Evidence that the Human Foamy Virus Genome Is DNA

Abstract

The genomes of the spumaviruses, of which human foamy virus (HFV) is the prototype, are very similar to those of other complex retroviruses. However, in some aspects of the viral replicative cycle, HFV more closely resembles pararetroviruses such as hepatitis B virus. Previous work indicated that HFV extracellular particles contain apparently full-length double-stranded DNA, as well as RNA. We have further characterized the amount of DNA in particles and the role that this DNA has in viral replication. Experiments with the reverse transcriptase inhibitor 3′-azido-3′-deoxythymidine (AZT) suggest that reverse transcription is largely complete before extracellular virus infects new cells. In addition, we have been able to show that DNA extracted from virions can lead to production of virus after transfection. Taken together, these data suggest that complete, or nearly complete, proviral-length DNA is present in viral particles and that this DNA is sufficient for new rounds of viral replication.

Human foamy virus (HFV) is the prototype virus of the Spumavirus genus of retroviruses. It was originally designated as a human virus because the first isolation was from a human cancer-derived cell line (1); however, recent analyses show that HFV is of chimpanzee origin (11). The spumaviruses are complex retroviruses whose genomes contain the three hallmark retroviral genes, gag, pol, and env, as well as two or more genes encoding nonstructural proteins. One of these is a transactivator protein called Bel1 or Tas, which is required for transcription from the long terminal repeat (LTR) and internal promoters and absolutely required for infection (17, 18). The genomic organization of foamy viruses is identical to that of other retroviruses with respect to the LTRs at the ends of the proviral DNAs, a primer binding site required for negative-strand DNA synthesis, and the order of the structural proteins (reviewed in reference 25). The viral genomic sequences are so like those of other retroviruses that there was no reason to suspect that foamy viruses would significantly diverge from other retroviruses in any aspects of replication. Therefore, it has been surprising to discover that many details of the foamy virus replicative pathway substantially differ from those of conventional retroviruses such as murine leukemia virus or human immunodeficiency virus (HIV).

In some respects, foamy viruses more closely resemble pararetroviruses such as hepatitis B virus (HBV). For example, as in HBV, HFV polymerase protein is not synthesized as a polyprotein containing structural protein determinants, suggesting that Pol incorporation into particles occurs in a different manner from that of retroviruses (6, 16, 34). The Gag protein of HFV is not efficiently processed by protease (PR) into matrix, capsid, and nucleocapsid (NC) components (8, 14, 21). Additionally, HFV particles do not bud from cells unless viral glycoproteins are present (3), analogous to what is observed for HBV. But perhaps the feature of HFV that is most divergent from retroviruses is the presence of large amounts of apparently full-length double-stranded DNA in extracellular virions, which we found in an earlier study (34). In that study, we found that long viral DNA was in roughly 10 to 15% of the viral particles. This is in contrast to the case with human HIV, where it was shown that while about 0.1 to 1.0% of the virions contain strong-stop DNA, less than 0.001% of the virions contain full-length DNA (32) and this DNA is not required for infectivity (2). This suggested, as in the case of the hepadnaviruses, that a large amount of reverse transcription is completed before, or shortly after, the time when virions bud from infected cells. However, the role of the DNA in HFV replication was not determined in this prior study.

To examine whether the presence of DNA in extracellular virions is important for viral replication, several experimental approaches were taken. Using quantitative PCR methods (13), we have been able to show that about 20% of the particles contain long DNA. Some of the virion-associated DNA is infectious when transfected into tissue culture cells and can lead to the production of infectious virus. Finally, the effects seen with the reverse transcriptase (RT) inhibitor 3′-azido-3′-deoxythymidine (AZT) are most consistent with a DNA genome, in that pretreatment of cells has only a small effect on viral infectivity but treatment of infected cells abolishes production of viral infection. Taken together, these data indicate that preformed DNA is an important component of de novo HFV infection.

MATERIALS AND METHODS

Tissue culture methods.

