J. Mol. BioE (1977) 117, 473495

Polyoma Virus Defective DNAs I. Physical Maps of a Related Set of Defective Molecules (D76, D91, D92) ELSEBET LUND-~, MIKE FRIED AND BEVERLY E. GRIFFIN Imperial Cancer Research Fund P.O. Box 123, Lincoln’s Inn Fields London WCZA 3PX, England (Received 27 January

1977, and in revised form 29 July

1977)

Three related polyoma virus species, designated D92 (92% the size of full-lengtll polyoma virus DNA), D91 (91%) and D76 (76%) have been analysed and their structures compared with that of polyoma virus A2 DNA. Three independent depurination fingerprinting and methods (restriction endonuclease cleavage, DNA-DNA hybridization) were used in the analysis. The defective DNAs appear to be: (1) entirely composed of viral sequences (no host DNA sequences were detected): (2) made up in part of long continuous sequences of DNA which appear identical to sequences of A2 DNA (D92 contains continuous sequences from 1 to 72 map units on the physical map of A2 DNA; that is, it contains the entire late region and part of the early region of the viral DNA. D91 and D76 contain those same sequences except for a 1% deletion around 18 map units) : (3) made up in part of rearranged viral sequences. Several interesting features were noted about the rearranged sequences present, in the defective DNAs. Sequences from the region around 67 map units were found linked to other (non-contiguous) regions of the DNA. Sequences from about 72 map units were linked to sequences from about 1 map unit. Multiple copies of sequences from 67 to 72 map units (from around the origin of DNA replication) were found (4 copies in D91 and D92, and 2 copies in D76).

1. Introduction Infection of cells with high titres of polyoma virus (high virus to cell ratio) results in the formation of virus particles which contain non-infectious (defective) supercoiled molecules that are heterogeneous in size (Thorne, 1968; Thorne et al., 1968; Blackstein et al., 1969; Fried, 1974). With continued high multiplicity passage! virus stocks become richer in particles which contain defective DNA. Most defective DNAs are shorter than infectious polyoma DNA (Thorne et al., 1968; Blackstein et al., 1969; Fried, 1974), contain restricted portions of the viral genome (Robberson & Fried, 1974; Griffin C%Fried, 1975) and in some cases may also contain host sequences covalently linked to viral sequences (Lavi & Winocour, 1974). The sequences retained in defective viral molecules, those reiterated and those arranged should prove useful in studying the virus-cell interactions as well as the evolution of the viral genome. t Present address: University of Wisconsin Medical Center, Department of Physiological Chemistry, 1215 Linden Drive, 589 Medical Science Buildings, Madison, Wise. 53706, U.S.A4. 473

474

E. LUND,

M. FRIED

AND

B. E: GRIFFIN

Several polyoma defective species have been cloned in the presence of wild-type helper virus (Fried, 1974). These defectives have been found to be non-infectious, to interfere with production of infectious virus and to be resistant to cleavage by the EcoRI restriction endonuclease. An electron microscopic study of heteroduplexes formed between EcoRI linear wild-type viral DNA and circular defective molecules has shown that the cloned polyoma defective molecules are not simply deletion mutants, but contain both regions which are homologous and apparently nonhomologous to full-length infectious polyoma DNA (Robberson & Fried, 1974). This paper presents the structures of a related set of defective molecules (D91, D92, and D76) which have been deduced by restriction endonuclease, pyrimidine fingeranalyses. The biological significance of these print and DNA-DNA hybridization structures

is discussed.

A preliminary

report

on the structure

of these defectives

has been presented elsewhere (Fried et al., 1974).

2. Materials and Methods (a) Virus

stocks

The preparation of low multiplicity virus from polyoma wild-type (A2) large-plaque strain and the clonal isolation of different defective polyoma virus stocks has been described previously (Fried, 1974). The original notations for the defective virus isolates were changed by Robberson & Fried (1974) and their notations are used here. The original clonal isolate of defective DNA, so-called “parental” D92, was recloned (as above) to separate the 2 predominant defective species, D92 and D91. During this recloning a third defective, D76, appeared in the plaque isolate which contained the D91 species. (b) Preparation

of viral

DNAe

Confluent secondary mouse embryo cells were infected with polyoma wild-type (A2) virus, or with clonal isolates of defective virus plus infectious helper virus at a multiplicity of 5 to 10 plaque-forming units per cell in Dulbecco’s modified Eagle’s medium containing 1 pg 13H]thymidine/ml (spec. act. 2 rCi/rg; Radiochemical Centre, Amersham). After about 4 days incubation at 37”C, the viral DNA was extracted by the selective procedure of Hirt (1967). 32P-labelled viral DNA was prepared from infected 3T6 mouse cells essentially as described previously (Griffin et al., 1974) ; the concentration of [32P]orthophosphate (Radiochemical Centre, Amersham) varied from 0.25 to 1.0 mCi per 90-mm dish. In some cases, the viral DNAs were purified directly from the Hirt supernatant by equilibrium centrifugation in ethidium bromide/caesium chloride gradients (Radloff et al., 1967) as previously described (Griffin & Fried, 1976); in other cases, the Hirt supernatant was extracted with phenol, the DNA precipitated with ethanol (2 vol.) and then purified by equilibrium centrifugation (as above). After isolation, supercoiled viral DNA was further purified either by neutral sucrose density gradient centrifugation (Fried, 1974) for wild-type A2 DNA or by electrophoresis in 1% agarose slab gels (20 cm x 20 cm) (see below) for viral DNA preparations which contained both wild-type and defective DNA molecules. (c) Digestion

with restriction

endonucleases

(i) Enzymes Restriction endonucleases HpaII, HaeIII and Hind111 were purified according to the procedures of Sharp et al. (1973) (for HpaII), Middleton et al. (1972) (for HaeIII) and Old et al. (1975) (for HindIII). EcoRI, BumI, HhI, Barn1 and XbaI were generous gifts from Drs R. Kamen, M. Matthews, Rich Roberts, Mr Alan Robbins and Mr Steve Barrett. (ii) Buffers and conditions used for enzyme digestions The cleavage of viral DNA with restriction enzymes EcoRI, HpaII, carried out as described by Griffin et al. (1974).

and Hind111 was

POLYOMA

VIRUS

DEFECTIVE

DN.4s

4 7 r,

All enzyme digestions (except with EcoRI) were carried out in 10 mn/r-Tris*HCl (pH 7.4), 6 mur-MgCl, and 1 maa-dithiothraitol for sufficient time and with sufficient enzyme to cleave the polyoma DNA to completion. Conditions for each batch of enzyme were worked out on 0.1 to 0.2 pg of A2 DNA per assay. EcoRI digestions was carried out in the above hrlffrr which also contained 0.1 M-N&I. (i 1~)Slab gel electrophoresis

