Quantitative Microscopy of Fluorescent Adenovirus Entry€¦ · Fluorescent viruses have long been...

Journal of Structural Biology 129, 57–68 (2000)doi:10.1006jsbi.1999.4201, available online at http://www.idealibrary.com on

Quantitative Microscopy of Fluorescent Adenovirus Entry

M. Y. Nakano and U. F. Greber1

Institute of Zoology, University of Zurich, Winterthurerstrasse 190, CH-8057 Zurich, Switzerland

Received August 30, 1999, and in revised form November 9, 1999

fasfqfflbgflrmtifltrtmncvatwds

bnc

cliLa

(nssEicaprpcIfirsnb

o22v1SmpiA9cHplci

6

c4tcpb

Fluorescence imaging of cells is a powerful toolor exploring the dynamics of organelles, proteins,nd viruses. Fluorescent adenoviruses are a modelystem for cargo transport from the cell sur-ace to the nucleus. Here, we describe a procedure touantitate adenovirus-associated fluorescence in dif-erent subcellular regions. CCD camera-captureduorescence sections across entire cells were de-lurred by a fast Fourier transformation, the back-round was subtracted images merged, and virusuorescence quantitated. The validity of the deblur-ing routine was verified by confocal laser scanningicroscopy, demonstrating that objects were nei-

her generated nor deleted. Instead, the homogene-ty of both the average intensity and the size ofuorescent particles was increased, facilitating au-omated quantification. We found that nuclear fluo-escence of wt adenovirus, but not of a virus mutants1, which fails to escape from endosomes, wasaximal at 90 min postinfection (p.i.). Surprisingly,uclear fluorescence decreased at 120 min, but in-reased again at 240 min p.i., suggesting that wtirus targeting to the nucleus may be multiphasicnd regulated. Interestingly, only the first nuclearransport period of wt but not ts1 virus coincidedith a significant increase of the peripheral andecrease of the cytoplasmic regions, indicative ofignal-dependent cell contraction. r 2000 Academic Press

Key Words: adenovirus transport; cytoplasm; de-lurring; fluorescent virus; image processing;ucleus; plasma membrane; quantitative micros-opy; thresholding.

INTRODUCTION

Fluorescence imaging of live and chemically fixedells is an increasingly powerful tool for the subcellu-ar analysis of cell and organelle dynamics, signal-ng, and gene expression (De Giorgi et al., 1999;udin and Matus, 1998; Raz et al., 1998; Tvarusk etl., 1999). Tagging the green fluorescent protein

1 To whom correspondence should be addressed. Fax: 141 1 635
822. [email protected]. T
57

GFP)2 and color variants thereof to a growingumber of proteins combined with sensitive noninva-ive imaging technologies has further broadened thecope of fluorescence cell analysis (Tsien, 1998).xamples of protein dynamics in time and space

nclude the nucleo-cytoplasmic trafficking of thealcium sensor calmodulin (Craske et al., 1999; Li etl., 1999), the transient membrane localization ofrotein kinase B during activation of the B celleceptor in lymphocytes (Astoul et al., 1999) or ofrotein kinase C following local changes in lipidomposition of cell membranes (Oancea et al., 1998).n addition, static analysis of GFP-expressing non-xed cells in the developing wing of drosophila hasecently led to the discovery of a new actin-basedtructure, which is implicated in long-range commu-ication between cells (Ramirez-Weber and Korn-erg, 1999).Fluorescent viruses have long been used as probes

f cell function (for review, see Stidwill and Greber,000). Initially, rhodamine-labeled adenovirus type(Ad) and reovirus particles were used to study

ector attachment to cultured cells (Defer et al.,990; Georgi et al., 1990; Persson et al., 1983).ubsequent virus entry and cell infection experi-ents have indicated that fluorophore-tagged Ad

articles can be prepared without affecting theirnfectivity (Greber et al., 1998; Greber et al., 1997).ds are nonenveloped, icosahedral particles of about0 nm diameter (Burnett, 1997). The main capsidomponent is the facette-associated hexon protein.exon is assembled and stabilized by various minorroteins and largely protects the double-strandedinear DNA genome, which is packed inside theapsid together with additional viral proteins, includ-ng the cysteine protease L3/p23. Hexon remains

2 Abbreviations used: Ad, adenovirus type 2; CCD, charge-oupled device; CLSM, confocal laser scanning microscopy; DAPI,8,68-diamidino-2-phenylindole; DIC, differential interference con-rast; FFT, fast Fourier transformation; FITC, fluorescein isothio-yante; FM, fluorescence microscopy; GFP, green fluorescentrotein; MT, microtubule; p.i., postinfection; PM, plasma mem-rane; r.o.i., region(s) of interest; SEM, standard error of mean;
R, Texas red; ts, temperature sensitive; wt, wild type.
1047-8477/00 $35.00Copyright r 2000 by Academic Press

All rights of reproduction in any form reserved.

ar(ipEidr(b(ecetTpvIevrpvtsr1

afdrerHacflcawmmrsntTfefncw

prtone

1ms((dp

faNcHmmNrpDT9(

Asd9Dw3pas(A(mT

Ic7vwgwfl(wswdFfoa

58 NAKANO AND GREBER

ssociated with the viral DNA until the particleseach the nucleus of a target cell and dissociateGreber et al., 1996). The base of the capsid vertecess built of the penton and the protruding fibers andrimarily functions during virus entry (Greber, 1998).ntry occurs by a dual receptor mechanism involv-

ng high affinity attachment of the terminal knobomain of the fiber protein to the immunoglobulin-elated cell surface receptor coxsackie–Ad receptorsee Bergelson, 1999), followed by contacts of pentonase with fibronectin/vitronectin binding integrinsWickham et al., 1993). An acid-assisted endosomescape mechanism releases virus particles into theytosol (Greber, 1998). In addition, virus escape fromndosomes requires a fully processed capsid struc-ure lacking in ts1 (Miles et al., 1980; Weber, 1976).s1 has a point mutation in the L3/p23 protease genereventing protease packaging and processing of theiral capsid structure (for review, see Greber, 1998).ncoming ts1 virions are confined to endosomes andventually degraded in lysosomes. Like cytosolic wtirus, ts1 is transported to and from the perinuclearegion by microtubule-dependent motors, as re-orted in live cell experiments using fluorescentiruses (Suomalainen et al., 1999). wt Ad but not ts1hen associates with the nuclear envelope, where itpecifically attaches to nuclear pore complexes andeleases its DNA into the nucleoplasm (Greber et al.,996, 1997).Fluorescent virus particles are not only useful for

