HIV-1 protein expression from synthetic circles of DNA mimicking the extrachromosomal forms of viral DNA.

We have constructed circular forms of human immunodeficiency virus type 1 viral DNA in vitro that closely resemble the single and double long terminal repeat circular forms of unintegrated viral DNA formed in the nuclei of infected cells. We have analyzed viral protein expression after transient transfection of these circular DNAs into HeLa cells and compared it with expression from a transfected linearized plasmid containing an integrated provirus. Both circular forms are expressed, as judged by the appearance of extracellular p24, and expression is trans-activated by human immunodeficiency virus type 1 Tat. Viral p24 production, however, is approximately an order of magnitude lower than that obtained with transfected integrated viral DNA. Similar data were obtained when a luciferase reporter gene was substituted for the coding regions of the viral DNA. Positional effects of the transcriptional initiation and termination signals in the long terminal repeat appear to account for some of the low expression levels. These data suggest that unintegrated circular viral DNAs are transcriptionally active, although at low levels, and may contribute to overall viral replication in infected people under some conditions.

We have constructed circular forms of human immunodeficiency virus type 1 viral DNA in vitro that closely resemble the single and double long terminal repeat circular forms of unintegrated viral DNA formed in the nuclei of infected cells. We have analyzed viral protein expression after transient transfection of these circular DNAs into HeLa cells and compared it with expression from a transfected linearized plasmid containing an integrated provirus. Both circular forms are expressed, as judged by the appearance of extracellular p24, and expression is trans-activated by human immunodeficiency virus type 1 Tat. Viral p24 production, however, is approximately an order of magnitude lower than that obtained with transfected integrated viral DNA. Similar data were obtained when a luciferase reporter gene was substituted for the coding regions of the viral DNA. Positional effects of the transcriptional initiation and termination signals in the long terminal repeat appear to account for some of the low expression levels. These data suggest that unintegrated circular viral DNAs are transcriptionally active, although at low levels, and may contribute to overall viral replication in infected people under some conditions. Unintegrated circular extrachromosomal forms of human immunodeficiency virus type 1 (HIV-1) 1 DNA are constantly produced during infection of cell culture and in vivo (1)(2)(3). They represent terminal products of the reverse transcription process in which newly synthesized double-stranded linear viral DNA is integrated into the host genomic DNA or is circularized to form extrachromosomal HIV-1 DNA. The circular forms contain either a single copy or a tandem double copy of the long terminal repeat (LTR) region, which contains the viral transcriptional initiation and termination control elements (4,5).
Despite their nuclear localization (6) and their presence at the site of HIV-1 replication in cell culture (1,2) and in vivo (3), it is not clear whether the circular forms are able to sustain viral replication. By using HIV-1 integrase defective viruses and different cell types, we and others have recently suggested that unintegrated DNA is transcriptionally active (7-9), although at low levels and for a short period of time. Because of the high amounts of unintegrated HIV-1 DNA at sites of viral replication in infected people, it is important to understand whether these forms could contribute to the overall amount of virus produced during the course of the disease.
In an attempt to answer this question, we measured the amount of viral p24 protein produced from synthetic circles of HIV-1 DNA constructed to closely mimic the extrachromosomal forms of HIV-1. We show that these forms indeed support the production of viral proteins with varying efficiencies, depending in part upon the relative position of the LTR, and can express infectious virus.

In Vitro Construction of Extrachromosomal Forms of HIV-1 DNA-
The HIV-1 molecular clone pLW/C has been described previously (7,10). Briefly, pLW/C contains the complete proviral genome of HIV-1 flanked by host genomic DNA cloned in a derivative of the pSP65 vector (Promega, Madison, WI). All reading frames are open, and transfection of pLW/C into target cells produces infectious virus that is competent for the infection of peripheral blood lymphocytes and macrophages (7).
For the construction of the 1LTR circle (c1LTR) of the LW/C virus (depicted in Fig. 1a), 30 cycles of polymerase chain reaction (PCR) were performed on DNA extracted from LW/C-infected peripheral blood lymphocytes three days after infection, using the primer pair NE-AS/EE-S and Vent DNA polymerase (New England Biolabs, Beverly, MA) (to avoid base misincorporation), following the manufacturer's instructions. This PCR amplifies the entire LTR and some flanking viral sequences from the one LTR circular viral DNA formed during infection, and provides flanking EcoRI sites for subsequent cloning operations. After digestion of this fragment with EcoRI and agarose gel purification of the 1LTR-amplified band, it was subcloned into the EcoRI site of the PCR II plasmid (Invitrogen, San Diego, CA) to produce the TA1LTR plasmid. A portion of the insert containing a single LTR and some flanking viral sequences were then excised with NarI and EspI (which are unique sites into the viral genome) and ligated to the large viral DNA fragment previously excised from pLW/C using the same restriction enzymes, thus reconstituting the complete genome of the LW/C virus in the form of c1LTR, which is identical to the natural 1LTR circular DNA of LW/C present during the course of the infection.
