Search

Parvovirus LuIII encapsidates single-stranded DNA of either plus or minus polarity with equal frequency, whereas the rodent parvoviruses MVMp and H-1 encapsidate minus strand DNA only. A full-length, infectious clone of LuIII was constructed and the complete nucleotide sequence of the genome was determined. Comparison of the LuIII sequence with those of MVMp and H-1 revealed that these viruses are virtually identical with respect to the genomic organization, location of regulatory signals, mRNA splicing patterns, and amino acid sequences of viral proteins. However, two regions of the LuIII sequence differ significantly from those of the rodent parvoviruses. At mu 92, LuIII has only one copy of a sequence found as a direct repeat in MVMp and H-1. Upstream of this sequence, at mu 89, there is an A-T-rich region, 47 nucleotides in length, unique to the LuIII genome. This A-T-rich region could represent a signal responsible for the totally different encapsidation patterns observed for these viruses.

Replication of parvoviruses requires cis signals located in terminal palindromes that function as origins of replication in conjunction with trans-acting viral and cellular proteins. A gel retardation assay was used to identify proteins in crude nuclear extracts of bovine parvovirus (BPV)-infected bovine fetal lung cells that interact with the hairpinned left end (3' OH terminus of the viral minus strand in the flop conformation) of BPV. Three specific DNA-protein complexes formed. One complex was shown to involve a BPV structural protein(s) by inhibiting its formation when antiserum specific for these BPV proteins was used. By specific competition with serum containing antibodies against the BPV nonstructural proteins, a second complex was shown to involve a BPV nonstructural protein. A third complex contained protein of cellular origin and was also formed with extracts of uninfected bovine fetal lung cells. DNA competition assays suggest that the viral proteins do not bind to the right hairpin, which differs in sequence and secondary structure from the left terminus, or to a BPV terminus that lacks the first 52 nucleotides, preventing formation of the stem of the hairpin. The cellular protein is regulated in a cell cycle-dependent fashion, with its binding activity increased in uninfected, actively dividing cells compared with contact-inhibited cells. Since autonomous parvovirus replication requires an S-phase factor for progeny formation, the terminal binding protein demonstrated here is a candidate for this factor.

All linear DNA molecules face special problems in replicating their 5' ends, as DNA polymerases add nucleotides only to pre-existing strands with free 3'-OH groups. Parvoviruses, a group of small animal viruses with a linear single-stranded DNA genome, cope with this problem by having palindromic terminal sequences that can fold back on themselves to form hairpin structures essential in priming DNA replication. The 3' terminal sequence that initiates replication becomes reversed in orientation during the process, and if the palindrome is imperfect, two different, reverse-complementary terminal sequences are generated. The relative abundances of the terminal sequence orientations at each end of the DNA molecules can be measured and give information about the replication process. From such clues, we developed a "kinetic hairpin transfer model" based on differential rates of hairpin formation and inversion processes depending on the conformations of the 3' termini. Numerical studies showed that this simple idea can account for the diverse pattern of DNA distributions observed in the family Parvoviridae. In this paper, we simplify the model to a set of coupled linear first-order ordinary differential equations in order to delineate its essential properties by Perron-Frobenius theory. Secondly, we examine our assumption of linear kinetics by modeling enzyme catalysis of the component steps of the hairpin transfer process. We show that the rate-determining step of the process is the binding of initiation complex to the self-priming hairpin structures. Furthermore, we find that if the replication machinery is saturated by DNA substrate late in an infection, the differential equations become non-linear but the steady-state DNA distribution is still given by the solution of our original linear equations.

The sequence homologies among the linear single-stranded genomes of several mammalian parvoviruses have been studied by electron microscopic analysis of the heteroduplexes produced by reannealing the complementary strands of their DNAs. The genomes of Kilham rat virus, H-1, minute virus of mice and LuIII, which are antigenically distinct non-defective parvoviruses, have considerable homology: about 70% of their sequences are conserved. The homologous regions map at similar locations in the left halves (from the 3' ends) of the genomes. No sequence homology, however, is observed between the DNAs of these nondefective parvoviruses and that of bovine parvovirus, another non-defective virus, or that of defective adenoassociated virus, nor between the genomes of bovine parvovirus and adenoassociated virus. This suggests that only very short, if any, homologous regions are present. From our results, we predict an evolutionary relationship among Kilham rat virus, H-1, minute virus of mice and LuIII. It is interesting to note that, although LuIII was originally isolated from a human cell line and is specific for human cells in vitro, its genome has sequences in common only with the rodent viruses Kilham rat virus, minute virus of mice and H-1, and not with the other two mammalian parvoviruses tested.

