Correlative microscopy for structural microbiology Stuart C Howes1, Roman I Koning
1,2 and Abraham J Koster
1,2
Understandinghowmicrobesutilizetheirenvironmentisaidedby visualizingthemintheirnaturalcontextathighresolution.
Correlativeimagingenablesefficienttargetingandidentificationof labelledviralandbacterialcomponentsbylightmicroscopy combinedwithhighresolutionimagingbyelectronmicroscopy.
Advancesingeneticandbioorthogonallabelling,improved workflowsfortargetingandimagecorrelation,andlarge-scaledata collectionareincreasingtheapplicabilityofcorrelativeimaging methods.Furthermore,developmentsinmassspectroscopyand softX-rayimagingareexpandingthecorrelativeimagingmodalities available.Investigatingthestructureandorganizationofmicrobes withintheirhostbycombinedimagingmethodsprovidesimportant insightsintomechanismsofinfectionanddiseasewhichcannotbe obtainedbyothertechniques.
Addresses
1LeidenUniversityMedicalCentre,DepartmentofMolecularCell Biology,POBox9600,2300RCLeiden,TheNetherlands
2NetherlandsCentreforElectronNanoscopy,InstituteofBiology, LeidenUniversity,POBox9502,2300RALeiden,TheNetherlands
Correspondingauthor:Koster,AbrahamJ(A.J.Koster@lumc.nl)
CurrentOpinioninMicrobiology2018,43:132–138
ThisreviewcomesfromathemedissueonThenewmicroscopy EditedbyArianeBriegelandStephanUphoff
ForacompleteoverviewseetheIssueandtheEditorial Availableonline4thFebruary2018
https://doi.org/10.1016/j.mib.2018.01.009
1369-5274/ã2018TheAuthors.PublishedbyElsevierLtd.Thisisan openaccessarticleundertheCCBY-NC-NDlicense(http://creative- commons.org/licenses/by-nc-nd/4.0/).
Introduction
All infections have their origins in the invasion and proliferation of microbes in their host. Visualization of bacteriaandvirusesintheirnativestateandwithintheir naturalenvironmentgeneratesvaluablestructural,func- tionalandorganizationalinformationaboutinfectionand disease. Observation of infectious agents inside their hosts bringsabout several challengesdue to thediffer- enceinlengthscalesbetweentheenvironmentalcontext, namely host cells and tissues, and the relatively small bacteria and viruses. Correlation of multiple imaging technologiesallowsforthisscalediscrepancytobeover- come. Often light microscopy(LM) is used to identify andlocalizetheobjectsofinterestinsidelargevolumes, and electron microscopy (EM) is used to image their
structural details. Although combinations of different LM and EM techniques can be used to image micro- organisms,commonstepsinanycorrelativeLMandEM (CLEM)workflowincludelabellingandidentificationof the microbes, localization and navigation to these microbeswithintheelectronmicroscope,andhigh-reso- lutionimaging.CLEM imagingprovidesvaluableinfor- mationabouttheinteractionsbetweenhostandpathogen in situ at a macromolecular level, which is often not obtainablebyothertechniques.
Identifyingorganismsand molecules of interestbyspecificlabelling
Fluorescent labelling of structures is important for CLEM imaging and can be achieved in several ways (Figure1).Non-geneticlabels,wherechemicallyconju- gatingfluorescentdyesto structuressuch aslipids,pro- teoglycans, or nucleotides, or to illuminate specific cell activitiessuchasmitochondrialmembranepotential[1], can target whole microbes, certain structures, or func- tionalsites insidemicrobes.Celleretal.haveusedlipid andpeptidoglycandyesandfluorescentlightmicroscopy (fLM) to demonstrate the formation of novel cellular compartmentalization assembliesas well as sitesof cell wall formation within Streptomyces coelicolor [2]. This chemicallabellingisstraightforwardandgeneratesrobust signal,but itsspecificity andapplicabilityislimited.
Geneticallyencodingafluorescentlabelthatisfusedto theproteinofinterestisthemostcommonlyusedtech- niquefortargetingspecificproteins,duetotheubiquity of optimized labels and the availability of extensive protocols[3].Geneticlabelsnoware easiertointroduce withcurrentgenomeeditingtechnologies[4],providing identityinformationthatcomplementstheultrastructural informationprovidedbyEM.Positiveidentificationand targetingofaparticularmicrobeinsitubyfLMhasbeen used with great success [5,6,7] and the reader is referredto excellent existing reviews and protocols for furtherinformation[8,9,10,11,12,13].Briefly,thegen- eral approach is to insert the gene for a fluorescent protein,oftengreenfluorescentprotein(GFP)oravari- ant,fusedtoaproteinofinterestthroughashort,flexible linker.Fluorescentimagingof theGFP-tagged protein, expressedbythemicrobe,isthenusedforidentification and targetingof suitable regionsfor morphological EM imaginginsidethehost.Straussetal. preciselylocalized HIV-1assemblyandbuddingfrominfectedcellsbylive- cellimaging,and used cryo-electron tomography(cryo- ET)toresolvethestructureofindividualvirusparticles andtheir membrane tethers[14].One should beaware that fluorescent protein tagging might interfere with
expression, localization or functionality of the labelled protein[15]andpossible incompatibilitieswithstaining methodsusedforEM[16].
