Foamy Virus Integration

Abstract

It had been suggested that during integration of spumaretroviruses (foamy viruses) the right (U5) end of the cDNA is processed, while the left (U3) remains uncleaved. We confirmed this hypothesis by sequencing two-long terminal repeat (LTR) circle junctions of unintegrated DNA. Based on an infectious foamy virus molecular clone, a set of constructs harboring mutations at the 5′ end of the U3 region in the 3′ LTR was analyzed for particle export, reverse transcription, and replication. Following transient transfection some mutants were severely impaired in generating infectious virus, while others replicated almost like the wild type. The replication competence of the mutants was unrelated to the cleavability of the newly created U3 end. This became obvious with two mutants both belonging to the high-titer type. One mutant containing a dinucleotide artificially transferred from the right to the left end was trimmed upon integration, while another one with an unrelated dinucleotide in that place was not. The latter construct in particular showed that the canonical TG motif at the beginning of the provirus is not essential for foamy virus integration.

Reverse transcription and provirus integration are hallmarks of retrovirus replication (15, 40). Like orthoretroviruses, foamy viruses (FVs) strictly depend on integration into the host cell genome mediated by their pol gene-encoded integrase (IN) (12, 25). The FV replication pathway, however, significantly deviates from that of orthoretroviruses and resembles in some aspects that of the reverse transcribing but nonintegrating hepadnaviruses (13, 24, 29, 33). Furthermore, provirus establishment appears to be unprecedented among exogenous retroviruses (12).

Orthoretroviral integration is initiated by IN-mediated removal usually of 2 nucleotides (nt) at both 3′-OH ends of the linear double-stranded DNA, resulting in the provirus precursor (4, 5). This reaction leaves the provirus to conventionally start with a TG and to terminate with a CA dinucleotide (4, 5). The proviruses of FVs also start and terminate with these dinucleotides (12, 25, 30). It has been suggested that a nick is produced at the right (U5) end of the linear FV DNA between the subterminal CA and a terminal AT dinucleotide, resulting in provirus termination with CA, while the left (U3) end remains uncleaved, retaining the corresponding TG motif (12). An unusual structure present in all FVs at the border between the polypurine tract (ppt) and U3 was claimed to be responsible for the proposed asymmetrical feature of the cleavage reaction (12).

By characterization of two-long terminal repeat (LTR) forms of circular viral DNA, different U3-end mutants, and corresponding proviral integrations sites, we wanted to prove this suggestion and to delineate more precisely this fundamental peculiarity in FV replication. Molecular clones of the prototypic FV (PFV) served as study tools.

MATERIALS AND METHODS

Cells and viruses.

Baby hamster kidney cells (BHK-21), BHK/LTR(HFV)lacZ cells (38), the human fibroblastoid HT1080 cells, kidney 293 cells, and 293T cells (10) were grown in Eagle's minimal essential medium or Dulbecco's modified minimal essential medium containing 5 to 10% fetal calf serum, antibiotics, and G418 (Invitrogen) when appropriate. Plasmid-derived viruses harvested after CaPO₄-mediated transfection of 293T cells were abbreviated with the plasmid name lacking the “p.” Cell-free (0.45-μm-pore-size filtrate) virus preparations were titrated on BHK/LTR(HFV)lacZ cells, which express an endogenous lacZ gene under the control of the PFV-LTR (38, 42). At least four independent titration experiments were performed for each mutant.

Recombinant DNA.