nalyzing the cell biology of virus entry, but also foracilitating analysis of vector tropism, which is oftenetermined by the availability and localization ofeceptors on the cell surface (Walters et al., 1999). Toffectively harness the potential of fluorescent vi-uses, quantitative procedures are indispensible.ere, we describe a protocol to quantitate Ad-

ssociated fluorescence in different regions of fixedells at various times postinfection (p.i.). Stacks ofuorescent microscopy images of infected cells wereaptured with a CCD camera, deblurred, and mergednd the fluorescence was quantitated in r.o.i., whichere generated by thresholding of a fluorescencearker image. Our results show that both wt Ad andutant ts1 progressively enriched in the perinuclear

egion, albeit with different kinetics. While the endo-omal ts1 virus reached quasi-steady state levelsear the nucleus at 30 min p.i. cytosolic wt virus wasransported to the nucleus in two subsequent waves.he first wave lasted until 90 min p.i. and was

ollowed by a period of virus transport to the periph-ry, followed by the second nuclear transport periodrom 120 to 240 min p.i. Surprisingly, the first, butot the second nuclear transport period of wt Adoincided with significant cell contraction, which
as reversed at 120 min p.i. These results raise the q
ossibility that nuclear transport of wt Ad may beegulated by virus-induced cell signaling. We expecthat the protocol described here is applicable to anyther fluorescent object of regular shape, includingaked and lipid-enveloped viruses and cell organ-lles.

MATERIALS AND METHODS

Cells and viruses. HeLa, A549, and TC7 cells were grown on2-mm glass coverslips (0.13–0.16 mm thickness, Assistant, Ger-any) in DMEM–F12 medium in the presence of 7% fetal bovine

erum and 5% CO2 and infected with purified Ad as describedSuomalainen et al., 1999). Labeling of isolated Ad with Texas redTR) and quantification of fluorophore incorporation into SDS-issociated Ad proteins by SDS–PAGE and phosphoimaging wereerformed as described (Greber et al., 1998).Microinjections. Purified Ad–TR (0.8 mg/ml) was diluted 20-

old into 0.01 M Tris–HCl, pH 7.4, 0.12 M KCl containing 1 mg/mlldehyde-fixable 10-kDa Dextran–FITC (Molecular Probes, LeidenE), centrifuged at 10 000g for 5 min, and microinjected into the

ytoplasm of HeLa cells in RPMI medium containing 20 mMepes–NaOH, pH 7.4, and 0.2% BSA, using the Eppendorficroinjection system at constant ejection pressure (injector 5242,anipulator 5120). Microinjection needles were pulled by aarishige needle puller (model PB-7, Nikon, Switzerland). To

educe virus uptake by the natural pathway, HeLa cells wereretreated with 0.4 µg/ml of recombinant Ad fiber knob inMEM–F-12 medium containing 0.2% BSA for 30 min at 37°C.his procedure effectively decreased virus uptake by more than0%, as indicated by binding studies with radioactively labeled Addata not shown).

Virus uptake and cell staining. Briefly, 0.16 µg of purifiedd–TR (approximately 1.4 3 108 particles) were attached to theurface of 7 3 104 cells in cold RPMI–BSA medium for 60 min asescribed (Suomalainen et al., 1999). Unbound virus (about 90 to5% of the input virus) was washed off and cells were incubated inMEM–BSA medium at 37°C for indicated amounts of time. Cellsere quickly washed in PBS at 37°C and fixed immediately in.3% pFA in PBS for 12 min at room temperature. Cells wereermeabilized with 0.5% Triton X-100 in PBS for 2 min to reduceutofluorescence, quenched in PBS containing 25 mM NH4Cl, andtained with an anti-plasma membrane Ca21 ATPase 1 antiserumStauffer et al., 1995) and goat anti-rabbit IgG conjugated tolexa488 (Molecular Probes). The nucleus was stained with DAPI

0.5 µg/ml, Sigma, Switzerland) for 5 min and the sample wasounted onto a glass slide in 85% glycerol containing 20 mM

ris–HCl, pH 8.8, and 2 mg/ml para-phenylenediamine.Analysis of isolated virus by fluorescence microscopy (FM).

solated Ad–TR was diluted to 8 µg/ml into Hanks’ mediumontaining 2 mg/ml BSA to yield a virus particle density of about.2 3 106 per µl (see Greber et al., 1998). Twelve microliters ofirus solution was placed on a microscope slide and furnishedith a coverslip (18 3 18 mm), such that the spacing between thelass surfaces was approximately 35 µm. The coverslip was sealedith nail polish and virus particles were analyzed immediately byuorescence microscopy using a 1003 oil immersion objective

Leica, Switzerland) and the CCD camera specified below. Imagesere recorded in five subsequent planes at 0.5-µm intervals

tarting from the bottom of the coverslip. Bottom and top imagesere discarded and the remaining three images either mergedirectly to a new raw image or were individually processed by fastourier transformation (FFT) and then merged using the sum

unction of the MetaMorph software (see below). Fluorescentbjects were further selected by a thresholding function andnalyzed in regions of 200 3 250 pixels (30 3 37.5 µm) by
uantitation of pixel intensities and object sizes.

owiompaoO(edds1etssCatuRdfflnFssittFFea

adaccsflhwoaadaorett

fflwtdu

F

mcFPmwavfflpAtichaaptt

cpfl(aooti(jannaai1ohgimsotlo

59QUANTITATIVE FLUORESCENCE MICROSCOPY FOR VIRUS ENTRY

Image acquisition and processing. Specimens were analyzedn an inverted motorized Leica microscope (DMIRBE) equippedith a plan 1003 objective (N.A. 1.3 PL Fluotar) and a back-

lluminated MicroMax-controlled CCD camera (800 3 1000 pixelsf 15 3 15 µm, Princeton Instruments, Visitron GmbH, Ger-any), which was operated in 12-bit mode at 225°C. All acquired

ixel values were below saturation. Samples were illuminated by100-W HBO lamp (Leica) at full power for 400 ms. Excitation

ccurred through bandpass excitation filters for DAPI (330/80,mega XF03); Alexa 488 (475/40, Omega XF100), and Texas red