Additionally,geneticallyencoding(small)chemicallabels ontoproteins,whicharelatercoupledtosyntheticfluor- ophores,hastheadvantagethattheseinterferelesswith protein functionality. They also tolerate better the embeddingandstainingprotocolsforhighpressurefreez- ing and freeze substitution to circumvent fluorescent quenching oftenobservedforGFP[17,18].
Geneticmanipulationtointroducethelabelislimitedto proteins.Toexpandbeyondproteinlabelling,alternative bioorthogonal approaches have been developed. The advantage of bioorthogonallabelling isthatnon-protein biomoleculessuchasnucleicacids,glycansandlipidscan be labelled [19]. Bioorthogonal labelling utilizes the insertion of small chemical moieties, which are inert (and therebyinvisible)tonormal biochemicalreactions, and theirsubsequentselectivereactionto incorporatea specifictag[20,21].Bioorthogonalfluorescenttagshave beenusedtoidentifyE.coli,intactorpartiallydegraded, in regionsusing generalmorphology thatis sufficiently distinct from the host [22]. Genetic code expansion to incorporate unnatural amino acids allows proteins that containtheunnaturalaminoacid(inpracticeallproteins
frommodifiedmicrobesareassumedtobelabelled)from a single organism to be identified within a mixture of species [23,24].The drawback is theextensive genetic manipulationthatisrequiredtoexpandthegeneticcode usedbythecell.Introductionoflabelsthatarecapableof diaminobenzidine (DAB) polymerization,which can be directlydetectedintheEM,canbeusedwhentargeting within the EM micrograph needs to be more accurate thantheLM-EMcorrelationaccuracy(seesectionInte- grating information to overlay and annotate volumes) [21].
Another approach to identifyorganisms and their sub- cellularcontentincorrelationwithEMimagingischem- icalisotopedetection,eitherbynano-secondaryionmass spectrometry(NanoSIMS)[25,26,27]orEnergy-disper- sive X-rayspectroscopy(EDX)[28].These techniques areusedforcorrelativeidentificationbutnottargetingof structures.Specificisotopesorelementscanbeimaged byNanoSIMSwithsensitivitiesintheparts-per-million range[27]andalateralresolutionofaround50nm[29].
All elements and isotypes can be detected, and it is especiallyusefulincaseswhereanisotopecanbecon- strained to a particular species by pre-culturing in stable isotopes. It allows for following the movement ofmineralswithinamicrobialcommunity[30],intracel- lulardrugtrafficking[31]orcanbeusedinapulse-chase
Figure1
Chemical labelling
Bioorthogonal labelling
Genetic fluorescent protein labelling
Genetic tag labelling
Protein
Fluorophore
Protein tag
GFP
DNA
PG
lipids
DNA probe–N N
probe N N
probe
Current Opinion in Microbiology
Fluorescentlabellingstrategies.FluorescentlabellingstrategiesformicroorganismconstituentsincludechemicallabellingforDNA,lipidsandPG, bioorthogonallabellingforPG,lipidsandproteins,geneticfluorescentproteinfusionandgenetictaglabellingforproteins.
experimenttomonitorturnover[25].Additionally,mul- tiplexing is relatively straightforward as multiple ele- mentsandisotopescanbereliablydetectedwithmini- malcomplicationstosamplepreparation,thusallowing multiple microbes to be identified within the same volume.
Selectionofanappropriatelabellingtechniquetoidentify aparticular microbewill dependultimately onthebio- logicalsystemandquestionathand.Fromthenumerous techniques,fLMwilllikelybethemostcommonchoice in thefuture.NanoSIMS and bioorthogonaltechniques maybeimportantinthefutureandoffermanyimportant opportunities for further research. Exceptions for non- proteinbiomoleculesandcaseswheregeneticfusionsare notpossiblemayrequireanotherapproachsuchasimmu- nolabeling[32]or isotopedistinctions.
Imaginglargevolumes efficiently
LMguidedandtargetedimagingofspecificstructuresis paramountforefficientimaging,sincecurrentEMimag- ing is too slow to make imaging the whole volume feasible. Additionally, the volumes that are routinely imaged with LM cannot be directly imaged with EM.