and puri$cation

of DNA

from

gels

Preparative separation of wild-type and defective supercoiled DNA (form 1) ant1 circular DNA (form II) was obtained by electrophoresis on 1”) agarose slab gels (20 cm x 20 cm) at 5 V/cm for about 15 h in Tris-acetate buffer (pH 7.5; Sharp et al., 1973). ‘I’IK~ viral DNAs were located either by autoradiography or by visualisation of the cthidirtrll t)romide-stained gel in ultraviolet light. For isolation of the DNA, the excised gc,l bands rverc electrophoretically eluted into dialysis bags (in Tris-acetate buffer) and the elut,ed DNA \vas extracted once with an equal volume of phenol/2-amyl alcol~ol/cl~lorofor~n (25: 1: 24. by vol.), and twice with 1 vol. ether. DNA was precipitated by ethanol in tllc, prtrwnre of 10 to 20 pg of carrier yeast RNA/ml. The products obtained by restriction enzyme digestions were fractionat,ed by electroplloresis on 20 cm x 20 cm agarose (1.0 to 1.4%) gels or polyacrylamide/bisacryla:nid~~ slat] gels (4 t,o 7 Oh, depending on the size of the fragments t,o be separated, with acrglwmidribisacrylamide ratio of 19: 1, at 5 to 10 V/cm) in Tris-borate buffer, pH 8.3 (Dingmarl CY:Prarock. 1968). Ttlt: individual DNA fragments were locabed and isolat.ed as descri kc1 (ci) DNA-DNA

hybridization

in solution

Form 1 viral DNAs and restriction enzyme fragments following elut,ion from gels (srcy above) were further purified by neutral sucrose gradient centrifugation or by G75 Sephadex (Pharmacia AR) gel filtration to remove solubilized a,garose or polyacrylamide particles which were found to interfere with hybridizations. Whole mouse embryo DNA was prepared from nuclei of uninfected secondary mouse embryo cells as described by Sharp et al. (1974). All DNA preparations were mechanically sheared into fragments about 500 basepairs long by sonication at 0°C for 2 to 4 x 30 s at 3.4 amp using an ultrasonic disint,egrat,ol (Son&r, model LS75, Branson Instruments, Inc.). To follow the rate of reannealing of 32P-labelled A2 or defective HpaII DNA fragments, 100 to 300-~1 portions of the sheared DNA in standard hybridization buffer (0.01 X25 pg denatured calf Tris.HCL (pH 7.5), 1 mM-EDTA, 0.050/ sodium dodecyl sulphate, thymus DNA/ml and 0.2 to I.0 M-NaCl) were denatured by heating to 105°C for 10 min (iIt a glvcerol/water bath) and then incubated for various lengths of time at 68°C. Pipurts legends indicate the cases where unlabelled A2 defective viral, whole mouse embryo or calf thymus DNAs were included in the incubation mixture. Incubation was terminated by chilling in ice and t)he samples were kept frozen until all portions of the experiment had been collect’ed. probe DNA which had reannealed was determined b) The percentage of s2P-labelled mt~ssuring the amounts of radioactivity resistant to digestion with the single-stranded specific nucleasc S 1 (from Aspergillus oryzae ; Sutton, 1971). For digestion, each portion was diluted with I.0 ml of Sl buffer (to give a final concn of 0.05 M-potassium acetatct (pH 4.5), 0.2 3x-NaCl, 1 mM-ZnCl,, and 6 mM-b-mercaptoethanol), 30 pg denatured calf thpmns DNA was added/ml, end the solution was incubated for 30 to 45 min at 50°C’ wit,h a sufficient amount of Sl enzyme (usually 0.5 to 1.0 pl) to completely degrade 30 pg of single-stranded DNA. (Sl nuclease was a kind gift from Dr R. Kamen.) After chilling in ice, trichloroacetic acid was added to the above solution to a final conen of 596 and thtx acltl-precipitable material was collected on Whatman GF/C filters, which werf: tllc?rl wa,shed with 5% trichloroacetic acid followed by ethanol and dried. The radioactivit,). was determined by liquid scintillation counting in POPOP/t,oluene scintillation fluid using a Packard Trioarb counter. The background for each set of experiments in these assays was determined by SI digestlion of portions which were removed at time zero of hybridization. This blank (1 t.o 5(!$) was subt’racted from the values of all other samples before calculating thcl percentage of reannealed DNA. The maximum amount of radioactivity of each probe

476

E. LUND,

M. FRIED

AND

B. E. GRIFFIN

DNA that could be rendered resistant to Sl nuclease was between 90 and 95% of the input radioactivity as determined from samples that contained a sufficient concentration of unlabelled homologous DNA to obtain complete reannealing of the probe DNA. This value of Sl-resistant radioactivity was defined as 100% double-strandedness of the probe and was used for calculation of percentage double-stranded probe DNA in all other samples of the same experiment. The specific activity (32P cts/min per pg DNA) of the probe DNAs and the nucleic acid concentration (M, moles of nucleotides/l) of the various viral DNAs are given in the Figure legends for each experiment. Since the rate of reannealing of a given DNA sequence in solution is proportional to its initial concentration (Britten & Kohne, 1968) the amount of probe DNA sequence in a heterologous DNA can be estimated by comparison of the rate constants of reannealing (C&t values) obtained for the probe DNA in the presence of the same molar concentration of homologous and heterologous DNAs being tested (cf. Sharp et al., 1974). The results are presented as a plot of the reciprocal of the fraction of probe DNA remaining single-stranded, Ilfs,(1/(1-f,,)), W~T.YUS the product of probe nucleic acids concentrations (C,) and the time of hybridization (t), (mol nucleotides x s/l). In the cases where the nucleic acids concentrations of the viral tester DNAs were such that the actual lnolar concentrations of homologous and heterologous DNA being tested were different, the Cot values (in the presence of the heterologous DNA) were normalized to those of the homologous DNA by multiplication with the ratio between the molar concentrations of heterologous and homologous tester DNAs. (e) Depurinution

jkgerprint

analysis

The depurination of 32P-labelled DNA fragments and the 2-dimensional separation of the resulting pyrimidine tracts were carried out essentially as described by Ling (1972). The 1stdimensional separation was by electrophoresis on cellulose acetate strips at pH 3.5 (5000 V, 15 to 30 min) and the Bnd-dimensional separation by homochromatography on thin-layer DEAE-cellulose sheets (CEL 3000 DEAE/HR 2/15 Macherey-Nagel and Co., purchased from Camlab, Cambridge) (20 cm x 20 cm, or 20 cm x 40 cm) by ascending chromatography for 3 to 6 h at 60°C using homomix C (Brownlee & Sanger, 1969). The fingerprint patterns were detected by autoradiography.

3. Results (a) Analyses with restriction endonucleases (i) Goss analysis In a preliminary paper (Fried et al., 1974), the structures of the polyoma defective DNA species D92, D91 and D76 have been reported based on analysis with HpaII and Hind111 restriction endonucleases. The HpaII cleavage patterns of these defective DNAs as well as that of the non-defective A2 strain are shown in Figure 1. In summary the three defective DNAs produced fragments corresponding in size to the A2 Hap11 fragments 6, 1 and 3 (Griffin et al., 1974). By a number of criteria no differences were detected between these HpaII fragments and those of A2 DNA. In addition, four fragments were detected in the HpaII cleavage pattern of D92 and D91 DNA and two in the HpaII cleavage pattern of D76 DNA which were not present in the cleavage pattern of A2 DNA. In the case of D92, these fragments migrated with electrophoretic mobilities corresponding to apparent sizes of 20% (one-molar yield), 10.5% (one-molar yield) and 4.5% (two-molar yield)?. For D91, the fragments had apparent sizes of 19%, 10.5% (one-molar yield each) and 4.5% (two-molar yields). For D76, the fragments had apparent sizes of 19% and 5.5% (one-molar yield each) (see Table 1) t Unless otherwise stated, fragment the A2 wild-type

genome.

sizes of the defective

molecules

are given as percentages

of

POLYOMA

VIRUS

HpulI D92

D91

DEFECTIVE

cleavage

DN.4s

477

patterns 076

A2

Hpull

D91 ID92

/fpon -2 (21 *4%)

