Predictive microbial ecology

Crystal ball 2009

In this feature, leading researchers in the field of microbialbiotechnology speculate on the technical and conceptualdevelopments that will drive innovative research and opennew vistas over the next few years.

Mining for new enzymes

Amir Aharoni, Department of Life Sciences and theNational Institute for Biotechnology in the Negev (NIBN),Ben-Gurion University, Beer-Sheva, Israel.The size of the microbial world is beyond our imagination.In last year’s crystal ball article, Tom Curtis compared thesize of the microbial world to the size of the universepointing out that the number of microorganisms in theworld is billions of times larger than the number of stars inthe sky (Curtis, 2007). There is little doubt that such unlim-ited microbial biodiversity holds great promise for enrich-ing our repertoire of known enzymes, with many novelenzymes awaiting discovery.

The development of new sophisticated sequencingtechnologies, including the 454 pyrosequencer andSolexa technology, allow for accumulation of extremelylarge amounts of sequence information. Such rapidlyaccumulated information will allow microbiologists, bio-informaticians and system biologists to perform extensivesurveys on microbial lineages, lead to the description ofnew metabolic pathways and allow for the identification ofnew regulatory mechanisms. These data also representan invaluable resource in our quest for new enzymes. Stillthe question of how do we actually find these preciousneedles in the large haystack arises. Namely, what will bethe best way to find new enzymes in the huge space thatis the microbial world?

The discovery of new enzymes is of ultimate impor-tance for fundamental knowledge regarding enzyme evo-lution, enzyme structure-function relationships, basicmechanisms of enzyme catalysis and even for the identi-fication of novel protein folds. Novel enzymes could alsoserve as corner stones in catalysing industrial chemicalsynthesis reactions, thus serving as ‘clean’ alternatives forproducing chemicals by large-scale chemical synthesis.Currently, new highly robust enzymes are urgentlyneeded as biocatalysts in many different industries,including the pharmaceutical and agricultural industries.The dream of significantly expanding the repertoire ofknown enzymes, both for research and industrial applica-tions, is currently the subject of intensive research and will

keep many scientists busy in years to come (Ferrer et al.,2007).

One of the main methods for the identification of novelenzymes is by virtual sequence homology screens. Usingthis approach, new enzymes are identified by comparingvast number of sequences from genomic/metagenomicsources to the sequences of known enzymes (Ferreret al., 2007). In this process, the immense repertoire ofmillions of proteins predicted from the microbial sequencedatabase is compared with known enzymes to identifynew homologues. One of the obvious limitations of suchan approach is that it exclusively relies on existing geneannotations that are difficult to predict and often prone toerrors due to their reliance on machine-based techniques(Hallin et al., 2008). No doubt that improved annotationwill enable more accurate gene function predictions frommicrobial sequencing data. However, the main conceptualdrawback of such an approach is that truly novel enzymesthat are only remotely related to known existing enzymeswill never be found. Genetic drift can significantly alter anyobvious homology between functionally similar enzymes,thereby restricting the search to only related enzymeswith similar sequences.

An unbiased manner to mine natural microbial biodiver-sity for new enzymes is by functional screening for thedesired enzymatic activity. Such direct screening for newenzymes will allow the identification of novel enzymes thatare not related by sequence homology to any other knownenzyme. A prerequisite for any functional screeningprocedure is the availability of high-quality genomic/metagenomic libraries and the use of an adequate hostorganism that is able to express the target genes to yieldfunctional proteins. Currently, most functional screens relyon spatial separation between the different samples eitheron agar plates (by direct screening of colonies) or usingmicrotitre plates. To perform such screens for large librar-ies, access to heavy robotic systems usually accessibleonly for large laboratories, is a prerequisite. Even with theaid of robotics, however, the number of genes that can bescreened is on the order of ~104. Given the unlimitedamount of diversity in the microbial world, this number willallow sampling of only a tiny fraction of the functionalenzymes out there awaiting to be discovered.

To allow for a more efficient route for functional miningof new enzymes, we must adopt high-throughputapproaches that will allow us to rapidly screen 106–108

samples. These numbers are clearly far beyond reach of

Microbial Biotechnology (2009) 2(2), 128–156 doi:10.1111/j.1751-7915.2009.00090.x

© 2009 The AuthorsJournal compilation © 2009 Society for Applied Microbiology and Blackwell Publishing Ltd

any currently available robotic system and rely on bulkselections for enzymatic activities rather than screeningsamples individually. In recent years, high-throughputscreening/selections for enzymatic activities have beendeveloped based on sophisticated approaches linkinggenotype to phenotype (Aharoni et al., 2005). Novelmethodologies that allow us to maintain the linkagebetween the gene, the enzyme it encodes and the productit generates are based on cell-surface display technolo-gies (using phage, bacteria or yeast) or on in vitro com-partmentalization, using emulsion techniques (Aharoniet al., 2005). Different high-throughput screening (HTS)approaches have been developed for a number ofenzyme families, including proteases, esterases, phos-photriesterases, peroxidases, DNA/RNA polymerasesand glucosyltransferases. All of these approaches allowthe screening of extremely large libraries by multiplecycles of enrichment using flow cytometry or selectiveimmobilization of active clones. These methodologies,originally developed for directed evolution experiments,can be readily adopted for screening large genomic/metagenomic libraries. Still, despite the powers of HTSassays, applications of these technologies for miningmicrobial libraries will require library pre-enrichment andthe use of appropriate host organisms to increase ourchances of the identifying and isolating novel enzymes. Ibelieve that in the near future, we will see increasingefforts in applying powerful screening technique tosample ever larger fractions of the microbial world for thediscovery of new and exciting enzymes.

References

Aharoni, A., Griffiths, A.D., and Tawfik, D.S. (2005) High-throughput screens and selections of enzyme-encodinggenes. Curr Opin Chem Biol 9: 210–216.

Curtis, T. (2007) Theory and the microbial world. EnvironMicrobiol 9: 1–11.

Ferrer, M., Golyshina, O., Beloqui, A., and Golyshin, P.N.(2007) Mining enzymes from extreme environment. CurrOpin Microbiol 10: 207–214.

Hallin, P.F., Binnewies, T.T., and Ussery, D.W. (2008) Thegenome BLASTatlas-a GeneWiz extension for visualizationof whole-genome homology. Mol Biosyst 5: 363–371.

Food and gut microbes for thoughts

Fabrizio Arigoni and Harald Brüssow, Nestlé ResearchCentre, Lausanne, Switzerland.The ambulance returns to the hospital with a young motorcyclist who had lost the control of his vehicle on an icypatch on the road. The surgeon observes heavy injurieson one leg, the bones are shattered, severe torsions wereexerted on the knee and ankle and a deep open woundgoes down into the bone. A lengthy operation will beneeded and he knows that an infection with pathogenic

staphylococci and streptococci can annihilate his efforts.The surgeon hates amputation and prosthetics in such ayoung man as a personal defeat. The surgeon will try hisbest. Surprisingly, he does not disinfect the wound nordoes he apply antibiotics into the operation wound. After afirst cleaning of the wound, he sprays a mixture of com-mensal skin bacteria and starter bacteria used in meatfermentation (Staphylococcus carnosus and Lactobacillusplantarum) on the wound. It turned out that these bacteriawork quite well: they digest the dead tissue, the produc-tion of metabolites like lactic acid and hydrogen peroxidefights off Staphylococcus aureus from the wound andassists in the build-up of granulation tissue.

This scenario is of course science fiction but not somuch as you might think. US military surgeons observedin the 19th century that wounded soldiers left for one ortwo days on the battle field fared better than rapidlyrescued soldiers. Flies had the time to lay their eggs intothe wounds. The maggots, i.e. the fly larvae that hatchedfrom the eggs, make their living from living flesh. Theydigest the wound tissue by proteolytic enzymes. Notably,the saliva of the maggots also contains substances thathave a strong anti-bacterial activity including one againstS. aureus. Small wonder that maggot therapy – despite itsemotional disgust reaction – recently got FDA approvaland starts to become popular with wound surgeons.

Antibiotics, the big success story of mid-20th centurymedicine, were not mentioned in our fiction. The magicbullets, how Paul Ehrlich called the agents of early che-motherapy, are loosing their spell. We urgently need alter-natives. In addition to antibiotics, which are the brain childof the chemical industry, commensal bacteria mightbecome the future tools of a biological industry that useswhole organisms as ecological competitors not onlyagainst pathogens. Prominent scientists explore theimpact of commensal bacteria on host physiology (obesityand related diseases, Turnbaugh et al., 2006). Othersinvestigate the effects of commensals on the metabolismof food (soy isoflavons in the context of postmenopausaldisturbances, Bolca and colleagues 2007) or drugs (toexplain distinct pharmacokinetics, Sousa et al., 2008).There are already bacteria on the market with docu-mented clinical studies showing activity against diarrhoeaor inflammatory gut diseases or with generalized immu-nostimulatory activity. On the analytical side, it is muchmore difficult to define the mode of action of these bio-logicals than for antibiotics, which target a specific bio-chemical reaction. However, the pathogen will also beconfronted with the same dilemma when competing withthe commensal and there is some theoretical hope thatthe pathogen might less easily escape from control bycommensals than from antibiotics.

The Iron Curtain crossing Europe during the Cold Warperiod has also prevented the flow of ideas across this

Crystal ball 129

© 2009 The AuthorsJournal compilation © 2009 Society for Applied Microbiology and Blackwell Publishing Ltd, Microbial Biotechnology, 2, 128–156

ideological frontier. Biological research was not an excep-tion. The former Soviet Union relied more on biologicalapproaches against infectious diseases than on antibiot-ics. Antibiotic-resistant bifidobacteria and lactobacilli weresold in Russian pharmacies before probiotics becamepopular in the Western world. Even the above sciencefiction scenario was to a certain extent realized in theSoviet Union with a wound spray and wound dressingcontaining a cocktail of phages. The phages are directedagainst six major wound pathogens including S. aureus,Streptococcus pyogenes and Pseudomonas aeruginosa.We cannot, however, claim that the future is already now.The phage approach – despite its application to ten thou-sands of soldiers of the Red Army over half a century –has never been tested in controlled clinical trials fulfillingcurrent criteria of clinical science. The jury is thus still outwith respect to their efficacy.

If you object that our crystal ball gazing goes into thepast, you can easily modernize the commensal and phageapproach biotechnologically. Lactobacilli have been con-structed that express single chain antibodies with anti-rotavirus activity. Lactococci were modified for intestinaldelivery of human interleukin 10 providing a therapeuticalapproach for inflammatory bowel disease (Steidler et al.,2000). The bacterial carrier offers not only a safe passagethrough hostile environments like the stomach; commen-sals derived from defined body sites might re-home to theirnatural site. Proteins of medical interest could thus betargeted to particular anatomical sites and expressed insitu (for ex. Bifidobacterium breve expressing cytosinedeaminase for tumour-targeting enzyme/prodrug therapy,Hidaka et al., 2007). Phages are particularly attractive forbiotechnologists as they are not only genetically moretractable than bacteria but also known to reach practicallyall body sites, including those which you cannot easilytarget with conventional drugs. Filamentous phages wereextensively studied for foreign gene expression and phagedisplay technology was developed with them. Now takethe following scenario: modify such a phage such that itexpresses a cocaine-binding single chain antibody. Applythe recombinant phage to the nose of a drug-addict.Taking the privileged neuronal pathway of the first cranialnerve, the recombinant phage travels directly into thebrain of the subject. Here the phage binds the cocaine,which reaches the brain of the drug addict and preventsthe psychoactive processes including the self-administration drive for cocaine. This story sounds evenmore science fiction than the opening story. However, it isthe content of a 2004 PNAS paper demonstrating theseeffects in mice (Carrera et al., 2004).

Looking beyond the time horizon of grant applications isa daring exercise. Microbiology, which currently enjoysand suffers a data flood derived from genomics, transcrip-tomics, proteomics, metabolomics and metagenomic

analyses, needs long-term visions based on a soundtheoretical reasoning. Microbiologists cannot leave thefield to computational scientists expecting them to sort outnew ideas from the data accumulated by the various‘-omics’ approaches. A relationship as between theoreticaland experimental physicists must be developed in micro-biology where new theoretical concepts are tested withavailable data sets leading to new experiments testingrefined theories. The challenge of the antibiotic crisismight therefore also be a healthy push for biologicalapproaches not only against infectious diseases.

References

Bolca, S., Possemiers, S., Herregat, A., Huybrechts, I., Hey-erick, A., De Vriese, S., et al. (2007) Microbial and dietaryfactors are associated with the equol producer phenotypein healthy postmenopausal women. J Nutr 137: 2242–2246.

Carrera, M.R., Kaufmann, G.F., Mee, J.M., Meijler, M.M.,Koob, G.F., and Janda, K.D. (2004) Treating cocaine addic-tion with viruses. Proc Natl Acad Sci USA 101: 10416–10421.

Hidaka, A., Hamaji, Y., Sasaki, T., Taniguchi, S., and Fujimori,M. (2007) Exogenous cytosine deaminase gene expres-sion in Bifidobacterium breve I-53-8w for tumor-targetingenzyme/prodrug therapy. Biosci Biotechnol Biochem 71:2921–2926.

Sousa, T., Paterson, R., Moore, V., Carlsson, A., Abrahams-son, B., and Basit, A.W. (2008) The gastrointestinal micro-biota as a site for the biotransformation of drugs. Int JPharm 363: 1–25.

Steidler, L., Hans, W., Schotte, L., Neirynck, S., Obermeier,F., Falk, W., et al. (2000) Treatment of murine colitis byLactococcus lactis secreting interleukin-10. Science 289:1352–1355.

Turnbaugh, P.J., Ley, R.E., Mahowald, M.A., Magrini, V.,Mardis, E.R., and Gordon, J.I. (2006) An obesity-associated gut microbiome with increased capacity forenergy harvest. Nature 444: 1027–1031.

Predictions: evolutionary trajectories andplanet medicine

Fernando Baquero, Department of Microbiology, The Uni-versity Hospital Ramón y Cajal in Madrid (Madrid’s Insti-tute of Health), CIBERESP Madrid, Spain.

The age of synthesis: prediction of bacterialevolutionary trajectories

Conventional scientific wisdom dictates that evolution is aprocess that is sensitive to many unexpected events andinfluences, and is therefore essentially unpredictable. Onthe other hand, considering the bulk of recent knowledgeabout bacterial genetics and genomics, population genet-ics and population biology of bacterial organisms, and

130 Crystal ball


their sub-cellular elements involved in horizontal genetransfer, we should eventually face the possibility of pre-dicting the evolution of bacterial evolution. The importanceof such type of approach is self-evident in the case of theevolution of antibiotic resistance, and bacterial–host inter-actions, including infections. Prediction of bacterial evolu-tion could provide similar clues as weather prediction –higher possibilities of certainty in the closer and more localframes. Indeed, there is a local evolutionary biology basedon local selective constraints that shapes the possible localtrajectories, even though in our global world, some of theselocally originated trends might result in global influences. Inthe case of adaptive functions (as antibiotic resistancegenes in pathogenic bacteria), some of the elementswhose knowledge is critical for predicting evolutionarytrajectories are: (i) the origin and function of these genes inthe chromosome of environmental bacterial organisms; (ii)their ability to be captured (mobilized) by different geneticplatforms, and to enter in particular mobile genetic ele-ments; (iii) the ability of these mobile genetic elements tobe selected and spread among bacterial populations; (iv)the probability of intra-host mutational variation and recom-bination; (v) the probability of re-combinatorial eventsamong of these and other mobile elements, with conse-quences in selectable properties and bacterial host-range;(vi) the original and resulting fitness of the bacterial clonesin which the new functions are hosted, including its coloni-zation power and epidemigenicity; (vii) the results of inter-actions of these bacterial hosts with the microbioticenvironment in which they are inserted; and (viii) theselective events, as the patterns of local antibiotic con-sumption, or industrial pollution, and in general, the struc-ture of the environment that might influence the success ofparticular complex genetic configurations in which theadaptive genes are hosted (Baquero, 2004). Dealingsimultaneously with all these sources of evolutionary varia-tion is certainly a challenge. Such a type of complexstructure has evolved along all biological hierarchicallevels, creating specific ‘Chinese-boxes’ or ‘Russian-dolls’patterns of stable (preferential) combinations, for instanceencompassing bacterial species, phylogenetic subspe-cific groups, clones, plasmids, transposons, insertionsequences and genes encoding adaptive traits (Baquero,2008). Assuming a relatively high frequency of combinato-rial events, the existing trans-hierarchical combinationsare probably the result on the local availability of thedifferent elements (pieces) in particular locations (localbiology), the local advantage provided by particular com-binations, and also the biological cost in fitness of some ofthem. More research is needed to draw the interactivepattern of biological pieces in particular environments(grammar of affinities). Such a complex frame required forpredicting evolutionary trajectories (Martínez et al., 2007)will be analysed (and integrated) by considering heuristic

techniques for the understanding of multi-level selection.The application of new methods – based on covariance,and contextual analysis, for instance using Price’s equa-tion derivatives – should open an entirely new syntheticway of approaching the complexity of living world.

