Threshold Number of Provirus Copies Required per Cell for Efficient Virus Production and Interference in Moloney Murine Leukemia Virus-Infected NIH 3T3 Cells

Abstract

The gag-pol readthrough mutant of Moloney murine leukemia virus, MLV-B(CAG) (T. Odawara, H. Yoshikura, M. Oshima, T. Tanaka, D. S. Jones, F. Nemoto, Y. Kuchino, and A. Iwamoto, J. Virol. 65:6376–6379, 1991), was poorly complemented by a mutant encoding only Gag. This is because with all the genetic elements necessary for env expression present in MLV-B(CAG), insufficient Env protein was produced by the cells expressing MLV-B(CAG) for efficient virus production. Since the env mRNA expression per provirus in the MLV-B(CAG)- and wild-type-MLV-producing cells were the same and since the cells expressing the former contained eightfold fewer proviral copies, the insufficient Env expression by the former was found to be due to insufficient proviral copies in the cells. Examination of the cell clones having various proviral copies of Δwt MLV (M. Oshima, T. Odawara, T. Matano, H. Sakahira, Y. Kuchino, A. Iwamoto, and H. Yoshikura, J. Virol. 70:2286–2295, 1996) showed that mRNA level was proportional to the number of proviral copies while interference and virus production followed a sigmoid curve with a sharp rise at the threshold number of proviral copies of around four per cell. Multicycle infection probably continues until the threshold level of proviral copies is attained in natural infection too.

For successful replication, murine leukemia virus (MLV) requires Gag, Pol, and Env, which are processed to final products by the virus-encoded protease during virion assembly and maturation. The Env protein is translated from the spliced viral mRNA, whereas the Gag and Pol proteins are synthesized from the unspliced viral mRNA. Although the unspliced viral mRNA of Moloney MLV (Mo-MLV) carries the amber stop codon, UAG, at the junction of the gag and the pol genes, translational termination is suppressed once every 10 times to produce the Gag-Pol fusion protein, which contains protease and polymerase activities (13, 32). When the UAG codon at the gag-pol junction was changed to a glutamine codon, CAG, the mutant virus, MLV-B(CAG), produced only the Gag-Pol fusion protein from the unspliced viral RNA and became replication incompetent. The mutant could replicate if it was supplied in trans with the Gag precursor, Pr65^gag, which served as the substrate for the protease present in the Gag-Pol fusion protein (8, 21).

In our above-mentioned study (21), the mutant used for supplying Pr65^gag encoded Env also. In this study, we examined whether the Env protein encoded by the complementing virus was dispensable. We also constructed complementing viruses encoding Pr65^gag of Mo-MLV and Env with different host ranges and examined their ability to restore the infectivity of MLV-B(CAG). The results indicated that the Env protein produced by the cells expressing MLV-B(CAG) was insufficient and that the host range of the virus produced by the complementation was determined mainly by the Env protein encoded by the complementing virus rather than by that of MLV-B(CAG). The amount of env mRNA expressed per provirus was the same for MLV-B(CAG) and the wild-type MLV. However, the cells expressing MLV-B(CAG) contained 1 or at most 2 proviral copies per cell while the wild-type-MLV-infected cells contained around 10 proviruses per cell. Examination of the cell clones having different proviral copies of Δwt MLV (24), which had a 306-base deletion in the rt region in pol, revealed that although the mRNA level increased linearly as the function of the proviral copies, the interference and virus production increased abruptly at the boundary of four proviral copies per cell. The results indicated that the presence of proviral copies above the threshold level was necessary for establishing interference and active virus production. In natural infection, too, multicycle infection probably continues until the interference is established and the cells begin to produce the virus at maximum efficiency. Although schematic representations of the retrovirus replication cycle usually depict integration of a single copy of provirus, the number of proviral DNA copies in the infected cells appears important for virus production and interference, and it may play a regulatory role by modulating the stoichiometric parameters for virus assembly.

MATERIALS AND METHODS

Plasmid constructs.

The plasmid constructs used in this study are shown in Fig. 1. They were all derived from Mo-MLV infectious clone pA^rMLV-48 (18), an integrated provirus which was flanked by mouse cell sequences at the integration site (1). All of the constructs retained 1.7- and 0.1-kb cellular flanking sequences on the 5′ and 3′ ends of the viral genome, respectively, and had a drug selection marker, either Neo^r or Hyg^r, driven by the simian virus 40 (SV40) promoter downstream of the 3′ long terminal repeat (LTR) (Fig. 1). Constructs pGE^6.4-hmB and pΔwt-hmB have been described previously (24). pMLV-B(CAG)-neo was constructed by ligating the Neo^r gene to the previously described pArMLV-B(CAG) (21). pGEBstE-hmB was constructed by digestion at the BstEII site (nucleotide [nt] 5923 according to the numbering scheme of Shinnick et al. [28]) of pGE^6.4-hmB, and the gap was filled with 5 bases (TAACC) to stop the premature translation of env. pGE-am-hmB was constructed by replacing the HindIII-ClaI (nt 4894 to 7674) portion of pGE^6.4-hmB with the corresponding portion of 4070A amphotropic MLV clone 8-1 (kindly provided by S. K. Chattopadhyay, National Institute of Allergy and Infectious Diseases, Bethesda, Md. [4]). Although the HindIII-ClaI fragment of 4070A clone 8-1 was about 300 bases shorter than the corresponding portion of pGE^6.4-hmB, MLV-GE-am could complement MLV-B(CAG) (see below). pGE-xe-hmB was constructed by replacing the NdeI-NheI (nt 5401 to 7846) portion of pGE^6.4-hmB with the corresponding portion of xenotropic MLV clone NZB9-1 (kindly provided by R. R. O’Neill, National Institute of Allergy and Infectious Diseases [23]). pΔwt-neo was constructed by replacing the Hyg^r gene of pΔwt-hmB with Neo^r.

FIG. 1.

