Functional Characterization of the N Termini of Murine Leukemia Virus Envelope Proteins

Abstract

The function of the N terminus of the murine leukemia virus (MuLV) surface (SU) protein was examined. A series of five chimeric envelope proteins (Env) were generated in which the N terminus of amphotropic 4070A was replaced by equivalent sequences from ecotropic Moloney MuLV (M-MuLV). Viral titers of these chimeras indicate that exchange with homologous sequences could be tolerated, up to V17eco/T15ampho (crossover III). Constructs encoding the first 28 amino acids (aa) of ecotropic M-MuLV resulted in Env expression and binding to the receptor; however, the virus titer was reduced 5- to 45-fold, indicating a postbinding block. Additional exchange beyond the first 28 aa of ecotropic MuLV Env resulted in defective protein expression. These N-terminal chimeras were also introduced into the AE4 chimeric Env backbone containing the amphotropic receptor binding domain joined at the hinge region to the ecotropic SU C terminus. In this backbone, introduction of the first 17 aa of the ecotropic Env protein significantly increased the titer compared to that of its parental chimera AE4, implying a functional coordination between the N terminus of SU and the C terminus of the SU and/or transmembrane proteins. These data functionally dissect the N-terminal sequence of the MuLV Env protein and identify differential effects on receptor-mediated entry.

The retrovirus infection cycle begins with the attachment and internalization of the virus through the receptor presented on the surface of susceptible cells. The viral protein that mediates this process is the env gene product, which is glycosylated and proteolytically cleaved by the cellular machinery into the surface (SU) and the transmembrane (TM) proteins (40). The SU-TM heterodimers form trimers (21, 23) within the envelope protein (Env) complex. Following binding, the SU-TM oligomers are proposed to go through major conformational changes that trigger the fusion of the viral and cellular membranes, by analogy to the influenza virus hemagglutinin (HA) protein (28, 29). Five classes of murine leukemia virus (MuLV) (ecotropic, amphotropic, 10A1, polytropic, and xenotropic) have been isolated and classified based on their receptor usage (49, 53). The cDNA clones encoding the cellular receptors of these viruses have been isolated and characterized (1, 8, 37, 38, 57, 63). All of these receptors are multipass transmembrane proteins, suggesting a conserved mechanism of entry.

The SU proteins of MuLVs all share similar structural elements (24, 36, 43), and functional domains within the SU and TM proteins have been characterized. The receptor binding domain (RBD) resides within the N-terminal half of the SU protein (6, 9, 18, 39, 44, 45). The crystal structure of the Friend MuLV RBD has been determined to be an independently folded domain (22). The region of the RBD interacting with the receptor consists of two variable regions, named VRA and VRB (7). Further studies have shown that sequences within VRA are critical for specific interaction with the receptor (5, 27, 55). For amphotropic MuLV, the VRB plays a role in cooperating with VRA for receptor recognition (27, 45). The RBD is linked to a proline-rich region (PRR) and the C-terminal half of the SU protein. The C terminus of the SU protein is involved in the association between the SU and TM proteins (26, 41, 46, 47). The PRR influences receptor choice of polytropic, xenotropic, and 10A1 amphotropic viruses (6, 44), cell-cell fusion (2, 34), and the stability of the protein (26, 61). Despite these functions, the PRR can still tolerate large insertions (33). The TM protein anchors the Env protein complex on the viral membrane, and the ectodomain contains elements for coiled-coil trimerization (23). Key elements for membrane fusion also reside in the TM protein. The N terminus of TM encodes a hydrophobic “fusion peptide” (31), and the R peptide at the TM protein C terminus regulates the fusion process (30, 48, 50).

Despite the knowledge of the function of specific domains, the dynamic intra- and interdomain interactions which transmit the receptor binding signal are not understood. Evidence of these complex interactions is the fusion-defective phenotype of a mutation of histidine 8 (H8) within the extreme N terminus of the ecotropic MuLV SU protein (4). A similar mutation in the 4070A amphotropic MuLV SU protein also resulted in a defective Env protein which was complemented in trans by the addition of a soluble RBD containing this histidine (35). Chimeric MuLV Env proteins exchanging the sequence just before the first cysteine loop between ecotropic and amphotropic MuLVs also resulted in defective viruses (45). Cumulatively, these studies provide evidence for an indispensable role of the SU protein N terminus in the entry process. Disruption of this function by the insertion of ligands might explain the failure of receptor retargeting schemes. MuLVs bearing ligands within the N terminus of either ecotropic or amphotropic MuLVs (reviewed in references 12 and 15) could redirect binding to alternate receptors but frequently fail to mediate successful viral entry.