Wild-type HFV with a deletion of the unique 3’ (U3) region of the LTR, called HFV13, is derived from the molecular clone pHSRV13 (18). pHSRV13 DNA is transfected into BHK cells. After extensive cytopathic effects are noted (5 to 7 days after transfection), cell lysates and viral supernatants are combined and used to infect diploid human embryonic lung (HEL) fibroblast cells as previously described (36). For some experiments, virus was harvested from the supernatant of chronically infected H92 human erythroleukemia cells (37). Viral infectivity assays using FAB cells (BHK cells containing HFV LTR–β-Gal DNA) were done as previously described (36) except that cells were stained with X-Gal (5-bromo-4-chloro-3-indolyl-β-d-thiogalactopyranoside) for 4 h at 37°C and then left at room temperature overnight before blue cells were counted. FAB and BHK cells were grown in Dulbecco modified Eagle (DME) medium with 5% fetal calf serum, HEL and NIH 3T3 cells were grown in DME medium with 10% fetal calf serum, and H92 cells were grown in RPMI medium with 5% fetal calf serum. Transfection and virus harvest were done as described for HFV13. A Moloney murine leukemia virus (MLV) vector, LNCZ, containing the β-galactosidase (β-Gal) gene (19), was packaged with MLV Gag and Pol proteins and the vesicular stomatitis virus (VSV) G protein (33) and obtained from Dusty Miller (Fred Hutchinson Cancer Research Center). MLV(VSV) infections were done in the presence of 8 μg of Polybrene per ml.

For treatment with inhibitors, FAB cells were pretreated for 4 h with inhibitor and then infected with cell-free virus stocks suspended in medium containing inhibitors. Virus was left on the cells for 4 or 24 h, and then the medium was changed to fresh medium containing inhibitor. The cells were fixed at different times after infection and stained for β-Gal activity. Individual H92 cell clones derived from a culture persistently infected with HFV (37) were continuously cultured with 100 μM AZT for several months. Periodically, cell lysates were made and combined with culture supernatants and assayed on FAB cells for infectious virus. AZT was from Sigma, ddI was from L. Corey, (Fred Hutchinson Cancer Research Center), and lamivudine (3TC) was from R. Schinazi (Emory University, Atlanta, Ga.).

Transfections were done with Lipofectamine or Lipofectamine Plus (GIBCO BRL, Gaithersburg, Md.). FAB cells (1.2 × 10⁵) were plated in 35-mm-diameter tissue culture dishes. The next day, the cells were transfected in accordance with the manufacturer’s directions. The transfection mixture contained 5 μl of Lipofectamine and different amounts of virion DNA extracted from the cell-free supernatants of acutely infected HEL cells or chronically infected H92 cells. DNA was left on the cells for 3 h. A linear fragment of pHFV13 digested with EagI and SalI (12,070 bp) was used as a positive control. The 12-kb fragment was purified on agarose gels prior to transfection. FAB cells (8 × 10⁴) were plated in 12-well (17-mm-diameter) tissue culture dishes. The cells were passaged once to 35-mm-diameter dishes before being stained.

Cell-free supernatants were obtained from cell lysates plus supernatants of either acutely infected HEL cells or chronically infected H92 cells. The supernatants were filtered through 0.2-μm-pore-size filters or centrifuged at 2,700 × g for 30 min at 4°C to remove cell debris. Virus was pelleted through a 20% sucrose cushion containing standard buffer (SB; 100 mM NaCl, 10 mM Tris, 1 mM EDTA [pH 8.0], 20% sucrose) by centrifugation at 24,000 rpm in a Beckman SW50.1 rotor for 2 h. Pellets were resuspended in SB with 10 mM MgCl₂ at 1/1,000 the original supernatant volume. Samples were divided in half to isolate RNA and to isolate DNA in parallel. These concentrated virus samples were then treated with RQ-1 RNase-free DNase (Promega, Madison, Wis.) (1 μl per 50 μl of sample volume) at 37°C, for 1 h, before the viral suspensions were divided into two aliquots for extraction of RNA or DNA.

In some experiments, HFV particles obtained from pooling lysates and concentrated supernatants from HFV-infected HEL cells were purified on sucrose gradients after pelleting through sucrose. Pellets were resuspended in 200 μl of SB, treated with RQ-1 RNase free DNase, layered on top of a 5-ml 20 to 60% sucrose gradient (in SB), and centrifuged in a Beckman SW50.1 rotor for 2 h at 36,000 rpm. Five-hundred-microliter fractions were collected, and the refractive index of each sample was measured. Ten microliters of each fraction was assayed for infectivity by using FAB cells. Fractions with a density of 1.14 to 1.17 g/cm³, and containing the peak of infectious virus, were pooled, diluted 1:5 in SB, and centrifuged for 2 h at 24,000 rpm. The resulting pellet was resuspended in 200 μl of SB and used to isolate nucleic acid after RNase or DNase treatment as described below.

Nucleic acid extractions from HFV particles. (i) RNA.