The age of planet medicine

Because of the increasing, apparently unavoidable influ-ence of human species on the ecology of our planetEarth, and the necessary counteraction of a modifiedplanet on human health and style of life, the entire planetshould be considered as something requiring medicalcare. The future human medicalization requires theplanet medicalization. The refined medical methodologyshould be escalated to the planet dimension, starting bydefining the signs and symptoms of illness, studyingthe pathogenesis and pathophysiology of planet illness,trying to evaluate their possibilities to invade otherregions, establishing specific methods for diagnosisusing all available technologies, from genetics to imageanalysis, and try to make ecological and evolutionarypredictions. This will be followed by applying specificinterventions (not excluding surgery), treatments, or evenisolation procedures and intensive care technology, andrecommending or imposing prevention measures. As weare examining the safety of drugs or foods for humans,we should do the same for anything influencing theplanet. Of course microbiologists have a big role to startthis process – as medical and environmental microbiolo-gists know each other and are progressively closer, andbecause they have the tradition of being involved inglobal problems – international health. Microbiologistsare also mastering one of the key issues to start withPlanet Medicine, the microbial diversity. Changes inmicrobial diversity might constitute one of the basesof altered planet symptomatology (Baquero, 2003). Ofcourse for starting such a process a totally new way ofmeasuring microbial biodiversity should be developed,forgetting the stupid Linnean way of defining biologicalevolutionary units (species?). Bacteria cannot have10 000 species (or less) if Arthropods have 10 000 000.Something is wrong. Population biology and populationgenetics should be expanded from organisms to anyother evolutionary unit. What is clear is the importance ofmeasuring microbial biodiversity, as microbes are criticalin the basic functions of Nature, including nutrient recy-cling. Cyanobacteria constitute up to 70% of the totalphytoplankton mass, and are responsible for more that25% of the total free O2 and about an equivalent propor-tion of CO2 fixation. Why the changes in size of appar-ently little Amazonias do not appear every day in the laypress, as the big one? Are we unable to recognize thesymptoms leading to biological catastrophes?

Crystal ball 131


References

Baquero, F. (2003) Chaos, complexity, and evolution. ASMNews 69: 543–552.

Baquero, F. (2004) From pieces to patterns: evolutionaryengineering in bacterial pathogens. Nat Rev Microbiol 2:510–518.

Baquero, F. (2008) Modularization and evolvability in anti-biotic resistance. In Evolutionary Biology of Bacterialand Fungal Pathogens. Baquero, F., Nombela, C., Cassell,G.H., and Gutiérrez-Fuentes, J.A. (eds). Washington, DC,USA: ASM Press, pp. 233–248.

Martínez, J.L., Baquero, F., and Andersson, D.I. (2007) Pre-dicting antibiotic resistance. Nat Rev Microbiol 5: 958–965.

Back to the roots

Matthias Brock, Microbial Biochemistry and Physiology,Leibniz Institute for Natural Product Research and Infec-tion Biology, Hans Knoell Institute, Beutenbergstr. 11a,07745 Jena, Germany.The biochemical and functional analysis of proteinswith unknown functions can be a difficult task andneeds endurance and the knowledge of sometimes ‘old-fashioned’ methods. Even more, without a sequencedgenome, it takes a long time to identify the DNA sequencecoding for the protein of interest. Remembering my timeas a diploma student, working on the filamentous fungusAspergillus nidulans, it took me weeks or months to iden-tify the genomic sequence corresponding to a purifiedprotein, because the genome sequence was not publiclyavailable. One of the strategies in ‘former days’ was topurify the protein, blot it onto a membrane, performN-terminal protein sequencing, construct degenerateprimers, screen a genomic library, subclone fragmentsand sequence them. Nowadays, the protein can be frag-mented by trypsin, surveyed by MALDI-TOF analysesand, due to the increasing number of finished genomeprojects, the gene is subsequently identified by an auto-matic database search against the genome of interest.This methodology speeds up the procedure by severalweeks and allows a much higher throughput in the iden-tification of gene functions.

However, genome projects would not have been initi-ated to ease the identification of the coding sequence of asingle interesting protein. In fact, the sequencing ofgenomes has opened the era of ‘omics’, starting from‘gen’omics, continuing with ‘transcript’omics and enhanc-ing the informative value of ‘prote’omics. Now it is possibleto compare genomes of different organisms (which genesare universal and which are specific), to look for changesin transcript levels (e.g. after applying an environmentalstress) and to identify modifications of proteins and theirabundance under defined conditions. These massiveamounts of data (especially from transcriptomics) create a

‘Garden of Eden’ for bioinformatitians, who can performstatistical analyses on the data sets to evaluate theirsignificance in order to develop new methods for hierar-chical cluster analyses. The use of mutual informationmatrices on time response studies allows the identifica-tion of genes that can be grouped into regulons. Theseanalyses aim to suggest new hypotheses on the connec-tion of different pathways and their communication.The challenge of the biologists is to examine thesehypotheses experimentally to give a verification of thepredictions.

This sounds like a ‘beautiful new world’, because ‘life’becomes computable and it works well for pathways inwhich all genes and proteins involved are already known.However, these analyses and predictions are hamperedby several problems, in particular: (i) the countless ‘hypo-thetical proteins’ and genes of ‘unknown function’, and (ii)the genes and proteins, which were annotated by theiridentity to other already characterized proteins. Analysesdealing with those data rapidly reach a dead end and leadto a sobering conclusion: bioinformatics cannot substitutethe wet lab.

Transcriptional analyses, the prediction of the numberof genes within a genome, the comparison of genomes,the location of proteins and other features depend on acorrect genome annotation. However, the functionalannotation of proteins is very frequently misleading withsometimes profound consequences. A single example,although there are many more that could be listed: Themethylisocitrate lyase is a key enzyme of the fungalmethylcitrate cycle and specifically cleaves (2R,3S)-2-methylisocitrate into succinate and pyruvate. The enzymeis highly specific for its natural substrate and does notaccept isocitrate in its active site. Fungal isocitrate lyases,specific enzymes from the glyoxylate bypass, share about35–50% identity to fungal methylisocitrate lyases, but arehardly active with methylisocitrate. By the means of ‘iden-tity’ most fungal methylisocitrate lyases are incorrectlyannotated as isocitrate lyases. This is exemplified on theso-called isocitrate lyase 2 from Saccharomyces cerevi-siae, which was formerly denoted as a ‘non-functional’isocitrate lyase, because no activity was observed withisocitrate as a substrate (Heinisch et al., 1996). However,a re-characterization revealed significant methylisocitratelyase activity (Luttik et al., 2000), which led to the correc-tion of the annotation. Nevertheless, there are still severalmethylisocitrate lyases in fungal genome annotations thatare denoted as isocitrate lyases, but indeed representmethylisocitrate lyases. Although this might sound as aminor problem, the missing methylisocitrate lyase wouldlead to an incomplete methylcitrate cycle. Metabolic fluxanalyses, however, depend on the knowledge of all path-ways present within a cell and a missing pathway maylead to incorrect data evaluations.

132 Crystal ball


Genes of unknown function or hypothetical proteinscause even bigger problems. A strong upregulation ordownregulation of gene expression suggests an impor-tance under the applied condition. However, due to thecomplexity within a cell it is difficult to predict, whetherthe change is a direct cause from the applied stressor resulted from a subsequent adaptation mechanism.Systems biology tries to answer this question by hierar-chical clustering of the changes, which provides hints onadditional genes that may be involved in the samepathway. However, without a detailed molecular biologicaland/or biochemical analysis of each gene, its true functionremains unsolved. Therefore, ‘omic’-researchers need tocontinue to collect large data sets but should have in mindthat the characterization of genes in the laboratory is stillan essential approach, in particular for those of unknownfunction, but also for those with only a predicted function.

Therefore, I propose to give the characterization ofunknown function genes a higher priority. Currently, itseems as the value of such research is not appreciatedand getting financial support for this kind of research isdifficult. A few years ago I published a paper dealing withthe biochemical characterization of several enzymes andthe impact of a gene deletion on cellular physiology(Brock and Buckel, 2004). It turned out quite difficult to getit published, because one reviewer commented: ‘. . . themanuscript mainly deals with methods used in the sixtiesand seventies and it is questionable, whether such inves-tigations are still suitable for publication.’ Of course itmight be less ‘sexy’ for readers to become confronted with‘old methods’, but the time may have come to go backto the roots and to elucidate the function of genesof unknown function to drive the advancing field of bio-logical science, including computer-based technologies,forward.

Acknowledgements

I thank Bernhard Hube for the helpful feedback and discus-sion on this issue.

References

Brock, M., and Buckel, W. (2004) On the mechanism of actionof the antifungal agent propionate. Eur J Biochem 271:3227–3241.

Heinisch, J.J., Valdes, E., Alvarez, J., and Rodicio, R. (1996)Molecular genetics of ICL2, encoding a non-functionalisocitrate lyase in Saccharomyces cerevisiae. Yeast 12:1285–1295.

Luttik, M.A., Kotter, P., Salomons, F.A., van der Klei, I.J., vanDijken, J.P., and Pronk, J.T. (2000) The Saccharomycescerevisiae ICL2 gene encodes a mitochondrial2-methylisocitrate lyase involved in propionyl-coenzyme Ametabolism. J Bacteriol 182: 7007–7013.

The microbial reactome

Manuel Ferrer, CSIC – Institute of Catalysis, Madrid,Spain.

There is not one path anymore. Twenty years ago, youworked at the clean bench, you isolated new microbesable to grow on agar plates, and then you isolated singlegenes coding single enzymes for particular processes,you optimized them, and you wrote books or articles withconceptual and technical developments based on knownbiodiversity. If you were lucky, a small-to-mediumcompany approached you and continued to go furtherwith the applicability of such finding. Today you can be abiogeochemist with bioinformatics knowledge who writesbooks or articles trying to reveal the mysteries of microbialand genetic adaptation and diversity. It all sounds so . . .uncomplicated, doesn’t it? But, of course, THIS DIDN’THAPPEN overnight. It’s been especially in the past fifteenyears that a confluence of factors, mainly, technical devel-opments, has resulted in some young people turning theirbacks on sequencing. But above all, there is the sensethat biodiversity is at the centre of a vital scientific uni-verse, with microbes as its capital: we know the commu-nities, how diverse they are, but we are far fromunderstanding the individual members and functions, andhow each of them can be helpful, for example, to improvethe human condition. It is like our human society: thegovernment knows how many we are, but it does notknow how each individual lives, and how many consortia(friends and family, to cite some) we constitute.

The co-founding editor of Microbial Microbiology wroteto me and asked, wouldn’t I want entree into a crystal ball,to ‘predict’ the future and catch reader attention? Ofcourse, there’s more than a little romanticism to do this,there is a discernible sense that, as a young researcherput it: ‘those kinds of jobs – people predicting the future –exist, but just not for scientists’. However, I agreed,because I don’t know why anyone who wants to beinvolved in scientific understanding would not want to turntheir attention to future ideas. Following on from this, howdo I know I have a correct vision over the next few years?It’s hard to get a sort of accurate gauge on how you’redoing or what you will do, but you have to just take it onfaith that ‘there is a real biotech-market out there and anappreciation for what you’re doing’.

I am a chemist-turned-enzymologist and now a micro-biologist. Like all of you, I believe that microbes are impor-tant for the Earth System, playing a very important role inmaintaining the well-being of our global environment.Despite the obvious importance of microbes, very little isknown of their diversity, how many species are present inthe environment, and what each individual species does –i.e. its ecological function. Until recently, there were noappropriate techniques available to answer these impor-

Crystal ball 133


tant questions. The vast majority of these organismscannot be cultured in the laboratory and so are not ame-nable to study by the methods that have proven sosuccessful with known microorganisms throughout the20th century. It was only with the development of high-throughput technology to sequence DNA from the naturalenvironment that information began to accumulate thatdemonstrated the exceptional diversity of microbes inNature – in fact, most microbes are entirely novel andhave not previously been described.

A non-exhaustive list of questions that should beaddressed over the next few years includes: ‘is every-thing is everywhere?’, ‘do microorganisms exhibit bio-geographical patterns of distribution?’, ‘is the relativeabundance of a certain group of microorganisms neces-sarily linked to their importance in the community func-tioning?’, ‘which organisms are of pivotal importance inthe community?’, ‘how diverse are metabolic pathwaysand networks within the given ecosystem?’, ‘how domicrobes and protein-coding genes interact with eachother to lead to the overall system function?’, ‘how manyspecific microbes are responsible for the metabolism ofdifferent substrates?’, ‘how do environmental stimuliimpact ecosystem functioning and long-term system sta-bility?’, and finally ‘how can we improve the meta-genomictechnology for accommodating the needs of microbialbiologists and enzymologists?’.

To answer such questions, it should be noted that con-ceptual advances in microbial science will not only rely onthe availability of innovative sequencing platforms butalso on sequence-independent tools for getting an insightinto the functioning of microbial communities. I believethat is so because, over the last four years, in all confer-ences there was the very same question: ‘how can I getinformation about hypothetical genes and functions’? Thereasons are clear. First, every single cell or environmentalgenomic project added a huge number of putative genes,the function of which is often unknown and at bestdeduced from sequence comparison. Second, even thebest annotations only created hypotheses of the function-ality and substrate spectra of proteins which requireexperimental testing by classical disciplines such asphysiology and biochemistry. This highlighted the difficul-ties of making sense of environmental sequence data: asignificant proportion of the open reading frames couldnot be characterized because there were no similarsequences in the databases.

As we are primarily concerned with establishing thefunction of microorganisms in the environments andidentifying new enzymes with biotech-promiscuity, thereis an urgent need for characterizing protein functionsfrom environmental DNA or proteomes. Once functionhas been identified, it can be mapped to metabolicpathways or proteins involved in a particular process

(environmental or biotech-like), to determine the func-tional activity. To this end, I think that it will be helpful togenerate visualization tools capable of generating func-tional and dynamic knowledge, if possible, non-destructively and in real time. That is, tools that identifythe connected poles of activities (the so-called microbialreactomes) that shape the internal structure of an eco-logical niche, without a large-scale DNA sequence analy-sis. This is a straightforward concept since it constitutesthe direct link from DNA and genes to proteins and func-tions, a major hurdle in both Systems Biology and Bio-technology studies. By doing this, it will be possible tounravel gene functions and add valuable informationabout how microorganisms adapt to changing environ-mental influences, and how biotech processes can bedesigned by new microbial functions that can be checkedby visualizing directly the reactomes.

I think that methods for identifying at global-scale micro-bial reactomes are partially available, but we still needto solve many problems. For example, bioinformaticsmethods exist for isolating in silico microbial reactomes;but they rely on sequence data. Metatranscriptomics hasthe potential to describe how metabolic activities willchange, but still does not reflect the protein level and doesnot predict microbial functions (only upregulation or down-regulation). Metaproteomics gives valuable informationon how microbes respond to stress, but it is limited to thelow resolution and no direct functionality. Metabolomicshas become very popular recently by combining new ana-lytical and isotope analyses but, in the environmentalcontext, its use is very meager because the difficulty inidentifying and localizing metabolites. Finally, single-cellgenomics is increasing in importance but, once again, thesequencing of such cells will predict functions based onsequence data or, in the best case functional hypothesisof certain individual functions. If we ignore theseproblems, we increasingly waste significant financialresources and staff effort in order to achieve a final goal:reconstruct experimentally based reactomes in singlecells or complex communities.