In order to design a successful retroviral vector targeted to an alternate receptor, it is necessary to identify the key elements in receptor-mediated entry. The available structure of the Friend ecotropic RBD provides insights into some of the essential elements of the MuLV Env protein (22). In that structure, the N terminus of the SU protein prior to the first cysteine loop is spatially distinct from the receptor binding pocket formed by VRA and VRB. The data presented herein functionally dissect the N terminus of the SU protein. The results support the role of the SU protein N terminus in a postbinding stage of the entry process and define cooperation between the N terminus of the SU protein with the C terminus of the SU and/or TM protein.

MATERIALS AND METHODS

Construction of Env expression plasmids.

Constructs were named according to the parental virus, from the N terminus to the C terminus of the protein, and the position of the chimeric junctions. N-terminal junctions were numbered using roman numerals. The higher roman numerals indicate the exchange of larger ecotropic Env sequences into the N terminus of 4070A Env. The junctions for crossover sites II and IV were created through the introduction of a common restriction site in the parental ecotropic and amphotropic sequences through PCR. Crossover II generated an SspI site. The ecotropic sequence was amplified using primer 3807 (5′GATATACATATGGCCGTTAAACAGGGA) and oligonucleotide 7326 (5′GTAATATTATAGACTTGATGAGAC), which is complementary to nucleotides 5890 to 5911 (51) of Moloney MuLV (M-MuLV) RNA and which contained a 5′ tail encoding an SspI site (underlined). The PCR product was subjected to digestion with SfiI and SspI to obtain a 0.5-kb fragment. The amphotropic sequence was amplified with primer 1561 (5′GGCTCCGTCGACTAGAGC) and oligonucleotide 7327 (5′TTAATATTACCTGGAGAGTCACC), which encodes nucleotides 152 to 165 of 4070A (43) and also contains a 5′ tail encoding an SspI site (underlined). The PCR product was digested with SspI and EcoRI, and a 450-bp fragment was obtained. The two fragments were ligated into the vector pNCA-C Am (45), which was digested with SfiI and EcoRI. Crossover IV chimeras were generated in a similar scheme. Oligonucleotide 7724 (5′CACTAGTTGCCCATACCGTCTCCCG), complementary to M-MuLV nucleotides 5937 to 5957 of the RNA, and oligonucleotide 7329 (5′CCACTAGTGTCGTGGGAACTGTACAAG), containing nucleotides 213 to 230 of 4070A, both contain a 5′ tail with an SpeI site (underlined) and both were used with 3807 and 1561 for PCR amplification.

Crossover junctions I, III, and V were generated using overlapping PCR. The oligonucleotide pairs for junction V were 7728 (5′TGGCCACCAGGTCCACAG) and 7727 (5′CTGTGGACCTGGTGGCCAAAATTATATTTTGATC); the oligonucleotide pairs for junction III were 7328 (5′ACCTGGGAGGTCACCAACCTGATGAC) and 7325 (5′GTTGGTGACCTCCCAGGT), and the oligonucleotide pairs for junction I were 7323 (5′CAGCCCCCACCAGGTCT) and 7324 (5′AGACCTGGTGGGGGCTGGAGCCGGGCGAAGC), where the ecotropic sequence is underlined and the amphotropic sequence is in bold. Oligonucleotides 3807 (ecotropic) and 1561 (amphotropic) served as the outside primers. The ecotropic sequence and amphotropic sequence were amplified separately and then pooled together to amplify the chimeric product by annealing to each other. The PCR product was digested with SfiI and EcoRI and ligated into pNCAC-Am digested with SfiI and EcoRI.

All the chimeras that were generated in pNCAC-Am were then subsequently moved into pHIT123 (M-MuLV env [54]) by replacing the 2.1-kb PmlI-NheI fragment. A previously described chimera, AE4, which contains an amphotropic RBD and ecotropic Env C terminus beyond the PRR, is also used as a backbone for exchanging the N terminus (45). After constructing junctions IA to VA in pHIT123 vectors, the C terminus was also replaced to generate the EAE chimeras by replacing the AE4 XhoI-NheI 1.4-kb fragments. Constructs were named based on the chimeric junctions contained within the env gene. For example, IA contains an ecotropic sequence at the N terminus through junction I, followed by an amphotropic sequence, and I4 indicates that this construct contains two sections of ecotropic sequence, one at the N terminus through junction I and one at the ecotropic C-terminal sequence beyond crossover 4. The numbering of the amino acids is based on that of the mature envelope protein after the signal peptide cleavage (43); alanine 34 of the ecotropic protein and alanine 32 of the amphotropic precursor protein are positions 1. The whole env gene of each construct was confirmed by sequencing analysis using the Amplicycle sequencing kit (Perkin-Elmer).

Cell lines.

Canine D17 cells were grown in Dulbecco's modified minimal essential medium (Gibco) supplemented with 7% fetal bovine serum, 200 μM glutamine, and 5 U of penicillin-streptomycin/ml (Gibco) and were kept in a 37°C, 5% CO₂ incubator. TELCeB6 cells (16) were grown in the same medium plus 6 μg of blasticidin S/ml. D17 cells stably expressing gag-pol (D17 gag-pol) were a gift from L. O'Reilly. D17 gag-pol and human 293T cell lines were kept in the same conditions as the D17 cells, with the addition of 400 μg of G418/ml in the medium.