An equal volume of buffered phenol, containing 4 M guanidinium isothiocyanate, was added at a 2:1 ratio to the concentrated virus sample. Samples were extracted twice with chloroform-isoamyl alcohol (24:1). Nucleic acid was precipitated in the presence of 10 μg of carrier glycogen (RNase free), and pelleted nucleic acids were resuspended in the original sample volume in diethylpyrocarbonate (DEPC)-treated double-distilled water (ddH₂O). Samples were digested with RQ-1 RNase-free DNase (Promega) at 37°C for 1 h. The sample was then reextracted as described above and resuspended in the original volume of DEPC-treated ddH₂O.

(ii) DNA.

Sodium dodecyl sulfate was added to the concentrated virus sample at a final concentration of 0.5%. Samples were extracted twice with a 24:24:1 mixture of phenol-chloroform-isoamyl alcohol, and nucleic acids were precipitated with ethanol. Pelleted nucleic acids were resuspended in the original volume of ddH₂O and treated with RNase A (Sigma) at 37°C for 1 h. The sample was then reextracted as described above and resuspended in the original volume of ddH₂O.

(iii) RNA and DNA extraction from gradient-purified virions.

Particles were treated with RQ-1 RNase-free DNase as described above. Nucleic acid was extracted from gradient-purified virus by using the QIAmp HCV RNA isolation kit (Qiagen, Valencia, Calif.) as specified by the manufacturer. Samples eluted by this procedure contained both RNA and DNA. The samples were divided into two equal portions. One half was treated with RNase to obtain DNA, and the other was treated with DNase to obtain RNA. Both samples were then extracted with phenol-chloroform-isoamyl alcohol as described above. The samples of RNA or DNA were then precipitated with ethanol and resuspended in equal volumes of DEPC-treated ddH₂O.

Quantitative competitive PCR and RT-PCR.

The competitor DNAs used were derived as follows. Gag plasmid pCR/H3RR, containing the NC domain of gag (nucleotides [nt] 2623 to 3270), was constructed by inserting a 647-bp PCR fragment from a previously described gag deletion mutant cloned in the pCR II (2.1) vector (Stratagene, La Jolla, Calif.). The mutation in pCR/H3RR is a 55-bp deletion from nt 2074 to 2129 in the NC domain of gag (35). Plasmid pKS/dU3R, containing a portion of the U3 repeat (R) region of HFV (nt 316 to 868), was constructed by inserting a mutant 485-bp PCR fragment (digested at unique SmaI and HindIII sites within the primer sequences) into the pBluescript II KS+ vector (Stratagene). The mutation in pKS/dU3R is a 67-bp deletion, created by a two-step PCR mutation, between nt 660 and 727 in the U3 and R regions of the HFV genome. RNAs for use as competitors in the RT-PCRs were transcribed from pCR/H3RR and pKS/dU3R by using the enzymes SP6 and T7 RNA polymerase, respectively. Both RNA and DNA were quantitated by determining the optical density at 260 nm.

PCR.

Each PCR mixture consisted of 1× PCR buffer (Perkin-Elmer, Branchburg, N.J.), 1.5 mM MgCl₂, 0.1 mM deoxynucleoside triphosphate mix (Gibco/BRL, Grand Island, N.Y.), 1 U of Perkin-Elmer Taq polymerase, and 4 ng of each primer. Samples were denatured for 2 min at 95°C before thermal cycling was done. Temperatures for denaturing, annealing, and extension were 95°C, 44°C (U3-R) or 53°C (H3RR), and 72°C, respectively, at 1 min each for 30 cycles. The final extension reaction consisted of specific annealing at 72°C for 10 min, extension at 72°C for 1 min, and primer extension at 72°C for 10 min. The U3-R versus HFV PCR was done by using primers 316 (5′-CTGCCGGGATCAGAACATTGACAGA) and 868 (5′-AAGCTTCAGCGAGTAGTGAAG). The Gag versus HFV PCR was done by using primers NC#8 (5′-ACTTCTAGACCCTCTCAAGGACCAG) and PR#2 (5′-CATGGGTACCGTTGCCCCTGAATCCCAG). (the HFV complementary primer sequence is underlined.)

RT-PCR.

A constant amount of viral RNA and increasing amounts of competitor RNA transcripts were combined in reverse transcription-PCR reaction mixtures containing 2.5 U of avian myeloblastosis virus RT (U.S. Biochemicals, Cleveland, Ohio) and 6 U of RNase inhibitor (Boehringer Mannheim, Indianapolis, Ind.) at 42°C for 45 min. Reaction mixtures were then subjected to PCR as described above.

All PCR and RT-PCR products were electrophoresed on 2% Nusieve agarose gels in 1× Tris-borate-EDTA buffer.

RESULTS

Quantitation of RNA and DNA in particles.

