ASTROBIOLOGY

Volume 20, Number 10, 2020 Review Article

ª Mary Ann Liebert, Inc.
DOI: 10.1089/ast.2019.2135

Metabolomics as an Emerging Tool in the Search

for Astrobiologically Relevant Biomarkers

Lauren Seyler,1,2 Elizabeth B. Kujawinski,1 Armando Azua-Bustos,3,4 Michael D. Lee,2,5

Jeffrey Marlow,2,6,7 Scott M. Perl,8,9 and Henderson James Cleaves II2,10,11,12

Abstract

It is now routinely possible to sequence and recover microbial genomes from environmental samples. To the
degree it is feasible to assign transcriptional and translational functions to these genomes, it should be possible,
in principle, to largely understand the complete molecular inputs and outputs of a microbial community.
However, gene-based tools alone are presently insufficient to describe the full suite of chemical reactions and
small molecules that compose a living cell. Metabolomic tools have developed quickly and now enable rapid
detection and identification of small molecules within biological and environmental samples. The convergence
of these technologies will soon facilitate the detection of novel enzymatic activities, novel organisms, and
potentially extraterrestrial life-forms on solar system bodies. This review explores the methodological problems
and scientific opportunities facing researchers who hope to apply metabolomic methods in astrobiology-related
fields, and how present challenges might be overcome. Key Words: Biomarkers—Biosignatures—Metabolism—
Molecular fossils—Biogeochemistry. Astrobiology 20, 1251–1261.

1. Introduction Laboratory experiments and field observations have shown

that small organic compounds (e.g., amino acids, carbox-

L inking discrete molecular compounds to biological

activity in the search for extraterrestrial life has been a
priority since the time of the Viking lander (Klein et al., 1976).
ylic acids, and nucleotides), including molecules known to
be essential for terrestrial microorganisms (Miller, 1953;
Eschenmoser and Loewenthal 1992; Cleaves, 2012), can be
Mars has been the main target of this effort by virtue of its synthesized abiotically. The ability to distinguish biotic from
proximity to the Earth, and the perceived similarity of Mars to abiotic molecules is therefore crucial in the search for life on
Earth has heightened the interest, although the following dis- other worlds. Ideally, given that life seems to be a phenom-
cussion can be related to other extraterrestrial environments as enon that combinatorially generates and then explores huge
well. For over 40 years, attempts have been made to measure chemical spaces, these searches need to be as open as pos-
discrete organic compounds on Mars (Biemann et al., 1977; sible (Cleland, 2019).
Eigenbrode et al., 2018). Future missions to Mars and the outer Many advances in the rapid analysis of metabolites (small
planets will include mass spectrometers capable of at least organic molecules, generally <1000 Da) for medical diag-
constraining mass distributions of small organic compounds nostics are now being used in the environmental sciences,
(Sephton et al., 2018). Recently, the Curiosity rover confirmed enabling the correlation of environmental genetic information
the presence of organic compounds in Mars’ shallow subsur- with biochemical data (Abram, 2015; Cao et al., 2019).
face (Eigenbrode et al., 2018). The identities of these com- However, various phenomena currently complicate intentions
pounds are not yet known, and their source is likewise unclear. to link genomic and small-molecule data for astrobiological

1
Department of Marine Chemistry and Geochemistry, Woods Hole Oceanographic Institution, Woods Hole, Massachusetts, USA.
2
Blue Marble Space Institute of Science, Seattle, Washington, USA.
3
Department of Planetology and Habitability, Centro de Astrobiologı́a (CSIC-INTA), Madrid, Spain.
4
Instituto de Ciencias Biomédicas, Facultad de Ciencias de la Salud, Universidad Autónoma de Chile, Santiago, Chile.
5
Exobiology Branch, NASA Ames Research Center, Moffett Field, California, USA.
6
Department of Organismic and Evolutionary Biology, Harvard University, Cambridge, Massachusetts, USA.
7
Department of Biology, Boston University, Boston, Massachusetts, USA.
8
Geological and Planetary Sciences, California Institute of Technology/NASA Jet Propulsion Laboratory, Pasadena, California, USA.
9
Mineral Sciences, Los Angeles Natural History Museum, Los Angeles, California, USA.
10
Earth-Life Science Institute, Tokyo Institute of Technology, Tokyo, Japan.
11
School of Natural Sciences, Institute for Advanced Study, Princeton, New Jersey, USA.
12
Geographical Research Laboratory, Carnegie Institution of Washington.

1251
1252 SEYLER ET AL.

exploration. For example, the geological conditions that favor bolomic methodologies and propose how they could be
DNA survival in the environment are fairly well understood. tailored for astrobiological purposes.
The maximum lifetime of useful DNA sequence data does
not exceed one million years (1 Ma) even under the most 2. The ‘‘Omics’’ Revolution and the New Frontier
favorable circumstances on Earth (Allentoft et al., 2012; of the Metabolome
Hofreiter et al., 2015), although claims have been made for
Biology is currently undergoing a rapid and expansive
the recovery of much older (*100 Ma) sequences (Inagaki
‘‘omics’’ revolution (Kuska, 1998), which focuses on mul-
et al., 2005). Thus, the potential for linking genomic infor-
tiple classes of living systems’ molecular components and
mation to molecular biomarkers as a function of time is only
integrates different types of data to obtain a more compre-
possible for the very recent paleontological record even on
hensive understanding of biological systems as a whole. The
Earth, where life is abundant, and samples are easy to obtain.
list of omics disciplines is ever-increasing. Several omics
In the context of the detection of potential extraterrestrial
disciplines encapsulate steps in the central dogma of mo-
biochemistries (Giri et al., 2018), there are several pos-
lecular biology: DNA is transcribed to RNA, which is
sibilities for carbon-based life: (1) such life-forms use
translated to protein (Crick, 1958). Discoveries at each level
conventional nucleic acids as a repository of genetic infor-
of this information-transfer can help elucidate an organism’s
mation, in which case conventional sequencing methodolo-
holistic metabolic capabilities (Fig. 1).
gies may be useful for solar system exploration (Mojarro
With regard to astrobiology, these omics techniques have
et al., 2017); (2) such life-forms use alternative nucleic acids
provided insights into the survival capabilities of microbes
as a repository for genetic information (Cleaves et al., 2015)
living in Earth environments that resemble potentially
in which case novel sequencing methods would be required
habitable environments on other worlds (Rothschild and
(Carr et al., 2016); (3) such life-forms use non-nucleic acid
Mancinelli, 2001). As stated above, however, extraterrestrial
polymers as genetic molecules (Sowerby and Petersen,
life may not use DNA or RNA to store genetic information,
2002); or (4) such life-forms do not depend on conventional
and even if it does, the lability of nucleic acids provides a
notions of molecule-based genetic inheritance (Segré et al.,
narrow temporal detection window following an organism’s
2000). It should be noted that even if the first possibility is
demise. It may therefore be helpful to seek out compounds
true, it may be very unlikely that alien biochemistries would
or physical phenomena that may serve as longer lasting,
use the same genetic code or coded amino acids (or indeed
process-agnostic biomarkers on other worlds (Chan et al.,
that nucleic acids and proteins will be universally paired
2019). Analyzing the suite of small metabolic products of
biochemical features), have cognate genes or protein folds,
living organisms—the metabolome—may help constrain
or use similar metabolic pathways. All of these phenomena
and direct the search for such biomarkers.
may be deeply contingent aspects of terrestrial biological
Metabolomic investigations aim to characterize the com-
evolution.
plexity of biological molecules in samples with minimum
Even if conventional or nonconventional nucleic acids
preparation. This can be a daunting task considering that
prove too fragile to survive in extraterrestrial environments
cells may contain a mixture of many thousands of metabo-
(or to be made in the first place), many other types of bio-
lites, over a large range of abundances (Zamboni et al.,
molecules could provide information about past biology
2015). Only recently have analytical technologies been
(Briggs and Summons, 2014). In the four cases mentioned
specifically developed for this purpose, and many challenges
above, extraterrestrial metabolism could be based on similar
remain. Mass spectrometry (MS), in particular, now enables
or dissimilar pathway transformations and/or compounds,
simultaneous detection of thousands of putative metabolites
and novel metrics for distinguishing living versus degraded
from minimal amounts of sample (Bowen et al., 2011; Patti
versus abiogenic organic compounds would be required.
Metabolomics, the study of the intermediates and end
products of metabolism, or metabolites, produced by living
cells, is rapidly revolutionizing the understanding of bio-
chemistry (Aldridge and Rhee, 2014). Environmental me-
tabolomics, which uses the techniques of metabolomics to
describe the interaction between organisms and their envi-
ronment, offers promising tools for life detection in the
Solar System beyond the Earth. The techniques used to
identify metabolites in modern microbial communities can
assist in detecting quantifiable substances whose presence
are indicative of ancient or extinct life, that is, biomarkers,
depending on the preservation of the original biomarkers.
Furthermore, inasmuch as organic compound-based bio-
chemistry will almost certainly involve controlled transfor-
mations of specific, relatively low-molecular-weight
compounds (Hoehler et al., 2018), metabolomic techniques
offer an unbiased ( Johnson et al., 2018; Chan et al., 2019)
way to search for alien biochemical components, indepen-
dent of known biological pathways that are encoded by
genes and carried out by protein-based catalysts (enzymes). FIG. 1. The central dogma of molecular biology and its
Here we review some contemporary environmental meta- corresponding omics disciplines.
METABOLOMICS IN THE SEARCH FOR BIOMARKERS 1253