0% 0%

D92 1 D761 ‘D91 1

HpdI

-I (27 - 3%)

HpoII-3

(16 -8%)

.HpaII-4

(131.2%)

. Hpull-5 -HpuE-6

(7, 7%) (6, *7 %)

-HpaII-7

(5 ‘02%)

. HpaII-

8 ( Is*8%1

*5%

D9 11ID92

-

Blue

FIG. 1. An autoradiogram of the products obtained by cleavage of defective (D92, D91 and D76) and A2 DNAs (3aP-labelled) with HpaII. Fragments were separated on a slab gel (20 cm x 40 cm) composed of 3.0% polyacrylamide/O~lS”/c bisacrylamide and 0.6% agarose. Fragments from A2 DNA are shown (right); they are numbered and sizes given as previously reported (Griffin et al., 1974). D92, D91 and D76 DNAs gave fragments with mobilities identical to A2 fragments HpaII-1, 3 and 6. In addition, D92 DNA gave a 20% and a 10.5% fragment (each in l-molar yield) and a 4.6% fragment (in 2-molar yield). The largest defective specific fragment from D91 DNA was 19% (1 -molar yield) ; other fragments in D91 were identical to those found in D92. D76 DNA also gave a 19% fragment (l-molar yield) and a 5.5% fragment (l-molar yield).

HpaII

27.3 21.4 16.8 13.2 7.7 6.7 5.2 1.8

;---52kgii-Fragment length (%ofA2 genome)

HpaIIHpaII-2 HpaII-3 HpaII-4 HpaII-5 HpaII-6 HpaII-7 HpaII-8

cleavage products fragment

1

of

Xummary

HpaIIHpaII HpaII-3 HpaII HpaII-6 HpaII

27.3 20.0 16.8 10.5 6.7 4.5t D92

D92

D92

HpdI cleavage products fmgment

D92 *-

4.5%

10.5%

1 20%

of

endonuclease

ggment length (% of A2 genome)

of restriction

TABLE

27.3 19-o 16.8 10.5 6.7 4.q

Gagment length (% of A2 genome)

D91

D91

D91

*-

D91

HpaIIHpaII HpaII-3 HpaII HpaII-6 HpaII

4.5%

10.5%

1 19%

Hpa II cleavage products of fragment

data on A2 and defective BNAs

1

27.3 19.0 16.8 6.7 5.5

I Fragment length (% ofA genome)

D76

D76

D76 n

HpaIIHpaII HpaII-3 HpaII-6 HpaII

HpaII cleavage products fragment

5.596

1 19”/b

of

-7

480

E. LUND,

M. FRIED

AND

B. E. GRIFFIN

By further restriction enzyme analyses, as well as fine structure depurination analyses (see below) no differences could be detected between the 4.5% HpaII fragments of D91 and D92, between the 10.5% HpaII fragments of D91 and D92 and between the 19% HpaII fragments of D76 and D91. These fragments will therefore be referred to as D91/D92 HpaII 4*5%, D91/D92 HpaII 10.5% and D76/D91 HpaII 19%. The D76/D91 HpaII 19% f ra g merit and the D92 HpaII 20% fragment were found to be almost identical. Restriction enzyme analyses were carried out on the defective DNAs using HindIII and HhaI. D76 was cleaved by Hind111 and by HhuI into two fragments. D92 and D91 were each cleaved by Hind111 into three fragments, and by HhuI into two fragments. Polyoma A2 DNA is cleaved by these enzymes into two and three fragments, respectively. In the fragments unique to the defectives, Hind111 cleavage sites were found to lie in the D76/D91 HpuII 19% fragment and the D92 HpuII 20% fragment, and in the D91/D92 HpuII 10.5 “/A fragment. An HhaI cleavage site was found in the D76/D91 HpuII 19% fragment and the D92 HpaII 20% fragment. The number and size of fragments generated by cleavage of the three defective DNAs and A2 DNA with Hind111 and HhuI are summarized in Table 1. The fragments generated by redigestion of the isolated Hind111 and HhuI fragments from the various DNAs with HpuII are also presented in Table 1. From the restriction endonuclease analyses of HpaII fragments using HindIII and HhuI (which each cleave A2 HpaII-2 once) either individually or in combination, the D92 HpuII 200/b fragment appeared to contain sequences identical to A2 HpuII-2 starting from the Hind111 cleavage site at 1.4 map units and extending to the HpuII 2-6 junction at 19.9 map units. Similar results were obtained for the D76/D91 HpuII 19% fragment except that this fragment appeared to be missing 1 y. of the sequences which lie between the HhuI site at 14 map units and the HpuII 2-6 junction (see Griffin, 1977). This 1% deletion has been found to include one of the two XbaI sites present in polyoma virus DNA and located at 17.5 and 18.2 map units on the physical map (see Fig. 2) (Fried & Griffin, unpublished observations). Analysis of the D91/D92 HpuII 105% fragments showed them to contain HindIII, MboI, MboII, and Hph restriction endonuclease sites in the same relationship to one another as found in A2 HpaII-2 (or in the D92 HpuII 20% and the D91/D76 HpuII 19% fragments). This provided evidence that at least 40% of the D91/D92 HpuII 10.5% fragments had sequences which were identical to those which lie between about 1 and 5 map units in A2 HpuII-2 (Griffin, unpublished results). The analyses using either successive cleavage with a number of restriction enzymes (as reported above or as summarized in Table 1) or partial digestion with HpuII followed by redigestion (to completion) of isolated incomplete digestion products (as used by Griffin et al., 1974 for A2 DNA) have led to the order of the HpuII fragments of the defective DNAs. These data were used to construct the circular HpaII physical maps of D92, D91 and D76 shown in Figure 2. Analysis of the fragments produced by cleavage of the D91 and D92 DNAs with Bum1 gave results which were consistent with the proposed physical maps (Fig. 2). Bum1 cleaves A2 DNA into four fragments (Griffin, 1977). Bum1 cleaved D92 into five fragments, two of which (58.6% and 2*80/O) were identical in size to fragments produced from A2 DNA; the 2.8% fragment was present in greater than one-molar yield in the defective species. The other three fragments from D92 DNA (about 9.5%, 7.7% and 1.7%, the latter in greater than one-molar yield) clearly differed in

FOLYOMA

VIRUS

DEFECTIVE

DNAs

4x1

Mndrn

D76

0 H/ndrn

2. Physical maps of polyoma A2 wild-type DNA and the cloned dcfect,ives D92, 1191 FIG. and D76. The A2 map is divided into 100 units with the single E’coRl cleavage site at zero map units. The map units shown on the physical maps of the defective DNAs correspond t,o A2 unit,s. Thr% origins (0) of viral DNA replication are indicated on all the maps. AZ: Physical maps obtained by cleavage with restriction endonucleases HpnIl (8 cleavage sit,es, Griffin et nl., 1974, inner ring) and HaeIII (24 cleavage sites, Griffin, 1977, outer ring). The cleavagta sit)es for BamI, Hind111 and HhaI are shown. This map was modified from Griffin (1977). D92: Proposed physical map obtained by cleavage with restriction endonuclease HpctIl (7 cleavage sites) (Fig. 1). This species contains contiguous viral sequences from 1 t,o 72 map units and rearranged sequences from HpaII fragments 2, 3 and 6 as shown (see text,). D91: The proposed HpaII physical map is identical to that of D92 except for a del&on of approx. 1% of sequences between 14 and 19.9 map units in HpaII-2 (indicated by V). D76: Proposed physical map obtained by cleavage with restriction endonuclease HpnIl (5 cleavage sites) (Fig. 1). This species cont,ains viral sequences from 1 to 72 map units as in D91 and rearranged sequences from HpaII fragments 3 and 6 as shown (see text). In the defective DNA maps, the HpaII cleavage sites are indicated by continuous lines wit,hin thr circles. Broken lines show where non-contiguous viral sequences have been covalently linked (see text) and the numbers indicate the A2 HpaII fragments from whioh these sequences are derived.