Two antigenically indistinguishable strains, 171 and 308, of Kilham rat virus (KRV) have distinct host ranges and contain capsid proteins of identical size, but with different isoelectric points. The single-stranded DNA genomes of the viruses are also the same size but appear to have different secondary and tertiary structures. The genomes of the two strains have nearly identical cleavage maps of 11 restriction endonucleases, except for differences in five restriction sites in the region bracketed by 0.63 to 0.90 map units (from the 3' ends of the virus strands). However, there is a lack of extended heterology in the nucleotide sequence of the two virus genomes, as judged by electron microscopic analysis of the heteroduplex of the two virus DNAs. This suggests that very subtle differences in the sequences of the genome, and possibly of the capsid proteins, may play a role in the host specificity without affecting the antigenic similarity of KRV strains.

Infection of synchronized bovine fetal spleen cells with bovine parvovirus results in changes in the levels and patterns of DNA polymerases alpha and gamma during the cell cycle. The pattern of DNA polymerase alpha activity closely paralled viral DNA synthesis and the production of progeny virus, and levels, of this enzyme were threefold greater than in mock-infected cells during the period of maximal viral DNA synthesis. DNA polymerase gamma activity remained slightly elevated during viral DNA replication. Levels and patterns of DNA polymerase beta were similar in mock- and virus-infected cells.

Sequence analyses of the left and right termini of LuIII virus show they are nonidentical imperfect palindromes of 122 and 211 nucleotides, respectively. The left terminus of the minus strand of LuIII DNA, uniquely in the flip conformation, can assume a T-shaped structure. The right terminus of the minus strand of LuIII DNA can assume a U-shaped structure, and it exists in either the flip or flop conformation. The termini of LuIII shared a high degree of sequence homology and showed conserved secondary structure with those of the rodent parvoviruses MVMp and H-1. LuIII, like adeno-associated virus, encapsidates equal amounts of plus- and minus-strand DNA. However, the sequence data for LuIII virus demonstrate that identical termini are not required for this encapsidation pattern.

The distribution of terminal-sequence orientations in the viral DNA of bovine parvovirus (BPV), an autonomous parvovirus, was studied by end labeling and restriction enzyme digestion and also by cloning. The left (3') end of the minus strand of BPV was found in two alternative sequence orientations (designated as flip and flop, which are reverse complements of each other), with a 10-fold excess of flip. This is in contrast to the autonomous rodent parvoviruses which encapsidate minus-strand DNA with only the flip orientation at this end. The right (5') end of the minus strand of BPV contained both sequence orientations with equal frequencies, as in the rodent parvoviruses. Sequence inversions were also detected at both ends of the plus strand, which makes up about 10% of the encapsidated BPV DNA. Each terminus of BPV DNA had a characteristic ratio of flip to flop forms, and this ratio was restored in the progeny DNA resulting from transfection with genomic clones of different defined terminal conformations. Replicative-form DNA showed the same distribution of terminal-sequence orientations as the reannealed plus and minus virion DNAs, suggesting that the distribution of flip and flop forms observed in virion DNA is not due to selective encapsidation, but rather to the specific distribution of replicative forms. The current replication model for autonomous parvoviruses, which was based on the available data for the rodent parvoviruses, cannot account for the observed distribution of BPV DNA. An alternative model is suggested.

Autonomous parvoviruses are thought to uniquely encapsidate single-stranded DNA of minus polarity. In contrast, the defective adeno-associated viruses separately encapsidate equal amounts of plus and minus DNA strands. We reexamined the uniqueness of minus strand encapsidation for the autonomous parvoviruses. Although we found that Kilham rat virus and H-1 virus encapsidate varying but small amounts of complementary-strand DNA, it was unexpected to find that LuIII virus encapsidated equal amounts of plus and minus DNA. The extracted LuIII DNA possessed properties of double-stranded replicative-form DNA, including insensitivity to S1 endonuclease, cleavage by restriction enzymes, and conversion to unit-length, single-stranded DNA when electrophoresed under denaturing conditions. However, the inability of this DNA to form single-stranded DNA circles when denatured and then renatured in the presence of formamide and the lack of double-stranded DNA circle formation after treatment with exonuclease III and reannealing shows a lack of sequence homology of the 3' and 5' termini of LuIII DNA, in contrast to adeno-associated virus DNA. Digestion of LuIII double-stranded DNA with EcoRI and HincII and separation of plus and minus DNA strands on composite agarose-acrylamide gels identified a heterogeneity present only in the plus DNA strand. These results suggest that strand specificity of viral DNA encapsidation is not a useful property for differentiation between the autonomous and defective parvoviruses. Furthermore, encapsidation by LuIII of equal amounts of complementary DNA strands in contrast to encapsidation of minus strands by H-1 virus, when propagated in the same host cell type, suggests that selection of strands for encapsidation is a virus-coded rather than host-controlled event.

