University of Groningen
Metabolic heterogeneity in clonal microbial populations
Takhaveev, Vakil; Heinemann, Matthias
Published in:
Current Opinion in Microbiology DOI:
10.1016/j.mib.2018.02.004
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date: 2018
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
Takhaveev, V., & Heinemann, M. (2018). Metabolic heterogeneity in clonal microbial populations. Current Opinion in Microbiology, 45, 30-38. https://doi.org/10.1016/j.mib.2018.02.004
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
Metabolic
heterogeneity
in
clonal
microbial
populations
Vakil
Takhaveev
and
Matthias
Heinemann
Inthepastdecades,numerousinstancesofphenotypic diversitywereobservedinclonalmicrobialpopulations, particularly,onthegeneexpressionlevel.Muchlessis, however,knownaboutphenotypicdifferencesthatoccuron thelevelofmetabolism.Thisislikelyexplainedbythefactthat experimentaltoolsprobingmetabolismofsinglecellsarestillat anearlystageofdevelopment.Here,wereviewrecentexciting discoveriesthatpointoutdifferentcausesformetabolic heterogeneitywithinclonalmicrobialpopulations.These causesrangefromecologicalfactorsandcell-inherent dynamicsinconstantenvironmentstomolecularnoiseingene expressionthatpropagatesintometabolism.Furthermore,we provideanoverviewofcurrentmethodstoquantifythelevelsof metabolitesandbiomasscomponentsinsinglecells.
Address
MolecularSystemsBiology,GroningenBiomolecularSciencesand BiotechnologyInstitute,UniversityofGroningen,Nijenborgh4,9747AG Groningen,TheNetherlands
Correspondingauthor:Heinemann,Matthias(m.heinemann@rug.nl)
CurrentOpinioninMicrobiology2018,45:30–38
ThisreviewcomesfromathemedissueonMicrobialsystems biology
EditedbyTerenceHwaandUweSauer
https://doi.org/10.1016/j.mib.2018.02.004
1369-5274/ã2018TheAuthors.PublishedbyElsevierLtd.Thisisan openaccessarticleundertheCCBY-NC-NDlicense( http://creative-commons.org/licenses/by-nc-nd/4.0/).
Introduction
Inthelasttwodecades,wehaveobtainedampleevidence thatcellularphenotypesaredeterminednotonlybythe genotypeandenvironment,butcanbeinfluencedalsoby stochastic effects. These stochastic effects arise from
Brownian motion of low copy number biomolecules
involvedingeneexpression.Arangeofstudieshasshown howthestochasticityingeneexpressioncanleadtothese differentphenotypes,specifically,in termsof transcript andproteinlevels[1–3].
Bycontrast,likelyduetothelackofexperimentaltoolsto probe metabolism in single cells, still relativelylittle is knownhowthemolecularnoiseonthegeneexpression levelpropagatesintometabolism.Recentevidence, nev-ertheless,suggeststhatclonalmicrobialcellscandisplay
significant diversity in their metabolism, with the
extremecaseof subpopulationshaving distinctly differ-entactivitiesofmetabolic pathways[4–6].Furthermore, recentdiscoveriesshowthatevenunderconstant condi-tionsmetabolism of microbes is notstatic, but changes overtime,forinstance,alongthecelldivisioncycle[7] and with ageing [8,9]. The variability of metabolism withinclonal populations canpose challenges in eradi-catingpathogenicmicrobes,asindividualcellscanexhibit differentdegreesofantibiotictolerance[5].Inthesame way,the presence of metabolically low-performing var-iantsmaycompromisetheefficiencyofbiotechnological processesin industry[10].
Inthisreview,wegatherevidenceformetabolic hetero-geneity observed in clonal microbial populations. We
show recent examples where metabolic heterogeneity
arises from ecological factors, cell-inherent dynamics and stochastic effects (Figure 1). Novel experimental tools probing metabolism in single cells are necessary tofurtheruncovermetabolicdifferenceswithinmicrobial populationsandto ultimatelyunderstandhow such het-erogeneityoccurs.Thus,wealsoreviewthelatest devel-opments of respective tools (Table 1) and highlight insightsonmetabolicdifferencesthathavebeen discov-eredusingthesemethods.
Metabolic
heterogeneity
arising
from
the
influence
of
ecological
factors
Itiswellaccepted thatenvironmental factorscancause metabolic heterogeneity. Still, recent work has added new insights into what such ecological factors can be andhowtheycaninfluencecellularmetabolism. Remark-ably,ithasrecentlybeenfoundthatsurface immobiliza-tionofyeastcellsupregulatesgenesofglycolysisandcell wallbiogenesis,aswellasacceleratesethanolproduction and induces glycogen accumulation. Contrary to yeast
in planktonic state, the immobilized cells rapidly
ceasedividing and preserveviabilityfor more than two weeks[11].
Another common ecological factor that can stimulate
metabolic diversification is provided by intercellular interactionamong thevery cells in thepopulation[12]. Forinstance,withinayeastcolony,cellshavebeenfound to cooperatebyproducing differentanabolicprecursors and eventually exchanging them [13]. Such metabolic specializationofindividualcellshasbeenshownto deter-minetheirabilitytocopewithstress[14].Interestingly,
complex behavior can emerge from such intercellular
interaction. Cells in the periphery of a Bacillus subtilis biofilmare known to starve the interiorcells [15]. The latter,inturn,communicatetheirdemandforfoodusing
specific metabolites [15] and even electrical signaling [16]. These intercellular interactions have been found toresultinperiodiccessationofgrowthoftheperipheral cellsandnutrientflowintotheinterior[15].Toefficiently exploitscarcenutrients,eventwoneighboringbiofilmsof B.subtilishavebeenfoundtocoupletheirmetabolismvia theelectricalcommunicationandto undergoanti-phase growth oscillations[17].Thus,ecologicalfactors, such as surface attachment and intercellular interactions, which arepresentevenunderstrictly controlled experi-mentalconditions,mayleadtometabolicheterogeneity.
Metabolic
heterogeneity
arising
from
cell-inherent
dynamics
Experiencing identical environmental factors, two cells can still differin their metabolism,for instance, dueto cell-inherent dynamics. One type of such dynamics is associated with thecell division cycle. Recently, it has beenshowninsingleyeastcellsthatoscillationsinNAD (P)H and ATP levelsoccur in synchrony with the cell cyclebut areautonomousfromit[7].Transcriptomics
and ribosome profiling experiments on synchronized
populationssuggestthatthecellsperiodicallyfoster bio-massproductionalong thecellcycle. Particularly, mito-chondrialandribosomalbiogenesisgenesarefound upre-gulated in G1, thesame happensfor lipid biosynthesis genes in G2/M [18]. Thus, eukaryotic microbes have
oscillations in thelevelsof metabolites, andpotentially alsoin thebiosynthesisofbiomasscomponents.
