ContentslistsavailableatScienceDirect
Annals
of
Anatomy
j o ur na l h o me p a g e :w w w . e l s e v i e r . c o m / l o c a t e / a a n a t
INVITED
REVIEW
Advances
in
cryo-electron
tomography
for
biology
and
medicine
Roman
I.
Koning
a,b,∗,
Abraham
J.
Koster
a,b,
Thomas
H.
Sharp
a,∗aDepartmentofCellandChemicalBiology,LeidenUniversityMedicalCenter,2300RCLeiden,TheNetherlands bNeCEN,InstituteBiologyLeiden,LeidenUniversity,2300RALeiden,TheNetherlands
a
r
t
i
c
l
e
i
n
f
o
Articlehistory:Received4September2017
Receivedinrevisedform2February2018 Accepted5February2018 Keywords: Electronmicroscopy Cryo-electrontomography Energyfilter Phaseplate
Directelectrondetector
a
b
s
t
r
a
c
t
Cryo-electrontomography(CET)utilizesacombinationofspecimencryo-fixationandmulti-angle elec-tronmicroscopyimagingtoproducethree-dimensional(3D)volumereconstructionsofnative-state macromolecularandsubcellularbiologicalstructureswithnanometer-scaleresolution.Inrecentyears, cryo-electronmicroscopy(cryoEM)hasexperiencedadramaticincreaseintheattainableresolutionof 3Dreconstructions,resultingfromtechnicalimprovementsofelectronmicroscopes,improved detec-torsensitivity,theimplementationofphaseplates,automateddataacquisitionschemes,andimproved imagereconstructionsoftwareandhardware.Thesedevelopmentsalsogreatlyincreasedtheusability andapplicabilityofCETasadiagnosticandresearchtool,whichisnowenablingstructuralbiologiststo determinethestructureofproteinsintheirnativecellularenvironmenttosub-nanometerresolution. Theserecenttechnicaldevelopmentshavestimulatedustoupdateonourpreviousreview(Koning,R.I., Koster,A.J.,2009.Cryo-electrontomographyinbiologyandmedicine.AnnAnat191,427–445)inwhich wedescribedthefundamentalsofCET.Inthisfollow-up,weextendthisbasicdescriptioninorderto explaintheaforementionedrecentadvances,anddescriberelated3Dtechniquesthatcanbeappliedto theanatomyofbiologicalsystemsthatarerelevantformedicine.
©2018ElsevierGmbH.Allrightsreserved.
Contents
1. Introduction...83
2. Samplepreparation...83
2.1. Vitrificationofthinsamples...83
2.2. Highpressurefreezingforthicksamplesandwholecells ... 85
2.3. Cryo-sectioningthicksamples...85
2.4. Stabilityofvitreoussamplesintheelectronmicroscope ... 86
3. Datacollection...86
3.1. Tiltseriesacquisition...86
3.2. Phaseplatesimprovetomograminterpretability...86
3.3. Energy-filteringreducesnoise...88
3.4. Directelectrondetectors...89
3.5. Automateddatacollectionandstorage...89
4. Imageprocessing...90
4.1. Tomogramreconstructionautomation...90
4.2. Themissingwedgelimitsinterpretationoftomographicvolumes...91
Abbreviations:3D,three-dimensional;cryoEM,cryo-electronmicroscopy;CCD,charged-coupleddevice;CEMOVIS,cryo-electronmicroscopyofvitreoussections;CET, cryo-electrontomography;CLEM,correlativelightelectronmicroscopy;CMOS,complementarymetaloxidesemiconductor;CT,computedtomography;CTF,contrasttransfer function;DED,directelectrondetector;DQE,detectorquantumefficiency;FIB,focusedionbeam;fLM,fluorescencelightmicroscopy;GFP,greenfluorescentprotein;HPF, high-pressurefreezing;MRI,magneticresonanceimaging;NMR,nuclearmagneticresonance;PET,positronemissiontomography;PSF,point-spreadfunction;SBF,solidblock face;SEM,scanningelectronmicroscopy;SIRT,simultaneousiterativereconstructiontechnique;SPA,singleparticleanalysis;SXT,softX-raytomography;TEM,transmission electronmicroscope;VPP,voltaphaseplate;WBP,weightedback-projection.
∗ Correspondingauthorsat:DepartmentofCellandChemicalBiology,LeidenUniversityMedicalCenter,2300RCLeiden,TheNetherlands. E-mailaddresses:R.I.Koning@lumc.nl(R.I.Koning),T.Sharp@lumc.nl(T.H.Sharp).
Anatomyinvolvesimagingatdifferentlengthscales,ranging fromcompleteorganisms,metersinsize,downtosinglecellswith dimensionsontheorderofmicrometers,andevennanometer-scale sub-cellularstructures.Consequently,manydifferenttechniques arerequiredtovisualizeanatomicalprocessesoverthiswiderange (Fig.1).Tomographygeneratesthree-dimensionalreconstructions andisusedindiverseareasofscienceandmedicine,formingthe basisofcomputedtomography(CT),magneticresonanceimaging (MRI),andpositronemissiontomography(PET).WhilstCT,MRI, andPETcanimagewholeorganismsandachievesub-millimeter resolution,cryo-electrontomography(CET,alsoknownascryo-ET orelectroncryo-tomography;ECT)canimagethemicrometerto sub-nanometerlengthscales(Fig.1).Infact,CETistypicallyusedto visualizeindividualorganellesandbiomoleculeswithincells, yield-inginsightsintothemolecularandcellularpathologyofdiseases (Bauerleinetal.,2017)andpathogens(Wanetal.,2017).
ThemethodologyofCETisfundamentallysimilartoother tomo-graphictechniques; a volumeof material (molecule,virus, cell, organism,etc.)isimaged frommultipleangles(different orien-tations),andisthenreconstructedcomputationallytoforma3D volume(Fig.2).WhereasinMRI,CTandPETtheimagingsystem isrotatedfullyaroundthespecimen(i.e.thepatient),withCET thespecimenisrotateduptoalimitedangularrangewithinthe electronmicroscope(Fig.2A).Theresultingseriesoftilt-images (Fig. 2B) iscomputationally back-projected toreconstructa 3D volume(Fig.2C).
Inalltomographicimagingtechniques,themaximumallowable samplesizeandtheattainableresolutionareimportantpractical parameterstoconsider,asthesedeterminethelimitsofwhatcanbe observed.InCET,thefieldofviewistypicallyintheorderofmicrons. Thesamplethicknessshouldbelessthan500nminordertoallow electronstotransmitthroughthesamplewithlimited specimen-damaging effects due to the coulomb-force interactions of the imagingelectronbeamwiththeatomsofthesample.Therefore,in practice,wholecellsandtissuesmustbeslicedintosufficientlythin sectionsbeforeimaging, whilesmall biologicalspecimens, such aspurifiedproteins,virusesandbacteria,canbeimagedwithout sectioning.
TheresolutionofCETisdeterminedbyacombinationofthe electronradiationdosethatcanbeusedtoimagethesample with-outdetectabledamage,detectorsensitivity,samplethickness,and thequalityoftheelectromagneticlenses.Firstly,aselectron radi-ationstrongly interactswithatoms(Henderson,1995), andcan thereforebehighlydamagingforcryo-fixedspecimens,the imag-ingdosemustbekeptaslowaspossible.Moreover,inCETthis allowableelectrondosehastobedistributedovermanyangular viewsfrom which the3Dtomogram isreconstructed, resulting inindividualimageswithalowsignal-to-noiseratio.Therefore, detectorsthatefficientlydetectelectronswithlowintrinsicnoise levels, are paramountto obtain thehighest possible single-to-noiseratioandresolutioninthetomogram.Withincreasingsample thickness,electronsaremorelikelytoundergomultiplescattering
thatareusedforimaging areimperfect,which resultsinimage aberrationsandlossofresolution.Theselensimperfectionsare,to someextent,wellknownandcharacterizedfortransmission elec-tronmicroscopy,andcanthereforebecomputationallycorrected toachievethemaximumpossibleresolution(Haideretal.,2009).
Recentadvances incryo-electron microscopy (cryoEM)have givenrisetoa“resolutionrevolution”(Kuhlbrandt,2014),which hasincreasedthenumberofnear-atomicresolutionstructuresof proteinsandproteinassembliesthatemergedusingsingle parti-clecryoEM(Fernandez-LeiroandScheres,2016).Theadvancesthat enabledthisincludetechnicalimprovementsoftheelectron micro-scope,improveddetectorsensitivity,theimplementationofphase plates,automateddataacquisitionschemes,andimprovedimage reconstructionsoftwareandhardware,which,takenalltogether greatlyincreasetheusabilityandpracticalapplicabilityofCETasa diagnosticandresearchtool.
Thisreviewfollowsuponourearlierpublication(Koningand Koster,2009)andfocusesondescribingthetechnicaladvancesthat haveevolvedandemergedsince.Afterrecapitulatingthegeneral detailsofcryoEM,therecentimprovementsonCETaredescribed anddiscussedintheframeworkofthepracticalcryoETworkflow.
2. Samplepreparation
ForimagingsamplesusingCET,thefirststepistocryogenically fixthespecimenbyamethodreferredtoasvitrification. Depend-ingonsamplethicknesstherearetwovitrificationmethods(Fig.3). Samplesthatarethinnerthan10m,suchasisolatedprotein com-plexes,viruses,bacteriaorthincells,maybevitrifieddirectlyby plunge freezingintoacryogen(Dubochetetal.,1988).Samples uptoathicknessof200m,suchascellpelletsortissuebiopsies, requirefreezingathighpressure(so-calledhigh-pressurefreezing; HPF)for vitrification(Vanheckeetal.,2008).Afterthe vitrifica-tionstep,samplesthickerthan500nmrequirethinningpriorto imaging.Thinningcanbeperformedbycryo-ultramicrotomy( Al-Amoudietal.,2004b),usingadiamondknifetoslicethevitreous iceintosections,orafocusedionbeam(FIB)ofaheavymetal(e.g. gallium)toablatethematerialtoaspecifiedthickness.
