JOURNAL OF VIROLOGY, May 1993, P. 2779-27860022-538X/93/052779-08$02.00/0Copyright © 1993, American Society for Microbiology

Opening and Refolding of Simian Virus 40 and In VitroPackaging of Foreign DNA

M. C. COLOMAR, C. DEGOUMOIS-SAHLI, AND P. BEARD*Department of Virology, Swiss Institute for Experimental Cancer Research, 1066 Epalinges, Switzerland

Received 19 October 1992/Accepted 19 January 1993

Simian virus 40 (SV40) can be disassembled under mild conditions by reducing disulfide bonds in the capsidand removing calcium ions. The nucleoprotein complexes formed, analyzed by electron microscopy, werecircular and made up of 59 4 subunits, each with a diameter of about 10 nm. The complexes contained theviral DNA, histones, and the viral capsid proteins. The complexes had much-reduced infectivities comparedwith intact SV40. Addition of calcium ions to the disrupted virus caused the nucleoprotein complexes to refoldinto virus-like structures which sedimented at the same rate as intact SV40 and regained infectivity. Treatmentof the disrupted SV40 with a high concentration of salt dissociated the viral proteins from the DNA. Loweringstepwise the salt concentration, removing the reducing agent, and adding calcium ions allowed structures to bereformed, and these structures sedimented, like SV40, at 240S and were infectious. The plaque-forming abilityof the reconstituted particles was between that of the dissociated components and that of intact SV40. Theaddition of purified DNA of polyomavirus to the dissociated SV40 before the lowering of the salt concentrationshowed that virus-like structures could be formed from SV40 proteins and a foreign DNA.

Simian virus 40 (SV40) is a small nonenveloped viruscontaining a circular double-stranded DNA of 5,243 bp (9,20). SV40 infects lytically monkey kidney cells, transformsrodent cells in culture, and can induce tumors in susceptibleyoung rodents (for a review, see reference 26). SV40 DNAhas been used as a vector for the expression of foreign genesin eukaryotic cells (22). In the virion, SV40 DNA is associ-ated with the four host-cellular histones, H2A, H2B, H3, andH4. This nucleoprotein complex is enclosed in a capsidcomposed of the virus-coded proteins VP1, VP2, and VP3(26). The icosahedral outer viral shell is composed of 360molecules of the major capsid protein VP1 assembled into 72pentameric capsomers. Of these, 60 are hexavalent (that is,in the virus particle they contact 6 other pentamers) and 12,which are at the fivefold axes of the icosahedron, arepentavalent (11). The structure of SV40 at high resolutionhas recently been determined (14).The viral particles can be disrupted by treatment with an

alkaline pH or with the reducing agent dithiothreitol (DTT)and the calcium chelator EGTA (ethylene glycol-bis(P-ami-noethyl ether)-N,N,N',N'-tetraacetic acid) (2, 4, 17, 28).Attempts to form infectious DNA-protein complexes fromSV40 components have been made some time ago (3), butthe products did not resemble SV40 virions. Nevertheless,VP1 of polyomavirus, which is closely related to SV40,synthesized in Escherichia coli, can assemble into formswith the structural features of viral capsids (10). The obser-vation (17) that SV40 virions can be disassembled under mildconditions to reveal a circular structure with well-definedprotein subunits attached to the viral DNA made it seem tous worthwhile to reexamine the structure and stability of theopened virions and to test the possibility of refolding them tomake compact infectious particles.

MATERUILS AND METHODSCells and virus. SV40 virus (strain 777) and polyomavirus

(strain A2) and their host cells (CV-1 and 3T6, respectively)

* Corresponding author.

were used. The cells were grown as monolayers in Dulbec-co's modified Eagle medium containing 5% fetal calf serum.For infection, the medium was removed from the culturesand replaced by 0.4 ml of viral suspension per 9-cm-diameterplate. After 60 min at 37°C, 10 ml of medium with 5% fetalcalf serum was added, and infection was continued.Virus purification. SV40 virions were isolated from a crude

freeze-thawed lysate after centrifugation at 10,000 x g for 30min. Supernatant containing the virions was centrifuged onan SW27 rotor at 4°C at 20,000 rpm for 4 h. Virus wasresuspended in 1 mM phosphate buffer (pH 8.0) and purifiedby sedimentation through a 10-to-30% sucrose gradient in 10mM Tris-Cl, pH 8.0, in an SW40 rotor for 50 min at 30,000rpm. The fractions containing virus, as judged from theoptical density at 260 nm, were dialyzed against 1 mMphosphate buffer.

Purification of polyomavirus DNA. Polyomavirus DNAwas obtained by the Hirt extraction method (12). Viral DNAwas purified by centrifugation in an ethidium bromide-cesium chloride gradient.

Disruption of SV40 virions and refolding. SV40 virions at aconcentration corresponding to about 1 mg of viral DNA perml (the concentration was calculated from the optical densityat 260 nm) were incubated in 50 mM Tris-Cl buffer (pH 7.9)containing 150 mM NaCl, 1 mM EGTA, and 20 mM DTT ina final volume of 50 ,ul at 37°C for 1 h. To test for refolding,disrupted virions were diluted with increasing concentra-tions of CaCl2 up to 5 mM (final concentration) and incu-bated at 37°C for 30 min.

