- Email: [email protected]
Visualization of the inverted terminal repetition in adeno-associated virus DNA
J. Mol.
Biol.
(1974) 82, 267-271
LETTERS
TO THE
EDITOR
Visualization of the Inverted Terminal Repetition in Adeno-associated Virus DNA Adeno-associated virus linear, single polynucleotide chains contain an inverted terminal repetition which allows the formation of single-stranded circles when the DNA is exposed to annealing conditions. Under appropriate annealing conditions single-stranded circular dimers are formed, the majority of which have two projections separated by 180” visible in the electron microscope. We conclude that these projections represent the regions of self-complementarity (inverted terminal repetition) contained within the virus DNA. Measurements of the lengths of the projections indicate that the length of the inverted terminal repetition represents approximately 1*5o/0 of the genome.
Adeno-associated virus is a small, defective human virus that requires coinfection with adenovirus for a productive infection (Atchison et al., 1965 ; Hoggan et al., 1966). Two types of virions, which contain complementary (+ and -) singIe-stranded DNAs with a molecular weight of 1.4~ 106, are produced (Crawford et al., 1969; Rose et al., 1969; Mayor et al., 1969; Berm & Rose, 1970; Berns & Adler, 1972). We have been studying the structure of adeno-associated virus DNA in an attempt to understand the role of each DNA strand in viral development and particularly in the mechanism of DNA replication. Recently we reported that the DNA of this virus contains a limited number of nucleotide sequence permutations, all of whose start points occur within a small region representing less than 6% of the genome. We also found that the nucleotide sequence of this DNA contains a natural terminal repet’ition (12----12), representing approximately 1% of the genome, similar to those found in linear bacteriophage DNAs (Gerry et al., 1973). Additionally, Koczot et al. (1973) have reported evidence that the nucleotide sequence of adenoassociated virus DNA also contains an inverted terminal repetition which is probably of the type 12----2’1’ (where 2’ and 1’ are nucleotide sequences complementary to 2 and 1, respectively) similar to that reoently described for adeno-virus DNA (Garon et al., 1972; Wolfson & Dressler, 1972). Further characterization of the termini of adeno-associated virus DNA is of interest because of their probable significance in DNA replication. In this report we have been able to visualize in the electron microscope a region of this DNA, which we conclude corresponds to the inverted terminal repetition. The conclusion that the virus DNA contains an inverted terminal repetition rests on the fact that single polynuoleotide chains produced by the virus can cyclize under annealing conditions (Koczot et al., 1973). If the model of inverted terminal repetition described above and illustrated in Figure 1 is correct, and the inverted terminal repetition is of sufficient length, then the regions of annealed inverted terminal repetition should be visible as a projection of double-stranded DNA from the single-stranded circles. Estimates of the extent of self-complementary sequences in DNA from this virus have ranged from as high as 20% (Carter et al., 1972) to a very small percentage (Koczot et al., 1973). Visualization of projections shown to represent the inverted terminal repetition would allow a direct measurement of the length of the self-complementary regions. Because of the probability that the 267
268
K.
I.
BERNS
AND
T.
J.
KELLY
JR
Monomer
FIG. 1. The simplest by linear,
single
model polynucleotide
for the formation chains oontsining
of single-stranded oiroular monomers an inverted terminal repetition.
