Cell, Vol . 9, 559 -571, December 1976 (Part 1), Copyright (91976 by MIT

Characterization of Single-Stranded Viral DNA Sequences Present during Replication of Adenovirus Types 2 and 5 S . J . Flint, S . M . Berget, and Phillip A . Sharp Center for Cancer Research Massachusetts Institute of Technology 77 Massachusetts Avenue Cambridge, Massachusetts 02139

Summary Replication intermediates of adenovirus DNA apparently contain extensive stretches of singlestranded DNA. Such single-stranded viral DNA sequences homologous to different regions of the viral genome present in adenovirus-infected cells during viral DNA replication have therefore been characterized by hybridization to the separated strands of restriction endonuclease fragments of 32 P-labeled adenovirus types 2 and 5 DNA . Saturation hybridization experiments with infected cell DNA extracted at late times suggest that all regions of the adenovirus genome are represented in the single-stranded fraction, but at unequal frequencies . This nonuniform representation has been characterized in more detail with self-annealed, total cell DNA extracted 18 hr after adenovirus type 2 infection : the concentration of single-stranded sequences homologous to different regions of the viral genome was determined by comparing the rates of hybridization of 32 P-labeled, single-stranded DNA probes with such self-annealed 18 hr DNA to the rates of hybridization of the same probes with equal concentrations of their complements . This approach allows the concentration of singlestranded viral DNA sequences in excess of their complements to be determined . Such sequences can be represented by two concentration gradients across the viral genome : those homologous to the r strand increase in concentration from 27 .8-40 .9 units toward the right end, whereas sequences homologous to the I strand increase from an area 27 .8-40 .9 units toward the left end . The time course of synthesis of single-stranded viral DNA sequences relative to accumulation of total viral DNA during the productive cycle and their behavior following a shift of H5ts125-infected cells in which viral DNA replication has begun from a permissive to a nonpermissive temperature support the contention that these sequences are indeed generated as adenovirus DNA is replicated . These results are therefore discussed in terms of current models of adenovirus DNA replication . Introduction Human adenovirus type 2 (Ad2) and type 5 (Ad5) possess linear, double-stranded DNA genomes of molecular weight 23 x 106 daltons (Green et al .,

1967 ; van der Eb and van Kesteren, 1969), which contain inverted, terminal repetitions (Garon, Berry, and Rose, 1972 ; Wolfson and Dressier, 1972) 100140 nucleotides in length (Arrand, Keller, and Roberts, 1974), but neither cohesive ends like bacteriophage A DNA (Hershey, Burgi, and Ingraham, 1963 ; Strack and Kaiser, 1965) nor terminally redundant sequences like bacteriophage T2, T3, or T7 DNAs (Streisinger, Edgar, and Denhardt, 1964 ; MacHattie et al ., 1967 ; Ritchie et al ., 1967) . Consequently, adenoviruses cannot use the strategies adopted during replication by these bacteriophages to circumvent the inability of DNA polymerases to initiate DNA synthesis without a primer (Gilbert and Dressier, 1968 ; Watson, 1972 ; Enquist and Skalka, 1973) ; the precise mechanisms by which adenovirus DNA is replicated, especially those leading to synthesis of new 5' ends, are not obvious at the present time . Adenovirus replication is semi-conservative (Bellett and Younghusband, 1972 ; van der Vliet and Sussenbach, 1972) and appears to proceed by discontinuous DNA synthesis, for small pieces (10S12S) of viral DNA have been observed during replication (Horwitz, 1971 ; Bellett and Younghusband, 1972 ; van der Eb, 1973 ; Vlack, Rozijn, and Sussenbach, 1975 ; Winnacker, 1975) ; these are homologous to all regions of the viral genome (Winnacker, 1975) . Replicative intermediates, labeled for short periods of time with 3 H-thymidine, are characterized by the presence of replicating daughter strands of variable but never greater than genome length (Horwitz, 1971 ; Sussenbach, Ellens, and Jansz, 1973 ; van der Eb, 1973), an increased sedimentation rate in neutral sucrose gradients (Horwitz, 1971 ; van der Eb, 1973 ; Robin, BourgauxRamoisy, and Bourgaux, 1973 ; Sussenbach et al ., 1973), and a higher buoyant density in CsCI gradients than mature viral DNA (Horwitz, 1971 ; Pettersson, 1973 ; van der Vliet and Sussenbach, 1972 ; Sussenbach et al ., 1973) . The two latter properties, the behavior of replicative intermediates upon BND-cellulose chromatography (van der Eb, 1973 ; Sussenbach and van der Vliet, 1973) and their sensitivity to the single-stranded, DNA-specific endonuclease from Neurospora crassa (Pettersson, 1973 ; Robin et al ., 1973 ; Pearson, 1975), indicate that replicative intermediates contain extensive stretches of single-stranded DNA ; these can be chased into mature DNA molecules . When replicating adenovirus DNA is observed by electron microscopy, Y-shaped molecules containing single-stranded branches at the AT-rich end can be seen (Ellens, Sussenbach, and Jansz, 1974) ; this single-stranded DNA was originally reported to be homologous to one strand only of the Ad5 genome, the r strand (Sussenbach et al ., 1973) . On