Dissociation of SV40 viral components and reconstitution ofparticles in vitro. Disrupted virions were incubated in thedisruption mixture with an additional 850 mM KCl for 30 minat 37°C to separate viral proteins from DNA. Reconstitutionwas performed by lowering the salt concentration from 1mM to 150 mM by stepwise addition of 50 mM Tris-HCl (pH7.6)-l mM EGTA-20 mM DTT in 17-,ul portions (one-fourthof the original total volume) once every 10 min at roomtemperature. This method was based on one previouslyshown to reconstitute nucleosome core particles from his-

2779

Vol. 67, No. 5

2780 COLOMAR ET AL.

tones and DNA (6). Samples were dialyzed against 10 mMTris-Cl (pH 7.2)-150 mM KCl-1 mM CaCl2 at 4°C.

Reconstitution of particles with polyomavirus DNA. SV40virions (45,ug of DNA) were completely dissociated in 1 Msalt; then, 450 ng of purified polyomavirus DNA was added.Reconstitution was performed as described above. Recon-stituted particles were treated with 1 U of DNase I (Phar-macia) in the presence of 1 mM MnCl2 for 1 h at 37°C.

Velocity sedimentation. Virus and viral components were

layered onto linear 5-to-20% (or 5-to-30% in the case ofrefolded particles) sucrose gradients. Centrifugation was

performed in a BeckmanSW60.Ti rotor at 40,000 rpm at 4°C.The time of centrifugation and the buffer used to form thegradient varied among experiments as follows. Sedimenta-tion of disrupted virions lasted 45 min in a buffer containing50 mM Tris-HCl [pH 7.9], 150 mM NaCl, 1 mM EGTA, and20 mM DTT; disrupted virions refolded by addition of CaCl2were analyzed on a gradient containing 10 mM Tris-HCl, pH7.6, and 100 mM NaCl. Sedimentation of dissociated viralcomponents was in 50 mM Tris-HCl (pH 7.9)-i M KCl-1mM EGTA-20 mM DTT for 4 h. Particles reconstituted afterthe high-salt-concentration treatment and intact virions usedas a control were sedimented for 45 min in 10 mM Tris-HClbuffer (pH 7.2 or 7.6) containing 150 mM KCl and 1 mMCaCl2. Gradients were collected from the bottom. Sedimen-tation coefficients were estimated by comparison with 21Ssupercoiled SV40 or polyomavirus DNA and 240S intactSV40 virions.

Electron microscopy. Samples were adsorbed onto amy-

lamine-charged, Formvar-carbon-coated copper grids, es-

sentially as previously described (7). Staining was performedwith 1% phosphotungstate, pH 7.0, and rotary shadowingwas performed with platinum. Photographs were taken in thePhilips 400 electron microscope at a magnification ofx27,500.DNA and protein analysis. Fractions from sucrose gradi-

ents were analyzed by sodium dodecyl sulfate (SDS)-15%polyacrylamide gel electrophoresis, and proteins were

stained by silver (23).SV40 DNA was analyzed by 1% agarose gel electrophore-

sis and stained with ethidium bromide, and the fluorescentDNA bands were observed under UV at 310 nm. SV40 DNAlabelled with [3H]thymidine was subjected to agarose gelelectrophoresis, the gel was soaked in Amplify (Amersham)and dried, and the radioactivity was visualized by fluorog-raphy.

Analysis of polyomavirus DNA by slot blotting. Eighteenmicroliters of each fraction was incubated for 10 min at 65°Cin the presence of 0.2 N NaOH. Samples were then cooled to

room temperature and neutralized by adding 1 volume of 2 Mammonium acetate. The samples were applied to the nitro-cellulose with a Schleicher & Schuell Slot-Blotter. The filterwas placed in a vacuum oven at 80°C for 2 h, and hybridiza-tion was performed by the method of Johnson et al. (13). Theprobe used to analyze polyomavirus DNA in reconstitutedparticles was polyomavirus DNA cloned into pEMBL9+.The probe was labelled by linearizing the DNA with theHindIII restriction enzyme and labelling the ends with

[at-32P]dATP as described by Sambrook et al. (21) by using

T4 DNA polymerase (Boehringer Mannheim Biochemicals).

Plaque assay. Confluent CV-1 cells in petri dishes (diame-ter, 5 cm) were inoculated with 0.1 ml of sample (SV40virions or disrupted or reconstituted SV40 particles) diluted

in phosphate-buffered saline (PBS) buffer. Cells were incu-

bated for 1 h at 37°C and rinsed with PBS buffer, and 5 ml ofmedium containing 2% fetal calf serum was added. Incuba-

tion was continued for 12 h. The medium was then replacedby 10 ml of medium containing 0.9% Noble agar and 1% fetalcalf serum. Plaques were visualized by staining with neutralred and counted after 6 and 8 days.

RESULTS

The structure of disrupted SV40 particles. SV40 virionswere prepared from infected CV-1 cells. Purified virionswere disrupted by treatment with EGTA and DTT in Tris-HCl buffer (pH 7.9) at 37°C for 1 h. Disrupted virions wereadsorbed to Formvar-carbon-coated grids, stained, shad-owed, and analyzed by electron microscopy. The electronmicrographs (Fig. lb) revealed circular nucleoprotein struc-tures composed of subunits, similar to those previouslyobserved (17). The number of subunits per disrupted virus isshown in the histogram (Fig. la). The average was 59 + 4and was reproducible in several independent disruptionexperiments. The distribution of subunits seemed to beuniform along the viral DNA. The subunit diameter wasestimated to be 10 + 0.5 nm. The contour length of thecircular structures was 0.98 ± 0.08,um, shorter than that offree SV40 DNA (1.40 + 0.17 ,um) (data not shown). Theappearance of disrupted viruses is illustrated in Fig. lb. Forcomparison, intact virions observed as round compact struc-tures and tobacco mosaic virus, used as a standard, werealso added to the sample applied to this grid (Fig. lb). As acontrol, SV40 minichromosomes from infected cells wereobserved by the same technique and had a contour length of0.84 + 0.09 ,um. The minichromosomes contained on aver-age 23 + 1 nucleosomes as observed before (5), withdiameters estimated to be 12.5 + 1.7 nm (data not shown).

