Two Point Mutations Produce Infectious Retrovirus Bearing a Green Fluorescent Protein-SU Fusion Protein

Abstract

Two second-site mutations in Moloney murine leukemia virus envelope surface protein (SU) were previously shown to rescue infection of two different SU mutants, a fusion-defective point mutant and a fusion-defective modified SU that exhibits weak subunit association. We report here that they also rescue infection of a third defective SU, one modified by insertion of the green fluorescent protein (GFP) between serine 6 and proline 7. GFP-SU assembled into virions and showed a strong association with the transmembrane protein (TM). However, these virions were noninfectious. GFP-SU expression was not maintained within cells, suggesting that the protein was toxic. Addition of the second-site mutations rendered the GFP-SU virus infectious and resulted in prolonged expression of the modified envelope protein. This virus showed a slight reduction in receptor binding but not in envelope protein processing, suggesting that addition of the GFP sequences results in subtle structural changes. Extrapolating these data, we see that the fundamental problem with the GFP-SU envelope protein appears to be a folding problem, suggesting that the second-site mutations rescue GFP-SU primarily by a mechanism that involves stabilizing the envelope protein structure.

Deletion or replacement of the essential histidine at position 8 of the ecotropic Moloney murine leukemia virus (MoMLV) surface protein (SU) results in loss of infection. Histidine 8 mutants maintain a stable association with the transmembrane protein (TM) and bind to the ecotropic receptor but fail to mediate virus-cell fusion (2, 17). Fusion and infection can be rescued by adding two second-site amino acid changes: a glutamine-to-arginine change at position 227 (Q227R) and an aspartate-to-tyrosine change at position 243 (D243Y) (17).

These second-site mutations also rescue the phenotype of another defective envelope protein, ampho-eco SU (16, 18), a modified SU first designed and characterized by Cosset and colleagues to study the problems encountered in targeting infection of retroviral gene therapy vectors (4). ampho-eco SU contains an insertion of the 208 amino acids of the amphotropic receptor binding domain between peptide linkers placed at residues 6 and 7 of MoMLV SU. In addition to having a postbinding defect (4), ampho-eco SU associates more weakly with TM than does wild-type ecotropic and amphotropic SU (18).

The ability of the second-site mutations to suppress two different defects (i.e., a membrane fusion defect and a weak subunit association) suggested that they might act through more than one mechanism. It also raised the question of how much each mechanism contributed toward the rescue of infection by modified envelope proteins. Specifically, it was possible that they acted on ampho-eco SU only by stabilizing the subunit association. Here we demonstrate that the second-site mutations rescue infection by a modified SU of similar design to ampho-eco SU but having a strong subunit association. This third envelope protein mutant exhibits evidence of subtle misfolding, suggesting that rescue of modified SU by the second-site mutations can involve a mechanism that acts solely by stabilizing envelope protein structure.

With the intention of developing a useful reagent for visualizing virus entry in live cells by confocal microscopy, we constructed an envelope gene encoding a fusion between SU of ecotropic MoMLV and green fluorescent protein (GFP) from Aequoria victoria. The design of this GFP-SU fusion was similar to that used previously by Cosset to construct the original ampho-eco modified SU (4, 12), by us to reconstruct ampho-eco SU (16, 18), and by others to construct SU fusions containing different polypeptide insertions (3, 10, 15, 19). In brief, sequences encoding the enhanced version of GFP flanked by flexible peptide linkers (Ala₃SerGly₄) and containing NotI restriction enzyme sites were amplified, digested with NotI, and ligated into a unique NotI site engineered between codons 6 and 7 of the mature MoMLV SU. The 2,778-bp PmlI-EcoRI restriction fragments containing the entire env-gfp fusion gene from two correctly modified molecular clones were ligated in place of the corresponding sequences in pcDNA MoMLV, a plasmid encoding a recombinant retroviral genome containing the MoMLV gag, pol, and env genes under the control of the cytomegalovirus promoter (17). This recombinant lacks the encapsidation and retroviral long terminal repeat sequences. The resulting plasmid was named pcDNA MoMLV GFP-SU.

We used transient transfection of a prepackaging cell line, H1BAG (17), to produce GFP-SU pseudotype virus transducing a lacZ-containing retroviral genome (9). The lacZ-transducing units were quantified by end point dilution titration on ecotropic MoMLV receptor-positive mouse NIH 3T3 fibroblasts and receptor-negative HEK 293 cells. Wild-type MoMLV pseudotype supernatant contained greater than 10⁵ infectious particles per ml (Fig. 1A), whereas the GFP-SU pseudotype virions were noninfectious. Similar results were obtained in an independent experiment (data not shown). No infections were observed on receptor-negative HEK 293 cells in either experiment.

