Deciphering Metabolism, One Microbe at a Time

For the latest research published by Nature, visit: nature.com www.nature.com/research- analysis

region was close to carbon balance during continued to accumulate carbon because their forests has therefore long been uncertain. the period of the study — about as much car- growing seasons have lengthened as a result Gatti and colleagues’ atmospheric profiles bon was taken up by plants for growth as was of climate change. Mid-latitude forests have show that the uncertain future is happening emitted from decay processes. done so because they have been recovering now. However, the moisture and fertility of from past clearance, and because they have Amazonian forests changes substantially benefited from the increased availability of Scott Denning is in the Department of farther south and east. Dry seasons (periods nutrients (produced as a result of human Atmospheric Science, Colorado State with rainfall of less than 100 millimetres per activities, or mobilized in soils by climate University, Fort Collins, Colorado 80523, USA. month) get progressively longer, eventually warming). e-mail: [email protected] lasting for 5 months or more as the forest By contrast, increased carbon sequestration grades into savannah10. Gatti and colleagues by tropical forests must be driven largely by find that the drier forests in the northeastern an increase in photosynthesis associated with

and southeastern regions studied were close rising CO2 levels — but regional atmospheric to carbon balance during the wet season, profiling12 suggests that this carbon sink is but that carbon release from decomposition threatened by forest degradation and warm- and fire tended to exceed carbon uptake by photosynthesis during the dry season. The 1. Friedlingstein, P. et al. Earth Syst. Sci. Data 12, 3269–3340 “The results cast doubt on (2020). observed regional and seasonal patterns of 2. Gatti, L. V. et al. Nature 595, 388–393 (2021). carbon uptake in the northwest transition- the future ability of tropical 3. Fung, I. Annu. Rev. Earth Planet. Sci. 48, 1–20 (2020). 4. Pan, Y. et al. Science 333, 988–993 (2011). ing to carbon release in the drier east were forests to sequester large 5. Song, X.-P. et al. Nature 560, 639–643 (2018). consistent with the year-to-year variability of amounts of carbon dioxide.” 6. Aragão, L. E. O. C. et al. Nature Commun. 9, 536 (2018). the data — which revealed that greater carbon 7. Hayek, M. N. et al. Biogeosciences 15, 4833–4848 (2018). 8. Malhi, Y. et al. Biol. Conserv. 253, 108889 (2021). releases, associated with decomposition and 9. Liu, J. et al. Science 358, eaam5690 (2017). fire, occurred during hotter and drier years. ing. Another complication is that fossil-fuel 10. Marengo, J. A., Liebmann, B.,Kousky, V. E., Filizola, N. P. & Gatti and co-workers show that the transi- emissions must be quickly reduced to meet Wainer, I. C. J. Clim. 14, 833–852 (2001). 11. Ballantyne, A. P., Alden, C. B., Miller, J. B., Tans, P. P. & tion of eastern Amazon forests from carbon international climate targets, but it is not clear White, J. W. C. Nature 488, 70–72 (2012). sink to carbon source during the dry season how the CO2-driven carbon sinks of tropical 12. Hubau, W. et al. Nature 579, 80–87 (2020). is associated with strong regional warming forests will respond to a rapidly warming world 13. Kaushik, A. et al. Eos 101, https://doi. 13 org/10.1029/2020EO140276 (2020). trends. Eastern Amazon sites have warmed by in which CO2 levels are no longer rising . The as much as about 0.6 °C per decade during the future of carbon accumulation in tropical The author declares no competing interests. dry season over the past 40 years. This is more than three times the rate of global warming Microbiology and about the same rate as for the Arctic. Wet-season and western Amazonian forests have warmed, too, but at a much slower rate. Warming rates in the dry season for eastern Deciphering metabolism, Amazonia might have been amplified by deforestation and forest degradation. Gatti one microbe at a time et al. conclude that increases in fires, and in physiological stress, mortality and decom- William F. Kindschuh & Tal Korem position of trees in this area, are associated with increasing carbon loss from regional Small molecules produced and modified by gut microorganisms ecosystems. can influence human physiology. An atlas of metabolic outputs The authors have documented the acceler- ating transition of forests from carbon sinks to of diverse gut microbes offers new ways to decipher the sources using direct measurements of large- microbial mechanisms behind their production. See p.415 scale gradients of atmospheric gas concentrations. The overall pattern of deforestation, warmer and drier dry seasons, drought stress, The microorganisms in our gut can have which microbes, enzymes and interactions fire and carbon release in eastern Amazonia far-reaching effects — on our liver1, arteries2 are involved in the production and uptake seriously threatens the Amazon carbon sink. and potentially even on our behaviour3. One of a specific metabolite. This task is further Indeed, the results cast doubt on the ability way these microbes exert their effects is hindered by the complexity of microbial com- of tropical forests to sequester large amounts through the generation or consumption of munities such as the gut microbiome, studies