Cell 560

this basis, Sussenbach and his colleagues proposed that initiation of adenovirus DNA synthesis occurs at only the AT-rich end of the genome, conventionally the right end ; synthesis of a daughter r strand in the 5'-3' direction would then displace the parental r strand, whose replication would have to be considerably delayed to account for its presence in a single-stranded form . It is now clear, however, that termination of Ad2 and Ad5 replication occurs at or near both ends of the genome (Tolun and Pettersson, 1971 ; Schilling, Weingartner, and Winnacker, 1975) . It has also been inferred from the strand specificity of label incorporated at the termini, that there must be an origin of replication at each end of the adenovirus genome (Horwitz, 1976 ; E . L . Winnacker, personal communication) . It has also been reported that the single-stranded viral DNA sequences generated during replication represent at least 50% of each strand of the Ad2 genome (Lavelle et al ., 1975 ; Tolun and Pettersson, 1975) . In this context, we have characterized single-stranded viral DNA sequences present during replication of Ad2 and AD5 DNA, and in this paper, report their quantitation and distribution along the viral genome . We also describe the fate of single-stranded DNA sequences in cells infected by H5ts125, a temperature-sensitive mutant of Ad5 (Ensinger and Ginsberg, 1972) reported to be defective in initiation of viral DNA replication (van der Vliet and Sussenbach, 1975), following a shift from a permissive to nonpermissive temperature . Results Saturation Hybridization Experiments between 18 Hr Ad2-Infected Cell DNA and Separated Strands of Eco R1 Fragments of Ad2 DNA Recent reports suggest that replicative forms of adenovirus DNA contain single-stranded sequences that are homologous to at least 50% of both strands of the viral genome (Lavelle et al ., 1975 ; Tolun and Pettersson, 1975) . These experiments, which used hybridization of replicative intermediates to the separated strands of Ad2 DNA, could not detect all sequences ; nor are they sensitive to nonuniform representation of different sequences . Our first experiments were therefore designed to assay for all single-stranded viral DNA sequences present in infected cells, including any that might be present at low concentration . Total DNA was isolated from HeLa cells 18 hr after infection with 10 pfu per cell of Ad2 ; the procedures used during extraction were chosen to avoid conditions, such as high pH and shearing forces, that might generate single-stranded viral DNA . Increasing concentrations of such DNA were hybridized with the