et al., 2012; Zamboni et al., 2015). MS-based metabolomics organisms adapted to similar extreme environments may
have been used to demonstrate the existence of complete have many shared metabolites (Poli et al., 2017), which
pathways not detected by genome annotation (Tang et al., could act as unique signatures of that environment and its
2009), to demonstrate the function of theoretical pathways associated adaptive metabolisms.
(Peyraud et al., 2009), to discover novel metabolic pathways A reliance upon shared metabolic products results in a
(Fürch et al., 2009; Liu et al., 2016), and to describe novel conundrum for astrobiological assessment. Truly unbiased
biomarkers associated with disease (Sabatine et al., 2005; biosignature searches (i.e., independent of genes, proteins,
Wang-Sattler et al., 2012). Thus, even if we are ignorant of a or pathways) may either yield no shared metabolites that can
particular extraterrestrial biology’s ‘‘genetic’’ system, we be used for eventual targeted analyses, or there may be
may be able to infer something about its metabolic cap- simple, highly conserved metabolites associated with central
abilities via the distributions of its detected metabolites. metabolism (housekeeping pathways that are common
across all domains of life) that can yield many downstream
compositional possibilities. For example, low-molecular-
3. Metabolomics and the Search for Biomarkers
weight organic acids act as intermediates in multiple central
Although extraterrestrial metabolisms could be wildly metabolism pathways, including amino acid metabolism,
different from terrestrial ones, the search for biomarkers on nucleic acid synthesis, and carbon fixation, but these com-
other worlds might begin with the identification of mole- pounds can also be produced abiotically through geochem-
cules that are diagnostic of metabolic pathways in analog ical reactions (McDermott et al., 2015). These outcomes
environments on Earth. It might be expected that there make it difficult to distinguish between biotic versus abiotic
would be some overlap of at least some ‘‘switchboard’’ compounds. In addition, many, if not most, primary me-
compounds such as those found in the various incarnations tabolites are ions at physiological and environmental pH
of the tricarboxylic acid or other carbon-fixation cycles values, which facilitates their retention inside cells (Bar-
(Braakman and Smith, 2012a, b). Even et al., 2011), and many metabolites, especially the
Few extraterrestrial bodies in the Solar System host en- most abundant ones (e.g., amino acids, cofactors, and nu-
vironments are considered ‘‘clement’’ on Earth, although cleotides), are common across a wide variety of organisms
environments with conditions overlapping those considered (Peregrı́n-Alvarez et al., 2009). Secondary metabolites, in
‘‘extreme’’ on Earth may abound. Generally, environments contrast, are often more species-specific, and often more
are considered ‘‘extreme’’ on Earth when they are charac- likely to diffuse into the surrounding environment (Breitling
terized by conditions (e.g., of pH, temperature, salinity, et al., 2013; Covington et al., 2016).
pressure) that are inhospitable to humans (Rothschild and The molecules present in microbially inhabited environ-
Mancinelli, 2001). Metabolomic profiling of microbial mental samples are likely to be a mix of two types: (1)
communities in such environments can assist in under- compounds that are being actively metabolized and syn-
standing metabolite preservation as well as microbial com- thesized, and are therefore relevant to extant microbes and
munity adaptation to extreme conditions and response to their communities, and (2) compounds that are recalcitrant
environmental change (Blanchowitz et al., 2019). However, to microbial reuse, and may be ancient remnants of com-
characterizing organisms, genomes, and metabolomes from munities. This simple distinction may allow for the identi-
extreme environments is challenging due to limited genomic fication of active biology, as an abundance of chemically
data (Hedlund et al., 2014) and the difficulty of interpreting labile compounds may suggest the existence of extant or-
fragmented sedimentological and geobiological data ganisms. Active microbial communities might be expected
(Hodgson et al., 2018). to effectively maintain low steady-state concentrations of
No standard environmental metabolomic protocol is ideal certain metabolites, and compounds that can accumulate and
for applying metabolomics to all extreme environments; dominate geochemical analyses may do so because the
protocols are usually both sample and environment specific microbiome cannot consume them (Kleber, 2010).
(Riekeberg and Powers, 2017). In general, extremophilic Organisms construct themselves from many types of or-
communities and their mesophilic counterparts differ with ganic compounds, which are out of thermodynamic equilib-
respect to the number of microbial species present and their rium with the environment (Ornes, 2017). When organisms
phylogenetic diversity (Mesbah and Wiegel, 2012). Extreme die, their components remineralize. Thermodynamics (as well
settings typically contain a relatively low diversity of mi- as the kinetics of the associated reactions), rather than genetic
crobial species, all adapted to the dominant ecological stress capability, will ultimately constrain the ability of micro-
of their environment (Stahl et al., 1985; Liu et al., 2014; biomes to effect environmental organic transformations.
Sharp et al., 2014; Poli et al., 2017). While species diversity Thus, one might expect the geochemical context to determine
is low, these environments can have extremely high abun- the distribution of organic compounds in environmental set-
dances of selected species (Cowan et al., 2002; Brazelton tings, rather than the composition of the microbial community
et al., 2006; Kubo et al., 2011). The metabolome of a (Louca et al., 2019). For example, long-term geological
community of interest can be compared with that of the burial of organic carbon (e.g., petroleum, kerogen, coal) re-
dominant species observed in situ (if it can be cultured in quires a lack of oxidants during burial (Berner, 2003). Un-
isolation), or the metabolome can be searched for interme- derstanding the genomic, metabolic, and geochemical
diates or products suggestive of the metabolic pathway(s) processes that allow biological carbon to evade reminer-
under investigation (Zamboni et al., 2015). This allows for alization is therefore important. Specific organic molecules
novel species to be associated together based on common could be markers of the limits of remineralization under
metabolites and metabolomes, and can also aid in describing specific Eh/pH conditions. Assuming extraterrestrial organ-
key metabolic pathways (Maifiah et al., 2017). Distinct isms share some common metabolites with terrestrial ones,
1254 SEYLER ET AL.