Structures of viral constructs. Although shown only in the Δwt construct at the bottom of the figure, all of the MLV constructs had the same cellular flanking sequence derived from Mo-MLV integrated clone 48 (1), 1.7 kb and 0.1 kb on the 5′ and 3′ ends of the genome, respectively, and also had a neomycin resistance gene (Neo^r) or a hygromycin resistance gene (Hyg^r) driven by the SV40 promoter downstream. In the plasmid names, neo means that the construct had the Neo^r gene and hmB means that the construct had the Hyg^r gene. MLV-B(CAG) and MLV-GE^6.4 have been described previously (21, 24). MLV-GE^6.4 had a 2.4-kb deletion spanning from 30 bases upstream of the gag-pol junction (nt 2206) to near the end of rt in pol (nt 4600). As a result, Gag translation terminated 12 bases downstream from the deletion point (nt 4612 [arrow]), adding four unrelated amino acids instead of six C-terminal amino acids of p10. All of the constructs except the wild-type MLV, MLV-B(CAG), and Δwt MLV used this termination for the Pr65^gag synthesis. MLV-GEBstE had an insertion of 5 bases containing TAA stop codon at the BstEII site (nt 5923). In MLV-GE-am and MLV-GE-xe, the env portion of MLV-GE^6.4 was replaced by the corresponding region of the amphotropic virus clone 4070A (4) and that of the xenotropic virus clone NZB-9-1 (23), respectively (the replaced regions are indicated by thicker lines). Locations of the probes used in the hybridization analysis are shown by boxes numbered 1, 2, 3, and 4. Probe 1 (the ClaI-SacI fragment of clone 48, 3′-LTR probe) detects all of the mRNAs expressed from these constructs. Probe 2 (the BstEII-BamHI fragment of clone 48) was specific for ecotropic env. Probe 3 (the AflII-EcoRI fragment of clone 4070A) and probe 4 (the BglII-EcoRI fragment of clone NZB-9-1) were specific for amphotropic and xenotropic env, respectively. Nucleotide numbers are those used by Shinnick et al. (28). H, HindIII; C, ClaI; Nd, NdeI; Nh, NheI; RV, EcoRV; SD, splice donor; SA, splice acceptor. The region detected by RT-PCR is shown at the bottom.

Cell culture and transfection.

NIH 3T3 cells were grown in Dulbecco’s modified Eagle’s medium supplemented with 7% fetal calf serum. Subconfluent cultures of NIH 3T3 cells (2 × 10⁵ cells per 6-cm dish) were transfected with 20 μg of plasmid DNA by the standard calcium phosphate precipitation method (15). Selection with G418 (GIBCO BRL Life Technologies Inc., Grand Island, N.Y.) (300 μg/ml) or hygromycin B (Wako Life Science Inc., Osaka, Japan) (160 to 200 μg/ml) was started 48 h after transfection and continued for 3 weeks. For all the constructs, 50 to 100 colonies appeared in each dish.

The ecotropic virus was subjected to titer determination by the UV-XC assay (26). To determine the titers of the amphotropic and the xenotropic viruses in NIH 3T3 or MDTF cells, the UV-PG4 S⁺L⁻ assay devised by ourselves was performed; the procedure was the same as that of the UV-XC assay, except that feline PG4 S⁺L⁻ cell line (9) was used in place of the XC cell line and the foci (3) were counted instead of the plaques.

DNA analysis.

To analyze the extrachromosomal provirus, 2 × 10⁶ NIH 3T3 cells plated in 10-cm dishes were infected with the virus in the presence of 8 μg of Polybrene (Sigma Chemical Co., St. Louis, Mo.) per ml and the extrachromosomal DNA was extracted 18 h after infection as described previously (10). One-sixth of the extrachromosomal DNA recovered from each 10-cm dish was electrophoresed through a 0.9% agarose gel and blotted onto a Nitro-plus 2000 filter (Micron Separations Inc.).

The genomic DNA was prepared by treating the cells with 10 μg of proteinase K (Merck Laboratory, Darmstadt, Germany) per ml in 100 mM Tris (pH 8.0)–50 mM EDTA 24 h at 55°C followed by extraction with phenol-chloroform and ethanol precipitation. Then 20-μg portions of the DNAs were digested with EcoRV or HindIII, electrophoresed in a 0.8% agarose gel, and blotted onto a Nitro-plus 2000 filter. The BstEII-BamHI (nt 5923 to 6537 [probe 2 in Fig. 1]) fragment of Mo-MLV clone 48 (1), which was specific to the ecotropic env, was used to probe the Mo-MLV provirus.

RNA analysis.

RNAs of the transfected cells were isolated by the guanidinium thiocyanate-acid phenol method as described previously (6). The RNAs were electrophoresed through a formalin–1% agarose gel and blotted onto a Nitro-plus 2000 filter. The probes used to detect the viral mRNAs are shown in Fig. 1. For the quantitation of RNAs or DNAs, the radioactivity of each band on the hybridized filter was measured with a Bas 2000 bioimaging analyzer (Fuji Photo Co., Ltd., Tokyo, Japan).

Immunoblotting.

Proteins were extracted with RIPA buffer (150 mM NaCl, 20 mM Tris-HCl [pH 7.5], 0.5% sodium deoxycholate, 1% Triton X-100, 0.05% sodium dodecyl sulfate [SDS]) from the confluent cultures grown in 6-cm dishes. An aliquot (8 μg) of protein extract was electrophoresed through an SDS–10% polyacrylamide gel and blotted onto a Nitro-plus 2000 filter. Viral Env proteins were detected with a goat anti-gp70 polyclonal antibody (National Cancer Institute lot 79S000713). The antibodies bound to the filter were detected by the enhanced chemiluminescence Western immunoblotting detection system (Amersham International Plc., Little Chalfont, United Kingdom).

Immunoprecipitation.

Cells plated at 5 × 10⁵ cells per 6-cm dish were pulse-labeled with 60 μCi of l-[³⁵S]methionine in 2 ml of methionine-free minimal essential medium for 60 or 30 min at 37°C and chased in Dulbecco’s modified Eagle’s medium containing 7% fetal calf serum. The labeled cells were lysed with 800 μl of RIPA buffer, and 400 μl of each sample was subjected to immunoprecipitation with the anti-gp70 antibody as previously described (19). The immunoprecipitated proteins were electrophoresed through an SDS–10% polyacrylamide gel and exposed to Amersham Hyperfilm-MP X-ray film.

Estimation of virions released into the culture medium.