2.1. Vitrificationofthinsamples
Fig.1.Scalesandresolutionofmedicalimagingtechniques.
Specimensofdifferentlengthscalesrequirediverseimagingtechniqueswithparticularlimitsinimagingdepthsandresolutions.
Fig.2.Cryo-electrontomographymethodology.
(A)Inthetransmissionelectronmicroscope,aseriesofimagesisacquiredonadetectorfromasampleofinterest,asthesampleisrotatedtospecifiedangles.Thesampleis embeddedinvitreousiceandcanbeverydiverse,e.g.wholecells,cellularextracts,proteinsorotherbiomolecules.(B)Theresultingprojectionimagesofthespecimenfrom differentangularviewsyieldsaso-calledtilt-series.(C)Thistilt-seriesiscomputationallyback-projectedtoforma3Dvolume,thetomogram.
Themoststraightforwardwaytovitrifyistouse instrumen-tationdevelopedforplunge-freezing.Thesedevicesallowcontrol overcryogentemperature,aswellasparameterssuchasblottime, humidityandtemperatureinthevicinityofthegridbeforeand dur-ingplunging,whichimprovesreproducibilityandenablesfreezing oflivecellsculturedongrids.Thinsamplesarevitrifiedbydirectly plungingthemintoacryogenwithhighheatcapacityandthermal conductivity,suchasliquidethaneatitsmeltingtemperatureof ∼90K.Practically,forpurifiedproteinsamples,typically2–5lis appliedonthegridand,inordertomakethesamplesufficiently thinforvitrification,mostfluidis removedbyblottingwith fil-terpapertoleaveathinlayerofliquid(typically30–100nm)over
Fig.3. Cryoelectronmicroscopysamplepreparationflowchart.
SamplepreparationflowchartforcryoEMinwhichchoices(grey)forseveralpreparationandvisualizationtechniques(blue)dependon:(i)thesamplethickness,and(ii) thedesiredstructurestobevisualized.Samples(green)thinnerthan10m(e.g.proteins,viruses,bacteria)canbevitrifiedbyplunge-freezing,whilethickersamples(e.g. cellpelletsortissueblocks/biopsies)canbecryo-fixedbyHPF.Samplesthickerthan∼500nm(e.g.cellsandtissue)mustbethinnedpriortoimaginginordertoensure electrontransparency.Thinningcanbeperformedeitherbycryo-ultramicrotomyorcryo-FIB-milling.Finally,cryo-preservedsamplesthinnerthan500nmcanbeimagedby 2DcryoEM(projectionimagingorsingleparticleanalysis)or3DCET(tiltseriesacquisitionfortomographyorsub-tomogramaveraging).(Forinterpretationofthereferences tocolorinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle.)
liposomes.ForcryoEMofcellularsamplesthatarecultureddirectly onthegrid,goldgridsareusedasasubstratetoavoidthe cytotox-icityofcopper.
2.2. Highpressurefreezingforthicksamplesandwholecells High-pressurefreezing(HPF) canbe usedto vitrifysamples upto200minthickness(DahlandStaehelin,1989;McDonald and Auer, 2006; Moor and Riehle, 1968), and is essential for cryo-fixationofadherentcells, cellpellets,tissuesandbiopsies. WithHPF,highpressureisappliedtoa sampleduringfreezing, theexpansionofwaterduringcrystallizationiscounteracted(le Chatalier’s principle),inhibiting theformation of hexagonal ice crystals(thecommonformoficefoundinfreezers)andthereby promotingtheformationofvitrifiedice.
DedicatedHPFapparatuses(Studeretal.,2001)freezesamples within∼20ms at200MPausinga jetofliquid nitrogenat77K (−196◦C)inacontrolledandreproducibleway.Withtheaddition ofarapidtransfersystem(Verkade,2008),correlativelightelectron microscopyimaging(CLEM,seebelow)isfacilitatedbymakingit practicallypossibletovitrifythesamplewithin∼4safterimaging with(fluorescent)lightmicroscopy.
Arecentlyproposed,relativelystraightforwardmethodforHPF utilizesself-pressurization,inwhichthesampleiscontainedwithin aclosedcoppercapillarytube(LeunissenandYi,2009).Upon cool-ingthetubeinliquidnitrogen,theclosedtubefreezesfromthe outside,whichincreasesthepressureinthecenter.Thismethod excludestheneedforspecialistHPFequipment,althoughtheuse
ofcryo-protectantswithinthesamplesappearstobenecessary tomaximizethesuccess-rateofhigh-qualitysamplepreservation (Hanetal.,2012).
2.3. Cryo-sectioningthicksamples
Specimens thicker than 500nm, prepared by either plunge-freezingorHPF,requirethinningpriortoimagingwithCET.This thinningmustbeperformedbelow∼140K(∼−130◦C),the tem-peratureabovewhichvitreousicebeginstorecrystallize(known asdevitrification). Thinning canbeperformedusing adiamond knifetoslicethesampleintomanythinsections(e.g.50–150nm) usingultramicrotomyatcryogenictemperature,oftenreferredto as cryo-sectioningor cryo-electron microscopy of vitreous sec-tions(CEMOVIS) (Al-Amoudi et al.,2004a).CEMOVIScombines cryo-preservationwiththepossibilitytoimageslicesfromthick samples,andenableshighresolutionimagingofe.g.vitrifiedcells thatcouldnotbevisualizedotherwise(Al-Amoudietal.,2004b). Cryo-sectioningofvitreoussamplesistechnicallydemandingand canintroduce artifactswhich distort thesample, suchasblade marksandcompressionfractures(Richter,1994).Manyofthese cryo-sectionscanbetransferred ontoan EMgrid,and multiple sectionscanbeimagedbyCET.
1999). Recently, dedicated scanning electron microscopes have beendevelopedtofacilitateFIBmillingofvitrifiedbiological mate-rialpreparedbybothHPF(Haylesetal.,2010)andplunge-freezing (Markoet al.,2007)sampleswhich cannotbethinnedby cryo-sectioning(Fig.3).WhileFIBmillingresultsinasinglecellularslice, cryo-sectioningcanbeusedtoimagemultiple,sequentialsections. Galliumionsareexpectedtodepositonmilledsurfacesandinthe sections,andirregularitiesofthemilledsurfaceoftenoccur,which inextremecasescanleadtoproblemsduringimaging(Rigortetal., 2012a;RigortandPlitzko,2015).
2.4. Stabilityofvitreoussamplesintheelectronmicroscope
Aftercryo-fixation,duringstorage,transport,thinning,transfer intotheTEMandimageacquisition,thetemperatureofthesample mustbemaintainedbelow∼140K(∼−130◦C)toavoid devitrifica-tion.Inaddition,samplesmustbekeptfreefromcontamination, suchastheadhesionofsmallice-crystalsthatmayfloatinliquid nitrogen,atalltimes.Therefore,cryoEMgridsaretypicallykeptand transferredinclosedstorageboxesundernitrogenliquidorgasat ∼77K(−196◦C)beforebeinginsertedintotheTEM.
Mostelectronmicroscopesareusedforbothconventional(room temperature)EMandcryoEM.Theyusearemovablecryo-holder whenperformingcryoEM,which isinsertedintothesideofthe TEMcolumn(aso-calledside-entrysystem).Thecryo-holder com-prisesaDewarfilledwithliquidnitrogentomaintainthegridat 95K(−178◦C).TheDewarmustberefilledmanuallywithliquid nitrogeneveryfewhours,whichlimitstheabilitytoautomatedata collectionforaprolongedperiodoftime.Themechanicalstability ofthesampleisalsolimitedbythetemperatureinstabilityofthe cryo-holdercausedbydiminishingliquidnitrogenlevelsandheat flowbetweentheholderandthemicroscope.Thiscausesthe sam-pletodriftovertime,furtherlimitingtheabilitytoautomatedata collectionoverthecourseofhours.
Whenworkingwithaside-entrysystem,EMgridsareloaded individually,andthereforeeachnewsamplerequiresre-insertion ofthecryo-holderintotheTEM.Eachinsertiondeterioratesthe low-pressureenvironmentwithintheTEM-columnnearthe sam-ple,andsubsequentlyincreasesthechanceofcontaminationofthe EMgrid.Moreover,overtime thesamplethicknesscanincrease duetoicegrowthonthesamplewithintheTEMcolumn,limiting themeasuringtimethatcanbeusedtocollectatiltseries.
Duringthelastfiveyears,manyoftheselimitationshavebeen removedwiththedevelopmentandintroductionofnoveltypesof electronmicroscopesdedicatedtoperformcryoEMforaprolonged time-period.ThesetypesofTEMhaveanautomatedspecimen load-ingsystemthatholdsmultiplegridsandautomaticallytransfers themintoacontinuouslycryogenically-cooledandstablespecimen stagewithoutdisruptingthehighvacuumenvironment. Addition-ally,thequalityofthevacuumsystemofthesetypeof TEMsis significantlyenhanced,whichresultsinnegligibleicegrowthon thesampleoverthecourseofdays,allowingextendedautomated datacollection.