pulse-chase experiments, in which microbial communities are et al., 2015). Detection and identification of metabolites rest
exposed to a ‘‘pulse’’ of stable isotope-labeled compound on the choice of mass spectrometer used in these studies.
followed by a ‘‘chase’’ of the same compound in an unlabeled Targeted metabolomic studies emphasize sensitive quanti-
form, could provide a way to distinguish between active and fication over comprehensive analysis and thus often use
recalcitrant compounds, provided the organisms in the sample quadrupole-based mass detection or time-of-flight mass
metabolize the pulse feed rapidly and efficiently. Lability of spectrometers. These mass spectrometers are sensitive to
organic compounds can be also assessed with analytical changes in ion abundance, but have limited mass resolution
techniques discussed further on in this review, including nu- for detecting molecules with small changes in molecular
clear magnetic resonance (NMR) spectroscopy (Knicker, mass. Untargeted metabolomic studies, in contrast, often
2004) and Fourier transform ion cyclotron resonance mass leverage high mass resolution to deconvolute complex
spectrometry (FTICRMS) (D’Andrilli et al., 2015). mixtures and thus rely on Fourier transform-based mass
analyses such as Orbitrap or FTICRMS. In each of these
techniques, structural identification rests on tandem frag-
4. An Overview of Metabolomic Methods
mentation spectra of selected ions, while quantification,
Metabolomic experiments generally follow one of two where possible, requires authentic standards.
approaches: targeted or untargeted. In targeted metabolomic Annotation of metabolite ‘‘features’’ (ions with unique
studies, a suite of metabolites are preselected for analysis and mass/charge ratios and retention times) in MS-based un-
then quantified by using standards. In untargeted studies, as targeted metabolomic data begins through comparison of
many metabolites as possible are measured from samples observed mass with metabolite databases. Two types of such
without preconceived notions of which compounds may be databases exist (Longnecker et al., 2015). The first type
present (Patti et al., 2012; Baig et al., 2016). Untargeted contains chemical information (e.g., formula, mass, struc-
approaches would likely be most valuable to the discovery of ture, and physicochemical properties) for any compound
novel biomarkers in astrobiological applications. regardless of source. Examples include PubChem (Bolton
All metabolomic methods seek to detect and quantify et al., 2008) and ChemSpider (Pence and Williams, 2010),
metabolite profiles within organisms or communities. No one which include both predicted and experimentally observed
method is ideal for all applications or compound classes, and compounds, and databases such as the Human Metabolome
each has its advantages and disadvantages (Aldridge and Database (Wishart et al., 2007) and METLIN (Smith et al.,
Rhee, 2014; Kido Soule et al., 2015). NMR-based methods 2005) that include experimental metadata for compound
can structurally elucidate and quantify metabolites with contextualization. The best current example of the latter
minimal sample preparation, but require large samples due to type of database is MetaboLights (Steinbeck et al., 2012;
lower sensitivity than MS-based methods. (In addition, the Haug et al., 2013), which contains only known naturally
size and power requirements of NMR instruments likely occurring metabolites (as opposed to predicted ones).
preclude the possibility of including NMR on a spacecraft.) Mass measurement alone cannot provide metabolite in-
In contrast, MS-based methods can detect thousands of formation beyond molecular formula (Schrimpe-Rutledge
metabolites over a large abundance range, but are limited in et al., 2016), masking the enormous diversity of structural
their ability to provide structural information for annotation isomers (Meringer et al., 2013). Combined accurate mass
(Aldridge and Rhee, 2014). In brief, NMR-based methods and retention time may still not be sufficient to unambigu-
are ideal for structural characterization, while MS-based ously identify compounds of interest (Baig et al., 2016).
methods are better suited to detecting large suites of me- Tandem MS (e.g., MS/MS), which results in unique frag-
tabolites. A summary of methods is provided in Fig. 2. mentation patterns for each compound, is therefore required
MS-based methods require ionization of metabolites be- to assign metabolite identity with more confidence for
fore analysis. The most common ionization methods are compounds larger than 50 Da (Hoffman et al., 2014). Im-
electrospray ionization (ESI), chemical ionization (CI), portantly, each detected feature may not represent a distinct
matrix-assisted laser desorption ionization (MALDI), and metabolite, resulting in overestimates of the number of un-
desorption electrospray ionization (DESI). ESI ionizes known compounds present in a given sample (Dunn et al.,
molecules that occur as ions within aqueous or polar sol- 2013). Naturally occurring isotopologues may be present,
vent solutions, and thus is used to couple liquid chroma- metabolites may ionize as more than one adduct, or me-
tography with mass spectrometry (LC/MS). LC/MS allows tabolites may fragment or form noncovalent interactions
for the detection of the greatest number of metabolites with other compounds upon entering the mass spectrometer
( Jonsson et al., 2005; Schrimpe-Rutledge et al., 2016), and (Zamboni et al., 2015). Accurate identification of metabo-
can resolve metabolites in their native state from com- lites must therefore discriminate between metabolites of
plex mixtures containing thousands of analytes ranging over different nominal masses, metabolites with the same nomi-
10,000-fold differences in abundance (Aldridge and Rhee, nal mass but different molecular formulae, and metabolites
2014). This approach, however, does not ionize neutral with the same monoisotopic masses but different chemical
(nonionic) or volatile metabolites, such as hydrocarbons. structures (e.g., enantiomers and structural isomers) (Dunn
Instead CI such as that in gas chromatography is used to et al., 2013). A number of bioinformatic tools have recently
introduce these molecules into a mass spectrometer. Large been developed to help automate metabolite peak identifi-
biopolymers such as proteins or polysaccharides can be cation, with methods for peak picking and grouping related
ionized by MALDI or DESI before MS (MALDI-MS) ion peaks, nonlinear retention time alignment, relative
(Edward and Kennedy, 2005). quantification, metabolite identification, and statistical anal-
Many metabolites are polar and water-soluble, and thus ysis (Patti et al., 2012; Lynn et al., 2014; Schrimpe-Rutledge
amenable to LC/MS (Aldridge and Rhee, 2014; Kido Soule et al., 2016). Furthermore, the Metabolomics Standards
FIG. 2. A summary of metabolomic methods.

1255
1256 SEYLER ET AL.