Determination of viral titer.

Aliquots containing 10⁶ TELCeB6 cells were seeded in 60-mm-diameter dishes 1 day prior to transfection. A volume of 10 μg of plasmid DNA containing the env gene of interest was introduced into the TELCeB6 cells by using the CaPO₄ transfection kit (Stratagene). At 12 h posttransfection, sodium butyrate was added to the medium to a final concentration of 10 mM for 10 h. After induction, the medium was replaced with fresh medium lacking sodium butyrate. The supernatants containing virus were collected after 24 h, filtered through a 0.45-μm-pore-size filter, and used to infect 10⁵ D17 cells in 35-mm-diameter gridded plates in the presence of 8 μg of Polybrene/ml. At 48 h postinfection the cells were fixed and the viral titer was determined by X-Gal (5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside) staining as previously described (45). Titers are presented as LacZ infectious units (IU/milliliter).

Surface expression assay and binding assay.

The TELCeB6 cells that were used for the viral titer assays were released from the plate in 0.5 mM EDTA, washed with phosphate-buffered saline, resuspended in 500 μl of media plus 7% fetal bovine serum, and incubated with 10 μl of monoclonal antibody 83A25 (20). After incubation on ice for 1 h, the cells were washed with phosphate-buffered saline, incubated on ice with 250 μl of a 1:100 dilution of goat anti-rat phycoerythrin-conjugated antibody (BIOsource International, Burlingame, Calif.) for 1 h, and fixed in 1% paraformaldehyde. The fluorescence intensity of cell samples was assayed on a fluorescence-activated cell sorter (FACS) (Becton-Coulter, Fullerton, Calif.).

The viral binding assay was performed as previously described (32). Briefly, TELCeB6 cells or D17 gag-pol cells were transfected with env constructs as described above for the titer determination. At 48 h after transfection, the supernatants were collected, filtered, and incubated with 5 × 10⁴ 293T cells on ice for 1 h. Virus binding was scored by immunofluorescence for the presence of the Env protein on the target cell surface, as assayed in the SU protein expression assay.

To correlate virus binding with virus titer, 4070A pseudotyped viruses were harvested from TELCeB6 cells transfected with 10 μg of pHIT 456 (4070A env [54]), as described above. The virus containing supernatant was harvested, diluted 1:4, 2:3, 3:2, and 4:1 with media, and used to infect D17 cells to determine titer. Binding on 293T cells indicated that within this dilution range, the correlation between binding and titer was linear. The 1:4 dilution had binding similar to that of the mock transfection control; however, it still maintained a titer of 2 × 10⁴.

Detection of virus-associated protein.

Volumes containing 10 μg of env expression plasmids were transfected into D17 gag-pol cells using Lipofectamine (Gibco). At 18 h posttransfection, viral particles were labeled with 150 μCi of [³⁵S]Met (Tran³⁵S-label; ICN). Viral supernatants were collected and pelleted through 20% sucrose at 16,000 rpm for 30 min at 4°C in an Eppendorf centrifuge. Virus pellets were resuspended in 50 μl of protein lysis buffer (1% Triton X-100, 0.5% desoxycholate, 0.1% sodium dodecyl sulfate [SDS], 0.01 M sodium phosphate [pH 7.5], 0.1 M NaCl). Viral proteins were separated on SDS–10% polyacrylamide gels and transferred onto polyvinylidene difluoride membranes (Immobilon-P; Millipore). Env protein was detected by SU antibody 80S-019 (1:1,000 dilution; Quality Biotech), washed, and visualized with horseradish peroxidase-conjugated rabbit anti-goat immunoglobulin G (1:100,000; Pierce) in the presence of supersignal substrate (Pierce).

Aliquots of viral pellet from the same preparation were used for detecting virus-associated proteins by immunoprecipitation. In a final volume of 500 μl, 4 μl each of 80S-019 and 79S-842 (anti-SU antibodies; Microbiological Associates, Inc.) and 20 μl of 42-114 (anti-TM antibody [36a]) were used to detect Env proteins.

RESULTS

Generation of chimeric ecotropic/amphotropic envelope protein.