Thislikelyextendstoprokaryotesaswell.Forexample, in Caulobacter crescentus the level of cyclic di-GMP has beenfoundtooscillateandtocontrolthecellcycle[19]. Similarly, recent dynamic single-cell measurements of NAD(P)H levelsin Escherichiacolihaveshown thatalso thismetaboliteoscillatesalongthecellcycle[20]. Like-wise, the earlier reported non-Gaussian distribution of ATPlevelsacrosssinglecells[21]couldpossiblyhaveits origin in a cell cycle dynamics. Furthermore, precise measurements of cell size and signals from fluorescent proteinsinE.colihavesuggestedthattherateofprotein biosynthesishas substantialcellcycledependence[22].
Thus, there is emerging evidence that prokaryotic
microbesalsoexperiencemetabolicchangesduringtheir celldivisioncycle.
Underconstantenvironments,cellularageingrepresents
another temporal change of metabolic phenotype and
henceisanadditionalcauseformetabolicheterogeneity. Buddingyeastages witheverydivision(replicative age-ing),andoldercellsshowincreasedlevelsofpyruvateand TCAcycleintermediatesanddecreasedlevelsof amino acids[8].Furthermore,ithasbeenfoundinreplicatively ageingyeastthatproteinsinvolvedintranslationbecome
more abundant relative to their transcripts, which
MetabolicheterogeneityTakhaveevandHeinemann 31
Figure1
Ecological factors Inherent dynamics Molecular noise
Surface Cell-to-cell interaction Time Current moment Time Current moment Cell cycle Cell ageing Electrical signalling Metabolite exchange Surface Fluctuation in gene expression Shift in conditions Distinct subpopulations Unimodal population
Current Opinion in Microbiology
Sourcesofmetabolicheterogeneityinmicrobialpopulations.Ecologicalfactors.Environmentalfactors,suchassurfaceandintercellular communications,canaffectindividualcellsdifferentlyand,therefore,stimulatetheirmetabolicdivergence.Inherentdynamics.Inamicrobial population,singlecellsusuallylacksynchronyincelldivisioncycleandageing.Consequently,atagivenmomentoftime,thecellsareinvarious phasesofcorrespondingmetabolicchanges.Molecularnoise.Fluctuationsingeneexpressioncangeneratevariabilityofenzymelevelsand metabolicfluxes,includinggrowthrates,amongsinglecells.Togetherwithparticularfeedbackcircuits,suchheterogeneitymayleadto multi-stabilityofmetabolism,forinstance,uponshiftsinnutrientconditions.
systems
biology
Experimental methods to quantify the levels of metabolites and biomass components in single cells.
Method Target Features Examples
Methods directly exploiting intrinsic properties of metabolites or biomass components Mass spectrometry (single-cell ‘metabolomics’) Absolute amount of primary and secondary metabolites
+ simultaneous measurement of up to 30 compounds + high-throughput disruptive, that is, non-dynamic
Metabolites of central carbon metabolism [54,45,73], highly abundant lipid species and pigments [46,47]
Exploiting
autofluorescence of metabolites
Concentration of cofactors and pigments
+ non-disruptive, that is, dynamic
mostly non-selective
Fluorescent microscopy: NAD(P)H [7], chlorophyll [54] FLIM:
decomposed NADH and NADPH [48], fraction of protein-bound to free NAD(P)H [49]
Raman spectroscopy Concentration of abundant chemical bonds, biomass components and pigments + non-disruptive mostly non-selective (regarding chemical compounds)
Lipids [50–52], DNA and proteins [51], paramylon and chlorophyll [52], astaxanthin [54], global metabolic profiling [74], carbon– deuterium chemical bond [53]
Methods assisted by interaction of metabolites with engineered proteins Transcription factor-based sensors Concentration of primary and secondary metabolites + genetically encoded non-ratiometric delayed response (based on the expression of a fluorescent protein)
NADPH [75], NAD+/NADH [76],
malonyl-CoA [55], p-coumaric acid [56], e-caprolactam, d-valerolactam and butyrolactam [77] Single fluorescent protein-based sensors Concentration of primary metabolites + genetically encoded + immediate response +/ ratiometric or non-ratiometric Non-ratiometric:
Trehalose [78], NAD+/NADH [79]
Ratiometric:
NAD+/NADH [58], NADH [80], NADPH [57] FRET-based sensors Concentration of primary
metabolites, signaling molecules, metal ions
+ genetically encoded + immediate response + ratiometric mostly <2-fold response
Two fluorescent proteins:
Pyruvate [61], trehalose-6-phosphate [81], ATP [82], cAMP [83]
One fluorescent protein: NADP+[62]
Methods assisted by interaction of metabolites with engineered RNA
RNA-based sensors expressing fluorescent proteins Concentration of secondary metabolites (big heterocyclic compounds) + genetically encoded + 7–11-fold response delayed response (current concept) non-ratiometric
Theophylline, tetracycline and neomycin [66]
RNA-based sensors with Spinach-like aptamers Concentration of primary metabolites, signaling molecules (big heterocyclic compounds) + genetically encoded + 5–25-fold response + immediate response require external fluorophore non-ratiometric
Adenosine, ADP, S-adenosylmethionine, guanine and GTP [63], cyclic di-GMP and cyclic AMP-GMP [68], cyclic di-AMP [69], thiamine 50-pyrophosphate [67] in Microbiol ogy 2018, 45 :30–38 www.sci encedirect.com
suggests thatbiosyntheticactivitieschangeas cellsage. Moreover, features of starvation and oxidative stress arealso inducedatoldages [9].Surprisingly,yeastcells surviving without nutrients (chronological ageing) are unexpectedlyfoundtoreducethefractionofmetabolites containingnaturallyoccurringheavyisotopes[23]. Over-all,ageingof thebuddingyeastis relatedtochangesin metabolism and thus contributes to themetabolic het-erogeneity ofapopulation.
Metabolic
heterogeneity
arising
from
molecular
noise
Inadditiontoecologicalfactorsandcell-inherent tempo-ral behavior, stochasticity in molecular interactions can giveriseto heterogeneityamongcells.Geneexpression involves interactions of molecules present at low copy numbers, whichmakestherateof thisprocess proneto fluctuations [1–3]. If flux-limiting metabolic enzymes and transporters are subject to such fluctuations, the noise in gene expressionmay propagate intothefluxes throughmetabolicpathways,asindicatedinrecent stud-ies[24,25].Infact,ithasbeenshowninE.colithatnoise
in the expression of an individual catabolic enzyme
propagates even further through the whole metabolic
network finally affecting thegrowth rate of single cells [26].Thealteredgrowthrate,inturn,hasbeenfoundto change the expression of other unrelated genes [26].