A total of 2 × 10⁵ cells were plated in each 6-cm dish. On the following day, the medium was changed to 3 ml of fresh medium. After 12 h, the culture fluid was collected and filtered through a 45-μm-pore-size Millipore MILLEX-HA filter. RNA was extracted from 250 μl of each sample. One-tenth of the extracted RNA was diluted threefold serially and then subjected to reverse transcription and nested PCR (RT-PCR) (27). The primers used for the detection of the viral env gene (Fig. 1) were Mo-1 (nt 5571 to 5590; 5′-CCGGTGGTACCTCACCCTTA-3′) and Mo-4 (nt 6035 to 6016; 5′-ATGGTCCATGGTGGGCTAAC-3′) as the external primers and Mo-2 (nt 5761 to 5780; 5′-CATCCTCTAGACTGACATGG-3′) and Mo-3 (nt 5934 to 5915; 5′-CCATTGGTTACCTCCCAGGT-3′) as the internal primers. The RT-PCR products were electrophoresed in a 2% agarose gel and detected by ethidium bromide staining.

RESULTS

A virus encoding only the Gag protein could restore the infectivity of MLV-B(CAG) virus, but far less efficiently than a virus encoding both the Gag and Env proteins.

We have previously shown that a mutant of Mo-MLV, MLV-B(CAG), which encoded the Gag-Pol fusion protein but not the Gag precursor protein, was replication defective (21). We have also shown that MLV-B(CAG) could be complemented by another defective virus, MLV-GE^6.4, which encoded both Gag precursor and Env proteins (21). It is reasonable that the Gag protein encoded by MLV-GE^6.4 was necessary for the complementation. However, it was unknown whether the Env protein encoded by MLV-GE^6.4 had any functional significance in the complementation, since MLV-B(CAG) encoded the Env protein by itself. To examine the possible role of the Env protein encoded by MLV-GE^6.4, its env gene was inactivated by inserting 5 nt, TAACC, at the BstEII site (nt 5923) (Fig. 1). The new virus, MLV-GEBstE, was unable to produce the Env protein due to premature termination of translation at the UAA codon within the inserted nucleotides. The DNA constructs of MLV-GE^6.4 and MLV-GEBstE were independently transfected to B2 cells which carried MLV-B(CAG) provirus. The MLV-GE^6.4- and MLV-GEBstE-transfected cells produced comparable levels of the Gag precursor (Pr65) and the processed capsid (CA) protein in B2 cells (Fig. 2A). Culture supernatants of the transfected B2 cells were used to inoculate fresh NIH 3T3 cells to examine the structure of extrachromosomal viral DNA in the virus-infected NIH 3T3 cells by BstEII digestion, which discriminated MLV-B(CAG), MLV-GE^6.4, and MLV-GEBstE (Fig. 2C). Hybridization signals corresponding to MLV-B(CAG), which had been already carried by B2 cells, and the newly transfected virus, MLV-GE^6.4 or MLV-GEBstE, were clearly detected. There was no detectable signal indicating the mutation or recombination of the viral genes (Fig. 2B). Therefore, MLV-GE^6.4 as well as MLV-GEBstE was able to complement MLV-B(CAG) virus.

FIG. 2.

Complementation between MLV-B(CAG) and MLV-GE^6.4 or MLV-GEBstE. (A) Processing of Gag precursor by the complementation. The pMLV-B(CAG)-neo-transfected clone, B2, was supertransfected with pGE^6.4-hmB (lane 3) or pGEBstE-hmB (lane 4). After selection with hygromycin B (160 μg/ml) for 2 weeks, the stable transfectants were obtained and used for the assay. The cells were labeled for 3.5 h with 25 μCi of l-[³⁵S]methionine per ml. The cell lysates with the same radioactivity were immunoprecipitated with an anti-ecotropic virus serum (21) and electrophoresed in an SDS–10% polyacrylamide gel. Lanes: 1, NIH 3T3 cells; 2, B2 cells; 3, pGE^6.4-transfected B2 cells; 4, pGEBstE-transfected B2 cells. Molecular size markers (in kilodaltons) are shown on the left. (B) Southern blot analysis of Hirt’s supernatants of NIH 3T3 cells infected for 18 h with the viruses produced from B2 cells supertransfected with either pGE^6.4 (lanes 1 and 3) or pGEBstE (lanes 2 and 4). Undigested DNAs (lanes 1 and 2) or those digested with BstEII (lanes 3 and 4) were electrophoresed and hybridized with the probe, PstI-XhoI (nt 739 to 1560) fragment of Mo-MLV p8.2 (29) (hatched box in panel C). L8.8, 8.8-kb linear provirus with two LTRs; L6.4, 6.4-kb linear provirus with one or two LTRs; C8.8, 8.8-kb circular provirus with one or two LTRs; C6.4, 6.4-kb circular provirus with one or two LTRs. Molecular size markers (in kilobases) are shown on the right. (C) Fragments hybridizing with the probe in the BstEII-digested linear DNAs are indicated by thick lines. The 5.8- and 6.4-kb bands, migrating more slowly than the 5.2-kb band in lane 4 (panel B), were derived from circular DNAs with one or two LTRs. (D) Titer determination of viruses produced from the supertransfectants. The viruses produced from B2 cells supertransfected with either pGE^6.4 (A) or pGEBstE (B) were subjected to titer determination in NIH 3T3 (□), B2 (•), or GEBst2 (○) cells by the UV-XC assay. Small symbols (□ and • in panel b) indicate another titer determination. The average XC plaque count of duplicate dishes was multiplied by the dilution factor and plotted against the virus dilution. Theoretical three-hit (——), two-hit (–––), and one-hit (— · —) curves are shown.

To further compare the ability of MLV-GE^6.4 and MLV-GEBstE to complement MLV-B(CAG), culture supernatants of the cells containing the MLV-B(CAG) provirus and either of the complementing proviruses were collected and their virus titers were measured by the UV-XC assay (26) on untransfected NIH 3T3, B2, or pGEBstE-transfected NIH 3T3 cells (GEBst2 cells) (Fig. 2D). The titer determination kinetics of the mixture of MLV-B(CAG) and MLV-GE^6.4 were expected to be two-hit in NIH 3T3 cells (because both viral genomes were necessary) and one-hit in B2 cells and GEBst2 cells [because the B2 cells were constitutively producing Gag-Pol plus Env, which can complement MLV-GE^6.4 encoding Gag (and Env), and the GEBst2 cells were producing Gag, which can complement MLV-B(CAG) encoding Gag-Pol and Env]. The kinetics for the MLV-B(CAG)/MLV-GE^6.4 mixture in NIH 3T3 cells and in B2 cells were just as expected (Fig. 2D, panel a), but the kinetics in the GEBst2 cells was two-hit (Fig. 2D, panel a). Since the GEBst2 cells were much more sensitive than the normal NIH 3T3 cells, the MLV-GEBstE genome in the GEBst2 cells was considered to complement MLV-B(CAG) to some extent but not sufficiently that a single infection by MLV-B(CAG) could establish the infection in the cells. The titer determination kinetics of the MLV-B(CAG)/MLV-GEBstE was expected to follow two-hit kinetics in NIH 3T3 cells and one-hit kinetics in B2 cells and GEBst2 cells. In the actual titer determination, it was difficult to measure the kinetics in NIH 3T3 cells on account of the low infectivity; the kinetics was near two-hit in the B2 and GEBst2 cells (Fig. 2D, panel b).