Our previous studies (34) showed that about 10 to 15% of extracellular viral particles contain ca. 12.4 kb of double-stranded DNA as detected by Southern blotting and electron microscopy. DNA of this size is expected to contain the entire HFV genome. To more carefully quantitate the amounts of RNA and DNA in HFV virions, we used quantitative competitive PCR and RT-PCR (24). For this assay, competitor DNAs and RNAs were prepared in vitro. These competitors are identical to the viral sequence except for a small region between the PCR primers, which is deleted to allow differentiation of PCR products resulting from viral and competitor templates. When the level of DNA or RNA competitor and the level of test nucleic acid are the same, an equivalent amount of PCR product is seen for each.

Virus was prepared from the supernatants from chronically infected H92 cell lines or from acutely infected HEL cells. Viral particles were pelleted through sucrose cushions and treated with DNase before nucleic acids were extracted from the virions. Since HFV-infected cells contain a large amount of linear viral DNA, we wanted to minimize the amount of contaminating DNA from lysed cells. We also used virus that had been treated with DNase and then further purified on sucrose density gradients. After virions were lysed, equal aliquots of the samples were treated with DNase before isolation of RNA or with RNase before isolation of DNA. Two different sets of PCR primers were used. One set of primers amplifies sequences between the U3 and R regions of the viral LTRs. This set of LTR primers is expected to hybridize to two locations in two-LTR-containing viral DNA but only once in viral RNA (Fig. 1A). The second set of primers hybridizes to the 3′ end of the viral gag gene (Fig. 1B). Both sets of primers will hybridize only to DNA which has been extended after the first reverse transcriptase jump and do not measure strong-stop DNA created after extension from the primer binding site to the 5′ end of the RNA.

FIG. 1.

Quantitative competitive RT-PCR and PCR of foamy virus particle nucleic acids. (A) Map of the ends of two-LTR-containing linear HFV DNA showing the location of the U3-R primers. The boxes represent the ends of the viral DNA, with the ends of the viral RNA indicated above by a thick solid line. The lines flanked by arrowheads below the boxes indicate the locations of the U3-R primers used. Δ67, deletion in the competitor nucleic acids. Abbreviations: U5, unique 5′ region; PBS, primer binding site. (B) Map of the 5′ end of the HFV DNA showing the location of the gag primers. The location of the primer is indicated by the line flanked by arrowheads. Δ55, deletion in the competitor nucleic acid, from the H3RR mutant. (C) RT-PCRs (left) and PCRs (right) using the U3-R primers. Viral nucleic acids were resuspended after ethanol precipitation so that each 1-μl aliquot represents 1 ml of original cell supernatant. The amount of competitor RNA (RT-PCR) or DNA (PCR) used is shown above each lane. Lanes 1 contain competitor only in the reaction mix, and lanes 7 show reactions done without competitor. (D) RT-PCRs (left) and PCRs (right) using the gag primers. Nucleic acids and designations are as for panel C, except that no competitor is in lane 9 for RT-PCRs or lane 7 for PCRs. The competitor-only lanes for the RT-PCR reaction are not shown.

Using deleted competitor RNAs or DNAs, we were able to directly assess the relative concentrations of RNA and DNA in extracellular particles. We initially did titrations over a wide range of competitor RNA or DNA concentrations and then focused on the concentrations around the equivalence point in the PCRs. Experiments were done with virions from the supernatants of both persistently infected H92 cultures (Fig. 1) and from acutely infected HEL cultures (data not shown), with the same results. Figure 1 shows the results of one set of experiments. We compared the competition levels for RNA and DNA by using the U3-R primers and found that there were equivalent competition levels for the RNA (RT-PCR) (Fig. 1C, left panel) and DNA (PCR) (Fig. 1C, right panel) in the extracellular particles. In both cases, there was approximately 50% competition with 2 ng of competitor. Since full-length viral DNA contains two copies of the U3-R region and the RNA contains only one copy, this indicates that there is about twice as much RNA containing U3-R sequences as there is DNA. Using the primers which hybridize to the 3′ end of gag, we found equivalent signals from viral nucleic acid and competitor with 0.2 ng of competitor in the RT-PCRs (Fig. 1D, left panels) and 0.1 ng of competitor in the PCRs (Fig. 1D, right panel). This again indicates that there is about a 2:1 ratio of RNA to DNA in extracellular virions. This experiment was performed seven times, with variations in the level of competition. When the H3RR competitor was used, the ratio of RNA to DNA varied from 2:1 to 10:1, with an average of 4.6:1. We also did the competition experiment with virus that had been purified over a 20 to 60% sucrose gradient, from which gradient fractions of 1.4 to 1.7 g/ml were pooled. When fractions containing the peak of infectivity were pooled and repelleted, we found that we had an approximately 30-fold loss of both RNA and DNA. However, the ratio of RNA to DNA in the sucrose gradient virus did not vary from that in the nongradient purified virus. We found that again there was a ratio of RNA to DNA of 4:1 in the gradient-purified particles. Therefore, we conclude that about 20% of the particles released from HFV-infected cells contain DNA intact enough to have 3′ gag sequences.