and dimers
DNA contains both natural and inverted terminal repetitions, a comparison of their relative lengths would be helpful in constructing a model that could account for the presenceof both. We have been able to repeat the observation of Koczot et al. (1973) that linear adeno-associated virus single polynucleotide chains can cyclize when exposed to annealing conditions. A preparation containing both types of intact, linear, complementary single strands yielded many single-strandedcircles after partial annealing. In this preparation 96% of the linear single strands had no visible projection while 92% of the single-stranded circles did have projections. However, 28% of the singlestranded circles contained more than one apparent projection and the possibility remained that the projections observed on the circles were artifacts caused by a twisting of the circles when they were mounted for microscopy. Additionally, it was not possibleto determine whether the projections occurred at any specific site about the circle. Koczot et al. (1973) have previously reported such projections but were also concerned that they might have been artifacts. To resolve this question the formation of single-stranded circular dimers was promoted by annealing only one of the complementary strands at a relatively high concentration. If the projections previously observed on single-stranded circular monomers had represented the base-paired, inverted terminal repetition region of the DNA, the majority of single-stranded circular dimers so produced should have contained two projections separated by 180” and any circular dimer containing more than two projections should have at least one pair separated by 180” (Fig. 1). To perform this experiment adeno-associatedvirus 2, containing l*C-labeled DNA in which bromodeoxyuridine had been partially substituted for thymidine (40%), was grown in KB cells in spinner culture (Rose et al., 1969). The complementary strands of bromodeoxyuridine-labeled virus DNA can be separated by banding in a neutral isopycnic cesium chloride density gradient (Berns & Rose, 1970). The heavy single strands were isolated and annealed at a concentration of 20 pg/ml in 50% formamide at room temperature for 24 hours. Examples of the single-stranded circles observed (both monomers and dimers) are demonstrated in Plate I. Of 107 single-stranded circular monomers photographed, 82 had only one projection and seven had no visible projection, the remainder having more than one apparent projection (Table 1). Twenty-five single-stranded circular
PLATE I. Representative single-stranded circular monomers virus DNA with projections (arrows) observed in the electron mounted using the formamide-protein film method (Davis JEMlOOB electron microscope. Magnification 60.000 >
and dimers microscope. et OZ., 1971)
of adeno-associatacl The molecules wwv and observed ill th,x I frrcina
,L ?tih
LETTERS
TO
THE
TABLE
The number
of projections Type
1
per molecule
of DNA
Monomer Dimer
o
of adeno-associated lprojeo~
7 0
269
EDITOR
82 0
12 16
3
4
3 5
0 4
viru.s DNA
dimers were photographed; of these 16 had two projections and the remainder had either three or four (Table 1). Measurements of 71 circular monomers and 25 circular dimers yielded average lengths of 1.56 f 0.12 pm and 3.12 f 0.22 pm, respectively (Fig. 2). In Figure 3(a) we have plotted the shorter of the two measured interprojection distances as a fraction of the total measured length of each circular dimer. Every molecule but one had at least one pair of projections that were separated by a distance equal to 0.45 to 050 of the total length of the molecule. Three circular dimers had two such pairs. Other interprojection distances occurred randomly. We have plotted the relative interprojection distances measured for those dimers with only two projections in Figure 3(b). In this case all but one of the shorter interprojection distances represent more than 0.45 of the total molecular length. From these data we conclude that the majority of projections (at least 48 out of 61) occur in the relative positions expected along the length of single-stranded circular dimers and thus that the majority of the projections most likely represent the hydrogenbonded, inverted terminal repetition regions holding circular dimers together. The relative lengths of the projections were determined for 50 single-stranded circular monomers that had just one projection. These ranged from 1.0 to 25% of the total length with a mean of 15%. The value of 15% for the inverted terminal repetition is close to the value of 1.0% determined for the length of the normal
30-
25-
:1 I
I.0 Relotw
I P-?-L length
FIQ. 2. Histogram of the relative messwed lengths of single-stranded dimers, oonsidering the mean of the monomers (1.66 pm) to be 1.0. Lengths 8 map measurer (KeulYel & Easer) and 8 grating replica (54,864 lines/inah, 18
circular monomers were determined E. G. Fullem).
and wing
270
K. I. BERNS
AND
T. J. KELLY
JR
L” (a)
15-
(b)
IO -
5-
I II 0.10 Interprojection
I I I in 020 distance
O-30
0.40
(fraction
r 0 50 of totol
length)
FIQ. 3. (a) Histogram of the shorter interprojeotion distances between EU possible pairs of projeotions plotted as the fractional length of the involved single-stranded oiroular dimer. (b) Histogram of the shorter interprojeation distances measured on all single-stranded oiroul8r dimers having two projections, plotted BS the fraotioxml length of the involved dimer.
