Multiple reiteration of a 40-bp nucleotide sequence in the inverted terminal repeat of the genome of a canine adenovirus

VIROLOGY

159, 76-83 (1987)

Multiple Reiteration of a 40-bp Nucleotide Sequence in the Inverted Terminal Repeat of the Genome of a Canine Adenovirus SERGE SIRA, MOUNIR G. ABOUHAIDAR,* Departments

of Microbiology

YUAN-CHING

and *Botany, University

Received November

of Toronto,

21, 1986; accepted

LIU, AND JAMES B. CAMPBELL’ Toronto, Ontario, Canada M5S IA8

March 8. 1987

The DNA of a vaccine strain of canine adenovirus type 1 [ICHV vaccine; Connaught Laboratories, Ltd.; CAV-1 (CLL)] has been cloned in plasmid pAT153 in the form of subgenomic BamHl digestion fragments. Analysis of the nucleotide sequences of cloned terminal fragments has revealed an inverted terminal repeat (ITR) with a minimum length of 198 nucleotides, including a tandem reiteration of the 40-bp nucleotide sequence from positions 14 to 53. The ITRs had the 5’CATCATCAAT. . . sequence typical of adenoviruses and the highly conserved sequence ATAATATAC (nucleotides 9-l 7) of human strains. Additionally, one BarnHI A clone (left terminus) contained three sequential copies of the 40-bp sequence, and two BarnHI C clones (right terminus) contained at least seven. These did not appear to be artifacts of cloning, since evidence was obtained that the multiple reiterations also occurred in DNA isolated from intact virus. By analogy with human adenoviruses, the repetitive sequence in the CAV-1 (CLL) genome encompasses the entire nuclear factor I (NFI) binding site of the origin of DNA replication. Additionally, the 40-bp nucleotide sequence was found to contain the sequence AGG(N),GCCTAA (nucleotides 27-39) which closely resembles the concensus sequence of the human adenovirus NFI binding site (TGG(N),_,GCCAA; nucleotides 25-381. It appears, therefore, that the Connaught CAV-1 vaccine contains reiterated copies of an essential part of the adenoviral origin of DNA replication. A mechanism is proposed for the generation of multiple reiterations of sequences in the right ITR, given an initial single tandem repeat in the left ITR. o 1987 Academic press, inc.

cleotides of which are homologous with the corresponding sequences in human adenoviruses (Shinagawa et al., 1983). Several restriction endonuclease cleavage profiles and restriction maps have been published (Assaf et a/., 1983; Darai et a/., 1985; Whetstone, 1985; Hamelin et al., 1986) but otherwise little is known of the genetic organization. As a first step in genetic analysis of a vaccine strain of CAV-1, we have cloned the viral DNA in plasmid pAT153 in the form of subgenomic BarnHI digestion fragments. During analysis of cloned terminal fragments, we observed that several contained multiple reiterations of nucleotides 14-53. In the present paper, we report on the isolation of these clones and on the nature of these novel reiterations.

INTRODUCTION The adenovirus (Ad) genome consists of a linear double-stranded DNA molecule with a molecular weight of 20-30 X lo6 Da. Studies with human adenoviruses have demonstrated the presence of a 55. kDa protein linked to the 5’terminus of each DNA strand by a phosphodiester bond between a deoxycytidine nucleotide and the /3-hydroxyl group of a serine residue (Rekosh et a/., 1973; Carusi, 1977; Desiderio and Kelly, 1981). Additionally, the genomes of all adenoviruses analyzed so far possess an inverted terminal repetition (ITR) of the form abc. . . c’b’a’, which plays an important role in DNA replication (Shinagawa and Padmanabhan, 1980; Challberg and Kelly, 1982; Shinagawa et al., 1983). Canine adenovirus type 1 (CAV-l), or infectious canine hepatitis virus (ICHV), is a member of the genus Mastadenovirus (Koptopoulos and Cornwell, 1981). Although the genome has not been as well characterized as some of the human adenoviruses, it appears to have a typical adenoviral structure. It is not degraded by X exonuclease, indicating that the 5’ ends are blocked (Darai et al., 1985). The genome has been reported to possess an identical ITR of 160 bp, the first 17 nu’ To whom correspondence addressed. 0042-6822187

MATERIALS

$3.00

METHODS

Virus Infectious canine hepatitis virus vaccine [Connaught Laboratories, Ltd. (CLL), Willowdale, Ontario] was obtained through the courtesy of Dr. K. F. Lawson, Connaught Research Institute, who also provided the following history. The vaccine strain, which is no longer commercially produced, originated as the “W” strain of canine adenovirus type 1 (Cornell University) and had been passaged mainly in porcine kidney cells be-

and requests for reprints should be

Copvigtit 0 1987 by Academac Press. Inc. All rights of reproducton I” any form reserved