560/55, Omega XF101) and emitted light was collected throughither a single pass (DAPI, 425LP, dichroic 400, Omega XF03) or aouble pass emission filter (green and red, 528/30 and 633/60,ichroic 490-575, Omega XF53). The TR images were recordedequentially throughout the entire cell sample (z-stack) at 0.5- or-µm steps as indicated.Alexa 488, DAPI, and differential interfer-nce contrast (DIC) images were obtained from sections towardhe cell bottom or the middle to indicate cell and nuclear dimen-ions, respectively. Each TR image was FFT processed using theettings available in the MetaMorph package (Universal Imagingorp., Visitron, GmbH, Germany). The FFT procedure expressesspatially varying image brightness (represented as pixel intensi-

ies) as a summation of sine and cosine waves and is frequentlysed for image enhancement and measurement (Bracewell, 1989;uss, 1995). Modulations of the wave frequencies were used toetermine a continuous background image which was subtractedrom the raw image to deblur a defined signal, such as auorescent virus particle. We used a deblur value of 70%, whicheither removed nor added any discrete fluorescent particles (seeig. 2). The FFT-processed images were then merged using theummation function and the background was subtracted byetting the lowest pixel value to 0. Pixel values were determinedn either r.o.i. or objects of interest, which were defined by ahresholding function (see below). The validity of the FFT andhresholding parameters was tested by direct comparison of theM-processed images with the image acquired by CLSM (seeig. 3). CLSM was performed on a Leica DMIRBE microscopequipped with a 633 lens (N.A. 1.4), using step intervals of 0.1 µmnd nonsaturating recording conditions.Calculations of regions of interest. For quantification of virus-

ssociated fluorescence in subcellular r.o.i., the nucleus wasefined by DAPI staining and the plasma membrane by annti-Ca21 ATPase 1 antiserum (Stauffer et al., 1995). Raw fluores-ent images of the nucleus and the plasma membrane wereollected from regions near the bottom of the cell and binarizeduch that the resulting images perfectly matched the originaluorescence. The peripheral region of the cell was defined byomogeneously decreasing the binary plasma membrane imageith a 3 3 3 pixel kernel unsharp routine, which reduced the

bject by 1 pixel (MetaMorph software package). Repetitivepplication of this routine yielded a peripheral region with anverage band width of 3.5 µm. A similar procedure was used toefine a perinuclear region of 1.8 µm extending to the cytoplasmnd the nucleus by 0.9 µm. Cytoplasmic and nuclear regions werebtained by linear subtractions as defined above. Nondefinedegions, where the perinuclear region overlapped with the periph-ral region, were below 2% in every sample. Virus fluorescence inhis region was always below 5% of the total fluorescence andherefore not included in the analysis.

Statistics. Twenty randomly selected cells of each time pointrom at least three independent experiments were subjected to theuorescence quantification procedure described above. Resultsere expressed as mean values of indicated samples (n), including

he standard error of the mean (SEM). Significances of theifferences were calculated from one sided t tests at the indicated
ncertainty levels (P). a
RESULTS

luorescence Labeling of Isolated Ad and Analysisby Fluorescence Microscopy

We have prepared Texas red-labeled wt and ts1utant Ad following a protocol described for fluores-

ein and TR labeling of wt Ad (Greber et al., 1998).luorography and Coomassie blue staining of SDS–AGE-fractionated virus particles indicated thatore than 90% of the TR fluorescence was associatedith hexon in both wt and ts1 particles (Fig. 1). Onlyminor fraction of TR was associated with other

irion proteins. This result was markedly differentrom fluorescein-modified wt Ad, where 50% of theuorescein was incorporated into external capsidroteins other than hexon (Greber et al., 1998). Wtd contained on average 2000 TR molecules (about

wice as much as TR-labeled ts1) and was fullynfectious, as determined by plaque assays on A549ells (not shown). Earlier studies have shown thatexon of both wt and ts1 particles remains associ-ted with the viral genome until the particles haverrived at the cytosolic side of nuclear pore com-lexes (Greber, 1998; Suomalainen et al., 1999). Wehus conclude that these TR-labeled virus prepara-ions can be used to trace subcellular virus particles.

Isolated wt Ad particles were analyzed by fluores-ence microscopy. Ad–TR (8 µg/ml, 7.2 3 106/µl) waslaced between two glass surfaces for analysis byuorescence microscopy with a cooled CCD camera

15- 3 15-µm pixel size) as described under Materi-ls and Methods. Three successive z-stacked imagesf 200 3 250 pixels were captured through a 1003bjective lens at 0.5-µm intervals near the bottom ofhe sample and merged together to a combined rawmage (Fig. 2A). As shown in the overview pictureinset), the image contained about 28 discrete ob-ects, each consisting of 10 to 20 pixels. The imagelso contained a significant amount of blurred sig-als, probably due to out-of-focus information andonuniform illumination. Based on size, intensity,nd frequency, the fluorescent objects resembleduthentic virus particles, which had been analyzedn live virus entry experiments (Suomalainen et al.,999). We therefore subjected the individual imagesf the z-stack to fast Fourier transformation. FFTad previously proven useful in removing nonhomo-eneous background, which was due to nonuniformllumination, autofluorescence, or out-of-focus infor-

ation (Russ, 1995). The parameters were chosenuch that none of the objects were removed and nobject was added. The FFT-processed images werehen merged. The resulting image had significantlyess background information (Fig. 2B) and the pixelsf its objects were homogeneous, as indicated by the
verage pixel intensity of the FFT-processed image