Refolding of disrupted SV40 into virus-like structures.SV40 virions have a sedimentation coefficient of 240S (26).SV40 disrupted with EGTA and DTT was found to sedimentat 110S, a value similar to 104S, reported by Moyne et al.(17). The lower rate of sedimentation of disrupted virionswas expected on the basis of their less compact structures(Fig. lb). Polyacrylamide gel analysis showed no virionproteins to be lost upon disruption (Fig. 2) (17). In agreementwith this conclusion, measurements of mass by A. Engel(Biozentrum, Basel, Switzerland) using scanning transmis-sion electron microscopy (8) gave coinciding results forintact SV40 virions and for the circular structures obtainedby DT1T-EGTA treatment.To answer the question of whether disrupted SV40 virions

could be refolded to form compact infectious particles, apreparation with [3H]thymidine-labelled DNA was dividedinto two parts and one part was incubated with 5 mM CaCl2.The two samples were sedimented in parallel sucrose gradi-ents (see Materials and Methods). After incubation withCaC12, the sedimentation coefficient of disrupted SV40 in-creased from about 110 to 240S (Fig. 3a). Thus, the disruptedvirions, refolded by addition of CaCl2, sedimented similarlyto intact SV40 virions. The effect of adding CaCl2 to dis-rupted SV40 was also followed by electron microscopy.However, CaCl2 interfered with the technique, and it wasnot possible to obtain pictures at 5 mM CaCl2. Nevertheless,it was apparent that with increasing concentrations of CaCl2compact structures were formed, and at 1 mM CaCl2 sphereswith dimensions roughly equ;valent to those of SV40 virionsappeared (Fig. 3b).

Infectivities of disrupted SV40 virions and refolded parti-cles. SV40 virions, disrupted virions, and refolded particleswere purified on sucrose g; adients, and the peak fractionsdetected by gel electrophoresis and ethidium bromide stain-

J. VIROL.

SV40 OPENING AND REFOLDING 2781

a) 10-'

zLLJ

LU

LU

8-

6-

4-

2- I'I I I

III. I

10 20 30 40 50 60 70 80 90 100

NUMBER OF SUBUNITS PER DISRUPTED SV40

b )

FIG. 1. Intact and disrupted SV40 visualized by electron microscopy. (a) Graphical representation of the number of subunits perEGTA-DTT-disrupted SV40 virion. (b) A mixture of EGTA-DTT-disrupted SV40, intact SV40 virions without disruption (compact sphereunder these conditions), and tobacco mosaic virus used as a standard was adsorbed to the grid.

ing were used to infect CV-1 cells. Two other fractions of thesucrose gradient outside the 240 and 110S peaks, used ascontrols, and disrupted virions without sucrose gradientfractionation were used for infection. Infection was detectedby the synthesis of viral DNA as follows. At 24 h postinfec-tion, cells were labelled with [3H]thymidine, and incubationwas continued at 37°C. Viral DNA was extracted at 72 hpostinfection, subjected to agarose gel electrophoresis, andvisualized by fluorography. The gel pattern (Fig. 3c) showedtwo bands, with the upper band corresponding to hostcellular DNA and the lower band corresponding to super-coiled SV40 DNA. No viral DNA was detected after mockinfection (Fig. 3c, lane 1) or after infection with the twogradient fractions used as controls (Fig. 3c, lanes 3 and 5) orwith the 110S peak of disrupted SV40 (Fig. 3c, lane 6). ViralDNA replication was detected after infection with SV40virions (Fig. 3c, lane 2) or refolded particles (Fig. 3c, lane 4).

VP23VtP3

H

dis V M

I _92_ 66

13 B 45rZ. 31""lWI 3 1

I 0 121

I 14

i_2 3 4

FIG. 2. SV40 disrupted with EGTA and DTT was isolated bysucrose gradient sedimentation as described in Materials and Methodsand the legend to Fig. 3. The proteins were analyzed by electrophoresisthrough an SDS-15% polyacrylamide gel. Lanes 1 and 2, gradientfractions containing disrupted (dis) SV40; lane 3, SV40 virions withoutdisruption (V); lane 4, molecular weight markers (M). The positions ofVP1, VP2, VP3, and histones (H) are indicated on the left. Molecularsizes in kilodaltons are indicated on the right.