Such apparently growth-rate-related changes of gene
expression, however, could also be due to flux-sensing and flux-dependent regulation [27,28]. Overall, these findings indicatethatmolecular noisein theproduction ofevenoneenzymecanhaveglobaleffectsontheentire metabolism andexpressionofgenes.
As growth rate represents theultimate productof each cell’smetabolism,wecanuseitsdispersionamongcellsto assess themetabolic heterogeneity withinapopulation. Here,largescattersofgrowthratescanbefoundamong singlecellsinmicrobialpopulations[29–31].Cantherebe an advantage of such dispersion? Counterintuitively, according to experiments in both bacteria and yeast, having a wider distribution of single cell growth rates
does eventually increase the population growth rate
[29,30], which suggests that the latter is determined not only by the average cell division time. Besides, at lowpopulationgrowthrates,abiggerdispersionofsingle cellgrowthratesmakesthepopulationmoreadaptiveto rapid shifts to more favorable conditions [31]. Thus, intercellular metabolic heterogeneity, manifested in growthratevariability,canactuallyplayabeneficialrole increasingfitnessofthewholepopulation.
Metabolicheterogeneityamongcellsdoesnotnecessarily
have a unimodal form. For example, under the same
conditions, the microalga Chlamydomonas reinhardtii
appears in fast and slowlygrowingsubpopulations [32].
The extreme case of metabolic heterogeneity is the
occurrenceofnon-dividingcellsinagrowingpopulation. In diauxie, after depletingthe favorable carbon source, Lactococcuslactis formsasubpopulation of dormantcells [4]. Similarly, a shift from glucose to gluconeogenic carbon sources generates bistability in E. coli’s central carbonmetabolism,resultinginafractionofnon-growing or slowly growingcells [5,33,34].Furthermore, starved yeast cells provided with glucose or galactose form a subpopulationthatdoesnotmanagetobalanceglycolysis and thus undergoes growth arrest [6]. These findings show thatmetabolicheterogeneitycanbealsobimodal. Moreover,itseemsthatsuchmultiplestablephenotypes tendtoemergefromaunimodalpopulationduetoarapid changein theenvironment.
How can such multi-stability of metabolism emerge
within one population? The space of metabolic fluxes couldhaveseveralattractors,andinagivenenvironment cells would be situated around one of them [35]. A nutrient changecouldpositioncells onadifferent loca-tion in the space of metabolic fluxes, where there is comparable influence of two different attractors. If the cellshadslightlydifferentmetabolicfluxesdueto molec-ular noise, flux-sensing circuits, like E. coli’s positive
feedback FBP-Cra [28,36] and negative feedback
cAMP-Crp [37],woulddistribute thecells betweenthe two attractors. Thus, building on molecular noise and assistedbynutrientchanges,feedbackmechanismscould assigncellsofonepopulationtodifferentattractors,that is,differentmetabolicphenotypes.
Cellsdrawnintoanattractorofnoorslowgrowth repre-sentahighlyrelevantphenomenonastheycanturninto so-called persisters, that is, cells that exhibit tolerance towardsantibiotics[35].Recently,themetabolic pheno-type of E. coli’spersistershasbeen unraveled,whereit wasfoundthattheirproteomeischaracterizedbyaglobal stress response, and that these cells are metabolically active with upregulated catabolism, very slow buildup ofbiomassandreducedmetabolitepools[34].Persisters oftheGram-positivebacteriumS.aureushavebeenfound tohavedecreasedlevelsofATP[38],whichmayreflecta strongperturbationoftheirmetabolichomeostasis,which wasfoundtotriggerpersistenceinE.coli[34,35]. Inter-estingly,recentfindingsindicatethatpersistersoccurnot only in bacteria. Cells resistant to harsh environmental conditions,includingdrugtreatment,alsoappearinyeast [39].Overall,dormantand,particularly,persistercellsare anexampleofextrememetabolicheterogeneitythatcan bepresentin aclonalpopulationof microbes.
Experimental
tools
to
uncover
metabolic
properties
of
single
cells
Tofurtheruncovermetabolicheterogeneityinmicrobial populations and mechanisms generating it, single-cell measurementsofmetabolicpropertiesarerequired. How-ever,herewearestillconfrontedwithgrandchallenges.
MetabolicheterogeneityTakhaveevandHeinemann 33
First,unlikeproteins,metabolitescannotbefluorescently tagged.Second,eveniftheconcentrationofametabolite canbemeasured,inmostcasestheconcentrationdoesnot saymuchabouttheactivityof thecorresponding meta-bolicpathway.
Below, we illustrate recent exciting developments that havebeenmadetowardsquantifyingthelevelsof metab-olitesandbiomasscomponentsinsinglecells.Measuring fluxesthroughspecificmetabolicpathwaysinsinglecells is,however,stillunattainable.Onlytheresultingactivity of many metabolic pathways, for example, the flux to biomass,thatis,thegrowthrate,canbegauged,evenwith high precision. Next to determining single-cell growth ratesfrommicroscopictime-lapsedata[29–31],arecent technologyexploitsresonantmasssensorstomeasurethe buoyantmassandhencethegrowthrateofsinglecellsin high-throughput manner [40]. Another microresonator set-up has quantified the total mass of a single cell andunexpectedlyrevealeditsfluctuationsinthesecond range, which seem to be linked to water transport and ATPsynthesis[41].
Metabolicactivitycanalsobecrudelyassessedinsingle cells via the assimilation of externally provided com-pounds,thatis,theaccumulationofcertainatomsinside the cells. Particularly, nanometer-scale secondary ion massspectrometry(NanoSIMS)hasidentified heteroge-neity in the activityof CO2and N2fixation withinthe populationofChlorobiumphaeobacteroidesincubatedwith thelabelledgases[42].Furthermore,recentNanoSIMS experimentshaveshownthatS.aureuscells incorporate heavywaterwithmarkedlydifferentrates[43].Ofnote, the cells analyzed in these two studies were extracted fromnaturalhabitats,thustheidentifiedmetabolic vari-abilitycouldalsobeduetogeneticdifferences.Arecent NanoSIMSstudy,usingaclonalE.colipopulationgrown ona mixtureof isotope-labelled arabinoseand glucose, hasdisclosedthatindividualcellsassimilatethesesugars withdifferentpreferences[44].Thus,NanoSIMS with the help of labelled nutrients can assess the resulting assimilation flux in single cells and identify metabolic
heterogeneityamongthem.