separated strands of the six Eco R1 fragments of 32 P-labeled Ad2 DNA (see Figure 1) . The concentration of 32 P-labeled probe DNA in the reaction was relatively high, 0 .5-1 x 10 -2 µg/ml, to facilitate detection of low concentrations of single-stranded viral DNA sequences . It is clear from the results shown in Figure 2 that viral DNA sequences homologous to both the r and I strands (nomenclature of Sharp, Gallimore, and Flint, 1974) of each fragment are present 18 hr after infection ; the input 18 hr infected cell DNA was not treated in any way that might denature it before hybridization, and the DNA sequences detected in this experiment must therefore be single-stranded in the infected cell . Such sequences complementary to the I strand-that is, r strand sequences-of Eco R1 fragments B-F, which comprise the right-hand 40% of the Ad2 genome (see Figure 1), are more abundant than their complements, but this pattern is reversed over Eco R1 fragment A . Thus, although sequences homologous to a large part of, and perhaps all of, both strands of the Ad2 genome are present 18 hr after infection, all regions do not appear to be uniformly represented in the single-stranded DNA pool . The conditions of hybridization used here, in which the concentration of probe was sufficient to drive the reaction, allow the concentration of nucleic acid complementary to each single-stranded probe to be determined (Flint and Sharp, 1976) . This, however, requires knowledge of the fraction of each probe saturated by the input DNA . Although apparent saturation values are shown in Figure 2, it seems probable that these reflect the outcome of annealing unlabeled, single-stranded viral DNA not only to its 3 2 P-labeled complement, but also to 10

0

20

30 40 50

60

70

80 90

100

Adenovirus 2 58.5

70.7 75 .9 83.4 89.7

A 1 B I F I DIE I 40 .9

27.8

B I

D

I C

C Eco

R1

59 .0

I

Bam H1

A

Adenovirus 5 76 .7 84.0

A 4.0

24.2 26 .5

EI C IG A

C

Eco R1

86.4 90.4

58.8

I

B

B

I

F

I

DHpa I

Figure 1 . Restriction Endonuclease Maps of Ad2 and Ad5 DNA The viral genomes are represented by the solid horizontal lines, 0-100 units . Vertical bars and numerical coordinates above the line indicate sites of cleavage . These data are from Mulder et al . (1974) and R . J . Roberts et al ., manuscript in preparation .



Adenovirus Single-Stranded DNA Sequences 561

complementary unlabeled sequences present in infected cell DNA . In other words, 18 hr Ad2-infected cell DNA contains single-stranded viral DNA sequences that are complementary to one another . It was therefore not possible to use data of the sort shown in Figure 2 to quantitate the nonuniform representation of single-stranded viral DNA sequences homologous to different regions of the genome . Appearance of Total Viral DNA during Ad5 Infection Before continuing characterization of single-stranded viral DNA sequences present in infected cells while adenovirus replicates its DNA, we wished to determine whether 18 hr after infection with 10-20 pfu per cell was the most suitable time for study . It was also of interest to compare the time course of appearance of total viral DNA with that of singlestranded DNA sequences homologous to different regions of the viral genome . HeLa cells were therefore infected with 10-20 pfu per cell of Ad5, which was chosen for these experiments because cleav18 hr. DNA 100

Eco RIB

Eco RIA

Eco RIC

80 60 40

1% I

I

•

Eco RID

I

I

j

Eco RIE

I

I

Eco RIF

40 20 0

I

0 .1

I

0.2

.

0

I 0 .1

I 0.2

DNA concentration,

1 0

I 0.1

I 0.2

mg/ml

Figure 2 . Hybridization of 18 Hr Ad2-Infected Cell DNA to the Separated Strands of the Eco R1 Fragments of Ad2 DNA The strands of the six Eco R1 fragments of 32P-labeled Ad2 DNA were separated and purified as described in Experimental Procedures . 0 .5-1 .0 x 10-2 µg/ml of each strand were hybridized with the concentration of HeLa cell DNA, isolated 18 hr after Ad2 infection, shown on the abscissa for 24 hr at 68°C in 0 .10 M phosphate buffer (pH 6 .8) containing 1 .0 M NaCI and 0 .4% SDS . Reactions were analyzed by chromatography on hydroxylapatite . ( •-0) and (A-A) represent annealing to the r and I strands, respectively, of each fragment .