of fossil-fuel-derived CO2 in the future. small molecules, termed metabolites. Meas- of which have to take into account the large For decades, ecologists have been surprised uring metabolite levels, an approach called number of microbes, the interactions between that the fraction of fossil-fuel emissions metabolomics, has led to ever-increasing them, their diverse metabolic capabilities absorbed by land ecosystems has remained recognition of their importance. And yet and several hard-to-measure non-microbial fairly constant11, even though these emissions only rarely do we understand the underlying factors, such as host diet4. On page 415, Han have increased. Forests at high latitudes have mechanisms driving these levels: namely, et al.5 present a comprehensive approach to

News & views it achieved an accuracy of approximately 70%). a Gut bacteria b These results raise a note of caution regard- Comparison ing typical microbiome analyses, which often of genetic and rely on microbial-abundance estimates at the Metabolite A metabolic similarities genus and species levels, and thus might miss Metabolite B Phylogenetic tree Metabolomic tree crucial metabolic aspects of the microbial showing genetic showing metabolic community. Metabolite C similarities similarities Although the correspondence between An atlas of phylogeny and metabolism is imperfect, the metabolic outputs c Comparative authors present an analytical approach using genomics Gene the association between specific genes and combined with comparative metabolic outputs to obtain insights into metabolomics microbial metabolism. Han and colleagues paired their metabolomic analyses with d analyses of bacterial genomes to uncover the Comparison genes responsible for unexplained metabolic of in vitro and in vivo results capacities. The authors identify a previously unknown mechanism by which microbes of the phylum Bacteroidetes utilize the amino Figure 1 | An approach to studying microbial metabolism. a, Han et al.5 generated a reference library acids glutamine and asparagine. Nevertheless, of 833 small molecules (metabolites) relevant to gut-microbial metabolism, and used it to assess the the spe genes responsible for producing the metabolic output generated by 178 bacterial strains commonly found in the human gut. The authors thereby molecules putrescine and agmatine in several generated an atlas of metabolic outputs, as shown in this hypothetical example. Darker blue on the heat map species are not present in three species of Fuso- indicates higher levels of production. b, The authors present a range of approaches for studying microbial bacterium that the authors found to produce metabolism using these data. They investigated the correspondence between the evolutionary relationships these molecules — a result that demonstrates of different microbes (their phylogeny) and their metabolic output. Phylogeny and metabolism generally the limitations of this analytical method. correspond; however, the authors found some exceptions and divergences. c, Han and colleagues further show that a parallel comparison between microbial genomes and their metabolic outputs could Han and colleagues conclude by turning to suggest genes responsible for unexplained metabolic capacities. d, The authors also investigated the the most challenging aspect of their approach: correspondence between in vitro and in vivo microbial metabolism, identifying many metabolites produced assessing the correspondence between in vitro in both contexts. and in vivo metabolic output. Strains with a prominent metabolic capacity, such as Citro- bacter portucalensis, which produces agma- addressing this major challenge, by carry- The authors show that, although the two tine from the amino acid arginine, maintained ing out metabolic and