So it does not need to take a crystal ball to see that thebottleneck in meta-genomic technology, both for microbialand biotech point of view, will not be only the design ofpowerful assembler computer programs but rather thedevelopment of technologies that provide direct analysisof complex mixtures and entail detecting specificsubstrate-protein transformations among thousands ofother endogenous metabolites and proteins in order to geta clear picture of ‘who is doing what’. Some methods doexist for isolating single transformations from the naturalenvironment; but these are not relevant for reactomecoverage as they are not universal. Clearly, existingmethods for enzymatic activity detection based onchanges in spectroscopic properties should give rise to

134 Crystal ball


high-throughput chips than can be used to provide infor-mation on chemistry of reactions and identity of theproduct formed. This type of information will be extremelyuseful for ascribing functions to genetic sequences fromenvironmental samples, thus minimizing annotation mis-takes and suggesting biotechnological potential. I believein a future where any single genome or environmentalsequence project is done in parallel with chip-basedenzyme screening, so that annotations are experimentallydocumented at the time when the paper is written. Onlythrough obtaining holistic information can holistic hypoth-eses about ecosystem characteristics be formulated. Thequestion then is: ‘how many reaction and substrate typesshould one have in a single high-throughput chip to coverthe whole microbial metabolism’? As Shakespeare said inHamlet, ‘that is the question’! I think that this is the time tothink about it as progress to manage sequence informa-tion per se accelerates. All in all, it is clear that to accessthe microbes in their natural milieu and new enzymes fromthem, there is a strong need to elaborate a SystemsBiology concept based on the combination of multiplestrategies to understand the functioning of microbial com-munities as a whole, with metagenomic tools playing apivotal role (Fig. 1).

Microbial genomics as pursuit ofhappiness

Michael Y. Galperin, NCBI, NLM, National Institutes ofHealth, Bethesda, MD 20894, USA.Over the past dozen years, the availability of completegenomes brought a profound change to all aspects of

(micro)biological research. As noted by Ian Dunham(2000), during the ‘dark ages’ before the advent of genom-ics, our perception of the cell was akin to the medievalmaps of Earth with large areas marked ‘Here be dragons’.It is now quite common to describe enzymes, metabolicand signalling pathways that are missing in a given organ-ism, something that can be done only with completegenome sequences. Although close to a third of genes inany newly sequenced genome have unknown functions(and the rest have only more or less reliable functionalpredictions that are not going to be experimentally testedany time soon), we can safely assume that these genesdo not code for dragon skin or any other dragon bodyparts. We are even running out of candidate genes thatcould code for the vital force (aka the ‘living soul’, Hebrew:nephesh; Greek: psuche; French: élan vital) of the bacte-rial cell.

In contrast, microbial technology has remained rela-tively unaffected by the genomics data. Most biotechno-logical processes still remain the same as they were10-15 years ago. Genomic and metagenomic libraries arewidely used to search for useful enzymes but thosesearches rely more on activity than on genome sequencedata. It is easy to predict that in the course of the nextseveral years genomics will start making its way intoeveryday technology. The most immediate change will bethe recruitment of an ever-expanding range of organismsfor use in the bioremediation of environmental contami-nants and in production of various compounds, from bio-pharmaceuticals to biofuels.

Use of new organisms will result in dramatic progress inmetabolic engineering. We already know that many bio-

Fig. 1. A generalized Venn diagram withthree sets of activities, (i) sampling processingand cell separation, (ii) DNA sequencing,annotation and analysis and (iii) functional (interms of activity screens) and phylogeneticanalysis, and their intersections. Theinterconnection among activities is crucial toget a proper analysis of microbial reactomeswith Systems Biology implications.Ecosystem

functioning

Genome size

Read length

G + C content

Assembly

Gene prediction

Annotation and analysis

Sequencingtechnology

Sequencingcoverage

Computing

Binning

Storage

Sampling

Cell preparation

Functionalcomposition

Phylogenteiccomposition

Rate of evolution

Genome size

Read length

G + C content

Assembly

Gene prediction

Annotation and analysis

Sequencingtechnology

Sequencingcoverage

Enriched community

Bulk sample

Single cell

Pure cultures

DNA isolation

Co-cultures

Single cell

Substrate spectra

Physiology

Biochemistry

Gene evolution

Protein evolution

Shot-gun

454 pyroequencing

Completedgenomes

Unassambledscaffolds

Computing

Binning

StorageCommunity driven

Bioinformatics

Database

Crystal ball 135


chemical reactions can be catalysed by two or more dif-ferent enzyme variants. We also know that metabolicpathways in any given organism have evolved to optimizethe organism’s growth, not the overproduction of anyparticular metabolite that we might want it to produce.Incorporating foreign genes could be used to steer themetabolism in the needed direction, to remove inconve-nient by-products, to relieve feedback inhibition, and toadapt the metabolic pathway to particular environmentalconditions (t°, pH, salinity). For example, the flux throughthe standard glycolytic pathway of Escherichia coli couldbe manipulated by introducing the ADP-dependent phos-phofructokinase from, metal-independent aldolase and/or bisphosphoglycerate-independent phosphoglyceratemutase from Methanococcus maripaludis or other meso-philic archaea. The abundance of alternative enzyme ver-sions from exotic and poorly studied microorganisms willbe complemented by the abundance of suitable hostscapable of deriving energy from the solar light (photosyn-thetic bacteria, including cyanobacteria) or cheap sub-strates, such as natural gas, methanol, sawdust andtimber waste. This combination will bring metabolic engi-neering to an entirely new level, allowing the constructionof customized organisms for every ecological niche thatwould consume industrial waste and convert it into usefulproducts.

Very soon, microbial metabolic engineering will be usedto improve our food supply, solve the energy crisis andfight global warming. We already have completelysequenced genomes of several nitrogen-fixing endophyticbacteria that enable the rapid growth of sugarcane andvarious legumes (Krause et al., 2006; Fouts et al., 2008;Lee et al., 2008). Adapting such bacteria to corn, wheat,rice and soy will dramatically decrease the need forchemical fertilizers, allowing rapid growth of plant biomassand, as an added benefit, increased consumption of CO2.Plant foods derived this way could be enriched in essen-tial amino acids and vitamins without carrying the stigmaof ‘Frankenfoods’. This will decrease the need for animalprotein and provide yet another way to decrease the pro-duction of greenhouse gases.

The next step will be using bacteria to improve humanbodies. We already affect human gut microflora with ourfoods and change it by consuming yogurts, beer, brie,kimchi, and other products that contain live microbial cul-tures. Enriching yogurts with vitamin-producing bacteriawill go a long way towards eliminating various vitamindeficiencies. The next step will be introducing engineeredbacteria in human tissues and even human cells. If theplans to use specially constructed clostridia for curing(or at least slowing down) cancer (Wei et al., 2008) bringeven modest success, they will pave the way for furthergene therapy. If aphids, nematodes and fruit flies canafford carrying intracellular bacteria to supply them with

nutrients (Wernegreen, 2004), we sure can try thesame thing in order to cure hereditary diseases. Aphenylalanine-dependent symbiotic bacterium could beused to improve the life of patients with phenylketonuria.Lipid-degrading bacteria (mycobacteria?) might be usedto clear atherosclerotic plaques, tartrate-metabolizingbacteria to dissolve kidney stones, and lactate-consumingbacteria to relieve muscle fatigue. Microbes could also beengineered to maintain a healthy balance of neurotrans-mitters, replacing the morning cup of latte and secretingjust enough serotonin derivatives to keep the host con-stantly happy.

Having fought bacteria in the last century, we have nearlyexhausted the repertoire of available antibiotics and willhave to learn to co-exist with bacterial world. Knowledge ofbacterial genomics should allow us to separate friend fromfoe and harness them both for our own use.

Acknowledgements

The opinions expressed above do not reflect the views ofNCBI, National Library of Medicine, or the IntramuralResearch Program of the National Institutes of Health, whichsupports my research.

References

Dunham, I. (2000) Genomics – the new rock and roll? TrendsGenet 16: 456–461.

Fouts, D.E., Tyler, H.L., DeBoy, R.T., Daugherty, S., Ren, Q.,Badger, J.H., et al. (2008) Complete genome sequence ofthe N2-fixing broad host range endophyte Klebsiella pneu-moniae 342 and virulence predictions verified in mice.PLoS Genet 4: e1000141.

Krause, A., Ramakumar, A., Bartels, D., Battistoni, F., Bekel,T., Boch, J., et al. (2006) Complete genome of the mutual-istic, N2-fixing grass endophyte Azoarcus sp. strain BH72.Nat Biotechnol 24: 1385–1391.

Lee, K.B., De Backer, P., Aono, T., Liu, C.T., Suzuki, S.,Suzuki, T., et al. (2008) The genome of the versatilenitrogen fixer Azorhizobium caulinodans ORS571. BMCGenomics 9: 271.

Wei, M.Q., Mengesha, A., Good, D., and Anne, J. (2008)Bacterial targeted tumour therapy-drawn of a new era.Cancer Lett 259: 16–27.

Wernegreen, J.J. (2004) Endosymbiosis: lessons in conflictresolution. PLoS Biol 2: e68.

Note: This article is a US Government work and is in thepublic domain in the USA.

Optimizing rational vaccine design

Carlos A. Guzman, Department of Vaccinology andApplied Microbiology, Helmholtz Centre for InfectionResearch, Braunschweig, Germany.Vaccines are the most cost-efficient tool to preventinfectious diseases, and their therapeutic use, against

136 Crystal ball


both infectious and non-infectious (e.g. cancer, auto-immunity) diseases, is gaining considerable interest.However, despite major advances in the fields of microbialpathogenesis, immunology and vaccinology, there are stillmany diseases for which vaccines are not available or theavailable vaccines are inadequate in terms of efficacyand/or safety. This is particularly true for chronic or per-sisting infections. In the post genomic era all potentialantigens, which are coming into consideration for inclu-sion into a vaccine formulation, are well known. Thisknowledge has been exploited in the context of reversevaccinology-driven approaches, which in combinationwith comparative genomics enabled to select the mosthighly conserved and promising antigens for vaccinedesign. However, the advent of new vaccines againstdiseases such as AIDS, chronic hepatitis or malaria, aswell as improved vaccines against ‘old diseases’, such astuberculosis, is well overdue. It is obvious that extremelyoptimistic end-points for vaccination against these agents,such as the stimulation of sterilizing immunity, should bereplaced by more realistic goals, like the stimulation ofimmune responses able to delay disease onset or pro-gression. However, this is not the key issue. Where thenlay the most critical roadblocks preventing the develop-ment of effective immune interventions against the agentscausing these diseases?

The first roadblock is that our knowledge on the effectormechanisms responsible for the clearance of these patho-gens is by and large fragmentary. In-depth studies ofnatural infections represent the best strategy to accessthis knowledge. There are individuals who are refractoryto infection (e.g. multiple exposed uninfected individualsfor HIV) or develop slow progressing forms of disease(e.g. long-term non-progressors for HIV, chronicallyinfected patients without liver cirrhosis for hepatitis). Well-defined patient cohorts with different forms of diseasewere established in recent years, which are being char-acterized in terms of their genetic, microbiological andimmunological profiles. This is expected to lead to biom-arkers and molecular/phenotypic signatures associatedwith better prognosis, as well as to the identification of theeffector mechanisms responsible for microbial clearance.This knowledge base will considerably facilitate andaccelerate rational vaccine design.

Let us consider for an instant an ideal scenario in whichthe first roadblock has been overcome. It is exactly knownwhich antigens need to be included in the formulation andwhich kind of effector mechanism should be stimulated toconfer protection. Considering the present state of the art,a subunit vaccine will probably be the strategy of choice,as the replacement of whole cell vaccines or semi-crudeantigen preparations by well-defined antigens has dra-matically improved their safety profile. At this point we willface the second roadblock; namely the availability of tools

enabling the stimulation of predictable immune responsesof the adequate quality following vaccination. In fact,highly purified antigens are often less immunogenic thanmore complex preparations, rendering essential theirco-administration with potent adjuvants. These com-pounds also have immune modulatory properties, whichallow to fine tune the responses elicited. This is criticalissue since the stimulation of a wrong response patternmay even lead to more severe forms of disease. However,despite the fact that there are several adjuvants underdevelopment, the sad truth is that only a handful of themhave been licensed for human use (i.e. Alum, MF59 andMPL; Tagliabue and Rappuoli (2008). This is far worseif compounds exhibiting activity when administered bymucosal route are considered, from which only a fewcandidates are in the development pipeline (Rharbaouiand Guzmán, 2005; Ebensen and Guzmán, 2008).Hence, there is a critical need for novel adjuvants,particularly those exerting their biological activities whenadministered by mucosal route. This is very important, asmost pathogens enter the host via the mucosal tissues.Thus, the stimulation of an effective local response wouldalso enable to block infectious agents at their portal ofentry, thereby reducing their capacity to colonize and befurther transmitted to other susceptible hosts. It isexpected that in the coming years we will see a newgeneration of well-defined and highly efficient adjuvantscoming in the market. This will facilitate the developmentof a new generation of more effective vaccines, as theavailability of adjuvants exhibiting different biologicalproperties will allow efficient fine-tuning of the immuneresponses elicited according to specific clinicalneeds.

The third roadblock is related to the need to bridge thetranslational gap, as well as to current stringent regula-tions for vaccine testing (e.g. requirement of GMP gradematerial for phase I studies), which have in turn led to anexplosive increase in clinical development costs. To accel-erate translation novel strategies are needed for a rapidand cost-efficient screening, selection and prioritization ofthe most promising candidates. For certain pathogens themost widely accepted animal model are primates (e.g.HIV, HCV). However, one of the most significant issuesassociated with these animal models is that they do notcompletely reproduce the pathophysiology of human dis-eases. Reproducibility is also an issue, as they suffergreatly by the small number of animals that can be studiedat any time and by inter-individual variability, which limittheir statistical power. Furthermore, primate models areoften too expensive and fraught with ethical constraints.Thus, none of the existing models adequately address theneeds of the vaccine developer. Hence, there is a clearneed for cost-efficient small animal models to addressthese limitations.

Crystal ball 137


In this context, mice are ideally suited to perform theinitial validation of vaccine candidates in a cost-efficientmanner. However, the results obtained in mouse-basedsystems cannot always be extrapolated to humans. A verypromising alternative strategy consists in the engraftmentof components of the human immune system into immunecompromised mice (Shultz et al., 2007; Legrand et al.,2008). When these animals are engrafted with liver orcord blood derived stem cells, proper development of NKcells, B cells, dendritic cells and different T-cell subsets(e.g. CD4+, CD8+, Treg) is obtained. While still experienc-ing some limitations, these human/mouse chimeras arepermissive to infection by different infectious agents,including the HIV (Baenziger et al., 2006; An et al., 2007).However, there is still margin for further development,such as the improvement of adaptive cellular responses.It is also critical to ensure that they fulfil with the keyfeatures of good animal models, namely ensure theirreproducibility and an adequate high throughput, performthoroughly validation with known human vaccines, andmade them available at an acceptable cost respect totheir benefit. Nevertheless, these aspects will be fullyaddressed in the coming years, thereby enabling theirroutine application for vaccine preclinical validation. It isexpected that the use of these advanced animal modelsfor vaccine testing will result in increased predictability fortheir performance in humans, thereby enabling a rapidand efficient selection of the best candidates to be trans-ferred into the clinical development pipeline.

References

An, D.S., Poon, B., Ho Tsong Fang, R., Weijer, K., Blom, B.,Spits, H., et al. (2007) Use of a novel chimeric mousemodel with a functionally active human immune system tostudy human immunodeficiency virus Type 1 infection. ClinVaccine Immunol 14: 391–396

Baenziger, S., Tussiwand, R., Schlaepfer, E., Mazzucchelli,L., Heikenwalder, M., Kurrer, M.O., et al. (2006) Dissemi-nated and sustained HIV infection in CD34+ cord bloodcell-transplanted Rag2-/-gamma c-/- mice. Proc Natl AcadSci USA 103: 15951–15956.

Ebensen, T., and Guzmán, C.A. (2008) Immune modulatorswith defined molecular targets: cornerstone to optimizerational vaccine design. Hum Vaccin 4: 13–22.