For the construction of the 2LTR circle (c2LTR) of LW/C virus (depicted in Fig. 1b), pLW/C was amplified using the primer pairs C3-S/ NE-AS and C5-AS/EE-S in two separate PCR reactions (15 cycles) with Vent DNA polymerase (New England Biolabs). PCR products were digested with ClaI, purified on column with the Magic PCR preps (Promega), ligated together at the ClaI site, and digested with EcoRI. The appropriate band containing two tandem LTR (2LTR) was purified on an agarose gel and cloned into the EcoRI site of plasmid PCR II (Invitrogen) to produce the TA2LTR plasmid. The 2LTR insert was then excised from the TA2LTR plasmid with NarI and EspI and ligated to the large viral DNA fragment previously excised from pLW/C using the same restriction enzymes, thus reconstituting the complete genome of the LW/C virus in the form of c2LTR. The final product (c2LTR) differs from the natural 2LTR circular DNA of LW/C only by the presence of the ClaI site at the junction of the two LTRs.
To purify the constructed c1LTR and c2LTR, each final ligation product was loaded on an agarose gel. After electrophoresis, the gel was ethidium bromide-stained (circular DNA runs faster than linear DNA on an unstained gel) and each visible DNA band below 7.0 kilobase present in the gel was purified and tested for the presence of c1LTR and c2LTR by digestion with NarI or EspI, which produce a singly cut DNA of 9.0 and 9.65 kilobases only with the c1LTR and c2LTR, respectively (data not shown). The sequences of the primers used in the PCR were as follows: C3-S, 5Ј-CGCTAGCAGGTTATCGATACTGGAAGGGCTAATTCAC-3Ј (the bases corresponding to nucleotides 16 -35 in the LW/C viral sequence are underlined; the ClaI site is bold; sense orientation); NE-AS, 5Ј-GCTAGCACGAATTCGCTTTCAGGTCCCTGTTCGG-3Ј (the bases corresponding to nucleotides 682-660 in the LW/C viral sequence are underlined; the EcoRI site is bold; antisense orientation); EE-S, 5Ј-GCTAGCTAGAATTCCGAGCTGAGCCAGCAGCAGATG-3Ј (the bases corresponding to nucleotides 8738 -8760 in the LW/C viral sequence are underlined; the EcoRI site is bold; sense orientation); C5-AS, 5Ј-GATC-GAGAACCATCGATCTAGAGATTTTCCACACTGAC-3Ј (the bases corresponding to nucleotides 9596 -9576 in the viral sequence are underlined; the ClaI site is bold; antisense orientation).
Construction of the LTR-luc Plasmids-The luciferase reporter gene vectors (LTR-luc plasmids) (depicted in Fig. 2) were constructed by cloning the single and double LTRs into the pGEM-luc vector (Promega), which contains the luciferase gene but lacks a promoter and polyadenylation sequence, and into the pGL2-Basic vector (Promega), which lacks a promoter but contains the SV40 polyadenylation signal at the end of the luciferase coding sequence. Single or double LTRs of LW/C were excised from the TA1LTR and TA2LTR plasmids described above and inserted into the luciferase cassette. Briefly, a NsiI/BamHIdigested fragment containing the single LTR was isolated from the TA1LTR plasmid described above and subcloned, using the same restriction sites, into the pGEM-luc vector, producing the 1LTR-luc-Circle. A NsiI/NotI-digested fragment containing the double LTR was isolated from the TA2LTR plasmid described above and subcloned, using the same restriction sites, into the pGEM-luc vector, producing the 2LTR-luc-Circle. For the introduction of the second LTR at the 3Ј end of the luciferase gene, an EcoRI fragment containing the LTR was isolated from TA1LTR, filled in with the Klenow enzyme, and cloned into the Klenow enzyme-filled XhoI restriction site of 1LTR-luc-Circle and 2LTR-luc-Circle, producing the 1LTR-luc-1LTR-Circle and the 2LTR-luc-1LTR-Circle plasmids. To obtain the 1LTR-luc-1LTR-Linear DNA, the 1LTR-luc-1LTR-Circle plasmid was digested with NsiI/SfiI. For the construction of 1LTR-luc-pA a BamHI/XbaI-digested fragment containing the LTR was isolated from TA1LTR and cloned between the BglII/NheI sites of the pGL2-Basic vector. For the construction of 2LTRluc-pA, a KpnI/XhoI-digested fragment containing the 2LTR was isolated from TA2LTR and cloned between the KpnI and XhoI sites of the pGL2-Basic vector.