Initiative has developed a protocol for assigning confidence resolution spot sizes (down to 50 nm), but fragmentation
to metabolite identification (Sumner et al., 2007). results in low-mass secondary ions from *1 to 300 amu
Metabolomics provides opportunities for discovery, and (Fletcher and Vickerman, 2010). This is an extremely useful
challenges in data processing and management, similar to tool for studying the isotopic composition of the compo-
those experienced during the development of high- nents of individual cells (Marlow et al., 2014; Kopf et al.,
throughput DNA and RNA sequencing technologies. Like 2015), and when coupled with knowledge of relevant res-
high-throughput sequence data, untargeted metabolomic data ervoirs and fractionation factors, this information can point
sets are massive (gigabytes per sample) and far too complex toward metabolic pathways and/or interspecies interactions
to be analyzed manually. When simple mass-to-charge ratios (Orphan et al., 2009; Pasulka et al., 2018) by tracking rel-
are searched against metabolite databases, there are often ative abundances of C, H, N, O, and S stable isotopes.
surprisingly few hits (Zamboni et al., 2015). The situation is However, the identity of the molecules that possess these
not much better for MS/MS spectral matching: of the over 60 isotopic ratios (e.g., lipids, proteins, and metabolites) cannot
million molecules in the PubChem database, only 20,000 be obtained due to the fragmentation accompanying high-
include MS/MS spectral data ( Johnson and Lange, 2015). energy ion beams and the inefficiency of ionization (typi-
The distribution of fragment ions within MS/MS spectra cally only about one in 100,000 molecules is ionized by the
depends on instrument parameters (Herman et al., 2017), primary ion beam; Fletcher and Vickerman, 2010).
limiting the general applicability of these libraries within a To better analyze larger molecules more relevant to me-
given instrumental configuration. Consequently, a few per- tabolomics, softer ionization methods are needed. Several
cent, at most, of spectral features in an untargeted metabo- variants exist, but MALDI (Caprioli et al., 1997) is most
lomics experiment can presently be annotated (da Silva commonly used. Sample surfaces are coated with a chemical
et al., 2015), underscoring the large gaps in current under- matrix that absorbs laser light at a given wavelength, and the
standing of metabolism (Kind et al., 2009; Patti et al., 2012). sputtered material is sent through a mass spectrometer, de-
As more metabolite features are discovered and character- termining the fragment sizes and inferred composition of the
ized, the number of compounds that may serve as potential sample entrained within the matrix. MALDI approaches
biomarkers on other worlds will likely grow. generally result in relatively large analytical spot sizes (e.g.,
Metabolomic researchers are rising to meet analytical 50 mm; Cornett et al., 2007), but recent advances in instru-
challenges with improved databases, bioinformatic software, ment configuration and matrix application have enabled
and crowdsourcing platforms such as Global Natural Pro- 1.4 mm spatial resolution in a study of metabolite, lipid, and
ducts Social Molecular Networking (GNPS) (Wang et al., peptide distributions (Kompauer et al., 2016), approaching
2016). For example, fairly exhaustive databases of potential the scale of microbial cells.
molecular isomers can be generated, blanketing a given As secondary ion mass spectrometry- and MALDI-based
chemical formula space in silico (e.g., Meringer and Cleaves, techniques converge in terms of spatial resolution, submicron-
2017). It is increasingly possible to accurately simulate MS/ scale analyses of a wide range of molecular weight com-
MS fragmentation spectra (e.g., Bauer and Grimme, 2016; pounds with resolution amenable to isotopic studies may
Ruttkies et al., 2016) although comparisons with laboratory become possible. Maintaining the spatial arrangements of
data still suffer from the analytical dependencies described geobiological samples is critical for their interpretation, as
above and the increasing complexity of fragmentation pat- distributions of key parameters (e.g., lipid or metabolite type,
terns within large molecules. There is also still room for isotopic composition) with respect to pore space, conduits,
significant improvements in sample preparation standardi- mineral type, or texture may offer valuable information on
zation and analytical and data reduction methods (e.g., biogenicity. With this additional context, astrobiologists will
www.metabolomicsworkbench.org/about/index.php). gain new perspectives on observed features, enabling nu-
anced, environment-specific interpretation of metabolomic
data sets. While the deployment of these tools for space-
5. The Promise and Challenges of Spatially
based missions is infeasible currently due to size and power
Resolved Metabolomics
constraints, sample-return missions will undoubtedly benefit
The colocalization of metabolites, with each other and from their use.
with other observables such as cell-like morphologies, is
particularly important in an astrobiological context. MS
6. Integrating Metabolomics with Other Omics Data
imaging compiles mass spectra from individual locations
within a sample and rasters across an area of interest to build To understand the function of metabolites within cellular
a map of spectral features and their corresponding parent processes, it is necessary to develop metabolomic analysis
molecules (Dunham et al., 2017). The contextual informa- approaches that integrate data across different omics data
tion that can be gained from such determinations can give sets ( Johnson et al., 2016)—an effort that has some overlap
otherwise uninteresting molecules new meaning if present in with the field of systems biology (Kitano, 2002). A desired
clustered or out-of-equilibrium arrangements. MS imaging outcome of this type of data integration is the character-
allows unique evaluation of distinct spatial aspects of a ization of unknown metabolites within the context of known
sample’s chemical distribution to better infer the processes genes and proteins. As stated previously, it cannot be as-
involved in compound formation and/or preservation. sumed that alien life uses nucleic acids or the same genetic
The most salient trade-off in MS imaging is between code as terrestrial life, or even that protein folding could be
spatial resolution and the size of detected molecules. Na- predicted from an alien gene sequence. Indeed, predicting
noscale secondary ion mass spectrometry uses a high-energy the structure and by extension the function of terrestrial
primary ion (e.g., Cs+ or O-) beam to provide high spatial proteins remains an ongoing problem (for a complete
METABOLOMICS IN THE SEARCH FOR BIOMARKERS 1257