Our laboratory previously reported that two chimeric proviral constructs, AE1 and EA1, which exchanged the sequence immediately upstream of the first cysteine loop of VRA between ecotropic M-MuLV (eco) and amphotropic 4070A (ampho) Env proteins, were not viable (45). In order to more closely examine the interchangeability of the N termini prior to the VRA region, a series of five chimeric crossovers were generated, in which increasing lengths of segments of ecotropic M-MuLV sequence replaced the amphotropic 4070A N-terminal sequence. Sequence comparison of the M-MuLV and the 4070A N-terminal SU region and the chimeric crossovers are shown in Fig. 1. Higher roman numerals indicate increased ecotropic sequence introduced at the N terminus. Crossover points were chosen to examine key differences within the N termini between ecotropic and amphotropic 4070A envelope proteins. Crossover I examines the effect of the size difference between the longer M-MuLV N-terminal sequence (ASPGSSPH) with that of the shorter 4070A sequence (AESPH). Both viral SU proteins contain a critical histidine residue (H8 in M-MuLV and H5 in 4070A) prior to this junction, which has been defined to be required for viral fusion (4). Crossover II, in addition to the changes defined by crossover I, replaced F8 in 4070A with Y11 in M-MuLV. Crossover III was generated in order to observe the effect of the oppositely charged side chain of E16 in M-MuLV versus R13 in 4070A. There was less homology between M-MuLV and 4070A after crossover III. To examine this, two additional chimeras (IV and V) were generated. Crossover IV resulted in the loss of the second N-linked glycosylation site encoded in 4070A.

FIG. 1.

Generation of chimeric MuLV Env N terminus. Ecotropic envelope protein sequences were introduced to replace the equivalent amino acids of the amphotropic Env N terminus. Five chimeric ecotropic-amphotropic (EA) junctions at the N terminus were generated. They were examined in amphotropic 4070A and AE4 backbones, yielding five EA chimeras and five EAE chimeras. The constructs were numbered according to the location of junction points, starting from the N terminus. The alignment of ecotropic and amphotropic N termini is shown in the middle, and the N-linked glycosylation sites are boxed. The sequence of each chimeric N terminus is shown. The ecotropic sequence is underlined. The two boxes at the top illustrate the SU protein; the ecotropic sequence is shaded. The general features of the SU protein are indicated in the open box of the amphotropic part: VRA, VRB, the hinge region (H), and the PRR.

These N-terminal chimeras were also subcloned into a functional AE4 Env protein, which contained an intact RBD from 4070A joined with the C terminus of M-MuLV from the hinge-PRR (45). The parental AE4 was found to be viable, with growth kinetics similar to those of the wild-type 4070A when expressed in a provirus construct. These EAE constructs allow the examination of functional cooperation between the N and C termini of the Env protein.

N-terminal chimeras displayed differential effects in supporting viral entry.

The chimeric env gene was introduced into mammalian expression plasmids and transfected into TELCeB6 cells, expressing gag-pol of M-MuLV and a vector containing a packaging site (Ψ) plus the lacZ marker. Viral particles bearing the chimeric Envs were tested for their infectivity on canine D17 cells (Fig. 2). The titer of the wild-type amphotropic 4070A env expression plasmid (pHIT456) was 7.2 × 10⁴ IU/ml. Figure 2A shows the effect of introducing the M-MuLV N terminus (roman numeral exchanges) into the 4070A backbone (A series). Viruses bearing IA, IIA, and IIIA had titers equal to or better than that of the wild-type 4070A, indicating a conservation of function between M-MuLV and 4070A within the region up to crossover III. However, differential effects of these chimeric junctions on titer were observed. The viral titers conferred by IA and IIA were higher than that of wild-type 4070A by approximately 10-fold, and IIIA had a titer similar to that of the wild type. The titer of chimera IVA (T28eco/S27ampho) was down to only 2% of the wild-type titer, and VA (W38eco/P37ampho) had no detectable titer.

FIG. 2.

Infectivity with N-terminal chimeric envelope proteins. Viruses bearing chimeric Env proteins were examined for their abilities to infect canine D17 cells. Viral supernatants were collected from TELCeB6 cells transiently transfected with the env constructs indicated, diluted with media, and incubated with D17 cells in the presence of Polybrene. Shown are titers of N-terminal chimeras in an amphotropic 4070A Env backbone (A) and equivalent chimeras in an AE4 backbone (B). Titers are presented as LacZ infectious units per milliliter of viral supernatant. The average from a triplicate experiment is shown with the error bar indicating standard deviation. The experiment was performed five times, and a representative set is shown.

The second set of experiments examined the effects of introducing the N-terminal chimeras into the AE4 backbone, which contained the amphotropic RBD followed by the ecotropic C terminus from the hinge-PRR region (Fig. 2B). Although virus bearing the AE4 Env protein was viable within a replication-competent viral backbone, within a vector providing a single round of infection, the efficiency of lacZ transduction was 15-fold lower than that of the wild-type 4070A Env protein. By replacing the N terminus with an ecotropic sequence, chimeras I4 and II4 improved their titers fivefold over that of the parental AE4 chimera. This increase is similar to what was observed with EA chimeras IA and IIA, which yielded higher titers than their backbone sequence 4070A. Surprisingly, the titer of III4 was boosted to levels equivalent to that of the 4070A control, a 15-fold improvement over that of the parental AE4. This increase in the titer of III4 is in contrast to the results obtained with IIIA, which conferred a lower titer than IA and IIA. III4 differed from II4 by only two residues: I13eco/V10ampho and E16eco/R13ampho. The higher titer by III4 over those of I4 and II4 suggests the restoration of an interaction that was absent in either AE4, I4, or II4. Similar to IVA and VA, the titer of IV4 was reduced and the titer of V4 was equivalent to that of the background. This indicates that amphotropic sequences beyond junction III are important and cannot be replaced by the sequence from ecotropic MuLV.