(Tginopvs(lw

hsgtaotstga

trVl


9.36 6 0.74) compared to the raw image (707 6 130).o select for fluorescent objects and quantitate theirray levels and dimensions, we applied a threshold-ng function under conditions which neither addedor deleted any of these objects. The processedbjects (indicated as a binary image in Fig. 2C) wereredominantly homogeneous in terms of both grayalues and sizes, as indicated by a frequency analy-is of 13 different images containing 581 objectsFig. 2D). More than 85% of the objects had grayevels less than 40% of the maximal object value,

FIG. 1. SDS–PAGE analysis of TR-labeled wt and ts1 Ad. Apps1 Ad was separated in a 4–12% acrylamide gradient SDS gel andespectively (lanes 4 and 5), as described (Greber et al., 1998). Abouirion proteins are indicated on the right and relative molecular w

adder).

hich was 7736, and more than 80% of the objects m

ad sizes between 30 and 60% of the maximal objectize, which was 24 pixels. The average object had aray level of 2267 and a size of 11.35 pixels, respec-ively. The latter is equivalent to 0.25 µm2, which isbout 16 times larger than a mature virus particlebserved by electron microscopy. We concluded thathe fluorescent objects were single virus particles,ince they were of a quasi-regular appearance withhe highest fluorescence in the center and decreasingray values toward the periphery in both the rawnd the processed images (Figs. 2A and 2B). Further-

ately 10 µg of trichloroacetic acid-precipitated TR-labeled wt andzed by fluorography (lanes 1 and 2) and Coomassie blue staining,of nonlabeled, isolated Ad was loaded as a control (lanes 3 and 6).

of marker proteins on the left (Gibco-BRL, BENCHMARK protein

roximanaly

t 20 µgeights

ore, these objects were normally distributed in

totpa

µutgttch

F

tplacctdt1otl3aacdirc1tTds(ptit5Fjntedofldttos

FmsmRtoflfmm


erms of size and gray levels (Fig. 2D). The averagebject number per image was 44.7. This was within awofold range of the theoretical number of virusarticles per image, which was calculated to be 24.3,

FIG. 2. Population analysis of isolated wt Ad-TR particles byM and image processing. Isolated Ad–TR was placed between aicroscope slide and a glass coverslip and imaged by FM in three

ubsequent planes of focus at 0.5-µm intervals. Images wereerged using the summation function yielding the raw image (A).aw images were FFT-processed (B) or FFT-processed and further

hresholded giving a binarized image (C), which was used to selectbjects (virus particles). Bar equals 0.75 µm. (D) The number ofuorescent objects in the FFT-processed images plotted as a

unction of pixel intensities and apparent object size (inset). Theaximal value of all the pixels in an object was 7736 and theaximal object size was 24 pixels.

ssuming that the merged optical sections were about 3 t

m thick and the virus was homogeneously distrib-ted between the glass plates. Possibly, virus par-icles were somewhat enriched at the surface of thelass coverslip where the sampling occurred. Together,his population analysis shows that the FFT and thehresholding procedures do not delete objects but effi-iently remove background signals and thus generate aomogenous population of fluorescent virus particles.

M and CLSM Analysis of Microinjected Ad–TR

To test if the applied FFT and thresholding opera-ions were valid for an analysis of fluorescent virusarticles within target cells, we microinjected iso-ated TR-labeled Ad into the cytosol of HeLa cellsnd analyzed the TR signal by fluorescence micros-opy in the fixed cells 5 min postinjection. The sameells were also imaged by CLSM in order to assesshe reliability of the imaging and processing proce-ures (Fig. 3). Prior to microinjection, cells werereated with recombinant fiber knob (Louis et al.,994), which blocks the natural infectious pathwayf Ad (Bergelson, 1999). To facilitate the identifica-ion of injected cells, we coinjected fluorescein-abeled 10-kDa Dextran together with Ad–TR (Fig.A, e). Two of the injected cells are indicated byrrows and one noninjected cell is pointed out by anrrowhead in the DIC image (Fig. 3A, d). TR fluores-ence was recorded across the entire z-dimension asefined by dextran–FITC fluorescence using stepntervals of 0.5 µm for FM and 0.1 µm for CLSM,espectively. Since the actual diameter of the virusapsid is approximately 100 nm (Stewart et al.,993), the CLSM procedure most likely captured allhe fluorescent virus particles present in the cell.he images of an entire z-stack were merged andisplayed directly (Fig. 3A, a and f) or background-ubtracted (Fig. 3A, b) and further processed by FFTFig. 3A, c). The distribution and number of virusarticles were very similar in all cases, as seen whenhe thresholded fluorescent objects in one of thenjected cells are counted (large arrow, Fig. 3A, d). Inhe raw and FFT-processed CLSM images, we found5 and 58 virus particles, respectively, whereas theFT-processed FM image contained 53 discrete ob-

ects (Fig. 3A, a, c, and f). The difference in theumber of objects was most likely due to particleshat were close together and, thus, not resolved byither the microscopy or the image-processing proce-ures. No discrete objects were, however, generatedr deleted, and both the pixel values and the sizes ofuorescent particles had a normal distribution un-er all the conditions (Figs. 3B and 3C). Similar tohe processing of extracellular virus particles (Fig. 2),he FFT procedure yielded a homogenous populationf intracellular virus particles with respect to inten-ities and sizes. In the FM-generated images, more
han 80% of the particles had gray levels and sizes

gicw(3cpiF , and 8080 and 11 in the case of the CLSM–FFT image.