review, see Lee et al., 2017). Metagenomic analyses must develop more complete pictures of the metabolism of or-
also make assumptions about homology between organisms ganisms or communities of interest (Noecker et al., 2016;
that may or may not be true (Prakash and Taylor, 2012). As Huan et al., 2017; Nagler et al., 2018; Witting et al., 2018).
the integration of omics techniques enables the development In vivo labeling of metabolite pools with 13C, 15N, and/or
of a deeper and more fundamental understanding of terres- deuterium (2H) can be used to improve metabolite annota-
trial cellular metabolism (Ritchie et al., 2015), it may also tion accuracy (Birkenmeyer et al., 2005; Hegeman et al.,
provide clues to the functions of unknown metabolites as- 2007) and for metabolic network-wide elucidation of path-
sociated with metabolisms of astrobiological interest. ways (Kikuchi et al., 2004; Winder et al., 2011; Creek et al.,
There is a growing gap, however, between the large (and 2012). The combined use of omics techniques thus allows
rapidly increasing) amount of omics data and researchers for an expansion of our understanding of metabolism on
with skills to process and interpret these data (Barone et al., Earth, improving our ability to search for analogous bio-
2017). The lack of standard laboratory and data processing logical processes on other worlds.
procedures across databases and studies is a critical prob-
lem, as is the elucidation of the function of unannotated 7. Conclusions
‘‘hypothetical’’ genes that often dominate environmental
omics annotation pipelines ( Jiao et al., 2017). Like astrobiology, metabolomics is highly interdisciplin-
Ideally, metagenomic data would provide information ary and often requires collaboration across different research
about which compounds a microbial cohort is capable of areas to complete all stages of an experimental workflow.
biosynthesizing, and then MS analysis would be able to For those who are new to metabolomics or who are not
corroborate that a physiologically relevant subset of such specialized in certain informatics approaches, several ini-
compounds is in fact present. For example, identification of tiatives have been developed, including resources through
sterols based on the detection of unique mass species in the US National Institutes of Health (NIH) Common Fund
geological specimens has allowed estimation of divergence Metabolomics Program, and the Coordination of Standards
times in the tree of life as determined by phylogenetic in Metabolomics (COSMOS) (Salek et al., 2015; Johnson
comparison. Using such techniques, Brocks et al. (1999) et al., 2016). Both vendor-provided and open-access MS
estimated the earliest eukaryotes to have appeared at 2.7 Ga analysis software offer user-friendly options for beginners,
(although these data have since been questioned; French as well as advanced tools for experts ( Johnson et al., 2016).
et al., 2015), and the Ediacaran fossil Dickinsonia was Metabolomics has much to offer to the advancement of
identified as one of the first animals based on MS-derived astrobiology, particularly in cases in which noncanonical
molecular fossil data (Bobrovskiy et al., 2018). modes of information storage and metabolism may be
However, there are multiple challenges in interpreting present. Applications of metabolomic-derived tools will un-
combined environmental genomic and molecular data. doubtedly increase as data collection, storage, and analysis
Genomic data may be amplified in ways that skew inter- techniques improve. The integration of metabolomics with
pretation of the activity and abundance of organisms (Kim other omics data will also provide insight into the functions
and Bae, 2011; Shakya et al., 2013; Solonenko et al., 2013; of newly discovered metabolites and their connectivity in
Quince et al., 2017). Furthermore, it cannot be assumed that metabolic networks. These advances will help astrobiologists
the abundance of a given gene cohort is directly proportional search for biochemical signs of life on other worlds.
to the abundance of specific molecules (Chan et al., 2010).
Author Disclosure Statement
Gene annotation, including accurate identification of start
and stop codons, can be challenging (Pauli et al., 2014; No competing financial interests exist.
Mattick and Rinn, 2015; Borriss et al., 2017). The ability to
annotate genomes has greatly improved as a consequence of Funding Information
the increase of annotation databases such as NCBI’s RefSeq
This study was partially supported by the ELSI Origins
(O’Leary et al., 2016), UniProtKB (Bateman et al., 2017),
Network (EON), which is supported by a grant from the
NCBI’s Clusters of Orthologous Groups (COGs) (Galperin
John Templeton Foundation. The opinions expressed in this
et al., 2015), the Kyoto Encyclopedia of Genes and Gen-
publication are those of the authors and do not necessarily
omes (KEGG) (Kanehisa et al., 2016), Gene Ontology (GO)
reflect the views of the John Templeton Foundation. This
(Huerta-Cepas et al., 2017), and Protein Families (Pfam)
work was partially supported by a JSPS KAKENHI Grant-
(Finn et al., 2016). The growth of these databases has en-
in-Aid for Scientific Research on Innovative Areas ‘‘Hadean
abled vast improvements in the interpretation of environ-
Bioscience,’’ grant number JP26106003, and also partially
mental metagenomic, transcriptomic, and proteomic data
supported by Project ‘‘icyMARS,’’ funded by the European
(e.g., as reviewed in Konopka and Wilkins, 2012; Haider
Research Council, ERC Starting Grant No. 307496. A.A-B
and Pal, 2013; Franzosa et al., 2016).
thanks the contribution from the Project ‘‘MarsFirstWater,’’
Programs such as ModelSEED (Henry et al., 2010) and
funded by the European Research Council, ERC Con-
Computation of Microbial Ecosystems in Time and Space
solidator Grant No. 818602 and the HFSP Project UVE-
(COMETS) (Harcombe et al., 2014) have been developed to
nergy RGY0066/2018.
reconstruct metabolic community networks from multiple
genomes, and can be used to identify key metabolic
‘‘nodes’’ within these networks (Henry et al., 2016) and References
describe metabolic networks over time (Granger et al., Abram F (2015) Systems-based approaches to unravel multi-
2016). Researchers are developing data analysis pipelines species microbial community functioning. Comput Struct
that integrate metabolomes with other omics data sets to Biotechnol J 13:24–32.
1258 SEYLER ET AL.