genetic analyses of a generally correspond, this correspondence some of this capacity both in culture and in broad set of microbes commonly found in the is not perfect. For example, Han et al. report mice. In some cases, this led to effects reach- human gut. that Clostridium sporogenes and Clostridium ing beyond the gut (systemic effects). For The authors’ approach (Fig. 1) has been facil- cadaveris, two closely related species, example, agmatine levels were increased in itated by notable technical advances. Using have strikingly different metabolic pro- the urine of mice if the animals were colonized liquid chromatography–mass spectrometry files. By contrast, Atopobium parvulum and by C. portucalensis. (LC–MS), a technique that quantifies metab- Catenibacterium mitsuokai, two phylogenet- However, such a high level of correspond- olites on the basis of their polarity, mass and ically distant species, have similar metabolic ence between in vitro and in vivo data was not charge, Han and colleagues compiled a ref- profiles. observed for the overall metabolic output, as erence database of 833 metabolites that are Furthermore, although the authors could tested by the authors for two strains. For these relevant to microbial metabolism. They con- identify some strong species-specific asso- strains, there was only a moderate correlation firmed that these metabolites are detectable between the in vitro metabolic profile of the in biological samples, and that their measure- “To what degree is the strain and the profile measured from the intes- ment is consistent in several types of sample, tines or faeces of a mouse colonized by it. Fur- such as faeces or blood, and quantifiable over a evolutionary relationship thermore, no correlation was found between wide range of concentrations. The authors also between two microbes the in vitro profiles of these strains and the developed an analysis pipeline that enables related to their metabolic blood or urine profiles of mice colonized by compound identification and statistical analy- capacity?” them. This was the case despite the simpli- sis. With this infrastructure in place, Han et al. fied ‘mono-colonization’ scenario, in which measured metabolite levels in thousands of each mouse harboured only a single microbial samples from in vitro cultures of 178 micro- strain, without other members of the bacterial bial strains grown separately in multiple media ciations with the production of particular community and the complex effects that arise types, and from various tissues taken from metabolites, such as production of the from the interactions between them6. mice whose intestines were colonized by the molecule tyramine by Enterococcus faecalis, These results highlight a major challenge same strains, either alone or in communities metabolomic profiles were insufficient to left in the wake of this impressive endeavour, of five or six species. independently distinguish between mem- which is to use this extensive atlas of meta- Having compiled an atlas of single-microbe bers of different species. A machine-learning bolic measurements, taken in simplified set- in vitro metabolic outputs, Han and colleagues algorithm trained to identify a species on the tings, to provide accurate models of complex set out to address a long-standing question: basis of its metabolomic profile was correct community metabolism. This could be done to what degree is the evolutionary relation- only about 30% of the time, and even members experimentally — for example, by extending ship between two microbes (their phylo of different genera or families were not well the work performed by Han et al. to assess geny) related to their metabolic capacity? separated by the algorithm (in such analyses, combinatorial co-cultures — or by harnessing