The above titer determination experiments suggested that (i) Env expressed by B2 cells was insufficient for receptor interference because the cells were sensitive to the MLV-B(CAG)–MLV-GE^6.4 mixture; (ii) infection by one virion each of MLV-B(CAG) and MLV-GEBstE was not enough to establish infection in NIH 3T3 cells; (iii) although the MLV-GEBstE genome in GEBst2 cells expressed Gag at a level comparable to MLV-GE^6.4, the single infection of GEBst2 cells by MLV-B(CAG) was not enough to establish infection, requiring additional infection by MLV-B(CAG) or MLV-GE^6.4; and (iv) although the B2 cells expressed Gag-Pol and Env, the single infection with MLV-GEBstE supplying enough Gag failed to establish infection. Therefore, although the coinfection of MLV-B(CAG) and MLV-GE^6.4 and that of MLV-B(CAG) and MLV-GEBstE both provided the full set of needed proteins, the latter combination was somehow deficient. The difference was probably that although MLV-GE^6.4 encoded Env in addition to Gag, MLV-GEBstE encoded only Gag. It should be noted here that expression of Env, both SU-TM (gPr90) and SU (gp70), was much lower in the B2 cells transfected with pGEBstE-hmB than in those transfected with pGE^6.4-hmB (Fig. 2A).

Complementation of MLV-B(CAG) by the MLV-GE^6.4 type construct with amphotropic or xenotropic Env.

If Env encoded by MLV-B(CAG) was insufficient and that encoded by MLV-GE^6.4-type provirus was necessary, the host range of the viruses produced by the cells harboring MLV-B(CAG) and MLV-GE^6.4-type genome would be determined by the Env encoded by the latter. To test this hypothesis, an MLV-GE^6.4-type molecule with amphotropic or xenotropic env (MLV-GE-am or MLV-GE-xe) was constructed by replacing the env portion of MLV-GE^6.4 with that of the amphotropic MLV clone 4010A (4) or that of xenotropic MLV clone NZB-9-1 (23) (Fig. 1). If the hypothesis is correct, the host range of the MLV-B(CAG)–MLV-GE-am mixture or the MLV-B(CAG)–MLV-GE-xe mixture will be amphotropic or xenotropic, respectively. To determine the host range of the virus preparations, the following cells were used: NIH 3T3 cells permissive to ecotropic and amphotropic viruses, the feral mouse-derived MDTF cells sensitive to any of the above viruses except Mo-MLV (17) [therefore, Env encoded by MLV-B(CAG) cannot mediate this infection], and the feline PG-4 S⁺L⁻ cells sensitive to amphotropic and xenotropic MLVs but not to ecotropic MLV. The virus titers were determined by the UV-XC and the UV-PG4 S⁺L⁻ assays (see Materials and Methods). The former assay detects the cells expressing ecotropic MLV Env, and the latter assay detects the replication of amphotropic or xenotropic virus.

(i) Complementation between MLV-B(CAG) and MLV-GE-am.

The B2 cells were transfected with pGE-am-hmB (plasmid encoding MLV-GE-am) and selected for hygromycin-resistant cells. The total RNA was extracted and analyzed by Northern hybridization with the 3′-LTR probe and the ecotropic or amphotropic env-specific probe. The unspliced and spliced RNAs encoded by MLV-B(CAG) genome (Fig. 3A, lane 4; see also lanes 3 and 5 for comparison) and those encoded by MLV-GE-am were detected (lane 6; see also lanes 3 and 5 for comparison); the spliced mRNA of the latter was double banded, with the faster-moving band being of the expected size. No genome-sized recombinant between MLV-B(CAG) and MLV-GE-am was formed. The virus titer was determined by the UV-XC and the UV-PG4 S⁺L⁻ assays in NIH 3T3 and MDTF cells (Fig. 3B, panel a). The UV-XC assay and the UV-PG4 S⁺L⁻ assay produced nearly the same titer determination curves. The kinetics was two-hit in NIH 3T3 and MDTF cells, as expected. The titer in MDTF cells was much higher than in NIH 3T3 cells, indicating that the mixture had the overall amphotropic host range [note that MLV-B(CAG) plus MLV-GE^6.4 both encoding ecotropic Env failed to infect MDTF cells, which are resistant to Mo-MLV (Fig. 3B, panel b)]. Since similar titer determination curves were obtained in the UV-PG4 assay and in the UV-XC assay, almost all of the infected cells were considered to express the ecotropic Mo-MLV Env.

FIG. 3.

Complementation between MLV-B(CAG) and MLV-GE-am or MLV-GE-xe. (A) Panel a shows a Northern blot analysis of the cells producing MLV-B(CAG) and MLV-GE-am. A 10-μg portion of total RNA of the original B2 cells was loaded in lanes 1, 3, and 5, and the same amount of total RNA of B2 cells supertransfected with pGE-am was loaded in lanes 2, 4, and 6. The filters were probed with the 3′-LTR probe (lanes 1 and 2), the ecotropic env-specific probe (lanes 3 and 4), or the amphotropic env-specific probe (lanes 5 and 6). Panel b shows a Northern blot analysis of MDTF cells stably transfected with pMLV-B(CAG)-hmB alone or cotransfected with pMLV-B(CAG)-hmB and pGE-xe-neo. A 10-μg portion of total cellular RNA of MDTF cells transfected with pMLV-B(CAG)-hmB alone was loaded in lanes 7, 9, and 11, and the same amount of total RNA of the cells cotransfected with pMLV-B(CAG)-hmB and pGE-xe-neo was loaded in lanes 8, 10, and 12. The filters were probed with the 3′-LTR probe (lanes 7 and 8), the ecotropic env-specific probe (lanes 9 and 10), or the xenotropic env-specific probe (lanes 11 and 12). The band hybridizing with the xenotropic MLV env probe corresponding to the unspliced RNA (lane 12) was absent in the original constructs. Un, unspliced RNA; un GE, unspliced RNA of MLV-GE-am or MLV-GE-xe; sp, spliced RNA. (B) In panel a, the supernatant of the coculture of MDTF cells and the pGE-am-super-transfected B2 cells was subjected to titer determination in NIH 3T3 (□, ▪) and MDTF (○, •) cells by the UV-XC (open symbols) or UV-PG4 S⁺L⁻ (solid symbols) assays. The estimated virus titer is the number of plaques multiplied by the dilution factor; this value is plotted against the virus dilution. Theoretical two-hit (–––) and one-hit (——) curves are shown. Panel b shows titer determination of the mixture of MLV-B(CAG) plus MLV-GE^6.4 as a control; it produced no XC plaque in MDTF cells even at the lowest tested dilution, 2⁻¹. In panel c, the supernatant of MDTF cells cotransfected with pMLV-B(CAG)-hmB and pGE-xe-neo was subjected to titer determination in MDTF and in NIH 3T3 cells by the UV-PG4 S⁺L⁻ or the UV-XC assays.