AND

76

ADENOVIRUS GENOME 40-bp NUCLEOTIDE SEQUENCE

fore adaptation to a dog kidney cell line (DK 6722). It was received by us at the 41st DK passage level and was used following 2-3 further passages in DK cells. In the text, it is referred to as CAV-l(CLL). A virulent strain of canine adenovirus type 1, CAV1(Glaxo), was kindly provided by Dr. C. Povey, Langford, Inc., Guelph, Ontario. Growth and assay of virus Virus stocks were grown in DK cells maintained as monolayer cultures in or-MEM supplemented with 10% heat-inactivated fetal calf serum (FCS), penicillin (100 IU/ml), and streptomycin (100 pg/ml). Roller cultures at 37’ were infected with virus at a multiplicity of infection of approximately 10 PFUlcell. When cytopathic effects were extensive, the supernatants were harvested, clarified by low-speed centrifugation (1000 g) for 15 min, and stored at -70” until required. Infectious virus titers were estimated by plaque assay in DK cell monolayers (0.65% Noble agar/MEM/lO% FCS overlay; 7 days at 37” in a humidified, 5% CO, atmosphere before staining with 0.1% neutral red). Extraction

and analysis of viral DNA

Virus stocks were concentrated approximately 50fold by tangential-flow membrane ultrafiltration (Pellicon system, Millipore Corp.). Following adjustment to a density of 1.34 g/cm3 with cesium chloride, aliquots were centrifuged for 17 hr at 64,000 g (Beckman Ti60 rotor, 30,000 rpm, 17”). Two clearly separated bands resulted, and the lower (p = 1.34 g/cm3), containing complete virus particles, was removed and dialyzed against 10 mM Tris-HCI, pH 8.0, containing 1 mM EDTA and 2 mn/l2-mercaptoethanol. The concentration of virus particles was estimated by absorbance (OD) at 260 nm, I OD unit taken as being equivalent to 10” particles (Halbert et a/., 1985). Viral DNA was extracted by the method of Pettersson and Sambrook (1973). This involved treatment with proteinase K (Sigma), which removes most, but not all, of the genomic 5’-termi’nal pratein. This step was necessary prior to restriction endonuclease analysis, since the pre,sence of the intact protein interfered with the migration of term’inal restriction fragments in agarose gels (Sharp et al., 1976). Restriction enzymes were purchased from several supplie,rs, and were used according to the manufacturer’s recommendations. Molecular

cloning

of the genome of CAV-I (CLL)

Viral DNA (5 yg) was digested with BarnHI, extracted with phenol saturated with 0.1 IVI Tris-HCI, pH 8.0, precipitated with ethanol, and redissolved in distilled

77

H20. It was then made 100 mM with respect to NaOH, and incubated at 37’ for 3 hr. ,Foilawing addition of Tris-HCI, pH 7.6, to 50 mfl/l, the reaction mixture was neutralized with HCI. This resulted in a 5amHl DNA digest with the 5’-terminal protein completely removed. The DNA digest was precipitated with ethanol, redissolved, and ligated for 17 hr at 16” with an excess of 5’-32P-labeled BarnHI linker (dGGGATCCC; Boehringer Mannheim). Incubation buffer was 50 mM Tris-HCI, pH 7.5, containing 10 mM MgQ, 20 m&+dithiothreitol, 1 mM rATP, BSA (10 pg/ml), and 2 mM spermidine. The ligation products were then extracted with Tris buffer-saturated phenol, digested with.BamHI, folowing which a sample was removed for analysis by agarose gel electrophoresis and autoradiography. Excess linker was removed from the remainder by two ethanol precipitations in the presence’of 2 NI ammonium acetate, followed by Sephadex G-50 spin-down column chromatography. Plasmid pAT153 DNA (plasmid kindly provided by ,Dr. V. L. Chan) was linearized by digestion with BarnHI and dephosphorylated with calf intestine phosphatase (Boehringer Mannheim). This plasmid DNA was then ligated to the BarnHI-digested (alkalitreated) viral DNA and us,ed to transform Escherichia co/i HBl 01 (Maniatis et al., 1982). Ampicillin-resistant transformants were setected in 2 ml LB broth cultures in the presence of 50 pg/ml ampicillin. Plasmid DNA from the transformants was isolated by an alkaline lysis procedure (Birnboim and Daly, 1979) and putative clones were identified by BarnHI digestion and Southern hybridization. Southern

hybridization

BarnHI-digested plasmids containing putative viral inserts were electrophoresed on 0.7% ag,arose gels, transferred to nitrocellulose filters, and hybridized with a BamHl digest of CAV-l(CLL) DNA which had been labeled with r2P]dATP using the random primer labeling technique (Feinberg and Vogelstein, 1983). The hybridization conditions were as described by Maniatis et a/. (1982). Briefly, fitters were tresited for 4 hr at 45” with prehybridization solution consisting of’50% deionized formamide, 5~ SSPE (SSPE = 0.15 M NaCI, 0.01 M NaH2P04, and 0.001 M EDTA, pH 7.4), 5X Denhardt’s solution, 0.1% SDS, denatured calf thymus DNA (100 Kg/ml), and tRNA (100 rug/ml). Labeled probe was added and the filters were incubated at 45” for 20 hr and then washed successively in 2X SSC (SSC = 0.15 M NaCl and 0.015 n/l sodium citrate, pH 7.01, 0.5% SDS at room temperature; 2x SSC, 0.1 o/oSDS at room temperature; and tw,ice in 2X SSC, 0.5% SDS at 65”. The nitrocellulose fitters were then dried and exposed to X-ray film (Kodak XAR-5).

SIRA ET AL.