One and Two LTR Circular DNAs Are Templates for the
Production of p24 -To characterize the relative transcriptional potentials of extrachromosomal forms of HIV-1, we synthesized single LTR circles (c1LTR) and double LTR circles (c2LTR) of DNA from the macrophage tropic strain LW/C (7, 10) using the strategy depicted in Fig. 1 (a and b) and described under "Materials and Methods." After purification by agarose gel electrophoresis, the c1LTR, c2LTR, supercoiled pLW/C, and pLW/C linearized with PvuI (which cuts inside the ampicillin resistance gene in the pLW/C plasmid) were then separately transfected into HeLa cells, either alone or along with an expression plasmid carrying the tat gene of HIV-1. The appearance of p24 in the supernatant of the transfected cells indicated that c1LTR and c2LTR, supercoiled pLW/C and linearized pLW/C were all able to sustain viral protein expression (Table  I). Moreover, the recovery of p24 in the supernatant was significantly higher in the presence of Tat, which transcriptionally transactivates the HIV-1 LTR (11), indicating that Tat is functional with the extrachromosomal forms of HIV-1 reconstructed in vitro. The amount of p24 was proportional to the amount of transfected c1LTR DNA and c2LTR DNA (Table I). Supernatants derived from transfections in which 1 g of synthetic c1LTR or c2LTR was used contained competent viral particles, as shown by their ability to infect peripheral blood lymphocytes from a normal donor (data not shown), indicating that both transfected c1LTR and c2LTR were able to serve as a source of viral production.
Recovery of p24 in the supernatant was 2-4-fold higher after transfection with the c1LTR than with the c2LTR construct in either the presence or the absence of the Tat expression plas-mid. In addition, p24 production was approximately an order of magnitude higher after transfection with either linearized or circular pLW/C (where the two LTRs are separated by the viral genome and flanked by host genomic sequences) than with c1LTR or c2LTR (where the LTRs are flanked by viral sequences and, in the case of c2LTR, joined end-to-end) ( Table I).
Recently it has been shown that when Tat driven transcription initiates from the LTR, the HIV-1 poly(A) site present in the same LTR is occluded (12,13). This occlusion could be operative on the LTR in the c1LTR construct and on both LTRs in the c2LTR construct, since the presumptive initiation and termination signals are much closer to each other than in the integrated provirus or in the transfected pLW/C. The higher transcriptional activity of c1LTR compared to c2LTR could be due to interference between the two promoters in the 2LTR circular forms because of their proximity and similar orientation (14,15), as indicated in Fig. 1c.
To ascertain the efficiency of transfection, we performed Southern blot experiments on DNA extracted from the nuclei of the transfected cells, while the cytoplasmic fractions were used for p24 analysis. The results indicated that c1LTR, c2LTR and linearized pLW/C were transfected with comparable efficiency (data not shown). Circular pLW/C, however, was present at higher levels (3 times) in the transfected cells (data not shown). This partially explains the higher amount of p24 recovered in the supernatant of the circular compared to the linearized pLW/C-transfected cells.
The greater expression of p24 with c1LTR DNA compared to c2LTR DNA argues against the possibility that the p24 in the supernatant of the transfected cells is expressed from unpermuted integrations of the transfected DNA. If that were the case, one would expect a higher yield from the c2LTR DNA, than from the c1LTR DNA, since the c2LTR DNA would have optimal spacing between the upstream promoter and the downstream polyadenylation signal. In contrast, integration of the c1LTR DNA would give a viral transcription unit lacking a functional promoter or polyadenylation signal except in those cases where, by chance, integration occurred in the proper position to utilize a host cell promoter or polyadenylation signal.
