C. posadasii, the dimorphic fungal species that cause Valley fever. By analyzing the volatilomes during the two modes of growth of the fungus—mycelia and spherules—we observed that the life cycle plays a significant role in the volatiles produced by Coccidioides. In contrast, we observed no significant differences in the C. immitis versus C. posadasii volatilomes. These data suggest that life cycle, rather than species, should guide the selection of putative biomarkers for a Valley fever breath test.

Missing data is a significant issue in metabolomics that is often neglected when conducting data preprocessing, particularly when it comes to imputation. This can have serious implications for downstream statistical analyses and lead to misleading or uninterpretable inferences. In this study, we aim to identify the primary types of missingness that affect untargeted metabolomics data and compare strategies for imputation using two real-world comprehensive two-dimensional gas chromatography (GC × GC) data sets. We also present these goals in the context of experimental replication whereby imputation is conducted in a within-replicate-based fashionthe first description and evaluation of this strategyand introduce an R package MetabImpute to carry out these analyses. Our results conclude that, in these two GC × GC data sets, missingness was most likely of the missing at-random (MAR) and missing not-at-random (MNAR) types as opposed to missing completely at-random (MCAR). Gibbs sampler imputation and Random Forest gave the best results when imputing MAR and MNAR compared against single-value imputation (zero, minimum, mean, median, and half-minimum) and other more sophisticated approaches (Bayesian principal component analysis and quantile regression imputation for left-censored data). When samples are replicated, within-replicate imputation approaches led to an increase in the reproducibility of peak quantification compared to imputation that ignores replication, suggesting that imputing with respect to replication may preserve potentially important features in downstream analyses for biomarker discovery.