(ii) Complementation between MLV-B(CAG) and MLV-GE-xe.

MDTF cells were cotransfected with pGE-xe-neo and pMLV-B(CAG)-hmB and selected for hygromycin B-resistant cells. The cells contained MLV-B(CAG)-encoded RNAs (Fig. 3A, lane 10; see lanes 9 and 11 for reference), MLV-GE-xe-encoded RNAs (lane 12), and the genome-sized RNA hybridizing with the xenotropic env probe (lane 12). The last RNA species was considered to represent a recombinant between MLV-B(CAG) and MLV-GE-xe genomes. Its presence was confirmed by limiting dilution and isolation of the virus infectious to PG-4 S⁺L⁻ and MDTF cells but not to NIH 3T3 cells (data not shown). The replication-competent ecotropic MLV was not detected.

The supernatant was subjected to titer determination in NIH 3T3 and MDTF cells by the UV-XC and the UV-PG4 S⁺L⁻ assays. By the UV-PG4 S⁺L⁻ assay, the titer was near 10⁵ in MDTF cells and around 3 × 10³ in NIH 3T3 cells (Fig. 3B, panel c). The UV-XC assay of parallel cultures gave essentially the same pattern, but the titers were 1 order lower (Fig. 3B, panel c). The higher titer in MDTF cells than in NIH 3T3 cells in the UV-XC assay reflected the excess presence of the replication-competent recombinant virus with the xenotropic host range (see above), and the positive UV-XC assay in the Mo-MLV-resistant MDTF cells (Fig. 3B, panel c) indicated that the infection of virions with MLV-B(CAG) genome occurred by using the xenotropic Env. The positive UV-PG4 S⁺L⁻ assay in NIH 3T3 cells indicated that the Env encoded by MLV-B(CAG) was actually used for the infection of virions encoding the xenotropic Env in the NIH 3T3 cells.

These data suggested that when MLV-B(CAG) was expressed together with MLV-GE-am or MLV-GE-xe, virions with chimeric Env were produced and the host range of the majority of virions was determined by Env protein of MLV-GE-am or MLV-GE-xe, respectively.

Expression of the env mRNA and the Env protein by MLV-B(CAG) was lower than that by wild-type Mo-MLV due to a smaller number of proviral DNA copies.

As shown above, B2 cells harboring the MLV-B(CAG) provirus did not exhibit a significant level of receptor interference and capacity to produce sufficient Env for virus production. This was rather surprising because MLV-B(CAG) virus was equipped with all of the genetic elements required for expression of the env mRNA and production of the Env protein. Since it was possible that Env protein production by MLV-B(CAG) virus was affected by an unexpected mechanism, we compared the levels of viral transcripts and proteins in B2 cells and the wild-type Mo-MLV-infected NIH 3T3 cells. Hybridization analysis of total RNA showed that the level of viral RNA was eightfold higher in Mo-MLV-infected cells than in B2 cells (Fig. 4B). The ratio of the spliced viral RNA to the unspliced RNA was the same for both viruses (Fig. 4B, compare lanes 6 and 8). Then the level of Env protein expression was compared by Western blot analysis with the anti-gp70 antibody (Fig. 4C). The results indicated that about 16-fold less Env protein was present in B2 cells than in the cells infected with the wild-type MLV. The number of integrated proviral copies in the cells transfected with the viral constructs was also compared by DNA hybridization analysis. The results indicated that the average amount of proviral DNA in the wild-type-MLV-infected cells was eight times that in B2 cells (Fig. 4A). Thus, the level of viral RNA per proviral DNA copy was the same for both viruses, and the level of the Env protein per proviral DNA in the MLV-B(CAG)-transfected cells was about half of that in the wild-type-MLV-infected cells.

FIG. 4.

Env expression by wild-type Mo-MLV and by MLV-B(CAG). The Env expression was compared between the wild-type (Wt) Mo-MLV-infected cells and the B2 cells. DNA, RNA, and protein were extracted from the cells prepared from the parallel cultures. (A) Southern blot analysis with the probe specific for ecotropic env (probe 2 in Fig. 1). A 12-μg sample of genomic DNA was digested with EcoRV. The 3.5-kb fragment (from nt 4086 to 7606) was expected to be detected by the probe. For the wild-type-MLV-infected cells, DNA was serially diluted twofold and loaded in lanes 2 to 6 (6 μg in lane 2 and 6/16 μg in lane 6). For pMLV-B(CAG)-neo-transfected cells, 6 μg of the digested DNA was loaded in lane 7 and 3 μg was loaded in lane 8. A 6-μg portion of the digested DNA of untransfected NIH 3T3 cells was loaded in lane 1 as a control. Mo, Mo-MLV-derived 3.5-kb band; endo, endogenous virus-derived band. (B) Northern blot analysis with the 3′-LTR probe (probe 1 in Fig. 1) and a β-actin probe. A 16-μg portion of total cellular RNAs was loaded in lanes 2 (wild type) and 7 [MLV-B(CAG)]. Twofold serially diluted RNAs were loaded in lanes 3 to 6 for wild-type-MLV-infected cells and in lane 8 for pMLV-B(CAG)-transfected cells. Un, unspliced RNA; sp, spliced RNA. (C) Panel a shows a Western blot analysis with an anti-gp70 antibody. A 28-μg portion of protein was loaded in lanes 1 (untransfected NIH 3T3 cells), 2 (wild-type-MLV-infected cells), and 7 (B2 cells). Twofold serially diluted proteins from the wild-type-MLV-infected cells were loaded in lanes 3 to 6, and the twofold-diluted protein of B2 cells was loaded in lane 8. Panel b shows a Coomassie-blue stain of the gel prepared in parallel with that in panel a.