Stable Env expression requires the amphotropic sequence beyond S27.

In order to understand the differences in titer of the N-terminal chimeras, the protein expression levels and incorporation into virions were investigated. First, the surface expression of these chimeric envelope proteins on the producer TELCeB6 cells was examined by FACS analysis (Fig. 3). The SU protein was detected using monoclonal antibody 83A25, which recognizes a common epitope in the SU C terminus shared by both ecotropic and amphotropic MuLVs (20). The profile of the surface protein expression of the chimeras in comparison with mock-transfected cells is shown in Fig. 3.

FIG. 3.

Surface expression of chimeric envelope proteins on TELCeB6 cells. TELCeB6 cells were transiently transfected with vectors expressing chimeric Env proteins. The names of individual constructs are in the right upper corner of each panel. Each gray line represents the level of fluorescence of mock-transfected cells. The expression of each construct is shown by a black line. The x axis is the fluorescence intensity detected for cell surface-associated Env protein. The y axis represents the cell number counted.

Chimeras IA, IIA, IIIA, and IVA had cell surface expression levels similar to that of wild-type 4070A, suggesting that the shorter exchanges at the N terminus did not disrupt the assembly, transport, or stability of the SU-TM protein complex on the producer cell surface. The differences between the titers transduced by these chimeras are therefore due to a functional alteration. The slightly increased expression of IA was observed in only one of five assays. In contrast, chimera VA yielded a lower, broad, more heterogeneous peak, indicating that the protein expressed on the cell surface was not stable.

Surface expression of N-terminal chimeras I4, II4, III4, and IV4 was similar to the expression level of the parental AE4 chimera. The detected fluorescence on the cell surface of chimera AE4 was generally lower than that of the 4070A control. This apparent decrease in the surface expression of SU protein could have resulted if the monoclonal antibody was biased towards recognizing the amphotropic over the ecotropic SU protein C terminus. Similar to VA, V4 showed a lower and heterogeneous pattern of expression, reflecting the loss of viral titer.

The association of the SU proteins on viral particles was also examined. Virions were collected from D17 gag-pol-expressing cells transiently transfected with individual env genes. After partial purification through a sucrose cushion, the Env proteins were analyzed by a Western blotting (Fig. 4) or by immunoprecipitation (data not shown). The SU proteins of N-terminal chimeras with junctions I, II, III, and IV constructed in either 4070A or AE4 backbones were readily detectable on viral particles. The levels of TM proteins paralleled SU expression. Substitution of up to 26 N-terminal amino acids of the amphotropic SU sequence with ecotropic sequence did not affect the global conformation of the Env protein and allowed for incorporation into viral particles. The variations in viral titers (Fig. 2) could not be simply correlated with their surface expression (Fig. 3) or viral particle association (Fig. 4). In contrast, neither chimera with junction V (VA or V4) resulted in stable association with viral particles. This corresponded with the heterogeneous profile observed with the SU protein surface expression (Fig. 3). In addition, TM protein could not be immunoprecipitated from VA and V4 viral preparations. N-terminal exchanges through junction V altered the association of the SU-TM protein complex on virions, reflected in the loss of the viral titer.

FIG. 4.

Western blot analysis of virus-associated SU proteins. Env proteins were expressed and metabolically labeled in D17 gag-pol cells. Viruses were isolated, and virus-associated SU proteins were analyzed by Western blotting as described in Materials and Methods. Proteins were separated on SDS–10% polyacrylamide gels and probed with anti-SU antibody 80S-019. The individual constructs are indicated above each lane. Mock transfection served as an Env⁻ control for the D17 gag-pol cells.

The N-terminal residues are involved in the postbinding process of entry.