FIG. 3. Verification of the FFT and thresholding procedureslass coverslips into the cytoplasm with Ad–TR (40 ng/ml) containnjection. A set of three cells was analyzed for TR and FITC fluoresell were identified by FITC fluorescence (e), indicated by arrows aas generated by merging TR images recorded across the entire

c). The raw image of merged CLSM images of the entire z-stack is-D projections from a tilted image, including a calibration bar: 0/582; f, 0/1252. The number of fluorescent objects of one of the inixel intensities in the case of the FFT-processed FM image and thndicated in C. The maximal sum of all the pixel values in an oM–FFT image, 20 890 and 17 in the case of the CLSM raw image

by CLSM with microinjected Ad–TR. HeLa cells were microinjected oning 10 kDa lysine-fixable Dextran–FITC (1 mg/ml) and fixed 5 min postcence by both FM and CLSM (A). Two injected cells and one noninjectednd an arrowhead in the DIC image, respectively (d). The raw FM imagez-stack (a), background-subtracted (b), and further processed by FFTindicated in f. All the TR fluorescence images (a–c, f) are represented as

with the following pixel minima and maxima: a, 581/2008; b, 33/1088;jected cells (indicated by the large arrow in d was plotted as a function ofe raw and FFT-processed CLSM images (B). The apparent object size is

bject and the maximal object size were 2751 and 27 in the case of the

bWifip

T

trcA6swfiaDtt

Utcni(pntVeflsa

pate(n

oflsw


elow 20% of the maximal values (Figs. 3B and 3C).e conclude that FFT processing of fluorescence

mages of virus particles across the entire z-stack ofxed cells generates a homogenous population of virusarticles with respect to gray values and dimensions.

wo Waves of Nuclear Transport of Incomingwt Ad in HeLa Cells

We therefore used FFT-processed FM to determinehe subcellular distribution of Ad-associated TR fluo-escence in natural infections of human and monkeyells. Ad–TR, and as a parallel control mutant ts1d–TR, was bound to the cell surface in the cold for0 min and internalized for different times as de-cribed earlier (Greber et al., 1997). Defined r.o.i.ere indicated by a plasma membrane staining ofxed cells with an anti-PM Ca21 ATPase antibodynd anti-rabbit-Alexa 488 and the nuclear stainAPI. The plasma membrane stain perfectly matched

he outline of the cells imaged by Nomarski differen-ial interference contrast optics (data not shown).

FIG. 4. Analysis of incoming Ad–TR by FM and FFT processingr cold-bound and internalized for 90 min (right), followed by FMuorescence is shown in red. The plasma membrane is visualizecondary antibody (green) and the nucleus is indicated by DAPI s
e defined r.o.i. (B) where virus-associated fluorescence was quantitated
sing a thresholding function (described under Ma-erials and Methods), we operationally defined theell periphery as a 3.5-µm-wide ribbon and theuclear periphery as an 1.8-µm-wide region extend-

ng 0.9 µm from both sides of the nuclear peripheryFig. 4B). The cytoplasm was defined by the totallasma membrane region minus nuclear and peri-uclear regions. TR fluorescence was imaged acrosshe entire z-stack as shown in Fig. 4A for wt Ad.irus fluorescence in a given region was eitherxpressed relative to the total cell-associated virusuorescence (Figs. 5A and 5D) or expressed aspecific fluorescence intensity per region (Figs. 5Bnd 5E).At 0 min of entry (when the virus particles are

resent at the cell surface) the specific peripheralnd cytoplasmic fluorescence levels of both wt ands1 Ad were approximately equal and only slightlylevated in the perinuclear and nuclear regionsFigs. 5B and 5E). This was possibly due to aonplanar topology of the plasma membrane proxi-

R was bound to the surface of HeLa cells in the cold (0 min, A, left)age processing as described under Materials and Methods. Virusan anti-PM Ca21 ATPase 1 antibody and anti-rabbit-Alexa 488g (blue). Based on the plasma membrane and the nuclear regions,

. Ad–Tand imed bytainin

(see Fig. 5).

bnwfiw4


FIG. 5. Quantification of incoming Ad fluorescence in defined cellular regions of interest. Wt Ad–TR (A–C) or ts1 Ad-TR (D–F) wasound to the surface of HeLa cells and internalized for indicated times. Fluorescence images of virus particles, plasma membrane, anducleus were acquired and processed as indicated in Fig. 4. Virus-associated fluorescence was determined in r.o.i. and the mean valuesere expressed as fractions of total cell-associated virus fluorescence (A, D), including SEM (bars). In addition, r.o.i. were plotted as a

unction of infection time (C, F) and virus-associated fluorescence was normalized to the r.o.i. (B, E). The number of cells analyzed isndicated in C and F. The mean gray values of wt Ad–TR fluorescence across an entire cell varied from 244 260 to 2 430 968 and the cell sizeas between 53 927 and 70 215 pixels. In the case of ts1 Ad–TR the fluorescence varied from 104 362 to 179 917 and the cell size from
6 523 to 58 140 pixels. n.d. (not determined) indicates regions where the cytoplasmic and the perinuclear regions overlapped.

mna5ssdmmAwM3m6rtnas(ban6cwipl(swe(

rpccgdArtCtmfdiani(

ntn0tmAsmt

trebqvihrptTl3(o1thcibdohnFcosqaacpwc

tgpa


al to the nucleus. At 30 min postwarming, ts1 butot wt virus was removed from the cell peripherynd enriched in the perinuclear region (Figs. 5B andE). At 60 and 90 min p.i., wt Ad levels wereignificantly reduced at the cell periphery andtrongly increased near the nucleus, whereas the ts1istribution remained unchanged compared to 30in p.i. These results were in agreement with virusotility data from live cells, demonstrating that wtd was transported predominantly to the MTOC,hereas ts1 was moving to and away from theTOC at similar frequencies and efficacies between

0 to 90 min p.i. (Suomalainen et al., 1999). At 90in p.i., the nuclear levels of wt Ad reached about

4% of the total cell-associated virus fluorescence,epresenting a more than twofold increase comparedo 0 min p.i. (Fig. 5A). At 120 min p.i., however, theuclear fluorescence of wt Ad was reduced to 51%,nd the cytoplasmic and peripheral levels wereignificantly elevated compared to 90 min p.i.P 5 0.05). These results suggested a global redistri-ution of wt Ad toward the cell periphery between 90nd 120 min p.i. This outward flux of virus wasearly reversed at 240 min, resembling the status at0 min p.i. (P 5 0.01). As expected, virus fluores-ence in the cell periphery and cytoplasmic regionere significantly reduced at 90 min of wt Ad

nfection both in terms of absolute numbers and on aer-region basis (P 5 0.01). Likewise, the peripheralevels of the endosomal ts1 virus declined rapidlywithin 30 min) and the nuclear levels increasedomewhat compared to 0 min p.i. Similar resultsere obtained with two other cell lines, humanpitheloid A549 cells and monkey TC7 fibroblastsdata not shown).