Electronscantravelarelativelysmalldistancethrougha sample unscattered (0.1mm) compared to light (1mm).Thus, the physical limit of samplethickness forimagingismuchthinnerforEMthanforLM[33].To overcomethislimit,volumesaresampledusingsections or byblockface imaging,where thinlayers are sequen- tiallyremovedfromatissueblockandthenewlyexposed surfacesare imaged.Thissampling allowsthecellmor- phology and the structures surrounding the targeted microbetobeimagedwithinavolumethatislargerthan thephysicallimitof EMimaging.
ForEMimaging,traditionally,thinserialsectionsarecut fromablock of resin-embeddedsampleand are placed onto a grid with a supporting carbon layer for TEM.
Automated sectioning systems allow for the collection andtrackingofhundredsofsections(75nm)onsilicon wafers [34], or indium tin oxide (ITO) coated glass, whichareimagedbySEM.Thistechniqueiscalledarray tomographyand volumesare obtainedbycomputation- allyjoiningthe2Dimageswithaspacingcorresponding tothesectionthickness.The useof automatedsystems and intelligently combining low-resolution scans with targeted high resolution imaging only where necessary speedsupimagingmillimetresizedsamples[34,35].The advantageofserialsectioningforCLEMimagingisthat thesectionscanbeusedtotargetspecificregionsinthe blockifthesamplepreparationpreservesthefluorescent signalwithinthefinalsections, aswasdoneforvaccinia infectedcells [6],orbyonsectionlabelling[22].
SEMimagesthesurfaceofatissue block,whichcanbe removedby slicing witha diamondknife orion-beam
[36]. By iterating the imaging and the removal of the surface,largevolumescanbesampled.Thetimeneeded toprepareandimagethemanysurfacesforcescompro- misesbetweenthetotalvolumethat isimagedandthe finalresolution[37].Thenewsurfacecanbeexposed bycuttingwithadiamondknife,aswasdonetoinvesti- gatethe infectionofzebrafishwithMycobacterium mar- inum [38], or by using a focussed ion beam (FIB) to ablate the surface [39].Diamond knives are generally fastertoexposethenewsurface,whileremovingmore material with each slice, around 50nm. This allows largervolumestobeimaged,butwithloweraxialreso- lution.TheminimumthicknessforFIBmillingisusu- allydependentonthedepthofthebeamdamagecaused whileimaging.Timegainscanbemadebyusingmulti beamapproaches[40],particularlyforsamplesthathave largesurfaceareascomparedtotheirdepths,butitisless suitable for volumes that require many slices where exposing the new surfaces takes as long as imaging.
Blockface methods also do not allow for revisiting of regions of interest as the sections are lost after each removal.
Methodstoimagethickersamples,therebyreducingthe need for sectioning, utilize soft X-rays (high energy photons)withhigherpenetrationdepthsforimagingthan electrons.SoftX-rayswereusedtoimagewholeeukary- oticcellswhileresolving theirultrastructuralfeaturesat 50nm resolution in a correlative soft X-ray/fLM setup [41].ComparedtoserialEMimaging,samplehandlingis simplifiedinthistechniqueanditrequiresmuchlessdata acquisitiontime.However,theinfrastructureandinstru- mentationrequirementsaresignificant.
Micro-CT also allows for thick samples (a few milli- metres) to beimagedwith hard X-rays. In cases where X-ray dose is not a consideration, such as with metal stainedsamplesthataresimilarlyprocessedasforEM,it ispossible to reach sub-micrometrepixelsizesin these muchlarger volumes.This allows for more constrained targetingofthehigh-resolutionEMimagingtoasmaller volume[42].However,thecurrentresolutionlimitations preventimagingfeaturessmallerthan1mm,limitingits presentapplicabilityfor microbes.
ThetechniquesforimaginglargevolumesbyEMallows forthevisualizationofcompletecellularandtissuemor- phology.However,CLEMtargetedimagingaroundsites of microbe infection are highly desired for efficiency.
Futureimprovementswillmostlycomefrombetterauto- mationandintegrationofLMandEMtechniquesalong withimprovedspecimenpreparation,ratherthanfunda- mentalchangestotheintegratingworkflowsorchangesin thephysicalprinciplesutilized.Theseimprovementswill increasetheamountofdatacollecteddramatically,raising theadditionalchallengesofintegratingandanalysingthe largeamountsof data.
Integratinginformation tooverlay and annotate volumes
Combining the datasetsgenerated bymultiple imaging modalities for CLEM purposes is crucial. While it is conceptually simple to overlay two different datasets, accuracyandthetimeinvestmentneededareimportant issues whenworking with differentresolution datasets.