3. Datacollection
3.1. Tiltseriesacquisition
Electronsdamagebiologicalspecimens(Glaeser,1971)andas suchtheelectrondoseexposedtothesamplemustbeminimized. Thiscanbeachievedusingasearch,focusandimagescheme(a so-calledlow-dosescheme)duringdatacollection,wherebythe desiredimaginglocationsareidentifiedatlowmagnification(and hencelow electrondose)whilefocusingisperformedata loca-tionadjacenttotheareaofinterestpriortoimageacquisition.This
minimizestheamountofdose,andhencedamagethatthearea ofinterestisexposedtoandmaximizestheachievableresolution. Thetotaldoseusedforimagingmuststillbeverylow,typicallyless than100e−/Å2,toachievenanometerresolution(Fig.3,toprow). Becausetomographyisbasedonimagingthesameregionfrom multipleangles,thisdosehastobedistributed(fractionated)over thenumberofimagesthatisusedtorecordatiltimage.Tiltseries aretypicallycollectedfrom±60◦,withaseparationof2◦between images;eachtiltseriestherefore imagesthesamelocation∼61 times,resultinginadoseoflessthan2e−/Å2 foreachtiltimage. Duringtilting,theapparentthicknessofthesampleincreasesas 1/cos(),whereisthetiltangle.Samplesappear2×thickerat 60◦tilt,hencethedistancethatelectronsmusttransmitthrough, andthelikelihoodof multiplescatteringeventswiththeatoms withinthesample,alsoincreases.Consequently,thefractionated electrondoseofthetiltseriesresultsinlowcontrastandhigh lev-elsofnoiseintheindividualimages(Fig.3,bottomrow,leftimage), whichmakesaccuratealignmentofatiltserieschallenging.To facil-itateandimprovetheaccuracyofalignment,goldfiducialmarkers between5–25nmindiameteraregenerallyaddedtothesample(or onthespecimensupport)immediatelypriortovitrification.These fiducialmarkershavehighcontrastinthelow-doseimages,and canbecomputationallytrackedaspoint-likeobjectstoalignthe imagespriorto3Dreconstruction(Lutheretal.,1988;Walzetal., 1997).
Overall,theresultinglowdosethathastobeusedforimaging tominimizeradiation-inducedspecimendamageresultsinnoisy imagesinthetiltseries(Fig.4),whichhampersaccuratealignment ofatiltseries.Asaconsequence,cryo-electrontomogramswill typ-icallyexhibithighnoiselevels,lowcontrastandlimitedresolution. Technicalimprovementsthatincreasethesignal-to-noiseratioin theindividualimagesofatiltserieshaveadirecteffecton improv-ingthequalityoftheresultingcryo-tomograms.Duringthelast threeyears,thequalityofdatacollectedforcryo-tomographyhas improvedsignificantly,primarilybythreemajor developments: phaseplates,energyfiltersanddirectelectrondetectors.
3.2. Phaseplatesimprovetomograminterpretability
As cryoEMsamples are very thin (less than a few hundred nanometers)andarecomposedofrelativelylightelements (typi-callycarbon,hydrogen,oxygenandnitrogen),thecontrastforming mechanism canbevery welldescribed by (weak−) phase con-trastimaging.Phasecontrastisgeneratedbyphasedifferencesthat occurbetweenanunscatteredelectronwave(themajorityofthe electronsthatdonotinteractwiththesample)andtheelectron wavesthatinteractwithatomswithinthesample.Totranspose thisphaseshiftintoameasurablecontrast,imagesaretypically takenoutoffocus,whichincreaseslow-resolutionimagecontrast butnegativelyaffectsthehighresolutionintheimages.Theway thatthephase(andamplitude)oftheelectronwavesareinfluenced bytheobjective lensofthemicroscope(e.g.defocusing)canbe mathematicallydescribedbythecontrasttransferfunction(CTF) (Marabinietal.,2015).TheCTFdescribesaninverserelationship betweencontrastandresolution, andadditionallypredicts con-trast reversalsat certainfrequencies,resulting infringes in EM images.Fortunately,imagescanbe(partly)correctedfortheCTF, whichreliesonaccuratedefocusdetermination.CTFcorrectionin CETisnotalwaysatrivialtasksinceimagesaretiltedandfocus varieswithinindividualimages.Nonetheless,CTFcorrectioncan alsoincreasetheresolutionoftomograms(Fernandezetal.,2006; Turonovaetal.,2017;Xiongetal.,2009).
Fig.4. Radiationdamageandelectrondose.
Electronradiationdamageofcryo-samplesof(gammaproteo)bacteriaoccursdirectlyuponimagingandincreaseswithdose.High-resolutionatomicfeaturesarelostwithin afewelectronsperÅ2(e/Å2)whilevisibledamageoccurslater(toprow,totalelectrondose).Becauseofradiationdamagetheelectrondoseforimagingislimited,which
greatlyaffectsthesignal-to-noiseratiointypicalcryoEMimages(bottomrow,electrondoseperimage).
Fig.5. TheVoltaphaseplateincreasesthecontrastofelectrontomograms.
(A)Incomingelectronwaves(blue)haveacertainphaseandamplitude.Transmitted,non-diffracted,electrons(blue)donotinteractwiththespecimenandtheirphaseis unaffected.Incontrast,someoftheelectronwaveisdiffractedbythespecimen(red),whichretardsthephaseofthediffractedwavebyquarterofawavelength.Thissmall phaseshiftcausesonlyminimalcontrastofthemagnifiedspecimen(B).InthepresenceoftheVPPthediffractedwaveisretardedfurther,ideallybyanotherquarterofa wavelength(green),placingit180◦outofphasewiththenon-diffractedtransmittedwave(blue),resultinginmaximalnegativephasecontrast(C).Micrographsacquiredat
focusofvitrifiedliposomesandlacey-carbonfilmintheabsence(B)andpresence(C)oftheVPP,demonstratingthecontrastimprovement.
correction,comparedtoimagesacquiredusingdefocuscontrast. Electronsarediffractedbyinteractionwiththinbiological speci-mens,whichbehaveasweakphaseobjectsandretardthephaseof thediffractedelectronwavewithrespecttothetransmitted (non-diffracted)electronwavebyaquarterofawavelength(Fig.5A),but nottheamplitude,sothatverylittlecontrastisvisibleatlow fre-quency(Fig.5B).Unfortunately,thephaseinformationislostand onlyamplitudedataiscollectedbythedetectorintheformof inten-sity.So-calledZernike-typephaseplatesfurthershiftthediffracted electronwaves,generatinganapproximatehalf-wavelengthphase shiftwithrespecttothenon-diffractedelectrons,turningthisphase informationintomeasurableamplitudecontrastwithouttheneed todefocus(Fig.5C)(DanevandNagayama,2001).Thisresultsin
destructiveinterferenceandmaximalcontrastatlowresolution (Fig.5C),whichmakesprojectionimagesandreconstructed tomo-gramsmuchmorereadilyinterpretable(Fukudaetal.,2015;Sharp etal.,2016).
Fig.6.Energyfiltering.
Electronswithacertainenergy(here200keV±1eV)areemittedfromanelectronsource(FEG).Elasticinteractionswiththespecimendonotresultinenergyloss(black wave)whileinelasticinteractionsleadtolowerenergyelectronsthathavealongerwavelength(redwave).Aprismlenstransfersthisdifferenceintoashiftoftheelectrons thathavelostenergy,whichareremovedbyaslit.Insetshowsthatanenergyfilteredimageislessbrightbuthasmorecontrastthananunfilteredimage.(Forinterpretation ofthereferencestocolorinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle.)
phaseplate.ThetransmittedelectronspassthroughthislocalVolta potentialwithouttheirphasebeingaffected,whilstthediffracted electronsinteractwiththerestofthephaseplateandareretarded byafurtherquarter-wavelengthbyinteractionwiththecarbonof specificthickness,therebygeneratingahalf-wavephaseshiftwith respecttothetransmittedelectrons(Fig.5).Becausethe transmit-tedelectronbeamcreatesthelocalphaseshift,theVPPrequires minimalphysicalalignment,andtheVoltapotentialcanbe gen-eratedonthecarbonfilm whenand where desired, facilitating automateddatacollection.However,sincethephaseshiftis gen-eratedbyirradiationwithelectrons(knownasconditioning)the amountisdependentontheelectrondose,whichresultsin unsta-blephaseshifts(Danevetal.,2014;Sharpetal.,2017).Nevertheless, thevariablephaseshiftsstillincrease thecontrastofthe tomo-gram,andtheVPPhasalreadyshownpotentialforinsitustructure determinationofcomplementactivation(Sharpetal.,2017)and membraneattackcomplexporeformation(Sharpetal.,2016)on liposomes,aswellasthecellulardistributionofproteasomes, ribo-somesandnuclearporecomplexes(Asanoetal.,2015;Mahamid etal.,2016).
3.3. Energy-filteringreducesnoise
Whenelectronsareincidentonasampletheycaninteractwith thesampleineitheranelasticevent,wherenoenergyistransferred betweenelectronandspecimen,aninelasticevent,whereenergy islosttothespecimen,ortheelectronsdonotinteractwiththe sampleandaretransmittedunaffected(Egerton,2011). Inelasti-callyscatteredelectronscontributetothenoisewithintheimage anddamagethespecimenbybreakingmolecularbonds,localized heating,andevolutionofhydrogengas,amongothereffects(Baker
andRubinstein,2010).Theratioofscatteredversusunscattered electronscanbepredictedbasedonthemeanfreepathofthese electrons,theaveragedistancetravelledbetweentwosuccessive scatteringevents.Thismeanfreepathfor200–300keVelectrons isontheorderof200–300nm(FejaandAebi,1999;Grimmetal., 1996)andisdifferentforelasticandinelasticevents.Thelikelihood oftheseeventsthereforeshiftsdependingonsamplethickness.In cryoEMsamplesthatarelessthan100nmthick,mostelectrons passthroughthesampleunscattered,whiletherestmainlyscatter onlyonce,eitherelasticallyorinelastically.However,thelikelihood thatelectronsscattermorethanonceincreasesquicklywith speci-menthickness(Hanetal.,1995).Sincemultiplescatteredelectrons resultinnoiseintheimage,samplesarepreferablynotthickerthan themeanfreepath.Withsamplesthat havea thicknessofone totwotimesthemeanfreepath,suchasbacterialcellsandthin adherentcells,removaloftheinelasticallyscatteredelectronscan significantlyincreaseimagecontrast.Thiscanbeachievedbyusing anenergyfiltertoblockelectronsthathavelostorgainedenergy.
pri-Fig.7. Electrondetectors.