E. LUND,

482

M. FRIED

AND

B. E. GRIFFIN

size from the A2 fragments. D91 differed from D92 (by 1%) in the size of the largest Bum1 fragment. The EcoRI cleavage site at 0 map unit in A2 DNA (in HpaII-2, Griffin et al., 1974) was not present in any of the three defective DNAs (Fried, 1974). The single Hue11 cleavage site which has been located at 72.4 map units on the A2 physical map (Griffin, 1977) in HpaII-5 was also absent in D91, D92 and D76, whereas the single Barn1 cleavage site which has been located at 59 map units (Griffin & Fried, 1976) in HpaII-3 was present in the three defective DNAs. (ii) Detailed analysis HaeIII endonuclease cleaves polyoma A2 DNB into 24 fragments which have been ordered relative to the HpaII physical map (Griffin, 1977, see Fig. 2). For analysis of the fine structure of the defective DNAs, the cleavage patterns produced by Hue111 were studied. The products obtained when the HpaII fragments specific to the three defective molecules were redigested with HaeIII, as well as those obtained by Hue111 cleavage of A2 HpaII-2, 3 and 5 are shown in Figure 3(a) and (b) ; the results of these analyses are discussed below, and summarized in Table 2. (a) HaeIII cleavage of D76/91 HpaII 19% and 092 HpaII 20% fragments Both fragments gave one HaeIII product which appeared identical to A2 HaeIII-3 (see Griffin, 1977). Both had a fragment (about 3.7%) which was slightly smaller than A2 HaeIII-8 but like the A2 fragment contained a Hind111 cleavage site. D92 HpaII 20% produced a fragment identical in size to A2 HaeIII-7, whereas D91 HpaII 19% gave a slightly smaller fragment. No fragment corresponding to the HaeIII product in A2 HpaII-2 from the A2 HpaII 7-2 junction was found in the defective species. The results are summarized in Table 2 and allow conclusion about the composition of the defective fragments which are consistent with those reached above. (b) HaeIII cleavage of D91/092 HpaII 104% fragments Three fragments were obtained, one of which was identical in size to A2 HaeIII-14 and another to A2 HaeIII-17, both of which come from A2 HpaII-3. The third fragment (about 3*7%, in two-molar yield) was slightly smaller than HaeIII-8 from A2 DNA and indistinguishable in size from the smallest HaeIII fragments described in (a) above. Only one of the two 3.7% species contained a HindIII site (see Table 2). (c) HaeIII cleavage of D76 HpaII 5.5% and D91/92 HpaII 4.5% fragments The HaeIII cleavage pattern of these defective specific fragments showed three products which were common to both. Two of these had the same apparent sizes as A2 HaeIII fragments 14 and 17 and the third was intermediate in size between these two fragments. The 1 y. extra sequences in the D76 HpaII 5.5% fragment were found in a single Hue111 fragment (Table 2, Fig. 3). (b) 1. A2 HpaII-2 2. D92 HpaII 3. D91 H&I In (19: In slab

cleaved with HaeIII. 20% cleaved with HaeIII. 19% cleaved with HaeIII.

(a) fragments were separated by electrophoresis on a 4%/7% polyacrylamide/bisecrylamide 1) slab gel (20 cm x 40 cm). (b) fragments were separated by electrophoresis on a 7% polyacrylamide/bisaorylamide gel (10 cm x 20 cm).

POLYOMA

VIRUS

DEFECTIVE

L)NAs

4x3

12345678

4 % 7%

(a)

(b)

FIG. 3. Autoradiograms of isolated 3aP-labelled HpaII fragments of A2, D92, D91 and D76 DNAs (see Fig. 1) after cleavage with HaeIII. For sizes of A2 Hue111 fragments, see Griffin (1977). For sizes of the HneIII fragments obtained from the defective DNAs, see Table 2. (a) 1. D92 HpaII

20% cleaved with HaeIII.

2. A2 HJXZII-2 cleaved with HaeIII (contains in order to show size relationships). 3. D92 HpnII

10.5% cleaved with HaeIII.

4. D91 HpaII

10.5% cleaved

5. A2 HpaII-3

cleaved with H&II.

with

HaeIII.

6. D91/D92 H@I 4.5% cleaved with HaeIII. 7. A2 HpaII-5 cleaved with HaeIII. R. 1176 HpteII

5.5% cleaved

with HaeIII.

a small amount

of the material

from column

1

E. LUND,

484

M. FRIED

AND

B. E.

GRIFFIN

TABLE 2

Xummary of HaeIII restriction endonuclease data on HpaII fragments from A2 DNA (in part) and 092, D91 and D76 Fragments produced by cleavage with HaeIIIt

1.2% (HaeIII-1) HaeIII-3 HaeIII-8 HaeIII-7

A2 HpalI-2

D76/D91

20%

2.2% + 1.7% -

HpaII

19%

-

2.2% + 1.5yo

3.7% 3.7%$ HaeIII-14 1.2% (HaeIII-17)

D91/D92

HpaII

4.6%

1.4% HaeIII-14 1.2% (HaeIII-17)

-

-

3.7% HaeIII-3 5.0%

10.5%

5.5%

-

2.2% + 1.5yo -

HpaII

0~20/ + 5.8%

3.7% HaeIII-3 HneIII-7

D91/D92

D76 HpaII

(% of A2 genome) of fragments by redigestion of HaeIII products (1st column) with HindUI HhaI

0.3% (HaeIII-9) HaeIII-12 HaeIII-5 HaeIII-15 HaeIII-14 1.2% (HaeIII-17)

A2 HpaII-3

D92 HpzII

Length produced

0.2% f

5.8%

0.2% + 4.8% 2.2% + 1.5% -

1.4% HaeIII-14 1.2% (HaeIII-17) 1.0%

-

-

-

t The sizes of the HaeIII fragments ore as given (Griffin, 1977) (see Fig. 2 for the HaeIII map of A2 DNA). $ Only one of the 3.7% HaeIII fragments in the D91/D92 HpaII 10.5% fragments is cleaved by HindIII. -, Denotes that HaeIII fragment (1st column) is not further cleaved.