Legrand, N., Weijer, K., and Spits, H. (2008) Experimentalmodel for the study of the human immune system:production and monitoring of ‘human immune system’Rag2-/-gamma c-/- mice. Methods Mol Biol 415: 65–82.

Rharbaoui, F., and Guzmán, C.A. (2005) New generation ofimmune modulators based on toll-like receptor signaling.Curr Immunol Rev 1: 107–118.

Shultz, L.D., Ishikawa, F., and Greiner, D.L. (2007) Human-ized mice in translational biomedical research. Nat RevImmunol 7: 118–130.

Tagliabue, A., and Rappuoli, R. (2008) Vaccine adjuvants: thedream becomes real. Hum Vaccin 4: 347–349.

Metagenetics: spending our inheritance onthe future

Jo Handelsman, Department of Bacteriology, University ofWisconsin, Madison, WI 53706, USA.

Consider a proposal

Just as genetics indelibly shaped our understanding ofsolitary bacterial existence, so it can transform our under-standing of bacteria as they engage in community life.This will require a new application of mutant analyses in acommunity context. Let’s call this ‘metagenetics’, to high-light the concept of an analysis that transcends individuals(‘meta’ in Greek means ‘transcendent’). Metageneticsprovides a parallel with metagenomics – genetics andgenomics deal with single organisms and metageneticsand metagenomics both apply to analysis of a multi-genome unit, or community.

Consider the past

The glory of the last 50 years of microbiology is founded,in large part, on genetic analysis. Every aspect of cellularbacterial life has been cracked by the use of mutants.Metabolic pathways have been defined by analyses ofmutants blocked in various biochemical steps. Macromo-lecular synthesis, membrane function and chemotaxisyielded to the chisel of genetics. Later, the study of proteinstructure and function emerged based on precise, singleamino acid changes generated by point mutations.

The mark of microbial genetics extends beyond under-standing the working of the bacterial cell. The foundationfor microbial evolution was provided by the classic Luria–Delbrück fluctuation test and the Lederbergs’ replicaplating experiment. Both provided irrefutable evidencethat bacterial evolution depends on pre-existing mutationsthat are independent of selection pressure. The impact ofthese landmark experiments was felt throughout biologybecause they presented potent fortification for the Darwin-ian concept of evolution that pre-existing variation inpopulations that are acted upon by natural selection.

In the early days of bacterial genetics, classical crossesand complementation analysis were accomplishedthrough conjugation, transformation, and transduction,fostering associations between genes, functions, and ulti-mately proteins. Genetics lost some of its abstract natureand advanced to a new level when it became possible tophysically isolate genes by cloning. The advent of DNAsequencing generated a new depth of understanding ofthe nature of mutations, making mutant analysis morepowerful than ever. The satisfying level of precision pro-vided by molecular genetic analysis has created a goldstandard of proof in modern microbiology. This, in turn,

138 Crystal ball


has generated a two-class distinction of sub-fields ofmicrobiology. Sadly, ecology has been largely relegated tothe less desirable class by many of those who studysolitary bacterial life because they find the types of evi-dence and structure of arguments in ecological study tolack the precision to which they are accustomed. All ofthat can change.

Consider the future

Metagenetics will dissect ecological questions at a newlevel of precision. But unlike the early days of bacterialgenetics, the new field of metagenetics will be buttressedby vast databases of sequence from metagenomic analy-sis. Metagenomics will generate hypotheses to be testedwith genetics as well as the sequence information onwhich to base mutant construction.

Metagenetics will need to embrace both randommutagenesis, which is a way of giving voice to the bacte-ria, and directed mutant analysis, which is driven (andlimited by) the imagination of the investigator. In the first,we will mutagenize a pure culture and then screen therandom mutants for a community phenotype. In thesecond, we will create a defined mutant in a gene ofinterest and determine its phenotype in the communitycontext. We might, for example screen a randomlymutagenized population of bacteria for the loss of ability toinvade a community or we might construct a mutantlacking flagella and determine whether it is affected inthe ability to invade. Both approaches will be facilitatedby available genome maps, sequence information andextensive ‘-omics’ (transcriptomics, proteomics andmetabolomics) data.

Metagenetics on culturable community members maynot seem all that different from classical genetics exceptin the nature of the phenotype tested. But the boldadvance will issue from development of genetic tools tostudy unculturable members. For example, imagine thepower of knocking out all homologues of a particular genein all members of the community? What would happen ifwe knocked out all of the polysaccharide biosynthesisgenes in a community? Alternatively, what if we knockedthem out only in one family of bacteria? Highly specificconjugal vectors and sequence-based homing devicescan make these approaches reality. Metagenomics willfurnish the raw material for such studies – sequenceinformation from the unculturable members of the com-munity that will form the basis for generating hypothesesand genetic devices to make targeted changes.

Metagenetics, in concert with the other tools of theecologist, including statistics, modelling, microscopy,radioactive labels, chemical analysis and meta-omics,will elevate the level of rigor and precision with which wecan approach community-level microbial ecology. Just as

early bacterial genetics provided critical data beyondmicrobiology, to the entire field of evolution, microbialmetagenetics may advance the entire field of ecology byanswering questions at a level and with tools that are notpossible in macroecological systems. The lessons from50 years of bacterial genetics are powerful. Perhaps 50years from now we will be reflecting on the advancementof ecology by a parallel metagenetic approach.

Future shock from the microbe electric

Derek R. Lovley, Department of Microbiology, Universityof Massachusetts, Amherst, MA 01003, USA.How can the future not look bright when you are dealingwith a microbial process that can power a light bulb? Thestudy of microbial fuel cells and, more generally, microbe–electrode interactions is rapidly amping up, not only inpower production, but also in the number of investigatorsand areas of study.

The most intense focus has been on wastewater treat-ment and this is likely to continue for some time. It wasprobably safe to say 5 years ago that any compound thatmicroorganisms can degrade could be converted to elec-tricity in a microbial fuel cell, but if there was ever was anydoubt, this point has been proven over and over again ina plethora of recent studies. It is clear from this work thata major limitation in converting complex wastes to elec-tricity is the initial microbial attack on the larger, difficult toaccess molecules, just as it is in any other treatmentoption. It may well be that the intensive focus on thedegradation of complex organic matter in other bioenegyfields will soon make a contribution here.

However, there are other issues specific to microbialfuel cell technology. At present the rate that even simpleorganic compounds can be converted to electricity ismuch too slow for practical wastewater treatment. Forexample, columbic efficiency (i.e. the percentage of elec-trons available in the organic substrate that are recoveredas current) is often diminished by methane production,indicating that even relatively slow-growing methanogensare competing with the current-producing microorgan-isms. This is despite the fact that electron transfer tooxygen, the ultimate electron acceptor in microbial fuelcells, is much more thermodynamically favourable thanmethane production.

Some contend that the limitations to current productionin waste treatment can be solved with improved engineer-ing of microbial fuel cell design and that there is little needto focus on the microbiology of microbial fuel cells forwaste treatment because as better fuel cell designs aredeveloped, the appropriate microorganisms will naturallycolonize the systems and produce more power. That maybe, but it also seems likely that, going forward, the mecha-nisms for microbe–electrode interactions will become

Crystal ball 139


better understood and this could significantly informoptimal microbial fuel cell design.

Furthermore, it is likely that we will find that it is possibleto greatly increase the current-producing capabilities ofmicroorganisms. This is because there has been no pre-vious evolutionary pressure for microorganisms to opti-mally produce current. Many of the microorganisms thatfunction best in microbial fuel cells are dissimilatoryFe(III)-reducing microorganisms, which have evolved tospecialize in extracellular electron transfer to insoluble,extracellular electron acceptors. However, microorgan-isms reducing Fe(III) in sedimentary environments aretypically in direct contact with the Fe(III). In contrast, whenmicroorganisms are producing high current densities inmicrobial fuel cells, only a small fraction of the microor-ganisms in the anode biofilm are in direct contact with theanode surface. Most must transfer electrons over sub-stantial distances through the biofilm. It is not clear thatthere has ever been substantial selective pressure onmicroorganisms for such long-range electron transfer.Thus, there should be ample room for improvement.

Another unnatural request that we make on micro-organisms when they are asked to generate high currentdensities is the requirement to metabolize organic com-pounds very rapidly. The natural habitat of most of themicroorganisms that have been shown to be most effec-tive in current production is the subsurface or aquaticsediments. These are rather low-energy environments inwhich there has probably not been much selective pres-sure for rapid growth and metabolism. Other challenges toanode-reducing microorganisms include the necessity totolerate the low pH that can develop within the anodebiofilm. This results from the fact that protons as well aselectrons are released from organic matter oxidation.

Strains that can better respond to these unusualdemands of high density current production will certainlybe found or developed. Understanding what characteris-tics of these strains confer enhanced current-productioncapability may aid in fuel cell design and these strainsmay be beneficial in some applications. Strain improve-ment may include attempts to select better strains fromcomplex microbial communities as well as geneticengineering and adaptive evolution approaches. Somedegree of strain selection has taken place in previousstudies in which conditions conducive to high current den-sities have been established in microbial fuel cells and thesystems have been inoculated with sewage or some othercomplex community. The surprising result from a numberof laboratories is that such conditions frequently select forGeobacter sulfurreducens, or closely related strains. Purecultures of G. sulfurreducens can produce current densi-ties as high as any known pure or mixed culture. We havehad moderate success in genetically engineering strainsof G. sulfurreducens for higher rates of respiration and

extracellular electron transfer, guided by a genome-scalein silico metabolic model. However, electron transfer toelectrodes appears to be a complex process, and may notbe well enough understood to rationally engineer. Adap-tive evolution has proven to be a much more promisingapproach for strain development and major enhance-ments in power production with this tactic are forthcoming.

As with any optimization procedure, once one bottle-neck is relieved another emerges. As better current-producing strains of G. sulfurreducens have beendeveloped, it has been necessary to use exceedinglysmall anodes relative to cathode area in order to keepreactions at the cathode from limiting rates of electrontransfer at the anode. The ability of microorganisms toaccept electrons from a cathode to support anaerobicrespiration has already been demonstrated and studiesin a number of laboratories have found that aerobiccathodes selectively enrich for specific microorganismsthat might promote faster rates of electron transfer fromthe cathode to oxygen. This is likely to be an area ofintense interest in the near future. It will probably bepossible to develop microbes with superior capabilities foraccepting electrons from cathodes with the reduction ofoxygen with approaches similar to those discussed abovefor improving the current-producing capabilities of anode-reducing microorganisms.

What if engineering and microbiology do not overcomethe barriers to making microbial fuel cell technology suit-able for wastewater treatment? There are many otherpotential applications for microbe–electrode technology.One near-term application is harvesting electricity fromwaste organic matter or vegetation to power electronics inremote locations. Sediment microbial fuel cells that powermonitoring devices at the bottom of the ocean are alreadyfeasible. Self-feeding robots that run on microbial fuelcells have also been proven in prototype. There are manyother applications in which relatively low power require-ments can probably be met with microbial fuel cells. Forexample, there are already several organizations plan-ning to distribute in developing countries inexpensivemicrobial fuel cells that run on wastes and can providelighting or charge electronic devices. A number ofresearch teams are working on developing implantedmedical devices that use blood sugar as a fuel. It seemslikely that many other applications that require low levelsof electrical current but for which it is difficult to install orcontinually replace traditional batteries could be helpedwith microbial fuel cell technology. Future applicationsmay also include microbial transistors, circuits and elec-tronic computing devices, among others.

Environmental technology is likely to be another emerg-ing field for microbe–electrode interaction applications.Anodes are attractive electron acceptors for stimulatingthe degradation of contaminants in the subsurface

140 Crystal ball


because they can be emplaced as a permanent, high-potential, electron acceptor and can adsorb and concen-trate many contaminants to co-localize pollutants and theelectron acceptor. Current produced from electrodesdeployed in anoxic subsurface environments is likely toprove to be a good proxy for estimating rates of microbialmetabolism in those environments. Cathodic reactionsare also likely to see more application in bioremediationand waste treatment. The potential for stimulating micro-bial reduction of nitrate, U(VI), and chlorinated contami-nants with electrodes serving as the electron donor hasalready been demonstrated and field application of thesetechnologies are on the horizon.

One of the most exciting areas of future research isalmost certain to be the production of specialty chemicalswith cathodic microorganisms accepting electrons froman electrode. Fixation of carbon dioxide and its conver-sion into useful organic commodities powered by elec-trons supplied directly from an electrode may prove to beone of the most lucrative applications of microbe–electroninteractions in the near future. This process is clearlythermodynamically feasible, and the ability for microor-ganisms to accept electrons for anaerobic respiration hasalready been demonstrated. It just remains to be seenwhether the appropriate microorganisms for this applica-tion exist in nature or whether extensive metabolic engi-neering will be required.

In summary, it would be shocking if the continuedincreased intensity of study on microbe–electrode inter-actions did not shed light on additional applications aswell as illuminate more of the basic mechanisms by whichmicroorganisms electronically interact with electrodes.The future of this biotechnology looks very bright indeed.

Extensive referencing to recent research on microbe–electrode interactions can be found at the following websites:http://www.microbialfuelcell.orghttp://www.geobacter.org

Listening to microbial conversations

Michael J. McInerney, Department of Botany and Micro-biology, University of Oklahoma, 770 Van Vleet Oval,Norman, OK 73019, USA.The microbial world contains a vast and untapped res-ervoir of genetic diversity that could be used for theproduction of novel, biologically active molecules or todevelop new strategies to manipulate the activities ofmicrobial consortia. However, we need to understand howmicrobial species interact and communicate with eachother in order to manipulate their interactions. Withpyrosequencing and other technological breakthroughs,we are beginning to understand ‘who is there’ and ‘whatthey are capable of doing’. What we need to do is to get

better at understanding ‘what they are doing’ to exploitfully the diversity of the microbial world. We have madegreat strides in computational approaches that allow us toassign putative functions to many of the genes presentin microbial genomes; but, even with these tools, manycoding regions lack functional assignments. Manygenomes contain ‘cryptic’ or ‘orphan’ gene clusters withthe potential to produce novel and structurally complexchemicals (Challis, 2008; Fischbach et al., 2008). Thesechemicals do not have ‘housekeeping’ functions, butprobably function as signals mediating interactionsamong microorganisms and between microorganismsand eukaryotes (Straight et al., 2006; Dietrich et al., 2008;Fischbach et al., 2008). These studies suggest thatmicrobes are carrying on a conversation with each other.We must listen to and translate this conversation to under-stand how microbial species interact with each other.Once we understand what microbes are saying and why,then we can manipulate the conversations.

We should not be surprised that microbes conversebecause we know that microbial species work in teams orguilds. These interspecies interactions must be coordi-nated, which means that there must be specific signals. Inthe future, we will have a variety of high-throughput toolsto identify how microbial populations respond to eachother and what molecules are used. Such approachesmay be analogous to microarray technologies such asGeoChip (He et al., 2007). Computational approaches willbe available to identify the key regulatory componentsthat receive, translate and transmit the signal to action. Asthe regulatory networks are defined, we will be able toidentify the input chemical stimulus, its receptor and howthe signal is transmitted within the cell. Bioengineers canthen construct multi-component systems from libraries ofstandard interchangeable parts engineered from the com-ponents identified during the ecological screening pro-cess. Professor Endy and his colleagues (Canton et al.,2008) have developed BioBrick (http://partsregistry.org/),a standard biological parts inventory, which includesprotein coding sequences and regulatory elements forgene expression and signalling and have defined quanti-tative measures of performance that will allow bioengi-neers to use these parts reliably. Future efforts willcertainly expand on the parts and chassis (organisms)available for manipulation. By understanding how biosyn-thetic genes change, move about and recombine, we canunderstand the processes that generate small-moleculediversity.

Once we understand the microbial conversations, wewill have the ability to manipulate the response of a spe-cific microbe or of microbial communities. We will be ableto identify signals to turn on gene systems to produce newbiologically active molecules that could be used as anti-biotics or anticancer drugs. Additionally, we may identify

Crystal ball 141


http://www.microbialfuelcell.org

http://www.geobacter.org

http://partsregistry.org

signals to turn on specific functions in complex communi-ties. Understanding why microorganisms make biosurfac-tants may provide an approach to turn biosurfactantproduction on in oil wells to enhance oil recovery (Youssefet al., 2007). Alternatively, we should be able to disrupt themicrobial conversation to prevent unwanted interactionsinvolved in disease or corrosion.