78

DNA sequencing

of cloned termini

of CAV-1 (CLL)

DNA sequencing was performed directly on covalently closed plasmids using the dideoxy termination method of Sanger et al. (1977). The DNA was sequenced using both forward and reverse BarnHI primers (New England Biolabs) and [35S]dATP (NEN).

profiles

Figure 1 shows the BarnHI cleavage profiles of DNA from two strains of CAV-1, Glaxo and CLL. Digestion of CAV-l(CLL) DNA resulted in three bands, one of which (BarnHI C) was found to exist as a diffuse doublet. This was more clearly shown upon prolonged electrophoresis (Fig. 1B, lane 1). Plaque-purified stocks of CAV-l(CLL) produced more sharply defined BarnHI C bands, with mobilities similar to either one or the other doublet or with an intermediate mobility (e.g., Fig. 1 B, lane 2). The Glaxo strain DNA (lane 3) showed no evidence of a doublet in its BarnHI C band, although this had a slightly greater mobility than the corresponding CAV-1 (CLL) one. The Glaxo strain additionally revealed a small 4th band of approximately 500 bp (BarnHI D). Shinagawa et al. (1983) have also reported a 453-bp D fragment in CAV-1, strain Woe-4. As with BarnHI, digestion of CAV-l(CLL) DNA with C/al and Sa/l revealed doublet bands. Maps of the genome, derived from end-labeling and double-digestion data, showed that the doublets were in each case asA

1

2

3

A

I

B

1

2

3

FIG. 1. BarnHI digestion profiles of DNA from CAV-1, strains CLL and Glaxo. Lane 1, CAV-l(CLL); lane 2, plaque-purified CAV-l(CLL); lane 3, CAV-l(Glaxo). (A) Electrophoresis for approximately 2.5 hr. (B) Electrophoresis for approximately 5 hr. The doublet BarnHI C band in CAV-l(CLL) is clearly visible (lane 1). Molecularweight markers were derived from a Hindlll digest of XDNA.

,

A

BarnHI

#

A

BarnHI

a

Sal I

t

I

a

c

, ml.

e21

CLL

, w,,

%I

CLL

D \(

a

A

c

GM0

1CICV ozl cl-L RT

LT

RESULTS Restriction endonuclease and cleavage maps

aa

FIG. 2. Physical maps of the DNA of CAV-1, strains CLL and Glaxo. The subscripted letters refer to the two major components of the diffuse doublet band (see text for details). LT = left terminus; RT = right terminus. Map orientations were derived by designating BamHl A as LT on the basis that it has been reported to transform rat kidney ceils (Yoshida eta/., 1984) and the transforming (El) region of human adenoviruses is located on the left (Sussenbach, 1984).

sociated with the right-hand terminal fragments (Fig. 2). Apart from the doublets, the CAV-1 (CLL) maps are similar to those reported by Darai et a/. (1985) for CAV1 strain Behring H.c.c. 269. Studies with cloned CAVl(CLL) BarnHI fragments further digested with Sal1 clearly demonstrated that the BarnHI D fragment hybridized only with the left end of BarnHI C (data not shown). Hence, the Glaxo strain has a third BarnHI site about 500 bp downstream of the BarnHI 6 restriction site. Molecular

cloning

of the genome of CAV-1 (CLL)

The intact CAV-1 (CLL) genome was found to be susceptible to degradation by E. co/i 3’-exonuclease III but resistant to the action of 5’-X exonuclease. This is consistent with the observations of Darai et al. (1985) for the Behring H.c.c. 269 strain. In addition, the CAVl(CLL) genome was found to be inaccessible to the action of polynucleotide kinase both before and after treatment with alkaline phosphatase and was also refractory to ligation. The native CAV-l(CLL) genome (DNA-protein complex) also exhibited the limited electrophoretic mobility characteristic of an adenovirus genome with a terminal protein (Sharp et a/., 1976) (data not shown). Removal of this protein with proteinase K and alkali treatment (Materials and Methods) permitted blunt-end ligation of 5’-32P-labeled BarnHI linkers to the terminal fragments BarnHI A and BarnHI C of a BarnHI digest. In one experiment, ligation of the above-treated BarnHI fragments to the plasmid vector pATl53, and transformation of E. co/i HBl 01, resulted in the selection of three BarnHI A, six BarnHI B, and three BarnHI C transformants. The BarnHI-digested plasmids were transferred to a nitrocellulose filter and probed with a [32PldATP-labeled BarnHI digest of CAV-1 (CLL), with the results shown in Figs. 3A, B. Each of the inserts in lanes 3-14 hybridized with the probe, thereby dem-

ADENOVIRUS

GENOME

40-bp

onstrating their viral origin. Examination of Fig. 3A reveals that there are small but distinct size differences in the cloned BamHl C fragments (lanes 12-l 4), estimated to be approximately 300 bp using the X HindIll DNA digest as molecular weight marker. The cloned BarnHI C fragments appeared to comigrate with one or the other component of the BamHl C doublet apparent in the BamHl digest of viral DNA (lane 2). Nucleotide

sequence

analysis

From the isolated clones, the terminal fragments were sequenced as described under Materials and Methods. The suffixes “/A” and “/C” identify the BarnHI fragment from which the inserts were derived. The nucleotide sequence banding patterns for SS58/ A and SS8 l/C (Figs. 4A and 4B) were found to be identical for the majority of the sequence shown, thereby indicating the presence of an inverted terminal repetiKbp A 23.1 9.4 6.5 4.4