various computational and mathematical set-ups, thereby promoting the democratiza- interactions with the environment. Decoher methods7,8. Future work could further vali- tion of metabolomics. Altogether, this work ence typically occurs much faster in exper- date the utility of this data set for studying the lays a foundation for future work seeking to iments with oscillators that are directly human gut microbiome; extend the data set decipher microbial metabolism — an impor- tethered to their environment than in levitated to strains that are found in, and have probably tant step towards new therapeutics that target systems. adapted to, a specific host9; and expand the the microbiome. After a decade of effort by several groups data set to microbes and metabolites that are worldwide3, light-induced cooling of a levi- relevant to other human-associated micro- William F. Kindschuh and Tal Korem are at tated nanoparticle to the quantum ground bial communities, such as the vaginal and skin the Program for Mathematical Genomics and state was finally reported4 in 2020. But that microbiomes. the Department of Systems Biology, Columbia experiment relied on the quantum mode of Han and colleagues provide useful resources University, New York, New York 10032, USA. light bouncing between two highly reflective for the research community, including an e-mail: [email protected] mirrors, a set-up known as an optical cavity. extensive metabolomics data set consisting This approach comes with limitations: only

of thousands of samples, web resources with 1. Yoshimoto, S. et al. Nature 499, 97–101 (2013). particles with certain ranges of oscillation fre- which to explore it and analytical approaches 2. Koeth, R. A. et al. Nature Med. 19, 576–585 (2013). quency can be cooled in each set-up. Moreo- for studying microbial metabolism. Moreover, 3. Lynch, J. B. & Hsiao, E. Y. Science 365, 1405–1409 (2019). ver, it is challenging to control the operation 4. Bar, N. et al. Nature 588, 135–140 (2020). this work provides a truly open-source tech- 5. Han, S. et al. Nature 595, 415–420 (2021). of an optical cavity sufficiently well to hold a nical resource, with protocols, analysis pipe- 6. Coyte, K. Z., Schluter, J. & Foster, K. R. Science 350, particle stably, and then to cool it. lines and an extensive metabolite reference 663–666 (2015). Magrini et al. and Tebbenjohanns et al. 7. Venturelli, O. S. et al. Mol. Syst. Biol. 14, e8157 (2018). library, which the authors demonstrate to be 8. Chiu, H.-C., Levy, R. & Borenstein, E. PLoS Comput. Biol. used a completely different approach, dis- applicable, with minimal calibration, to dif- 10, e1003695 (2014). pensing with the optical cavity, and thus ferent machines. This resource could be used 9. Zhao, S. et al. Cell Host Microbe 25, 656–667 (2019). evading the associated problems. Their tech- by others as they pursue similar experimental The authors declare no competing interests. nique might therefore offer a more robust and straightforward way to prepare quantum Quantum physics states of mesoscopic objects (those between about 100 nanometres and one micrometre in size) . The authors’ approach (Fig. 1) is an exten- Feedback offers quantum sion of a method termed feedback cooling, in which continuous measurement of an oscilla- control of nanoparticles tor’s position enables a force (the feedback) to be applied that counters and damps the Tania S. Monteiro oscillator’s motion. Although feedback cooling has been extensively investigated, for some Precise measurements of the position of a levitating years there was considerable scepticism as to nanosphere have been used to control forces that damp whether this approach alone, without cavity the nanosphere’s motion — potentially opening the way to cooling, could reach the milestone of cooling a levitated particle to an average energy that quantum control of larger objects. See p.373 & p.378 corresponds to less than a single quantum of energy above the fundamental zero-point motion (the residual motion that an oscil- The Heisenberg uncertainty principle states independent studies in which they were able lator retains in the quantum ground state). that certain incompatible pairs of properties to track the position not of a single atom, The current studies demonstrate that this of a particle cannot be determined simulta- but of a nanosphere containing billions of milestone can indeed be reached using this neously with unlimited precision. It is often atoms, with a precision close to the Heisen- method. taught using a thought experiment for the berg limit (the minimum possible product of Several advances have paved the way to this case involving position and momentum: if the the uncertainties of the measured quantities). achievement. A feedback technique known as position of an atom is measured with light, the This enabled them to use a technique called cold damping, which applies a force that is pro- back-action of the scattered photons on the measurement-based quantum control to cool portional to the velocity of the particle, was in atom invariably disturbs the atom’s momen- the nanosphere from highly excited thermal the past few years5,6 shown to yield highly effi- tum. The back-action can be reduced by using states down to average energies that are close cient cooling. Importantly, the nanospheres less-energetic or fewer photons, but this also to the lowest energy state of the particle (the are naturally charged, which means that the reduces the precision of the measurement. quantum ground state). feedback force can be applied using an electric More specifically, the Heisenberg uncertainty The results of the two studies are a break- field7, rather than light — thus avoiding extra principle stipulates that the product of the through in optomechanics, the research photon back-action being exerted on the nano- uncertainties in measurements of position and field that aims to bring small mechanical spheres. The experimental set-ups in the two momentum must be greater than or equal to oscillators into quantum regimes through new studies also operate at ultrahigh-vacuum half of Planck’s constant, ħ. But this constant their interaction with light. In the subfield levels (about 10–12 of normal air pressure), is so tiny (1.05 × 10–34 joule seconds) that the of levitated quantum optomechanics, the largely eliminating heating and decoherence trade-offs between the back-action and impre- oscillator is a silica particle about the size associated with collisions of the nanosphere cision can be observed only in carefully con- of a virus (100–200 nanometres in diam with surrounding gas molecules. And both trolled experiments, typically using objects eter), and is trapped and controlled by light. studies benefited from improvements in at the size scale of atoms. This minimizes both unwanted heating of the efficiency with which scattered photons Now, Magrini et al.1 (page 373) and the particle and decoherence — the loss are collected to measure the position of the Tebbenjohanns et al.2 (page 378) report of the particle’s quantum behaviour through nanosphere8.