VOL. 67, 1993

2782 COLOMAR ET AL.

( a )1200-

1000 -

800-

E

m- 600-

400

200

0

( b)

( C)

240 S

o -(

110 S

10fraction number

0

1 2 3 4 5

ANM. tIo mo

FIG. 3. Sedimentation analysis of disruptedparticles. (a) SV40 virions labelled with [ H]trupted at pH 7.9 in an EGTA-DTT mixture as deand Methods (large circles). Refolded particleincubation of disrupted virus in 5 mM CaCl2 (smwere layered onto 10-to-30% sucrose gradients45 min at 40,000 rpm in an SW60 rotor. Fractiand the radioactivity was measured. The positabout 110 and 240S are indicated. SV40 insedimented to the position marked 240S (not shof disrupted SV40 refolded in 1 mM CaCl2 otmicroscopy. (c) Infectivity of disrupted and refSV40 virions and disrupted and refolded particsucrose gradients and used to infect CV-1 cellstion, cells were labelled with [3H]thymidine,continued at 37°C. At 72 h postinfection, viral Danalyzed by agarose gel electrophoresis. Lanecells; 2, infection with intact SV40 virus; 4, infipeak of refolded particles; 6, infection withdisrupted SV40 virions; 7, disrupted SV40 withfractionation. As controls, the bottom fractionsrefolded and disrupted particles were tested (latively). C and V, cellular and viral DNAs, resp

On the basis of the intensities of the bandsafter infection with the refolded particleefficient as that after infection with intacilow level of viral DNA replication was se

with disrupted but unpurified virions (Fig. 3c, lane 7). Thiscould be due to a small residual amount of incompletely

a disrupted SV40 (for example, see the gradient profile of6.a disrupted SV40 in Fig. 3a). We conclude that the infectivity

of SV40 was greatly reduced upon disruption of the virionswith EGTA and DTT. The loss of infectivity upon disruptioncould result from, among other possibilities, failure to bindto cell surface receptors, increased susceptibility to nu-cleases, or unsuitability as a replication template. In anycase, infectivity was restored in the refolded particlesformed upon incubation with CaCl2.Complete dissociation of SV40 into DNA and protein sub-

units. To attempt to dissociate the DNA and protein subunitsof disrupted SV40, we increased the ionic strength to 1 Msalt still in the presence of EGTA and DTT. This materialwas applied to a 5-to-20% sucrose gradient, and fractionswere collected and analyzed by agarose and SDS-polyacryl-amide gel electrophoresis to detect viral DNA and proteins

20 (Fig. 4a and b, respectively). Two forms of viral DNA were

observed (Fig. 4a). The higher sedimentation coefficientcorresponded to the supercoiled form (with a peak in frac-tion 5) and the lower sedimentation coefficient correspondedto the relaxed form (with a peak in fraction 9). The sedimen-tation rates observed were exactly the same as those ofprotein-free SV40DNA (not shown). The structural proteins

ri;x-z; of SV40 (Fig. 4b) sedimented more slowly than the viralDNA; their peak was in fractions 14 and 15. Under these

>w6*ws2 ?-><kw conditions, therefore, the viral DNA and protein subunitswere separated. There was no peak of protein in the DNA-

:-W-< i; containing fractions. However, since there was a smallamount of VP1 in every fraction we cannot exclude thatsome of it was associated with DNA.Can the dissociated viral components be induced to reas-

6 7 sociate? To test this possibility, the ionic strength of thesample the analysis of which is shown in Fig. 4a and b was

wC: decreased stepwise slowly to 150 mM in the presence ofV EGTA and DTT. This method was based on one previously

shown to reconstitute nucleosome core particles from his-tones and DNA (7) (see Materials and Methods). The reduc-

and refolded SV40 ing agent and EGTA were removed by dialysis at 4°C,thymidine were dis- calcium ions were added, and the samples were subjected to-scribed in Materials sucrose gradient sedimentation. Fractions were collecteds were obtained by and analyzed by agarose and SDS-polyacrylamide gel elec-iall circles). Samples trophoresis as described above. A single major peak inand centrifuged for which SV40DNA and the viral proteins sedimented togetherions were collected, was observed (Fig. 4c, lower panel, and Fig. 4d [fraction 7]).ions of the peaks at The histones, present in SV40 in much smaller amounts thanaown). (b) Examples the major capsid protein (Fig. 2), were seen only faintly inbserved by electron this gel. As a control, intact SV40 virions were sedimented inFolded SV40 virions. a parallel sucrose gradient. The viral DNA is visualized inles were purified on the upper gel of Fig. 4c. The peak of virions was in fractions. At 24 h postinfec- 7 (the material sedimenting to the bottom of the gradientand incubation was [fraction 1] is thought to be due to aggregates of viralNA was purified and particles). The sedimentation rate of the reconstituted parti-s: 1, mock-infected cles was, therefore, the same as that of the SV40 virions.the 11oS peakof20 Infectivity of reconstituted particles. To examine the infec-

out sucrose gradient tious capacity of the particles reconstituted from separatedofthe gradientsdwith SV40 components, two different assays (measurements ofines 3 and 5, respec- viral DNA replication in infected cells and a plaque assay))ectively. were used. CV-1 cells were infected with the same amount of

either intact SV40, disrupted virus treated with 1 M salt, or

particles reconstituted as for the analysis shown in Fig. 4. To,, DNA replication measure viral DNA replication, viral DNA was purified ates was almost as different times after infection, subjected to agarose gelt SV40 virions. A electrophoresis, and detected by ethidium bromide staining.en after infection (Fig. 5a). Viral supercoiled, viral relaxed, and contaminating

J. VIROL.

SV40 OPENING AND REFOLDING 2783

I 3 5 7 9 1 135 -758 9)c)- 0ar

° a >cn >XL) L) () 1)S

(a) _

1 2 3 4 5 6 7

(b) tMl 1 3 5

VP-3 -

V-3

3 5 9 3 1 5

( b )