(b) Depurination jingerprint

analyses

32P-labelled polyoma DNA, followed by twoChemical depurination of full-length dimensional separation of the resulting products produces a pattern (fingerprint) in which at least ten specific oligonucleotides present in only one-molar yield can be identified. These have been correlated with different regions of the DNA on the polyoma physical map (Griffin, 1977). This method has been used to characterize the defective DNAs discussed in this paper. The depurination fingerprints of the purified 19%, 10.5% and 4.5% defective specific HpaII fragments from D91 and the 5.5% fragment from D76 were compared

POLYOMA D76/D91

VIRI

JS DEFECTIVE

HpoJI-19%

*

DNAs

48.5

A2 HpaII-2

Elfxtrophoresls,

pH 3.5

Fro. 1. A comparison of fingerprints of A2 HpaII-2 and D76/D91 HpaII 19% fragments. The products were separated by the 2.dimensional procedures indicated in Materials and Mrt>hods. (The fingerprint of the D92 HpaII 20% fragment was indistinguishable from that, of the 1176/D91 HpaII 19% fragment.) The defective specific fragments lack a cytidine-rich oligomer indicated by (*) which is present in HpaII-2 (see text). In addition, the defective specific fragment contains the unique oligomer from A2 HpaII-5 (see Fig. 6) which is found near the origin of IIN. replication (Griffin, 1977). This oligomer is indicated by 6.

with depurination fingerprints of the eight HpaII fragments from A2 DNA, and with relevant fingerprints of A2 HaeIII fragments. Some of these patterns are shown in Figure 4. The fingerprints of each of the defective fragments were different from t’hosc of any A2 HpaII fragment (Griffin, 1977), but nonetheless could be analysed. (i) D76jD91 Hpa’II

19% and 092 HpaII

20% fragments

(Fig.

4)

The D76/D91 HpaII 19% fragments (or the D92 HpaII 20% fragment) were similar in most respects t,o A2 HpaII-2. One notable difference however was the presence of a large oligopyrimidine in the defective fragment which was not found in 82 HpaII-2. This has been shown to be identical in size (and mobility) to the largest oligomer in A2 HpaII-5, and has been further locatted in A2 HaeIII-14’ (see Griffin, 1977). Another difference was the absence in the defective fragments of a large

A2 HpalI-3

Fla.

cl91 ID92

10.5%

5

Electrophorests,

HpoII-

pH 3.5

A2 HpoII-5

POLYOMA

VIRUS

DEFECTIVE

‘hi7

DNAs

(greater than decamer size) cytidine-rich oligomer normally present in A2 Hpall-2. In more detailed studies (Lund et al., 1977), this oligomer has been shown to come from HaeIIl-8, a fragment which contains both the EcoRI site and one of the HindIll sites in A2 DNA (see Table 2 and Fig. 2). This technique suggests that some (but not all) sequences from A2 HpaII fragments 2 and 5 are present in the fragments from the defective species. (ii) D91/092

HpaII

10.5% fragments

(Fig.

5)

The D91/D92 HpaII 10.5% fragments appear from restriction endonucleasr analyses to be composed in part of sequences from A2 HpaII-2 and in part of sequences from HpaII-3 (Table 2). Like the D76/D91 HpaII 19% fragment discussed above, they contain the large oligopyrimidine from A2 HueIII-14’ (a subfragment of HpaII-5, see Fig. 2) and lack a cytidine-rich oligomer normally found in HpaII-2 near the EcoRI restriction site. In addition, the 1O*5o/ofragments contain sequences from A2 HpaII-3, which include an oligothymidine tract (T7 oL‘8) known to come from A2 HueIII-17 (see Griffin, 1977) and a pair of oligomers (9 or IO-nucleotides long) known to come from A2 HaeIII-14 (Lund et al., 1977). Thus, by this technique, the 10.5”/, fragments appear to contain sequences from A2 HpaII fragments 2, 3 and 5. (iii) D91/092

HpaII

4.5% and D76 HpaII

5.5% fragments

(Fig.

6’)

The depurination fingerprints from the D91/D92 HpaII 4.5% fragment and D76 HpaII 5.5”/0 fragment were shown qualitatively to be nearly identical. Both species contained thelarge oligopyrimidinefromA2 HueIII-14’, the pairofoligomers (9or lonucleotides long) from A2 HaeIII-14 and the T, ol‘ 8 oligothymidine tract from A2 HueIII- 17 (see above). In the defective species, a quantitative difference was however apparent in the pair of oligomers from A2 Hue111 fragment 14. This pair appear in equimolar yield in the defective HpaII 10.5% (above) and 4.5% fragments and in A2 HueIII-14 (Lund et al., 1977). In the 5.5% fragment, they appear in unequal amounts, the more cytidine-rich species being in higher yield. This might reflect an inherent error in thta technique but as it has been observed consistently, it probably reflects the presence in the 5.50,/, fragment of a repetition of the sequences found in the more cytidine-rich of the two species. (In addition, the 5.5% fragment was found to contain a C, tetramer and a C&T tetramer not present in the 4.5% fragment. Because of the small size of these oligomers, they could not be assigned to any specific region of the DNA.) Thus, the depurination technique suggests that the small defective specific HpalI fragments contain sequences from A2 HpaII-3 and 5. (c) Hybridization.

analyses

(i) Annlyses using A2 viral DNA probes The rate of reannealing of 32P-labelled A2 HpaII fragments was determined in the absence and in the presence of A2 and defective D92, D91 and D76 DNAs (see Fig. 7) as described in Materials and Methods. From the slopes of the lines obtained and Pra. 5. A comparison of fingerprints of A2 HpaII-3, A2 H&I-5 and D9l/D92 HpaII 10.69: fragments. The defective specific fragment lacks the cytidine-rich oligomer found in A2 HpaII-2 (see Fig. 4) and contains the unique oligomer (5) from A2 HpaII-5. In addition the defective specific fragment) contains 3 unique oligomers from A2 HpaII-3, indicated by 3, and at least 1 unique oligomer from A2 HpnIT-2, indicated by 2 (see Fig. 4).

488

E. LUND,

M. FRIED

AND

B. E. GRIFFIN

076 Hpatt-5.5%

4

D91 ID92

Electrophoresls,

HpoII-4.5%

DH 3.5

FIG. 6. A comparison of the fingerprints of D91/D92 HpaII 4.6% and D76 HpnII 5.5% fragments. Both species contain the unique oligomer (5) from A2 N@I-6 (see Fig. 5) and 3 unique oligomers (3) from A2 HpaII-3 (Fig. 5). Both of the 9 or lo-nucleotide long oligomers (designated-IX on the Figure) from HpaII-3 appear in equimolrtr yield in the D91/D92 HpaII 4.6% fragment but in unequal amounts in the D76 HpaII S.SOh fragment (see text). These oligomers are found in equimolar yields in the fingerprints of A2 HpaII-3 and D91/D92 HpaII 10.S”/o fragments (Fig. 4).

from the concentration of the various defective DNAs (relative to the concentration of the A2 DNA) the amount of probe DNA sequences in the defective species could be estimated. All three defective molecules, D92, D91 and D76, were found to contain approximately one copy of the DNA sequences from A2 HpaII fragments 1,3 (Fig. 7 (a) and (c) ) and 6 (data not shown). A similar result was obtained for A2 HpaII-2 (Pig. 7(b)). The rate of reannealing of A2 HpaII fragments 4, 5 and 7 was not affected by the addition of any of the defective DNAs, indicating that the defective D91, D92 and D76 DNAs lacked most (if not all) of these viral DNA sequences. As an example, the data for A2 HpaII-4 are shown in Figure 7(d). (A2 HpaII-8 was not tested in this analysis.)

POLYOMA

VIRUS

:01

0

4.0

0 /’.

i 3.0

4X!)