Infectivity of virion DNA.

To determine whether any of the virion-associated DNA is full length and capable of directing the synthesis of infectious virions, we prepared DNA from extracellular particles. These particles were treated with DNase before nucleic extraction was performed to minimize contamination with cellular DNA. DNA was then introduced into FAB cells by using Lipofectamine, or Lipofectamine Plus reagent, which we have found to be the most efficient way of transfecting these cells. Although some of the virion DNA appeared to be of proviral length by Southern blotting (34), it is possible that the DNA contains nicks or gaps, as in the case of HBV genomes (reviewed in reference 7). As a control, we used full-length two-LTR-containing linear DNA excised from the pHFV13 molecular clone. Rather than staining the cells directly, transfected cells were trypsinized and replated on larger plates to allow cell growth and viral spread and thus to amplify the signal. The results are given in Table 1. We found that we could detect infectivity in 1 ng of plasmid DNA and in 50 ng of virion DNA. Smaller amounts of either plasmid or virion DNA were negative for infectivity. Controls without added DNA or with 1,000 ng of irrelevant pCITE plasmid DNA yielded no blue cells. All of the positive plates showed clusters of cells staining positive for β-Gal (Fig. 2). At the higher levels of DNA, transfection with either control plasmid DNA (Fig. 2A) or virion DNA (Fig. 2B) led to formation of giant multinucleate syncytia and massive cytopathic effects. Even at low concentrations of virion DNA (Fig. 2C and D), many β-Gal-positive cells were seen in clusters, indicating the spread of infectious virus. Thus, virion DNA is about 1/50 as infectious as plasmid DNA. We have not been able to repeat the infectivity assays with sucrose gradient-purified DNA because of the ca. 30-fold loss resulting from such purification. It is not feasible to grow and purify enough virus to do this experiment.

TABLE 1.

Infectivity of DNA extracted from virions

pHFV13 plasmid DNAa (ng)	Infectivityb	Control plasmid DNA (ng)c	Infectivityb	Virion DNAd (ng)	Infectivityb
5.0	+	1,000	−	1,000	+
2.0	+			500	+
1.0	+			200	+
0.1	−			100	+
0.01	−			50	+
None	−			10	−
				1	−

^apHFV13 DNA was linearized with the restriction enzymes EagI and SalI, which releases the viral DNA, and the viral DNA was gel purified, before transfection. 

^bCells were transfected with Lipofectamine Plus in 12-well dishes. Two days later, all of the cells were trypsinized and passaged to six-well dishes. Two days later, the cells were stained with X-Gal. All transfections were done in duplicate. +, foci or blue cells were seen on the FAB cells after staining; −, no foci or blue cells were seen. 

^cThe control plasmid used was the cloning vector pCITE. 

^dDNA was extracted from virions released from HEL cells. Extracellular virions were treated with DNase before nucleic acid extraction with SDS-phenol. After extraction, nucleic acid was treated with RNase. 

FIG. 2.

Photographs of FAB cells transfected with plasmid or virion DNA. Plates obtained as described for experiment B in Table 1 were photographed at ×250 magnification after staining with X-Gal. Cells were transfected with 1 ng of linearized pHFV13 DNA (A), 1 μg of virion DNA (B), 100 ng of virion DNA (C), and 50 ng of virion DNA (D).

Effect of inhibitors on viral infection.

In the case of conventional retroviruses, viral particles contain essentially only RNA genomes, and reverse transcription is required as an early step after infection of new cells. Thus, infection is sensitive to inhibitors of reverse transcription, such as AZT. However, once cells are infected, reverse transcription is not required for production of infectious particles, so adding inhibitors to infected cells does not abrogate the ability to produce infectious virus (2) (Fig. 3, left). We reasoned that if the functional foamy virus genome was DNA rather than RNA, then inhibitor studies should give the opposite results (Fig. 3, right). That is, reverse transcription should be a late event in viral production and virions should be able to infect cells that had been pretreated with inhibitors. Conversely, virions produced from inhibitor-treated cells should not be infectious.

FIG. 3.