The resolution of conventionalfLM is a few orders of magnitudelowerthan EM,limitingitscorrelationaccu- racytotheEMdata,whichitgenerallyaround50–100nm using bead based approaches [12,43,44]. The use of photoswitchable proteinsthat allowfor super-resolution LM [6,45,46] in combination with fiducialmarkersto precisely overlay the images [12] can improve the correlationto10–20nm[46,47].Inidealcases,integrated microscopescanimprovetheaccuracytobetterthan5nm [48]. Thus, depending on the size of the microbe of interestandofthetargetedregion,aswellashowcrowded the environment is and theprecision needed, different fluorescent tags should be considered for correlative studiestoensurethattheoverlayissufficientlyaccurate.
Combining different LM and EM datasets has dual purposes. The first is wherefLM is usedto guide EM datacollection.Insomecases thespecimenshapeitself canbeusedforcorrelation,providingtheshapeisobvious andthetargetcanbeobservedinbothLMandEM[2].In many cases, it is necessary to locate fiducials (or other landmarks) that are visible in both modalities, such as fluorescentbeads. Thesecanserveas landmarkstocal- culatetherequiredcoordinatetransformationsandlocate the exact region of interest in theelectron microscope from the LM image with measurable accuracy [44].
Alternatively, thecommonlyusedEM stainuranylace- tate canbemade fluorescentunder cryoconditionsand can be directly correlated with the EM image [49].
Having theLM images available withinthe EM refer- ence frame is particularly useful when targeting rare events or whensetting up long EM acquisitionruns to ensure the appropriate regions are imaged. Specialized EMgridgeometry(e.g. findergrids)andpatterned sub- strates [7] allow for easy coordinate transformations [7,50]. Targetingis also criticalfor block faceor other destructive methods where it would notbe possible to revisit aregion. Commercial systemshave been devel- oped that use markers on the sample carrier or user- identified pointstoperformthegeometricaltransforma- tions[12,51].
ThesecondneedforcombiningLMandEMisforpost- acquisition visualization. For simple overlay of two- dimensional images, nonspecialized software may be sufficient. However, to combine datasets of different types,modalitiesdimensionsandsizes,specificimaging packagessuchasAmira(ThermoFisherScientific,Wal- tham, MA), Icy [52] with the eC-CLEM plugin [53], OMERO [54], ImageJ[55] with pluginTrakEM2 [56],
andothersbundledwiththeFijidistribution[57,58]are necessary.Especiallythehandlingofverylargeindivid- ualdatasets(e.g.asampleof50mm3recordedusing1nm3 voxel sizes with 8-bitvoxel values wouldrequire 113 terabytes of storage) underscores the need for smart algorithms and significant computing resources for CLEM imaging.
Analysis and interpretation of the combined datasets requiresthatfeaturesofinterestareaccuratelyidentified, and is a critical part of fully utilizing the information.
Segmentationandannotationoffeaturesremainsagreat challenge and using generalized algorithms is difficult, primarilybecausestructuralfeaturesdependonparticular samplepreparationandimagingconditions.Thisremains alabour-intensive,manualprocess.Automatedmethods are under development, and some advanced automatic segmentationmethodsareavailablefortracingandiden- tifying cells in neuronal tissue [59,60]. Automatically identifyingcellsinothertissues,orsub-cellularstructures is lagging. There has been some exciting progress in neural networks [61] that can be trained to identify particular features of interest, and should be easily applied to well defined structures, for example some viruses.Thelargestructuralvariationbetweenandwithin organelles and otherobjects (i.e.cytoskeletal elements, ribosomes,gapjunctions,andsoon)thatcouldsurround any givenmicrobe, makethis anextremelychallenging problem.
Conclusion
Effective correlative imaging, that is, using imaging modalities that span large size and resolution ranges, enables identification, localization and visualization of microbeswithintheultrastructuralcontextoftheirnatu- ral host environment. Localizing viruses or bacteria within their larger environment requires specific label- ling,samplepreservationthatservesbothimagingmodal- ities,andthinning(orrepeatedsurfaceremoval)forEM.
Automatedserialsectionsanddataacquisitionnowallow for larger volumes to be imaged and more easily inte- gratedwithLMdata,andisexpectedtoprovideawealth ofinformationtothemicrobiologyfield.Improvementsto LM that may be applied to microbial research in the immediatefutureincludetheuseoftwoobjectivelenses to increase the amount of light collected (and thereby improve the resolution) [62], and the increased use of singlemoleculelocalizationtechniques.Longterm,tech- niques such as microCT and soft X-rays may become feasibletolookatlargervolumesathighresolution,but currently onlyEM canprovide informationin the1nm resolutionregime.
For the future, CLEM imaging of microbes should be improved by taking LM tagging of microbes a step further.Specifically,newstrategiesfortaggingmicrobes while they are in their native environment need to be
developed.Incurrentstudies,infectedcellculturesand exvivomicrobe–hostsystemsaregenerallyused.Tagging ofmicrobes thatarealready insidetheirhost,that is,in vivoorinsitu,foridentificationandtargetedimaging,isa great challenge, but it is extremely important to get a realistic and unperturbed view of naturally occurring infections.Furthermore,largescalecorrelativeEMimag- ing that pushes towards the goal of millimetre scales should be taken to the next level in terms of speed, integrationofmodalities,reconstructionanddatamining.