Receptor determinants of 4070A Env have been mapped to the VRA and VRB regions (5, 27). However, the introduction of exogenous sequences outside of this region could alter the protein folding and thus alter the conformation of the receptor binding face. This is of particular interest for the IVA and IV4 chimeras, which displayed a wild-type level of surface expression yet conferred 45- and 5-fold, respectively, decreases in viral titer. The N-terminal chimeras were examined for their ability to bind acceptor cells. Viral binding studies were performed on D17 cells, the host cells used to determine the titer of the chimeras, as well as on 293T cells. Viruses used in the binding assay were collected from a transient transfection of chimeric env constructs into D17 gag-pol cells or TELCeB6 cells. Similar results were obtained independent of the source of the virus producer cell or the target cell. Figure 5 shows a representative result of the binding experiment. Strong receptor binding was detected for chimeras IA, IIA, IIIA, and IVA. Although IA and IIA displayed receptor binding equivalent to that of the wild-type 4070A Env protein, their titers were almost 10-fold higher than that of 4070A. The average binding level on acceptor cells (shown as the mean channel fluorescence) of IIIA was 1.1 channel numbers higher than that of the 4070A, whereas IVA was 1.1 channel numbers lower. With a titration of 4070A virus, a similar binding decrease of 1.1 channel numbers maintained approximately 70% of the titer (data not shown). Therefore, it is not believed that the decrease in the mean channel fluorescence for IVA can account for the loss of 98% of the viral titer. Chimera VA had no detectable binding, as predicted due to its low level of surface expression.

FIG. 5.

Binding of chimeric envelope proteins on target cells. Viral supernatants were collected from D17 gag-pol cells transiently transfected with chimeric env expression vectors, and the binding was tested on 293T cells. The level of binding was analyzed by flow cytometry as described for the surface expression assay. The basal level of fluorescence determined for each mock-transfected cell supernatant for binding is shown by a gray line. Binding intensities of each Env construct are shown by black lines. The mean channel number of the binding is shown at the upper right corner of each panel.

All N-terminal chimeras within the AE4 backbone had less binding than those within 4070A, possibly due to the variation in the 83A25 antibody recognition. The receptor binding of chimeras I4 and II4 was comparable to that of the parental backbone, AE4. The similar binding level suggests that the increase in titer by these chimeras over that of AE4 is binding independent. Similar to IIIA, III4 also displayed slightly improved binding. However, such an increase in binding probably could not account for the 10-fold increase in the titer of III4 over that of the parental AE4. IV4 had the same level of binding as parental AE4; however, the titer by IV4 was about fivefold lower than that of AE4. V4 failed to bind the receptor due to the defect in protein expression on the cell surface.

Summary.

Table 1 summarizes the results of the Env protein expression, binding, and titer studies. Exchange of chimeric junction I or II into either a 4070A or AE4 backbone did not alter the binding of Env to the receptor but yielded increases in viral titers over those of the parental constructs. Replacement of the N-terminal 9 amino acids of amphotropic SU protein with the N-terminal 12 amino acids of ecotropic M-MuLV SU protein therefore improves the efficiency of viral entry through the canine D17 amphotropic receptor. Interestingly, viruses bearing the IIIA and III4 Env proteins both showed slightly improved virus binding over their parental backbone constructs. However, the titer of IIIA was equivalent to that of wild-type 4070A, whereas the titer of III4 was greater than that of its parental AE4 Env protein. Chimeras IIIA and III4 had different sources for the C termini of the SU and TM proteins. This differential effect on the titer of IIIA versus that of III4 implies the existence of a functional interaction between the N and C termini of SU or TM proteins. The titers of IVA and IV4 were both significantly lower than those of their parental constructs. These decreases in titer did not correlate with a similar decrease in binding and suggest a loss of function at a postbinding step in viral entry. Further exchanging the N terminus to junction V resulted in a defect in Env cell surface expression. The N-terminal SU sequences are thus involved in different aspects of the function of the Env protein.

TABLE 1.

Summary of assays of chimeric Env proteins

Env protein	Surface expressiona	Bindingb	Titerc
4070A	++	4.90	7.2 × 104
IA	++	4.81	7.5 × 105
IIA	++	4.83	5.2 × 105
IIIA	++	6.01	7.1 × 104
IVA	++	3.81	1.6 × 103
VA	+/−	0.67	0
AE4	++	2.00	3.4 × 103
I4	++	2.03	1.7 × 104
II4	++	1.77	1.7 × 104
III4	++	2.90	7.2 × 104
IV4	++	2.02	7.0 × 102
V4	+/−	1.20	0

^aCell surface expression was determined by FACS analysis of TELCeB6 cells transfected with env expression vector. ++, wild-type level of Env expression on producer cells; +/−, low to trace amounts of detectable Env. 

^bMean channel number of 293T cell-conjugated fluorescence determined by FACS analysis. Tested cells were bound with Env-pseudotyped virus. 

^cTiter of Env-pseudotyped, lacZ-containing virus on D17 cells. An average of a triplicate experiment is shown. 

DISCUSSION

Entry of an enveloped virus is a dynamic process initiated upon receptor binding which ultimately yields the mixing of the viral and host cell membranes. Receptor binding triggers a series of structural rearrangements exposing the fusion apparatus. The post-receptor binding conformational changes require the alteration of a large number of intramolecular and intermolecular interactions within the retroviral Env proteins with spatial and temporal precision. It is critical to understand the steps required for functional retroviral entry, both by means of inhibiting the viruses as well as for the development of retroviral vectors. In order to define functional interactions within the large SU-TM protein complex, we have generated chimeras between two related Env proteins. This approach allows the inspection of the function of specific domains without making gross changes in the structure of the protein.