The level of ecotropic Env protein in B2 cells is elevated by the expression of heterologous amphotropic Env protein.

To find the possible mechanism for the twofold difference in the level of the Env protein per proviral DNA copy in B2 cells and the wild-type-MLV-infected cells, we compared the rate of Env protein synthesis and processing in these cells by pulse-chase experiments (Fig. 5). In the wild-type-MLV-infected cells, unprocessed gPr90 was gradually processed to gp70 (SU) over the course of 5 h (Fig. 5A, lanes 5 through 8). In contrast, in the MLV-B(CAG)-transfected cells, the unprocessed gPr90 was detected as a thinner band immediately after pulse-labeling and neither the unprocessed nor the processed Env protein was detectable thereafter (lanes 1 through 4). Thus, it appeared that the Env protein produced in the B2 cells was degraded or released from the cells relatively rapidly. These observations may account for the twofold difference in the level of Env protein per proviral DNA copy. We also carried out a similar analysis on the B2 cells transfected with or without the pGE-am-hmB construct which encodes the amphotropic Env protein. In pGE-am-hmB-transfected B2 cells, the amphotropic Env proteins, whose mobility in the gel was slower than its ecotropic counterpart (Fig. 5B, panel b, lanes 1 and 2), was expressed and processed to the SU protein (Fig. 5B, panel a, lanes 10 through 12). The ecotropic Env proteins were more abundant in the pGE-am-hmB-transfected B2 cells (Fig. 5B, panel a, lanes 10 through 12) than in the B2 cells containing only the MLV-B(CAG) provirus (lanes 7 through 9). The kinetics of the ecotropic Env protein processing in the pGE-am-hmB-transfected B2 cells also became similar to that in the wild-type-MLV-infected cells.

FIG. 5.

Pulse-chase experiments. (A) The cells harboring wild-type Mo-MLV and those harboring MLV-B(CAG) were pulse-labeled for 1 h with 30 μCi of l-[³⁵S]methionine per ml and chased for 1, 3, and 5 h. The lysates of 5 × 10⁵ cells were immunoprecipitated with the anti-gp70 antibody and analyzed in SDS–10% acrylamide gel. Lanes: 1 to 4, pMLV-B(CAG)-neo-transfected cells (B2 cells); 5 to 8, wild-type-MLV-infected cells; 9 to 12, NIH 3T3 cells. Lanes 1, 5, and 9, without chase; lanes 2, 6, and 10, chase for 1 h; lanes 3, 7, and 11, chase for 3 h; lanes 4, 8, and 12, chase for 5 h. Molecular size markers (in kilodaltons) are shown on the right. (B-a) pGE-am-hmB-supertransfected B2 cells together with the control cells were pulsed for 30 min and chased for 1 and 3 h. Lanes: 1 to 3, wild-type-MLV-infected cells; 4 to 6, NIH 3T3 cells; 7 to 9, pMLV-B(CAG)-transfected cells (B2 cells); 10 to 12, pGE-am-supertransfected B2 cells; 13 to 15, pGE-am-transfected NIH 3T3 cells. Lanes 1, 4, 7, 10, and 13, without chase; lanes 2, 5, 8, 11, and 14, chase for 1 h; lanes 3, 6, 9, 12, and 15, chase for 3 h. (B-b) Comparison of the mobilities of SU-TM and SU of Mo-MLV and amphotropic MLV. SU-TMam, SU-TM of amphotropic MLV-GE-am; SU-TMec, SU-TM of ecotropic Mo-MLV; SUam, SU of amphotropic MLV-GE-am; SUec, SU of ecotropic Mo-MLV.

Effects of the number of proviral DNA copies in NIH 3T3 cells on virus production and interference.

The above results suggested that the ability of MLV to produce the Env protein was a crucial determinant for establishing productive infection in NIH 3T3 cells and that the number of proviral copies capable of encoding the Env protein played a key role in controlling the ability. To test this hypothesis, we examined the relationship between the number of proviral copies and the level of viral gene products. For this purpose, we used the replication-defective Δwt virus, which contained a 306-base deletion in the pol gene (24) and was less prone to spontaneous reversion than the point mutant virus MLV-B(CAG). By transfecting the Δwt construct into NIH 3T3 cells repeatedly, we obtained NIH 3T3 cell clones harboring different numbers of the Δwt proviral copies. Among 11 clones, the number of proviral copies was the highest in clone 11 and was estimated to be eight (Fig. 6A). Clones 1, 3, 7, and 33 contained about four copies of the proviral DNA per cell, clones 8 and 9 contained three copies, clones 31 and 32 contained two copies, and clones 2 and 6 contained only one copy (Fig. 6A). The numbers of proviral copies in clones 6 and 11 were ascertained by DNA digestion with HindIII, which cut the transfected plasmid at a single site: the detected band represents the fragment between the single-cut site (nt 4894) and the nearest cellular site downstream of the provirus. A single band was detected for B2 and clone 6, while at least eight bands were detected for clone 11 (Fig. 6B). The level of the viral transcripts in these clones was examined by hybridization analysis. The intensity of the hybridization signals corresponding to the unspliced and the spliced viral RNAs in each clone was normalized with respect to the intensity of the signal of the control actin mRNA. The normalized levels of the viral RNA in these clones were proportional to their proviral copy number (Fig. 6C). The level of the Env protein expressed in clones 1, 3, 7, 8, 9, 11, and 33 appeared to be correlated with the number of proviral copies in these clones (Fig. 6D). However, the amount of Env present in clones 2, 6, 31, and 32 appeared disproportionately small (Fig. 6D).

FIG. 6.