Conventional molecular cloning techniques (1, 36) were used to introduce mutations at the ppt-U3 junction of the infectious clone pcHSRV2 (38) as shown in Table 1. Briefly, a 2.0-kb ClaI/ApaI fragment of pcHSRV2 spanning the 3′ LTR and adjacent sequences was subcloned, and recombinant PCRs (21) were carried out on a 1.03-kb BglII/XbaI subfragment. The modified ClaI/ApaI fragments were reinserted into pcHSRV2. The bet mutants of pHSRV2-M4 and -M5 (2) were introduced into pcHSRV2 by exchange of a 2.84-kb NheI fragment encompassing these mutations. Some of the altered ppt-U3 junction sequences were also inserted into the replication-deficient FV vector pMH123. To do this, 5.38-kb AflII/MluI and 5.82-kb MluI/EcoRI fragments of pMH118 (23) were ligated with a 2.73-kb EcoRI/HindIII fragment from pMH92 (18) and a 0.44-kb pcHSRV2 HindIII/AflII (38) fragment. This four-fragment ligation generated the wild-type vector pMH123. pMH123/TJ-09, -10, and -11 were created by substituting the 0.44-kb HindIII/AflII fragment of the pTJ ppt-U3 junction mutations (Table 1) for the corresponding wild-type fragment. DNA sequencing of all amplimers was carried out to exclude undesired nucleotide exchanges.

TABLE 1.

Mutations introduced at the ppt-U3 junction and corresponding oligonucleotide primer pairs

Plasmid	ppt-U3a
pcHSRV-2 (wild type)	5′-GAGGG TGTGGTGGAA-3′
pTJ-01	5′-GAGGG TAGTGGTGGAA-3′
pTJ-02	5′-GAGG ATGTGTGGTGGAA-3′
pTJ-03	5′-GAGG TGTGTGGTGGAA-3′
pTJ-04	5′-GAGGG TGTGTGGTGGAA-3′
pTJ-05	5′-GAGGG TGGTGGAA-3′
pTJ-06	5′-GAGGG TATGGTGGAA-3′
pTJ-07	5′-GAGGG TATGTGGTGGAA-3′
pTJ-08	5′-GAGGG ATTGGTGGAA-3′
pTJ-09	5′-GAGGG ATGTGGTGGAA-3′
pTJ-10	5′-GAGGG ATTGTGGTGGAA-3′
pTJ-11	5′-GAGGG ACTGGTGGAA-3′

^aMutated nucleotides are shown in boldface type.

Protein detection.

Protein expression in lysates from transfected 293T cells and in partially purified viral particles was detected by immunoblotting as described earlier (16, 17).

Amplification and sequencing of circular viral DNA forms.

DNA was extracted from 293T or 293 cells, respectively, after transfection or infection with plasmid-derived virus. Total cellular RNA-depleted DNA was prepared using current standard protocols. A PCR for two-LTR circles was performed with 0.5 μg of template DNA with the primers 2LC3 (5′ TCT GCG GCT AGT ATA ATC A) and 2LC5 (5′ TTA GCC TTG CTA AGG GAG) under conditions essentially as described by Saïb et al. (35). The amplimers were inserted into pCR2.1 of the Topo TA cloning kit (Invitrogen). DNA sequence analysis was done with AmpliTaqFS and the ABI 310, 377, or 3100 automated DNA sequence analysis systems (Applied Biosystems) and primers 741 (5′ TGG TAC CGA GCT CGG ATC CAC) and 742 (5′ TCT AGA TGC ATG CTC GAG CGG).

Sequencing of integration junction sites.

The junctions between proviral and cellular DNA were amplified by a modified ligation PCR as described previously (12, 18). Briefly, 2.5 μg of genomic DNA was digested overnight with HaeIII. After heat inactivation of the enzyme, self-ligation, and overnight digestion with AvrII, a nested PCR was performed with primer pairs 489-188 and 571-572 (18). After purification the amplimer sequences were determined directly or after shotgun cloning into pCR2.1 with primer 572 as described above.

Virion DNA analysis.

After partial purification of virions by ultracentrifugation through sucrose cushions, DNA was extracted from the sedimented material after DNase treatment. A PCR was performed with primers 196 and 197, which hybridize in the IN domain (37), and with primers and conditions used to amplify two-LTR circles as described above.

Vector transfer.