Aldridge BB and Rhee KY (2014) Microbial metabolomics: natural products and microbial biotransformation products
innovation, application, insight. Curr Opin Microbiol 19: from the human microbiota. mSystems 4:e00387-19.
90–96. Caprioli RM, Farmer TB, and Gile J (1997) Molecular imaging
Allentoft ME, Collins M, Harker D, et al. (2012) The half-life of biological samples: localization of peptides and proteins
of DNA in bone: measuring decay kinetics in 158 dated using MALDI-TOF MS. Anal Chem 69:4751–4760.
fossils. Proc R Soc Lond B 279:4724. Carr CE, Mojarro A, Tani J, et al. (2016) Advancing the search
Baig F, Pechlaner R, and Mayr M (2016) Caveats of untargeted for extra-terrestrial genomes. In Aerospace Conference,
metabolomics for biomarker discovery. J Am Coll Cardiol IEEE, Big Sky, Montana, USA pp 1–15.
68:1294–1296. Chan EKF, Rowe HC, Hansen BG, et al. (2010) The complex
Bar-Even A, Noor E, Flamholz A, et al. (2011) Hydrophobicity genetic architecture of the metabolome. PLoS Genet 6:e1001198.
and charge shape cellular metabolite concentrations. PLoS Chan MA, Hinman NW, Potter-McIntyre SL, et al. (2019).
Comput Biol 7:e1002166. Deciphering biosignatures in planetary contexts. Astrobiology
Barone L, Williams J, and Micklos D (2017) Unmet needs for 19, 1075–1102.
analyzing biological big data: a survey of 704 NSF principal Cleaves HJ (2012) Prebiotic chemistry: what we know, what we
investigators. PLoS Comput Biol 13:1–8. don’t. Evol Educ Outreach 5:342.
Bateman A, Martin MJ, O’Donovan C, et al. (2017) UniProt: Cleaves HJ, Meringer M, and Goodwin J (2015) 227 Views of
the universal protein knowledgebase. Nucleic Acids Res 45: RNA: is RNA unique in its chemical isomer space? Astro-
D158–D169. biology 15:538–558.
Bauer CA and Grimme S (2016) How to compute electron Cleland CE (2019) The Quest for a Universal Theory of Life.
ionization mass spectra from first principles. J Phys Chem A Cambridge University Press, Cambridge, England.
120:3755–3766. Cornett DS, Reyzer ML, Chaurand P, et al. (2007) MALDI
Berner RA (2003) The long-term carbon cycle, fossil fuels and imaging mass spectrometry: molecular snapshots of bio-
atmospheric composition. Nature 426:323–326. chemical systems. Nat Methods 4:828–833.
Biemann K, Oro J, Toulmin III P, et al. (1977) The search for Covington BC, McLean JA, and Bachmann BO (2016) Com-
organic substances and inorganic volatile compounds in the parative mass spectrometry-based metabolomics strategies
surface of Mars. J Geophys Res 82:4641–4658. for the investigation of microbial secondary metabolites. Nat
Birkenmeyer C, Luedemann A, Wagner C, et al. (2005) Meta- Prod Rep 34:6–24.
bolome analysis: the potential of in vivo labeling with stable Cowan DA, Russell NJ, Mamais A, et al. (2002) Antarctic dry
isotopes for metabolite profiling. TRENDS Biotechnol 23:28–33. valley mineral soils contain unexpectedly high levels of mi-
Blachowicz A, Chiang AJ, Elsaesser A, et al. (2019) Proteomic crobial biomass. Extremophiles 6:431–436.
and metabolomic characteristics of extremophilic fungi under Creek DJ, Chokkathukalam A, Jankevics A, et al. (2012) Stable
simulated Mars conditions. Front Microbiol 10:1013. isotope-assisted metabolomics for network-wide metabolic
Bobrovskiy I, Hope JM, Ivantsov A, et al. (2018) Ancient pathway elucidation. Anal Chem 84:8442–8447.
steroids establish the Ediacaran fossil Dickinsonia as one of Crick FHC (1958) ‘‘On Protein Synthesis’’. In Symposia of the
the earliest animals. Science 361:1246–1249. Society for Experimental Biology, Number XII: The Biolo-
Bolton EE, Cavanaugh M, Clark K, et al. (2008) PubChem: gical Replication of Macromolecules, edited by FK Sanders,
integrated platform of small molecules and biological activ- Cambridge University Press, Cambridge, pp. 138–163.
ities. Ann Rep Comput Chem 4:217–241. D’Andrilli J, Cooper WT, Foreman CM, et al. (2015) An ul-
Borriss R, Danchin A, Harwood C, et al. (2018) Bacillus sub- trahigh-resolution mass spectrometry index to estimate nat-
tilis, the model Gram-positive bacterium: 20 years of anno- ural organic matter lability. Rapid Commun Mass Spectrom
tation refinement. Microb Biotechnol 11:3–17. 29:2385–2401.
Bowen BP, Fischer CR, Baran R, et al. (2011) Improved ge- da Silva RR, Dorrestein PC, and Quinn RA (2015) Illuminating
nome annotation through untargeted detection of pathway- the dark matter in metabolomics. Proc Natl Acad Sci U S A
specific metabolites. BMC Genomics 12:S6. 112:12549–12550.
Braakman R and Smith E (2012a) The compositional and Dunham SJB, Ellis JF, Li B, et al. (2017) Mass spectrometry
evolutionary logic of metabolism. Phys Biol 10:011001. imaging of complex microbial communities. Acc Chem Res
Braakman R and Smith E (2012b) The emergence and early 50:96–104.
evolution of biological carbon-fixation. PLoS Comp Biol 8: Dunn WB, Erban A, Weber, RJM, et al. (2013) Mass appeal:
e1002455. metabolite identification in mass spectrometry-focused un-
Brazelton WJ, Schrenk MO, Kelley DS, et al. (2006) Methane- targeted metabolomics. Metabolomics 9:S44–S66.
and sulfur-metabolizing microbial communities dominate the Edward JL and Kennedy RT (2005) Metabolomic analysis of
lost city hydrothermal field ecosystem. Appl Environ Mi- eukaryotic tissue and prokaryotes using negative mode MALDI
crobiol 72:6257–6270. time-of-flight mass spectrometry. Anal Chem 77:2201–2209.
Breitling R, Ceniceros A, Jankevics A, et al. (2013) Metabo- Eigenbrode JL, Summons RE, Steele A, et al. (2018). Organic
lomics for secondary metabolite research. Metabolites 3: matter preserved in 3-billion-year-old mudstones at Gale
1076–1083. crater, Mars. Science 360:1096–1101.
Briggs DE and Summons RE (2014) Ancient biomolecules: Eschenmoser A and Loewenthal E (1992) Chemistry of poten-
their origins, fossilization, and role in revealing the history of tially prebiological natural products. Chem Soc Rev 21:1–16.
life. BioEssays 36:482–490. Finn RD, Coggill P, Eberhardt RY, et al. (2016) The Pfam
Brocks JJ, Logan GA, Buick R, et al. (1999) Archean molecular protein families database: towards a more sustainable future.
fossils and the early rise of eukaryotes. Science 285:1033– Nucleic Acids Res 44:D279–D285.
1036. Fletcher JS and Vickerman JC (2010) A new SIMS paradigm
Cao L, Shcherbin E, and Mohimani H (2019) A metabolome- for 2D and 3D molecular imaging of bio-systems. Anal
and metagenome-wide association network reveals microbial Bioanal Chem 396:85–104.
METABOLOMICS IN THE SEARCH FOR BIOMARKERS 1259