FIG. 1.

Addition of the second-site changes (Q227R and D243Y) to the GFP-SU fusion protein renders the virus infectious. (A) H1BAG prepackaging cells were transfected with pcDNA MoMLV encoding a wild-type envelope protein or with pcDNA MoMLV GFP-SU or pcDNA MoMLV GFP-SU Q227R D243Y, each encoding modified SU. Replicate sets of quadruplicate wells of ecotropic MoMLV receptor-positive mouse NIH 3T3 fibroblasts or receptor-negative HEK 293 cells were exposed to 10-fold serial dilutions of a 3-ml aliquot of each cell culture supernatant of the H1BAG cells taken 24 h posttransfection. The titer was calculated from the end point dilution. No infection of HEK 293 cells was observed. i.f.u., infection- forming units. (B) The subunit association of the envelope protein does not appear to be weakened by insertion of the GFP sequences. A 7-ml aliquot of each of the virus supernatants described for panel A was then subjected to ultracentrifugation at 30,000 rpm in a Beckman SW41 rotor at 4°C for 2 h through a 25% sucrose cushion to purify virions. Viral proteins were resolved by sodium dodecyl sulfate-PAGE and transferred to nitrocellulose membranes. For the left panels, membranes were cut into two pieces at the position of the 45-kDa molecular mass marker and the upper portion was probed with anti-SU antiserum (1:100 dilution of serum ID 80S000018; Quality Biotech, Inc.), while the lower portion was probed with anti-capsid (CA) antibodies (1:10,000 of serum ID 81S000263; Quality Biotech, Inc.). All primary antibodies were detected by horseradish peroxidase-conjugated secondary antibodies at a 1:10,000 dilution. (C) Transient expression of envelope proteins in transfected H1BAG cells. After the 48-h harvest of supernatant, transfected H1BAG cells were lysed and the cell lysates were subjected to sodium dodecyl sulfate-PAGE and immunoblotting with anti-SU antiserum. WT SU, virus containing wild-type MoMLV envelope protein; GFP-SU, virus containing the GFP-SU fusion protein; GFP-SU Q227R D243Y, virus containing GFP-SU fusion protein with the second-site mutations added.

To determine if the modified SU assembled into virions, we performed Western blot analysis on virus particles purified from 7 ml of each cell supernatant by high-speed centrifugation through a sucrose cushion. This analysis would also show if the GFP-SU association was compromised, because centrifugation through sucrose reveals weak subunit interactions by separating shed modified SU from virions prior to polyacrylamide gel electrophoresis (PAGE) (16, 18). Wild-type MoMLV pellet (Fig. 1B) and producer cell lysate (Fig. 1C) contained the expected 70-kDa SU species, while the GFP-SU virus pellet (Fig. 1B) and producer cell lysate (Fig. 1C) contained an approximately 95-kDa anti-SU reactive species, in agreement with the expected size of 92 kDa for the modified SU. These data indicated that the fusion protein assembled into virions and maintained a relatively strong association with TM. This finding was in contrast to the weak association previously observed for the ampho-eco SU and suggested that infection of GFP-SU virions failed for reasons other than the stability of the SU-TM association.

Thus, GFP-SU presented another test case for investigating the mechanism of action of the second-site mutations. Plasmid pcDNA MoMLV GFP-SU Q227R D243Y was constructed by placing the second-site changes into pcDNA MoMLV GFP-SU. H1BAG prepackaging cells were transiently transfected with the plasmid. The supernatant from cells transfected with pcDNA MoMLV GFP-SU Q227R D243Y and wild-type MoMLV supernatant contained comparable amounts of infectious virus (Fig. 1A), indicating that the addition of the Q227R and D243Y changes to GFP-SU restored infection to wild-type levels. Western blot analysis of the virus pellet with anti-SU antiserum and anti-CA antiserum revealed the presence of both proteins as expected (Fig. 1B). Modified SU was also present in producer cell lysate (Fig. 1C).