Currently LM imaging, SEM/TEM imaging, volume reconstruction,dataminingandvisualizationareseparate processeswhich are sequentially knittedtogether. The particularratelimitingstepvariesbyapplication,butitis oftenin thereconstruction,dataminingor visualization steps, rather than in theactual imaging,a trendthat is likelyto increaseasimagingbecomesincreasinglyauto- mated.Whilelessimagingtimeisalwaysdesirable,inthe futureefficientlymakingsenseofallthedatawillrequire more effort thanactually obtaining theraw data. All of thesesteps shouldbe approachedholistically, andinte- gratedintoasingleworkingsystem,inordertodrastically increase overall speed and usability. Taken together these developments should enable efficient, functional andstructuralidentificationofinfections invivo.
Acknowledgements
ThisworkwassupportedinpartbyGenmab(Utrecht,theNetherlands)and theEuropeanUnionthroughtheHorizon2020ProgrammegrantiNEXT (grantagreementno.653706).
References andrecommendedreading
Papersofparticularinterest,publishedwithintheperiodofreview, havebeenhighlightedas:
ofspecialinterest
ofoutstandinginterest
1. GacM,BigdaJ,VahlenkampTW:Increasedmitochondrial superoxidedismutaseexpressionandloweredproductionof reactiveoxygenspeciesduringrotavirusinfection.Virology 2010,404:293-303.
2. CellerK,KoningRI,WillemseJ,KosterAJ,vanWezelGP:Cross- membranesorchestratecompartmentalizationand
morphogenesisinStreptomyces.NatCommun2016,7 ncomms11836.
3. TsienRY:Thegreenfluorescentprotein.AnnuRevBiochem 1998,67:509-544.
4. KomorAC,BadranAH,LiuDR:CRISPR-basedtechnologiesfor themanipulationofeukaryoticgenomes.Cell2016http://dx.
doi.org/10.1016/j.cell.2016.10.044.
5. GurskayaNG,VerkhushaVV,ShcheglovAS,StaroverovDB, ChepurnykhTV,FradkovAF,LukyanovS,LukyanovKA:
Engineeringofamonomericgreen-to-redphotoactivatable fluorescentproteininducedbybluelight.NatBiotechnol2006, 24:461-465.
6.
PeddieCJ,DomartM-C,SnetkovX,O’TooleP,LarijaniB,WayM, CoxS,CollinsonLM:Correlativesuper-resolutionfluorescence andelectronmicroscopyusingconventionalfluorescent proteinsinvacuo.JStructBiol2017,199:120-131.
Demonstrationofsuper-resolutionLM inresin-embeddedsectionsof vacciniainfectedHeLacells.UnstainedSEMsectionswereimagedusing backscatteredsignalatvariablepressurestofacilitatecollectionwithout adjusting the sample between modalities, allowing for very precise correlations.
7. RussellMRG,LernerTR,BurdenJJ,NkweDO,Pelchen- MatthewsA,DomartM-C,DurganJ,WestonA,JonesML, PeddieCJetal.:3Dcorrelativelightandelectronmicroscopyof culturedcellsusingserialblockfacescanningelectron microscopy.JCellSci2017,130:278-291.
DescribestheuseofpatternedsubstratesandserialblockfaceCLEMof MycobacteriumtuberculosisandHIV-1.
8. HauserM,WojcikM,KimD,MahmoudiM,LiW,XuK:Correlative super-resolutionmicroscopy:newdimensionsandnew opportunities.ChemRev2017,117:7428-7456.
9. KopekBG,Paez-SegalaMG,ShtengelG,SochackiKA,SunMG, WangY,XuCS,vanEngelenburgSB,TaraskaJW,LoogerLLetal.:
Diverseprotocolsforcorrelativesuper-resolution
fluorescenceimagingandelectronmicroscopyofchemically fixedsamples.NatProtoc2017,12:916-946.
10. HamptonCM,StraussJD,KeZ,DillardRS,HammondsJE, AlonasE,DesaiTM,MarinM,StormsRE,LeonFetal.:Correlated fluorescencemicroscopyandcryo-electrontomographyof virus-infectedortransfectedmammaliancells.NatProtoc 2017,12:150-167.
Aprotocolsarticlethatdescribesstate-of-the-artmethodsusedinCLEM workflows.
11. BykovYS,CorteseM,BriggsJAG,BartenschlagerR:Correlative lightandelectronmicroscopymethodsforthestudyofvirus–
cellinteractions.FEBSLett2016,590:1877-1895.