To measure metabolite levels in single microbial cells, mass spectrometry can be utilized (Table 1). Here,
through microarrays for mass spectrometry (MAMS),
single or few cells have been isolated in ‘wells’ with subsequent matrix-assisted laser desorption/ionization (MALDI)oftheircontent.Withthistechnique,19 metab-olitesofcentralcarbonmetabolismcouldbeidentifiedin tens of single yeast cells. On the basis of such data, two subpopulations with either low or high amounts of fructose-1,6-bisphosphate could be distinguished [45].
Recent experiments employing MAMS have allowed
high-throughput screening of thousands of individual C. reinhardtii cells with measuring 22 highly abundant
lipidsandpigments[46].Withanothertechnique,aerosol time-of-flightmassspectrometry(ATOFMS),individual cellsofthesamealgaexperiencingfourdaysofnitrogen limitationhavebeenanalyzedatthefrequencyof50Hz. Ithasbeenfoundthatontheseconddaythevariability amongcellsinthelevelofonelipidis40%biggerthanon any other day, suggesting a subpopulation with slower response to nitrogen deprivation [47]. Given the high throughputpossibilityof thesesingle-cellmass spectro-metrictechniques, it should now bepossible to screen individualcellsofentiremicrobialpopulations.However, thenumberof identified metabolitesis still rather lim-ited,and theirlevelsareonly qualitativelyassessed. Single-cell concentrations of some metabolites can be assessedbyexploitingtheirintrinsicspectroscopic prop-erties, particularly their autofluorescence. In a recent
study using the autofluorescence of NAD(P)H, it has
been shown that metabolite oscillations exist in single cellsofbuddingyeast[7].AssessingNAD(P)H concen-trationviaitsautofluorescencehasrecentlybecome pos-sible even in the much smaller cells of E. coli, whose metabolism seems to oscillate as well during the cell divisioncycle[20].
By means of fluorescence-lifetime imaging microscopy (FLIM),itisevenpossibleto decomposethesignals of
NADPH and NADH in single-cellmeasurements [48].
Furthermore, the FLIM-phasor technique could
deter-minetheratiobetweenprotein-boundtofreeNAD(P)H molecules.Recently,ithasbeenshown thatsingle cells ofdiversebacteria,includingE.coli,B.subtilisand Staph-ylococcus epidermidis, modulate this ratio in response to environmental conditions. Particularly, increased free NAD(P)Happearswith antibiotictreatment[49]. WithRamanspectroscopy,utilizingtheinelastic scatter-ing of light by chemical bonds, it is also possible to measure a number of biomolecules in single microbial cells.Coherentanti-StokesRamanspectroscopy(CARS) couldquantifyneutrallipidsinindividualyeastcells[50]. WithstimulatedRamanscattering(SRS)microscopyitis feasibleto measure DNAas wellas proteins and lipids [51].SRS,beingafasttechnique,hasrecentlyrevealed variabilityinlipid,paramylonandchlorophyllcontentin the motile microalga Euglena gracile [52]. In labelling experimentswithheavywater,single-cellRaman
micro-spectroscopy has been elegantly applied to estimate
bacterialviabilityunderantibioticinfluence,specifically, bytrackingsubstitutionofC–HbyC–Dbondin macro-molecules[53].
Through application of different single-cell techniques on the very same cell, interesting insights have been obtained. For instance, first, fluorescent and Raman microscopieswere usedto determineconcentrations of thepigmentschlorophyllandastaxanthininsinglecellsof
the algaeHaematococcus pluvialis. Thereafter, the abun-dances of13primary metabolitesweremeasuredin the same cellsusingMAMS. Here,acrossthepopulationof cells beingin various stagesof encystment,it hasbeen
found that ATP/ADP ratio anti-correlates with the
ratio between astaxanthin and chlorophyll contents
[54].Thus,beforedisruptingcellsformassspectrometry, they can be studied spectroscopically to maximize the information gain.
If dynamicsingle-cell measurementsof metabolites are necessary, but the metabolites are not fluorescent or cannotbeaccessedviaRamanspectroscopy,then geneti-callyencodedbiosensors,selectivelyreportingmetabolite concentrations,arethemethodofchoice.Here,inrecent years, three different approaches exploiting interaction of targetmetaboliteswith proteinshave beenexplored: transcription factor-based, single fluorescent protein-basedandFRET-(i.e.Fo¨rsterResonanceEnergy Trans-fer) basedbiosensors (Table1).
Transcriptionfactorsthatbindspecificmetaboliteshave beenengineeredintosensorsdrivingtheexpressionofa fluorescent protein and thus providing a proxy for the metaboliteconcentration.Withsuchtranscription factor-based biosensors, populations of E. coli,Corynebacterium glutamicumandSaccharomycescerevisiaehaverecentlybeen sortedinsearchofsubpopulationswiththehighestlevel of industrially desired metabolites [55,56]. Relying on fluorescent protein expression, these sensors, however, cannot be used for tracking fast changes in metabolite concentration.
Contrarily, another sensor concept,where ametabolite binding domain isfused to asingle fluorescent protein andinduceschangesinitsfluorescence,allowsreporting transientmetaboliteconcentration.Recently,sensors
uti-lizing this concept have been developed for NADPH
[57] and theratio of NADH and NAD+ [58].Grafting
GFPintoammoniumtransceptorshasproduced
biosen-sorswhosefluorescentresponsecorrelateswiththe trans-port activity, however, this may represent sensing of extracellular ammonium concentration only rather than the flux itself [59]. The disadvantage of some of such sensors, and also of all transcription factor-based ones, is that their read-out (fluorescence) can be easily con-foundedbysensorexpressionlevels.
FRET-based sensors are a solution against such
con-founding effects, because they are intrinsically ratio-metric. In most of these sensors, a metabolite binding domain links two fluorescent proteins and affects the
efficiency of FRET between them upon metabolite
binding.WithaFRET-basedsensorexpressedin Myco-bacterium smegmatis, it has been found that maintaining high ATP levels during antibiotic exposure correlates withcells’abilitytoresumegrowthinnormalconditions
[60]. A sensor homologous to the previous one has
recentlydisclosedATPoscillationsinsinglecellsofyeast [7]. Interestingly, a pyruvate FRET-sensor has been used to infer the flux of this metabolite into the mito-chondria,withtheapplicationofaninhibitorwhichstops thepyruvateinfluxintothecell[61].FRET-based sen-sors can also rely on one fluorescent protein. Here, a sensor dimerizesupon metabolitebinding and causesa decrease in steady-state fluorescence anisotropy, thus reportingtheconcentrationofthesubstrate.Suchsensor hasbeendevelopedforNADP+[62].FRET-sensorsare excellent tools offering real-time dynamic read-outs of metabolite concentration. However, the set of protein domains able to bind metabolitesand appropriately re-orientthefluorescentproteinstoenableFRETseemsto belimitedinnature[63].