We then attempted to establish stable producer cell lines for GFP-SU and GFP-SU Q227R D243Y pseudotype virions. Plasmids pcDNA MoMLV GFP-SU and pcDNA MoMLV GFP-SU Q227R D243Y were transfected into HEK 293/DHFR cells, a human embryonic kidney cell line stably expressing a packageable, recombinant MoMLV genome containing a mutant dihydrofolate reductase (DHFR) gene in place of the gag, pol, and env genes (13). The transfected cells were grown in medium containing the aminoglycoside G418 to select for the neomycin resistance gene present on the plasmids. The green fluorescence of live cells was monitored by fluorescent microscopy and became detectable within 24 h posttransfection in both cell populations (Fig. 2A and B), a phenomenon typical of transient expression. However, GFP-SU-transfected cells were typically less fluorescent than cells transfected with GFP-SU Q227R D243Y at 24 h. After 2 weeks, very few of the G418-resistant colonies of GFP-SU-transfected cells fluoresced green (Fig. 2A, middle panels). In contrast, many of the cells in the G418-resistant colonies of pcDNA MoMLV GFP-SU Q227R D243Y-transfected cells continued to express robust green fluorescence on their surfaces (Fig. 2B, middle panels). By the end of 12 weeks of selection, almost none of the GFP-SU-transfected cells fluoresced green, whereas green fluorescent cells were abundant in the GFP-SU Q227R D243Y-transfected cells (Fig. 2A and B, bottom panels).

FIG. 2.

Stable production of pseudotype virions. (A and B) Cells transfected with plasmid encoding GFP-SU without (A) or with (B) the second-site mutations fluoresced green during the first 24 h posttransfection, but only cells transfected with GFP-SU plus the second-site mutations continued to show robust green fluorescence at later times posttransfection. HEK 293/DHFR cells were transfected with 15 μg of pcDNA MoMLV GFP-SU Q227R D243Y or pcDNA MoMLV GFP-SU and then grown in medium containing 1 mg of G418/ml to select for the plasmids. Micrographs were taken at 24 h posttransfection and after 2 and 12 weeks of selection. The fluorescence emission of GFP (left panels) and phase-contrast images (right panels) of live cells were visualized using a 40× water-immersible objective on an epifluorescent microscope (Zeiss Axiophot). Images were captured using a Kodak digital camera. (C) Western blot analysis of lysates from transfected cell populations at 5 weeks posttransfection using anti-SU antiserum (1:100 dilution of serum ID 80S000018; Quality Biotech, Inc.). For comparison, lysates from cells transfected with wild-type SU were also analyzed. (D) Assembly of GFP-SU Q227R D243Y into virus particles and infectivity of pseudotype virions. Two clonal cell lines, clones 1 and 2, expressing the highest levels of green fluorescence were isolated from the population of HEK 293/DHFR cells stably expressing pcDNA MoMLV GFP-SU Q227R D243Y after 5 weeks of selection with G418. Virions were purified by ultracentrifugation (30,000 rpm in a Beckman SW41 rotor at 4°C for 2 h) through a 25% sucrose density cushion from the cell culture supernatant of each clonal line and from control cells producing wild-type envelope pseudotype of MoMLV DHFR. Replicate sets of HEK 293 cells stably expressing ecotropic receptor cDNA were exposed to 10-fold serial dilutions of virus supernatant. Forty-eight hours later, infected cells were selected by growth in methotrexate-containing medium which required expression of the mutant dhfr gene transduced by the virus, and the titer was calculated from the end point dilution after 21 days of selection. For the left panels, the upper portion of the membrane was probed with anti-SU antiserum, while the lower portion was probed with anti-CA antibodies as described in the legend to Fig. 1. Values under each lane show the titer in infectious units per milliliter. For the right panel, the upper portion of the membrane was stripped of anti-SU antibodies and reprobed with rabbit anti-GFP antiserum (1:1,000 dilution; Clontech). All primary antibodies were detected by horseradish peroxidase-conjugated secondary antibodies at a 1:10,000 dilution. WT SU, recombinant MoMLV DHFR particles containing wild-type SU; GFP-SU Q227R D243Y, virus from the two clonal HEK 293 DHFR cell lines stably expressing pcDNA MoMLV GFP-SU Q227R D243Y.