2.3 2.0

FIG. 3. Analysis of recombinant plasmid DNAs. (A) Recombinant plasmid DNAs (300 ng) were digested with BarnHI and electrophoresed In a 0.7% agarose gel at 80 mA for 6 hr in the presence of ethldlum bromide. (El)The gel was transferred to a nltrocellulose filter by the method of Southern (1975) and hybndlzed with a [32P]dATP random primer-labeled CAV- 1(CLL) BarnHI digest. The filter was subsequently washed, dned. and exposed to X-ray film as described under Materials and Methods. Lanes: 1, X HindIll DNA digest as a molecular weight marker; 2. BamHl digest of CAV-l(CLL) DNA; 35, BamHl A clones SSl l/A, SS58/A. and SS78/A. respectively; 611, BamHI I3 clones: 12- 14. BarnHI C clones SS24/C. SS45/C. and SS8 1/C, respectively

NUCLEOTIDE

SEQUENCE

79

tion (ITR). On the other hand, SS45/C is seen to contain a multiply reiterated nucleotide sequence of 40 bp (R,R,) within the ITR (Fig. 4C). A summary of the nucleotide sequence data obtained is shown in Fig. 5. It is evident that the length of the ITR of CAV- 1(CLL) is 198 bp long, as determined from the identical terminal nucleotide sequences present in SS58/A and SS81/C (Fig. 5B). This includes a 40-bp nucleotide sequence which is a perfect repetition of the nucleotide sequence from position 14 to 53. The 5’-terminal sequence of the BarnHI C fragment SS45K is also depicted in Fig. 5. The first 56 nucleotides are identical with the first 53 of SS58/A and SS81/C with an additional 3 bp, AAT, at position 1 1 (or 14). instead of the singly repeated 40-bp nucleotide sequence (nucleotides 14-53) of SS58/A and SS8l/C, however, SS45/ C contains at least seven apparently identical copies (R,-R,; Figs. 4 and 5) reiterated sequentially. Another clone, SSO5/C, from another experiment, contained at least 10 copies of the same sequence. Clone SS 1 l/A was identical with SS58/A, with a tandem repeat (i.e., two copies), but SS78/A contained three copies. Clone SS24/C was presumably truncated since it did not contain the common ITR sequence. Table 1 summarizes the observed structures of the isolated clones. DISCUSSION With the exception of SS24/C, the cloned terminal BamHl subgenomic fragments of CAV-l(CLL) described here all contain the 5’-CATCATCAAT (nucleotides 1- 10) sequence typical of adenoviruses and the highly conserved sequence ATAATATAC (nucleotides 9-l 7) present in the genome of human adenoviruses (Shinagawa et al., 1983; Fig. 5). In addition, clones SS58/A and SS81/C contained the sequence TGACGT (nucleotides 163- 168) found at or near the boundary of the ITR of human adenoviruses (Shinagawa et al., 1983). The 198-nucleotide ITR sequence found in CAV1(CLL) corresponds closely to the 160-nucleotide ITR sequence of CAV- 1(Woe-4) reported by Shinagawa et al. (1983) (Fig. 5A), the difference in length being due largely to the tandem repeat of nucleotides 14-53. The remaining differences were as follows: at position 55, CAV-1 (woe-4) has an adenine residue which is missing in CAV-l(CLL); at position 69, a cytosine residue in CAV- 1(Woe-4) is replaced by a thymine residue in CAV1(CLL); and at position 123, CAV-l(Woc-4) has a cytosine residue which is missing in CAV-l(CLL). There have been a number of reports of adenovirus mutants in which the genomes were characterized as possessing enlarged ITRs as a direct result of the acquisition of DNA sequences from one end of the genome (Hammarskjeld and Winberg, 1980; Deuring et

80

SIRA ET AL.

--I

FIG. 4. Sequence analysis of termini of CAV-l(CLL) DNA. (A) A clone containing the BarnHI A terminal fragment (SSSS/A); (B) and (C) clones containing the BarnHI C terminal fragment (SSSl/C and SS45/C). The sequences shown were obtained with a forward BarnHI primer, using [35S]dATP (Sanger et a/., 1977). Each sample was loaded onto the sequencing gel in 5 lanes, left to right, as A, G, C, T, and A. The large arrows demarcate the start of the viral genome and the end of the ITR. R, = reiterated sequence, where n = copy number.

a/., 1981; Gluzman and Van Doren, 1983; Brusca and Chinnadurai, 1983; lshiyama et a/., 1986; Haj-Ahmad and Graham, 1986). Schwarz eta/. (1982) isolated Ad1 2 mutants with enhanced growth properties which possessed additional sequences of regularly increasing size in a tandem arrangement at the right end of the genome. The additional sequences were an exact repetition of a terminal sequence, and the junction site for the repeat was highly conserved. Their observations are similar to ours in that the reiterated sequence in CAV-l(CLL) is consistently the 40-bp nucleotide sequence located between nucleotides 14 and 53, and the junction site is highly conserved. In our system, however, the reiterated sequence is found at both ends of the genome. The report of Larson and Tibbetts (1985) is similar to ours in that they described the generation of a novel stock of variant Ad3 genomes which con-

tained tandem repetitions of viral DNA sequences near the left and right ends of the genome; however, they observed that the junction sites of the sequence reiterations were effectively random in the region harboring the variant sequences and that the reiterated sequences were not always identical. Of the BarnHI DNA fragments cloned and analyzed, SS58/A and SS81K contained a single tandem reiteration of nucleotide sequence 14-53. Since they were cloned from a DNA population having apparently heterogeneous termini, we have no way of determining whether or not they were derived from the same genotype, although this is a possibility. Clone SS45/C contained at least 7 reiterated copies of the 40-bp nucleotide sequence. It would appear that this is not an isolated case (and therefore a possible artifact), since analysis of clone SSO5/C from another experiment re-