Luciferase Activity from Reporter Gene Expression Vectors for Analysis of c1LTR and c2LTR-To further evaluate the transcriptional ability of pLW/C, c1LTR and c2LTR, we constructed luciferase reporter gene expression vectors (LTR-luc plasmids), depicted in Fig. 2a and described under "Materials and Methods," in which the position of the LTR(s) with respect to the luciferase coding sequence correspond to that of pLW/C (1LTRluc-1LTR-Circle), c1LTR (1LTR-luc-Circle) and c2LTR (2LTRluc-Circle), with respect to the LW/C coding sequence. Seventytwo hours after transfection of the plasmids into HeLa cells the intracellular levels of luciferase activity were measured. The results (Table II) were in good general agreement with the results obtained using p24 expression as a measure of transcriptional activity. The only exception was with linearized DNA containing LTRs at both ends (1LTR-luc-LTR linear), which, unlike linearized pLW/C, was much less active than its circular counterparts. One difference between the two linear DNAs was that the 1LTR-luc-LTR linear lacked any DNA sequences flanking the two LTRs. The host cell flanking regions adjacent to the LTR in pLW/C-linearized DNA may strongly influence LTR activity, possibly by providing a relatively nonspecific anchor for transcriptional factors. In all cases, cotransfection of the same set of plasmids with a tat expression plasmid increased the level of luciferase activity.
The similarity of the patterns of transcriptional activity of the LTR-luc plasmids and the p24 expression plasmids sug- FIG. 1. a and b, synthetic representation of the construction of 1LTR circle (c1LTR) (a) and 2LTR circle (c2LTR) (b) of LW/C virus as described under "Materials and Methods." c, diagram of the relative position of the viral genes along with the tataa box and polyadenylation sequence of the LTR of HIV-1 in c1LTR, c2LTR, and pLW/C. gests that the differences observed among the circular and the linear forms are due to the positions of the LTRs themselves and are not affected by the presence or absence of viral regulatory sequences between the LTRs. To better understand the influence of the relative LTR positions on transcriptional activity, we prepared further LTR-luc plasmids, shown in Fig. 2b and described under "Materials and Methods," in which the downstream LTR, which normally acts as a termination site for transcription, was replaced by a heterologous SV40 termination signal.
One construct contained an LTR upstream of the reporter gene and a SV40 termination signal downstream (1LTR-luc-pA). The activity in the absence of Tat was increased about 3-fold by replacing the downstream LTR, and was increased about 6-fold when Tat was supplied. This suggests that the downstream LTR has a moderate negative effect on the activity of the upstream LTR, as has been demonstrated previously in similar systems (16,17). A second set of constructs contained two tandem LTRs upstream from the reporter gene and either a LTR (2LTR-luc-1LTR) or a SV40 (2LTR-luc-pA) termination signal downstream. The presence of a second LTR in tandem with the upstream LTR in general only slightly reduced activity; the one exception was a 4-fold reduction in the activity of 2LTR-luc-pA compared to 1LTR-luc-pA in the presence of Tat. Replacement of the downstream LTR in the 2LTR-luc-1LTR construct with the SV40 termination sequence again resulted in an increase in activity, although the increase was only about 2-fold.
In order to rule out the possibility that the transfected LTRluc plasmids exhibit different transcriptional activity because the LTR(s) are competing for a limited pool of transcription factors, we cotransfected, with the LTR-luc plasmids, the TA2LTR plasmid (described under "Materials and Methods"), which lacks the luciferase coding sequence as carrier DNA to normalize the amount of LTR DNA. There was no difference in luciferase activity relative to the LTR-luc plasmids analyzed, whether or not the TA2LTR was present in the transfected DNA (data not shown), indicating that competition for a limited pool of transcription factors was not influencing the results. DISCUSSION During the last few years, several groups have shown that the unintegrated circular forms of HIV-1 are present specifically in the nucleus of infected cells and are markers of active transport to the nucleus of the preintegration complex (18 -21). Furthermore, unintegrated HIV-1 DNA is abundant at the sites of viral replication in vivo (3,22). The possibility that unintegrated forms of HIV-1 DNA are able to sustain viral production is thus of great importance, because this DNA could serve as a reservoir for virus production and thereby contribute to the progression to AIDS. To address this question under well defined conditions, we studied the ability of circular forms of  a HeLa cells were transfected by the calcium phosphate method (Profection mammalian transfection system; Promega) with either 100 ng or 1 g of DNA in six-well plates at 80% confluence. Where indicated, a 2:1 molar ratio of pRBK-Tat plasmid was co-transfected with the indicated DNA. Seventy-two hours after transfection, supernatants were analyzed for p24 using an antigen capture enzyme-linked immunosorbent assay kit (Coulter Corp., Miami, FL).