Acknowledgements

I thank Matthew F. Traxler for his insights into microbialcommunication.

References

Canton, B., Labno, A., and Endy, D. (2008) Refinement andstandardization of synthetic biological parts and devices.Nat Biotechnol 26: 787–793.

Challis, G.L. (2008) Mining microbial genomes for newnatural products and biosynthetic pathways. Microbiol 154:1555–1569.

Dietrich, L.E.P., Teal, T.K., Price-Whelan, A., and Newman,D.K. (2008) Redox-active antibiotics control gene expres-sion and community behavior in divergent bacteria.Science 312: 1203–1206.

Fischbach, M.A., Walsh, C.T., and Clardy, J. (2008) Theevolution of gene collectives: how natural selection driveschemical innovation. Proc Natl Acad Sci USA 104: 4601–4608.

He, Z., Gentry, T.J., Schadt, C.W., Wu, L., Liebich, J., Chong,S.C., et al. (2007) GeoChip: a comprehensive microarrayfor investigating biogeochemical, ecological, and environ-mental processes. ISME J 1: 67–77.

Straight, P.D., Willey, J.M., and Kolter, R. (2006) Interactionsbetween Streptomyces coelicolor and Bacillus subtilis: roleof surfactants in raising aerial structures. J Bacteriol 188:4918–4925.

Youssef, N., Simpson, D.R., Duncan, K.E., McInerney, M.J.,Folmsbee, M., Fincher, T., and Knapp, R.M. (2007) In-situbiosurfactant production by injected Bacillus strains in alimestone petroleum reservoir. Appl Environ Microbiol 73:1239–1247.

Microbial biotechnology meetsenvironmental microbiology

J. Colin Murrell, Department of Biological Sciences, Uni-versity of Warwick, Coventry, UK.Thomas J. Smith, Biomedical Research Centre, SheffieldHallam University, Sheffield, UK.The isolation of commercially valuable bacteria from theenvironment has been a cornerstone of microbial biotech-nology for many decades. The environment has yieldedorganisms capable of producing valuable fermentationproducts such as alcohols and amino acids, strains ableto produce diverse pharmacologically active secondarymetabolites, as well as microorganisms that can affect

highly selective chemical transformations and convertrecalcitrant pollutants into non-toxic metabolites. AsMicrobial Biotechnology (sister journal to the now wellestablished Environmental Microbiology) celebrates itsfirst birthday, we would like to speculate upon the way inwhich the ongoing revolution in the characterization ofuncultured environmental microorganisms may facilitatediscovery of valuable microbial enzymes and pathwaysthat are currently beyond reach. The great majority ofmicroorganisms in natural environments have never beenobtained in pure culture and represent an importantsource of microbial diversity that biotechnologists cannotignore. Cloning of environmental DNA (metagenomics)has already emerged as a rich source of new biocatalystsfor production of bulk and high-value chemicals (reviewedin Steele et al., 2009). Metagenomics is reliant on thecloning of genes from complex samples that contain DNAfrom all manner of organisms, some relevant to the bio-technologist but most otherwise. Hence methodology forspecifically increasing the abundance of functional genes(genes encoding key target enzymes) of interest would beof great value in increasing the proportion of the relevantbiodiversity that could be accessed. In the sphere ofenvironmental microbiology, stable isotope probing (SIP)techniques employ enrichment cultures containing a 13C-labelled growth substrate, in which the DNA of organismsgrowing on the labelled substrate becomes enriched inthe heavy isotope and can be separated from bulk envi-ronmental DNA by means of CsCl density gradientcentrifugation. Originally developed to identify organismsactively metabolizing one-carbon compounds via analysisof 16S rRNA and functional genes, SIP has since beenapplied to characterize microorganisms utilizing a widerange of microbiological growth substrates (reviewed inDumont and Murrell, 2005; Friedrich, 2006). In principle,SIP is ideal for increasing the abundance of target genesfor subsequent direct cloning or amplification by means ofPCR. A report from Daniel and co-workers (Schwarz et al.,2006) was the first that indicated the feasibility of suchapplications. Daniel and co-workers focused on glyceroldehydratase, a key enzyme during biosynthesis of thevaluable product propane-1,3-diol. SIP with glycerol-13C3

led to an increase of up to 3.8-fold in the frequency ofrecovery of glycerol dehydratase genes per megabaseof cloned environmental DNA, compared with parallelmetagenomics experiments where SIP enrichment wasnot used. While the increase in sensitivity that SIP yieldedin this pilot study was modest, we predict that throughcareful manipulations of enrichment conditions, SIP andrelated techniques can be developed into a key tool ingene mining. DNA-SIP would give access to valuablefunctional genes that are present at very low abundancein inhospitable extreme environments or at low levels incomplex ecosystems and which are below the threshold

142 Crystal ball


of detection of current technology. This would generate apool of potentially novel target genes that could then bescreened in expression libraries or used in gene shufflingexperiments in order to generate novel biocatalysts. Inaddition, heavy DNA in the SIP experiments will becomeenriched in the genomes of target organisms, thus allow-ing focussed or targeted metagenomics, and isolation ofpotentially novel catabolic (or biosynthetic) gene clustersof biotechnological relevance.

References

Dumont, M.G., and Murrell, J.C. (2005) Stable isotopeprobing – linking microbial identity to function. NatureReviews Microbiol 3: 499–504.

Friedrich, M.W. (2006) Stable-isotope probing of DNA:insights into the function of uncultivated microorganismsfrom isotopically labeled metagenomes. Curr Opin Biotech-nol 17: 59–66.

Schwarz, S., Waschkowitz, T., and Daniel, R. (2006)Enhancement of gene detection frequencies by combiningDNA-based stable-isotope probing with the construction ofmetagenomic DNA libraries. World J Microbiol Biotechnol22: 363–367.

Steele, H.L., Jaeger, K.-E., Daniel, R., and Streit, W.R. (2009)Advances in recovery of novel biocatalysts from meta-genomes. J Mol Microbiol Biotechnol 16: 25–37.

Building bugs

Sven Panke, Department for Biosystems Science andEngineering, ETH Zurich, Basel, Switzerland.Biotechnology is one field that has profited immenselyfrom the advent of ‘-omics’ technologies. For example,fluxomics brought a better understanding of the flow ofmetabolites through the cellular metabolic network,transcriptomics and proteomics helped us appreciate themultiple consequences of gene overexpression, andgenomics allowed us to catalogue mutations in high-performer strains and transfer them to new strains. But,while our understanding of the system-wide effects of ourcurrent rather subtle strain modifications continuouslygrows, this change in scope does not yet extend to themanipulation of bio(techno)logical systems. Broadlyspeaking, we still paste a few genes into a plasmid, insertit into our pet-strain, and hope for the best.

However, if it is system-wide consequences that weneed to take into consideration, it is most probablysystem-wide action that we need to take to design trulyeffective biotechnological systems. I argue that a majorline of research in the next years and decades will dealwith our enabling of biological system engineering andproviding the corresponding arsenal of tools. I predictenabling on three levels: technical, theoretical andorganizational.

The first step in this transformation to system levelmanipulation is easy to spot: de novo DNA synthesis. The

technology as such is not new, but it has become now socheap that it is about to make the crucial step out of theindustrial laboratories into the world of academic researchas a routine tool. Moreover, the success of de novo DNAsynthesis by assembling entire genes from oligonucle-otides has re-ignited the search for novel DNA synthesistechnologies that might in the future help to bring costsdown further and directly accessible DNA sequenceslonger. Currently, the price of a bp in a synthesized genehalves every 2–3 years, and it is only a question of timewhen the full force of this technology will drive the art ofcloning out of our laboratory.

Of course, the next step will then be to go from singlegenes to novel entire pathways or even novel genomicsections or entire genomes. The required methods are notyet routinely available, but improvements in DNA synthe-sis technology and the recruitment of the proper biologicaltools – such as homologous DNA recombination toassemble DNA fragments in vivo and in vitro – suggestthat the corresponding problems will be solved ratherquickly (Gibson et al., 2008).

But where a slow assembly process used to allow thestep-by-step verification of underlying scientific assump-tions, a 50 kbp sequence that will be delivered 4 weeksfrom now does no longer allow such luxury. It will becomeexceedingly important (i) to integrate all available informa-tion into the sequence already at the start; (ii) to usepredictive tools to substitute for the missing information;(iii) to develop the corresponding experimental technologyto obtain the remaining indispensable data rapidly; and(iv) to make sure that the host that is to receive the DNAsequence can read out the information in a predictableand reliable fashion.

The first point is at its heart an organizational challenge.The design-relevant information for one promoter, oneribosome binding site, one RNAse site or one transcrip-tional terminator sequence might be available in the lit-erature, but locating and exploiting it is currently anachievement in itself, and it is even more so for the 50genes on the ordered DNA sequence. To make it availablefor engineering – that is a rational selection of a standardelement based on quantitative criteria – this informationneeds to be made available centrally, such as it is the goalof the Registry of Standard Biological Parts (http://partsregistry.org). Of course, the ‘standard’ part would notonly encompass requirements for the presentation andcompleteness of data and information, but it will extend tothe data’s generation, preferably as a part of the opera-tions of such a facility.

Reliable standardization will also be of crucial impor-tance in the reliable use of computational tools to predictthe behaviour of the functions encoded on our artificialDNA sequence (Marchisio and Stelling, 2008). But evenmore, just as CAD technology helps designing anything

Crystal ball 143




from houses to mechanical engineering artifacts by hidinga huge body of knowledge behind the interface, a biotech-CAD will help to recruit the system design knowledge thatis available from, for example, electrical engineering into abest practice for biological systems design: What is themost effective way to engineer an oscillation? Or a regu-latory circuit that makes signal output dependent on theconcomitant availability of two signals (an AND gate)?

Clearly, much of the work that is required for thepredicted transformation to systems level is in a senserepetitive: for example, results on ribosome binding sitestrength for many sites need to be verified under variousgrowth conditions and with sufficient redundancy to bestatistically relevant. Or long DNA sequences need to beassembled step-by-step from shorter DNA elements.This work is in principle excellently suitable to auto-mation. Its reduction to micrometer dimensions and itsintegration into microfluidic systems is then the crucialstep that will make it affordable and allow the requiredparallelization.

While all of the three points above are well underwayalready today, the future of (iv) is much less clear. From anengineering point of view, the notion that every designedDNA sequence requires a tailor made host to interact withacts as a real deterrent. It will be much more attractive tohave hosts available that provide required resources (e.g.protein synthesis) but other than that behave orthogonalto (not or hardly influenced by) the introduced DNAsequence. We are currently far from understanding thecentral rules of orthogonal design in biotechnology, but itseems safe to say that it will depend on our ability tomanipulate chemical interfaces to remove and introduceinteractions at will. Already a range of techniques is avail-able that points the way to orthogonal engineering, eitherby removing unwanted interactions through genomereduction (Posfai et al., 2006) or working with in vitrosystems (Jewett et al., 2008), or by engineering novelorthogonal interactions by designing smart selectionschemes and then recruiting evolution (Rackham andChin, 2005).

In my view, to truly flourish, systems biotechnology willneed the future toolbox of synthetic biology. The corre-sponding changes will turn biotechnology into a true engi-neering discipline and finally produce in full the industrywe have been dreaming of for the last 30 years.

References

Gibson, D.G., Benders, G., Andrews-Pfannkoch, C., Den-isova, E.A., Baden-Tillson, H., Zaveri, J., et al. (2008) Com-plete chemical synthesis, assembly, and cloning of aMcyoplasma genitalium genome. Science 319: 1215–1220.

Jewett, M.C., Calhoun, K.A., Voloshin, A., Wuu, J.J., andSwartz, J.R. (2008) An integrated cell-free metabolic

platform for protein production and synthetic biology. MolSyst Biol 4: 220.

Marchisio, M.A., and Stelling, J. (2008) Computational designof synthetic gene circuits with composable parts. Bioinfor-matics 24: 1903–1910.

Posfai, G., Plunkett, G., Feher, T., Frisch, D., Keil, G.M.,Umenhoffer, K., et al. (2006) Emergent properties ofreduced-genome Escherichia coli. Science 312: 1044–1046.

Rackham, O., and Chin, J.W. (2005) A network of orthogonalribosome-mRNA pairs. Nature Chem Biol 1: 159–166.

Removal of organic toxic chemicals in therhizosphere and phyllosphere of plants

Juan L. Ramos, Lázaro Molina and Ana Segura, ConsejoSuperior de Investigaciones Científicas, Estación Experi-mental del Zaidín, Department of Environmental Microbi-ology, Granada, Spain.With the ever-growing increase in quality of life standardsand awareness about environmental issues, remediationof polluted sites has become top priority. Because of thehigh economic cost of physicochemical strategies forremediation, the use of biological tools to clean up con-taminated sites has turned out to be a very attractiveoption. The use of microorganisms associated to plantroots (rhizoremediation) and leaves in the removal of soiland air contaminants is an area in which success isexpected in the near future.

In recent years knowledge has been gathered on theremoval of contaminants by microbes living in plantniches. Plants provide a series of overlapping nichesfor microbial development, and culture enrichmentapproaches and new ‘-omic’ technologies have demon-strated that the number of microbes in the rhizosphere(soil around the roots) and phyllosphere (leave surfaces)of plants is larger than expected. On the other hand,metabolite analysis and stable isotope probe techniques,as well as other approaches have shown that microbesassociated to plants are metabolically active (Fig. 1). Theability of the microorganisms to proliferate to high densi-ties in the plant’s niche depends on the plant providing anappropriate surface for the microbes’ development and,most importantly, on providing nutrients that fulfil thecarbon, nitrogen and other elements demands, as well asenergy needs. Looking at microbes as bioremediationcatalysts, one can say that proliferation of microbes tohigh cell densities in the plant niches acts as a multiplierand can lead to an increase in the efficiency of pollutantremoval if the resident microbes are endowed with theappropriate catabolic potential.

The above positive view of bioremediation contrastswith some attempts that have been made to re-introducemicroorganisms in soils for pollutant removal, which haveturned out to be utterly unsuccessful. For the design of a

144 Crystal ball


successful rhizoremediation strategy it is necessary tofulfil at least two minimal requirements: microbes have tobe able to proliferate in the root/leave system and cata-bolic pathways need to be operative. With the advent ofmicro-array technology, global approaches in expressionof genes in the plant’s environment are coming to light.Several recent papers (Matilla et al., 2007; Attila et al.,2008) have demonstrated that almost 200 promoters arespecifically induced in different strains of the genusPseudomonas in the presence of root exudates or plantroots. These studies have revealed the mechanismsunderlying microbe–plant interactions and we predict thatthis knowledge will contribute to recognize the bestplant–bacteria combination and establish the optimalinduction of catabolic pathways in sites undergoingrhizoremediation. To further support our positive view ofprospects in bioremediation we can state that someproducts present in natural root exudates can act asinducers of different catabolic pathways for the degrada-tion of contaminants. Although plants produce a vastamount of secondary metabolites, not all plants canproduce every product and these are often generated

only during a specific developmental period of the plant.We predict future studies on root/leaf bacterial meta-bolomes and transcriptomes of plant-bacteria interac-tions during remediation to establish the best ways tointroduce catabolic pathways in sites undergoingremediation. Having said this, successful rhizospherecolonization does not only depend on the interactionsbetween the plant and the microorganism of interest, butalso on the interactions with other microorganisms. Newtechniques to study population changes have greatlyimproved over the last few years and they are and will beused to determine the changes that the introduction ofnew microorganisms in the ecosystem will cause andhow it might affect the sustainability of the ecosystem inlong run.