81

ADENOVIRUS GENOME 40-bp NUCLEOTIDE SEQUENCE 40 20 30 10 (A) CATCATChAT---AATATACAGGACAAAGAGGTGTGGCCTAAATGTTGTTTT:iTT

(B)

. .. ....... ... .. ... ......... .......... .. ..... ......... CATCATCAAT---AATATACAGGACAAAGAGGTGTGGCCKWAATGTTGTTTTTTTT . . . . . . . . . . ::::::::::::::::::::::::::::i::::::::::::::

(C) CATCATCAATAATAATATACAGGACAAAGA$GTGTGGCWATGTTGTTTTTTTT 20 40 50 10 60 (A)

---------------------------------------TV :

:::::::::::::

fB) ATACAGGACAAAGAGGTGTGGCCTL~ATGTTGTTTTTTTTT-AA~L~AGTTTTTGTT ::::t::::::::::::::::::::::::::::::::::: 100 (C) ATACAGGACAMGAGGTGTGGCCTAAATGTTGTTTTTTTT96 [RZ]

::::::::::::::::::::::::::::::::::::::::

ATACAGGACMAGAGGTGTGGCCTMATGTTGTTTITTTT 136 :....................................... ....................................... ATACAGGACAAAGAGGTGTGGCCTAAATGTTGTTTTTTTT176 :::::::::::::::::::::::::::::::::::::::: ATACAGGACAAAGAGGTGTGGCCTAAATGTTGTTTTTTTT216 :::::::::::::::::::::::::::::::::::::::: ATACAGGACAMGAGGTGTGGCCThUTGTTGTTTTTTTT 256 .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. ATACAGGACAAAGAGGT272 [R71

[R31 [R41 [R5] [R6]

::::::::::::I:::::::::::::::::::::::::::::::::::::

(B) TGATTGTTTTGACAAGGTCACACCCTGTTCAGGGCGTTTCCCACGGGAAA 110 130 140 120 150 140 150 120 130 160 (A) GACCATGACGTCAATTGGGTGTTTTTGTGGACTTTGGCCCG :t.' i:::::::::::::::::::::::::::::::::: : (B) G;&ATGACGTCA$TGGGTGTT;;TGTGGACTTTGGCCGG 190 198 FIG. 5. Nucleotide sequences (as read 5’ to 3’) of the ITRs of (A) CAV-1 (Woe-4) (from Shinagawa et a/. (1983); (B) CAV-1 (CLL), clones SS58/A and SS81/C; (C) CAV-l(CLL), clone SS45/C. The reiterated sequence found in SS58iA. SS81/C, and SS45JC corresponds to the nucleotide sequence from positions 14-53, inclusive, of the Woe-4 strain. (“:” Indicates homologous sequences; R, = reiterated sequence, where n = copy number.)

vealed the presence of at least 10 copies of the same sequence (Table 1). It is evident that CAV-l(CLL) consists of a mixed population of virions, some of which contain considerably more than two copies of the 40-bp nucleotide sequence in their right-hand ITR. This provides an explanation for the diffused doublet bands seen by electrophoresis in the right terminal fragments (Figs. 1 and 2). Presence of these multiple inserts did not appear to affect the viability of the virions for the following reasons. First, viral DNA was obtained from cesium chloride-purified virions, and second, electrophoresis of BarnHI digests of DNA from plaque-purified isolates revealed sharply defined single BarnHI C bands with slightly differing mobilities. This latter point also provides additional evidence that the multiple inserts are not artifacts of cloning. Sequence analysis of the three BarnHI A clones has revealed the presence of only 2 or 3 copies of the 40bp sequence. Clearly, analysis of a total of three clones is inadequate to conclude that multiple repeats to the extent found in the right terminus (BarnHI C) do not occur in the left; however, the observed restriction en-