SV40 VIRIONS

RECONSTITUTEDPARTICLES

( C ) 1v1

r -!S-L

;.,JIs.r . A73

*1

( d ) ". 3 D 7

*'t'j ': _A

vp-1 --VP-3- -I....... RECONSTITUTED

PARTICLES

FIG. 4. Sedimentation analysis of DNA and capsid proteins ofSV40 virions after complete dissociation and after reconstitution ofviral components. Disrupted particles were incubated with an addi-tional 850 mM KCl at 37'C for 30 min. Samples were applied to a5-to-20% sucrose gradient. After centrifugation for 4 h at 40,000 rpmin an SW60 rotor, fractions were collected. Viral DNA and proteinswere analyzed by gel electrophoresis and visualized by staining withethidium bromide and silver, respectively. Fraction numbers aremarked at the top. Fractions 1 and 19 correspond to the bottom andthe top of the gradient, respectively. (a) Agarose gel electrophoresisof the SV40 DNA in the dissociated-virion preparation. M, sizemarker of SV40 DNA; r and s, relaxed and supercoiled forms ofSV40 DNA, respectively. (b) SDS-polyacrylamide gel electrophore-sis of SV40 proteins. M, marker of SV40 capsid proteins. (c)Sedimentation analysis of DNAs of reconstituted particles (lowergel) and those of intact virions (upper gel). Samples were applied toa 5-to-20% sucrose gradient and centrifuged at 40,000 rpm for 45min. (d) Viral proteins in reconstituted particles.

cellular DNAs were observed (Fig. 5a). Production of DNAfrom reconstituted virus was readily detectable at 48 and 72h postinfection. In contrast, when the dissociated virus wasused for infection little or no viral DNA was seen at 72 hpostinfection. Both supercoiled and relaxed forms of SV40DNA were observed. There was a noteworthy differencebetween infection with intact SV40 and infection with thereconstituted particles. The SV40 stock used contained asmall proportion of defective DNA molecules shorter thannormal, as often seen with SV40 (25). When cells wereinfected with reconstituted particles, the proportion of

1 2 3 4 5BBottom

-5

FRACTION

I NUMBER

i- 15

-18_ Top

FIG. 5. (a) Analysis of viral DNA replication following infectionby disrupted SV40 virions or reconstituted particles. Samples wereused to infect CV-1 cells, and viral DNA was extracted (12) 48 h(lanes 4 and 5) and 72 h (lanes 6 and 7) postinfection, treated withRNase A, and analyzed by agarose gel electrophoresis. For apositive control we infected CV-1 cells with intact virions (lane 3).For this sample, DNA was extracted after 24 h because after 48 hcell lysis had begun and the amount of DNA obtained overloadedthe gel. M, size marker of SV40 DNA. c, vr, and vs, cellular, viralrelaxed, and viral supercoiled DNAs, respectively. (b) Sedimenta-tion analysis of reconstituted particles with polyomavirus DNA andSV40 proteins. Samples were applied to a 5-to-20% sucrose gradientand centrifuged at 40,000 rpm for 45 min in an SW60 rotor. Fractionswere collected, and polyomavirus DNA was analyzed by slot blothybridization (see Materials and Methods). Lanes: 1, intact polyo-mavirus; 2, polyomavirus DNA; 3, reconstituted hybrid particlestreated with DNase I; 4, reconstituted hybrid particles; 5, polyoma-virus DNA treated with DNase I.

shorter viral DNA molecules present in the population alongwith full-length molecules was increased. Since the recon-stitution experiments used the same viral stock as that forthe intact virions used to infect CV-1 cells (Fig. 5a, lane 3),it seems either that in vitro reconstituted particles encapsi-date preferentially shorter SV40 DNA molecules rather thanfull-length viral genomes or that the DNA is shortenedduring infection (see Discussion).The plaque assay showed that dissociation of the compo-

nents of SV40 reduced the infectious titer by a factor of 107(Table 1). Reconstitution of particles from the componentsraised the infectivity by a factor of 104 (Table 1). Theinfectivity of the reconstituted particles is, therefore, be-tween that of the separated components and that of intactSV40.

In vitro reconstitution of particles with foreign DNA. To

(a) M

r-S -

'Cvr

Vs

VOL. 67, 1993

2784 COLOMAR ET AL.

TABLE 1. Plaque assay of SV40 infectivity

No. of plaquesbDNA concn(mg/ml)' Intact Disrupted Reconstituted

virions virions particles

10-1 3410-2 410-310-4 1710-5 510-6 110-7 110-8 161o-9 610-10 1

aEstimated from the optical density at 260 nm of the initial viral stock.b Average of two experiments without correction for loss during disruption

or reconstitution.