An important problem is that of reducing pollutants thatare associated to air particles. The main limitation prob-ably comes from the bioavailability of the pollutant depos-ited on the leaves’ surfaces. Most organic contaminantsare highly hydrophobic compounds that dissolve poorly inwater and many of them can form complexes with air-borne particles; this lack of bioavailability may lower

Bacterial community modifications

Bacteria-plant relationships

Transcriptomics

TGGE, DGGE, 16S libraries

Plant-bacteria relationships:

-Signals

-Inducers

Metabolomics

20.0022.0024.0026.0028.0030.0032.0034.0036.0038.0040.0042.0044.00

500000

1000000

1500000

2000000

2500000

3000000

3500000

4000000

Time-->

Abundance

TIC: ANA6H.D

Phenomics

Production of biosurfactants

Carbon source utilization

Catabolic properties

Fig. 1. Molecular approaches in rhizoremediation and phylloremediation.

Crystal ball 145


removal efficiency. Three recent papers have dealt withthe degradation of air pollutants, namely, toluene, phenoland phenanthrene (Molloy, 2006; Sandhu et al., 2007;Waight et al., 2007) and studies on the bioavailability ofpollutants and on the range of pollutants to be degradedwill appear in the next few years. We also envisageadvances in unveiling the strategies used by microbes toenhance the bioavailability of hydrophobic compounds[i.e. polycyclic aromatic hydrocarbons (PAHs)] via theproduction of biosurfactants, extracellular polymericsubstances or formation of biofilms. We also envisionresearch in the area of air decontamination to reveal thefull remediation potential of microbes in an area wherethere is little study.

It has been argued that beneficial plant endophytes,bacteria that colonize the internal tissues of the plantwithout causing negative effects, could be an alternativein bioremediation since microbes would be somehowphysically protected from adverse changes in the environ-ment. However, successful remediation by endophyticbacteria requires the transport of the contaminant to theplant’s interior. Research in this area will reveal whether ornot endophytes are of interest in bioremediation. Baracand colleagues (2004) showed improvement in toluenephytoremediation using engineered endophytic bacteria.The authors transferred the toluene-2 monooxygenase(TOM) pathway to an endophytic Burkholderia strain, anatural endophyte of yellow lupine. Although the authorsshowed that the strain was not maintained in the endo-phytic community, there was horizontal gene transfer ofthe tom genes to different members of the endogenousendophytic community, demonstrating new avenues tointroduce desirable traits into the community.

The use of plant/microorganisms in biorremediation hasalso got certain drawbacks; pollutants above a certainlevel can be toxic for the plant and may limit plant growthin polluted sites (van Dillewijn, 2008). Another limitationcomes from the fact that the plant can take up somecontaminants and transform them into other chemicalswhose toxicity would need to be tested. We predict reme-diation technologies involving plant/microbes risk assess-ment assays to come into consideration.

Acknowledgements

Research activities of the authors have been supported bytwo projects from Junta de Andalucía (CV11767 and CVI-344) and by EC Project SYSMO (GEN2006-27750-C5-5-E/SYS).

References

Attila, C., Ueda, A., Cirillo, S.L.G., Cirillo, J.D., Chen, W., andWood, T.K. (2008) Pseudomonas aeruginosa PAO1 viru-lence factors and poplar tree response in the rhizosphere.Microb Biotechnol 1: 17–29.

Barac, T., Taghavi, S., Borremans, B., Provoost, A., Oeyen,L., Colpaert, J.V., et al. (2004) Engineered endophytic bac-teria improve phytoremediation of water-soluble, volatile,organic pollutants. Nat Biotechnol 22: 583–588.

van Dillewijn, P., Couselo, J.L., Delgado, E., Corredoira, A.,Wittich, R.M., Ballester, A., and Ramos, J.L. (2008) Bio-remediation of 2,4,6-trinitrotoluene by bacterial nitroreduc-tase expressing transgenic aspen. Environ Sci Technol 42:7405–7410.

Matilla, M.A., Rodríguez-Herva, J.J., Ramos, J.L., andRamos-González, M.I. (2007) Genomic analysis revealsthe major driving forces of bacterial life in the rhizosphere.Genome Biol 8: R79.

Molloy, S. (2006) Environmental microbiology: phenol andphyllosphere. Nat Rev Microbiol 4: 880–881.

Sandhu, A., Halverson, L.J., and Beattie, G.A. (2007) Bacte-rial degradation of airborne phenol in the phyllosphere.Environ Microbiol 9: 383–392.

Waight, K., Pinyakong, O., and Luepromchai, E. (2007) Deg-radation of phenathrene on plant leaves by phyllospherebacteria. J Gen Appl Microbiol 53: 265–272.

Visualizing bacterial surfaces in real time

Antonello Covacci and Rino Rappuoli, Novartis Vaccinesand Diagnostics, Via Fiorentina 1, 53100 Siena, Italy.The study of the network of proteins, protein complexes,sugars and surfaces organelles has been frustrated bythe complexity of bacterial surfaces and the necessity torely on molecular coordinates. The logical consequenceto develop crystallographic methods for large cellularcomponents and the alternative use of cryo-electronmicroscopy to solve organelles structure (flagellum, TypeIV pili) has been fundamental but inherently slow. Devel-opment of new vaccines may also depend on the identi-fication of complexes located at the surface and exposedprotective molecules. Prediction of both is part of routinegenome analysis using dedicated algorithms. Antibodystaining and FACS analysis is a potent tool to visualizeexposed molecules while it is heavily dependent fromantibody specificity, affinity and outer layer penetrationeffect (detection of hidden structures by antibody penetra-tion during preparation of the sample).

We suggest it is now possible to merge high-resolutionfluorescence and scanning confocal microscopy withsortase tagging (Popp et al., 2007) to generate nativemolecules bearing a chromophore. This will potentiallyallow image rendering of the bacterial surface to visualizethe dynamics of protein topology during growth and infec-tion in real time. Science and Nature have both expressedtheir wonder in 2008 about recent advances in lightmicroscopy (Chi, 2008). In addition, white light sourcesare moving from lasers to inexpensive mass production ofLED (http://www.lens.unifi.it). White light is a necessarystep to pulse a sample with a wide range of light wave-lengths to simultaneously collect signals in the visible

146 Crystal ball


http://www.lens.unifi.it

spectrum from excited dyes. The resulting live image canintegrate all labelled proteins in a single picture and with adifferent colour tag providing realistic 3D coordinates. Thecrucial step in closing the loop is in vivo tagging of a targetprotein. This in theory can provide a repertoire of onestrain-one tagged protein with the final goal to colour-codeall the proteins of a bacterial species.

Sortases are bacterial enzymes that predominantlycatalyse the attachment of surface proteins to the bacte-rial cell wall (Telford et al., 2006; Popp et al., 2007). Othersortases polymerize pilin subunits for the construction ofthe covalently attached pili of the Gram-positive bacteria(Telford et al., 2006). The sortase recognition sequenceof Staphylococcus aureus sortase A, LPXTG, whenengrafted near the C-terminus of proteins without naturalsortase specificity, should be part of a sortase-catalysedtranspeptidation reaction using artificial glycine-basednucleophiles. The chemical modification of such sub-strates with fluorophores allows modifications of proteinsin in vitro and in vivo conditions. This method can beefficiently scaled-up for high-throughput data capture.Once the expected wave of new microscopes will beavailable and tagging with fluorophores will be pervasive,this technology should be ready also for fast and unex-pensive real-time expression analyses.

References

Chi, K.R. (2008) The year of sequencing. Nat Methods 5:11–14.

Popp, M.W., Antos, J.M., Grotenbreg, G.M., Spooner, E., andPloegh, H.L. (2007) Sortagging: a versatile method forprotein labeling. Nat Chem Biol 3: 707–708. Epub 2007Sep 23.

Telford, J.L., Barocchi, M.A., Margarit, I., Rappuoli, R., andGrandi, G. (2006) Pili in gram-positive pathogens [Review].Nat Rev Microbiol 4: 509–519.

Coral microbiology

Eugene Rosenberg, Department of Molecular Microbiol-ogy and Biotechnology, Tel Aviv University, Tel Aviv69978, Israel.The young field of coral microbiology is driven primarily bya desire to understand coral diseases, which are causingworldwide damage to coral reefs. The field is particularlyattractive to microbiologists who enjoy combining labora-tory research with field work. In this case, the field worktakes place in the most beautiful surroundings.

Up to now, most of the research in coral microbiologyhas been concerned with isolating potential pathogens(reviewed by Rosenberg, et al., 2007) and comparing themicrobial communities of healthy and diseased corals(e.g. Bourne et al., 2008). These studies have utilized

both classical culture techniques and modern molecularmethods and provide the necessary background for futurecoral microbiology.

What lies ahead? Taking inspiration from medical andenvironmental microbiology, coral microbiologists will inthe near future (i) begin to understand the mechanisms ofcoral disease, (ii) discover the environmental factorswhich contribute to diseases and their spread (reservoirsand vectors), and (iii) define the positive role of microbesin the health of the coral holobiont. One area of researchthat has been largely ignored, but may be very important,is coral virology, including bacterial, algal and coralviruses.

In my opinion the most important and achievable goal ofcoral microbiology is to develop practical technologies forcontrolling the spread of coral diseases. Application of thetechniques that have been developed for land animalsand plants will not be sufficient for the treatment of coraldiseases in coral reefs. It is going to take a creativebreakthrough- and it is difficult to predict when that willoccur.

References

Bourne, D., Iida, Y., Uthicke, S., and Smith-Kuene, C. (2008)Changes in coral-associated microbial communities duringa bleaching event. ISME J 2: 350–363.

Rosenberg, E., Koren, O., Reshef, L., Efrony, R., and Zilber-Rosenberg, I. (2007) The role of microorganisms in coralhealth, disease and evolution. Nat Rev Microbiol 5: 355–362.

Measurements versus understanding:the (metabol)omics dilemma

Uwe Sauer, Institute of Molecular Systems Biology, ETHZurich, Switzerland.The mantra of biology is more data, if possible mea-suring everything, at high resolution and throughput.Everyone who reviews research or PhD proposals is bom-barded with statements on the use of top-notch-omicsmethods. Much rarer are clear visions on how the antici-pated results will aid in finally understanding a particularphenomenon. When reviewing upshots of the proposedresearch (e.g. publications), we often complain aboutdescriptive data gathering. Now, one could argue that thiswill change once the current proposals start to spin offpublications. Still, glancing at old proposals, includingthose of this crystal glazer, we were hopeful at the timethat the anticipated results would indeed provide us withthe missing insights into our various subjects. Do we livein an extremely lucky, and thus unlikely time where isfinally all downhill or is there a conceptual problem?

Of course there is a conceptual problem because thesheer amount of data, let alone their non-linear dynamic

Crystal ball 147


relationships, challenges our intuition and logical reason-ing far beyond their capabilities. It is computational analy-sis, stupid. Historically, the more data mantra is perfectlyunderstandable. In the molecular age, we have only beenable to glimpse at tiny fragments of the whole for decades.With the emergent transcriptomics and proteomicsmethods a dream became true. Blinded by the suddenlyavailable potential, the initial flood of papers was primarilydescriptive. Attracted by this potential, computer scientistslater developed bioinformatics methods that help nowto sift through piles of expression data, identifyingsets of co-regulated genes, regulons and functionalcorrelations.

Much of the bioinformatics success is related to the factthat this research still is, for the most part, in a discoverymode to identify involved molecular components andstructures of genetic networks. Understanding, however,implies the ability to accurately predict the non-lineardynamic behaviour. For this we need computationalmodels that represent relevant biological mechanisms ina quantitative fashion, enabling what-if simulations andpredictions of behaviour in not-yet-studied situations.While computational modelling has not been the pride ofthe biological toolbox so far, the last couple of years havebrought forward a number of promising applications thatmake intelligent use of transcriptomics and proteomicsdata. Obviously further technical developments are still tocome, in particular for proteomics, but even with standardtechnology a single PhD student can generate piles of‘-omics’ data today. The times they are a changin’ forbiology. Development of models and computationalmethods to analyse and integrate such ‘-omics’ dataand to design key follow-up experiments for unravellingcomplex mechanisms becomes the key challenge. Oneindication that resistance to the change is dwindling is thelaunch of a new section on computational biology in thetraditional ASM flagship Journal of Bacteriology (Zhulin,2009).

For the more recent addition to the ‘-omics’ arsenal –metabolomics – the technical challenges are perhapseven greater because of the chemical heterogeneity (andsometimes extreme similarity) of metabolites, their rapidturnover, chemical instability, dynamic range and oftenunknown structure. Nevertheless, I have no doubt thatthese problems will be solved eventually. The question is:will the experience from the above ‘-omics’ historypromote faster intelligent use of metabolomics data? Inone incarnation, metabolomics focuses on profiling of asmany as possible metabolites to identify functional biom-arkers for biological traits, with medical and plant metabo-lomics at the forefront. This line of research is primarily inthe discovery mode and has clearly learned its lesson –suitable bioinformatics methods are available and are rou-tinely used.

An entirely different matter is the nascent field ofquantitative metabolomics. As the catalytic interactionsbetween metabolites and enzymes are known for majorparts of metabolic networks, the focus is not discoverybut monitoring sometimes only subtle response ofknown system components to perturbations. Conse-quently, the data contain important functional andmechanistic information, but these are not immediatelyobvious. What does an increase in one metabolitesignify, and what does it mean when changes occur indistant parts of the network? In sharp contrast to tran-scriptomics and proteomics, metabolite concentrationsare not directly linked to genes. Instead, the concentra-tion of a given metabolite is determined by the presenceand in vivo activity of its cognate enzymes, their kineticand regulatory parameters, the pathway flux and otherfactors. As the most informative metabolites are typicallyconnected to many different enzymes, changes in theirconcentrations are extremely difficult to trace back toparticular events.

An obvious molecular interpretation approach is kineticmodels of metabolism, and the lack of such data formodelling was a key motivation for metabolomics methoddevelopment in the first place. Here is the prediction:although there are still significant analytical and work-flowproblems to be solved for quantitative microbial metabo-lomics, I predict a dramatic increase in the availability ofsuch data in the near future. Profiling metabolite data arealready generated in vast amounts and there are no con-ceptual problems for high-throughput (semi)quantitativemetabolomics. As cost, effort and time per single analysisare only a fraction of those for other ‘-omics’ techniques,metabolomics time-course and large-scale screeningdata sets will soon outnumber those from gene-based‘-omics’ by far.

This development will create a dilemma because welack currently appropriate concepts, beyond simple corre-lation analyses, to obtain mechanistic insights from theexpected metabolomics data. Unless the experiments arespecifically designed for this purpose, kinetic modellingwill not be able to exploit large-scale metabolomics datato a significant extent, simply because metabolite dataalone are insufficient. Different from gene-based ‘-omics’,logical reasoning and the current bioinformatics/statisticsmethods will also not be overly useful. Unless my predic-tion is far off, the gap between our technical capacity formetabolomics data generation and our ability for digestingthem will soon become huge. Thus, the call is open forintelligent computational methods. My guess (and thatis all it is) is that methods enabling identification of themost probable conditions or mutants from metabolomicsscreens for specific follow-up analyses, as well as thedesign of such further experiments, have initially thegreatest potential for obtaining mechanistic insights.

148 Crystal ball


References

Zhulin, I.B. (2009) It is computation time for bacteriology.J Bacteriol 191: 20–22.

Engineered exploitation of microbialpotential

Ian Thomspon, Department of Engineering Science, Uni-versity of Oxford, Begbroke Science Park, Sandy Lane,Yarnton OX5 1PF, UK.We are at a critical point in our history, faced with thechallenges induced by our own large-scale activities,over many years, which are leading to dramatic changesin our climate and an urgent need for remedial mea-sures. Added to these concerns is the continual growthin populations and the pressure this puts on resourcesand environmental quality. These multiple stresses havestimulated the need to improve efficiencies in currenttechnologies and the search for alternatives for sustain-able energy, securing reliable water supplies, treatingwaste and the generation of sustainable products. Oneof the positive features of this alarming situation is theincreasing awareness among microbiologist and non-experts alike, of the potential of microorganisms as pro-viders of some of the remedies. For instance, issues ofenvironmental quality and energy have impacted on thewaste industry in such a way it is now seen to be aresource opportunity, and anaerobic digestion is consid-ered to be the way forward for treatment and sustainableenergy generation.