zyme digestion profiles support this statement. For example, we found no indication of any diffuse or doublet bands in the left terminal fragments (CM A, BarnHI A, and SalI B bands). These fragments are all larger than BamHl C (Fig. 11, however, and a small insertion in these of about 300 bp (corresponding to a reiteration of 7 or 8 copies) would be difficult to detect electrophoretically. Nevertheless, the small Sac1 D fragment (1970 bp), identified as being located at the left terminus of CAV-l(CLL), did not reveal the presence of any diffused doublets (data not shown). The question arises as to how these DNA reiterations occurred. Haj-Ahmad and Graham (19&6) have proposed a model to explain the behavior of an Ad5 mutant with a direct repeat of left-end DNA segments that underwent further rearrangements at high frequency. Figure 6 shows how t’his model can be extended to account for the multiple reiterations occurring in BarnHI C fragments such as those occurring in SS45K and sso5/c. Briefly, the model requires a parental, genome with a tandem repetition of part of the ITR at one end (e.g., the 40-bp nucleotide sequence; R, and R2 in Fig. ‘6). During replication of the genome’the displaced’strand forms a panhandle structure (Daniel, 1978; Stow, 1982) in which the complementary R’, region which normally hybridizes with RI hybridizes instead with Rp. Excision of the unhybridized 3’end by 3’-exonuclease digestion, followed by 3’-extension and normal’replication, would result in a genome with tandem’ repeats in both ITRs (Fig. 6A). During subsequent replication of thisgenome, the panhandle form show:n in Fig. 65 would occu.r if R’,were to hybridize once again with R2 instead of the more likely RI. The result would be a genome with 2 copies of sequence R in one ITR, and 3 in the other. Repetition of this process ‘through succeedi,ng’ generations, would produce genomes with 2 copi’es of R in one ITR, and 4, 5, 6, etc.,‘in the other. The above mechanism provides an explanation for the existence of a genome with a tsndem repeat in one ITR and a variable number of repeats in the other. This is consistent with our suggestion that the reiterated TABLE 1 REITERATEDSEQUENCESIN BarnHI SUBGEN~MICCLONES

Clone

Terminus

SSl l/A

Left

SS58/A SS78lA SSO5lC SS24lC SS45lC SS81K

Left Left

Right Right Right Right

Copies of 40-bp sequence 2 2 3 310 No ITR &7 2

SIRA ET AL.

82

sequence in the genome of CAV-l(CLL) is restricted to a few copies in the left ITR with more highly variable numbers in the right. Since adenovirus DNA replication is considered to initiate at either terminus, however (Challberg and Kelly, 1982), the mechanism does not explain how this polarity of multiple insertions can be maintained. The explanation may be that multiple insertions do, in fact, occur in the left ITR but these DNA molecules are not packaged into viable virions since the insertion has interfered with the packaging signal located between nucleotides 290 and 390 (Hammarskjold and Winberg, 1980). Indeed, lshiyama et al. (1986) have isolated equine adenovirus mutants with insertions of 100 bp which were packaged, whereas those containing 200 bp in the left ITR were not. Answers to the question as to what mechanism regulates the number of repetitions remain speculative, however, and the proposal does not account for the origin of the first tandem sequence in CAV-l(CLL) or for the generation of the additional AAT sequence (positions 1 l13) in SS45/C. An important consideration concerns the effect of the repetitive inserts on the replication of the virus, both in vivo and in vitro. It is interesting that the repeated sequence occurs in the region of the ITR that, in human adenoviruses at least, has been identified as being essential for DNA replication (Hay and McDougall, 1986). It has bean shown for Ad2 that the terminal 18-bp nucleotide sequence is capable of functioning as the minimal origin of DNA replication in vitro (Tamanoi and Stillman, 1983; Van Bergen eta/., 1983; Challberg and Rawlins, 1984; Rawlins eta/., 1984; Lally et al., 1984), although additional sequences between nucleotides 19 and 48 greatly enhanced the efficiency of replication. Nucleotides 19-48 of the Ad2 genome have been identified as containing a binding site for a cellular protein, nuclear factor I ‘(NFI) (Nagata et al., 1983; Guggenheimer et a/., 1984; Leegwater et al., 1985). Based on analysis of NFI binding sites in cellular DNA and other viral genomes, a concensus sequence of TGG(N)B-,GCCAA (nucleotides 25-38 of the Ad2 ge-

5

3’exo S’ext

RI 32 R2 RI' /I I: II

RI Rz -ii A

5

5’3

S’exo 3’ext

Ad2 CAV-1

NFI-BINDING SITE 10 40750 120,.30 ~*-CATCATCAATAATATACCTTATTTTGGATTCAAZZTATGATAATCA&X... .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. : : :: ::: : ::: : : 5'-CATCATCAATAATATACAGGACAAAGAGGTGTGGCCT~TGTTGTTTTTTTT... -Reiterated sequence-

FIG. 7. Terminal nucleotide sequences of human Ad2 [from Sussenbach (1984)] and CAV-l(Woc-4 and CLL), showing the origin of replication identified for human strains. The human adenovirus NFI binding site and the consensus sequence required for NFI binding, TGG(N)B--/GCCAA, are identified, with the corresponding sequence shown for CAV-1. See text for further details.

nome) has been established (Gronostajski et a/., 1985; Hay and McDougall, 1986). The NFI is believed to interact with the guanine residues of the consensus sequence (Gronostajski et a/., 1985), and Hay and McDougall (1986) have recently demonstrated the importance of the two adenine residues (nucleotide positions 37 and 38) in the Ad2 origin of DNA replication. In addition, Pruijn et al. (1986) have identified an NFIII binding site in the Ad2 origin of replication (nucleotides 36-56) which overlaps that of the NFI binding site. By analogy with the human adenovirus system, the repetitive sequence of CAV-1 (CLL) (nucleotides 14-53) encompasses the entire NFI binding site and most of the NFIII (Fig. 7). It is apparent that it contains the essential components of the NFI consensus sequence: AGG(N),GCCTAA (nucleotides 27-39). With reference to the NFIII binding site, CAV-1 (CLL) does not possess the frequently conserved human adenovirus sequence at positions 39-50. It would appear, therefore, that the genomes of CAVl(CLL) virions contain, in their ITR regions, reiterated copies of an essential part of the adenoviral origin of DNA replication. On the basis of the clones analyzed, the genomes contain a minimum of two copies of this sequence (left and right termini). Indications were obtained that further reiteration in the left terminus is limited (possibly to 3 copies), whereas some genomes may contain 7-l 0 or more copies in the right terminus. The effect of these reiterations on the physiological and biological properties of the virus remains undetermined.