DNAs

c (b)

l

2-o I .o

DEFECTIVE

piI/

.

ry --h--

A-

-.A

2.2 4.0 3.0 2.0 I-0

i” L

1111111,111,

,,,,,,,/,,,, 5

IO

5

IO

Cot x IO’ (mol nucleotldes x s/l)

PIG. 7. Reannealing A:! DNAS.

of A2 HpnII

fragments

in t,he presence of defective

D76, D91 and 092 01

Portions of 32P-labelled A2 HpaII DNA fragments (spec. act. 1 x lo6 a?? cts/min per pg DNA) and u&belled viral DNAs were denatured and allowed to reanneal at 68°C for various times in 200 ~1 of standard hybridization buffer containing 0.2 M-NaCl. The rates of reassociation were monitored by nuclease Sl digestion and results were plotted as described in Materials and Methods. Each portion contained 4.5~ 10m9 M (mol nucleotides/l (32P-labelled fragment DNA). The reassociation kinetics (self-annealing) of A2 HpaII-1 (a), A2 HpnII-2 (b), A2 HpaII-3 (c) and A2 HpnII-4 (d) DNAs are indicated by (-A-A-). The reannealing of the same concentrations of fregmentDN,4 inthepresenceof 5 x ~O-‘M-A~DNA (a), 9.6 x 10-8~-D92DXA(~1, 3.3 x lo-‘M1191 DNA (J) or 1.5x 1O-7 M-D76 DNA (*) is also shown. This concentration of A2 UK.4 corresponds to a Wfold, 24.fold, 19-fold and 15.fold molar excess of homologous DNA seyuencns of A2 HpaII fragments 1,2,3 and 4, respectively. In the case of the defective DNAs the C”& values have been normalized to those of A2 DNA to compensate for the differences in concentration and molecular weight of the DNAs (see Materials and Methods).

From the analyses previously described, it appeared that the DNA sequences corresponding to A2 HaeIII-14 and 17 (subfragments of A2 HpaII-3) were present in several copies per molecule of defective DNA (four copies in D91 and D92 and two copies in D76 DNA). With the hybridization conditions used, the failure to detect an increase in the copy number of A2 HpaII-3 sequences is presumably a reflection of the location of the additional HpaII-3 sequences in the defective DNAs. In D91, D92 and D76 DNAs, they appear to be interspersed between other non-contiguous viral DNA sequences (for detailed consideration see Discussion). (ii) Analyses

using defective specijc

DNA

probes

In order to determine whether the defective specific HpaIL fragments contained both viral and host DNA sequences, similar hybridization experiments were carried out using these fragments as probes and measuring the rate of reannealing in the absence and in the presence of homologous (D91) and heterologous (A2 and whole

E. LUND,

490

M. FRIED

AND

B. E. GRIFFIN

mouse embryo) DNAs. The results are presented in Figure 8 and Table 3. Analysis of the data showed that: (1) D76/D91 HpaII 19% and D91/D92 HpaII 4.5% fragments both contained viral DNA sequences. Clearly, the rate of reannealing of each of these fragments was increased in the presence of A2 DNA (Fig. 8(a) and (b), respectively). Furthermore the reassociation kinetics observed for these fragments in the presence of A2 DNA

5-c 4.c 3.c 2.c I*C

--’

P

I

I

I

I

(b)

I

Cd)

5-c 4.0 3.0 2.0 I.0 I

I

I

5

I

IO

15

I

5

IO

Cot x IO5 (mol nucleotides x s/l)

FIG. 8. Reannealing defective DNAs.

of defective

specific

HpaII

DNA

fragments

in the presenoe of A2 and

The conditions of hybridization are described in the legend to Fig. 7. (a) and (b). The reassociation kinetics of 32P-labelled D91/D92 Spa11 4.5% and D76/D91 HpaII 19% fragment DNAs, respectively (speo. act. 3 x 10 5 32P cts/min per pg DNA) at a ooncn of 2.4 x 10e8 M (mol nucleotides/l) alone (A) and in the presence of 4.5 x 10m6 M-unlabelled D91 DNA (a) or A2 DNA (0). This concentration of homologous D91 DNA corresponds to a 39-fold molar excess of D76/D91 HpaII 19% DNA sequences and IS-fold molar excess of D91/D92 HpaII 4.6% DNA sequences, assuming that each molecule of D91 DNA contains 2 copies of the 4.5% fragment. (c) and (d). The reassociation kinetics of 32P-labelled D91/D92 HpaII 10.5% and A2 HpaII-4 fragment DNAs, respectively (spec. act. 3 x lo- 5 32P cts/min per pg DNA) at a concn of 1.2 x lOwa M (A). Also shown is the reannealing of the same concentration of probe DNAs in the presence of 1.6 x 1O-B w-unlabelled D91 (17) or A2 (a) viral DNAs. The concentrations of unlabelled homologous DNAs cor:eapond to a 15.fold molar excess of D91/D92 HpaII 10.5% DNA sequences (c) and to an la-fold molar excess of A2 HpaII-4 DNA sequences (d), respectively. In the oases, where A2 DNA is the “heterologous” DNA the C,t values have been normalized to those of the homologous D91 DNA (see Materials and Methods).

POLYOMA

VIRUS

DEFECTIVE

TABLE

3

of denatured D91/092 HpaII 10.5% and 45% fragmeds of homologous and heterologous DNAs

Reannealing

D91/D92

HpaII

irk the presence

y. of probe DNA rertnnealed in the presence of: (c) (4 (b) (4 D91 + CT WME DNA CT DNA A2 + CT DNA DNA

Probe DNA

1)91/D92 HpaII

491

DNAs

10.6% 4.5%

25

81

60

27

23

80

64

35

Portions containing 3 x low9 M (mol of nucleotides/l) of denatured 32P-labelled D91/92 HpaII 10.5% or D91/D92 HpaII 4.5% fragment DNAs (speo. act. 6.4 x lo5 cts/min 32P per rg DNA) in 360 ~1 of standard hybridization buffer (0.6 M-N&~) were allowed to reanneal for 5.3 h at 68°C:. The reassociation kinetics of the 32P-labelled probes were analysed in the presence of (a) 4.2 x lo- 3 nlcalf thymus (CT) DNA; (b) 4.2x 1O-2 M-CT DNA and 1.7 x lo-’ M-D91 DNA; (c) 4.2~ 10-3MCT DNA and 1.7 x lo-’ ~622 DNA; or (d) 4.6 x 10y3 M-whole mouse embryo (WME) DNA.

with, but do not prove that both of these fragments are composed of viral DNA sequences. (2) The rate of reannealing of D91/D92 HpaII 10.5% was also increased by the addition of A2 DNA, but to a lesser extent than observed with the defective specific 19% fragment, suggesting that only part of the probe DNA sequences were of vira,l origin (compare (b) and (c) of Fig. 8). To illustrate this further, the reassociation kinetics of D91/D92 HpaII 10.5% and A2 HpaII-4 (13% of full-length A2 DNA) were compared under identical hybridization conditions. As shown in Figure 8(c) and (d), the fragments alone reannealed at very similar rates, whereas in the presence of A2 DNA the wild-type DNA fragment (HpaII-4) was found to reanneal about twice as fast as the defective species. (3) The two defective specific HpaII fragments (10.5~~ and 45%) were also allowed to reanneal in the presence of whole mouse embryo DNA at a concentration which would allow the detection of repetitive, but not unique, host DNA sequences in these fragments. Since no increase in the rate of reannealing was observed (Table 3) it was concluded that, neither of these fragments contained repetitive host sequences. (4) In agreement with the HaeIII restriction analyses, the defective specific HpaII fragments appeared to be composed of sequences which are present in more than one copy per defective DNA molecule. This is illustrated by the much larger increase in the rate of reannealing of each of these fragments in the presence of the homologous D91 DNA than in the presence of equimolar amounts of heterologous A2 DNA (fol example, see Fig. 8(a)). were

consistent

entirely

4. Discussion The structures three independent tion mapping and appears to contain on the A2 physical similar continuous

of three related defective polyoma DNAs have been deduced using methods of analysis (restriction endonuclease cleavage, depurinaDNA-DNA hybridization). One defective species, designated D92. continuous viral sequences from approximately 1 to 72 map units map (see Fig. 2). The two others, designated D91 and D76, contain sequences from this region of the viral genome except that around

492

E.