age with Eco R1 generates only three fragments (Figure 1) . Infected cells were harvested 4, 8 .5, 13, 17 .5, and 22 hr later, and total DNA was isolated and purified as described in Experimental Procedures . The concentration of total Ad5 DNA at each time was determined by measuring the rate of reassociation of denatured, fragmented, 32 P-labeled Ad5 DNA in the presence of control DNA (salmonsperm) and in the presence of denatured, fragmented, infected cell DNA (Gelb, Kohne, and Martin, 1971) ; the magnitude of the increase in the rate of reannealing in the presence of infected cell DNA was used to calculate the concentration of total Ad5 DNA at each time, with the result shown in Figure 4 . This increases slowly from a value of 24 copies (a value reassuringly close to the input) at 4 hr to about 1250 copies per diploid quantity of cell DNA at 17 .5 hr . In the next 4 .5 hr, the concentration of Ad5 DNA increases nearly 10 fold, as shown in Figure 4 . Appearance of Single-Stranded Ad5 DNA Sequences Homologous to Different Regions of the Genome during Infection To measure the concentration of single-stranded Ad5 DNA homologous to different regions of the genome, we have also adopted a kinetic method : infected cell DNA, which was not denatured or sheared, was hybridized with the separated strands of the three Eco R1 fragments of Ad5 DNA . Figure 3 shows the kinetics of annealing of these probes with the five infected cell DNA samples ; each probe was also annealed with an equal concentration of its complement and a known amount of denatured, fragmented, unlabeled Ad5 DNA . The results of such reconstruction experiments, given in Table 1A, indicate that single-stranded probes are suitable for use in these experiments and are discussed in relation to the method of analysis (see Experimental Procedures) . No hybridization to any of the six probes was observed with DNA isolated 4 hr after infection, indicating that the extraction procedure described below did not artificially generate single-stranded viral DNA sequences . Inspection of Table 2 and Figure 3 reveals that single-stranded Ad5 DNA sequences complementary to the 1 strand of Eco R1 fragments B and C and to both strands of Eco R1 fragment A are first present at 8 .5 hr and increase in concentration thereafter . Little hybridization to the r strands of Eco R1 fragments B and C was detected, even with 17 .5 hr and 22 hr infected cell DNA, suggesting that sequences homologous to the r strand of the right end of the genome (see Figure 1) are much more abundant than their I strand complements at all times when single-stranded Ad5 DNA sequences are present . A similar pattern was observed in the saturation hybridization



Cell 562

w

2 .6

2 .6

2.2

2 .2

I .8

1 .8

1 .4

1 .4

1 .0 0 .2 0.4 2.6

N .,N

1 .0

0.6 0.8 -

7.4

2 .2

-

5.8

18

-

4.2

Eco R1 Br

1.4

0.2 0.4 0 .6 0.8

2.6

1 .0 0.2 0.4 2 .6

0.6 0.8

1.0

Eco R1 Cr

0.2 0.4 0 .6

1 .2

1.0

1.2

Eco R1 Cl

9.0

2 .2

0.8

7.0

5.0

1.4

1.0 °-A I& 0.4 0.8

3.0

1.2

1 .6

~

1 .0

.moo

2.0 t / < t"2

0.4 0.8

1 .2

1 .6~ 2.0

P>

Figure 3 . Annealing of Infected Cell DNA to the Separated Strands of Eco R1 Fragments of 32P-Labeled Ad5 DNA Total DNA was isolated from Ad5-infected HeLa cells 4, 8 .5, 13, 17 .5, and 22 hr after infection as described in Experimental Procedures. Without further treatment, samples of such DNA were annealed with 32P-labeled DNA, spec . act . 0 .5 x 106 cpm/µg, of the r and I strands of each of the three Eco R1 fragments of Ad5 DNA in 0 .10 M phosphate buffer (pH 6 .8) containing 1 .0 M NaCl and 0 .5% SDS . The concentrations of probe and infected cell DNA in each reaction are given in Table 2 . Samples were withdrawn from the reaction mixtures after various times of incubation at 68°C, and the fraction of probe that had entered hybrid at each time was determined by chromatography on hydroxylapatite . Each probe was also annealed in the presence of an equal volume of its complement (A L) and 0 .75 x 10-2 µg/ml of denatured, fragmented, unlabeled Ad5 DNA (x-x) . was calculated as described in the text for the standard reaction of single-stranded probe with its complement . Annealing of each probe in the presence of 4, 8 .5, 13, 17 .5, and 22 hr Ad5-infected cell DNA is represented by (0-0), (s-•) , (o-o), (∎- M), and (AA), respectively .