Infection of human cells with wt Ad has beeneported to trigger actin rearrangements that are inart mediated by the Rho family GTPases rac anddc42 (Li et al., 1998), known regulators of the actinytoskeleton (Tapon and Hall, 1997). We investi-ated whether or not any of the r.o.i. were altereduring Ad infection. As indicated in Fig. 5C, the wtd infection significantly decreased the cytoplasmicegion from 40% of the total cell region at 0 min p.i.o 35–37% at 30–90 min p.i. (P 5 0.01, Fig. 5C).oncomitantly, the cell periphery increased from 36

o 39–40% (P 5 0.01).After 120 min p.i. both cytoplas-ic and peripheral regions were indistinguishable

rom the corresponding regions at 0 min p.i. The dataemonstrate that wt Ad-infected cells, but not ts1-nfected cells, have a tendency to contract between 0nd 90 min p.i. This effect is specific to wt Ad and isot observed with mutant ts1-infected cells, indicat-

ng that the experimental procedures are adequateFig. 5F).

Altogether, our results reveal two surprising and p

ot-yet described differences between the wt and thes1 Ad infection. First, nuclear transport of wt butot ts1 virus occurrs in two distinct waves, one fromto 90 min, followed by a period of virus transport

oward the cell periphery, and another at 120 to 240in p.i. The second surprising result is that the wtd infection significantly reduces the apparent cellize between 0 and 90 min p.i. Both of these resultsay reflect a different signaling potential of wt and

s1 Ad.

DISCUSSION

Here we presented a fluorescence microscopy pro-ocol to quantitate fluorescently tagged Ad in cellular.o.i. at different times of infection. We have utilizedlectronically controlled fluorescence microscopy com-ined with a back-illuminated CCD camera to ac-uire complete sets of optical sections of fluorescentirus particles through infected cells. Individualmages were processed by FFT and merged to give aomogenous representation of virus-associated fluo-escence without generating or deleting any virusarticles. This was confirmed by CLSM analysis ofhe same sample in the absence of image processing.he signal of a single virus particle is clearly in the

ower end of the intensity spectrum, reaching 60 to00 pixel units in the case of the 12-bit CCD image4096 gray levels) and 3 to 20 pixel units in the casef the 8-bit CLSM image (256 gray levels). Using a2-bit (or higher) CCD camera for signal detection isherefore preferable to an 8-bit CLSM chip, since theigher dynamic range of the CCD image signifi-antly increases the resolution of neighboring lowntensity pixels and thus enhances the accuracy ofackground subtraction and quantification proce-ures. Detection and amplification of low-intensitybjects by photomultipliers used in CLSM also en-ances background signals, including electronicoise. This problem is documented in Fig. 3, whereM is directly compared to CLSM. The FM image isonsiderably smoother than the CLSM image. More-ver, quantification of low-intensity CLSM objects,uch as fluorescent virus particles, will always re-uire image processing in order to remove cellularutofluorescence (see, for example, Fig. 3A, f). Last,n FM configuration including a sensitive CCDamera is clearly less expensive than a CLSM,articularly if the latter is operated at multipleavelengths from different laser units or with a

ooled photomultiplier to minimize electronic noise.The procedure described here reliably accounts for

he sum of all virus-associated fluorescence across aiven cell without the necessity to count individualarticles, which is particularly difficult when virusesccumulate, for example, near the nucleus. We ex-
ect that this protocol complements live cell analy-

ssapIcsmiiahapsaiflpsac

fwpfrpsptpdkw1fpeaw1dtiqrieetsmbu

tmwttbpnr

eswud1pnmApupsmptgcTv(

swdAaa1aw(Am(tsdcar

pc


is, which is typically conducted in single opticalections of flat cell regions, and will provide anlternative to traditional biochemical fractionationrocedures and quantitative electron microscopy.mportantly, this protocol is not limited to fluores-ence microscopy but can be readily applied to deblurtacks of CLSM images, as indicated in Fig. 3. Weust, however, emphasize that the image-process-

ng procedure described here is not applicable torregularly shaped objects, since we do not take intoccount the optical properties of the microscope andave not corrected the shape of the objects with anppropriate point-spread function. Consequently, therocedure does not reflect the real fluorescence inten-ities of the objects. However, the procedure doesccurately reproduce the relative fluorescence levelsn subcellular regions, since it corrects for localuorescence quenching by generating a homogenousopulation of virus objects. We believe that theimplicity of the present protocol to quantify virus-ssociated fluorescence within a cell will be appli-able to various aspects of quantitative cell biology.In this study we also described the preparation of

ully infectious Texas red-labeled virus particleshich are exclusively modified in the major coatrotein hexon. Hexon-associated fluorescence is idealor tracing incoming virus particles, since hexonemains associated with the viral genome until thearticles reach the nuclear membrane and disas-emble at nuclear pore complexes after about 90 min.i. Here we have concentrated on the analysis of twoypes of viruses, wt and mutant ts1 Ad, in theeripheral and the nuclear regions of HeLa cells atifferent times of infection. Both wt and ts1 Ad arenown to enter in an integrin-dependent mannerith identical kinetics and efficacies (Greber et al.,996; Wickham et al., 1993). Unlike ts1 virus, whichails to escape from endosomes, wt Ad effectivelyenetrates the endosomal membrane shortly afterndocytosis and is transported as a naked particle in

microtubule- and dynein-dependent manner to-ard the nuclear membrane (Suomalainen et al.,999). Using appropriate fluorescent markers forefining the plasma membrane and the nucleus andranslating the corresponding images into binarymages has allowed us to automate a nonbiaseduantification routine of virus fluorescence in the.o.i. Our results show for the first time that wt viruss slowly cleared from the cell periphery, whereas thendosomal ts1 is rapidly transported from the periph-ry to a perinuclear compartment. The data suggesthat wt Ad may escape from peripheral early endo-omes consistent with a mildly acidic pH require-ent for membrane rupture. Cytosolic wt virus may

e confined to a peripheral region until it is able to
tilize the microtubule-based minus-end-directed s
ransport machinery. Once virus engages with thisachinery, it is effectively transported to the nucleus,here it is enriched up to 90 min p.i. Ts1 virus, on