Diagram of the strategy of, and expected results from, AZT experiments. The left side of the diagram shows the expected results if the genome is RNA as in conventional retroviruses; the right side shows the predicted results for a DNA-containing virus. Symbols: ×, virus cannot replicate; √, virus can replicate.

We tried a series of RT inhibitors for their effects on viral infection and infectious virus production. We could not show any inhibition with the drug lamivudine (3TC) at concentrations known to inhibit HIV (data not shown). This is not surprising, given that the catalytic site sequence YMDD, which is thought to be important for lamivudine inhibition (28), is YVDD in HFV RT. Contrary to the findings in a published report (27), we were unable to demonstrate any effect on HFV inhibition with the inhibitor ddI, even at concentrations as high as 10 μM (data not shown). This may also be caused by the sequence being YVDD rather than YMDD (10). However, we did find that AZT was a potent inhibitor of HFV production, and detailed analyses were done with this drug. We used a high concentration of AZT in our studies, generally 100 μM, to ensure that any lack of an effect on viral infectivity was significant. This concentration is not toxic for FAB or H92 cells (data not shown).

Next, we analyzed the ability of AZT to prevent HFV13 infection of FAB cells which had been pretreated with AZT (Table 2). We found that in over a dozen different experiments, AZT had only modest effects on the ability of untreated virus to infect the cells, which had been pretreated with AZT for 4 h. In these experiments, viral infection was measured directly by staining the FAB cells with chromophore at different times after infection. The decreases ranged from 50- to 70-fold in one experiment to 3- to 7-fold in the majority of the experiments. The average difference in infection between untreated and cells treated with 100 μM AZT in these experiments was 0.17 at 48 h after infection. We examined the effect of AZT on a control retroviral vector, MLV encoding β-galactosidase, which was pseudotyped with VSV G protein. This allowed us to compare the effects of inhibitor in the hamster-derived FAB cells, as well as mouse NIH cells (Table 3). We found that AZT decreased infectivity, as measured by chromophore staining of the cells, by 300- to 1,000-fold, as was observed in similar experiments which measured the effects of AZT on HIV infectivity (4). In this experiment, the effect of AZT on HFV was about a 10-fold decrease in infectivity.

TABLE 2.

HFV infection of cells pretreated with AZT^a

Time after infection (h)	Titer(s) relative to no-AZT controlb
	25 μM AZT	100 μM AZT
24		0.25, 0.28
30		0.3, 0.09
48	0.022, 0.09 0.20, 0.62, 0.22, 0.06 (0.2)c	0.19, 0.13, 0.013, 0.05, 0.17, 0.27, 0.41, 0.22, 0.25, 0.30, 0.21, 0.12, 0.094, 0.18, 0.22 (0.17)
60	0.21	0.17
70		0.11, 0.08

^aCells were stained with X-Gal at the indicated times, and blue cells were counted. The ratio of blue cells in treated plates to those in untreated plates was determined. Each number represents the average of at least two plates in each experiment. 

^bCells were pretreated for 4 h with the indicated concentration of AZT or were left untreated. Cells were infected with between 100 and 1,000 infectious units of HFV, as measured by the FAB assay, in the continual presence of the drug. 

^cNumbers given in parentheses indicate average values. 

TABLE 3.

MLV(VSV) infection of AZT-pretreated cells^a

Cell type	Virus	No. of β-Gal+ colonies
		No AZTb	25 μM AZT	100 μM AZT
FAB	MLV	2 × 105	1.3 × 103 (0.0065)c	2 × 102 (0.001)
	HFV	9 × 103	7 × 102 (0.07)	1.6 × 103 (0.17)
NIH 3T3	MLV	3.5 × 105	7 × 102 (0.002)	4 × 102 (0.001)

^aCells were pretreated for 4 h with the indicated concentration of AZT or were left untreated. They were then infected with different dilutions of an MLV vector containing the β-Gal gene, pseudotyped with the VSV G protein or with HFV13, and incubated in the continual presence of the drug. 

^bFAB cells were stained 48 h after infection, and NIH cells were stained 72 h after infection. 

^cNumbers in parentheses indicate the ratio of the titer of AZT-treated cells to that of untreated cells. 

In several experiments, a time course of infection showed little change in the ratio of infectivity from 1 to 3 days after infection. At 3 days after infection, the control cells exhibited abundant syncytia formation, indicative of a spreading infection, while the AZT cultures did not have any syncytia. As discussed below, this is because AZT prevents formation of infectious particles. HFV derived from another infectious molecular clone, PHFV-2 (11), yielded similar results (data not shown).