(A)TypicallayoutsofCCD(chargecoupleddevice)cameraandCMOS(complementarymetaloxidesemiconductor)directelectrondetector(DED).CCDcamerasdetect electrons(e−)viaaluminescentscintillatorlayerthatconvertselectronsintolight(␥).ThislightistransferredviaafiberopticcouplingintoaCCDchip,whichconvertsthe lightintoavoltageonasquarearrayofpixelsthatisreadoutinafewseconds,resultinginadigitalimage.(B)Incontrast,adirectelectrondetectorCMOSchipdirectly detectsincomingelectronsandconvertsthemintoavoltageoveranarrayofpixelsthatcanbereadoutathighspeed.
maryelectronbeamoperateinaso-calledzero-lossimagingmode, removingthebackgroundnoiseduetoinelasticallyscattered elec-trons.Consequently,thecontrastoftheresultantimageisincreased by∼16%comparedtounfilteredimages(Fukudaetal.,2015).Apart frombeingusefulforthickerspecimens,itisalsousefulin tomogra-phy,where,upontilting,theslab-shapedspecimeneffectivelygets thickerathightiltangles.Zero-lossenergyfilteringcanbeused togetherwithphaseplateimaging,andprovidean∼68%increase incontrastwhenusedincombination(Fukudaetal.,2015).
3.4. Directelectrondetectors
Electronmicrographswereoriginallyacquiredonphotographic film,whichneedstobedevelopedandscannedbeforedata anal-ysisortomogramreconstruction.Digitalcamerasaremuchmore convenientduetotheirnear-instantaneousreadout.Thefirst dig-italdetectorsforEMwerecharged-coupledevice(CCD)cameras (Dierksenetal.,1995;Kosteretal.,1992;KrivanekandMooney, 1993;SpenceandZuo,1988)(Fig.7A).CCDcamerasincorporate aluminescentscintillatorlayerthatconvertsincidentelectronsto photons,andthescintillatorlayerisconnectedtoaCCDchipby fiber-coupledoptics(Fig.7A),which spreadstheelectronsignal overa largeareadependingonthethicknessofthescintillator, resultingin lower resolutionimages. Thisspreading of the sig-nalcanbedescribedbya so-calledpoint-spreadfunction(PSF). Anotherimportanteffectofthisdesignisthatnotallincoming elec-tronsareregisteredasphotonsonthechip.Thelikelihoodthatan incidentelectronisactuallyregisteredisdescribedbythedetector quantumefficiency(DQE).Ideally,allelectronsaredetectedandno signalcomingfromthesampleislost.Furthermore,itisimportant thatdigitalcamerashaveahighdynamicrange,candetectbothlow andhighlevelsofelectrons,andtheirresponseoverthisbrightness rangeislinear.So,overall,gooddetectorscanbedefinedbyhaving asmallPSF,ahighDQEandideallyalinearsensitivityoverahigh dynamicrange.
TheprimaryreasonfortheresolutionrevolutionimpactingEM istheadventofdirectelectrondetectors(DEDs)(Xuongetal.,2007). DEDsdirectlyconvertelectronsintoanelectricalsignalforoutput detection(Fig.6B).Theyveryefficiently detectelectrons,which resultsin a highDQE,much betterthan film(McMullanet al., 2014).Therefore,alsothePSFnowdependsonscatteringof elec-tronsinsidethedetectoritself,whichcanbeminimizedbymaking thedetectorasthinaspossible.AnotheradvantageofDEDsistheir fastreadout,whichcanreachafewhundredframespersecond. Insteadofsingleimages,moviesofthesamplecanberecorded. Thisisusefulsinceirradiationbytheelectronbeamandthe
result-ingbeamdamageofthesamplecanresultinimagemovementand unpredictablelocaldistortionswithinthesampleitself.By record-ingmovies,thesemovementsanddistortionscanbecorrectedbya posteriorialignmentanddistortion-correctionoftheframes,which compensatesforsamplemovementandincreasestheresolutionof theimage.Anotheradvantageofthefastreadoutisseenwhenalow electrondoseisusedandthusindividualelectronscanbedetected. Infact,singleelectronscanbelocalizedwithsub-pixelaccuracy, so-calledsuper-resolutionmode,whichalsopositivelyaffectsthe DQEofthecamera(McMullanetal.,2014).InSPA,nownearatomic resolutionmapsareregularlyproducedofproteinandprotein com-plexes(Merketal.,2016).Intomography,theresolutionincrease resultingfromDEDshasenabledvisualizationofproteinsinside cellsandalsonear-atomicresolutionmapsusingsub-tomogram averaging(Schuretal.,2016).
3.5. Automateddatacollectionandstorage
Tomographicdatacanbeacquiredautomaticallybyseveral pro-grams,ofwhichserialEM(Mastronarde,2005),UCSFtomography (Zhengetal.,2004),TOM(Nickelletal.,2005),andXplore3D/Tomo4 (Thermo-Fischer,formerlyFEIcompany)aremostcommonlyused. Althoughthesearedevelopedfromdifferentperspectives(e.g.type ofsample,typeofmicroscope,typeofusers),theseprogramsin essenceperformthesameroutines(Kosteretal.,1992;Zieseetal., 2002).Withthecurrentimprovementsinhardware,tomographic acquisitionsoftwareistargetedtowardsautomatedrecordingof multipletilt-seriesoverseveraldays,withthesupportof movie-moderecordingbydirectelectrondetectorcamerasandtheVolta phaseplate.
Automateddatacollectionand movie-modesuper-resolution imaginghasfar-reachingconsequencesforimageprocessingand datastorage.Thenumberofdatasetsthatcanberecordedperday increasesonefactor,whilethesizeofasingletomogramtiltseries dataset,typically∼1Gbcaneffectivelyincreasebetweenoneortwo factors,witha4×increasegoingfromnormaltosuper-resolution mode and typically 7–30× increase due tothe frames used in movierecording.TheselargedatasetsneedCTFcorrection,movie alignmentandtomographicimageprocessing.Standarddesktop computershardlysuffice,andworkstationsareoftenemployedfor computation.
Fig.8.LayoutofsingleanddualaxistiltseriesinFourierspace.
Effectofangularsamplingandmissingwedgeontheresolutionintomograms.TheeffectoftheangularsamplingandthemissingwedgeintomographyshowninFourier spaceslices(a),surfacerenderingsofareconstructedthin-shelledsphere(b),andYZslicesthroughthecorrespondingtomogram(c).Theleftcolumnshowsthesituation whenthereisfullangularsamplingalongonetiltaxis(theX-axis).
ThetopimageshowsthatFourierspaceissampledequallyintheYZplane.Thesurfacerenderingandacentralsliceofthetomogramindicatenoanisotropicresolution inthereconstructedsphereinthosedirections.Themiddlecolumnshowsthesituationwhenthereislimitedtiltangle(from−60◦to+60◦)alongasingleaxis.Theresult
willbeamissingwedgeinFourierspacethatcanbeobservedinthesurfacerenderingandasdiminishingdensityatthetopandbottomofareconstructedsphere(b)and anelongationintheYZslice.Therightcolumnshowsthesituationwhentwotomogramsarecombinedfromtwoorthogonaltiltaxes(dual-axistomography)withlimited rotationbetween−60◦and+60◦.InFourierspace,themissingwedgeisreducedtoamissingpyramid,andtheanisotropyintheresolutionalongtheY-axisisreduced(from
G.vanTendeloo,D.vanDyckandS.J.Pennycook“HandbookofNanoscopy”,chapter36,page1313.2012.CopyrightWiley-VCHVerlagGmbH&Co.KGaA.Reproducedwith permission).
structuredatabases,suchastheproteindatabase(PDB)(Bernstein etal.,1977)andotherimagingmodalities(Patwardhanetal.,2014; Patwardhanetal.,2017).
4. Imageprocessing
4.1. Tomogramreconstructionautomation
Typically, reconstruction of cryo-electron tomograms con-sists of four steps; pre-processing, tilt-series alignment, post-processing,andreconstruction.Preprocessinginvolvestruncation ofextremedatavaluescausedbyX-raysorocclusionbytheEMgrid. Tilt-seriesalignmentisnecessaryasthesampleisnotrotated per-fectlyduringacquisition.Obviouswhole-imagemovementscaused byimperfectstagebearingscanbealignedusingcross-correlation oftheimages.Smallerandlocaldistortionsarecaused bybeam inducedmovement,specimendriftanddefocuseffects.For low-contrastCET tilt series,fine alignment is usuallyperformed by trackinghigh-contrastfiducialmarkers,5–25nmgoldbeadsthat areaddedtothesample priortovitrification specificallytoaid alignment(Lutheretal.,1988;Walzetal.,1997).However, alter-nativemethods,suchascross-correlationandpatchtracking,are availablethatdonotrequiretheadditionoffiducialstothesample (Amatetal.,2010;Guckenberger,1982).Accuratefinealignment
iscritical,aspooralignmentresultsinprojectionmismatchand limitstheresolutionoftheresultingtomogram.Micrograph post-processing includes CTF correction and noise filtering. Finally, 3DreconstructionscanbeachievedbyeitherFourier-space algo-rithms,suchas weightedback-projection (WBP)(Radermacher, 1988),orreal-spacemethods,suchassimultaneousiterative recon-structiontechnique(SIRT)(Gilbert,1972).WBPismorecommon, and involves taking a Fourier transform of each image before placingtheseslicesintoa3DFouriervolumeattheangles spec-ifiedduringtiltseriesacquisition.TheFouriercomponentsofeach slicearethenweighteddependingontheirfrequencytoprevent oversamplingoflow-resolutiondetails.Next,aninverseFourier transform back-projects the volume into real-space, yielding a tomographicvolume.ReconstructionusingWBPisgenerallyfaster thanSIRT,althoughWBPtomogramscanbedominatedby high-frequencynoise,whereasreal-spacereconstructionmethods,such asSIRT,generallytakelongerdue totheiterativenatureofthe method,althoughnoiseissuppressed.