terminal repetition in adeno-associatedvirus DNA described by Gerry et al., 1973, and would thus be compatible with the possibility that this DNA contains a terminal repetition whose sequence is symmetrical (122’1’----122’1’). Two other ah&natives, however, are not ruled out: (1) different molecules have either a natural or an inverted terminal repetition, but not both, or (2) the two types of terminal repetition are at different sites (12---2’1’12). If the projections observed do indeed represent the region of terminal repetition, the lengths would correspond to 49 to 100 nucleotide pairs, a length which has been observed previously using this technique of visualizing DNA molecules (Davis et al., 1971). Uncertainty as to the absolute length of the region of inverted terminal repetition arisesfrom the fact that the projections may be longer than the absolute base-paired region because nearby basesmay be held in apposition although unpaired or, conversely, the projections may not represent the full length of the base-paired region because of undetected folding. However, in agreement with Koczot et al. (1973) we do not fmd any evidence for the extent of self annealing (up to 20%) suggestedby Carter et al. (1972). In summary, we have presented evidence for projections on single-stranded circular dimers of adeno-associated virus DNA which we conclude correspond to the region of inverted terminal repetition contained within the DNA becauseof the
LETTERS
TO THE
EDITOR
271
specific locations of the projections along the contour of the molecules. These observations are oompatible with an inverted terminal sequencerepetition of the type 12----2’1’, which would allow base pairing of complementary regions in the anti-
parallel configuration. We thank D. Nathans, H. 0. Smith, B. Weiss and T. C. Pinkerton for helpful suggestions with regard to the manuscript. This work wez zupported by grants from the U.S. Public Health Service (noz lROlAI10843 and lROlCA13394) and the National Science Foundation (no. GB34163). One of us (K. I. B.) is a Hughes Medical Investigator. The other author (T. J. K.) is a reoipient of a Research Career Development Award from the National Institutes of Health (no. lK04CA70220-01). Department of Miorobiology The Johns Hopkins University School of Medicine Baltimore, Md 21206, U.S.A. Received
KENNRTE I. BERNS THOMAS J. KELLY JR
21 May 1973 REFERENCES
Atchison, R. W., C&o, B. C. t Hammon, W. McD. (1906). Science, 194, 764-760. Berns, K. I. & Adler, S. (1972). J. Virool. 9, 394-396. Berns, K. I. & Rose, J. A. (1970). J. Viral. 6, 693-699. Carter, B. J., Khoury, G. & Rose, J. A. (1972). J. ‘CTdroZ.10, 1118-1126. Crawford, L. V., Follett, E. A. C., Burdon, M. G. & McGeoch, D. J. (1969). J. Gen. Viral. 4, 37-46. Davis, R. W., Simon, M. & Davidson, N. (1971). In Met&da im Elw%(mology (Grossman, L. & Moldave, K., eds), vol. 21, pp. 413-428, Academia Press, New York. Garon, C. F., Berry, K. W. & Rose, J. A. (1972). Prw. Nat. Acud. Sci., U.S.A. 69, 23912396. Gerry, H. W., Kelly, T. J., Jr t Berns, K. I. (1973). J. Mol. Bid. 79, 207-226. Hoggan, M. D., Blacklow, N. R. & Rowe, W. P. (1966). Proc. Nat. Ad. hi., U.S.A. 65, 1467-1471. Koczot, F. J., Carter, B. J., Garon, C. F. & Rose, J. A. ( 1973). Proc. Nat. Ad. Sci., U.S.A. 70, 216-219. Mayor, H. D., Torikai, K., Melnick, J. L. & Mandel, M. (1969). Soierace, 166,1280-1282. Rose, J. A., Berns, K. I., Hoggan, M. D. & Koozot, F. (1969). Proc. Nat. Acud. Soi., U.S.A. 64, 863-869. Wolfson, J. & Dreesler, D. (1972). Pmt. Nat. Ad. Sci., U.S.A. 69, 3064-3067.