he other hand, initially moves in endosomes towardhe nucleus without delay, presumably in a microtu-ule-dependent manner, and later, at 30 to 90 min.i., travels to the cell periphery and the nucleus atearly equal efficiencies, in agreement with earlieresults (Suomalainen et al., 1999).Our results reveal two novel features of virus

ntry, a biphasic nuclear transport of wt Ad and aignificant cell contraction during the early phase oft Ad entry. The first nuclear transport phase lastsntil 90 min p.i. and is followed by periphery-irected transport of virus fluorescence from 90 to20 min. p.i. Whether this transport representsarticles that are broken up or not stably bound touclear pore complexes is unknown. From 120 to 240in p.i., a second nucleus-directed transport wave ofd occurs, leading to a reduction of peripheral virusarticles and an enrichment of nuclear particles. Thenderlying mechanism for this biphasic virus trans-ort is not known. Conceivably, virus-induced cellignaling affects either the attachment or detach-ent of actin or MT-dependent motors from cargo or,

erhaps less likely, regulates motor efficacy. Alterna-ively, MT-associated proteins (MAPs) may be tar-ets for Ad signaling and thus regulate the frequen-ies of virus attachment and detachment to MTs.his principle has recently been demonstrated foresicular transport regulated by the axonal MAP tauTrinczek et al., 1999).

The second unexpected result of our analysis is aignificant cell contraction within the first 90 min oft Ad infection. Since this effect is transient andoes not occur with ts1 Ad, it might be related to wtd-induced cell signaling. Whether the small GTP-ses rac and cdc42, which regulate cell spreadingnd have been implicated in Ad endocytosis (Li et al.,998), are involved in cell contraction can now benalyzed. It will be interesting to test if integrins,hich are upstream of rac in the Ad entry pathway

Li et al., 1998) are involved in cell contraction.lternatively, the transient induction of the raf/itogen-activated kinase pathway by incoming Ad

Bruder and Kovesdi, 1997) could promote cell migra-ion and, perhaps, interfere with cell contraction andpreading, analogously to the plasminogen-depen-ent activation of the ras pathway in fibrosarcomaells (Nguyen et al., 1999). Future experiments willddress the nature and mechanisms of early cellesponses to Ad infections.With the discovery of an increasing number of

rimary and secondary receptors for viruses, fluores-ent virus particles are expected to become useful for
tudies of virus uptake, delivery to the replication

stfpssdirv

tlZaififCvwa

A

B

B

B

B

C

D

D

G

G

G

G

G

G

L

L

L

L

L

M

N

O

P

R

R

R

S

S

S

S


ite, and also uncoating of the viral genome. For geneherapists, fluorescent virus vectors are importantor evaluating vector binding to cells ex vivo anderhaps even facilitate an in vivo analysis of vectorpread and clearance from different tissues or theite of administration. The latter process is often illefined and the responsible cells are poorly character-zed. Quantitative procedures to measure virus fluo-escence may thus improve a rationale design ofectors for gene-based therapies.

We thank R. Stidwill for expert help with CLSM data acquisi-ion and discussions, S. Keller for fluorography of Texas red-abeled adenovirus, and T. Bachi (Elektronenmikroskopischesentrallabor, University of Zurich) for access to the CLSM. We arelso grateful to J. Chroboczek for recombinant baculovirus express-ng Ad fiber knob, M. Suomalainen and L. Trotman for purifiedber knob, D. Guerini (Swiss Federal Technical University, Zurich)or the generous gift of affinity-purified anti-plasma membranea21 ATPase 1 antibody, and U. Ziegler (University of Zurich) andarious members of the laboratory for helpful discussions. Thisork was supported by the Swiss National Science Foundationnd the Kanton of Zurich.

REFERENCES

stoul, E., Watton, S., and Cantrell, D. (1999) The dynamics ofprotein kinase B regulation during B cell antigen receptorengagement, J. Cell. Biol. 145, 1511–1520.

ergelson, J. M. (1999) Receptors mediating adenovirus attach-ment and internalization, Biochem. Pharmacol. 57, 975–979.racewell, R. N. (1989) The Fourier transform, Sci. Am. 260,86–89, 92–95.

ruder, J. T., and Kovesdi, I. (1997) Adenovirus infection stimu-lates the Raf/Mapk signaling pathway and induces interleu-kin-8 expression, J. Virol. 71, 398–404.

urnett, R. M. (1997) The structure of adenovirus, in Chiu, W.,Burnett, R. M., and Garcea, R. L. (Eds.), Structural Biology ofViruses, pp. 209–238. Oxford Press, Oxford.raske, H., Takeo, T., Gerasimenko, O., Vaillant, C., Torok, K.,Petersen, O. H., and Tepikin, A. V. (1999) Hormone-inducedsecretory and nuclear translocation of calmodulin: Oscillationsof calmodulin concentration with the nucleus as an integrator,Proc. Natl. Acad. Sci. USA 96, 4426–4431.efer, C., Belin, M.-T., Caillet-Boudin, M.-L., and Boulanger, P.(1990) Human adenovirus–host interactions: Comparative studywith members of subgroups B and C, J. Virol. 64, 3661–3673.e Giorgi, F., Ahmed, Z., Bastianutto, C., Brini, M., Jouaville,L. S., Marsault, R., Murgia, M., Pinton, P., Pozzan, T., andRizzuto, R. (1999) Targeting GFP to organelles, Methods Cell.Biol. 58, 75–85.eorgi, A., Mottola-Hartshorn, C., Warner, A., Fields, B., andChen, L. B. (1990) Detection of individual fluorescently labeledreovirions in living cells, Proc. Natl. Acad. Sci. USA 87,6579–6583.reber, U. F. (1998) Delivery of animal virus DNA into the nucleus(review article), in Seymour, L., Kabanov, A., and Felgner, P.(Eds.), Self-Assembling Complexes for Gene Delivery: FromChemistry to Clinical Trial, pp. 89–114, Wiley, Sussex, UK.reber, U. F. (1998) Virus assembly and disassembly: The adeno-virus cysteine protease as a trigger factor (review article), Rev.Med. Virol. 8, 213–222.reber, U. F., Nakano, M. Y., and Suomalainen, M. (1998)
Adenovirus entry into cells: A quantitative fluorescence micros-
copy approach, in Wold, W. S. M. (Ed.), Adenovirus Methods andProtocols, Vol. 21, Meth. Mol. Med., pp. 217–230, HumanaPress, Totowa, NJ.reber, U. F., Suomalainen, M., Stidwill, R. P., Boucke, K.,Ebersold, M., and Helenius, A. (1997) The role of the nuclearpore complex in adenovirus DNA entry, EMBO J. 16, 5998–6007.reber, U. F., Webster, P., Weber, J., and Helenius, A. (1996) Therole of the adenovirus protease in virus entry into cells, EMBOJ. 15, 1766–1777.