Fig.9. Sub-tomogramaveraging.
(A)Fromatomographicvolume,sub-volumescontainingsingleparticlesofthesamebiologicalstructureareextractedfromtomographicvolumes.Inthiscase,theparticle isamembraneattackcomplex(MAC)poreperforatingaliposome.Theparticlesarealigned(B)andaveraged(C)togenerateaninitialmap.(D)Particlesaretheniteratively re-alignedtothepreviousmapandaveragedtogenerateimprovedmaps(i–iv).(E)Thefinalmapwiththehighestresolutionisachievedwhenaniterationdoesnotgenerate anyimprovement,andcanbeinterpretedwiththeuseofcrystalstructures(EMD-3289(Sharpetal.,2016)).
manual processing to achieve the best quality. Reconstructing tomogramsfromatiltserieswasoriginallyaninteractive, labor-intensiveprocessinwhichmanystepsweremanuallyperformed andchecked,especiallygoldfiducial finealignment,inorderto generateahigh-qualitytomogram.Sincemanualtomogram recon-structionisratherinefficientandslowerthandataacquisitiontime, automatedtiltseriesimagealignmentandtomogram reconstruc-tionproceduresarehighlydesirable.Recently,severalautomated fiducial markedbased andmarker-free finealignment schemes havebeendeveloped(Castano-Diezetal.,2010;Hanetal.,2015; Sorzanoetal.,2009)andimplementedinseveralpackagessuchas IMOD(MastronardeandHeld,2017),Protomo(NobleandStagg, 2015)andUCSFtomo(Zhengetal.,2011)(foranextensivelistof tomographypackagesseehttps://en.wikibooks.org/wiki/Software ToolsForMolecularMicroscopy),aswellasindualaxistiltseries (Winkler andTaylor,2013).Furthermore,severalpackageswith additionaldatabasestorageareavailable(Dingetal.,2015;Zheng etal.,2007).
4.2. Themissingwedgelimitsinterpretationoftomographic volumes
Duetothegeometryofboththespecimen/EMgridand objec-tivelensofthemicroscope,tiltseriesaretypicallycollectedover anangularrangebetween±60◦(Fig.1).Thisresultsina“missing wedge”ofdata(inFourierspace),causinganisotropicresolutionin thetomogram,whichisseenaslossofresolutionintheaxial direc-tion(Fig.8)(Diebolderetal.,2015;Sharpetal.,2017).Anisotropic resolutionisalsoinfluencedbythemissinginformationbetween sequentialtiltimages, whichhaveangularstepsof2–3◦ (Koster etal.,1997).Anisotropycanbereducedusingdual-axis tomogra-phy,wherebytwotiltseriesareacquiredwitharelative90◦inplane rotation(Penczeketal.,1995).Thetworesultingtomogramsare thenalignedandcombinedintoasingletomographicvolume.This procedurereducesthemissingwedgeintoa“missingpyramid”, whicheliminatesanisotropyespeciallyintheplaneorthogonalto
thetiltaxis(Mastronarde,1997),butdoesnotremoveblurringof thereconstructionalongthedirectionoftheelectronbeam. Dual-tilttomographyrequiresthatthesameregionisimagedtwiceand thereforeitisnecessarytofractionatetheelectrondoseoverboth tiltseriesinordertopreventincreasedradiationdamage. There-fore,eachofbothdualaxistiltseriesreceiveshalfofthetotaldose comparedtoasingleseriesandthustheseeitherhavevery poor-contrastimagesorhalfthenumberofimages.
Onemethodtoeliminatethemissingdataentirelyisimaging arod-shapedsampleinaholderthatcanberotatedbyafull180◦ (PalmerandLowe,2014).Alternatively,incasethetomogram con-tainsmanyidenticalcopiesofaproteincomplexorparticle,the missingdataoftheseparticlescanbefilledinbyvolumeaveraging, knownassub-tomogramaveraging.
4.3. Sub-tomogramaveragingandtemplatematching
meth-Fig.10.Surfacerenderingofcellularcryoelectrontomograms.
(A)Aslicethroughacryo-electrontomogramofacellshowsallkindsofstructuresin2D.(B)Somestructuraldensitiescanbesurfacerenderedbytemplatematching,e.g. actin(blue),whileothersaredrawninbyhand:lipidmembranes(yellow),microtubules(green),mitochondria(purple),andintermediatefilaments(red).(Forinterpretation ofthereferencestocolorinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle.)
ods,evennearatomicresolutionstructurescanbeachieved(Schur etal.,2016).
Apartfromaveragingsub-volumesfromatomograminorder toincreaseresolution,onecanlookforknownstructuresinside a tomogram. The structure of many molecular machines from cellsaredetermined(e.g.fromX-raydiffraction,cryoEMandNMR techniques)andavailable(e.g.fromtheProteinDataBank(PDB); www.rcsb.org(Bermanet al.,2000)or theElectronMicroscopy Database(EMDB)(Lawsonetal.,2016)).Byutilizinga template-matchingapproach, an extensivesix-dimensional(three spatial dimensionsandthree rotationaldimensions)cross-correlational searchbetweenthemolecularvolumeandthecellulartomogram canbeperformedtofindthepositionandorientationofcertain proteinmoleculesinsidepartsofacell(Nickelletal.,2006). Molec-ularcrowdingandlowtomogramresolutionlimittheapplicability ofthistechnique(Becketal.,2009),buthugeimprovementshave beenmadeusingthestate-of-the-artimagingtechniquesdescribed herein(Mahamidetal.,2016).
4.4. Tomogramsegmentation
Asinmedicaltomographicreconstructions,theinterpretation of3Dvolumedataisdependentonaspecialistwitha familiar-itywiththe specimenand knowledgeof thepotentialartifacts presentintomograms.Interpretationisdifficultfortheuntrained eyebecauseoneisaccustomedtolookingatsurfacesandnotat noisyslicesfromvolumes.Therefore,surfacerepresentationofthe volumesintobiologically-relevantsegmentsgreatlyaidsthe inter-pretationand 3D viewof the volumefor both the trained and untrained eye(Fig.10).A particularproblem with(automated) segmentationofcryo-electrontomogramsisthatthevolumesare noisy(duetothelimitedelectrondose),anisotropicallydistorted (duetothetiltgeometry/missingwedge), andthat density val-uesarenotlinearwiththeoriginalsignal(duetotheCTF,phase contrastandreconstructionalgorithms).Inpractice,structuresof interestareoftendrawnbyhandintothevolume,whichisboth highlytime-consumingandsubjective.Automatedsegmentation isdesirablebecauseitislessuserintensiveand,moreimportantly, itprovidesanobjective,reproducible waythatallows interpre-tationandquantificationofresults.Theseobjective,reproducible segmentedtomogramscanalsobedepositedin databasessuch asEMDB(Lawson etal.,2016).Identifyingparticularstructures withincellularvolumesislimitedbytheresolutionofthe tomo-gram.However,certainlargerstructures,suchasactinfilaments andmicrotubules,weresuccessfullytracedandsegmented auto-maticallybytemplatematchingofrod-likeshapes(Rigortetal.,
2012b;Rusuetal.,2012).Also,someinitialattemptson segmen-tationofsurfacesofliposomesweremade(Koningetal.,2013). Eventemplatematchingofspecificproteinsinsidecellsis possi-ble(Asanoetal.,2015).TheincreasingqualityofCETimagingand theresultingtomograms,combinedwithdevelopmentand imple-mentationofbetterandfastertemplatematching,neuralnetwork algorithmsandincreasingcomputationalresourcesmakesit possi-bletoautomaticallysegmentand/ortemplate-matchmanyofthe largerstructuresinsidecellulartomograms(Chenetal.,2017),and developmentsinrelatedfields,suchasmulti-dimensionaltransfer functionvolumerendering(Knissetal.,2002),arealsoexpectedto yieldimprovementsintomogramvisualization.
5. Relatedtechniques
AsidefromtheCETtechniquedescribedabove,whichisbased on 2D imaging at different angles, there are several related newly-developingcryotechniquesthatareworthmentioningin the context of three-dimensional imaging of cells and tissues inan anatomicsetting:cryoserial block-facescanningelectron microscopy(cryoSBF-SEM),whichcombinesserialsectioningand imaging of surfaces with scanning electron microscopy (SEM), cryosoftX-raytomography(cryoSXT),inwhichX-raysareused for imaging,and cryocorrelativelight and electronmicroscopy (cryoCLEM),which combinesfluorescentlightmicroscopy(fLM) imagingfortargetingspecificsiteswithCET.
5.1. CryoSBF-SEM
Fig.11.Cryo-CLEM.Cryocorrelativelightandelectronmicroscopy.
(A)Cryo-fluorescentlightmicroscopy(cryo-fLM)imagingoffluorescentlytaggedStreptomycesbacterialcellscanaidthetargetingofcertainstructures.(B)ImagingcryoEM ofthesameareaallowstargetingoflabeledstructuresintheelectronmicroscope.(C)HighermagnificationcryoEMimagingandaccurateoverlaydeterminesthespecific siteforCETimaging.(D)Overlayofthecryoelectrontomogramsliceandthecryo-fLMlightsupthelabeledstructuresinsidethecell.