The N termini of the MuLV SU proteins are moderately homologous. Prior to the first cysteine loop, 20 out of the 44 amino acids are identical. The N-terminal region of the SU protein has been implicated in a postbinding step. A conserved histidine (H8 of M-MuLV, H5 of 4070A) was reported to play a critical role in the fusion process (4). All of our chimeric constructs maintained a histidine residue. The N terminus of the SU protein is frequently selected as a site for ligand insertion in order to retarget retroviral vectors. Although a linker-insertion between positions 3 and 4 in M-MuLV resulted in a viable virus (26), insertion of cell-targeting ligands at the N terminus has proven more difficult (14, 62, 65) and requires manipulation of a spacer region (58). These chimeric envelope proteins can bind to both the retroviral receptor and the alternative receptors. However, entry through the alternative receptor cannot be completed. These observations support the role of the N-terminal SU protein sequence in the postbinding entry process.

In order to finely map the critical residues within the SU protein N terminus, a series of five junction points were generated. The N terminus of the amphotropic 4070A RBD was replaced with an equivalent sequence from the ecotropic M-MuLV RBD. For comparison, this was performed within the wild-type 4070A backbone as well as a chimeric AE4 construct. Several key observations were obtained. First, the amphotropic 4070A entry was not compromised with exchange through ecoV17/amphoV14 (crossover III). Second, the sequence beyond T28 in ecotropic Env and T26 (crossover V) in 4070A Env is important for wild-type levels of protein expression. Third, the sequence through T26 (crossover IV) is required for functional entry without affecting receptor binding. Thus, a portion of the N terminus appears to play an important role in an early postbinding stage of entry, transmitting the binding signal to the rest of the Env protein. Most interestingly, the increased titer of chimera III4 over that of its parental AE4 chimera suggests a functional cooperation between the N terminus and the C-terminal part of the Env protein.

Figure 6 shows a molecular model of the amphotropic Env RBD based on the available Friend MuLV RBD structure, which was built by the X look program (42). The N terminus is located beneath the receptor binding face formed by VRA and VRB, at the base of the structure. This distinct spatial location of VRA-VRB explains why the binding of most of the N-terminal chimeras was not altered. The crossover junctions within the N terminus are individually colored. The sequence before junction I, and thus that of the conserved H5, is not within the structure. Chimeras to junction I (IA and I4) have increased the titer over those of the parental 4070A and AE4, respectively, without affecting binding to cells expressing the amphotropic receptor. The ecotropic N-terminal peptide to junction I is larger than the amphotropic peptide and may differentially display the conserved histidine to interact with the fusion machinery within the Env protein complex. Minimally, the exchange to either junction I or II did not deregulate the fusion process and no syncytia were observed in an XC cocultivation assay with any of our chimeric Env expression cells (data not shown). The model initiates with the two amino acids prior to junction II. These residues are shown in red and form the N-terminal half of the first β-strand. Chimeras through crossover II also contain titers higher than those of their backbone Envs. Junction I and junction II differ in a conservative change from F8 in 4070A to Y11 in the ecotropic sequence. This change may have subtle effects; however, in general, chimeras with junctions I and II behave similarly. The increased titer of amphotropic virus bearing the ecotropic N terminus to junction I or II has a direct application for retroviral vectors.

FIG. 6.

N-terminal sequences on a molecular model of the amphotropic Env RBD. The proposed RBD structure was generated using the X look program, which is based on the X-ray crystal structure of the Friend MuLV RBD (22). VRA and VRB are shown in cyan and blue, respectively, to illustrate the spatial correlation with the N terminus. The sequence before crossover I (A1-H5) is not within the structure. Y8 and N9 within crossover II are shown in red. V10 to V14, between junctions II and III, are green. The region between III and IV (T15-T26) is orange. Sequences beyond crossover IV and before crossover V are yellow. Amino acids between junction III and junction IV form a loop adjacent to VRA. The amino acids that could mediate contacts between VRA and this loop are labeled.

Junction III is at the end of the first β-strand. Exchanging an ecotropic sequence up to junction III is tolerated in the amphotropic Env protein, suggesting that the function mediated by these sequences is conserved between the two strains of virus. The mechanism of the increase in binding of IIIA and III4 remains to be elucidated. The sequence between junction II and junction III is mostly buried within the β-barrel at the base of the structure. In fact, the only surface-accessible side chain is oppositely charged between 4070A and M-MuLV (E16 in M-MuLV, R13 in 4070A) and may be critical in the differential function of the chimeras bearing this junction. In the amphotropic backbone, the titer of IIIA is not as high as those of IA and IIA. In contrast, when junction III is in the AE4 Env backbone, which contains the ecotropic sequence beyond PRR, the titer of III4 is greater than those of the parental AE4, I4, or II4 construct. This indicates that the charged E16 side chain of the ecotropic N-terminal SU protein might coordinate a critical interaction with the C terminus of either its cognate SU protein or TM protein, which is required for a post-receptor-binding step in the entry process. This potential interaction would be lost in IIIA and regained in III4. On the modeled structure, the E16- or R13-charged side chain is exposed at the center of a surface which can be in contact with another domain.