A total of 1.8 × 10⁶ 293T cells were transfected with FV vectors and the Env expression plasmid pCenv-1 (17-20). Vector transfer was analyzed by inoculating 10⁴ HT1080 cells with 1 ml of cell (0.45-μm-pore-size filtrate) supernatant harvested 48 h after transfection. The percentage of cells successfully transduced with the marker gene coding for enhanced green fluorescent protein was determined 4 days after inoculation by flow cytometry on a FACScan (Becton Dickinson) as described previously (17-20).

RESULTS

Sequence analysis of wild-type virus two-LTR circles.

Two-LTR circles are considered to be dead-end by-products of the retroviral replication cycle since they do not yield viral progeny (4, 5, 15). These molecules are formed in the nucleus by processes occurring independently of the viral IN activity (4, 5, 15). It has been suggested that incorrectly or unprocessed ends prevent regular binding of IN to the attachment sites followed by integration of linear viral DNA. Instead, such DNA species may be prone for ligation to two-LTR circles (4, 5). The foamy proviral sequence starts with the motif TGTGGTGG (12). By sequencing representative numbers of amplimers derived from two-LTR circle junctions isolated from wild-type FV-infected cells, we intended to provide further evidence for the asymmetric cleavage of FV proviral ends.

A total of 41 molecular clones of wild-type FV two-LTR circles were analyzed. The TGTGGTGG motif of the proviral 5′ end was found in 19 clones (Table 2). Twelve clones harbored remainder sequences of various lengths derived from the ppt attached to U3, and four clones had U3 deletions of various lengths. One clone had an insertion of 22 bp derived from the FV pol gene, and the remaining five clones had insertions between U3 and U5 of 1 to 3 nt whose origin could not be assigned. The U5 ends appeared to be more heterogeneous (Table 2). Seven clones were processed at the U5 end and terminated with the CA dinucleotide. Eleven clones showed the unprocessed U5 end (CAAT), and eight had an incompletely cleaved end (CAA). Small deletions upstream in the U5 region were found in three clones. Six clones had polynucleotides of various lengths attached, which were derived from the pbs. The remaining clones had the insertions already described above.

TABLE 2.

Wild-type PFV two-LTR junction sequences

No. of clonesa	U5	pbs and ppt	U3
wt	CATGACAAT	TGGCGCCCAACGTGGGG − AGAGAGGAAGTAACGAGGAGAGGG	TGTGGTGG
1	CAT		TGTGGTGG
2	CATGA		TGTGGTGG
2	CATGACA		TGTGGTGG
1	CATGACA	t + 58 ntb + AGAGAGGAAGTAACGAGGAGAGGG	TGTGGTGG
4	CATGACA		—e
8	CATGACAA		TGTGGTGG
1	CATGACAAT		TGTGGTGG
6	CATGACAAT	GGAGAGGG	TGTGGTGG
3	CATGACAAT	AGGAGAGGG	TGTGGTGG
1	CATGACAAT	gtacAGGAGAGGG	TGTGGTGG
4	CATGACAAT	T	TGTGGTGG
1	CATGACAAT	TAGGAGAGGG	TGTGGTGG
1	CATGACAAT	TGGCGCCCAACGTGGGGC	TGTGGTGG
3	CATGACAAT	a	TGTGGTGG
1	CATGACAAT	aa	TGTGGTGG
1	CATGACAAT	ata	TGTGGTGG
1	CATGACAAT	T gaccaatttctgattttttcaad	TGTGGTGG

^aNumber of individual clones analyzed from a total of 41. wt, wild type.

^bAn additional 58 nt adjacent to the ppt were found inserted in this clone.

^cThe origin of these inserted nucleotides (in italics) is unclear.

^dInsertion derived from the FV pol gene.

^eDeletion of 123 to 160 nt from U3.