12. SchorbM,GaechterL,AvinoamO,SieckmannF,ClarkeM, BebeacuaC,BykovYS,SonnenAFP,LihlR,BriggsJAG:New hardwareandworkflowsforsemi-automatedcorrelativecryo- fluorescenceandcryo-electronmicroscopy/tomography.J StructBiol2017,197:83-93.
Technicalpaperdescribingequipment andmethodsforrobustCLEM experimentswithimprovedreliabilityandthroughput.
13. OuHD,DeerinckTJ,BushongE,EllismanMH,O’SheaCC:
Visualizingviralproteinstructuresincellsusinggenetic probesforcorrelatedlightandelectronmicroscopy.Methods 2015,90:39-48.
14. StraussJD,HammondsJE,YiH,DingL,SpearmanP,WrightER:
Three-dimensionalstructuralcharacterizationofHIV-1 tetheredtohumancells.JVirol2015,90:1507-1521.
15. SwuliusMT,JensenGJ:ThehelicalMreBcytoskeletonin EscherichiacoliMC1000/pLE7isanartifactoftheN-terminal yellowfluorescentproteintag.JBacteriol2012,194:6382-6386.
16. BramaE,PeddieCJ,JonesML,DomartM-C,SnetkovX,WayM, LarijaniB,CollinsonLM:Standardfluorescentproteinsasdual- modalityprobesforcorrelativeexperimentsinanintegrated lightandelectronmicroscope.JChemBiol2015,8:179-188.
17. PerkovicM,KunzM,EndesfelderU,BunseS,WiggeC,YuZ, HodirnauVV,SchefferMP,SeybertA,MalkuschSetal.:
Correlativelight-andelectronmicroscopywithchemicaltags.
JStructBiol2014,186:205-213.
18. LissV,BarlagB,NietschkeM,HenselM:Self-labellingenzymes asuniversaltagsforfluorescencemicroscopy,super- resolutionmicroscopyandelectronmicroscopy.SciRep2016, 5:17740.
19. HangHC,WilsonJP,CharronG:Bioorthogonalchemical reportersforanalyzingproteinlipidationandlipidtrafficking.
AccChemRes2011,44:699-708.
20. HangHC,YuC,KatoDL,BertozziCR:Ametaboliclabeling approachtowardproteomicanalysisofmucin-typeO-linked glycosylation.ProcNatlAcadSci2003,100:14846-14851.
21.
NgoJT,AdamsSR,DeerinckTJ,BoassaD,Rodriguez-RiveraF, PalidaSF,BertozziCR,EllismanMH,TsienRY:Click-EMfor imagingmetabolicallytaggednonproteinbiomolecules.Nat ChemBiol2016,12:459-465.
Describestheuseofinnovativebio-orthogonalapproachestolabelDNA, RNA,phospholipids,andpeptidoglycansynthesisinavarietyofsystems.
22. vanElslandDM,BosE,deBoerW,OverkleeftHS,KosterAJ,van KasterenSI:Detectionofbioorthogonalgroupsbycorrelative lightandelectronmicroscopyallowsimagingofdegraded bacteriainphagocytes.ChemSci2016,7:752-758.
23. UttamapinantC,HoweJD,LangK,Bera´nekV,DavisL,MaheshM, BarryNP,ChinJW:Geneticcodeexpansionenableslive-cell andsuper-resolutionimagingofsite-specificallylabeled cellularproteins.JAmChemSoc2015,137:4602-4605.
24. ChinJW:Expandingandreprogrammingthegeneticcodeof cellsandanimals.AnnuRevBiochem2014,83:379-408.
25. TakadoY,KnottG,HumbelBM,MasoodiM,EscrigS,MeibomA, CommentA:Imagingthetime-integratedcerebralmetabolic activitywithsubcellularresolutionthroughnanometer-scale detectionofbiosyntheticproductsderivingfrom13C-glucose.
JChemNeuroanat2015,69:7-12.
26. KoppC,Domart-CoulonI,EscrigS,HumbelBM,HignetteM, MeibomA:Subcellularinvestigationofphotosynthesis-driven carbonassimilationinthesymbioticreefcoralPocillopora damicornis.MBio2015,6:e02299-e2314.
27. WirtzT,PhilippP,AudinotJ-N,DowsettD,EswaraS:High- resolutionhigh-sensitivityelementalimagingbysecondary ionmassspectrometry:fromtraditional2Dand3Dimagingto correlativemicroscopy.Nanotechnology2015,26:434001.
Explains thephysicalprinciples behindNanoSIMS anddemonstrates someearly,excitingapplicationsofthisrelativelynascenttechnique.
28. PenenF,MalherbeJ,IsaureM-P,DobritzschD,BertalanI, GontierE,LeCoustumerP,Schaumlo¨ffelD:Chemical bioimagingforthesubcellularlocalizationoftraceelements byhighcontrastTEM,TEM/X-EDS,andNanoSIMS.JTrace ElemMedBiol2016,37:62-68.