Sequences between junction III and junction IV form a loop right under VRA (Fig. 6, orange). It has been shown that the amphotropic sequence between E50-G64 is essential for recognizing the amphotropic receptor (5). Within this motif, two amino acids, Y60 and V61, have been shown to be critical for receptor recognition (55, 56). On the modeled structure, the region composed of the first disulfide loop of VRA is in extremely close contact with the loop formed by the sequence between junctions III and IV and could be directly affected by receptor binding. There is low homology between the ecotropic and amphotropic MuLV sequences within the region between junctions III and IV, including the absence of an N-linked glycosylation site in M-MuLV. The lowered titers conferred by chimeras bearing junctions IV may be caused by the structural incompatibility between this region and VRA. The region between junctions III and IV can, therefore, be directly involved in the pathway transmitting the receptor binding signal to the rest of the SU-TM protein complex. Alternatively, the lower titer may be due to the loss of the second glycosylation site. In the absence of a 3-dimensional structure of the entire SU-TM protein complex, the intra- and/or intermolecular interaction mediated by this motif can only be speculated. Further studies to identify second-site revertants (42, 64) which compensate junction IV chimeras are under way and have high significance.

The amino acids between junctions IV and V are found on the opposite side of the structure shown in Fig. 6, at the base of the RBD forming a loop between the second β-strand and VRA (Fig. 6, yellow). As shown in Fig. 1, there is almost no homology between M-MuLV and 4070A within this region. Exchanges through junction V yielded a lower, heterogeneous level of SU protein on the cell surface. Virus produced from these cells lacked TM protein, failed to bind to receptor, and had no detectable titer. Further investigation into the function of this loop can only be investigated by more limited exchanges within the IV-V region. Oligomerization is an essential step in the intracellular transport and processing of viral proteins (19, 25). The distinct location of the junction IV-V loop in the amphotropic RBD molecular model could alter protein multimerization. Previously, a single chimeric MuLV isolate which exchanged an N-terminal fragment of M-MuLV larger than junction V into an amphotropic chimeric envelope was reported to be viable (45). In light of the current studies, the nature of this infectious virus (EAE1/6) was reexamined. Sequence analysis of this construct indicated that the viable virus corresponded with the parental AE6. This result is now consistent with the functional chimeric junctions defined in the present study.

In the absence of the structure of the MuLV SU-TM protein complex, the influenza virus surface HA proteins serve as a model system to predict the function of the N-terminal SU residues (reviewed in references 52 and 59). In the native conformation, the N terminus of the HA1 protein of influenza virus is distal to the RBD and forms a β-strand buried within the HA2 protein (60). This first β-strand ends with a histidine (H17) which forms hydrogen bonds with the HA2 N-terminal residues. Interestingly, mutations of this histidine raise the pH required to trigger the fusion-competent conformation (17, 59). If MuLV is homologous to HA, then the N-terminal residues near H8 or H5 of the MuLV Env protein may form a stable complex with either the TM protein or the C terminus of the SU protein. The N termini replaced with the longer ecotropic sequence in chimeras I and II might present this histidine differently, forming a more stable complex than the wild-type protein, and thus release more free energy from the metastable conformation upon receptor binding (13). During low-pH-induced conformation rearrangement, the first β-strand of HA dissociates from the N terminus of HA2, leaving the histidine accessible (11). For HA-mediated entry, the density of HA is important for the formation of the fusion pores (10). In both MuLV and feline leukemia virus (FeLV), soluble envelope protein fragments containing the histidine at the N terminus can assist the entry of virus-bearing defective Env proteins lacking this histidine (3, 35). It is possible that the soluble Env protein fragment can assist in the complex formation through the histidine in order to form the functional fusion pore.

The results of these studies provided a fine map of the N-terminal domain of the M-MuLV SU protein. Five chimeric junctions were created within a 39-amino-acid segment. Interestingly, four different phenotypes were identified: increased titers with junctions I and II, cognate recognition of the N terminus with either the C terminus of SU or TM protein (junction III), a post-receptor binding defect (junction IV), and an intracellular transport defect (junction V). The ability to successfully attach alternative ligands to this domain, therefore, requires careful maintenance of its structure and function. These differential effects observed between homologous MuLVs point to the importance of this region in maintaining the native state of the viral envelope protein complex.