In summary, in none of the clones did we identify an additional 2 nt attached to the U3 end, and conversely the AT dinucleotide was readily observed at uncleaved U5 ends. These results strongly support the suggestion that 2 nt is cleaved from the right end of the linear viral DNA while the left end remains unprocessed. Deletions of U3 were found in only four cases, while cleavage of U5 was identified 18 times. Thus, the analysis of the two-LTR circle sequences also revealed the relative stability of the U3 end and a greater heterogeneity of the U5 end. These observations may also reflect the lack of left-end processing by FV IN. Furthermore, out of 41 clones only 3 were identified with correctly processed or unprocessed LTR junction sequences, which supports the hypothesis that a cDNA with untidy ends is not further used for integration and is prone to two-LTR circle ligation.

Design of virus mutants.

Eleven clones bearing mutations at the ppt-U3 junction sequence were constructed and analyzed for protein expression, reverse transcription, and export of viral DNA, as well as replication competence following transient transfection. The mutants contain either small insertions, deletions, or sequence alterations (Table 1). For clarity, all mutations are described for the coding strand, although during cleavage the 3′-OH end of each DNA strand is attacked. The wild-type start of U3 is characterized by the redundant TG motif. This motif was changed to harbor another 5′TG with (pTJ-02 and -03) or without (pTJ-04) altering the adjacent ppt sequence. In addition, we deleted the ultimate TG in pTJ-05 and pTJ-11. In the latter plasmid the human immunodeficiency virus (HIV) type 1-derived AC start of U3 was substituted for the FV terminal TG motif (39, 41). pTJ-01 and pTJ-06 to -10 bear other alterations. To force a cleavage reaction at the left end of the linear viral DNA the ultimate nucleotides (AT) of the U5 end were joined to the U3 end in pTJ-10. pTJ-01 and pTJ-06 to -09 harbor modifications of this dinucleotide. In pTJ-07, TA was changed to AT. pTJ-01 has a similar mutation combined with the deletion of the first T of U3. The U3 region of pTJ-06 also starts with TA as a substitute for the wild-type terminal TG motif. This wild-type terminus was replaced by AT from pTJ-08. pTJ-09 has an A upstream of TG, which results in U3 starting with AT (Table 1).

Protein expression, reverse transcription, and replication competence of FV U3-end mutants.

All mutants remained competent to express viral proteins after transfecting 293T cells. Gag and Pol proteins were detected in both lysates from transfected cells and partially purified extracellular virions (data not shown). Reverse transcription of a pregenomic RNA is a late event in the replication cycle of FVs and occurs in virus-producing cells. As a consequence, DNA is the functionally relevant virion genome (26, 34, 43). The presence of this molecule species in wild-type virus and mutants was shown by PCR with a primer pair to amplify a stretch of the IN domain (data not shown). Two-LTR circles were found neither in wild-type nor in any of the mutant-derived extracellular particles (data not shown). This is consistent with the view that their formation depends on cellular factors in the nuclei of infected cells (35).

We analyzed the transient-replication competence of the mutants by transfecting 293T cells and assaying cell-free supernatant on BHK/LTRlacZ cells 72 h later. Since all mutations involving the ppt-U3 junction disrupt or alter the bel-2 open reading frame, which is the major exon encoding the Bet protein (2), we therefore used the pcHSRV-2 derivatives M4 and M5 as controls in titration experiments, which similarly truncate Bet without significantly affecting in vitro virus titers (2).

As shown in Fig. 1 the mutants could be classified in three groups according to cell-free virus titer yield. The mutants TJ-02, -03, -05, and -07 gave very poor titers—on the average 4 to 5 orders of magnitude lower than those for wild-type HSRV2. TJ-01, -04, and -06 yielded intermediate titers approximately 2 orders of magnitude lower than wild-type virus. In contrast, the group of TJ-08 to -11 developed titers comparable to wild-type virus or the Bet⁻ controls M4 and M5.

FIG. 1.

Amounts of cell-free virus produced by cells transfected with proviral plasmids and titer determination by the LacZ assay (39). The results from four independent experiments and standard deviations (error bars) are shown.

Analysis of vector transfer rates.