Biol.
(1974) 82, 267-271
LETTERS
TO THE
EDITOR
Visualization of the Inverted Terminal Repetition in Adeno-associated Virus DNA Adeno-associated virus linear, single polynucleotide chains contain an inverted terminal repetition which allows the formation of single-stranded circles when the DNA is exposed to annealing conditions. Under appropriate annealing conditions single-stranded circular dimers are formed, the majority of which have two projections separated by 180” visible in the electron microscope. We conclude that these projections represent the regions of self-complementarity (inverted terminal repetition) contained within the virus DNA. Measurements of the lengths of the projections indicate that the length of the inverted terminal repetition represents approximately 1*5o/0 of the genome.
Adeno-associated virus is a small, defective human virus that requires coinfection with adenovirus for a productive infection (Atchison et al., 1965 ; Hoggan et al., 1966). Two types of virions, which contain complementary (+ and -) singIe-stranded DNAs with a molecular weight of 1.4~ 106, are produced (Crawford et al., 1969; Rose et al., 1969; Mayor et al., 1969; Berm & Rose, 1970; Berns & Adler, 1972). We have been studying the structure of adeno-associated virus DNA in an attempt to understand the role of each DNA strand in viral development and particularly in the mechanism of DNA replication. Recently we reported that the DNA of this virus contains a limited number of nucleotide sequence permutations, all of whose start points occur within a small region representing less than 6% of the genome. We also found that the nucleotide sequence of this DNA contains a natural terminal repet’ition (12----12), representing approximately 1% of the genome, similar to those found in linear bacteriophage DNAs (Gerry et al., 1973). Additionally, Koczot et al. (1973) have reported evidence that the nucleotide sequence of adenoassociated virus DNA also contains an inverted terminal repetition which is probably of the type 12----2’1’ (where 2’ and 1’ are nucleotide sequences complementary to 2 and 1, respectively) similar to that reoently described for adeno-virus DNA (Garon et al., 1972; Wolfson & Dressler, 1972). Further characterization of the termini of adeno-associated virus DNA is of interest because of their probable significance in DNA replication. In this report we have been able to visualize in the electron microscope a region of this DNA, which we conclude corresponds to the inverted terminal repetition. The conclusion that the virus DNA contains an inverted terminal repetition rests on the fact that single polynuoleotide chains produced by the virus can cyclize under annealing conditions (Koczot et al., 1973). If the model of inverted terminal repetition described above and illustrated in Figure 1 is correct, and the inverted terminal repetition is of sufficient length, then the regions of annealed inverted terminal repetition should be visible as a projection of double-stranded DNA from the single-stranded circles. Estimates of the extent of self-complementary sequences in DNA from this virus have ranged from as high as 20% (Carter et al., 1972) to a very small percentage (Koczot et al., 1973). Visualization of projections shown to represent the inverted terminal repetition would allow a direct measurement of the length of the self-complementary regions. Because of the probability that the 267
268
K.
I.
BERNS
AND
T.
J.
KELLY
JR
Monomer
FIG. 1. The simplest by linear,
single
model polynucleotide
for the formation chains oontsining
of single-stranded oiroular monomers an inverted terminal repetition.