test whether SV40 capsid proteins can associate with a DNAother than that of SV40, we also tried to reconstitute in vitrovirus-like particles made up of SV40 proteins and polyoma-virus DNA. Polyomavirus is another papovavirus similar insize and structure to SV40. Purified supercoiled polyomavi-rus DNA was incubated with 100-fold molar excess of SV40,completely dissociated as for the analysis shown in Fig. 4,and reconstitution was performed as described above. Poly-omavirus DNA was present as a small proportion of the totalDNA in the reconstitution mixture, and so it would notsignificantly change the ratio of proteins to DNA. Does partof it become associated with histones and SV40 capsidproteins?The reconstituted particles were analyzed by sucrose

gradient sedimentation, along with intact polyomavirus vir-ions and free polyomavirus DNA. Polyomavirus DNA in thegradient fractions was detected by slot blot hybridization(Fig. 5b) using a polyomavirus DNA probe (see Materialsand Methods), and the SV40 DNA was detected by agarosegel electrophoresis and ethidium bromide staining. Polyoma-virus virions sediment similarly to SV40 at about 240S (Fig.5b, lane 1) (26), while free DNA, which sediments at 21S,was found near the top of the gradient (Fig. Sb, lane 2).Reconstituted particles containing polyomavirus DNA sedi-mented in a broader peak (Fig. Sb, lane 4); however, themajority sedimented at about the same position as intactpolyomavirus virions. After treatment of the reconstitutedmixture with DNase I, the DNase I-resistant particles withpolyomavirus DNA and SV40 proteins sedimented in asingle sharp peak (Fig. Sb, lane 3) close to the position ofintact polyomavirus virions (Fig. Sb, lane 1) and at the sameposition as the bulk of the reconstituted particles whichcontained SV40 DNA (not shown). Thus, reconstitutedparticles with polyomavirus DNA sedimenting at about 240Swere resistant to DNase I, while the other, presumably lesscompletely reconstituted, particles sedimenting outside thispeak (Fig. 5b, lane 3, fractions 1 to 6 and 13 to 19) were not.The infectivity in mouse cells of the polyomavirus DNA

reconstituted with SV40 proteins was tested by a DNAreplication assay similar to that shown in Fig. 5a. It wasfound to be low, about the same as that of naked polyoma-virus DNA (unpublished results). SV40 can bind to and entermouse cells, so the reason for this low infectivity is unclear.

DISCUSSION

In this report, we describe the dissociation of SV40 virionsby the calcium chelator EGTA and the reducing agent DTTat physiological ionic strength. This dissociation led to theformation of circular nucleoprotein complexes. These com-plexes sedimented at 110S, significantly slower than intactvirions, which sediment at 240S, yet contained the sameamounts of the four histones and the three viral structuralproteins as intact virions (Fig. 2). The circular structureswere composed of subunits, with the number of subunits percomplex on the average being 59 + 4.EGTA and DTT probably act by disrupting bonds linking

adjacent capsomers. Although it is well established thatSV40 virions are sensitive to these agents (1, 17), the exactroles of Ca2' and disulfide bonds in maintaining the struc-ture of the virion are not clear. The SV40 structure (14)suggests that bound Ca2+ could form a bridge between twopentamers, and it is also possible that disulfides form be-tween cysteines of neighboring pentamers. The capsid ofpolyomavirus (a virus closely related to SV40) is a highlypolymeric structure of VP1 protein subunits, which arecovalently linked by disulfide bridges (29). VP2 and VP3 arebound to this structure by hydrophobic bonds.What do the subunits in the disassembled virions repre-

sent? We cannot give a clear answer to this question. Wecounted an average of 60 subunits in several preparations ofdisrupted virus. This is higher than the average number ofnucleosomes in SV40 chromatin, which is 23. It is likely thatthe majority of subunits we see are capsomers attached toDNA and histones. A minority could be nucleosomes.

Earlier work showed that treatment of SV40 virus with analkaline pH released VP1-containing capsomers, leaving anucleoprotein complex indistinguishable by electron micros-copy from the SV40 minichromosome extracted from in-fected cells (4). The nucleoprotein complex contained DNA,the viral capsid protein VP3, and the four histones. Simi-larly, when SV40 virus disrupted with EGTA-DTT at aneutral pH was incubated at pH 9.8, the complexes appearedby electron microscopy to contain nucleosomes (17). Re-moval of VP1 apparently allows the nucleosomes to be seen.Studies of the polyomavirus and SV40 capsid structuressuggest that each pentamer is associated with an internalprotein (VP2 or VP3) which probably forms bridges to theminichromosome (11, 14). Therefore, most of the subunits indisassembled SV40 that we observed probably correspondedto pentamers of VP1 associated with the internal protein(VP2 or VP3) along a circular structure composed of DNAand histones.Moyne et al. (17) disrupted SV40 virions under similar

though not identical conditions and obtained circular com-plexes with a higher number of beads (mean, 93 + 17 permolecule). The reason for the difference from our observa-tions is not clear, but it may result from the differentmethods used to purify SV40 virions: while Moyne et al. (17)purified the virions in a cesium chloride gradient, we usedsucrose gradient sedimentation and, therefore, did not ex-pose the virus to high salt concentrations.

Since the disassembled viral particles were obtained undermild conditions, it is possible that they mimic an intermedi-ate stage of the viral growth cycle, such as uncoating orassembly, or they may be templates for gene expression.Disruption of SV40 virions by EGTA-DTT at a neutral pHgave nucleoprotein complexes that were more efficient tem-plates for transcription than SV40 minichromosomes iso-lated from nuclei of infected cells and almost as efficient as