29. SakaSK,VogtsA,Kro¨hnertK,HillionF,RizzoliSO,WesselsJT:
Correlatedopticalandisotopicnanoscopy.NatCommun2014, 5:3664.
30. SheikAR,BrussaardCPD,LavikG,LamP,MusatN,KrupkeA, LittmannS,StrousM,KuypersMMM:Responsesofthecoastal bacterialcommunitytoviralinfectionofthealgaePhaeocystis globosa.ISMEJ2014,8:212-225.
31. JiangH,PassarelliMK,MunroPMG,KilburnMR,WestAA, DolleryCT,GilmoreIS,RakowskaPD:High-resolutionsub- cellularimagingbycorrelativeNanoSIMSandelectron microscopyofamiodaroneinternalisationbylung macrophagesastheevidencefordrug-induced phospholipidosis.ChemCommun2016,53:1506-1509.
32. BosE,HussaartsL,vanWeeringJRT,EllismanMH,deWitH, KosterAJ:VitrificationofTokuyasu-styleimmuno-labelled sectionsforcorrelativecryolightmicroscopyandcryo electrontomography.JStructBiol2014,186:273-282.
33. HelmchenF,DenkW:Deeptissuetwo-photonmicroscopy.Nat Methods2005,2:932-940.
34.
HayworthKJ,MorganJL,SchalekR,BergerDR,HildebrandDGC, LichtmanJW:ImagingATUMultrathinsectionlibrarieswith WaferMapper:amulti-scaleapproachtoEMreconstructionof neuralcircuits.FrontNeuralCircuits2014,8:68.
Describesautomatedsoftwaretomakeimagingofthousandsofsections feasibleandefficienttherebyimprovingandexpandingtheuseofATUM.
35. HildebrandDGC,CicconetM,TorresRM,ChoiW,QuanTM, MoonJ,WetzelAW,ScottChampionA,GrahamBJ,RandlettO etal.:Whole-brainserial-sectionelectronmicroscopyinlarval zebrafish.Nature2017,545:345-349.
36. TitzeB,GenoudC:Volumescanningelectronmicroscopyfor imagingbiologicalultrastructure.BiolCell2016,108:307-323.
37.
XuCS,HayworthKJ,LuZ,GrobP,HassanAM,Garcı´a-Cerda´nJG, NiyogiKK,NogalesE,WeinbergRJ,HessHF:EnhancedFIB- SEMsystemsforlarge-volume3Dimaging.Elife2017,6:1-36.
Demonstration of closed-loop FIB-milling for reliable sectioning and imagingofunsupervisedsystems,criticalforincreasingthevolumesize neededformanybiologicalquestions.
38. HosseiniR,LamersGEM,SoltaniHM,MeijerAH,SpainkHP, SchaafMJM:Efferocytosisandextrusionofleukocytes determinetheprogressionofearlymycobacterial pathogenesis.JCellSci2016,129:3385-3395.
39. VillingerC,NeusserG,KranzC,WaltherP,MertensT:3Danalysis ofHCMVinduced-nuclearmembranestructuresbyFIB/SEM
tomography:insightintoanunprecedentedmembrane morphology.Viruses2015,7:5686-5704.
40. EberleAL,MikulaS,SchalekR,LichtmanJ,KnotheTateML, ZeidlerD:High-resolution,high-throughputimagingwitha multibeamscanningelectronmicroscope.JMicrosc2015, 259:114-120.
41. ElgassKD,SmithEA,LeGrosMA,LarabellCA,RyanMT:Analysis ofER-mitochondriacontactsusingcorrelativefluorescence microscopyandsoftX-raytomographyofmammaliancells.
JCellSci2015,128:2795-2804.
42. KarremanMA,RuthensteinerB,MercierL,SchieberNL,SoleckiG, WinklerF,GoetzJG,SchwabY:FindyourwaywithX-ray.
MethodsCellBiol2017,140:277-301.
43. KukulskiW,SchorbM,WelschS,PiccoA,KaksonenM, BriggsJAG:Precise,correlatedfluorescencemicroscopyand electrontomographyoflowicrylsectionsusingfluorescent fiducialmarkers.MethodsCellBiol2012,111:235-257.
44. SchorbM,BriggsJAG:Correlatedcryo-fluorescenceandcryo- electronmicroscopywithhighspatialprecisionandimproved sensitivity.Ultramicroscopy2014,143:24-32.
45. LiuB,XueY,ZhaoW,ChenY,FanC,GuL,ZhangY,ZhangX, SunL,HuangXetal.:Three-dimensionalsuper-resolution proteinlocalizationcorrelatedwithvitrifiedcellularcontext.
SciRep2015,5:13017.