ACKNOWLEDGMENTS 5’3

RI’ RP’

This work was carried out under contract for The Minister of Natural Resources, Ontario (J.B.C.) and was additionally supported by a grant from the Natural Sciences and Engineering Research Council of Canada (M.G.A.).

REFERENCES

II ::

FIG. 6. Schematic representation of the ITRs of CAV-1 (CLL) during replication and the mechanism of amplification of reiterated sequences near the right-hand terminus. R, and R, represent the copies of the original tandem sequence. The genome-associated protein blocks the 5’-terminus. See text for further details.

ASSAF, R., MARSOLAIS,G., YELLE, J., and HAMELIN, C. (1983). Unambiguous typing of canine adenovirus isolates by deoxyribonucleic acid restriction-endonuclease analysis. Canad. J. Con-p. Med. 47,

460-463. BERKNER,K. L., and SHARP, P. A. (1983). Generation of adenovirus transfection of plasmids. Nucleic Acids Res. 11, 6003-6020.

by

ADENOVIRUS GENOME 40-bp NUCLEOTIDE SEQUENCE BIRNBOIM,H. C., and DOLY,J. (1979). A rapid alkaline extraction procedure for screening recombinant plasmid DNA. Nucleic Acids Res. 7, 1513.

BRUXA, J. S., and CHINNADURAI, G. (1983). Structure and reversion of an adenovirus type 2 mutant containing a duplication of the left end of the genome at the right end. L&o/logy 129, 381-392. CARusl, E. A. (1977). Evidence for blocked 5’-termini in human adenovirus DNA. virology 76, 380-394. CHALLBERG, M. D., and KELLY,T. J. (1982). Eukaryotic DNA replication: Viral and plasmid model systems. Annu. Rev. Biochem, 61, QOl934.

CHALLBERG, M. D., and RAWLINS, D. R. (1984). Template requirements for the initiation of adenovirus DNA replication, Proc. Nat/. Acad. SC;. USA 81, 100-104. DANIEL,E. (1976). Genome structure of incomplete particles of adenovirus. /. Viroi. 19, 685-708. DARAI,G., ROSEN,A., DELIUS,H., and FL~GEL,R. M. (1985). Characterization of the DNA of canine adenovirus by restriction enzyme analysis. intervirology 23, 23-28. DESIDERIO, S. V., and KELLY,T. J. (1981). Structure of the linkage between adenovirus DNA and the 55,000 molecular weight terminal protein. J. Mol. Viol. 145, 319-337. DEURING,R., KLOTZ.G., and DOERFLER, W. (1981). An unusual symmetric recombinant between adenovirus type 12 DNA and human cell DNA. Proc. Nat/. Acad. Sci. USA ,78, 3142-3146. FEINBERG, A., and VOGELSTEIN, B. (1983). A technique for radiolabeling DNA restriction endonuclease fragments to high specific activity. Anal. Biocheq

132, 6.

GLUZMAN, Y., and VAN DOREN,K. (1983). Palindromic adenovirus type 5-simian virus 40 hybrid. J. Krol. 45, 91-T 03. GROMOSTAJSKI, R. M., ADHYA,S,, NAGATA,K., GUGGENHEIMER, R. A., and’ HURWI”TZ, J. (1985). Site-specific DNA binding of nuclear factor 1: Analysis of cellular binding sites. MO/. Cell. Biol. 5, 964-971. GUGG&$IEIM&,R. A., STILLMAN,B. W., NAGATA,K., TAMANOI,F., and HudWiri, J. (1Q84). DNA sequences required for the in vitro repiication of adenovirus DNA. Proc. Nat/. Acad. Sci. USA 81, 30693073. HAJ-ARMRD,Y., and GRAHAM,F. L. (1986). Characterization of an aden&us type 5 mutant carrying embedded inverted terminal repeats. Virbbgy 163, 22-34. HALBZ~T,D. N., CUTT,I. R., and SHENK,T. S. (1985). Adenovirus early region 4 encodes functions required for efficient DNA replication, late gene expression, and host cell shut off. 1. Viroi. 56, 250-257. HAMELIN,C,, JOUVENNE, P., and ASSAF,R. (1986). Genotypic characterizs?ion of type-2 variants of canine adenovirus. Amer. J. Vet. Res.,47, 625~630.