Nucleic acid-based sensors might be an alternative, as theycouldbedevelopedfor anymetabolite dueto effi-cientinvitroselectionofaptamers,namelySELEX[64], andversatilityofnaturallyoccurringriboswitches[65].In
recent RNA-based sensors, mRNA encoding a
fluores-cent protein contains an aptamer and a self-cleaving ribozyme so that, upon metabolite binding, the former modulates the latter, eventually controlling translation [66]. In another design, an invitro selected aptameror riboswitchbindsametaboliteandenablesanother apta-mer,specificallySpinachorSpinach2,toactivate fluores-cenceofanexternallyaddedchemical.Studies
employ-ing such sensors have reported a large cell-to-cell
variabilityintheconcentrationofS-adenosylmethionine [63] and thiamine 50-pyrophosphate [67] in cells of an E. colipopulation.RNA-basedsensorscanleadto 5–25-fold increases in fluorescence upon metabolite binding [63,66–69]whichisdramaticallyhigherthanrather mod-estresponsesofFRET-basedsensors.Newfluorophores with different spectral properties[70] and recently dis-covered alternative fluorophore-binding aptamers, like baby Spinach [71] and Broccoli [72], may further fuel developmentofRNA-basedsensors.Suchfurtherworkis necessaryasthecurrentlyavailableRNA-based metabo-lite sensors can measure only large heterocyclic com-pounds, and unfortunately not yetthe small and nega-tively-chargedmetabolitesof primarymetabolism.
Future
avenues
As shown in this review, the cells of clonal microbial populationscanexhibitsignificantmetabolic heterogene-ity, which in the most extreme case leads to the co-existenceofgrowingandnon-growingcells.Manyopen questions exist,forexample,onthesourceofmetabolic oscillations during the cell division cycle and on the originsofageing-associatedchangesinmetabolism. Fur-ther, the mechanisms causing microbial populations to
end up in metabolically different phenotypes remain
mostly unknown.
MetabolicheterogeneityTakhaveevandHeinemann 35
Keytounravelthesequestionswillbemethodstoprobe metabolism on the single cell level. Here, significant advanceshavebeenmadeinrecentyears.Nevertheless, manyofthecurrenttoolsarestillattheproof-of-concept level,meaningthatthoroughinvivovalidationis neces-sarybeforethesetoolscanberoutinelyappliedforactual research.Duetothetheoreticallyunlimitedversatilityof nucleic acid-based metabolite sensors, we expect that theywillbecomethestandardtoolto dynamically visu-alize metabolite levels in single cells, although, admit-tedly, still many technicalproblems need to besolved withsuchsensors.
The ultimate challenge, however, will be to devise
methods to sense or visualize the functional output of metabolism,thatis,themetabolicfluxesthroughspecific metabolicpathways,insinglecells.Measuringmetabolic fluxes in single cells,ideallyin dynamic fashion, repre-sents an extreme technical challenge, and even onthe conceptuallevel itis notclear how thiscould bedone. Onepossibilitymightbetoexploitflux-signaling metab-olites, whose concentration strictly correlates with the corresponding metabolic flux [27,36], and develop bio-sensorsfor these metabolites to obtain dynamic single-cellfluxmeasurements.
Author
contribution
VT and MH conceived the study and wrote the
manuscript.
References
and
recommended
reading
Papersofparticularinterest,publishedwithintheperiodofreview, havebeenhighlightedas:ofspecialinterest ofoutstandinginterest
1. ElowitzMB,LevineAJ,SiggiaED,SwainPS:Stochasticgene expressioninasinglecell.Science(80-)2002,297:1183-1186. 2. GoldingI,PaulssonJ,ZawilskiSM,CoxEC:Real-timekineticsof geneactivityinindividualbacteria.Cell2005,123:1025-1036. 3. RajA,vanOudenaardenA:Nature,nurture,orchance:
stochasticgeneexpressionanditsconsequences.Cell2008, 135:216-226.
4. SolopovaA,vanGestelJ,WeissingFJ,BachmannH,TeusinkB, KokJ,KuipersOP:Bet-hedgingduringbacterialdiauxicshift. ProcNatlAcadSci2014,111:7427-7432.
5. KotteO,VolkmerB,RadzikowskiJL,HeinemannM:Phenotypic bistabilityinEscherichiacoli’scentralcarbonmetabolism.Mol SystBiol2014,10736-736.
6. VanHeerdenJH,WortelMT,BruggemanFJ,HeijnenJJ, BollenYJM,Planque´ R,HulshofJ,O’TooleTG,WahlSA, TeusinkB:Lostintransition:start-upofglycolysisyields subpopulationsofnongrowingcells.Science(80-)2014, 343:1245114.
7.
PapagiannakismetabolicoscillationsA,NiebelrobustlyB,WitEC,gateHeinemanntheearlyM:andAutonomouslatecell cycle.MolCell2017,65:285-295.
ThispapershowsthatinsinglecellsofyeastthelevelsofNAD(P)Hand ATPoscillateduringthecellcycleandalsoinitsabsence.Thisstudy exploitsmicrofluidics,metaboliteautofluorescenceandaFRET-based sensor.
8. KameiY,TamadaY,NakayamaY,FukusakiE,MukaiY:Changes intranscriptionandmetabolismduringtheearlystageof replicativecellularsenescenceinbuddingyeast.JBiolChem 2014,289:32081-32093.
9. JanssensGE,MeinemaAC,Gonza´lezJ,WoltersJC,SchmidtA, GuryevV,BischoffR,WitEC,VeenhoffLM,HeinemannM:Protein biogenesismachineryisadriverofreplicativeaginginyeast. Elife2015,4:e08527.
10. XiaoY,BowenCH,LiuD,ZhangF:Exploitingnongenetic cell-to-cellvariationforenhancedbiosynthesis.NatChemBiol2016, 12:339-344.
11. NagarajanS,KruckebergAL,SchmidtKH,KrollE,HamiltonM, McInnerneyK,SummersR,TaylorT,RosenzweigF:Uncoupling reproductionfrommetabolismextendschronologicallifespan inyeast.ProcNatlAcadSci2014,111:E1538-E1547.