We next examined the ability of FAB cells that were treated with AZT, and then infected, to produce infectious virus when assayed on untreated cells (Table 4). We found that in contrast to the results shown in Table 2, AZT had a marked effect on the inhibition of infectious virus production, which ranged from 100-fold to greater than 3,000-fold, in different assays. This is again different from the results with conventional retroviruses. We also examined the effect of AZT on infectious virus production from persistently infected H92 cell clones (Table 5). We grew these cells in 100 μM AZT and determined the titer of the virus after 6 days to 8 weeks. Again we saw variations in the assays, but in all cases, inhibition of infectious virus production was greater than 150-fold. Since the control MLV β-Gal vector cannot replicate, we were unable to use it in this assay.

TABLE 4.

Effect of AZT on ability of virus to infect untreated FAB cells^a

Time after infection (h)	AZT concn (μM)	Titer(s)b
80	25	0.01
48	100	<0.0003
	25	0.0002, <0.002
	50	0.0002
	100	0.0004

^aCells were pretreated with AZT 4 h before infection at a multiplicity of infection of <0.1. At the indicated times, the cells were scraped and added to supernatants and the supernatants were freeze-thawed. Lysates plus supernatants were assayed on FAB cells. Untreated control preparations had AZT added to the same final concentration as that of AZT-treated preparations, before titers were determined on FAB cells. 

^bTiters are expressed relative to that of the no-AZT control. 

TABLE 5.

Effect of AZT on ability of virus from chronically infected H92 cells to infect untreated FAB cells^a

Length of treatment with 100 μM AZT (days)	Cloneb	Titerc
6	A14	<0.007
48	A1	<0.0017
51	A1	0.006
56	A1	0.00015
	A2	<0.0004
	A5	0.002

^aAt the indicated times, cells were scraped and added to supernatants and the supernatants were freeze-thawed. Lysates plus supernatants were assayed on FAB cells. Untreated control preparations had AZT added to the same final concentration as that of AZT-treated preparations, before titers were determined on FAB cells. 

^bSingle-cell clones were expanded, and the presence of HFV was tested by determining the titers on FAB cells and by Southern blotting. Individual clones were continually treated with AZT for the times indicated, when aliquots were removed and lysates plus supernatants were assayed. 

^cTiters are expressed relative to that of the no-AZT control. 

DISCUSSION

Retroviruses and pararetroviruses share a requirement for RT activity in their life cycles. However, the timing of reverse transcription is very different for these two viral groups. Conventional retroviruses such as HIV have RNA genomes, and reverse transcription occurs primarily after infection of cells. Linear viral DNA is found in the cytoplasm and is transported into the nucleus, where it is integrated into the genome as proviral DNA. Proviral DNA is present for the life of the infected cell and is the template for viral transcription and translation. In contrast, virus budded from hepadnavirus (pararetrovirus)-infected cells contains gapped circular DNA which is converted to covalently closed circular DNA (cccDNA) after infection, a step which is thought to require RT activity. This cccDNA does not usually integrate into the genome. Infection is maintained by small numbers of cccDNAs per cell, since the infected cell is a mature hepatocyte which does not normally divide (30).

The genomes of foamy viruses clearly place them in the retrovirus family, yet they share more features with the hepadnavirus group of pararetroviruses than other retroviruses do. The work presented here lends credence to the idea that the foamy virus group is unique among the retroviruses, and its replication strategy has evolved in a different direction more closely related to that of the hepadnaviruses. Thus, the most recent classification of retroviruses into seven genera (mammalian type B, mammalian type C, avian type C, type D, bovine leukemia virus–human T-cell leukemia virus, Lentivirus, and Spumavirus) is somewhat misleading, with Spumavirus being much more distantly related to the other six. While the nature of the genome of other members of the Spumavirus genus has not been characterized, both feline and bovine foamy viruses also synthesize RT from a spliced mRNA rather than as a Gag-Pol fusion protein (5, 12) and therefore are likely to activate RT at a similar stage in viral maturation as HFV. However, similar analyses of the virion genomes have not as yet been done.