HAMM@%J&.b, M. L., and WINBERG,G. (1980). Encapsidation of adenovirus 16 DNA is directed by a small DNA sequence at the left end of the genqnie. Cal/ 20,787-795. HAY,R..T., and,Mc!DoueA~~,I. M. (1986). Viable viruses with deletions in the left inverted terminal repeat define the adenovirus origin of DNA repli,cation. 1. C&n. Viral, 67, 321-332. ISHIYA~~A,. T.,‘.%INAGPIWA, M., SATO,G., FUJINAGA, K., and PADMANABHAN, R. (1986) Generation of packaging-defective DNA molecules of equine’addnovirus. Virology 151, 66-76. KOPTOP~UL~~, G., and CORNWELL, H. J. C. (198 1).Canine adenoviruses: A review. Vet. Buli. 5i, 135-142. LALLY,C., DCIRPE~R, T., GR~GER,W., ANTOINE,G., and WINNACKER, E-L. (1984); A size analysis of the adenovirus replicon. En/lBCI/. 3, 3337337. LARSON,. P. L., and TIBBE~~S,C. (1985). Spontaneous reiterations of DNA sequences near the ends of adenovirus type 3 genomes. Virobgy

1*7, 187-2’00.

83

P. A. J., VARYDRIEL,W., and VAN DERVLIET,P. C. (19853. Recognition site of nuclear factor I, a sequence-specific DNAbinding protein from HeLa cells that stimulates adenovirus DNA replication. E#BO J. 4, 1515-i 521. MANIATIS,T., FRtTSCH,E. F., and SAMBROOK, J. (1982). “Molecular Cloning: A Laboratory Manual,” pp. 383-386. Coid Spring Harbor Laboratory, Cold Spring Harbor, NY. NAGATA,K., GUGGENHEIMER, R. A., and HURWITZ,J. (1983) Specific binding of a cellular DNA replication protein IO the arigin of replication of adenovirus DNA. Proc. Natl. Acad. Sci. USA 80, 6 1376181. PETTERSSON, U., and SAMBROOK, J. (1973). Amount of viral DNA in the genome of ceils transformed by adenovirus type 2. J. Mol. Bioi. 73,125-l 30. PRUIJN, G. J. M., VANDRIEL,W., and VANDERVLIET,P. C. (1986). Nuclear factor Ill, a novel sequence-specific DNA-binding protein from HeLa cells stimulating adenovirus DNA replication. Nature (London) 322,

LEEGWATER,

656-659.

RAWLLNS, D. R., ROSENFELD, P. J., WIDES,R. J., CHAL~BERG, M. D., and KELLY,T. J. (1984). Structure and function of the adenovirus origin of replication. CeN 37, 309-319. REKCJSH, D. M. K., RUSSELL, W. C., BELLES, A. J. D., and ROBINSON, A. J. (7973). Identification of a protein linked to the ends of adenovirus DNA. CeN 11, 283-295. SANGER,F., NIGKLEN, S., and GOULSON, A. R. (1977). DNASequencing with chain terminating inhibitors. Proc. Nat\. Aoad. Sci. USA 74, 5463-5467. SCHWA!%,E., REINKE,C., YAMAMOTO,N., and ZUAHAUSEN,H. (1982). Terminal rearrangements in the genome of adenovirus type 12 mutants adapted to growth in two human tumor call lines. Virology 116,284-296. SHARP,P. A., MOORE,C., and HAVERTY,J. (1976). The infectivity of adenovirus 5 DNA-protein complex. I/irology 76442-456. SHINAGAWA. M., ISHIYAMA, T., PADMANABHAN, R.. FUJIRIAGA, K., KAMADA, M., and SATO,G. (1983). Comparative sequence analysis of the inverted terminal repetition in the genomes of animal and avian adenoviruses. Virobgy 125, 491-495. SHINAGAWA, M., and PADMANABHAN, R. (1980). Comparative sequence analysis of the inverted terminal repetitions from different adenoviruses. Proc. Nati Acad. SC/. USA 77, 3831-3835. SOUTHERN, E. M. (1975). Detection of epecific sequences among DNA fragments separated by gel electrophoresis. 1. Ma6. Biol. 98, 503517. STOW,N. D. (1982). The infectivity of adenovirus genomes lacking DNA sequences from their left-hand termini. Nu&ic Acids Res. IO, 5105-5119. SUSSENBACH, J. 8. (1984). The structure of the genome. In “The Adenoviruses” (t-l. S. Ginsberg, Ed.), pp. 35-l 24. Plenum, New YorW London. TAMANOI,F., and STILLMAN,B. W. (1983). Initiation ofadenovirus DNA replication in vitro requires a specific DNA sequence. Pro-c. tVa?l. Acad. Sci. USA 80, 6446-6460.

VAN BERGEN,B. G. M., VAN DERLEY,P. A., VAN DRIEL.,W., VAN MANSFELP, A. D. M., and VAND~RVLIET,P. C. (1983). Replication of origin containing adenovirus DNA fragments thst do not carry the terminal protein. Nucleic Acids Res. 1 I, 1975-1989. WHETSTONE, C. A. (1985). Should the criteria for species distinction in adenoviruses be reconsidered? Evidence from’ canine adenoviruses 1 and 2. lnfervirology 23, 116-l 2.0. YOSHIDA,K., REN,C. S., YAMASHITA, T., and FUJINAGA, K. (1984). The transforming gene of canine adenovirus type 1. tr: “SV40, Polyoma, and’ Adenoviruses,” p. 107. Cold Spring Harbor Laboratory, Cold Spring Harbor, NY.