12. CampbellK,Herrera-DominguezL,Correia-MeloC,ZelezniakA, RalserM:Biochemicalprinciplesenablingmetabolic cooperativityandphenotypicheterogeneityatthesinglecell level.CurrOpinSystBiol2017http://dx.doi.org/10.1016/j. coisb.2017.12.001.
13. CampbellK,VowinckelJ,Mu¨llederM,MalmsheimerS, LawrenceN,CalvaniE,Miller-FlemingL,AlamMT,ChristenS, KellerMAetal.:Self-establishingcommunitiesenable cooperativemetaboliteexchangeinaeukaryote.Elife2015,4. 14. CampbellK,VowinckelJ,RalserM:Cell-to-cellheterogeneity
emergesasconsequenceofmetaboliccooperationina syntheticyeastcommunity.BiotechnolJ2016,11:1169-1178. 15. LiuJ,PrindleA,HumphriesJ,Gabalda-SagarraM,AsallyM,
LeeDYD,LyS,Garcia-OjalvoJ,Su¨elGM:Metabolic co-dependencegivesrisetocollectiveoscillationswithin biofilms.Nature2015,523:550-554.
16. PrindleA,LiuJ,AsallyM,LyS,Garcia-OjalvoJ,Su¨elGM:Ion channelsenableelectricalcommunicationinbacterial communities.Nature2015,527:59-63.
17.
SagarraLiuJ,Martinez-CorralM,Garcia-OjalvoR,PrindleJ,Su¨elGM:A,LeeCouplingDYD,LarkinbetweenJ,Gabalda-distant biofilmsandemergenceofnutrienttime-sharing.Science(80-) 2017,356:638-642.
ThispapershowsthatneighbouringbiofilmsofB.subtiliscommunicate viaelectricalsignallingandgrowwithanti-phaseoscillatingspeeds.
18.
BlankJohnsonHM,CD,PerezBankaitisR,HeVA,C,MaitraKennedyN,MetzBKetR,al.:HillTranslationalJ,LinY, controloflipogenicenzymesinthecellcycleofsynchronous, growingyeastcells.EMBOJ2017,36:487-502.
ThispapershowsthatthetranslationalefficiencyofmRNAsencoding lipogenicenzymeschangesthroughoutthecellcycle.
19. LoriC,OzakiS,SteinerS,Bo¨hmR,AbelS,DubeyBN,SchirmerT, HillerS,JenalU:Cyclicdi-GMPactsasacellcycleoscillatorto drivechromosomereplication.Nature2015,523:236-239. 20. ZhangZ,Milias-ArgeitisA,HeinemannM:Dynamicsingle-cell
NAD(P)Hmeasurementrevealsoscillatorymetabolism throughouttheE.colicelldivisioncycle.SciRep2018,8:2162. 21. YaginumaH,KawaiS,TabataKV,TomiyamaK,KakizukaA,
KomatsuzakiT,NojiH,ImamuraH:DiversityinATP
concentrationsinasinglebacterialcellpopulationrevealedby quantitativesingle-cellimaging.SciRep2014,4:6522. 22. WalkerN,NgheP,TansSJ:Generationandfilteringofgene
expressionnoisebythebacterialcellcycle.BMCBiol2016, 14:11.
23. LiX,SnyderMP:Yeastlongevitypromotedbyreversing aging-associateddeclineinheavyisotopecontent.npjAgingMech Dis2016,2:16004.
24. NikolicN,BarnerT,AckermannM:Analysisoffluorescent reportersindicatesheterogeneityinglucoseuptakeand utilizationinclonalbacterialpopulations.BMCMicrobiol2013, 13:258.
25. WelkenhuysenN,BorgqvistJ,BackmanM,BendriouaL, Gokso¨rM,AdielsCB,CvijovicM,HohmannS:Single-cellstudy
linksmetabolismwithnutrientsignalingandrevealssources ofvariability.BMCSystBiol2017,11:59.
26.
StochasticityKivietDJ,NgheofP,metabolismWalkerN,BoulineauandgrowthS,Sunderlikovaatthesingle-cellV,TansSJ: level.Nature2014,514:376-379.
Thispapershowsthatfluctuationsintheexpressionofsinglecatabolic enzymespropagateintochangesofacell’sgrowthrate,which,inturn, affecttheexpressionofgenes,includingunrelatedones.
27. LitsiosA,OrtegaA´D,WitEC,HeinemannM:Metabolic-flux dependentregulationofmicrobialphysiology.CurrOpin Microbiol2018,42:71-78.
28. KochanowskiK,VolkmerB,GerosaL,HaverkornvanRijsewijkBR, SchmidtA,HeinemannM:Functioningofametabolic fluxsensorinEscherichiacoli.ProcNatlAcadSci2013, 110:1130-1135.
29. HashimotoM,NozoeT,NakaokaH,OkuraR,AkiyoshiS, KanekoK,KussellE,WakamotoY:Noise-drivengrowthrate gaininclonalcellularpopulations.ProcNatlAcadSci2016, 113:3251-3256.
30. CerulusB,NewAM,PougachK,VerstrepenKJ:Noiseand epigeneticinheritanceofsingle-celldivisiontimesinfluence populationfitness.CurrBiol2016,26:1138-1147.
31. RochmanN,SiF,SunSX:Togrowisnotenough:impactof noiseoncellenvironmentalresponseandfitness.IntegrBiol 2016,8:1030-1039.
32. DamodaranSP,EberhardS,BoitardL,RodriguezJG,WangY, BremondN,BaudryJ,BibetteJ,WollmanFA:Amillifluidicstudy ofcell-to-cellheterogeneityingrowth-rateandcell-division capabilityinpopulationsofisogeniccellsofChlamydomonas reinhardtii.PLoSOne2015,10:e0118987.
33. AmatoS,OrmanM,BrynildsenM:Metaboliccontrol ofpersisterformationinEscherichiacoli.MolCell2013, 50:475-487.
34.
RadzikowskiHeinemannM:JL,BacterialVedelaarpersistenceS,SiegelD,isOrtegaanactiveA´D,SchmidtsSstressA, responsetometabolicfluxlimitation.MolSystBiol2016, 12:882.
ThispapercharacterisesthemetabolicphenotypeofE.colipersisters. Particularly, it shows that they are metabolically active and have decreasedmetabolitepools.
35. RadzikowskiJL,SchramkeH,HeinemannM:Bacterial persistencefromasystem-levelperspective.CurrOpin Biotechnol2017,46:98-105.
36. KotteO,ZauggJB,HeinemannM:Bacterialadaptationthrough distributedsensingofmetabolicfluxes.MolSystBiol2010, 6:355.
37. HermsenR,OkanoH,YouC,WernerN,HwaT:Agrowth-rate compositionformulaforthegrowthofE.colionco-utilized carbonsubstrates.MolSystBiol2015,11801-801.