Western blot analysis of cell lysates at 5 weeks posttransfection confirmed that GFP-SU Q227R D243Y was expressed in the transfected cell population, but GFP-SU levels were too low to be detected (Fig. 2C). Since the levels of GFP-SU and GFP-SU Q227R D243Y in cell lysates and virions from transient transfections were comparable (Fig. 1B and C and 2A and B), the GFP-SU protein does not appear to be turned over more rapidly. Taken together, the loss of green fluorescence and of protein expression within the first weeks posttransfection suggest that accumulation of GFP-SU protein during sustained overexpression is cytotoxic. Regardless of the reason for the inability of cells to maintain GFP-SU overexpression, the addition of the second-site mutations is sufficient to overcome the problem.

After the second week of selection, the G418-resistant colonies were pooled and maintained under selection in 1 mg of G418/ml for more than 5 weeks to select for stable expression. Single cell clones were then isolated from the stable population by limiting dilution. The two clonal lines exhibiting the highest level of green fluorescence were identified using a plate-reading fluorimeter (HTS 7000 Bioassay Reader; Perkin-Elmer) and designated clones 1 and 2. Western blot analysis on particles extracted from clone 1 or 2 supernatant by ultracentrifugation through a sucrose cushion were compared to virions from an equal volume of supernatant from a wild-type ecotropic MoMLV that transduces the DHFR genome (MoMLV DHFR; gift of D. Williams) (13). Virus from clones 1 and 2 contained an approximately 95-kDa SU species (Fig. 2D, left panels). The presence of the GFP sequences in the 95-kDa species was verified using anti-GFP antiserum (Fig. 2D, right panels). Supernatants from clone 2 and the MoMLV producer cells contained comparable numbers of virus particles (i.e., a comparable capsid protein content). Virions from clone 1 contained lower levels of capsid but comparable amounts of modified SU, suggesting that this cell line produced about half as many virions but that these virions contained a greater density of envelope protein.

Quadruplicate wells of HEK 293 cells and of HEK 293 cells stably expressing exogenous ecotropic virus receptor (7) were exposed overnight to 10-fold serial dilutions of clone 1 or 2 cell culture supernatant or to the MoMLV DHFR supernatant. Infected cells were selected by growth in the presence of 150 μM methotrexate (Sigma) and 8% dialyzed donor calf serum. Clone 1 and 2 supernatants exhibited a titer within 10- to 20-fold that of the MoMLV DHFR supernatant containing wild-type envelope protein (Fig. 2D, left panels). No infection of parental HEK 293 cells that lack an ecotropic MoMLV receptor was observed (data not shown).

We first identified the second-site mutations Q227R and D243Y by their ability to suppress the fusion defect of a substitution for histidine 8 (17). Based on the crystal structure of the receptor binding domain of the closely related Friend 57 murine leukemia virus (MLV) (6), we proposed a mechanism of action by which the Q227R change twists the carbon chain just enough to place the tyrosine from the D243Y change into the position normally occupied by histidine 8 (17). Thus, Q227R acted by introducing a structural change, while D243Y acted by providing a functional replacement for the missing histidine.

In the second case studied, virions containing ampho-eco SU bound to the targeted amphotropic receptor but were 10,000-fold less infectious than wild-type amphotropic MoMLV (4). The association of ampho-eco SU with TM was weaker than that of wild-type ecotropic and amphotropic SU (4, 16, 18). The second-site mutations not only increased the rate of infection via the amphotropic receptor to within 10-fold that of of wild-type amphotropic SU but also stabilized the association of ampho-eco SU with TM (16, 18). This unexpected stabilization of subunit association raised the question of how much the structural changes resulting from the second-site changes might be contributing to the rescue of infection versus how much the replacement of the histidine 8 fusion function contributed, if at all. Their ability to rescue the third mutant, GFP-SU, presents a clearer case. It suggests that the second-site mutations can rescue a modified envelope protein through a mechanism that does not involve stabilizing a weak subunit association.

Ager et al. showed that infection using modified SU could be increased by placing the insertion at position 1 of SU (1). Although we did not construct a GFP insertion at position 1, it seems reasonable that this approach might also rescue GFP-SU infection. If so, do the two strategies act through the same mechanism? Addressing this issue will likely require structural information for at least two versions of a modified SU, one made with the second-site mutations and the other made by placing the insertion at position 1. A modified SU containing a clinically relevant ligand instead of GFP or the amphotropic receptor binding domain would be a more appropriate modified SU for such a study.