The ppt-U3 junction sequence mutations of pTJ-09, -10, and -11 were also introduced into the replication-incompetent FV vector pMH123 (Fig. 2). Cotransfection of 293T cells with vector and an env-expressing plasmid allowed determination of enhanced green fluorescent protein-transducing units in cell-free supernatants as an alternative estimation to the transiently produced replication-competent virus and the blue-cell assay. Moreover, we speculated that wild-type FVs could have developed a particular strategy of attenuation by permitting cleavage only at one end of the linear DNA to eventually reduce the integration rate. Conversely, if cleavage at the left end were forced or allowed, as in mutant pTJ-10, this should lead to an increase in replication-incompetent vector transfer rates. However, this was not what we observed (Fig. 2). Compared with wild-type pMH123 vector, all mutants analyzed produced slightly less transduction. The reductions were less than threefold and may, therefore, have escaped the detection in the blue-cell assay. The experiment shown in Fig. 2 revealed a slight gradient: wild-type vector resulted in the highest transduction rate, followed by pTJ-9, with 1 nt to be cleaved; pTJ-10 allowed a TA dinucleotide cleavage; and, finally, the cleavage-resistant mutant pTJ-11 had the lowest transduction rate.

FIG. 2.

Vector constructs and transfer rates. 293T cells were transfected with the replication-deficient vector pMH123 or its derivatives with mutations of the 3′ LTR along with an Env-expressing plasmid. The transduction rates of HT1080 target cells expressed as mean percentages (± standard deviation) from four experiments indicated that the mutants transduced recipient cells less efficiently than wild-type virus-derived vector.

Two-LTR circle sequence of virus mutants.

Two-LTR circle junctions were amplified and sequenced after infection of cells with cell-free virus supernatants or after transfection of 293T cells. The latter way of analysis became necessary with the intermediately and the poorly replicating mutants pTJ-01 to pTJ-07. The results are summarized in Fig. 3. In particular, the least-efficient mutants, pTJ-02 and -03, showed a high rate of severe U3 deletions compared with wild-type junction sequences. This may indicate a relative instability of the modified U3 termini. In general, more U3-end alterations were seen in the group of the poorly replicating mutants, those which replicated close to wild-type levels. These U3-end alterations were of the type to contain either additional ppt sequences or deletions shortening U3. Mutants with remaining ppt sequences probably arose from aberrant plus-strand synthesis (32).

FIG. 3.

Overview of the results obtained upon sequencing the two-LTR junctions of mutant viruses. The figures on the left indicate the numbers of clones with unaltered sequences (first line), isolated U5-end deletions (second line), isolated U3 deletions (third line), deletions of both ends (forth line), and insertions regardless of additional deletions (fifth line) of the total number of clones analyzed. The two-LTR sequences are available as supplementary information at http://www.uni-wuerzburg.de/virologie/PFV2LTRsequences.html.

Analysis of mutant proviral integration sites.

The two-LTR circle analyses (Fig. 3) did not formally answer the question of whether any left-end DNA cleavage had occurred in the mutants upon integration. This was of particular interest when looking at the high-titer mutants pTJ-10 and pTJ-11, which were designed according to previously performed in vitro IN cleavage assays (31). The U3 end of pTJ-10 contains a nucleotide stretch that was readily cleaved, while pTJ-11 was modeled after a cleavage-resistant oligonucleotide (31).

Therefore, we determined the proviral junction sequences in cells transduced with vectors pMH123/TJ-09, pMH123/TJ-10, and pMH123/TJ-11, selected for G418 resistance, and subjected the culture to single cell cloning. As shown in Table 3 the wild-type pMH123 vector produced the viral integration terminus (TGTGGTGG) already known from previous studies (12, 25, 30). Curiously, TJ-9 and TJ-10 proviruses started with the same viral sequence. To generate these proviruses, in TJ-9 a terminal A:T nucleotide pair and in TJ-10 a terminal TA:AT dinucleotide pair were removed from the linear viral DNA. TJ-11 proviruses started with the motif ACTGGTGG, which is identical to the U3 sequence of this mutant (Table 3).

TABLE 3.

