INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

The packaging of genomic RNA into retroviral particles is a complex and highly specific process involving interactions between viral structural proteins and cis-acting sequences in the viral RNA, called psi () or encapsidation signals (29). The packaging sequences for several retroviruses have been mapped. For Moloney murine leukemia virus (30), spleen necrosis virus (46), bovine leukemia virus (20), and type D retroviruses (44), these essential cis sequences are in close proximity to the 5' end in the untranslated leader region of the genomic RNA, next to the splice donor site. In Moloney murine leukemia virus, psi can extend into the gag open reading frame (ORF) (5, 31). A more complex situation is found in retroviruses such as Rous sarcoma virus (RSV) and human immunodeficiency virus type 1 (HIV-1). For RSV, sequences upstream of the major splice donor site are involved in packaging, raising the question of how the spliced transcripts are distinguished from the genomic RNA in the encapsidation process (3, 21, 35). Direct-repeat sequences flanking the v-src gene enhance packaging by exerting an influence on several steps of the viral life cycle (41, 42). For HIV-1, sequences involved in packaging are located between the major splice donor site and the gag ORF (1, 12, 28). However, additional sequences in the env ORF enhance the efficiency of the process (7, 22, 37).

Secondary structures for psi sequences have been predicted, and a stem-loop structure common to a variety of retroviruses has been described (2, 17, 23, 47). Controversy exists as to whether dimerization plays an important role in RNA packaging (3, 6, 13, 26, 34); this idea is supported by the fact that the dimer linkage structure (DLS) of retroviruses, which mediates stable and noncovalent intermolecular linkage of the RNA monomers, coincides with the psi sequence.

Foamy viruses are a subfamily of retroviruses which may constitute good candidate vectors for gene delivery because of their broad tropism, their benign nature, and their greater packaging potential compared with that of other retroviruses (8, 18, 38, 40). However, the packaging sequence(s) has not been determined in detail. Recently, Russell and Miller (38) reported the construction of human foamy virus (HFV)-based vectors carrying genes for neomycin phosphotransferase and alkaline phosphatase. These constructs, with deletions in the gag, pol, and env ORFs, could be successfully transferred to recipient cells when cotransfected with wild-type helper HFV DNA, implying that the deleted sequences were not involved in transfer.

To investigate which sequences are necessary for gene delivery, a series of HFV-based vectors with deletions in the viral genome were constructed and stably transfected cell lines were generated. After transfection with a plasmid carrying helper HFV, it was found that sequences in the pol gene were needed, in addition to the leader region, to allow efficient transfer by helper HFV.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Recombinant DNA. Previously described laboratory procedures were followed for the manipulation of DNA, plasmid preparation, and cloning (39). The constructs used in the transfer experiments are depicted in Fig. 1 and were subjected to restriction enzyme digestion for verification of their construction. All constructs were derived from HSRV1 (36), an infectious molecular clone of HFV with a deletion in the U3 region. The positions cited refer to the proviral DNA. The start site of transcription (+1) of viral RNA coincides with position 778 of HSRV1.

View larger version (22K): [in this window] [in a new window]   FIG. 1.   Constructs used in the packaging experiments with respect to the pFOV-7 helper virus. Numbers indicate the proviral DNA of HSRV1. The U3 region is 777 bp long; thus, the start site of transcription (+1) of viral RNA coincides with position 778. IRES-hygror indicates the gene encoding hygromycin resistance under the control of the poliovirus IRES. SV40-gusA denotes the gusA gene under the control of the SV40 early promoter. Solid lines indicate HSRV1 sequences present in the vector constructs. Transfer was assayed by GUS (vector) and -Gal (helper HFV) expression per milliliter. Values shown are the means of at least three independent experiments. The ratios of numbers of GUS-positive colonies to numbers of -Gal-positive colonies are given in percentages. The asterisk in plasmid pIH-1, BamHI denotes the site of mutation.

Cells and DNA transfections. Stably transfected vector cell lines were generated with Lipofectin (GIBCO BRL) from BHK/Bel-1 cells, which constitutively express the viral transactivator Bel-1 (8). They were selected in Dulbecco's modified Eagle's medium containing 5% fetal calf serum, 0.5 mg of G-418 (GIBCO BRL) per ml, and 400 µg of hygromycin (Sigma) per ml. The selection medium was changed daily until colonies of hygromycin-resistant cells were observed. These were subsequently pooled, and stable transfection was verified by staining for GUS expression.