5.2. CryoSXT
Incryosoft X-raytomography (cryoSXT),likeCET,a3D vol-umeofbiologicalmaterialisalsoreconstructed fromaseriesof projectionviews,thoughwithseveraldifferences.X-raysareused forimaging,andsincetheyhavelargerpenetrationdepthfor vit-rified biological samples than electrons, complete cells can be imaged.ThesoftX-rayshaveatunedenergythatfallswithinthe so-called‘waterwindow’(between284–543eV)whichoptimizes theabsorptiondifferenceof X-raysbetweenoxygenandcarbon elements,which isthecontrastingmechanism.TheX-raybeam isfocused ontothesample andthedetector byzone-plates, X-raydiffractingelements.Thecurrentresolutionisintheorderof 25–35nm.CorrelativecryofLM/cryoSXTtechniquesarealso devel-oped and typicalexamples of cryo-SXTimaging are wholecell reconstructionsoftheyeastSaccharomycescerevisiae.Foramore in-depthreview,seeCarzanigaetal.(2014).
5.3. Cryo-correlativelightandelectronmicroscopy
TherecentdevelopmentsinCETenableimagingof macromolec-ularandproteinstructuresinsidecryogenicallypreservedcellular landscapes(Mahamidetal.,2016).Suchhigh-resolutiondatasets relyonhigh-magnificationtomograms,whichlimitstheareathat canbeimaged.Specificpositionsorareasofinterestcantherefore bedifficulttoidentifywithinthecell.Onesolutionistoemploy cryo-correlativelightandelectronmicroscopy(cryoCLEM),which combinesCET withcryo-fLMof thesame sample(Fig.11).This ispossibleby usinglabelling techniquesforfLM,eithergenetic tagging(suchasGFP)orbychemicallabelingusingspecificdyes totargetandidentifystructuresofinterest,eitherinatwo-step
LM/EMapproach(Celleretal.,2016;Hamptonetal.,2017)oran integratedsetup(Faasetal.,2013;Wangetal.,2017).Currently withcryo-CLEM,theresolutiongapbetweenfLM(∼400nm)and CET(∼4nm)istwoordersofmagnitude,althoughlocalizationcan beperformedwitha precisionof∼40nmwithfLM(Schorband Briggs,2014).Nevertheless,targetingspecificmoleculesinsidea crowdedlandscaperequiredhighercryo-fLMimagingtechniques, suchassuper-resolutionfLM(Wolffetal.,2016)foraccurate local-ization,andfuturedevelopmentsaretargetedtowardsintegrating theseintocryo-CLEM. Thiswillneed adaptationofLMsystems and thedevelopmentof probesspecificallyfor super-resolution cryoCLEM,butwillallowmultidimensionalfunctionalimagingof cellularlandscapesatnanoscaleresolution.
6. Conclusionsandperspectives
Despite these recent advances, CET is still relatively labor-intensiveandslowcomparedtoSPAcryo-EM.Datacollectionof cryo-EMislargelyautomated,aftersettingimagingparametersand semi-automatedtargetingofimagingpositionsontheholey sup-port.CETshouldbemadefasterby(i)automatedtargetingusing CLEMbytheintegrationwithfluorescentlightorramanmicroscopy techniques,and(ii)fastertiltseriesacquisition,whichnow typ-icallytakesapproximatelyanhour, byimplementationof more stablestagetiltbehavior,adaptedacquisitionsoftwareandfaster cameras,inordertoallowaspeedincreaseofcircaafactorof100 (Migunovet al.,2015).Furthermore,tomographicimage recon-structionshouldbefullyautomatedandintegratedwithacquisition andalsorobustinordertopreventlargedatastoragerequirements ofintermediaterawfiles.Automaticsegmentation,quantification andvisualizationby3Dsurfacerenderingshouldbedevelopedto aidinterpretationofdata.InordertomakeCETmorewidely appli-cable,designatedCETmicroscopesystems,comparabletomedical imagingdevices(CAT,MRI,CT)mightbethewayforward.
Dueto the advanceswe have described, the ability to per-formmolecularimagingofsubcellularstructuresatnanoscalelevel invivoandinsitumeansthattheapplicabilityofCETtobiologyand medicinehasneverbeenhigher.Futuredevelopmentswillonly increasetherelevancyofthistechniquetodiagnostics.Wewilluse thesetechniquestoimagebiologicalprocessesintheirnative envi-ronment,suchascellular,viralandbacterialinfectionsandimmune complexesbindingtocells,toelucidatethemolecularmechanisms ofdisease.
Acknowledgements
This work has been supported by iNEXT, project number 653706,fundedbytheEuropeanUnion’sHorizon2020research andinnovationprogramme undergrantagreement759517,the DutchfoundationSTWProjects13711,NanosurfAGandSmartTip B.V.,and12713,MicroscopyValley,theCouncilforChemical Sci-ences(CW)oftheNetherlandsOrganizationforScientificResearch (NWOgrant700.57.010).Theauthorsacknowledgethesupportand theuseofresourcesofInstruct,aLandmarkESFRIproject.Wethank RobHoeben,TonRabelink,TonRaapandPeterNibbering(LUMC) forcriticalreadingofthemanuscript.
References
Al-Amoudi,A.,Chang,J.J.,Leforestier,A.,McDowall,A.,Salamin,L.M.,Norlen,L.P., Richter,K.,Blanc,N.S.,Studer,D.,Dubochet,J.,2004a.Cryo-electronmicroscopy ofvitreoussections.EMBOJ.23,3583–3588.
Al-Amoudi,A.,Norlen,L.P.,Dubochet,J.,2004b.Cryo-electronmicroscopyofvitreous sectionsofnativebiologicalcellsandtissues.J.Struct.Biol.148,131–135. Amat,F.,Castano-Diez,D.,Lawrence,A.,Moussavi,F.,Winkler,H.,Horowitz,M.,
2010.Alignmentofcryo-electrontomographydatasets.MethodsEnzymol.482, 343–367.
Asano,S.,Fukuda,Y.,Beck,F.,Aufderheide,A.,Forster,F.,Danev,R.,Baumeister,W., 2015.Proteasomes.Amolecularcensusof26Sproteasomesinintactneurons. Science347,439–442.
Baker,L.A.,Rubinstein,J.L.,2010.Radiationdamageinelectroncryomicroscopy. MethodsEnzymol.481,371–388.
Bauerlein,F.J.B.,Saha,I.,Mishra,A.,Kalemanov,M.,Martinez-Sanchez,A.,Klein,R., Dudanova,I.,Hipp,M.S.,Hartl,F.U.,Baumeister,W.,Fernandez-Busnadiego,R., 2017.InsituarchitectureandcellularinteractionsofPolyQinclusions.Cell171, 179–187,e110.
Beck,M.,Malmstrom,J.A.,Lange,V.,Schmidt,A.,Deutsch,E.W.,Aebersold,R.,2009. VisualproteomicsofthehumanpathogenLeptospirainterrogans.Nat.Methods 6,817–823.
Berman,H.M.,Westbrook,J.,Feng,Z.,Gilliland,G.,Bhat,T.N.,Weissig,H.,Shindyalov, I.N.,Bourne,P.E.,2000.Theproteindatabank.NucleicAcidsRes.28,235–242. Bernstein,F.C.,Koetzle,T.F.,Williams,G.J.,MeyerJr.,E.F.,Brice,M.D.,Rodgers,
J.R.,Kennard,O.,Shimanouchi,T.,Tasumi,M.,1977.Theproteindatabank:a computer-basedarchivalfileformacromolecularstructures.J.Mol.Biol.112, 535–542.
Carzaniga,R.,Domart,M.C.,Collinson,L.M.,Duke,E.,2014.Cryo-softX-ray tomogra-phy:ajourneyintotheworldofthenative-statecell.Protoplasma251,449–458.
Castano-Diez,D.,Scheffer,M.,Al-Amoudi,A.,Frangakis,A.S.,2010.Alignator:aGPU poweredsoftwarepackageforrobustfiducial-lessalignmentofcryotilt-series. J.Struct.Biol.170,117–126.
Celler, K., Koning, R.I., Willemse, J., Koster, A.J., van Wezel, G.P., 2016. Cross-membranesorchestratecompartmentalization andmorphogenesisin Streptomyces.Nat.Commun.7,ncomms11836.
Chen,M.,Dai,W.,Sun,S.Y.,Jonasch,D.,He,C.Y.,Schmid,M.F.,Chiu,W.,Ludtke, S.J.,2017.Convolutionalneuralnetworksforautomatedannotationofcellular cryo-electrontomograms.Nat.Methods14,983–985.
Dahl,R.,Staehelin,L.A.,1989.High-pressurefreezingforthepreservationof biolog-icalstructure:theoryandpractice.J.ElectronMicrosc.Tech.13,165–174. Dai,W.,Fu,C.,Raytcheva,D.,Flanagan,J.,Khant,H.A.,Liu,X.,Rochat,R.H.,
Haase-Pettingell,C.,Piret,J.,Ludtke,S.J.,Nagayama,K.,Schmid,M.F.,King,J.A.,Chiu, W.,2013.Visualizingvirusassemblyintermediatesinsidemarinecyanobacteria. Nature502,707–710.
Danev,R.,Buijsse,B.,Khoshouei,M.,Plitzko,J.M.,Baumeister,W.,2014.Volta poten-tialphaseplateforin-focusphasecontrasttransmissionelectronmicroscopy. Proc.Natl.Acad.Sci.U.S.A.111,15635–15640.