J. VIROL.

SV40 OPENING AND REFOLDING 2785

free SV40 DNA (1). The early viral promoter was active, butthe late promoter was not (24).The disassembled SV40 structures were much less infec-

tious than the intact virus. Incubation with Ca2+ ions causedthese structures to fold back into a compact form whichsedimented in a sharp peak at the same position as intactvirus and was infectious, as judged by measuring the repli-cation of viral DNA in infected cells. The disassembly ofSV40 virions could be reversed, therefore, at least by thesecriteria.We wanted to test whether the proteins forming the

subunits of disassembled SV40 could be induced to dissoci-ate from and reassociate with the viral DNA. Cremisi et al.(5) showed that SV40 chromatin can be disrupted by increas-ing the ionic strength. By incubating the disassembled viruswith 1 M salt, the virion DNA and proteins dissociatedcompletely, sedimenting separately in a sucrose gradient(Fig. 4a and b). They could be reassembled into a nucleo-protein complex by a stepwise lowering of the salt concen-tration. Removal of the disulfide-reducing agent and EGTAand addition of calcium ions enabled the nucleoproteincomplex to fold into a particle sedimenting at the sameposition as intact virions, 240S (Fig. 4c and d). Reconstitu-tion of particles was about 30% efficient, as quantified bysedimenting the reconstituted particles through a sucrosegradient, collecting fractions, and scanning the intensity ofthe band of viral DNA sedimenting at 240S. Most of thematerial not in the 240S peak was in aggregates at the bottomof the tube (data not shown). The reassembled virus-likeparticles were found to be infectious by several assays:plaque-forming ability in CV-1 cells (Table 1), the synthesisof viral structural proteins measured by immunofluorescence(data not shown), and production of viral DNA in infectedcells (Fig. 5a).The plaque assay showed that, although the infectivity of

the reassembled particles was 104-fold higher than that of theseparated components, reconstitution of infectivity was stilla lot less efficient than the reconstitution of physical parti-cles. One reason that reconstituted particles are less infec-tious than SV40 is likely to be that they contain preferen-tially defective DNA. However, this would account for onlypart of the difference, since particles which contain full-length DNA are also formed. The reassembled particleswere, therefore, imperfect. It is possible that the proteins inreassembled particles are randomly positioned along theDNA and interfere with the start of viral replication. Inter-estingly, viral DNA which replicated following infection byreassembled particles contained a significant population oflower-molecular-weight DNA, in addition to that of fullgenomic length. This suggests that the reassembly selectedfor the viral DNA molecules (a small fraction of the initialpopulation) which were shorter than full length. Preferentialpackaging by SV40 of DNA molecules shorter than fulllength has been noticed previously (15).An alternative explanation is that the shorter DNA mole-

cules were generated during reassociation, or after infection,as a result of increased sensitivity of the reassembled parti-cles to DNase cleavage. We think this explanation is lesslikely, since there is no evidence for DNA damage during thein vitro steps (Fig. 4). The distinct shorter DNA bands seenafter infection with reassembled particles migrated like thosefrom the original SV40 preparation used but were relativelymore intense.The ability to reassemble in vitro infectious particles

raised the idea of reconstituting particles with a foreignDNA. Oppenheim et al. (18) recently described a sequence

important for encapsidation in vivo, which is located nearthe replication origin of SV40 DNA. Since both SV40 (27)and polyomavirus (16) can encapsidate host cell DNA duringthe lytic cycle, the requirement for such a sequence seemsnot to be absolute. We used free DNA from polyomavirusbecause of its similarity in size and structure to SV40 (19).The sequences of SV40 and polyomavirus DNA are overallnot closely related (26), but there are similar control ele-ments in the ori regions. The reassociation of virion proteinsand DNA in vitro may not have the same sequence require-ments as the process of encapsidation during viral replica-tion in infected cells. The reconstituted particles (pseudo-types) containing polyomavirus DNA and SV40 proteinswhich were resistant to DNase I sedimented slightly moreslowly than intact polyomavirus (Fig. Sb) and at the samerate as SV40 (results not shown). Incompletely reconstitutedparticles sedimenting in other positions were sensitive todigestion by DNase I. Thus, DNase-resistant reconstitutedparticles sedimenting to the correct position were obtained;in addition, other structures were also formed, and theywere formed in greater quantities than with SV40 DNA.The conditions we describe for disruption of SV40 and

reconstitution should facilitate a systematic approach tounderstanding the stabilizing interactions between the com-ponents of the virion and how viral components combine invivo to form the viral structure. These experiments showthat it is possible to reconstitute in vitro infectious virus-likeparticles from an animal virus. This could be a first steptowards a method to encapsidate any DNA.

ACKNOWLEDGMENTS

We are very grateful to B. Hirt for useful discussions and criticalreading of the manuscript. We thank A. Engel of the Biozentrum forcarrying out measurements of mass by scanning transmission elec-tron microscopy and for providing tobacco mosaic virus. We thankH. Bruggmann and B. Bentele for competent technical help.

This work was supported by grants from the Swiss NationalScience Foundation.

REFERENCES1. Brady, J. N., C. Lavialle, and N. P. Salzman. 1980. Efficient

transcription of a compact nucleoprotein complex isolated frompurified simian virus 40 virions. J. Virol. 35:371-381.

2. Brady, J. N., V. D. Winston, and R. A. Consigli. 1977. Dissoci-ation of polyoma virus by the chelation of calcium ions foundassociated with purified virions. J. Virol. 23:717-724.

3. Christensen, M., and M. Rachmeler. 1976. Studies on the invitro formation of infectious DNA-proteins aggregates fromSV40 components. Virology 75:433-441.

4. Christiansen, G., T. Landers, J. Griffith, and P. Berg. 1977.Characterization of components released by alkali disruption ofsimian virus 40. J. Virol. 21:1079-1084.