View larger version (37K): [in this window] [in a new window]   FIG. 2.   Schematic presentation of the protocol followed to test for transfer. DMEM, Dulbecco's modified Eagle's medium; FCS, fetal calf serum.

Transfer of the construct by helper HFV. On day 7 posttransfection with plasmid pFOV-7, supernatant fluid collected from the stably transfected vector cell lines was assayed for the presence of the helper virus and the construct (Fig. 2). The supernatant was filtered through a 0.45-µm-pore-size membrane (Acrodisc), and 1, 0.1, or 0.01 ml in duplicate was plated onto BHLL cells. One replicate was assayed for the presence of helper HFV by counting -galactosidase (-Gal)-positive foci, and the other was assayed for GUS expression to determine the transfer of the construct.

Staining for GUS and -Gal activities. Staining for GUS and -Gal expression has been described elsewhere (8, 9).

RNase protection assay. To generate the antisense probe, sequences from positions 1118 to 1655 of proviral DNA were amplified by PCR with primer pair P20-P22 and cloned as a MunI/HindIII fragment into plasmid pSPT18 (Boehringer Mannheim Biochemicals) digested with EcoRI/HindIII. The plasmid was digested with SspI at position 1392 of HSRV1 and transcribed with SP6 polymerase (Boehringer-Mannheim) in the presence of ³²P. This procedure resulted in a radiolabelled riboprobe 272 nucleotides (nt) long. Of those, 263 nt were complementary to full-length helper HFV from positions 1393 to 1655 or 159 nt (from positions 1393 to 1551) for the transcripts with internal deletions. Total cellular RNA was extracted from stably transfected cells with the RNeasy kit (Qiagen), and 10 µg was ethanol precipitated along with 4 × 10⁵ cpm of gel-purified riboprobe. RNase protection assays were carried out with an RPA II kit (Ambion). Plasmid pBR322 digested with MspI provided a size marker for electrophoresis. A linearized plasmid for the human -actin gene (Ambion) generated a radiolabelled transcript of 218 nt, 127 nt of which were complementary to cellular RNA to provide an internal control.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

A series of HFV-based vectors was constructed; each construct carried the IRES-hygro^r cassette and the SV40-gusA cassette (Fig. 1). To investigate the transfer of the constructs by replication-competent helper HFV, the vectors were stably transfected into BHK/Bel-1 cells (8). The derived vector cell lines were then transfected with plasmid pFOV-7 carrying replication-competent helper HFV (38). The supernatant was plated on recipient BHLL cells (9) 1 week after transfection (Fig. 2), when the HFV-induced cytopathic effect of the vector producing-cell lines was maximal. Since BHLL cells carry the gene for -Gal under the control of the HFV LTR, helper virus infection was monitored by the addition of X-Gal, while vector transfer was assayed by incubation with X-GlcA.

Delineation of sequences in the leader and gag ORF required for transfer of HFV vectors. Plasmid pIH-1 generated a vector with sequences from positions 1 to 6927 of proviral DNA. When BHK/Bel-1 cells stably expressing this construct were transfected with helper virus DNA and the supernatant was plated on BHLL cells, vector-transduced, GUS-positive foci approximated 10% of -Gal-positive foci resulting from infection by helper HFV (Fig. 1). This result demonstrated that the cis-acting sequences essential for transfer were present. The criterion for transfer adopted was that the construct titer should constitute at least 1% of the helper virus titer. Constructs with a titer of less than 1% that of helper HFV were described as being poorly transferred. To investigate the contribution of the 5' end of the genome, mutants pIH-2, pIH-3, and pIH-4, with internal deletions of proviral sequences from positions 1551 to 2153, 1274 to 2153, and 1173 to 2153, respectively, were constructed. Transfer was observed for mutants with deletions in gag (pIH-2 and, to a lesser extent, pIH-3). However, no transfer was observed for construct pIH-4, in which part of the leader region was also deleted (Fig. 1), suggesting that sequences crucial for transfer lie upstream of position 1274.