LUND,

M.

FRIED

AND

B. E. GRIFFIN

18.0 map units approximately 1 y0 of the DNA has been deleted. Thus, these three defective DNAs contain the entire late region of the polyoma genome (25 to 71 map units) (Kamen & Shure, 1976) but are missing much of the early region. These findings agree with the heteroduplex results of Robberson & Fried (1974), which suggested approximately 70% homology between D91/D92 DNA and wild-type A2 DNA. In addition, all three of the defective molecules appear to contain non-contiguous viral sequences which are derived from portions of A2 HpaII fragments 2, 3 and 5. For D91 and D92 there appears to be a tandem repeat of about 15% of these mixed sequences. Each of the repeating units is composed of (1) a duplication of sequences which extend from about 67 to 72 map units (sequences from A2 HpaII-3 and 5), and (2) sequences from about 1 to 6 map units in HpaII-2 (see Fig. 2). This location of sequences from HpaII-2 is in agreement with those found in heteroduplex structures previously observed (Fig. 5(c) and (d) of Robberson t Fried, 1974). Tandem repeats of Pequences have been observed in other defective molecules of polyoma virus (Griffin & Fried, 1975; Lund et al., 1977). The arrangements of the sequences in the defective molecules show three interesting features. One feature is the linkage of sequences normally found at approximately 67 map units in A2 DNA (in A2 HpaII-3) to other regions of the genome. For D76, D91 and D92 these sequences are joined to (non-contiguous) sequences at 72 map units in A2 HpaII-5 ; in the latter two defectives they are also joined to sequences at 6 map units in HpaII-2. In other defective species (Griffin & Fried, 1975; Lund et al., 1977), sequences at 67 map units are linked to sequences elsewhere in the genome. It is possible that these sequences in HpaII-3 may represent a “hot spot” for recombination with other parts of the viral genome. On the other hand, they may be indispensable for the maintenance of the viral DNA (e.g., needed for viral DNA replication or for binding of capsid protein to form virus particles) and thus always be retained while other (dispensible) viral sequences can be deleted. A second feature is the linkage of sequences from about 72 map units (in HpaII-5) to sequences at about one map unit (in Hp.zII-2). This linkage appears to be present in other independently isolated defective species (Fried, unpublished results) but is not a feature of all polyoma defective molecules (Griffin & Fried, 1975). The fact that the same non-contiguous regions become covalently linked in a number of cases might suggest that some sequence homology exists between these two regions. Since both of these regions can be isolated in small restriction endonuclease fragments (HaeIII-14’ and the region between the EcoRI and one of the Hind111 sites) (Griffin, 1977), this hypothesis can be tested using direct sequencing methods. A third interesting feature is the presence of multiple copies of the sequences from around the HpaII 3-5 junction (about 67 to 72 map units). These sequences are found four times in D91 and D92 and twice in D76 (Fig. 2). The origin of viral DNA replication has been previously mapped at 71 f3 map units (Griffin et al., 1974). The only sequences common to the polyoma defectives previously studied (Griffin & Fried, 1975) and the D92, D91 and D76 defective species are those found between 67 and 72 map units. Since both sets of defectives replicate efficiently and have at least 99% of sequences which correspond to polyoma wild-type DNA sequences, the viral origin of replication can presumably be further defined to lie within this region. The three methods used for analysis each give limited amounts of information about the identity of the sequences in the defective species : (1) Analysis with restriction endonucleases with few cleavage sites in the viral

POLYOMA

VIRUS

DEFECTIVE

DNAs

493

genome (for example, HpaII, HhaI, HindIII, BumI, EwRI, XbaI and BamI), has proved useful for identifying long regions of continuous viral sequences. Treatment with an enzyme such as Hue111 which cleaves the viral DNA at many sites (Griffin, 1977) allows for a more detailed analysis of shorter regions of continuous viral sequences (e.g. in the defective specific HpaII fragments). Regions of rearranged sequences (or deletions and additions) do not always however give interpretable results when analysed by multicleavage enzymes like HaeIII. (2) Depurination fingerprint analyses allow for the identification of very short) stretches of viral sequences. This method is limited to pyrimidine tracts that occm only once (unique) or rerely in the viral genome. Two such unique tracts lie in the viral sequences that are alwa,ys retained in the defective DNAs, and are sometimes found in regions that have rearranged viral sequences. These tracts are located in wild-type A2 DNA around the HpaII 3-5 junction (Griffin, 1977), close to or within the region containing the origin of viral DNA replication. One of these is the large (1T to 20-long) oligopyrimidine from HaeIII-14’, a subfragment of HpaII-5, and the other is the T, 01‘ a oligothymidine tract in HaeIII-17, a subfragment of Hpa.II-3 (see Fig. 2). Pyrimidine tracts of small DNA fragments have been used to detect, duplication, and deletions of short stretches of sequences in some cases (see Results). (3) DNA-DNA hybridization in solution is theoretically capable of identifying viral sequences present in the defective DNAs independent of rearrangements. It appears, however, that the detection of short stretches of rearranged sequences is very dependent, on the size of the sheared DNAs used in the analysis (see below). Whereas the hybridization data prove to be consistent with the other methods of analysis in identifying the presence of long regions of continuous viral sequences (portions of HpaII-2, and all of HpaII-6 and 1, one copy of each) and the absence of other sequences (HpaII-4 and 7), there appears to be a discrepancy in the case of A2 HpaII-3 sequences. The hybridization data indicate that these sequences are found in only one-molar yield in the defective DNAs. However, the Hue111 restriction analyses and depurination fingerprints show that a fraction of the HpaII-3 sequences are present in several copies in each of the defective species. From the restriction endonuclease analysis it appeared that the “repeated” A2 HpaII-3 sequences consisted of sequences from HaeIII-17, 14 and (part of) 15, and correspond to about 150 to 175 base-pairs of DNA (Griffin, 1977). Within the defective specific fragments D91/D92 HpaII 45% and D91/D92 HpaII 1O*5o/o these sequences appear to be covalently linked, with sequences from A2 HaelII-15 being joined to DNA sequences derived from A2 HpaII-2 or HpaII-5 (see Fig. 2). The evidence for t’his covalent linkage comes both from restriction endonuclease and depurination anai,vses of the defective specific HpaII fragments. Thus, in the defective DNAs the “extra” HpalI-3 sequences are interspersed between otherwise non-cont:guous viral DNA sequences, Ln the hybridization experiment performed, all DNA preparations were sheared to fragments about 500 base-pairs in length. According to the structural arrangements outlined above, that means that on the average less than one-third of t,he sequences from each of the sheared defective DNA fragments, which actually contain rearranged portions of HpaII-3, will show homology with the A2 HpaII-3 probe. It seems likely then that homologous annealing between such defective DNA strands (which a,re present in much higher concentration than strands of the probe DNA) will predominate because the relatively short heteroduplexes (between A2 HpaII-3

494

E.

LUND,

M.

FRIED

AND

B.

E.