Env expression of Δwt-transfected clones. Env expression between the wild-type-Mo-MLV-infected cells and Δwt-transfected clones was compared. DNA, RNA, and protein were extracted from the cells prepared at the same time. (A) Southern blot analysis with the ecotropic env-specific probe (probe 2 in Fig. 1). The genomic DNAs were digested with EcoRV, and the detected region was the fragment between nt 4086 and 7606. For each clone, 6 μg of EcoRV-digested DNA was loaded. For the wild-type-MLV-infected cells and also for clone 11, DNAs serially diluted twofold were also loaded onto the gel. The clone numbers are indicated above the lanes. Analysis of the DNA from the uncloned NIH 3T3 cells transfected with pΔwt-hmB and selected for hygromycin B resistance (uncloned) is also shown. The open arrowhead indicates the Δwt provirus copy, and the solid arrowhead indicates the endogenous provirus copy. (B) The genomic DNAs of Δwt-transfected clone 6 and 11, B2, and NIH 3T3 cells were digested with HindIII, which cut the transfected plasmid at the single site, nt 4894 in the Mo-MLV sequence, and probed with the ecotropic env specific probe. The endogenous virus-derived bands are shown by solid arrowheads, and the bands labelled with open arrow heads indicate the fragments containing the 3′ half of the MLV (from nt 4894 to the 3′ end). (C) Northern blot analysis with the 3′-LTR probe and a β-actin probe. A 16-μg portion of total cellular RNA was loaded in each lane. Twofold serially diluted RNAs were loaded for the wild-type-MLV-infected cells. The clone numbers are indicated above the lanes. Un, unspliced RNA; sp, spliced RNA; ac, actin. (D) Panel a shows a Western blot analysis with the anti-gp70 antibody. A 28-μg portion of proteins was loaded in each lane. Twofold serially diluted proteins from the wild-type-MLV-infected cells were also loaded on the gel. Panel b shows a SYPRO Orange (Bio-Rad) protein stain of the gel before blotting.

The UV-XC assay was carried out on the clones. Clone 11 generated extensive XC cell fusion, indicating efficient Env expression by this clone, while clones 2, 6, 31, and 32 did not induce a detectable level of XC cell fusion. Although the other clones, 1, 3, 7, 8, 9, and 33, were able to induce XC cell fusion, the fusion was less extensive than that generated by clone 11. Representative photographs are shown in Fig. 7B. The numbers of virions released by these clones were compared by measuring the level of viral RNA in the filtered culture supernatant (Fig. 7A). The titer of the viral RNA in 25 μl of the culture fluids was on the order of 3⁷ for wild-type-Mo-MLV-infected NIH 3T3 cells; 3⁴ for clone 11; 3³ for clones 1, 3, 7, and 33; 3² for clones 8 and 9; and 3⁰ for clones 2, 6, 31, and 32. The lower titer in clone 11 relative to that in the wild type in spite of the similar numbers of proviral copies was probably due to the mutation in rt, which resulted in the aberrant processing of Gag (24).

FIG. 7.

XC assay and the viral RNA released in the culture supernatants of Δwt-transfected clones. (A) RNA was extracted from 250 μl of filtered culture supernatant, and 1/10 was used for the RT-PCR assay, which detected the 5′ part of env (see Fig. 1). The RNA samples were diluted threefold serially before being used in RT-PCR. (B) Confluent cultures of Δwt-transfected clones were subjected to the UV-XC assay. The cells were stained with crystal violet. The clone number is indicated on the upper left of each photograph.

The levels of receptor interference displayed by the clones were compared by examining their resistance to focus induction by Moloney murine sarcoma virus. As shown in Table 1, clone 11 was almost as resistant as Mo-MLV-infected NIH 3T3 cells. On the other hand, clones 2 and 6 were as susceptible as the uninfected NIH 3T3 cells, exhibiting no significant level of receptor interference. Clones 1, 3, 7, 8, 9, and 33 showed a considerable but incomplete level of receptor interference.

TABLE 1.

Receptor interference by NIH 3T3 cell clones

Δwt clone	MSV titer (FFU) (%) ina:
	Expt 1	Expt 2
2	1,875 (96)
6	3,750 (192)
31		600 (47)
32		105 (8)
8	75 (8)	150 (12)
9		135 (11)
1	108 (6)
7	15 (0.8)
33		6 (0.5)
3		3 (0.2)
11	3 (0.2)	12 (1)
Wtb	0 (0)
NIH 3T3	1,950 (100)	1,275 (100)

^aMurine sarcoma virus rescued by Mo-MLV was subjected to titer determination in Δwt-transfected clones. The mean results of duplicate titer determinations are shown. 

^bWt, wild-type Mo-MLV-infected NIH 3T3 cells. 

These results demonstrated that more than four MLV provirus copies were necessary for establishing productive infection characterized by expression of a high level of Env, production of a large amount of viral particles, and resistance to superinfection by receptor interference.

DISCUSSION

The gag-pol readthrough mutant, MLV-B(CAG), encoding Gag-Pol fusion and Env proteins, was complemented by MLV-GE^6.4, encoding Gag and Env. It was also complemented by another mutant, MLV-GEBstE, encoding only Gag but far less efficiently. When it was complemented by MLV-GE^6.4-type virus with amphotropic or xenotropic Env, the host range of the virus preparation as a whole became amphotropic or xenotropic, respectively. It was thus suggested that Env encoded by MLV-B(CAG) was insufficient and Env encoded by MLV-GE^6.4-type virus was necessary.

To find why the Env expression by MLV-B(CAG) provirus alone was insufficient, we first compared the transcription level between the B2 cells harboring MLV-B(CAG) and the wild-type-MLV-producing cells. The amount of transcript per provirus was the same for the both, but since the B2 cells had eightfold fewer proviral copies, they expressed eightfold less mRNA. This suggested that the number of proviral copies per cell was crucial for producing Env required for the virion production. To test this possibility, we used an MLV mutant Δwt, which had a small in-frame deletion in the rt region [we used this mutant instead of MLV-B(CAG) on account of its genetic stability]. With repeated transfection of the cells with Δwt, we obtained the cell clones with different proviral copy numbers.

As summarized in Fig. 8A, although the amount of mRNA increased linearly as a function of proviral copy number, the interference and the virion release increased rather abruptly when the number of viral copies exceeded three or four per cell; i.e., there was a threshold of proviral copies per cell required for establishing the interference and the efficient virion production. It was also noticed that the amount of Env product per cell did not increase linearly but increased rather abruptly at the above-mentioned threshold proviral copy number.

FIG. 8.