46. ChangY,ChenS,TochevaEI,Treuner-langeA,Lo¨bachS, Søgaard-AndersenL,JensenGJ:Correlatedcryogenic photoactivatedlocalizationmicroscopyandcryo-electron tomography.NatMethods2014,11:737-739.
47. SochackiKa,ShtengelG,vanEngelenburgSB,HessHF, TaraskaJW:Correlativesuper-resolutionfluorescenceand metal-replicatransmissionelectronmicroscopy.NatMethods 2014,11:305-308.
48. HaringMT,LivN,ZonnevylleAC,NarvaezAC,VoortmanLM, KruitP,HoogenboomJP:Automatedsub-5nmimage registrationinintegratedcorrelativefluorescenceand electronmicroscopyusingcathodoluminescencepointers.
SciRep2017,7:43621.
49. TuijtelMW,MulderAA,PosthumaCC,vanderHoevenB, KosterAJ,Ba´rcenaM,FaasFGA,SharpTH:Inducing fluorescenceofuranylacetateasadual-purposecontrast agentforcorrelativelight-electronmicroscopywith nanometreprecision.SciRep2017,7:10442.
50. BeckwithMS,BeckwithKS,SikorskiP,SkogakerNT,FloTH, HalaasØ:SeeingaMycobacterium-infectedcellinnanoscale 3D:correlativeimagingbylightmicroscopyandFIB/SEM tomography.PLoSOne2015,10:e0134644.
51. WackerI,ChockleyP,BartelsC,SpomerW,HofmannA, GengenbachU,SinghS,ThalerM,GrabherC,Schro¨derRR:Array tomography:characterizingFAC-sortedpopulationsof zebrafishimmunecellsbytheir3Dultrastructure.JMicrosc 2015,259:105-113.
52. deChaumontF,DallongevilleS,ChenouardN,Herve´ N,PopS, ProvoostT,Meas-YedidV,PankajakshanP,LecomteT,Le MontagnerYetal.:Icy:anopenbioimageinformaticsplatformfor extendedreproducibleresearch.NatMethods2012,9:690-696.
53. Paul-GilloteauxP,HeiligensteinX,BelleM,DomartM-C,LarijaniB, CollinsonL,RaposoG,SalameroJ:eC-CLEM:flexible multidimensionalregistrationsoftwareforcorrelative microscopies.NatMethods2017,14:102-103.
Improvedsoftwaretofacilitatethecorrelationofdatasets.
54. AllanC,BurelJ-M,MooreJ,BlackburnC,LinkertM,LoyntonS, MacDonaldD,MooreWJ,NevesC,PattersonAetal.:OME RemoteObjects(OMERO):aflexible,model-drivendata managementsystemforexperimentalbiology.NatMethods 2012,9:245-253.
55. SchneiderCA,RasbandWS,EliceiriKW:NIHImagetoImageJ:
25yearsofimageanalysis.NatMethods2012,9:671-675.
56. CardonaA,SaalfeldS,SchindelinJ,Arganda-CarrerasI, PreibischS,LongairM,TomancakP,HartensteinV,DouglasRJ:
TrakEM2softwareforneuralcircuitreconstruction.PLoSOne 2012,7.
57. KeeneDR,TufaSF,WongMH,SmithNR,SakaiLY,HortonWA:
CorrelationofthesamefieldsimagedintheTEM,Confocal, LM,andMicroCTbyimageregistration.MethodsCellBiol2014, 124:391-417.
58. SchindelinJ,Arganda-CarrerasI,FriseE,KaynigV,LongairM, PietzschT,PreibischS,RuedenC,SaalfeldS,SchmidBetal.:Fiji:
anopen-sourceplatformforbiological-imageanalysis.Nat Methods2012,9:676-682.
59. BoergensKM,BerningM,BocklischT,Bra¨unleinD,DrawitschF, FrohnhofenJ,HeroldT,OttoP,RzepkaN,WerkmeisterTetal.:
webKnossos:efficientonline3Ddataannotationfor connectomics.NatMethods2017,14:691-694.
60. StafflerB,BerningM,BoergensKM,GourA,vanderSmagtP, HelmstaedterM:SynEM,automatedsynapsedetectionfor connectomics.Elife2017,6.
61. ChenM,DaiW,SunSY,JonaschD,HeCY,SchmidMF,ChiuW, LudtkeSJ:Convolutionalneuralnetworksforautomated annotationofcellularcryo-electrontomograms.NatMethods 2017,14:983-985.
Applicationofneuralnetworkstoidentifyingfeatureswithintomograms, important for reducing labour time and improving reproducibility in annotations.
62. HuangF,SirinakisG,AllgeyerES,SchroederLK,DuimWC, KromannEB,PhanT,Rivera-MolinaFE,MyersJR,IrnovIetal.:
Ultra-highresolution3Dimagingofwholecells.Cell2016, 166:1028-1040.