i, C. J., Heim, R., Lu, P., Pu, Y., Tsien, R. Y., and Chang, D. C.(1999) Dynamic redistribution of calmodulin in HeLa cellsduring cell division as revealed by a GFP-calmodulin fusionprotein technique, J. Cell. Sci. 112, 1567–1577.

i, E., Stupack, D., Bokoch, G. M., and Nemerow, G. R. (1998)Adenovirus endocytosis requires actin cytoskeleton reorganiza-tion mediated by Rho family Gtpases, J. Virol. 72, 8806–8812.

i, E., Stupack, D., Klemke, R., Cheresh, D. A., and Nemerow,G. R. (1998) Adenovirus endocytosis via alpha(V) integrinsrequires phosphoinositide-3-oh kinase, J. Virol. 72, 2055–2061.

ouis, N., Fender, P., Barge, A., Kitts, P., and Chroboczek, J.(1994) Cell-binding domain of adenovirus serotype 2 fiber, J.Virol. 68, 4104–4106.

udin, B., and Matus, A. (1998) GFP illuminates the cytoskeleton,Trends Cell. Biol. 8, 72–77.iles, B. D., Luftig, R. B., Weatherbee, J. A., Weihing, R. R., andWeber, J. (1980) Quantitation of the interaction between adeno-virus types 2 and 5 and microtubules inside infected cells, Virol.105, 265–269.guyen, D. H., Catling, A. D., Webb, D. J., Sankovic, M., Walker,L. A., Somlyo, A. V., Weber, M. J., and Gonias, S. L. (1999)Myosin light chain kinase functions downstream of Ras/ERK topromote migration of urokinase-type plasminogen activator-stimulated cells in an integrin-selective manner, J. Cell. Biol.146, 149–164.ancea, E., Teruel, M. N., Quest, A. F., and Meyer, T. (1998) Greenfluorescent protein (GFP)-tagged cysteine-rich domains fromprotein kinase C as fluorescent indicators for diacylglycerolsignaling in living cells, J. Cell. Biol. 140, 485–498.

ersson, R., Svensson, U., and Everitt, E. (1983) Virus receptorinteraction in the adenovirus system. II. Capping and coopera-tive binding of virions on HeLa cells, J. Virol. 46, 956–963.amirez-Weber, F. A., and Kornberg, T. B. (1999) Cytonemes:Cellular processes that project to the principal signaling centerin Drosophila imaginal discs, Cell 97, 599–607.az, E., Zlokarnik, G., Tsien, R. Y., and Driever, W. (1998)Beta-lactamase as a marker for gene expression in live ze-brafish embryos, Dev. Biol. 203, 290–294.uss, J. C. (1995) The Image Processing Handbook, 2nd ed., CRCPress, Boca Raton, FL.

tauffer, T. P., Guerini, D., and Carafoli, E. (1995) Tissue distribu-tion of the four gene products of the plasma membrane Ca21pump. A study using specific antibodies, J. Biol. Chem. 270,12184–12190.

tewart, P. L., Fuller, S. D., and Burnett, R. M. (1993) Differenceimaging of adenovirus: Bridging the resolution gap betweenx-ray cystallography and electron microscopy, EMBO J. 12,2589–2599.

tidwill, R. S., and Greber, U. F. (2000) Intracellular virustrafficking reveals physiological characteristics of the cytoskel-eton (review article), News Physiol. Sci., in press.

uomalainen, M., Nakano, M. Y., Boucke, K., Keller, S., Stidwill,
R. P., and Greber, U. F. (1999) Microtubule-dependent minus

T

T

T

T

W

W

W


and plus end-directed motilities are competing processes fornuclear targeting of adenovirus, J. Cell Biol. 144, 657–672.

apon, N., and Hall, A. (1997) Rho, Rac and Cdc42 GTPasesregulate the organization of the actin cytoskeleton, Curr. Opin.Cell Biol. 9, 86–92.

rinczek, B., Ebneth, A., Mandelkow, E.-M., and Mandelkow, E.(1999) Tau regulates the attachment/detachment but not thespeed of motors in microtubule-dependent transport of singlevesicles and organelles, J. Cell Sci. 112, 2355–2367.

sien, R. Y. (1998) The green fluorescent protein, Annu. Rev.Biochem. 67, 509–544.

varusk, W., Bentele, M., Misteli, T., Rudolf, R., Kaether, C.,
Spector, D. L., Gerdes, H. H., and Eils, R. (1999) Time-resolved
analysis and visualization of dynamic processes in living cells,Proc. Natl. Acad. Sci. USA 96, 7950–7955.alters, R. W., Grunst, T., Bergelson, J. M., Finberg, R. W., Welsh,M. J., and Zabner, J. (1999) Basolateral localization of fiberreceptors limits adenovirus infection from the apical surface ofairway epithelia, J. Biol. Chem. 274, 10219–10226.eber, J. (1976) Genetic analysis of adenovirus type 2. III.Temperature sensitivity of processing of viral proteins, J. Virol.17, 462–471.ickham, T. J., Mathias, P., Cheresh, D. A., and Nemerow, G. R.(1993) Integrin-alpha-v-beta-3 and integrin-alpha-v-beta-5 pro-mote adenovirus internalization but not virus attachment, Cell
73, 309–319.

Quantitative Microscopy of Fluorescent Adenovirus Entry€¦ · Fluorescent viruses have long been...

Documents

Transcript of Quantitative Microscopy of Fluorescent Adenovirus Entry€¦ · Fluorescent viruses have long been...