These interpretations are based on the assumption that neither the fusion of GFP nor the presence of the second-site mutations alters the pathway of infection or creates a new binding domain that changes the cell surface receptor being used for entry. In most cell types, ecotropic MoMLV entry is sensitive to lysosomal inhibitors, presumably due to the buffering effects of these agents on the pH of endosomes. For example, the presence of 100 μM chloroquine or NH₄Cl has been shown to reduce MoMLV infection of NIH 3T3 cells by 90 (14) and 95% (8), respectively. We quantified the effect of chloroquine on GFP-SU Q227R D243Y infection. In two independent titrations, chloroquine reduced infection of GFP-SU Q227R D243Y pseudotype virus by 95 and 97.5%, respectively (Table 1). In these experiments, infection by wild-type envelope pseudotype was reduced by comparable amounts (97.5 and 95%, respectively). These results suggest that the addition of the GFP or of the Q227R and D243Y changes did not alter the path of virus entry.

TABLE 1.

Chloroquine inhibits GFP-SU Q227R D243Y MLV infection of host cells^a

Expt	Titer by endpoint dilution after infection with wild-type Moloney MLV with indicated treatment		% Inhibition	Titer by endpoint dilution after infection with GFP-SU Q227R D243Y MLV with indicated treatment		% Inhibition
	NTb	CHQc		NTb	CHQc
I	2 × 105	5 × 103	97.5	2 × 104	1 × 103	95.0
II	1 × 105	5 × 103	95.0	2 × 104	5 × 102	97.5

^aNIH 3T3 cells were pretreated with 100 μM chloroquine and then exposed to the BAG virus pseudotyped with the appropriate envelope protein for 1 h. Chloroquine treatment was maintained during virus exposure as well as for 3 h after removal of the virus. Control cells were subjected to a similar protocol without the addition of chloroquine. 

^bNT, no treatment (control cells). 

^cCHQ, chloroquine. 

To determine if entry occurred via the ecotropic MoMLV receptor, we examined the receptor interference pattern. A chronically infected NIH 3T3 cell line was established by exposing naive cells to a replication-competent, wild-type ecotropic MoMLV stock called OA virus (11) and culturing for 5 weeks of virus spread. The chronically infected cells (NIH 3T3/OA) were then exposed to dilutions of GFP-SU Q227R D243Y pseudotype virus produced by transient transfection of H1BAG cells. Two independent titrations were done in which titers were calculated from the end point dilution of a 10-fold dilution series and measured in terms of lacZ-transducing units per milliliter. For NIH 3T3 cells, the titers were 1.5 × 10⁶ and 1.0 × 10⁶ U/ml for the first and second titrations, respectively, whereas for NIH 3T3/OA cells, the titers were 5.0 × 10² U/ml for both titrations. Thus, NIH 3T3/OA cells were 3,000- and 2,000-fold less susceptible than naive NIH 3T3 cells to GFP-SU Q227R D243Y virus. This receptor interference or resistance to superinfection indicated that the observed infection occurred via an interaction with the ecotropic MoMLV receptor.

Notably, ecotropic OA interfered 10 times more strongly with virions containing the fusion protein than with wild-type MoMLV virions. NIH 3T3/OA cells were 200-fold less susceptible than naive NIH 3T3 cells to a wild-type MoMLV envelope pseudotype virus transducing the lacZ gene (2.5 × 10¹ versus 5 × 10³ lacZ-transducing units per ml, respectively). Davey et al. (5) used an interference assay in Xenopus oocytes expressing an exogenous ecotropic receptor and mutant envelope proteins of ecotropic Friend 57 MLV as an indirect measure of relative affinity for the receptor. They found that a lower binding affinity correlated with a lower ability to interfere with the binding of wild-type Friend MLV SU (5). Their work suggests that the difference in interference results from a decrease in the binding of GFP-SU Q227R D243Y to the ecotropic receptor.

The ability of cells to produce a transient burst of GFP-SU protein immediately following transfection, but not to maintain expression at detectable steady-state levels suggests that this modified protein is toxic. However, its toxic effects do not explain why GFP-SU pseudotype particles are noninfectious. The small reduction in receptor binding of GFP-SU Q227R D243Y indicates there are subtle structural changes in the envelope protein. Extrapolating these data, we conclude that the fundamental problem with the GFP-SU envelope protein appears to be a folding problem that is too subtle to interfere with precursor cleavage into SU and TM or to weaken the subunit association but that might be sufficient to abolish receptor binding and result in loss of infection. Thus, in the case of the GFP-modified SU, the second-site mutations act by stabilizing envelope protein folding. The improved folding not only would restore envelope protein function but also provides a plausible explanation for the mechanism by which toxicity is alleviated.