38. ConlonBP,RoweSE,GandtAB,NuxollAS,DoneganNP,ZalisEA, ClairG,AdkinsJN,CheungAL,LewisK:Persisterformationin StaphylococcusaureusisassociatedwithATPdepletion.Nat Microbiol2016,1:16051.
39. YaakovG,LernerD,BenteleK,SteinbergerJ,BarkaiN:Coupling phenotypicpersistencetoDNAdamageincreasesgenetic diversityinseverestress.NatEcolEvol2017,1:16. 40. CermakN,OlcumS,DelgadoFF,WassermanSC,PayerKR,
MurakamiMA,KnudsenSM,KimmerlingRJ,StevensMM, KikuchiYetal.:High-throughputmeasurementofsingle-cell growthratesusingserialmicrofluidicmasssensorarrays.Nat Biotechnol2016,34:1052-1059.
41.
Martı´nez-Martı´n D,BeerliC,MercerJ,Fla¨schnerGerberC,G,Mu¨llerGaubDJ:B,InertialMartinS,picobalanceNewtonR, revealsfastmassfluctuationsinmammaliancells.Nature 2017,550:500-505.
Thispaperpresentsatooltomeasurethetotalmassofasinglecellwith picogramsensitivityandmillisecondresolution.Thismethodrevealscell massfluctuationsinthesecondrangethatseemtobelinkedtoATP synthesisandwatertransport.
42. ZimmermannM,EscrigS,Hu¨bschmannT,KirfMK,BrandA, InglisRF,MusatN,Mu¨llerS,MeibomA,AckermannMetal.: Phenotypicheterogeneityinmetabolictraitsamongsingle cellsofararebacterialspeciesinitsnaturalenvironment quantifiedwithacombinationofflowcellsortingand NanoSIMS.FrontMicrobiol2015,6:243.
43. KopfSH,SessionsAL,CowleyES,ReyesC,VanSambeekL,HuY, OrphanVJ,KatoR,NewmanDK:Traceincorporationof heavywaterrevealsslowandheterogeneouspathogen growthratesincysticfibrosissputum.ProcNatlAcadSci2016, 113:E110-E116.
44.
LittmannNikolicN,S,SchreiberKuypersF,MMM,DalCoAckermannA,KivietM:DJ,Cell-to-cellBergmillerT,variation andspecializationinsugarmetabolisminclonalbacterial populations.PLoSGenet2017,13:1-24.
ThispapershowsthatinclonalE.colipopulationgrownonamixtureof glucoseandarabinose,individualcellsassimilatethesemetaboliteswith differentpreferences.Theassimilationofthesugarsismeasuredwith NanoSIMS.
45. Iba´n˜ezAJ,FagererSR,SchmidtAM,UrbanPL,JefimovsK, GeigerP,DechantR,HeinemannM,ZenobiR:Mass
spectrometry-basedmetabolomicsofsingleyeastcells.Proc NatlAcadSci2013,110:8790-8794.
46. KrismerJ,SobekJ,SteinhoffRF,FagererSR,PabstM,ZenobiR: ScreeningofChlamydomonasreinhardtiipopulationswith single-cellresolutionbyusingahigh-throughputmicroscale samplepreparationformatrix-assistedlaserdesorption ionizationmassspectrometry.ApplEnvironMicrobiol2015, 81:5546-5551.
47. CahillJF,DarlingtonTK,FitzgeraldC,SchoeppNG,BeldJ, BurkartMD,PratherKA:Onlineanalysisofsingle Cyanobacteriaandalgaecellsundernitrogen-limited conditionsusingaerosoltime-of-flightmassspectrometry. AnalChem2015,87:8039-8046.
48. BlackerTS,MannZF,GaleJE,ZieglerM,BainAJ,SzabadkaiG, DuchenMR:SeparatingNADHandNADPHfluorescenceinlive cellsandtissuesusingFLIM.NatCommun2014,5:3936. 49. BhattacharjeeA,DattaR,GrattonE,HochbaumAI:Metabolic
fingerprintingofbacteriabyfluorescencelifetimeimaging microscopy.SciRep2017,7:3743.
50. ChumnanpuenP,BrackmannC,NandySK,
ChatzipapadopoulosS,NielsenJ,EnejderA:Lipidbiosynthesis monitoredatthesingle-celllevelinSaccharomyces cerevisiae.BiotechnolJ2012,7:594-601.
51. LuF-K,BasuS,IgrasV,HoangMP,JiM,FuD,HoltomGR, NeelVA,FreudigerCW,FisherDEetal.:Label-freeDNAimaging invivowithstimulatedRamanscatteringmicroscopy.Proc NatlAcadSci2015,112:11624-11629.
52. WakisakaY,SuzukiY,IwataO,NakashimaA,ItoT,HiroseM, DomonR,SugawaraM,TsumuraN,WataraiHetal.:Probingthe metabolicheterogeneityofliveEuglenagraciliswith stimulatedRamanscatteringmicroscopy.NatMicrobiol2016, 1:16124.
53.
Metabolic-activity-basedTaoY,WangY,HuangS,ZhuassessmentP,HuangofWE,antimicrobialLingJ,XuJ:effects byD2O-labeledsingle-cellramanmicrospectroscopy.Anal Chem2017,89:4108-4115.
ThispapershowsthatusingsinglecellRamanmicrospectroscopyand heavywateritispossibletoassesstheviabilityofpathogenicmicrobes bytrackingtheshiftfromC–HtoC–Dbond.
54. FagererSR,SchmidT,Iba´n˜ezAJ,PabstM,SteinhoffR, JefimovsK,UrbanPL,ZenobiR:Analysisofsinglealgalcellsby combiningmassspectrometrywithRamanandfluorescence mapping.Analyst2013,138:6732.
55. LiS,SiT,WangM,ZhaoH:Developmentofasynthetic malonyl-CoAsensorinSaccharomycescerevisiaeforintracellular metabolitemonitoringandgeneticscreening.ACSSynthBiol 2015,4:1308-1315.
56. SiedlerS,KhatriNK,Zsoha´rA,KjærbøllingI,VogtM,HammarP, NielsenCF,MarienhagenJ,SommerMOA,JoenssonHN: Developmentofabacterialbiosensorforrapidscreeningof
MetabolicheterogeneityTakhaveevandHeinemann 37
yeastp-coumaricacidproduction.ACSSynthBiol2017, 6:1860-1869.
57.