Franzosa EA, Hsu T, Sirota-Madi A, et al. (2016) Sequencing Inagaki F, Okada H, Tsapin AI, et al. (2005) Microbial survival:
and beyond: integrating molecular ‘omics for microbial the paleome: a sedimentary genetic record of past microbial
community profiling. Nat Rev Microbiol 13:360–372. communities. Astrobiology 5:141–153.
French KL, Hallmann C, Hope JM, et al. (2015) Reappraisal of Jiao D, Han W, and Ye Y (2017) Functional association pre-
hydrocarbon biomarkers in Archean rocks. Proc Natl Acad diction by community profiling. Methods 129:8–17.
Sci U S A 112:5915–5920. Johnson CH, Ivanisevic J, and Siuzdak G (2016) Metabolomics:
Fürch T, Preusse M, Tomasch J, et al. (2009) Metabolic fluxes beyond biomarkers and towards mechanisms. Nat Rev Mol
in the central carbon metabolism of Dinoroseobacter shibae Cell Biol 17:451–459.
and Phaeobacter gallaeciensis, two members of the marine Johnson SR and Lange RM (2015) Open-access metabolomics
Roseobacter clade. BMC Microbiol 9:209. databases for natural product research: present capabilities
Galperin MY, Makarova KS, Wolf YI, et al. (2015) Expanded and future potential. Front Bioeng Biotechnol 3:22.
microbial genome coverage and improved protein family an- Johnson SS, Anslyn EV, Graham HV, et al. (2018) Finger-
notation in the COG database. Nucleic Acids Res 43:D261– printing non-terran biosignatures. Astrobiology 18:915–922.
D269. Jonsson P, Bruce SJ, Moritz T, et al. (2005) Extraction, inter-
Giri C, Jia TZ, Cleaves II, et al. (2018) Life-detection tech- pretation and validation of information for comparing sam-
nologies for the next two decades. arXiv arXiv:1810.06026. ples in metabolic LC/MS data sets. Analyst 130:701–707.
Granger BR, Chang Y-C, Wang Y, et al. (2016) Visualization Kanehisa M, Sato Y, Kawashima M, et al. (2016) KEGG as a
of metabolic interaction networks in microbial communities reference resource for gene and protein annotation. Nucleic
using VisANT 5.0. PLoS Comp Biol 12:e1004875. Acids Res 44: D457–D462.
Haider S and Pal R (2013) Integrated analysis of transcriptomic Kido Soule ME, Longnecker K, Johnson WM, et al. (2015)
and proteomic data. Curr Genomics 14:91–110. Environmental metabolomics: analytical strategies. Mar
Harcombe WR, Riehl WJ, Dokovski I, et al. (2014) Metabolic Chem 177 (Part 2):374–387.
resource allocation in individual microbes determines ecosys- Kikuchi J, Shinozaki K, Hirayama T (2004) Stable isotope la-
tem interactions and spatial dynamics. Cell Rep 7:1104–1115. beling of Arabidopsis thaliana for an NMR-based metabo-
Haug K, Salek RM, Conesa P, et al. (2013) MetaboLights—an lomics approach Plant Cell Physiol 45:1099–1104.
open access general-purpose repository for metabolomics stud- Kim K-H and Bae J-W (2011) Amplification methods bias
ies and associated meta-data. Nucleic Acids Res 41:D781–D786. metagenomic libraries of uncultured single-stranded and
Hedlund BP, Dodsworth JA, Murugapiran SK, et al (2014) double-stranded DNA viruses. Appl Env Microbiol 77:7663–
Impact of single-cell genomics and metagenomics on the 7668.
emerging view of extremophile ‘‘microbial dark matter.’’ Kind T, Scholz M, and Fiehn O (2009) How large is the me-
Exremophiles 18:865–875. tabolome? A critical analysis of data exchange practices in
Hegeman AD, Schulte CF, Cui Q, et al. (2007) Stable isotope chemistry. PLoS One4:e5440.
assisted assignment of elemental compositions for metabo- Kitano H (2002) Systems biology: a brief overview. Science
lomics. Anal Chem 79:6912–6921. 295:1662–1664.
Henry CS, Bernstein HC, Weisenhorn P, et al. (2016) Microbial Kleber M (2010) What is recalcitrant soil organic matter? En-
community metabolic modeling: a community data-driven viron Chem 7:320–332.
network reconstruction. Cell Phys. DOI: 10.1002/jcp.25428 Klein HP, Horowitz NH, Levin GV, et al. (1976) The Viking
Henry CS, DeJongh M, Best AA, et al. (2010) High-throughput biological investigation: preliminary results. Science 194:99–
generation, optimization, and analysis of genome-scale met- 105.
abolic models. Nat Biotechnol 28:977–982. Knicker H (2004) Stabilization of N-compounds in soil and
Herman S, Khoonsari PE, Aftab O, et al. (2017) Mass spec- organic-matter-rich sediments—what is the difference? Mar
trometry based metabolomics for in vitro systems pharma- Chem 92:167–195.
cology: pitfalls, challenges, and computational solutions. Kompauer M, Heiles S, and Spengler B (2016) Atmospheric
Metabolomics 137:79. pressure MALDI mass spectrometry imaging of tissues and
Hodgson DM, Bernhardt A, Clare MA, et al. (2018) Grand cells at 1.4-mm lateral resolution. Nat Methods 14:90–96.
challenges (and great opportunities) in sedimentology, stra- Konopka A and Wilkins M (2012) Application of meta-
tigraphy, and diagenesis research. Front Earth Sci. https:// transcriptomics and -proteomics to analysis of in situ physi-
doi.org/10.3389/feart.2018.00173. ological state. Front Microbiol 3:184.
Hoehler TM, Som SM, and Kiang NY (2018) Life’s Require- Kopf SH, McGlynn SE, Green-Saxena A, Guan Y, Newman
ments. In Handbook of Exoplanets, edited by HJ DeegJuan, DK, and Orphan VJ (2015) Heavy water and 15N labeling
and A Belmonte, Cham, Switzerland, pp. 1–22. with NanoSIMS analysis reveals growth rate-dependent me-
Hoffman T, Krug D, Hüttel S, et al. (2014) Improving natural tabolic heterogeneity in chemostats. Env Microbiol 17:2542–
products identification through targeted LC-MS/MS in an 2556.
untargeted secondary metabolomics workflow. Anal Chem Kubo K, Knittel K, Amann R, et al. (2011) Sulfur-metabolizing
86:10780–10788. bacterial populations in microbial mats of the Nakabusa hot
Hofreiter M, Paijmans JL, Goodchild H, et al. (2015) The future spring, Japan. Sys Appl Microbiol 34:293–302.
of ancient DNA: technical advances and conceptual shifts. Kuska B (1998) Beer, bethesda, and biology: how ‘genomics’
BioEssays 37:284–293. came into being. J Natl Cancer Inst 90:93.
Huan T, Forsberg EM, Rinehart D, et al. (2017) Systems bi- Lee J, Freddolino PL, and Zhang Y (2017) Ab initio protein
ology guided by XCMS online metabolomics. Nat Methods structure prediction. In From Protein Structure to Function
14:461–462. with Bioinformatics, edited by DJ Rigden, Springer Press,
Huerta-Cepas J, Forslund K, Coelho LP, et al. (2017) Fast pp. 3–35.
genome-wide functional annotation through orthology as- Liu J, Hua Z-S, Chen L-X, et al. (2014) Correlating microbial
signment by EggNOG-Mapper. Mol Biol Evol 34:2115–2122. diversity patterns with geochemistry in an extreme and
1260 SEYLER ET AL.