Left (U3)-end proviral junction sequences of FV mutants

Construct	Sequence
	Cell	U3-LTRd
pMH123/wt	CATTCAGTAT	TGTGGTGG
	ACAATGGCCC	TGTGGTGG
	TACTATCAAG	TGTGGTGG
	TTATCATTAG	TGTGGTGG
	ACACTAACAC	TGTGGTGG
	TTTTGCCATG	TGTGGTGG
pMH123/TJ-9a	GTCTCCGATC	TGTGGTGG
	TTCAGCATGG	TGTGGTGG
	AAAGGTAATG	TGTGGTGG
	GGGTTAGATG	TGTGGTGG
	TGTCAATGTA	TGTGGTGG
	CAGCTGTGAA	TGTGGTGG
pMH123/TJ-10b	GTGGAGGAGC	TGTGGTGG
	CTCCTAATAC	TGTGGTGG
	AACAATGGCC	TGTGGTGG
	CGGTTTCACC	TGTGGTGG
	GAGATGGAGT	TGTGGTGG
	CTCCACATAC	TGTGGTGG
pMH123/TJ-11c	GATGTTGGAC	ACTGGTGG
	CAGAACCCCC	ACTGGTGG
	ATGCTTACAG	ACTGGTGG
	AGGGTAGGGG	ACTGGTGG
	AAAGGGGACC	ACTGGTGG
	TTTAAAACAG	ACTGGTGG

^aU3: ATGTGGTGG.

^bU3: ATTGTGGTGG.

^cU3: ACTGGTGG.

^dMutated nucleotides are in boldface.

DISCUSSION

In all orthoretroviruses tailoring of viral cDNA termini is the initial IN-mediated step in provirus establishment (4, 5, 9, 14, 15, 40). Its function, however, is poorly understood. Two explanations for the cleavage reaction have been suggested: to produce defined ends of the linear unintegrated DNA (5) and to enable correct binding of the IN enzyme to the ends of the linear viral DNA (11). Initial work on extended deletions in the murine leukemia virus (MLV) system suggested cleavage at one end of the DNA to be required for cleavage of the other end (28). A later study performed with HIV delivered contradictory results (6). Moreover, in vitro assays with recombinant HIV IN showed the two terminal base pairs to be completely dispensable for any consecutive IN reaction (22). However, this contrasts again with the MLV system, where cleavage of the right end was shown to be essential for integration (7). Thus, IN-mediated cleavage may be virus specific.

At least in HIV there are good indications that the right-end dinucleotide is involved in first-strand reverse transcription (3). The analysis of right-end mutations should tell whether in FVs this end serves a similar role. The function of the U3-end dinucleotide of orthoretroviral cDNA is less well defined, and respective studies utilizing mutation analyses do not, to our knowledge, exist.

Our data suggest that spumaretrovirus integration requires cleavage at the right U5 end, while the viral IN does not modify the left U3 end. This is reminiscent of the situation in the yeast retrotransposon TY1, where neither end is cleaved at all (27). Although in all FVs isolated from different species the U3 end is strictly conserved (12), PFV tolerated alterations without significantly affecting the replication competence. DNA end flexibility was previously noted in the MLV system too (8). Several of our mutants exhibited a more severe impact on the replication competence. These differences, however, appeared to depend on the stability of the altered U3 ends and were, furthermore, also shown to be unrelated to its cleavability. When introduced into a vector, the uncleavable pTJ-11 transduced recipient cells almost as efficiently as the cleavable pTJ-10 mutant, and no difference was noted in the transient-replication competence between both mutants. This indicates that (i) an elongated U3 end is trimmed to wild-type length provided that cleavage is permitted and (ii) although phylogenetically conserved among retroviruses, the proviral TG start dinucleotide is not essential for FV integration.

An attractive explanation for the asymmetric FV DNA cleavage consists of the necessity of the U5 dinucleotide for first-strand reverse transcription. Subsequently, this dinucleotide has to be cleaved off to allow proper integration. In contrast, the FV U3 end may be as such a good substrate for integration which does not require further trimming.