Danev,R.,Glaeser,R.M.,Nagayama,K.,2009.Practicalfactorsaffectingthe per-formanceofa thin-filmphaseplatefortransmissionelectronmicroscopy. Ultramicroscopy109,312–325.
Danev,R.,Nagayama,K.,2001.TransmissionelectronmicroscopywithZernike phaseplate.Ultramicroscopy88,243–252.
Denk,W.,Horstmann,H.,2004.Serialblock-facescanningelectronmicroscopyto reconstructthree-dimensionaltissuenanostructure.PLoSBiol.2,e329. Diebolder,C.A.,Beurskens,F.J.,deJong,R.N.,Koning,R.I.,Strumane,K.,Lindorfer,
M.A.,Voorhorst,M.,Ugurlar,D.,Rosati,S.,Heck,A.J.,vandeWinkel,J.G.,Wilson, I.A.,Koster,A.J.,Taylor,R.P.,Saphire,E.O.,Burton,D.R.,Schuurman,J.,Gros,P., Parren,P.W.,2014.ComplementisactivatedbyIgGhexamersassembledatthe cellsurface.Science343,1260–1263.
Diebolder,C.A.,Faas,F.G.,Koster,A.J.,Koning,R.I.,2015.ConicalFouriershell corre-lationappliedtoelectrontomograms.J.Struct.Biol.190,215–223.
Dierksen,K.,Typke,D.,Hegerl,R.,Walz,J.,Sackmann,E.,Baumeister,W.,1995. Three-dimensionalstructureoflipidvesiclesembeddedinvitreousiceandinvestigated byautomatedelectrontomography.Biophys.J.68,1416–1422.
Ding,H.J.,Oikonomou,C.M.,Jensen,G.J.,2015.Thecaltechtomographydatabaseand automaticprocessingpipeline.J.Struct.Biol.192,279–286.
Dubochet,J.,Adrian,M.,Chang,J.J.,Homo,J.C.,Lepault,J.,McDowall,A.W.,Schultz, P.,1988.Cryo-electronmicroscopyofvitrifiedspecimens.Q.Rev.Biophys.21, 129–228.
Egerton,R.F.,2011.ElectronEnergy-LossSpectrosopyintheElectronMicroscope. Springer.
Faas,F.G.,Barcena,M.,Agronskaia,A.V.,Gerritsen,H.C.,Moscicka,K.B.,Diebolder, C.A.,vanDriel,L.F.,Limpens,R.W.,Bos,E.,Ravelli,R.B.,Koning,R.I.,Koster,A.J., 2013.Localizationoffluorescentlylabeledstructuresinfrozen-hydrated sam-plesusingintegratedlightelectronmicroscopy.J.Struct.Biol.181,283–290. Feja,B.,Aebi,U.,1999.Determinationoftheinelasticmeanfreepathofelectronsin
vitrifiedicelayersforon-linethicknessmeasurementsbyzero-lossimaging.J. Microsc.193,15–19.
Fernandez,J.J.,Li,S.,Crowther,R.A.,2006.CTFdeterminationandcorrectionin elec-troncryotomography.Ultramicroscopy106,587–596.
Fernandez-Leiro,R.,Scheres,S.H.,2016.Unravellingbiologicalmacromoleculeswith cryo-electronmicroscopy.Nature537,339–346.
Fukuda,Y.,Laugks,U.,Lucic,V.,Baumeister,W.,Danev,R.,2015.Electron cryoto-mographyofvitrifiedcellswithaVoltaphaseplate.J.Struct.Biol.190,143–154. Ghosal,D.,Chang,Y.W.,Jeong,K.C.,Vogel,J.P.,Jensen,G.J.,2017.Insitustructureof theLegionellaDot/IcmtypeIVsecretionsystembyelectroncryotomography. EMBORep.18,726–732.
Gilbert,P.,1972.Iterativemethodsforthethree-dimensionalreconstructionofan objectfromprojections.J.Theor.Biol.36,105–117.
Glaeser,R.M.,1971.Limitationstosignificantinformationinbiologicalelectron microscopyasaresultofradiationdamage.J.Ultrastruct.Res.36,466–482. Grange,M.,Vasishtan,D.,Grunewald,K.,2017.Cellularelectroncryotomography
andinsitusub-volumeaveragingrevealthecontextofmicrotubule-based pro-cesses.J.Struct.Biol.197,181–190.
Grimm,R.,Typke,D.,Barmann,M.,Baumeister,W.,1996.Determinationofthe inelasticmeanfreepathinicebyexaminationoftiltedvesiclesandautomated mostprobablelossimaging.Ultramicroscopy63,169–179.
Guckenberger,R.,1982.Determinationofacommonorigininthemicrographsoftilt seriesinthree-dimensionalelectronmicroscopy.Ultramicroscopy9,167–173. Haider,M.,Hartel,P.,Muller,H.,Uhlemann,S.,Zach,J.,2009.Currentandfuture aberrationcorrectorsfortheimprovementofresolutioninelectronmicroscopy. Philos.Trans.AMath.Phys.Eng.Sci.367,3665–3682.
Hampton,C.M.,Strauss,J.D.,Ke,Z.,Dillard,R.S.,Hammonds,J.E.,Alonas,E.,Desai, T.M.,Marin,M.,Storms,R.E.,Leon,F.,Melikyan,G.B.,Santangelo,P.J.,Spearman, P.W.,Wright,E.R.,2017.Correlatedfluorescencemicroscopyandcryo-electron tomographyofvirus-infectedortransfectedmammaliancells.Nat.Protoc.12, 150–167.
Han,H.M.,Huebinger,J.,Grabenbauer,M.,2012.Self-pressurizedrapidfreezing (SPRF)asasimplefixationmethodforcryo-electronmicroscopyofvitreous sections.J.Struct.Biol.178,84–87.
2013.Cryo-electrontomographyanalysisofmembranevesiclesfrom Acineto-bacterbaumanniiATCC19606T.Res.Microbiol.164,397–405.
Koning,R.I.,Koster,A.J.,2009.Cryo-electrontomographyinbiologyandmedicine. Ann.Anat.191,427–445.
Koster,A.J.,Chen,H.,Sedat,J.W.,Agard,D.A.,1992.Automatedmicroscopyfor elec-trontomography.Ultramicroscopy46,207–227.
Koster,A.J.,Grimm,R.,Typke,D.,Hegerl,R.,Stoschek,A.,Walz,J.,Baumeister,W., 1997.Perspectivesofmolecularandcellularelectrontomography.J.Struct.Biol. 120,276–308.
Krivanek,O.L.,Friedman,S.L.,Gubbens,A.J.,Kraus,B.,1995.Animagingfilterfor biologicalapplications.Ultramicroscopy59,267–282.
Krivanek,O.L.,Mooney,P.E.,1993.Applicationsofslow-scanCCDcamerasin trans-missionelectronmicroscopy.Ultramicroscopy49.
Krueger,R.,1999.Dual-column(FIB–SEM)waferapplications.Micron30,221–226. Kuhlbrandt, W., 2014. Biochemistry. The resolution revolution. Science 343,
1443–1444.
Lanio,S.,Rose,H.,Krahl,D.,1986.Testandimproveddesignofacorrectedimaging magneticenergyfilter.Optik73,56–63.
Lawson,C.L.,Patwardhan,A.,Baker,M.L.,Hryc,C.,Garcia,E.S.,Hudson,B.P., Lagerst-edt,I.,Ludtke,S.J.,Pintilie,G.,Sala,R.,Westbrook,J.D.,Berman,H.M.,Kleywegt, G.J.,Chiu,W.,2016.EMDataBankunifieddataresourcefor3DEM.NucleicAcids Res.44,D396–D403.
Leunissen,J.L.,Yi,H.,2009.Self-pressurizedrapidfreezing(SPRF):anovel cryofix-ationmethodforspecimenpreparationinelectronmicroscopy.J.Microsc.235, 25–35.
Luther,P.K.,Lawrence,M.C.,Crowther,R.A.,1988.Amethodformonitoringthe collapseofplasticsectionsasafunctionofelectrondose.Ultramicroscopy24, 7–18.
Mahamid,J.,Pfeffer,S.,Schaffer,M.,Villa,E.,Danev,R.,Cuellar,L.K.,Forster,F., Hyman,A.A.,Plitzko,J.M.,Baumeister,W.,2016.Visualizingthemolecular soci-ologyattheHeLacellnuclearperiphery.Science351,969–972.
Malac,M.,Beleggia,M.,Kawasaki,M.,Li,P.,Egerton,R.F.,2012.Convenientcontrast enhancementbyahole-freephaseplate.Ultramicroscopy118,77–89. Marabini,R.,Carragher,B.,Chen,S.,Chen,J.,Cheng,A.,Downing,K.H.,Frank,J.,
Gras-succi,R.A.,BernardHeymann,J.,Jiang,W.,Jonic,S.,Liao,H.Y.,Ludtke,S.J.,Patwari, S.,Piotrowski,A.L.,Quintana,A.,Sorzano,C.O.,Stahlberg,H.,Vargas,J.,Voss,N.R., Chiu,W.,Carazo,J.M.,2015.CTFchallenge:resultsummary.J.Struct.Biol.190, 348–359.
Marko,M.,Hsieh,C.,Schalek,R.,Frank,J.,Mannella,C.,2007.Focused-ion-beam thinningoffrozen-hydratedbiologicalspecimensforcryo-electronmicroscopy. Nat.Methods4,215–217.
Mastronarde,D.N.,1997.Dual-axistomography:anapproachwithalignment meth-odsthatpreserveresolution.J.Struct.Biol.120,343–352.
Mastronarde,D.N.,2005.Automatedelectronmicroscopetomographyusingrobust predictionofspecimenmovements.J.Struct.Biol.152,36–51.