A physical map has been constructed for bovine parvovirus (BPV) replicative form (RF) DNA synthesized in vitro. Analysis of restriction enzyme digestion products for 12 enzymes by neutral gel electrophoresis allowed the mapping of 26 cleavage sites between the 3' and 5' termini of the in vitro RF DNA. Cleavage sites for these enzymes were not detected within the 3'-terminal hairpin. HindIII, SalI, SmaI, SstII and XorII did not cleave BPV DNA, while 12 other enzymes cleaved the genome at one to four sites. Eleven additional enzymes were shown to cleave the genome at 8 to 20 sites. Comparison of the restriction site locations for BPV with those of other autonomous parvoviruses suggests that there may be conservation within the genome of certain restriction sites.

We have developed a nuclear lysate system from infected, synchronized cells capable of synthesizing unit-length parvoviral DNA in vitro. It was necessary to supplement the nuclear lysates with the polyamines, spermidine and spermine, to prevent degradation of template and product DNAs. In this system RF, RI, and single-stranded progeny DNAs were synthesized. Label incorporated in viral RF DNA in vivo appeared first in RI DNA and then in single-stranded DNA during incubation in vitro. By sedimentation the product DNAs were identical to those found in infected cells. Their viral identity was confirmed by hybridization. The addition of ribonucleotides, RNase, or alpha-amanitin did not affect parvoviral DNA synthesis in this system. The results with the specific inhibitors of mammalian DNA polymerases, aphidicolin, N-ethylmaleimide, and 2',3'-dideoxythymidine 5'-triphosphate indicated that DNA polymerase alpha was required for synthesis of parvoviral DNA in the nuclear lysates. This requirement was confirmed by experiments with antibody to bovine DNA polymerase alpha.

Since parvoviruses apparently do not possess a DNA polymerase activity, one or more of the host cell DNA polymerases must be responsible for replicating the single-stranded DNA genome. We have focused on determining which polymerase, alpha, beta, or gamma (pol alpha, pol beta, or pol gamma, respectively), is responsible for the first step in bovine parvoviral DNA replication: conversion of the single-stranded DNA genome to a parental replicative form (RF). In this study, we used aphidicolin, a specific inhibitor of DNA pol alpha, to assay for the requirement of pol alpha activity in parental RF formation in vivo. Synchronized cell cultures were infected with bovine parvovirus with or without aphidicolin, and the products of viral replication were separated on agarose gels and identified by Southern blot analysis. We found that complete inhibition of viral DNA synthesis resulted when 20 microM aphidicolin was present throughout the infection. In addition, viral DNA synthesis was inhibited by as little as 1 microM aphidicolin, whereas lower concentrations (0.1 and 0.01 microM) resulted in partial inhibition of the replication process. Using 32P-labeled bovine parvovirus as the input virus we differentiated parental RF from daughter RF and progeny DNA synthesis. We conclude that DNA pol alpha is required for the production of RF during bovine parvovirus replication in vivo and that this requirement is most likely for the conversion of bovine parvovirus input single-stranded DNA to parental RF. These results do not rule out a possible role for DNA pol gamma in the first step, nor do they rule out a role for pol alpha or pol gamma in later stages of the replication cycle.

The DNAs encapsidated by parvoviruses show distinctly different patterns with respect to the ratio of plus-to-minus strands and sequence heterogeneity at the ends. A kinetic model, based on differential rates of hairpin transfer at 3' termini, is described and shown to account for all known parvoviral DNA distributions.

DNA polymerase gamma, purified from fetal bovine liver, replicated virion single-stranded DNA from bovine parvovirus to a unit-length double-stranded DNA molecule. This product was not nicked and was covalently linked to the 3' hairpin primer. The reaction was inhibited by dideoxythymidine 5'-triphosphate, but was unaffected by ATP or aphidicolin. Double-stranded viral DNA was not a functional template for purified DNA polymerase gamma.