Relationship between the number of proviral copies and virus gene expression. (A) The relative amount of viral transcripts in the cells (•) and the viral RNA in the culture supernatants (▵) were plotted against the proviral copy number for Δwt-transfected clones. The relative titers of the MSV obtained on these cells (the indicator of the interference) were also plotted (○). The number of proviral copies was deduced from the density of the band representing the provirus (the density of the endogenous virus band was used for the normalization) (Fig. 6A). The amount of viral RNA in the cells was expressed by the relative density in the Northern blot analysis (Fig. 6C). The relative viral RNA release from the cells was expressed as a percentage of that from clone 11 (Fig. 7A). (B) Schematic representation of the hypothesis. When there is only one provirus copy in the cell, even if the expression per viral copy is the same, Env expression is insufficient because there is a smaller chance of multimerization of Env protein and its higher rate of degradation. Such cells may be infected with the virus repeatedly until the interference is established and the cells begin to produce virions at the maximum rate.

Fan et al. (7) previously examined virus production in the producer cell clones which were obtained by low- or high-multiplicity infection followed by cell cloning 6 h later. They found no correlation between virus production and the number of viral copies in the cells. Their conclusion entirely contradicts the conclusion obtained here. It should be noted, however, that under the conditions used by Fan et al. (7), the possible expansion of proviral copies within the cloned cells was not rigorously precluded and at least one clone (A9) which initially contained a single copy was later found to contain multiple copies (2, 7).

Hwang and Gilboa (12) reported that the rEnv-Neo^r plasmid, which was constructed by replacing env by the Neo^r gene, expressed 10- to 50-fold-higher levels of vector-specific RNA when introduced by retroviral infection than when introduced by transfection and that expression per provirus introduced by transfection was variable depending upon the integration site. The reduced gene expression after transfection was due to partial methylation (12). In our experiments with the pA^rMLV-48-derived constructs, however, the RNA expression per provirus introduced by transfection was almost invariable for all the transfectant mouse cell clones and also for all the mutant constructs used in the experiments; i.e., there appeared to be no influence of integration site on the proviral expression. The conclusion was in contradiction to that of Hwang and Gilboa (12). In our experiments, however, all of the constructs derived from pA^rMLV-48 retained the mouse cell DNA sequences at the natural integration site on both sides of the provirus. The 5′-flanking cellular sequence was as long as 1.7 kb, and it was cloned from the chromosome, where the provirus was actively transcribed. The introduction of the provirus together with the flanking cellular sequence as a set probably allowed proviral expression irrespective of the site of the integration. Since the rEnv-Neo^r used by Hwang and Gilboa consisted of plasmid pBR322 and the retroviral gene, the transfected DNA was directly flanked by the cellular sequence at the site of integration. Therefore, the LTR would easily be subject to methylation at the integration site.

The maximum number of copies of MLV proviruses in productively infected NIH 3T3 cells was reportedly around 10 (2, 5, 20, 24), and our study showed that interference was established around 4 to 8 copies per cell. It is therefore inferred that the maximum number of provirus copies was determined by the interference. In Mo-MLV transgenic mice, leukemic cells were found to contain multiple proviral copies while other somatic cells continued to have a single copy (14). This apparent contradiction may be explained by postulating that the viral receptor is down regulated in nontarget tissues (this assumption is not entirely baseless, because exogenously infected MLV targets the lymphatic cells of the mice). If the receptor expression is low, Env protein and consequently proviral copies needed for establishing the interference will be low. Thus, in such cells, the viral expression will be barely detectable on account of the small number of proviral copies. Proviruses in such cells will be in apparent latency. If this argument is valid, the proviral copy number per cell in vivo must be regulated primarily by the balance between the expression level of viral receptors and the expression level of the virus. In this respect, it is interesting that the extrachromosomal copies of human immunodeficiency virus (HIV) DNA indicative of a further integration event accumulated after tumor necrosis factor alpha activation in the persistently infected cell line (16). It was also reported that the superinfection actually took place in cloned cell lines actively producing HIV-1 (25). It may be worthwhile examining whether the number of HIV proviral copies per cell in lymphatic cells or macrophages is kept low in the latent phase and increases after transition to the overt phase.

The above conclusion that active virus production and the establishment of interference required multiple proviral copies in the cells does not mean that infection by multiple virions was necessary to establish the virus infection in NIH 3T3 cells. Actually, the titer determination curve of Mo-MLV follows a single-hit kinetics. Each of the cells infected with a single virion supposedly produces a low level of virions at the beginning. Since the interference is not established in such cells, the released virions will infect the same cells again. The process will be repeated until the interference is established.

The presence of a threshold number of proviral copies needed for interference and active virus production is probably due to the fact that structural proteins of viruses function as multimers (11, 30, 31). If the formation of multimers is dependent on the concentration, a twofold decrease of the concentration results in a fourfold decrease in the number of dimers and an eightfold decrease in the number of trimers. Thus, formation of multimers does not follow a linear dose-response curve but a sigmoid dose-response curve. Therefore, the structural proteins must be produced in large amounts. To achieve this, lytic viruses, such as poliovirus and SV40, make large amounts of structural proteins at the cost of host cell death. However, in retroviruses which are not cytopathic, expression of the structural proteins has to be regulated at a level that is just sufficient. Our data suggests that the regulation is achieved by that of the number of proviral copies per cell and that the number of proviral copies is controlled by the interference, whose full expression is set around eight proviral copies in NIH 3T3 cells. Thus, the interference may play an important regulatory role in the expression of Mo-MLV, although the level of transcription, which could differ between cell types and integration sites, may be an added factor.

An unanswered question in the complementation between MLV-B(CAG) and MLV-GEBstE is why the MLV-B(CAG) proviral copy number did not increase as a result of the repeated infection in the initially infected cells to attain a level sufficient for Env expression. One explanation among others is the following. For proper replication, retroviruses require expression of Gag-Pol and Gag at an appropriate ratio, which is around 1:10 (13, 22). The 1:1 presence of MLV-B(CAG) and MLV-GEBstE in the initially infected cells will result in the production of Gag-Pol and Gag in an equal ratio; i.e., Gag-Pol will be present in relative excess. If the MLV-B(CAG) copy number increases to elevate the expression of Env, Gag-Pol will be present in far greater excess. Thus, in this combination, the optimum complementation can never be obtained. Meanwhile, in a complementation between MLV-B(CAG) and MLV-GE, the optimum ratio between Gag and Gag-Pol can be reached by increasing the copy number of MLV-GE encoding both Gag and Env. Actually, the MLV-B(CAG)- and MLV-GE^6.4-producing cells expressed MLV-GE^6.4 in about 10-fold excess over MLV-B(CAG) (21). Therefore, for construction of packaging cell lines with any combination of viral components, it is important to provide each gene in an amount and in a proportion optimal for the virion formation.