experiments with Ad2 DNA described above, and when kinetic experiments were repeated with selfannealed, Ad5-infected cell DNA samples (data not shown) . The rate of each reaction was calculated at all time points at which >10% but 10, and such cases are indicated by parentheses in Table 2 . Even in these worst cases, the values shown in Table 2 are probably only 25-40% in error (see Experimental Procedures) . At 8 .5 hr after infection, very low levels of singlestranded DNA sequences, 1 .5-3 .1 copies per diploid quantity of cell DNA, can be detected . Although these values are low, they are clearly positive since the probes used are single-stranded and do not anneal in the absence of complementary sequences, as when incubated with 4 hr infected cell DNA . After this time, the concentration of all single-stranded sequences that can be detected increases until 17 .5 hr, after which time little change is observed (Table 2) . The time course of synthesis of single-stranded viral DNA complementary to the I strand of Eco R1 fragment C and the r strand of Eco R1 fragment A are shown in Figure 4 and compared to the concentration of total Ad5 DNA at each time . Although the levels of sequences complementary to these two probes are apparently rather different (but see below), maximal accumulation of both occurs between 13 and 17 .5 hr after infection ; most total Ad5 DNA, on the other hand, is made between 17 .5 and 22 hr after infection in this system . This pattern is consistent with, but by no

Adenovirus Single-Stranded DNA Sequences 563

means proves, a role for the single-stranded viral DNA sequences that we have detected in total DNA isolated from infected cells in the replication of adenovirus DNA . It should be noted that in contrast to the other four probes, the kinetics of hybridization of infected cell DNA to the separated strands of Eco R1 fragment A deviate appreciably from apparent secondorder before half the probe has formed hybrid (Figure 3) . This suggests that sequences complementary to the I strand of Eco R1 fragment A, for example, are more abundant than their complements over only part of Eco R1 fragment A . One would suppose that abundant single-stranded r strand sequences are homologous to the right part of Eco R1 fragment A, since only r strand sequences were detected with Eco R1 fragments B

and C, which constitute the right-hand 25% of the genome . This interpretation is also consistent with the results of the saturation experiment with 18 hr Ad2-infected cell DNA . The concentrations of single-stranded viral DNA sequences complementary to the r and I strands of Eco R1 fragment A given in Table 2 and Figure 4 were calculated assuming that all sequences complementary to each strand were present and are therefore undoubtedly underestimated . Characterization of Viral Single-Stranded DNA Sequences Present in Excess of Their Complements 18 Hr after Ad2 Infection Both experiments described so far indicate that some regions of the adenovirus genome are preferentially represented in the single-stranded viral DNA

Table 1 . Reconstruction Experiments with 32 P-Labeled, Single-Stranded DNA Probes Total Probe Concentration ug/ml x 103

32P-Labeled Probe ug/ml x 103

:

Al

6 .29

0 .56 ± 0 .05

9 .18

0 .36

Ar

6 .29

0 .53 ± 0 .04

9 .18

0 .36

BI

4 .85

0 .17 ± 0 .02

5 .44

Cl

2 .59

0.11 f 0 .01

BI

1 .93

Cl

1 .56

Probe

tv Probe/tr

Concentration Unlabeled DNA Complementary to Probe ug/ml x 103

Unlabeled Viral DNA Mg/Ml x 10 2 Calculated

Added

3 .30

0 .86

0.75

3 .31

0 .87

0.75

0 .15

0 .82

1 .01

0 .75

2 .86

0 .10

0 .28

0 .79

0 .75

3 .65 ± 0 .07

8 .03

0 .88

7 .01

11 .0

10 .0

3.69 ± 1 .10

6 .66

0 .86

5 .81

11 .0

10 .0

(A) Eco R1

(B) Eco R1

Barn H1

DI

1 .89

2 .64 ± 0 .11

5 .59

0 .89

5 .00

13 .0

10 .0

El

2 .19

1 .16 ± 0 .30

5 .24

0 .48

2 .51

8 .3

10 .0

Fl

1 .78

1 .23 ± 0 .28

4 .23

0 .52

2 .20

8 .9

10 .0

BI

2 .74

7 .76 ± 0 .22

16 .60

1,30

21 .00

15 .0

10 .0

Cr

1 .85

4 .42 f 0 .56

10 .90

0 .75

8 .20

9 .0

10 .0

Dr

2 .20

2.72 ± 0 .36

8 .75

0 .68

6 .00

9 .1

10 .0

DI

2 .20

2 .84 ± 0 .36

8 .75

0 .71

6 .25

9 .5

10 .0

The rates of reactions in which each probe was annealed in the presence of an equal concentration of its complement were calculated at each time point at which samples were analyzed, and the average, , was calculated . An average value, , was similarly calculated for reactions in which each probe was annealed in the presence of 0 .75 X 10-2 ug/ml unlabeled Ad2 DNA (A) or 10 -1 µg/ml unlabeled Ad2 DNA (B) . These DNAs were denatured and fragmented at 100°C in 0 .3 M NaOH before annealing . The incubation and assay conditions are described in Experimental Procedures . Eco R1 endonuclease cleaves Ads DNA into three fragments-76 .9, 15 .8, and 7 .23%, and Ad2 DNA into six fragments-59 .2, 12 .2, 10 .2, 7.4, 6 .1, and 4 .9% of the genome in length . Barn H1 cleaves Ad2 DNA into four fragments-41 .0, 27 .8, 18 .1, and 13 .1% of the genome in length . The concentration of 32P-labeled probe strand is listed in the first column . As double-stranded total Ad2 and Ad5 DNA was added to each reaction, the total concentration of probe strand (third column) was calculated by summing the concentration of unlabeled DNA strand homologous to the probe (the value given in the seventh column multiplied by the fraction of the viral genome represented by the probe and divided by 2) and 32P-labeled probe DNA concentration . The measured ratio : (second column) was then corrected to reflect the relative rate of renaturation of a hypothetical reaction with a DNA concentration twice that of the total probe strand concentration shown in the second column by multiplying the latter value by the ratio of 32 P-labeled to total probe concentrations (column 1 : column 3) . The concentration of strand complementary to the probe strand is the product of the values in the third and fourth columns . The concentrations of unlabeled viral DNA added to the reaction shown in the sixth column were calculated from the values given in the fifth column by dividing the latter by the fraction of the viral genome represented by the probe and multiplying by a factor of 2 .



Cell 564

fraction . It is also clear, however, that sequences complementary to one another are present in this fraction . To describe this nonuniformity more fully, we have therefore isolated infected cell DNA 18 hr after Ad2 infection and have subjected it to a cycle

of self-annealing . Only those sequences in excess of their complements were therefore measured in hybridization experiments with the separated strands of restriction endonuclease fragments of 32 P-labeled Ad2 DNA . This serotype was chosen

Table 2 . Concentration of Single-Stranded Ad5 DNA Sequences Complementary to the Separated Strands of the Eco R1 Fragments of Ad5 DNA at Different Times after Infection Concentration SingleStranded Viral DNA Complementary to Probe x 10 3

Infected Cell DNA Input mg/ml

Copies SingleStranded DNA/Diploid Quantity Cell DNA

Probe

:

32 P-Labeled Probe DNA µg/ml x 103

8 .5 Hr DNA Ar

0 .23

3 .15

0 .72

0 .20

(1 .5)

0 .33

3 .15

1 .23

0 .20

(2 .2)

0 .20

(2 .6)

Al Br BI Cr CI

No 0 .10 No

significant 2 .43 significant

hybridization 0 .25 hybridization

0 .11

1 .30

0 .14

0 .20

(3 .1)

Ar

1 .76

3 .15

5 .55

0 .20

12 .7

Al

1 .44

3 .15

4 .52

0 .20

10 .0

0 .20

30 .0

15 .0

13 Hr DNA

Br BI Cr Cl

No 1 .21 No

significant 2 .43 significant

hybridization 2 .94 hybridization

0 .51

1 .30

0 .67

0 .20

11 .10

3 .15

34 .97

0 .20

(75)