With such high hopes riding on microbial potential, it isreassuring to know that in recent decades we haveinvested significant funding in techniques for improvingour knowledge of the microbial world. This includesincreasingly elaborate and sophisticated molecularmethods for detecting unculturable populations andrapid sequencing for improving genetic understanding.However, we are now at a critical point whereby weurgently need to translate this vast mountain of knowl-edge and microbial system insights into solutions to thedemanding global challenges. Furthermore, if new oreven established microbial technologies are going to haveany significant impact on climate change or any other ofour global problems, they must be effective at very largescales and importantly, be controllable. Control ofmicrobial potential en masse is in fact an engineeringchallenge.

Manipulation of microbial biomass on a large scaleand in a controlled manner is by no means an easy task.However, mankind has had some notable successes inharnessing the potential of microbial processes, mostnotably the effective exploitation of microbial communi-ties in municipal sewage systems and agriculture.

Although such systems are very effective and haveserved us well for generations, they do not representexamples of where key insights from new genetic infor-mation have been exploited to develop novel environ-mental technologies for solving mankinds’ problems.This has yet to be achieved. Furthermore, the suc-cesses in terms of harnessing of microbial potentialwere achieved by Victorian engineers and agricultural-ists, not microbiologists. The kind of opportunity todayfor such life-changing exploitation would be the identifi-cation of microbial gene sequences/strains that enableproblematic CO2 to be converted directly by algae intomethane or even clean fuels. However, even if we couldisolate or generate such strains the challenge of effec-tive exploitation would have to be resolved, most prob-ably by engineers. This is because in order to harnesslight, algae have to grow on the surface of ponds andsuch two-dimensional growth leads to self-shading. Thesolutions to such issues require effective collaborationwith good engineers to develop algae holding systemsor bioreactors, which enable maximal light exposure inthree dimensions, reducing the foot-print and improvingyield. Other scenarios whereby cross-disciplinary col-laboration will eventually lead to more effective microbialexploitation include hybrid approaches for treating tracelevels of contaminants (such as hormones) in drinkingwater, by employing a combination of high affinity nano-filters, which concentrate the contaminant, making it bio-available to catabolic strains.

Encouragingly, there are signs that microbiologists arebeginning to open their minds in terms of developing newphysical technologies for exploiting cells en masse, in amore controlled manner. These approaches include novelapproaches for moving bacteria through soil and manipu-lating biofilm formation by electrokinetics (Andrews et al.,2006), stimulating biodegradation by manipulating bacte-rial genomes in situ employing ultrasound (Song et al.,2007), and the application of nanomaterials for in situdetection and stimulating cell activity (Chien et al., 2008).Although early days, such novel approaches provide hopethat the microbial potential can be engineered and morereliably harnessed on a large scale. This will require a newgeneration of microbiologists who have more cross-disciplinary training, who embrace the opportunities thatphysical and engineered techniques offer, and who havethe imagination to consider complementary approachesfor the limited array of microbial cell manipulation methodswe traditionally employ (e.g. pH, temperature, concentra-tion). This is very good news as the quicker we realize thatsequencing more genomes is not the only option forresolving our problems, the quicker we can generatesome effective solutions. Such critical advances willbe accelerated by employing more systems biologyapproaches, linking information from the cell to the whole

Crystal ball 149


community, an approach which again will require multidis-ciplinary training, in this instance computing science andmathematics.

The Victorians may have not realized when they devel-oped sewage and clean water systems the extent theyhad harnessed the potential of microbial communities tosolve their problems. However, what they achieved andwhat we need to learn again is that the solution to many ofour current problems is going to come from effective engi-neering of microbial systems, as this is the only way tocontrol more effectively and provide the scale-up requiredto have significant global impacts.

References

Andrews, J.S., Mason, V.P., Thompson, I.P., Stephens, G.M.,and Markx, G.H. (2006) Construction of artificial structuredmicrobial consortia (ASMC) using dielectrophoresis: exam-ining bacterial interactions via metabolic intermediateswithin environmental biofilms. J Microbiol Methods 64:96–106.

Chien, S.-F., Chen, S.-H., and Lin, L.-C. (2008) Nanomateri-als enhanced gene expression in yeast cells. J Nanoma-terials 2008: 391497.

Song, Y.Z., Hahn, T., Thompson, I.P., Mason, T.J., Preston,G.M., Li G., et al. (2007) Ultrasound–mediated DNA trans-fer. Nucleic Acids Res 35: e129.

Human biome biotechnology and thepersonalization of odour profiles*

Kenneth Timmis, Environmental Microbiology Laboratory,Helmholtz Centre for Infection Research, and Institute ofMicrobiology, Technical University of Braunschweig,Braunschweig, Germany.Personalized medicine, the monitoring, prevention andtreatment of disease in an individual, specifically tailoredto his or her specific genomic make-up, is rapidly gaininginterest in the medical, scientific and general population,as our understanding of genetic susceptibilities to diseaseand health-relevant causal relationships between our indi-vidual genetically determined physiologies and environ-mental factors advances.

The current basis of personalized medicine, the geneticdiversity of humans and the resulting diversity of suscep-tibilities of individuals to disease, is of course only part ofthe equation, because it is well appreciated that we, likeall animals and plants, are covered by second outer‘skins’ of populations of phylogenetically and physiologi-cally diverse microbes, which add and integrate metabolic

functions that profoundly influence our physiology andhealth. We are organized ‘biomes’ of interacting commu-nities of human and microbial cells and tissues. The nextlevel of personalized medicine will therefore integrate thegenetic and physiological diversity of our microbial biomepartners.

Disease is a negative aspect of health and living, andcurrent personalized medicine is principally targeted atdisease prevention and treatment, particularly in suscep-tible individuals. However, personalized medicine willalso be applied to the positive side of health and living –lifestyle medicine for healthy people – and will thus beapplied to anyone at any time, in most instances over alarge part of the lifespan. Just as the treatment ofchronic disease is often of more commercial importancethan that of acute disease, so chronic health is particu-larly attractive commercially for personalized lifestylemedicine.

Human biome biotechnology will contribute to person-alized medicine, including lifestyle medicine, in a varietyof spheres; the one I elect to deal with here is our desireto smell nice for personal satisfaction and to please/attract others. The commercial importance of this isreflected in the high value of the odour ingredients (inperfumes, colognes, body lotions, deodorants, scentedsoaps and the like) that form the core of the body carebusiness.

Our smell – the olfactory perception of volatile com-pounds emitted from our skin, body hair and bodily ori-fices – is determined by a number of rather variablefactors (see Anesti et al., 2004; Eggert et al., 1998–1999;Jacob et al., 2002; Roberts et al., 2008; Wood and Kelly,2009; Yamazaki et al., 1998–1999), and citations therein),such as (i) the ‘personal’ composition of volatiles andnon-volatiles that we secrete onto our skin surface, whichdepends on our particular physiology (our individualgenomic programme), (ii) the age-related changes incomposition of these secretions, (iii) the temporary/periodical secretion of additional chemicals resulting fromchanges in health/hormonal balance/recent food intake/etc., (iv) the washing of skin with detergents, many ofwhich are perfumed, (v) the application of body care prod-ucts, many of which are also perfumed (see Wood, 2009),and (vi) the composition and activity of our microbial ‘skin’that interacts with our natural secretions and applied prod-ucts, metabolizing and thereby changing their composi-tion, and also creating further volatiles, either metabolitesfrom the secretions, or that are purely microbially derived(see e.g. Anesti et al., 2004). It is this mix of volatiles ofendogenous human secretions, endogenous microbialskin secretions, metabolites resulting from the microbialtransformation of human secretions, and perfume supple-ments, which structures our individual odour profile at anymoment in time.

*Inspired by Ann Wood, Don Kelly and Joan Timmis, anddedicated to the memory of Rose Timmis, who activated andcultivated my olfactory appreciation of the enormous diversityof smells wafting on the air of the English countryside, flowergarden, vegetable garden and dining table.

150 Crystal ball


It is also this individual diversity of odour-structuringparameters that leads to the same perfume/cologne ondifferent women/men smelling subtly (or non-subtly) dif-ferent (‘Gosh: what is that awful perfume you’re wearingtoday?’ ‘What do you mean: you said how nice it was onFelicity!’). This has led to the current practice of theempirical testing of the whole range of options in theperfumery in order to find the product that most pleasesand matches an individual’s preferences and skin odourcharacteristics. Products that modify our own odour aregenerally applied daily, usually after showering, and aresubject to change – physical, chemical and biological –on the skin surface – over the course of the day,changes that are both generic and individually deter-mined by our particular genome-determined physiology,skin flora and daily activities, such that the smell createdat the time of application changes in time, and differentlyon different people (Ann Wood: ‘and some disappearmuch faster than others, making them very expensiveodour-rentals!’).

Elucidation of the key interactions and causal relation-ships that determine personal odour, and how it can bemodified to achieve a desired quality, will ultimately allowthe development of rational design of enhanced odourprofiles. Replacement of the current hit-and-miss empiri-cism of odour selection by human chemistry:microbialecology-based procedures to create customized odourprofiles, and to optimize their temporal developmentover the course of the day, may become a significantactivity of the lifestyle branch of the personalized medi-cine industry.

The scientific progress necessary to accomplish thiswill include elucidation of the key human genomic andphysiological determinants, the ecological interactions ofthe skin microbial flora with the epidermal surface andits secretions, the individual variations in these, whichunderlie personal odour specificity, and the manner inwhich these interact with and modify relevant personalcare products. This will involve not only analysis of thespecifics of different anatomical areas of the bodysurface (axillae, feet, ears, neck, etc.) exhibiting differentsecretion characteristics, but also different microscopicniches within such areas that characterize the spatialdifferentiation of the epidermal tissue structure and theirdiverse ecological chemistries. Functional metagenom-ics of such micro-niches will be a key componentof this research. At the core of the functional genomicsof the skin biome will be the development of ultra-sensitive analytical procedures to identify and quantifyvolatiles at odour-relevant concentrations, namely‘odouromics’.

Such studies will result in major advances in ourunderstanding of physiological and ecological determi-nants and triggers of production of specific volatiles, and

identification of the principal players mediating changesin such volatiles and the agents of regulation of change.This will lead to identification of compounds, microbesand procedures that enable modification of volatile com-position. This in turn will lead to the formulation ofpersonalized products (e.g. creams containing volatiles,volatile precursors/metabolites, inducers of volatile for-mation, inhibitors, prebiotic and/or probiotic microbes,etc.) that modulate skin chemistry and microbial florasuch that production/maintenance of production of spe-cific volatiles are favoured and formation of undesirablevolatiles by the skin biome are disfavoured, and that,in combination with specific perfumed products, createcustom odour profiles in individuals that match theiraspirations.

Underpinning all of this will be newly developed power-ful, but simple to operate and interpret odouromics instru-mentation to analyse odour profiles, and skin microbefunctionalities, that, in combination with personal genomicprofiles, will produce individual assessments for the cus-tomization of commercial products.

The other aspect of personal odour is, of course, itsolfactory perception, which also varies from person toperson. As a consequence, self-perception of one’s ownsmell can be quite different from the perception of thesame smell by someone else. This is not an issue if onebuys perfume for self-satisfaction, but may be a major oneif the goal is to please or attract others. Thus, a furtherlevel in personalizing odour profiles may well be thetuning to a partner’s preferences.

The aspect of attracting others inevitably leads toanother issue, namely that some skin volatiles may act aspheromones: the composition of our volatiles also contrib-utes to our non-visual sexual attraction to others. Thisaspect will also undoubtedly be of some interest to thepersonal care branch of the personalized medicineindustry.

It is perhaps worth noting that this research may notonly result in the ability to custom design personal olfac-tory images but also lead to a clearer understanding ofthe fundamental relationships between our volatile emis-sions and physiological and mental states. This in turnwill lead to improved diagnostic procedures in clinicalmedicine for certain disease states and syndromes(the chemical analysis of volatiles in breath is alreadyused for such purposes: e.g. see Salerno-Kennedy andCashman (2005), as well as advances in dermatologyitself, but also perhaps for other purposes, such as liedetection (in judicial investigations, job interviews, rela-tionship conflicts, etc.), aptitude assessments; partner-ship compatibility assessments. And all of this will drivethe development of new technology and instrumentation.Only the imagination limits the range of applications. Goto it, biotech!

Crystal ball 151


References

Anesti, V., Vohra, J., Goonetilleka, S., McDonald, I.R.,Sträubler, B., Stackebrandt, E., et al. (2004) Moleculardetection and isolation of facultatively methylotrophic bac-teria, including Methylobacterium podarium sp. nov., fromthe human foot microflora. Environm Microbiol 6: 820–830.

Eggert, F., Müller-Ruchholtz, W., and Ferstl, R. (1998–1999).Olfactory cues associated with the major histocompatibilitycomplex. Genetica 104: 191–197.

Jacob, S., McClintock, M.K., Zelano, B., and Ober, C. (2002).Paternally inherited HLA alleles are associated withwomen’s choice of male odor. Nat Genet 30: 175–179.

Roberts, S.C., Gosling, L.M., Carter, V., and Petrie, M. (2008)MHC-correlated odour preferences in humans and the useof oral contraceptives. Proc Biol Sci 275: 2715–2722.

Salerno-Kennedy, R., and Cashman, K.D. (2005) Potentialapplications of breath isoprene as a biomarker in modernmedicine: a concise overview. Wien Klin Wochenschr 117:180–186.

Wood, A.P. (2009) Personal care products. In Handbook ofHydrocarbon and Lipid Microbiology, Volume 3. Conse-quences of Microbial Interactions with Hydrocarbons, Oilsand Lipids. Timmis, K.N. (ed.). In press.

Wood, A.P., and Kelly, D.P. (2009) Skin microbiology, bodyodour and methylotrophic bacteria. In: Handbook of Hydro-carbon and Lipid Microbiology, Volume 3. Consequencesof Microbial Interactions with Hydrocarbons, Oils andLipids. Timmis, K.N. (ed.). In press.

Yamazaki, K., Singer, A., and Beauchamp, G.K. (1998–1999)Origin, functions and chemistry of H-2 regulated odorants.Genetica 104: 235–240.

The future is artificial

Svein Valla, Department of Biotechnology, NorwegianUniversity of Science and Technology, Trondheim,Norway.The obvious danger of trying to predict the future inscience and also allow the prophecy to be put on theInternet is that a few years later one would most likelywish it was never written. Alternatively, the predictionsmay be stated in such vague ways that they can beinterpreted to always fit what actually happens in thefuture, a technique heavily used by politicians. Manywould also say that this is the method used by the famousFrench doctor and mathematician Nostradamus (1503–1566), enabling him to keep his many believers for cen-turies. I have decided not to use such tricks and insteadface the judge when the time comes. One reason I am notso worried is that I think it is hard to overestimate the paceof progress in the times we are living now. The develop-ments in molecular biology are currently very much drivenby new methodologies, and anyone scanning the mostprestigious journals will have to admit that even within asingle calendar year lots of methods are published that anindividual would not even think of. What would be theresponse in 1975 to someone suggesting to sequence

the entire human genome, to produce human growthhormone in E. coli, or to scan the scientific literature for aparticular topic in one second? Therefore, if we think asufficient number of years back (not very long) in time wewill see that things we could not even imagine has actuallyhappened. As I see it we now have or can already foreseeso many tools available to dissect and manipulate livingorganisms that I think we can safely say: ‘anything’ willbecome possible in the future. The problem is thereforemore, when exactly will each predictable developmentbecome a reality?

At the moment one of my favourite and perhaps tooobvious predictions is that we soon will see a boom in thenew field of Synthetic Biology. A few years ago it was arather costly project to synthesize genes. Now it is socheap that an average scientist can buy his own favouritegene by paying from his or her own pocket. As this is nowtrivial, and we also know how to link DNA fragmentstogether it will soon also be cheap to make very longstretches of DNA, as a custom-made service. This devel-opment will certainly come, because the synthesis of anentire plasmid was published as early as in 1990 (Man-decki et al., 1990), a synthetic polio-virus DNA in 2002(Cello et al., 2002), and synthesis of the entire Myco-plasma genitalium chromosome was recently reported(Gibson et al., 2008).