and dimers
DNA contains both natural and inverted terminal repetitions, a comparison of their relative lengths would be helpful in constructing a model that could account for the presenceof both. We have been able to repeat the observation of Koczot et al. (1973) that linear adeno-associated virus single polynucleotide chains can cyclize when exposed to annealing conditions. A preparation containing both types of intact, linear, complementary single strands yielded many single-strandedcircles after partial annealing. In this preparation 96% of the linear single strands had no visible projection while 92% of the single-stranded circles did have projections. However, 28% of the singlestranded circles contained more than one apparent projection and the possibility remained that the projections observed on the circles were artifacts caused by a twisting of the circles when they were mounted for microscopy. Additionally, it was not possibleto determine whether the projections occurred at any specific site about the circle. Koczot et al. (1973) have previously reported such projections but were also concerned that they might have been artifacts. To resolve this question the formation of single-stranded circular dimers was promoted by annealing only one of the complementary strands at a relatively high concentration. If the projections previously observed on single-stranded circular monomers had represented the base-paired, inverted terminal repetition region of the DNA, the majority of single-stranded circular dimers so produced should have contained two projections separated by 180” and any circular dimer containing more than two projections should have at least one pair separated by 180” (Fig. 1). To perform this experiment adeno-associatedvirus 2, containing l*C-labeled DNA in which bromodeoxyuridine had been partially substituted for thymidine (40%), was grown in KB cells in spinner culture (Rose et al., 1969). The complementary strands of bromodeoxyuridine-labeled virus DNA can be separated by banding in a neutral isopycnic cesium chloride density gradient (Berns & Rose, 1970). The heavy single strands were isolated and annealed at a concentration of 20 pg/ml in 50% formamide at room temperature for 24 hours. Examples of the single-stranded circles observed (both monomers and dimers) are demonstrated in Plate I. Of 107 single-stranded circular monomers photographed, 82 had only one projection and seven had no visible projection, the remainder having more than one apparent projection (Table 1). Twenty-five single-stranded circular
PLATE I. Representative single-stranded circular monomers virus DNA with projections (arrows) observed in the electron mounted using the formamide-protein film method (Davis JEMlOOB electron microscope. Magnification 60.000 >
and dimers microscope. et OZ., 1971)
of adeno-associatacl The molecules wwv and observed ill th,x I frrcina
,L ?tih
LETTERS
TO
THE
TABLE
The number
of projections Type
1
per molecule
of DNA
Monomer Dimer
o
of adeno-associated lprojeo~
7 0
269
EDITOR
82 0
12 16
3
4
3 5
0 4
viru.s DNA
dimers were photographed; of these 16 had two projections and the remainder had either three or four (Table 1). Measurements of 71 circular monomers and 25 circular dimers yielded average lengths of 1.56 f 0.12 pm and 3.12 f 0.22 pm, respectively (Fig. 2). In Figure 3(a) we have plotted the shorter of the two measured interprojection distances as a fraction of the total measured length of each circular dimer. Every molecule but one had at least one pair of projections that were separated by a distance equal to 0.45 to 050 of the total length of the molecule. Three circular dimers had two such pairs. Other interprojection distances occurred randomly. We have plotted the relative interprojection distances measured for those dimers with only two projections in Figure 3(b). In this case all but one of the shorter interprojection distances represent more than 0.45 of the total molecular length. From these data we conclude that the majority of projections (at least 48 out of 61) occur in the relative positions expected along the length of single-stranded circular dimers and thus that the majority of the projections most likely represent the hydrogenbonded, inverted terminal repetition regions holding circular dimers together. The relative lengths of the projections were determined for 50 single-stranded circular monomers that had just one projection. These ranged from 1.0 to 25% of the total length with a mean of 15%. The value of 15% for the inverted terminal repetition is close to the value of 1.0% determined for the length of the normal
30-
25-
:1 I
I.0 Relotw
I P-?-L length
FIQ. 2. Histogram of the relative messwed lengths of single-stranded dimers, oonsidering the mean of the monomers (1.66 pm) to be 1.0. Lengths 8 map measurer (KeulYel & Easer) and 8 grating replica (54,864 lines/inah, 18
circular monomers were determined E. G. Fullem).
and wing
270
K. I. BERNS
AND
T. J. KELLY
JR
L” (a)
15-
(b)
IO -
5-
I II 0.10 Interprojection
I I I in 020 distance
O-30
0.40
(fraction
r 0 50 of totol
length)
FIQ. 3. (a) Histogram of the shorter interprojeotion distances between EU possible pairs of projeotions plotted as the fractional length of the involved single-stranded oiroular dimer. (b) Histogram of the shorter interprojeation distances measured on all single-stranded oiroul8r dimers having two projections, plotted BS the fraotioxml length of the involved dimer.