heterogeneous environment of mine tailings. Appl Environ with BONCAT and nanoSIMS-based methods. Environ Mi-
Microbiol 80:3677–3686. crobiol 20:671–692.
Liu X, Lu YF, Guan X, et al. (2016) Characterizing novel Patti GJ, Yanes O, and Siuzdak G (2012) Metabolomics: the
metabolic pathways of melatonin receptor agonist agomela- apogee of the omics trilogy. Nat Rev Mol Cell Biol 13:263–
tine using metabolomic approaches. Biochem Pharmacol 269.
109:70–82. Pauli A, Valen E, and Schier AF (2014) Identifying
Longnecker K, Futrelle J, Coburn E, et al (2015) Environmental (non-)coding RNAs and small peptides: challenges and
metabolomics: databases and tools for data analysis. Mar opportunities. BioEssays 37:103–112.
Chem 177 (Part 2):366–373. Pence HE and Williams A (2010) ChemSpider: an online
Louca S, Scranton MI, Taylor GT, et al. (2019) Circumventing chemical information resource. J Chem Ed 87:1123–1124.
kinetics in biogeochemical modeling. Proc Natl Acad Sci Peregrı́n-Alvarez JM, Sanford C, and Parkinson J (2009) The
U S A 116:11329–11338. conservation and evolutionary modularity of metabolism.
Lynn K-S, Cheng M-L, Chen Y-R, et al. (2014) Metabolite Genome Biol 10:R63.
identification for mass spectrometry-based metabolomics Peyraud R, Kiefer P, Christen P, et al. (2009) Demonstration of
using multiple types of correlated ion information. Anal the ethylmalonyl-CoA pathway by using 13C metabolomics.
Chem 87:2143–2151. Proc Natl Acad Sci U S A 106:4846–4851.
Maifiah MH, Creek DJ, Nation RL, et al. (2017) Untargeted Poli A, Finore I, Romano I, et al. (2017) Microbial diversity in
metabolomics analysis reveals key pathways responsible for extreme marine habitats and their biomolecules. Micro-
the synergistic killing of colistin and doripenem combination organisms 5:25.
against Acinetobacter baumannii. Sci Rep 7:45527. Prakash T and Taylor TD (2012) Functional assignment of
Marlow JJ, Steele JA, Ziebis W, et al. (2014) Carbonate-hosted metagenomics data: challenges and applications. Brief
methanotrophy represents an unrecognized methane sink in Bioinform 13:711–717.
the deep sea. Nat Commun 5:5094. Quince C, Walker AW, Simpson JT, et al. (2017) Shotgun
Mattick JS and Rinn JL (2015) Discovery and annotation of metagenomics, from sampling to analysis. Nature Biotechnol
long noncoding RNAs. Nat Struct Mol Bio 22:5–7. 35:833–844.
McDermott JM, Seewald JS, German CR, et al. (2015) Path- Riekeberg E and Powers R (2017) New frontiers in metabo-
ways for abiotic organic synthesis at submarine hydrothermal lomics: from measurement to insight. F1000Res 6:1148.
fields. Proc Nat Acad Sci U S A 112:7668–7672. Ritchie MD, Holzinger ER, Li R, et al. (2015) Methods of
Meringer M and Cleaves HJ (2017) Exploring astrobiology integrating data to uncover genotype-phenotype interactions.
using in silico molecular structure generation. Proc R Soc Nat Rev Genetics 16:85–97.
Lond A 375:20160344. Rothschild L and Mancinelli R (2001) Life in extreme envi-
Meringer MH, Cleaves HJ, and Freeland SJ (2013) Beyond ronments. Nature 409:1092–1101.
terrestrial biology: charting the chemical universe of a-amino Ruttkies C, Schymanski EL, Wolf S, et al. (2016) MetFrag
acid structures. J Chem Inf Model 53:2851–2862. relaunched: incorporating strategies beyond in silico frag-
Mesbah NM and Wiegel J (2012) Life under multiple extreme mentation. J Cheminform 8:3.
conditions: diversity and physiology of the halophilic alka- Sabatine MS, Liu E, Morrow DA, et al. (2005) Metabolomic
lithermophiles. Appl Env Microbiol 78:4074–4082. identification of novel biomarkers of myocardial ischemia.
Miller SL (1953) A production of amino acids under possible Circulation 112:3868–3875.
primitive Earth conditions. Science 117:528–529. Salek RM, Neumann S, Schober D, et al. (2015) COordination of
Mojarro A, Ruvkun G, Zuber MT, et al. (2017) Nucleic acid Standards in MetabOlomicS (COSMOS): facilitating integrated
extraction from synthetic Mars analog soils for in situ life metabolomics data access. Metabolomics 11:1587–1597.
detection. Astrobiology 17:747–760. Schrimpe-Rutledge AC, Codreanu SG, Sherrod SD, et al.
Nagler M, Nägele T, Gilli C, et al. (2018) Eco-metabolomics (2016) Untargeted metabolomics strategies—challenges and
and metabolic modeling: making the leap from model sys- emerging directions. J Am Soc Mass Spectrom 27:1897–1905.
tems in the lab to native populations in the field. Front Plant Segré D, Ben-Eli D, and Lancet D (2000) Compositional ge-
Sci 9:1556. nomes: prebiotic information transfer in mutually catalytic
Noecker C, Eng A, Srinivasan S, et al. (2016) Metabolic model- noncovalent assemblies. Proc Nat Acad Sci 97:4112–4117.
based integration of microbiome taxonomic and metabolomic Sephton MA, Hunter Waite J, and Brockwell TG (2018) Searching
profiles elucidates mechanistic links between ecological and for life with mass spectrometry. Astron Geophys 59:3–23.
metabolic variation. mSystems 1:e00013-15. Shakya M, Quince C, Campbell JH, et al. (2013) Comparative
O’Leary NA, Wright MW, Brister JR, et al. (2016) Reference metagenomic and rRNA microbial diversity characterization
sequence (RefSeq) database at NCBI: current status, taxo- using archaeal and bacterial synthetic communities. Env
nomic expansion, and functional annotation. Nucleic Acids Microbiol 15:1882–1899.
Res 44:D733–D745. Sharp CE, Brady AL, Sharp GH, et al. (2014) Humboldt’s spa:
Ornes S (2017) Core concept: how nonequilibrium thermody- microbial diversity is controlled by temperature in geother-
namics speaks to the mystery of life. Proc Nat Acad Sci USA mal environments. ISME J 8:1166–1174.
114:423–424. Smith CA, O’Maille G, Want EJ, et al. (2005) METLIN: a
Orphan VJ, Turk KA, Green AM, et al. (2009) Patterns of 15N metabolite mass spectral database. Ther Drug Monit 27:747–
assimilation and growth of methanotrophic ANME-2 archaea 751.
and sulfate-reducing bacteria within structured syntrophic Solonenko SA, Ignacio-Espinoza JC, Alberti A, et al. (2013)
consortia revealed by FISH-SIMS. Environ Microbiol 11: Sequencing platform and library preparation choices impact
1777–1791. viral metagenomes. BMC Genomics 14:320.
Pasulka AL, Thamatrakoln K, Kopf SH, et al. (2018) Inter- Sowerby SJ and Petersen GB (2002) Life before RNA. Astro-
rogating marine virus-host interactions and elemental transfer biology 2:231–239.
METABOLOMICS IN THE SEARCH FOR BIOMARKERS 1261

Stahl DA, Lane DJ, Olsen GJ, et al. (1985) Characteriza- Zamboni N, Saghatelian A, and Patti GJ (2015) Defining the
tion of a Yellowstone hot spring microbial community by metabolome: size, flux, and regulation. Mol Cell 58:699–706.
5S rRNA sequences. Appl Environ Microbiol 49:1379–
1384.
Steinbeck C, Conesa P, Haug K, et al. (2012) MetaboLights: Address correspondence to:
towards a new COSMOS of metabolomics data management. Lauren Seyler
Metabolomics 8:757–760. Department of Marine Chemistry and Geochemistry
Sumner LW, Amberg A, Barrett D, et al. (2007) Proposed Woods Hole Oceanographic Institution
minimum reporting standards for chemical analysis. Meta- 86 Water Street
bolomics 3:211–221. Woods Hole, MA 02543
Tang YJ, Yi S, Zhuang W-Q, et al. (2009) Investigation of USA
carbon metabolism in ‘‘Dehalococcoides ethenogenes’’ strain
195 by use of isotopomer and transcriptomic analyses. J E-mail: [email protected]
Bacteriol 191:5224–5231.
Wang M, Carver JJ, Phelan VV, et al. (2016) Sharing and Submitted 7 July 2019
community curation of mass spectrometry data with Global Accepted 1 May 2020
Natural Products Social Molecular Networking. Nature Bio- Associate Editor: Lewis Dartnell
tech 34:828.
Wang-Sattler R, Yu Z, Herder C, et al. (2012) Novel bio-
markers for pre-diabetes identified by metabolomics. Mol Sys Abbreviations Used
Biol 8:615. CI ¼ chemical ionization
Winder CL, Dunn WB, and Goodacre R (2011) TARDIS-based DESI ¼ desorption electrospray ionization
microbial metabolomics: time and relative differences in ESI ¼ electrospray ionization
systems. Trends Microbiol 19:315–322. FTICRMS ¼ Fourier transform ion cyclotron resonance
Wishart DS, Tzur D, Knox C, et al. (2007) HMDB: the Hu- mass spectrometry
man Metabolome Database. Nucleic Acids Res 35:D521– LC/MS ¼ liquid chromatography with mass
D526. spectrometry
Witting M, Hastings J, Rodriguez N, et al. (2018) Modeling MS ¼ mass spectrometry
meets metabolomics—the WormJam consensus model as MALDI ¼ matrix-assisted laser desorption ionization
basis for metabolic studies in the model organism Cae- NMR ¼ nuclear magnetic resonance
norhabditis elegans. Front Mol Biosci 5:96.