Does this mean that we can soon make synthetic catsand dogs from scratch? Certainly not soon, but possiblysomewhat later! Infectious viruses can already be made,and this alone can lead to enormous impacts in scienceand on the society. The new viruses may be used benefi-cially to study viral pathogenesis in model systems or toanalyse their surviving capacity in natural environments.Potentially they may also be used by terrorists to createdevastating epidemic diseases, and are we really pre-pared for this very frightening possibility in the nearfuture? Some believe that this will never happen, but I feelthat this potential problem should be analysed and dis-cussed much more.

If one moves one level up synthetic bacteria is the nextobvious step. This has already been achieved to someextent (see above and Lartigue et al., 2007), but right nowit is far from trivial. I cannot see any overwhelming tech-nical obstacles for further progress in this field, so manysynthetic bacteria will be soon be created. Potentially theycan be used to make super-bugs for all kinds of specificpurposes, ranging from ‘traditional’ cell factories to savingthe worlds climate (http://blog.litfuse.com.au/2008/03/15/synthesising-new-organisms-to-save-the-world/), andalso perhaps purposely to cause damage to the society,as discussed above for viruses. In the first years to comesuch synthetic bugs will have to rely heavily on thecomplex regulatory networks already known to functionin existing living organisms, as it appears unrealistic to

152 Crystal ball


http://blog.litfuse.com.au/2008/03/15/synthesising-new-organisms-to-save-the-world

http://blog.litfuse.com.au/2008/03/15/synthesising-new-organisms-to-save-the-world

design completely new and well-functioning networksfrom scratch. However, in a longer time-frame (20 years?)I am convinced that this will also become possible.

If I am going to be really brave I might predict that thedevelopments in Systems and Synthetic Biology willbecome a major turning point in the history of man. It may(together wit other developments) lead to an understand-ing and control of fundamental processes such as agingand consciousness. Such breakthroughs would haveenormous and obvious impacts on society, and it is hardto see why it should never become possible. Again, it ismostly a matter of when. However, in a five two ten-yearsperspective synthetic bugs will dominate and as usualthese technologies can be applied both to the benefit andharm of mankind. What generally worries me is that weare getting in control of stronger and stronger forces, suchthat our traditional ways of thinking about our roles asscientists may soon become outdated. Al Gore has alsonoticed this by stating that man has become a force ofnature. Another prediction is therefore that scientists inthe future must even more than in the past think about theconsequences of what they are doing. I believe that thiswill become a major issue in the next 10 years!

References

Cello, J., Paul, A.V., and Wimmer, E.T. (2002) Chemicalsynthesis of poliovirus cDNA: generation of infectious virusin the absence of natural template. Science 297: 1016–1018.

Gibson, D.G., Benders, G.A., Andrews-Pfannkoch, C.,Denisova, E.A., Baden-Tillson, H., Zaveri, J., et al. (2008)Complete chemical synthesis, assembly, and cloning of aMycoplasma genitalium genome. Science 319: 1215–1220.

Lartigue, C., Glass, J.I., Alperovich, N., Pieper, R., Parmar,P.P., Hutchison, C.A., et al. (2007) Genome transplantationin bacteria: changing one species to another. Science 317:632–638.

Mandecki, W., Hayden, M.A., Shallcross, M.A., and Stotland,E. (1990) A totally synthetic plasmid for general cloning,gene-expression and mutagenesis in Escherichia coli.Gene 94: 103–107.

Mining the microbes – the humanmicrobiome as model

Willem M. de Vos, Laboratory of Microbiology at Wagenin-gen University, Wageningen, The Netherlands, andDepartment of Basic Veterinary Sciences at Helsinki Uni-versity, Helsinki, Finland.While the number of completed microbial genomes issteadily increasing, a further explosion of genomicsequencing has been developing with the deployment ofpost-Sanger sequencing instruments that couple amazing

technology and extreme throughput with fancy and imagi-native names (Mardis, 2008). Assuming that ABC issueson the accuracy, bioinformatics and costs are solved, theimpact of this next-generation sequencing technology onthe future developments in microbial biotechnology maywell be beyond our present imagination. What can beenvisaged now is already quite wild and includes manytechnological applications for (meta)genome discovery,SNP analysis and transcriptome profiling. These can beimplemented in all phases of discovery, optimization andquality control, while being applicable to all colours of thebiotechnology rainbow, ranging from red to green andwhite to blue. However, the most exciting applications thatare specific for the microbial world still have to come andrelate to the mining of the global microbial diversity andfunction. This requires more than only metagenomesequencing as can be illustrated for the largest microbialecosystem that is closest to our heart: our microbesinside.

Intestinal microbes dominate our body since birth andoutnumber our own cells by one or more orders of mag-nitude (Zoetendal et al., 2008). There are a variety ofwell-funded metagenome sequencing projects in allcorners of the world that aim to characterize the role ofthese intestinal microbes in health and disease (seeMullard, 2008). These projects on the microbiome of thehuman intestine and other body parts have recently beenunited in a large global initiative, the International HumanMicrobiome Consortium that is providing a platformfor data sharing between the US National Institutes ofHealth and the European Commission (see URL: http://cordis.europa.eu/search/index.cfm?fuseaction=news.document&N_RCN=30004). This platform may alsoprovide a means to generate the needed coordinationon data quality and protocols that are highly relevantwith the implementation of next-generation sequencingtechnologies.

What will be the outcome of this massive sequencingoperation? Exposing the function of specific microbes andtheir genes will be obvious targets, in the line with recentexamples of intestinal microbes (Konstantinov et al.,2008; Sokol et al., 2008). However, this will require a longprocess of functional discovery that can be initiated by themicrobiome sequencing effort but also requires growthand isolation of the intestinal microbes that are fastidious,anaerobic or require cell–cell contact for growth. Signifi-cant progress has been made here with the developmentof throughput culturing devices (Ingham et al., 2007).However, there is ample room for improvement, notablyat the level of microbial interactions and varying growthconditions. A shorter time horizon can be foreseen formicrobiome-based microbial diagnostics that will allowrapid insight in intestinal microbial diversity. Specific high-throughput systems are already in place (Zoetendal et al.,

Crystal ball 153


http://cordis.europa.eu/search/index.cfm?fuseaction=news

http://cordis.europa.eu/search/index.cfm?fuseaction=news

2008) and can be expanded to be of use for nutritionaland pharmaceutical interventions as well as to discrimi-nate between health and disease state of the intestinaltract.

Given the abundance and vast coding capacity of theintestinal microbiome, it will be of great interest to followthe developments in the International Human MicrobiomeConsortium and see whether it can serve as a model forother ecosystems where next-generation sequence tech-nology is applied to expose the coding capacity and func-tion of the microbial world on our planet.

References

Ingham, C.J., Sprenkels, A., Bomer, J., Molenaar, D., vanden Berg, A., van Hylckama Vlieg, J.E., and de Vos, W.M.(2007) The micro-Petri dish, a million-well growth chip forthe culture and high-throughput screening of microorgan-isms. Proc Natl Acad Sci USA 104: 18217–18222.

Konstantinov, S.R., Smidt, H., de Vos, W.M., Bruijns, S.C.,Singh, S.K., Valence, F., et al. (2008) S layer protein A ofLactobacillus acidophilus NCFM regulates immature den-dritic cell and T cell functions. Proc Natl Acad Sci USA 105:19474–19479.

Mardis, E.R. (2008) Next-generation DNA sequencingmethods. Annu Rev Genomics Hum Genet 9: 387–402.

Mullard, A. (2008) The inside story. Nature 453: 578–580.Sokol, H., Pigneur, B., Watterlot, L., Lakhdari, O., Bermúdez-

Humarán, L.G., and Gratadoux, J.J. (2008) Faecali-bacterium prausnitzii is an anti-inflammatory commensalbacterium identified by gut microbiota analysis of Crohndisease patients. Proc Natl Acad Sci USA 105: 16731–16736.

Zoetendal, E.G., Rajilic-Stojanovic, M., and de Vos, W.M.(2008) High throughput diversity and functionality analysisof the gastrointestinal tract microbiota. Gut 57: 1605–1615.

Predictive microbial ecology

Jizhong Zhou, Institute for Environmental Genomics andDepartment of Botany and Microbiology, University ofOklahoma, Norman, OK, USA.Microbial ecology is a science that studies the relation-ships between microorganisms and their biotic and abioticenvironments. The ultimate goal of microbial ecology is topredict who is where with whom doing what, why andwhen. To achieve this predictive goal, computational mod-elling and simulation of microbial community dynamicsand behaviour at both the structural and functional levelsis essential. However, in contrast to the intensive model-ling studies in plant and animal community ecology,the modelling of microbial community behaviour withindifferent environmental contexts has not been initiated.The vast majority of the community studies in microbialecology are still at a descriptive level rather than ata quantitative and predictive level. One of the major

problems is the lack of reliable experimental informationon community-wide spatial and temporal dynamics ofmicrobial community structure, activities and functions.Because microorganisms are difficult to directly count,obtaining enough data required for modelling is extremelydifficult and even impossible with conventional moleculartechniques such as PCR-based cloning, in situ hybridiza-tion and quantitative PCR. Although applications of suchconventional molecular techniques over the last twodecades have provided new insights into microbial diver-sity and structure, they have failed to provide community-wide quantitative information in a rapid fashion formodelling and predicting microbial community dynamics.Lack of sufficient reliable data prevents microbial ecolo-gists from addressing quantitatively important ecologicaland environmental questions such as competition, stabil-ity, succession, and adaptation in microbial communities.Such questions are extremely important for the appliedmanipulation of microbial communities for desired func-tions related to human health, food, energy production,environmental cleanup, and industrial and agriculturalpractices.

With the recent development and application of large-scale high-throughput sequencing (see Sogin et al., 2006;Huber et al., 2007; Hamady et al. 2008) and associatedmetagenomics technologies such as GeoChip relatedfunctional gene arrays (He et al., 2007; Zhou et al., 2008;Handelsman et al., 2007), community-wide spatial andtemporal information on microbial community functionalstructure and activities can be rapidly obtained. This willbe the first time that microbial ecologists will be not frus-trated by the paucity of experimental data to addressecological questions. In turn, microbial ecologists mayvery well be overwhelmed by massive amounts ofmetagenomics data and will be perplexed as to how toutilize and interpret the data within ecological and envi-ronmental contexts. Here, I will argue that we are in thenew era of transforming microbial ecology from descrip-tive studies to a quantitative and predictive science.

Using high-throughput metagenomics technologies andcomputational modelling, it is possible to tackle somefundamental ecological questions (Zhou et al., 2004),which have been difficult to address before such as: (i)How do the phylogenetic and functional structure ofmicrobial communities change across various spatial(from micrometers, kilometers to tens of thousands kilo-meters) and temporal scales and what are the forcesshaping their diversity and structure? (ii) How can thecomplex ecological networks in microbial communities beidentified and whether are they important to ecosystemfunctioning? (iii) What is the molecular basis for functionalstability and adaptation of microbial communities? (iv)How is the functional stability of a microbial communityrelated to its genetic and metabolic diversity as well as

154 Crystal ball


environmental disturbance? (v) Can the functional stabil-ity and future status of a microbial community be pre-dicted based on the metabolic functional conservationand differentiation of individual microbial populations? (vi)Can a microbial community be manipulated to achieve adesired stable function by manipulating the metabolictraits of the community? (vii) How can the information bescaled from molecules to populations, to communities,and to ecosystems for understanding ecosystem behav-iours and dynamics? (viii) Can the molecular-level under-standing of microbial community structure improve ourpredictive power of the ecological and evolutionaryresponses of microbial communities to environmentalchanges, especially global climate changes?

Addressing the above questions in a quantitative andpredictive way requires rigorous experimental designingand systematic intensive sampling of the microbialsystems studied. Selection of experimental systems withappropriate complexity and replications could be veryimportant. I believe that two general strategies can beemployed. One is to focus on surveying complex naturalmicrobial systems by using high-throughput metagenom-ics technologies to systematically compare the common-ality and differences of microbial community diversitypatterns, metabolic capacities, and functional activitiesacross various spatial and temporal scales. While suchsurvey-based approach provides rich information onmicrobial community diversity patterns and dynamics, itcould be difficult to establish detailed definitive mechanis-tic linkages between microbial diversity and ecosystemfunctioning because the microbial systems in natural set-tings are generally very complex. Another complementarystrategy is to establish well-controlled laboratory systemssuch as bioreactors with simplified communities to sys-tematically examine the responses of microbial commu-nities to environmental changes and the impacts of theirresponses on ecosystem functioning. Such laboratorysystems are important to establish cause-and-effect rela-tionships, because they have great advantages in termsof system controls, monitoring, data collection, replica-tions and modelling. Determination of cause-and-effectrelationships is much easier with simpler, engineered,laboratory-based bioreactor systems than with complexnatural communities, as input and output parameterscan be controlled, along with environmental conditions.Although the community in a controlled system is not anatural community, such systems would offer the bestopportunity to acquire mechanistic understanding of thefundamental principles of interactions among variousmicroorganisms and the molecular level ecological andevolutionary responses of microbial communities to envi-ronmental changes. Therefore, well-controlled laboratoryengineered systems will be critical to predictive microbialecology studies.

Predictive microbial ecology requires not only high-throughput experimental tools but also high performancecomputational capabilities. System-level understandingof the dynamic behaviour of microbial community struc-ture, functions and their relationships to ecosystem func-tioning faces several grand computational challenges.First, microbial diversity is extremely high. The number ofgenes in a genome or populations in a community farexceeds the number of sample measurements due tohigh cost of measurements. It is difficult to apply classicalmathematical tools such as differential equations to simu-late high-throughput metagenomics data because no solesolution can be obtained for the constructed models. Newmathematical theories and approaches are neededto deal with such dimensionality problems. Second,metagenomic data from analyses of transcriptomes,proteomes and metabolomes, as well as physiologicaland geochemical data, are heterogeneous. Synthesizingvarious types of large-scale data together to make bio-logical sense is also difficult. Rapid high performanceparallel computational tools are needed for data process-ing, computation and visualization. In addition, becausethe dynamic behaviours of biological systems at variouslevels (cells, individuals, populations, communities, andecosystems) are measured on different temporal andspatial scales, linking cellular-level genomic informationto ecosystem-level functional information for predictingecosystem dynamics is even more challenging. Novelmathematical framework and computational tools areneeded for achieving systems-level understanding andprediction of microbial community dynamics, behaviourand functional stability.

With the rapid continuing advances of metagenomics-based high-throughput experimental technologies andassociated high performance computational tools, micro-biologists should be able to perform more quantitativemodelling studies of microbial systems as macroecolo-gists have done since last half century. There is no doubtthat the era of quantitative predictive microbial ecology iscoming.

References

Hamady, M., Walker, J.J., Harris, J.K., Gold, N.J., and KnightR. (2008) Error-correcting barcoded primers for pyrose-quencing hundreds of samples in multiplex. Nat Methods5: 235–237.

Handelsman, J., Tiedje, J.M., Alvarez-Cohen, L., Ashburner,M., Cann, I.K.O., DeLong, E.F., et al. (2007) Committee onMetagenomics: Challenges and Functional Applications.Washington, DC, USA: National Academy of Sciences.

He, Z., Gentry, T.J., Schadt, C.W., Wu, J., Liebich, S.C.,Chong, Z., et al. (2007) GeoChip: a comprehensivemicroarray for investigating biogeochemical, ecologicaland environmental processes. ISME J 1: 67–77.

Crystal ball 155


Huber, J.A., Mark Welch, D., Morrison, H.G., Huse, S.M.,Neal, P.R., Butterfield D.A., and Sogin M.L. (2007) Micro-bial population structures in the deep marine biosphere.Science 318: 97–100.

Sogin, M.L., Morrison, H.G., Huber, J.A., Welch, D.M., Huse,S.M., Neal, P.R., et al. (2006) Microbial diversity in thedeep sea and the underexplored ‘rare biosphere’. Proc NatlAcad Sci USA 103: 12115–12120.

Zhou, J.-Z, Kang, S., Schadt, C.W. Garten, C.T., Jr (2008)Spatial scaling of functional gene diversity acrossvarious microbial taxa. Proc Nat Acad Sci USA 105: 7768–7773.

Zhou, J.-Z., Thompson, D.K., Xu, Y. and Tiedje, J.M. (2004)Microbial Functional Genomics. Hoboken, NJ, USA: JohnWiley & Sons.

156 Crystal ball


Documents

Predictive microbial ecology