terminal repetition in adeno-associatedvirus DNA described by Gerry et al., 1973, and would thus be compatible with the possibility that this DNA contains a terminal repetition whose sequence is symmetrical (122’1’----122’1’). Two other ah&natives, however, are not ruled out: (1) different molecules have either a natural or an inverted terminal repetition, but not both, or (2) the two types of terminal repetition are at different sites (12---2’1’12). If the projections observed do indeed represent the region of terminal repetition, the lengths would correspond to 49 to 100 nucleotide pairs, a length which has been observed previously using this technique of visualizing DNA molecules (Davis et al., 1971). Uncertainty as to the absolute length of the region of inverted terminal repetition arisesfrom the fact that the projections may be longer than the absolute base-paired region because nearby basesmay be held in apposition although unpaired or, conversely, the projections may not represent the full length of the base-paired region because of undetected folding. However, in agreement with Koczot et al. (1973) we do not fmd any evidence for the extent of self annealing (up to 20%) suggestedby Carter et al. (1972). In summary, we have presented evidence for projections on single-stranded circular dimers of adeno-associated virus DNA which we conclude correspond to the region of inverted terminal repetition contained within the DNA becauseof the
LETTERS
TO THE
EDITOR
271
specific locations of the projections along the contour of the molecules. These observations are oompatible with an inverted terminal sequencerepetition of the type 12----2’1’, which would allow base pairing of complementary regions in the anti-
parallel configuration. We thank D. Nathans, H. 0. Smith, B. Weiss and T. C. Pinkerton for helpful suggestions with regard to the manuscript. This work wez zupported by grants from the U.S. Public Health Service (noz lROlAI10843 and lROlCA13394) and the National Science Foundation (no. GB34163). One of us (K. I. B.) is a Hughes Medical Investigator. The other author (T. J. K.) is a reoipient of a Research Career Development Award from the National Institutes of Health (no. lK04CA70220-01). Department of Miorobiology The Johns Hopkins University School of Medicine Baltimore, Md 21206, U.S.A. Received
KENNRTE I. BERNS THOMAS J. KELLY JR
21 May 1973 REFERENCES
Atchison, R. W., C&o, B. C. t Hammon, W. McD. (1906). Science, 194, 764-760. Berns, K. I. & Adler, S. (1972). J. Virool. 9, 394-396. Berns, K. I. & Rose, J. A. (1970). J. Viral. 6, 693-699. Carter, B. J., Khoury, G. & Rose, J. A. (1972). J. ‘CTdroZ.10, 1118-1126. Crawford, L. V., Follett, E. A. C., Burdon, M. G. & McGeoch, D. J. (1969). J. Gen. Viral. 4, 37-46. Davis, R. W., Simon, M. & Davidson, N. (1971). In Met&da im Elw%(mology (Grossman, L. & Moldave, K., eds), vol. 21, pp. 413-428, Academia Press, New York. Garon, C. F., Berry, K. W. & Rose, J. A. (1972). Prw. Nat. Acud. Sci., U.S.A. 69, 23912396. Gerry, H. W., Kelly, T. J., Jr t Berns, K. I. (1973). J. Mol. Bid. 79, 207-226. Hoggan, M. D., Blacklow, N. R. & Rowe, W. P. (1966). Proc. Nat. Ad. hi., U.S.A. 65, 1467-1471. Koczot, F. J., Carter, B. J., Garon, C. F. & Rose, J. A. ( 1973). Proc. Nat. Ad. Sci., U.S.A. 70, 216-219. Mayor, H. D., Torikai, K., Melnick, J. L. & Mandel, M. (1969). Soierace, 166,1280-1282. Rose, J. A., Berns, K. I., Hoggan, M. D. & Koozot, F. (1969). Proc. Nat. Acud. Soi., U.S.A. 64, 863-869. Wolfson, J. & Dreesler, D. (1972). Pmt. Nat. Ad. Sci., U.S.A. 69, 3064-3067.