View larger version (25K): [in this window] [in a new window]   FIG. 3.   Assay for recombination between construct pIH-2 or pIH-3 and helper HFV used in the transfer experiments. Vector cell lines for each construct were transfected with helper HFV DNA, and the supernatant was used to infect BHK/Bel-1 cells. This procedure allowed transcription from the LTR of the constructs and subsequent expression of the IRES-hygror cassette. Primer pair P20-P21 was used to amplify DNA from hygromycin-resistant colonies. Due to the sizes of their internal deletions, PCR fragments of 537 and 260 bp would be expected if plasmids pIH-2 and pIH-3 were transferred without recombination. Helper HFV would generate a fragment of 1,138 bp. Lanes: 1, PCR product of cells exposed to helper HFV and construct pIH-3; 2, PCR product of cells exposed to helper HFV and construct pIH-2 (in both cases, no fragment corresponding to helper HFV could be detected); M, molecular weight marker X174 cut with HaeIII. The sizes of the bands are, from top to bottom (in base pairs): 1,353, 1,078, 872, 603, 310, 281, 271, 234, 194, 118, and 72.

Additional sequences in the pol ORF are required for the transfer of HFV vectors. Plasmids pIH-5 and pIH-6 were constructed to include proviral sequences from positions 1 to 6372 and 1 to 6318, respectively. These constructs could be transferred (Fig. 1). Plasmid pIH-6 lacks a sequence, described as a second polypurine tract (PPT), which might play a role in reverse transcription (25, 43). However, this sequence was dispensable for the transfer of vector genomes. In contrast to these plasmids, plasmid pIH-7, with sequences up to position 5894, was found to be poorly transferred by helper HFV, constituting less than 1% of the values obtained for helper HFV (Fig. 1). We generated two other constructs, pIH-8 and pIH-9. Plasmid pIH-8 includes HSRV1 sequences from positions 1 to 2817. Plasmid pIH-9 includes proviral sequences from positions 1 to 7786 but has an internal deletion from positions 5893 to 6766. Both constructs behaved in a manner similar to that of plasmid pIH-7, indicating that they also lacked important sequences.

Construction of an HFV vector containing minimal cis-acting sequences. To examine the possibility that further sequences in the gag or pol ORF were necessary for transfer by HFV, an internal deletion in the proviral sequences from positions 1551 to 6011 of plasmid pIH-6 was constructed. The resulting construct, pIH-10, was poorly transferred, whereas plasmid pIH-11, containing HSRV1 sequences up to position 6317 and having an internal deletion from positions 1551 to 5638 of the proviral DNA, was efficiently transferred. Another construct, pIH-12, which resulted from insertion of proviral sequences from positions 4678 to 5428 into construct pIH-10, was only poorly transferred (Fig. 1). These results demonstrate the importance of the pol sequences after position 5638 of HSRV1, which cannot be substituted by other upstream sequences. In the viral RNA, this position corresponds to position 4861. These results are in keeping with the findings of Russell and Miller (38) and, furthermore, define for the first time the minimal cis-acting sequences required for vector transfer.

Stability of vector RNA. We addressed the question of whether the vector cell lines produced the same amounts of RNA from transferable and nontransferable constructs to exclude the possibility that the lack of transfer resulted from an instability in the vector cell lines of RNAs lacking the pol sequences. To investigate this question, an RNase protection assay was performed. The riboprobe was generated from positions 1393 to 1655 of the proviral sequences (Fig. 4A), representing positions 616 to 878 of HFV RNA, and hybridized only to unspliced RNA, since the major splice donor site of HFV lies at position 51 (33). Cellular RNA was extracted from a set of vector cell lines generated from three transferable and three poorly transferable constructs. These two groups of constructs included examples with or without an internal deletion (Fig. 4B).

View larger version (31K): [in this window] [in a new window]   FIG. 4.   RNase protection assay. (A) Localization of the riboprobe in the gag ORF. Numbers indicate the proviral DNA of HSRV1. The expected sizes of the protected RNA fragments for constructs with and without internal deletions protected by the riboprobe are indicated. nt, nucleotides. (B) RNA from stably transfected vector cell lines was compared with RNA from BHK/Bel-1 cells and RNA from BHK/Bel-1 cells producing helper HFV (FOV-7). Vectors pIH-10, pIH-2, and pIH-12 yielded a protected fragment of 159 nt (C*). Vectors pIH-1, pIH-7, and pIH-6 yielded a protected fragment of 263 nt (C). A fragment (127 nt) of human -actin RNA (hu) served as an internal control. M, molecular weight marker pBR322 cut with MspI. The sizes of the DNA fragments are indicated in nucleotides.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The successful use of vectors based on replication-competent and replication-incompetent HFV to deliver heterologous genes has been reported (8, 18, 38, 40). However, these vectors still encode large parts of viral sequences, limiting the space available for the incorporation of foreign DNA. To exploit further foamy viruses as vectors for gene delivery, we investigated what sequences in the viral genome were necessary for the construction of a minimal HFV-based vector. Because the titer of viruses released from foamy virus-infected cells is low, we used an indirect approach to monitor whether helper HFV could transfer constructs carrying reporter genes. The experiments described here, therefore, cannot strictly distinguish among the impairment of nucleoplasmic transport, the packaging process, and subsequent steps of the replication cycle after the infection of recipient cells (e.g., reverse transcription and integration).