Mastronarde,D.N.,Held,S.R.,2017.Automatedtiltseriesalignmentand tomo-graphicreconstructioninIMOD.J.Struct.Biol.197,102–113.
McDonald,K.L.,Auer,M.,2006.High-pressurefreezing,cellulartomography,and structuralcellbiology.Biotechniques41,137,139,141passim.
McMullan,G.,Faruqi,A.R.,Clare,D.,Henderson,R.,2014.Comparisonofoptimal performanceat300keVofthreedirectelectrondetectorsforuseinlowdose electronmicroscopy.Ultramicroscopy147,156–163.
Merk,A.,Bartesaghi,A.,Banerjee,S.,Falconieri,V.,Rao,P.,Davis,M.I.,Pragani,R., Boxer,M.B.,Earl,L.A.,Milne,J.L.,Subramaniam,S.,2016.Breakingcryo-EM res-olutionbarrierstofacilitatedrugdiscovery.Cell165,1698–1707.
Migunov,V.,Ryll,H.,Zhuge,X.,Simson,M.,Struder,L.,Batenburg,K.J.,Houben,L., Dunin-Borkowski,R.E.,2015.Rapidlowdoseelectrontomographyusingadirect electrondetectioncamera.Sci.Rep.5,14516.
Moor,H.,Riehle,U.,1968.Snap-freezingunderhighpressure:anewfixation tech-niqueforfreeze-etching.Proc.FourthEurop.Reg.Conf.Elect.Microsc.2. Nickell,S.,Forster,F.,Linaroudis,A.,Net,W.D.,Beck,F.,Hegerl,R.,Baumeister,W.,
Plitzko,J.M.,2005.TOMsoftwaretoolbox:acquisitionandanalysisforelectron tomography.J.Struct.Biol.149,227–234.
Nickell,S.,Kofler,C.,Leis,A.P.,Baumeister,W.,2006.Avisualapproachtoproteomics. Nat.Rev.Mol.CellBiol.7,225–230.
Noble,A.J.,Stagg,S.M.,2015.Automatedbatchfiducial-lesstilt-seriesalignmentin AppionusingProtomo.J.Struct.Biol.192,270–278.
Palmer,C.M.,Lowe,J.,2014. Acylindricalspecimenholderforelectron cryo-tomography.Ultramicroscopy137,20–29.
ribosomes.Structure20,1508–1518.
Radermacher,M.,1988.Three-dimensionalreconstructionofsingleparticlesfrom randomandnonrandomtiltseries.J.ElectronMicrosc.Tech.9,359–394. Richter,K.,1994.Cuttingartefactsonultrathincryosectionsofbiologicalbulk
spec-imens.Micron25,297–308.
Rigort,A.,Bauerlein,F.J.,Villa,E.,Eibauer,M.,Laugks,T.,Baumeister,W.,Plitzko,J.M., 2012a.Focusedionbeammicromachiningofeukaryoticcellsforcryoelectron tomography.Proc.Natl.Acad.Sci.U.S.A.109,4449–4454.
Rigort,A.,Gunther,D.,Hegerl,R.,Baum,D.,Weber,B.,Prohaska,S.,Medalia,O., Baumeister,W.,Hege,H.C.,2012b.Automatedsegmentationofelectron tomo-gramsforaquantitativedescriptionofactinfilamentnetworks.J.Struct.Biol. 177,135–144.
Rigort,A.,Plitzko,J.M.,2015.Cryo-focused-ion-beamapplicationsinstructural biol-ogy.Arch.Biochem.Biophys.581,122–130.
Rusu,M.,Starosolski,Z.,Wahle,M.,Rigort,A.,Wriggers,W.,2012.Automatedtracing offilamentsin3DelectrontomographyreconstructionsusingSculptorandSitus. J.Struct.Biol.178,121–128.
Schorb, M., Briggs, J.A.,2014. Correlated cryo-fluorescence and cryo-electron microscopywith highspatial precisionand improvedsensitivity. Ultrami-croscopy143,24–32.
Schur,F.K.,Obr,M.,Hagen,W.J.,Wan,W.,Jakobi,A.J.,Kirkpatrick,J.M.,Sachse,C., Krausslich,H.G.,Briggs,J.A.,2016.AnatomicmodelofHIV-1capsid-SP1reveals structuresregulatingassemblyandmaturation.Science353,506–508. Sharp,T.H.,Faas,F.G.,Koster,A.J.,Gros,P.,2017.Imagingcomplementbyphase-plate
cryo-electrontomographyfrominitiationtoporeformation.J.Struct.Biol.197 (2),155–162.
Sharp,T.H.,Koster,A.J.,Gros,P.,2016.HeterogeneousMACinitiatorandpore struc-turesinalipidbilayerbyphase-platecryo-electrontomography.CellRep.15, 1–8.
Sorzano,C.O.,Messaoudi,C.,Eibauer,M.,Bilbao-Castro,J.R.,Hegerl,R.,Nickell,S., Marco,S.,Carazo,J.M.,2009.Marker-freeimageregistrationofelectron tomog-raphytilt-series.BMCBioinform.10,124.
Spence,J.C.H.,Zuo,J.M.,1988.AlargedynamicrangeparalleldetectionCCDsystem forelectrondiffractionandimaging.J.Sci.Instrum.59.
Studer,D.,Graber,W.,Al-Amoudi,A.,Eggli,P.,2001.Anewapproachforcryofixation byhigh-pressurefreezing.J.Microsc.203,285–294.
Studer,D.,Humbel,B.M.,Chiquet,M.,2008.Electronmicroscopyofhighpressure frozensamples:bridgingthegapbetweencellularultrastructureandatomic resolution.Histochem.CellBiol.130,877–889.
Tanaka,M.,Tsuda,K.,Terauchi,M.,Tsuno,K.,Kaneyama,T.,Honda,T.,Ishida,M., 1999.Anew200kVomega-filterelectronmicroscope.J.Microsc.194,219–227. Turonova,B.,Schur,F.K.M.,Wan,W.,Briggs,J.A.G.,2017.Efficient3D-CTFcorrection forcryo-electrontomographyusingNovaCTFimprovessubtomogramaveraging resolutionto3.4A.J.Struct.Biol.199(3),187–195.
Unverdorben,P.,Beck,F.,Sledz,P.,Schweitzer,A.,Pfeifer,G.,Plitzko,J.M.,Baumeister, W.,Forster,F.,2014.Deepclassificationofalargecryo-EMdatasetdefinesthe conformationallandscapeofthe26Sproteasome.Proc.Natl.Acad.Sci.U.S.A. 111,5544–5549.
Vanhecke,D.,Graber,W.,Studer,D.,2008.Close-to-nativeultrastructural preserva-tionbyhighpressurefreezing.MethodsCellBiol.88,151–164.
Verkade,P.,2008.MovingEM:therapidtransfersystemasanewtoolforcorrelative lightandelectronmicroscopyandhighthroughputforhigh-pressurefreezing. J.Microsc.230,317–328.
Vidavsky,N.,Akiva,A.,Kaplan-Ashiri,I.,Rechav,K.,Addadi,L.,Weiner,S., Scher-tel,A.,2016.Cryo-FIB-SEMserialmillingandblockfaceimaging:largevolume structuralanalysisofbiologicaltissuespreservedclosetotheirnativestate.J. Struct.Biol.196,487–495.
Walz,J.,Typke,D.,Nitsch,M.,Koster,A.J.,Hegerl,R.,Baumeister,W.,1997. Elec-trontomographyofsingleice-embeddedmacromolecules:three-dimensional alignmentandclassification.J.Struct.Biol.120,387–395.
Wan,W.,Kolesnikova,L.,Clarke,M.,Koehler,A.,Noda,T.,Becker,S.,Briggs,J.A.G., 2017.StructureandassemblyoftheEbolavirusnucleocapsid.Nature551, 394–397.
Wang,S.,Li,S.,Ji,G.,Huang,X.,Sun,F.,2017.Usingintegratedcorrelativecryo-light andelectronmicroscopytodirectlyobservesyntaphilin-immobilizedneuronal mitochondriainsitu.Biophys.Rep.3,8–16.
Winkler,H.,Taylor,K.A.,2013.Marker-freedual-axistiltseriesalignment.J.Struct. Biol.182,117–124.
Xiong,Q.,Morphew,M.K.,Schwartz,C.L.,Hoenger,A.H.,Mastronarde,D.N.,2009. CTFdeterminationandcorrectionforlowdosetomographictiltseries.J.Struct. Biol.168,378–387.
Xuong,N.H.,Jin,L.,Kleinfelder,S.,Li,S.,Leblanc,P.,Duttweiler,F.,Bouwer,J.C.,Peltier, S.T.,Milazzo,A.C.,Ellisman,M.,2007.Futuredirectionsforcamerasystemsin electronmicroscopy.MethodsCellBiol.79,721–739.
Zheng,Q.S.,Braunfeld,M.B.,Sedat,J.W.,Agard,D.A.,2004.Animprovedstrategyfor automatedelectronmicroscopictomography.J.Struct.Biol.147,91–101. Zheng,S.Q.,Branlund,E.,Kesthelyi,B.,Braunfeld,M.B.,Cheng,Y.,Sedat,J.W.,Agard,
D.A.,2011.Adistributedmulti-GPUsystemforhighspeedelectronmicroscopic tomographicreconstruction.Ultramicroscopy111,1137–1143.
Zheng,S.Q.,Keszthelyi,B.,Branlund,E.,Lyle,J.M.,Braunfeld,M.B.,Sedat,J.W.,Agard, D.A.,2007.UCSFtomography:anintegratedsoftwaresuiteforreal-time elec-tronmicroscopictomographicdatacollection,alignment,andreconstruction.J. Struct.Biol.157,138–147.