GRIFFIN

and the defective DNA) will be strand displaced to leave homoduplexes. The same argument would explain the failure of the hybridization experiments to detect “extra” copies of some of the sequences from A2 HpaII-2 and 5 in the defective DNAs (see Figs 2 and 7). Similarly, in the reciprocal hybridization experiments (that is, using the defective specific HpaII fragments as probes), it was observed that the increase in the rate of reannealing of the D91/D92 HpaII 1O6O/o fragment in the presence of A2 DNA was only about one-half that observed for the comparably-sized wild-type fragment, A2 HpaII-4, under identical conditions (see Fig. 8(c) and (cl)). However, reannealing to higher Cot values (data not shown) indicated that most (if not aZE)of the probe DNA sequencesactually hybridized to A2 DNA. Furthermore, if only half of the defective probe DNA sequenceshad been of viral origin a hyperbolic curve rather than a straight line would have been obtained in the plot of Figure S(c). Since the D91/D92 HpaII 10.5% fragment appears to contain viral DNA sequencesfrom several noncontiguous regions of the wild-type polyoma DNA (i.e. from A2 HpaII-2, 3 and 5, see above) it is suggestedthat under our conditions of hybridization the formation of A2 DNA homocluplexes would prevail over heteroduplex formation (between A2 and defective DNAs) and this would result in the relatively low rate of defective probe reannealing. With these findings in mind, it should be emphasized that a “negative result” in a reassociation test, for example the lack of increase in the rate of reannealing of a probe in the presenceof a given test DNA, doesnot necessarily mean that none of the probe sequencesare present in that DNA. As demonstrated here, additional methods may be needed before it can be established with certainty that all of the probe DNA sequencesare really absent in the DNA being analysecl. The hybridization data indicate that none of the defective specific HpaII fragments contain repetitive host DNA sequences(section (c), (2) above). The data clo not, however, rule out the presence of unique DNA sequences.Furthermore, considering the arguments presented above on the limits of detection by hybridization assays,it is clear that short repetitive host DNA sequenceswould have been difficult (if not impossible) to identify by this method. Interpretation of the sequencespresent in polyoma virus DNAs which contain rearrangements can be misleading if only data obtained by nucleic acid hybridization with DNA shearedto about 500 base-pairs are considered.It should bepointed out that conclusions about the absence of specific viral sequencesor the presence of unique host DNA sequencesin several simian virus 40 (SV40) defective DNAs appear to rely on such hybridization conditions (Brockman et al., 1973; Rozenblatt et al., 1973; Khoury et al., 1974; Frenkel et al., 1974; Lee et al., 1975). The defective speciesdescribed here were isolated after only two high-multiplicity passages(Fried, 1974) and they can be seento contain large amounts of continuous vira.1 sequenceswith no detectable amount of host DNA. These defective molecules may well represent an early stage in the evolution of a particular type of defective species. Furthermore, HpaII restriction patterns similar to those of D92 (or D91) have also been observed in independent plaque-purified virus isolates after limited passageat high multiplicity, which suggestthat the rearrangements found in these defective speciesmay have a unique type of selective advantage. It is clear that other types of defective molecules can also arise in virus stocks (Griffin t Fried, 1975; Lund et al., 1977). The mechanism for the generation of defective molecules is not

POLYOMA

VIRUS

DEFECTIVE

DNAn

495

clear, although the presence of exact tandem duplications of sequences is compatible with their generation by deletion from large oligomeric forms of DNA (Cuzin et al.. 1970: Martin et al., 1976). SV40 defective molecules also have some of the features described here. For a comparison of polyoma virus and SV40 defective species, see Fried & Griffin (1977). The defective species described here do not appear to be capable of transformation but are able to replicate efficiently and to be encapsidated in the presence of helper virus. (Studies on the transcription of these defectives are being carried out.) They provide some insight into the viral sequences which are essential for various viral functions. The types of rearrangements which may lead to selective advant,ages for a particular species can be deduced by determining the structures of different classes of defective molecules. REFERENCES

Blackstein,

M. E., Stanners, C. P. & Farmilo, A. J. (1969). J. Mol. Biol. 42, 301-313. Britten, 1~. J. & Kohne, D. E. (1968). Science, 161, 529-540. Brockman, W. W., Lee, T. N. H. & Nathans, D. (1973). Virology, 54, 384-397. Brownlee, U. G. dz Sanger, F. (1969). Eur. J. Rio&em. 11, 395-399. Cuzin, F., Vogt’, M., Dieckmann, M. & Berg, P. (1970). J. MoZ. Bid. 47, 317-333. Dingman, c‘. W. & Peacock, A. C. (1968). Biochemistry, 7, 659-668. Frenkel, N., Lavi, S. & Winocour, E. (1974). J’irology, 60, g-20. l+iod, M. (1974). ,J. Viral. 13, 939-946. Fried, M. & Griffin, B. E. (1977). In Advances in Cancer Research (Klein, G. & Weinhouse, S. eds). vol. 24, pp. 67-113, Academic Press, New York. Fried, M., Griffill, B. E., Lund, E. & Robberson, D. L. (1974). Cold Spring Harbor Symp. Quad. Biol. 39, 45-52. Griffin, B. E. (1977). J. Mol. Biol. 117, 447-471. Griffin, B. E. & Fried, M. (1975). Nature (London), 256, 175-179. Griffin, B. E. & Fried, M. (1976). In Methods in Cancer Research XII (Busch, H., ed.), vol. 12, pp. 49.-86, Academic Press, New York. Griffin, B. E., Fried, M. & Cowie, A. (1974). Proc. Nat. Acad. Sci., U.S.A. 71, 207752081. Hirt,, B. (1967). .J. Mol. Bid. 26, 365-369. Kamen, R. & Share, H. (1976). Cell, 7, 361-371. Khoury, G., Fareed, G. C., Berry, K., Martin, M. A., Lee, T. N. H. & Nathans, D. (1974). .J. Mol. Biol. 87, 289-301. Lavi. S. & Winocour, E. (1974). Virology, 57, 296-299. Lclc, T. N. H., Brockman, W. W. & Nathans, D. (1975). C’irology, 66, 53-69. Ling, V. (1972). .J. Mol. BioZ. 64, 87-102. Lund, R., Fried, M. & Griffin, B. E. (1977). J. Mol. BioZ. 117, 497-513. Mart,& M. A., Hawley, I’. M., Byrne, J. C. & Garon, C. F. (1976). I’iroloyy, 71, 28-40. Middleton, J. H., Edgell, M. H. & Hutchison, C. A. III. (1972). J. I/iroZ. 10, 42-50. Old, R., Murray, K. & Roizes, G. (1975). J. Mol. Biol. 92, 331-339. Radloff. R., Bauer, W. & Vinograd, J. (1967). Proc. Nat. Acad.Sci., U.S.A. 57, 1514--1521. Robberson, D. J,. & Fried, M. (1974). Proc. Nat. Acad. Sci., U.S.A. 71, 3497-3501. Rozenblatt, S., Lavi, S., Singer, M. F. & Winocour, E. (1973). J. Viral. 12, 501-510. Sharp, 1’. A., Sugdon, B. 85 Sambrook, J. (1973). Biochemistry, 12, 3055-3063. Sllarp, 1’. A., Pettersson, U. 8: Sambrook, J. (1974). J. Mol. BioZ. 86, 709-726. Sutton, Mr. D. (1971). Biochim. Biophys. Acta, 240, 522-531. Tltorne. H. V. (1968). J. Mol. BioZ. 35, 215--226. Tl~orncb. H. V.. Evans. .J. & Warden, D. (1968). Nature (London), 219, 728-730.