b The Tat expression plasmid pRBK-Tat is an episomal mammalian expression vector that carries the tat gene of HIV-1 under the control of the Rous sarcoma virus LTR. The construct was derived by insertion of the tat gene into the BamHI site of the pRBK plasmid (Invitrogen). c ND, not done. d DNA was transfected either after linearization with PvuI (a unique site in pLW/C, present in the ampicillin resistance gene) or in its supercoiled form. The values in parenthesis represent the amount of p24 recovered after transfection of the supercoiled DNA.
HIV-1 DNA synthesized in vitro to be expressed and compared their relative transcriptional activities using both viral p24 and luciferase activity as reporter genes. We demonstrated that extrachromosomal forms of HIV-1 can direct expression of viral proteins and infectious virus in cell culture. The p24 expression was surprisingly high, sometimes exceeding 20% of the activity seen with transfected linear infectious viral DNA containing host flanking regions. A second set of constructs, using luciferase as a reporter gene, confirmed the results obtained from p24 expression.
In comparison with linearized pLW/C, the circular forms of HIV-1 possess a lower transcriptional activity, probably because of the positional effect of the LTR(s) present in the construct. The 1LTR circular form of HIV-1 DNA was more active than its 2LTR counterpart. This difference appears to be due to interference between the two LTRs rather than to any effect from HIV-1 coding sequences, as shown with the LTR-luc plasmids. Positional effects of the LTR may thus partially explain the low ability of extrachromosomal forms of HIV-1 DNA to direct viral protein synthesis. In fact, addition of the SV40 polyadenylation signal at the end of the coding sequence of the luciferase gene partially restored the activity of the 2LTR promoter, compared with the 1LTR promoter.
It is possible that in vivo unintegrated forms of DNA are present in a state which differs from that we analyzed, and that low levels of expression by HIV-1 extrachromosomal DNA in cell culture (7-9) may be further due to HIV-1 viral protein or cellular factors complexed to the unintegrated DNA. However, although total unintegrated viral DNA (linear and circular) present in nuclear extracts of infected cells has been shown to be associated with the matrix and integrase proteins of HIV-1 (23), unintegrated circular forms of HIV-1 are not associated with integrase, matrix, or capsid proteins (6). The apparent lack of association of viral proteins with unintegrated circular DNA suggests that these forms are structurally and functionally similar to the synthetic DNA circles described here. With respect to the apparent very low level of transcriptional activity reported for unintegrated HIV-1 DNA in cell culture (7)(8)(9), it is important to note that (a) these forms of DNA possess a halflife that is short (less than 1 day) when compared to that of integrated DNA (life span of the infected cell) (22), since they do not possess an origin of replication and, accordingly, are unable to self-replicate; (b) the integrated provirus, because it is in a replicating state, is likely to produce a high level of viral protein, obscuring low levels of expression from unintegrated DNA (24,25); (c) despite the presence of large amounts of unintegrated HIV-1 DNA at sites of intense viral replication (3,26,27), less unintegrated HIV-1 DNA is present during the course of the infection compared to the integrated proviral DNA (22,28,29), suggesting that the majority of the newly produced virus usually derives from the integrated provirus; and (d) promoter interference and/or occlusion could negatively activate transcription from unintegrated HIV-1 as shown by data with the LTR-luc plasmids. However, the evidence presented here showing that a variety of forms of unintegrated episomal viral DNA are all expressed in cell culture suggests that they have an intrinsic ability to support viral replication in vivo and could therefore actively contribute to the overall amount of produced virus. a HeLa cells were transfected as described in Table I with 1 g of reporter plasmid DNA. For the transfection of 1LTR-luc-1LTR-Linear, 0.5 g of DNA was used. Where indicated, a 2:1 molar ratio of pRBK-Tat plasmid was co-transfected with the reporter plasmid. Seventy-two hours after transfection cells were lysed in a solution containing 1% Triton X-100, 2 mM dithiothreitol, 25 mM Tris, pH 7.8, 2 mM CDTA, 10% glycerol and analyzed for luciferase activity using a Bertholdt luminometer.
b RLU, relative light units. Activities of each sample were normalized to 1 g of protein concentration by using the Bradford assay (Bio-Rad).