6 .67

3 .15

20 .98

0 .20

45

significant

hybridization 0 .20

78

17 .5 Hr DNA Ar Al Br BI

No 3 .00

2.43

7 .29

Cr

No

significant

hybridization

C1

11 .84

1 .30

15 .40

0 .20

Ar

8 .54

3 .15

26 .90

0 .20

58

Al

7 .93

3 .15

24 .98

0 .20

54

Br

No

significant

hybridization

BI

13 .53

2.43

32 .88

Cr

No

significant

hybridization

CI

11 .84

1 .30

15 .40

(348)

22 Hr DNA

0 .20

(336)

0 .20

(348)

Reactions between 32P-labeled DNA, spec . act . 0 .5 x 106 cpm/µg, of the r and I strands of the three Eco R1 fragments of Ad5 DNA (see Figure 1) and 0 .2 mg/ml infected cell DNA were performed and analyzed as described in the text to give the ratios of (probe + equal concentration of its complement) to (probe + infected cell DNA) shown in the second column . These values and the concentrations of single-stranded probe shown in the third column were used to calculate the concentration of single-stranded DNA sequences complementary to each probe shown in the fourth column . Knowing the molecular weight of each single-stranded probe (A, 8 .82 x 106; B, 1 .91 x 106 ; and C, 0 .84 x 10 6 ) and the concentration of infected cell DNA in each reaction (0 .2 mg/ml), and assuming that a diploid quantity of cell DNA is equivalent to 3 .9 + 10 12 daltons (Sober, 1968), we calculated, from the values shown in the fourth column, the concentrations of single-stranded sequences complementary to each probe shown in the sixth column . In these calculations, it was assumed that sequences complementary to all of each probe that entered hybrid are represented in infected cell DNA . The values shown in parentheses indicate those cases where the rates of the standard and experimental reactions were significantly different (see second column) ; these values are likely to be 40% in error (see Experimental Procedures) . The data upon which these calculations are based are shown in Figure 3 .



Adenovirus Single-Stranded DNA Sequences 565

for the experiments because the use of Eco R1 fragments B-F and Barn H1 fragments B-D (see Figure 1) permits reasonable sampling of all regions of the genome . The kinetics of annealing of the separated strands of these eight fragments of 32 P-labeled Ad2

DNA to self-annealed, 18 hr infected cell DNA were measured and compared to the rates of standard reactions containing equal concentrations of the r and I strands of each fragment, as discussed above . Reconstruction experiments were also performed, with the results shown in Table 1 ; once again the calculated concentration of unlabeled Ad2 added to the reaction is very similar to the amount actually added . The data obtained with self-annealed, 18 hr infected cell DNA are given in Table 3 and summarized in Figure 5 . No single-stranded Ad2 sequences homologous to the I strand could be detected at the right end of the genome (Figure 5), confirming the indications described above that sequences homologous to the r strand are more abundant over the right part of the adenovirus genome . Interestingly, single-stranded Ad2 DNA sequences homologous to the r strand decrease in concentration from about 400 copies between 90100 units to only about 20 copies per diploid quantity of cell DNA between 28 .5 and 40 .9 units . A reverse pattern is seen at the left-hand of the adenovirus genome (Figure 5), such that concentration gradients of single-stranded viral DNA sequences homologous to the r and I strands meet within Bam H1 fragment D . This latter fragment was 0 0 500

10

Q - 400 Z 0
1% of the normal amount of viral progeny DNA when infected cells are maintained at a nonpermissive temperature (Ensinger and Ginsberg, 1972) . Furthermore, labeling studies have indicated that incorporation of thymidine into DNA decreases rapidly following a shift of H5ts125infected cells to restrictive temperatures, 39-41 °C (Levine et al ., 1974 ; van der Vliet and Sussenbach, 1975 ; Carter and Ginsberg, 1976) . In agreement with these results, we find that incorporation of labeled thymidine into viral-specific DNA drops to 50% of the preshift value within 30 min and to