Our data show that mature extracellular particles contain apparently full-length DNA molecules of about 12.4 kb in length that probably contain two LTRs. We had previously suggested that about 10 to 15% of the particles contain such DNA and that there were sufficient DNA-containing particles to account for viral infectivity (34). Using a more quantitative assay, we confirmed that about one-fifth of extracellular particles contain DNA containing U3-R and gag sequences. This might appear to be a large amount of RNA in a virus with a DNA genome. Much of this RNA could be due to reverse transcription which is aborted at an early stage. There is no information as to how efficient reverse transcription is after infection of cells by conventional retroviruses, since it is not easy to purify uncoated cytoplasmic particles after reverse transcription, but before nuclear transport, and examine them for nucleic acid content. Clearly, an integrated provirus does not result from every infecting retroviral particle, and it may be that reverse transcription occurring early in infection is also quite inefficient. We have no further information about the structure of the HFV DNA and do not know whether the DNA molecules contain nicks. However, a fraction of the virion DNA can lead to production of viral particles when transfected into susceptible tissue culture cells. The 50-fold-lower infectivity compared to that of plasmid-derived DNA could indicate that many of the packaged DNA molecules contain nicks or gaps which need to be repaired prior to integration. In fact, gapped DNA intermediates have been reported in HFV-infected cells, probably as a result of positive-strand DNA initiation at more than one site (15, 31). If HFV nicks are normally repaired by RT, this could account for the low infectivity seen in our assay. It is not known whether DNA extracted from extracellular hepadnavirus particles is infectious because of the lack of suitable tissue culture infectivity assays.

These studies show that virion DNA is abundant and can lead to the production of infectious virus. The result of experiments with AZT, a potent inhibitor of RT, are consistent with DNA as the major functional genomic molecule. When AZT is added to chronically or lytically infected cells, little or no infectious virus is produced. If RNA can serve as a viral genome as well as DNA, then one would expect only a modest effect on viral infectivity since at least 60% of the virions contain RNA. It should be noted that there are no published reports showing the size of the HFV-packaged RNA, and it is possible that a large fraction of the RNA is less than full length. The results of pretreatment of cells with AZT show that wild-type levels of virus are never produced, and the highest level of infection we have seen is 30 to 40% of that in untreated controls. In some experiments, the level of virus after AZT treatment is much lower (Table 2). This could indicate that upon infection, limited reverse transcription is needed to produce DNA that is competent for integration. The fact that virus produced from AZT-treated cells has little or no infectivity (Table 4) is also inconsistent with a major role for RNA genomes in infectivity. However, AZT could be having some other effect on viral production. We have looked at viral protein in the supernatants of AZT-treated or untreated HFV-infected H92 clones and have not found any differences, even after months of treatment (6a). However, it is not possible to absolutely rule out a role for RNA in viral infection with the present data. While this manuscript was being prepared, similar, but less extensive, results with AZT treatment of HFV-infected cells were reported (20). We have not as yet found another inhibitor of HIV RT which prevents HFV replication. This may be due to the differences in the sequence at the site sensitive to HIV RT inhibitors. HFV could prove useful as a screen of RT inhibitors for their effects on a virus which efficiently infects human cells, including T cells and macrophages, and which naturally lacks the methionine in the YMDD motif as do a majority of AZT-resistant HIV particles (28).

Foamy viruses differ from conventional retroviruses in both the nucleic acid used as a genome and their mode of expression of the Pol protein, and these could be intimately related. In conventional retroviruses, Pol is expressed as a Gag-Pol fusion protein, and activation of protease (PR) and cleavage of Pol from the complex are relatively late events in the viral life cycle, occurring during or after budding (8, 21). In general, Gag-Pol proteins do not exhibit full RT activity (9). In the case of HFV, the primary Pol translation product is a PR-RT-integrase (IN) protein (23, 34). HFV PR has limited activity in the viral life cycle. Only one cleavage in the Gag protein, near the C terminus, is seen in mature virions (14). IN is efficiently cleaved from PR-RT-IN, and a PR-RT protein can readily be detected in infected cells. This protein is probably localized in intracellular virions, which are extremely abundant in infected cells and highly infectious (37). It is not yet known whether PR and RT are further cleaved from PR-RT or whether this fusion protein contains both activities. It is possible that premature reverse transcription and concomitant accumulation of DNA in released particles occur in HFV infection because the lack of Gag domains on the Pol protein allows early activation of the PR-RT-IN protein during viral assembly.

Foamy virus-infected cells contain many hundreds of copies of viral DNA (6a, 29). By using the retroviral paradigm, it has been assumed that this DNA is cytoplasmic linear DNA, as seen after acute infection by HIV and other retroviruses (22, 26). However, most HFV does not bud from tissue culture cells. Since budded virions contain large amounts of DNA, it is reasonable to assume that the cell-associated HFV DNA is actually in intracellular viral particles, which are present in large numbers both at intracellular membranes and in the cytoplasm (3). This intracellular pool of particles may be analogous to the cccDNA of hepadnaviruses and provides a way to recycle DNA into the nucleus for integration. Although very little is known about how foamy viruses are maintained in vivo for long periods of time, pools of intracellular DNA containing virions which escape immune surveillance could be a reservoir for persistent infection in infected animals.