TaoSuNR,etZhaoal.:GeneticallyY,ChuH,WangencodedA,ZhufluorescentJ,ChenX,ZousensorsY,ShiM,revealLiuR, dynamicregulationofNADPHmetabolism.NatMethods2017, 14:720-728.
Thispaperpresentsaratiometricgeneticallyencodedbiosensor mea-suringNADPHconcentration.
58. ZhaoY,HuQ,ChengF,SuN,WangA,ZouY,HuH,ChenX, ZhouHM,HuangXetal.:SoNar,ahighlyresponsiveNAD +/NADHsensor,allowshigh-throughputmetabolicscreening ofanti-tumoragents.CellMetab2015,21:777-789.
59. DeMicheleR,AstC,Loque´ D,HoCH,AndradeSLA,LanquarV, GrossmannG,GehneS,KumkeMU,FrommerWB:Fluorescent sensorsreportingtheactivityofammoniumtransceptorsin livecells.Elife2013,2013:e00800.
60. Maglica,O¨ zdemirE,McKinneyJD:Single-celltrackingreveals antibiotic-inducedchangesinmycobacterialenergy metabolism.MBio2015,6:e02236-e2314.
61. SanMartı´n A,CeballoS,Baeza-LehnertF,LerchundiR, ValdebenitoR,Contreras-BaezaY,Alegrı´a K,BarrosLF:Imaging mitochondrialfluxinsinglecellswithaFRETsensorfor pyruvate.PLoSOne2014,9:e85780.
62. CameronWD,BuiCV,HutchinsonA,LoppnauP,Gra¨slundS, RocheleauJV:Apollo-NADP+:aspectrallytunablefamilyof geneticallyencodedsensorsforNADP+.NatMethods2016, 13:352-358.
63. PaigeJS,Nguyen-DucT,SongW,JaffreySR:Fluorescence imagingofcellularmetaboliteswithRNA.Science(80-)2012, 335:1194.
64. EllingtonAD,SzostakJW:InvitroselectionofRNAmolecules thatbindspecificligands.Nature1990,346:818-822. 65. MellinJR,CossartP:Unexpectedversatilityinbacterial
riboswitches.TrendsGenet2015,31:150-156.
66. TownshendB,KennedyAB,XiangJS,SmolkeCD: High-throughputcellularRNAdeviceengineering.NatMethods 2015,12:989-994.
67. YouM,LitkeJL,JaffreySR:Imagingmetabolitedynamicsin livingcellsusingaSpinach-basedriboswitch.ProcNatlAcad Sci2015,112:E2756-E2765.
68. KellenbergerCA,WilsonSC,Sales-LeeJ,HammondMC: RNA-basedfluorescentbiosensorsforlivecellimagingofsecond messengerscyclicdi-GMPandcyclicAMP-GMP.JAmChem Soc2013,135:4906-4909.
69. KellenbergerCA,ChenC,WhiteleyAT,PortnoyDA,
HammondMC:RNA-basedfluorescentbiosensorsforlivecell imagingofsecondmessengercyclicdi-AMP.JAmChemSoc 2015,137:6432-6435.
70. SongW,StrackRL,SvensenN,JaffreySR:Plug-and-play fluorophoresextendthespectralpropertiesofspinach.JAm ChemSoc2014,136:1198-1201.
71. WarnerKD,ChenMC,SongW,StrackRL,ThornA,JaffreySR, Ferre´-D’Amare´ AR:Structuralbasisforactivityofhighly efficientRNAmimicsofgreenfluorescentprotein.NatStruct MolBiol2014,21:658-663.
72. FilonovGS,MoonJD,SvensenN,JaffreySR:Broccoli:rapid selectionofanRNAmimicofgreenfluorescentproteinby fluorescence-basedselectionanddirectedevolution.JAm ChemSoc2014,136:16299-16308.
73. SchmidtAM,FagererSR,JefimovsK,BuettnerF,MarroC, SiringilEC,BoehlenKL,PabstM,Iba´n˜ezAJ:Molecular phenotypicprofilingofaSaccharomycescerevisiaestrainat thesingle-celllevel.Analyst2014,139:5709-5717.
74. TengL,WangX,WangX,GouH,RenL,WangT,WangY,JiY, HuangWE,XuJ:Label-free,rapidandquantitative
phenotypingofstressresponseinE.coliviaramanome.Sci Rep2016,6:34359.
75. SiedlerS,SchendzielorzG,BinderS,EggelingL,BringerS,BottM: SoxRasasingle-cellbiosensorforNADPH-consuming enzymesinEscherichiacoli.ACSSynthBiol2014,3:41-47. 76. KnudsenJD,CarlquistM,Gorwa-GrauslundM:
NADH-dependentbiosensorinSaccharomycescerevisiae: principleandvalidationatthesinglecelllevel.AMBExp2014, 4:1-12.
77. ZhangJ,BarajasJF,BurduM,RueggTL,DiasB,KeaslingJD: Developmentofatranscriptionfactor-basedlactam biosensor.ACSSynthBiol2017,6:439-445.
78. NadlerDC,MorganSA,FlamholzA,KortrightKE,SavageDF: Rapidconstructionofmetabolitebiosensorsusing domain-insertionprofiling.NatCommun2016,7:12266.
79. HungYP,YellenG:Live-cellimagingofcytosolicNADH-NAD+ redoxstateusingageneticallyencodedfluorescent biosensor.MethodsMolBiol2014,1071:83-95.
80. ZhaoY,JinJ,HuQ,ZhouHM,YiJ,YuZ,XuL,WangX,YangY, LoscalzoJ:Geneticallyencodedfluorescentsensors forintracellularNADHdetection.CellMetab2011, 14:555-566.
81. PerozaEA,EwaldJC,ParakkalG,SkotheimJM,ZamboniN:A geneticallyencodedFo¨rsterresonanceenergytransfer sensorformonitoringinvivotrehalose-6-phosphate dynamics.AnalBiochem2015,474:1-7.
82. ImamuraH,HuynhNhatKP,TogawaH,SaitoK,IinoR, Kato-YamadaY,NagaiT,NojiH:VisualizationofATPlevelsinside singlelivingcellswithfluorescenceresonanceenergy transfer-basedgeneticallyencodedindicators.ProcNatlAcad Sci2009,106:15651-15656.
83. KlarenbeekJ,GoedhartJ,VanBatenburgA,GroenewaldD, JalinkK:Fourth-generationEpac-basedFRETsensorsfor cAMPfeatureexceptionalbrightness,photostabilityand dynamicrange:characterizationofdedicatedsensorsfor FLIM,forratiometryandwithhighaffinity.PLoSOne2015,10: e0122513.