5. Cremisi, C., P. F. Pignatti, 0. Croissant, and M. Yaniv. 1976.Chromatin-like structures in polyoma virus and simian virus 40lytic cycle. J. Virol. 17:204-211.

6. Drew, H. R., and A. A. Travers. 1985. DNA bending and itsrelation to nucleosome positioning. J. Mol. Biol. 186:773-790.

7. Dubochet, J., M. Ducommun, M. Zollinger, and E. Kellenberger.1971. A new preparation method for dark-field electron micros-copy of biomacromolecules. J. Ultrastruct. Res. 35:147-167.

8. Engel, A., W. Baumester, and W. 0. Saxton. 1982. Massmapping of a protein complex with the scanning transmissionelectron microscope. Proc. Natl. Acad. Sci. USA 79:4050-4054.

9. Fiers, W., R. Contreras, G. Haegeman, R. Rogiers, A. van deVoorde, H. van Heuverswyn, J. van Herreweghe, G. Volckaert,and M. Ysebaert. 1978. Complete nucleotide sequence of SV40DNA. Nature (London) 273:113-120.

10. Garcea, R. L., D. M. Salunke, and D. L. D. Caspar. 1987.Site-directed mutation affecting polyoma virus capsid self-as-

VOL. 67, 1993

2786 COLOMAR ET AL.

sembly in vitro. Nature (London) 329:86-87.11. Harrson, S. C. 1990. Principles of virus structure, p. 37-62. In

B. N. Fields and D. M. Knipe (ed.), Virology, 2nd ed., vol. 1.Raven Press, New York.

12. Hirt, B. 1967. Selective extraction of polyoma DNA frominfected mouse cell cultures. J. Mol. Biol. 26:365-369.

13. Johnson, D. A., J. W. Gautsch, J. R. Sportsman, and J. H.Edler. 1984. Improved technique utilizing nonfat dry milk foranalysis of proteins and nucleic acids transferred to nitrocellu-lose. Gene Anal. Tech. 1:3-8.

14. Liddington, R. C., Y. Yan, J. Moulai, R. Sahli, T. L. Benjamin,and S. C. Harrison. 1991. Structure of simian virus 40 at 3.8-Aresolution. Nature (London) 354:278-284.

15. Menck, C. F. M., M. James, A. Benoit, and A. Sarasin. 1990.Constraints in simian virus 40 (SV40) encapsidation, as deter-mined by SV40-based shuttle viruses. J. Gen. Virol. 71:143-150.

16. Michel, M. R., B. Hirt, and R. Weil. 1967. Mouse cellular DNAenclosed in polyoma viral capsids (pseudovirions). Proc. Natl.Acad. Sci. USA 58:1381-1388.

17. Moyne, G., F. Harper, S. Saragosti, and M. Yaniv. 1982.Absence of nucleosomes in a histone-containing nucleoproteincomplex obtained by dissociation of purified SV40 virions. Cell30:123-130.

18. Oppenheim, A., Z. Sandalon, A. Peleg, 0. Shaul, S. Nicolis, andS. Ottolenghi. 1992. A cis-acting DNA signal for encapsidationof simian virus 40. J. Virol. 66:5320-5328.

19. Rayment, I., T. S. Baker, D. L. D. Caspar, and W. T. Mu-rakami. 1982. Polyoma virus capsid structure at 22.5-A resolu-tion. Nature (London) 295:110-115.

20. Reddy, V. B., B. Thimmappaya, R. Dhar, K. N. Subramanian,B. S. Zain, J. Pan, P. K. Ghosh, M. L. Celma, and S. M.

Weissman. 1978. The genome of simian virus 40. Science200:494-502.

21. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecularcloning: a laboratory manual, 2nd ed. Cold Spring HarborLaboratory, Cold Spring Harbor, N.Y.

22. Subramani, S., and P. J. Southern. 1983. Analysis of geneexpression using simian virus 40 vectors. Anal. Biochem. 135:1-15.

23. Switzer, R. C., C. R. Merril, and S. Shifrin. 1979. A highlysensitive stain for detecting proteins and peptides in polyacryl-amide gels. Anal. Biochem. 98:231-237.

24. Tack, L. C., and P. Beard. 1985. Both trans-acting factors andchromatin structure are involved in the regulation of transcrip-tion from the early and late promoters in simian virus 40chromosomes. J. Virol. 54:207-218.

25. Tai, H. T., C. A. Smith, P. A. Sharp, and J. Vinograd. 1972.Sequence heterogeneity in closed simian virus 40 deoxyribonu-cleic acid. J. Virol. 9:317-325.

26. Tooze, J. (ed.). 1981. DNA tumor viruses, 2nd ed. (revised).Cold Spring Harbor Laboratory, Cold Spring Harbor, N.Y.

27. Trilling, D. M., and D. Axelrod. 1970. Encapsidation of free hostDNA by simian virus 40 pseudovirus. Science 168:268-271.

28. Varshavsky, A. J., S. A. Nedospasov, V. V. Schmatchenko, V. V.Vakayev, P. M. Chumakov, and G. P. Georgiev. 1977. Compactform of SV40 viral minichromosomes is resistant to nuclease:possible implications for chromatin structure. Nucleic AcidsRes. 4:3303-3325.

29. Walter, G., and W. Deppert. 1974. Intermolecular disulfidebonds: an important structural feature of the polyoma viruscapsid. Cold Spring Harbor Symp. Quant. Biol. 39:255-257.

J. VIROL.