Constructs carrying the genes hygro^r and gusA were introduced into the viral genome, and stably expressing vector cell lines were generated. After transfection of these vector cell lines with helper HFV, the supernatant was assayed for the packaged construct by infection of recipient cells and staining for the reporter gene. From these experiments it was evident that functional HFV vectors must contain two distinct regions of the viral genome. The first of these two regions, located at the 5' end of the proviral genome upstream of position 1274 (RNA, 497), by analogy to other retroviruses probably constitutes a common DLS-encapsidation packaging signal (6, 10, 15, 20). Three sites (SI, SII, and SIII) in the leader region are important for full dimerization of HFV RNA in vitro (16). A deletion mutant (pIH-4) lacking SII and SIII was not a functional vector. However, deletion mutants pIH-2 and pIH-3, both of which include the complete DLS, could be transferred (Fig. 1), as proved by sequence analysis after transfer to recipient cells. The observation that plasmid pIH-1, BamHI, carrying a mutation in the SII palindrome which impaires dimerization in vitro, also inhibited transfer by helper HFV indicates that these two functional sites are closely linked. The splice donor site of HFV is located at position 51 of viral RNA (33). Therefore, the putative packaging sequence in the leader region of viral RNA is present only on unspliced genomic RNA.

The second region needed for efficient transfer by helper HFV lies in the pol ORF between positions 5638 and 6317 of proviral DNA (4861 and 5540, respectively, of viral RNA). Deletion mutants which lacked these sequences could not be transferred, whereas constructs which included these sequences constituted functional vectors (Fig. 1). It has been reported that the pol ORF of foamy viruses possesses a second central PPT, a feature shared only by the lentiviruses (reviewed in reference 24), and a dual mode of initiation of plus-strand DNA synthesis has been suggested (25, 43). This is the case for HIV-1, for which a PPT duplication has been described (45), and mutation of this central PPT reduces the infectivity of the virus (19). In HSRV1, the central PPT is located at positions 6340 to 6351. However, since pIH-6 and pIH-11 could be efficiently transferred in the absence of this sequence, its influence in the context of replication-defective vectors is, at most, minor.

Additional env sequences which are required to produce vectors based on HIV-1 have been described, although subsequent investigations could not define a specific sequence (7, 22, 37). For RSV, direct-repeat sequences flanking the v-src gene are also essential cis-acting sequences (41, 42). The situation with HFV seems to be similar in that two sequences from different parts of the genome must be present simultanously to enable efficient transfer by helper HFV. The mechanism by which the region in pol exerts its influence is unclear, but the RNase protection assay carried out showed that a potential instability or a low rate of synthesis of RNA lacking this region was not the explanation. Like other complex retroviruses (14), HFV has a transcriptional transactivator. In order to allow cytoplasmic expression of unspliced and singly spliced mRNAs encoding the structural proteins, some complex retroviruses, such as HIV-1, require a posttranscriptional transactivator (Rev) and the presence of a specific recognition element (Rev-responsive element) within these mRNAs. No Rev- or Rev-responsive element-like functions are required for HFV gene expression (27). Moreover, it was shown recently that the Pol protein of HFV is not required for packaging (4). These findings beg the question of how and where the unspliced genomic RNA of HFV is recognized and packaged. A possible function of the sequences in the pol ORF is that they work in a manner similar to that of a constitutive transport element, as described for Mason-Pfizer monkey virus (11), allowing unspliced RNA to effectively reach the cytoplasm. So far, we have not investigated whether the sequences in the pol ORF can act as a constitutive transport element or whether they are responsible for other functions.

Recently, it became clear that the life cycle of foamy viruses is different from that of other retroviruses and that it follows a pathway not unlike that of hepadnaviruses (48). The detailed contributions of the pol sequences to the replication process in general and to packaging in particular remain to be determined. However, with the identification of the minimal cis-acting sequences described here and the construction of a useful vector (pIH-11) with only about 3.6 kb of proviral sequences remaining for the delivery of heterologous genes, it will be possible to fully exploit the potential packaging capacity of HFV as a vector.