Involvement of the C-Terminal Disulfide-Bonded Loop of Murine Leukemia Virus SU Protein in a Postbinding Step Critical for Viral Entry

ABSTRACT A role for the C-terminal domain (CTD) of murine leukemia virus (MuLV) Env protein in viral fusion was indicated by the potent inhibition of MuLV-induced fusion, but not receptor binding, by two rat monoclonal antibodies (MAbs) specific for epitopes in the CTD. Although these two MAbs, 35/56 and 83A25, have very different patterns of reactivity with viral isolates, determinants of both epitopes were mapped to the last C-terminal disulfide-bonded loop of SU (loop 10), and residues in this loop responsible for the different specificities of these MAbs were identified. Both MAbs reacted with a minor fraction of a truncated SU fragment terminating four residues after loop 10, indicating that while the deleted C-terminal residues were not part of these epitopes, they promoted their formation. Neither MAb recognized the loop 10 region expressed in isolated form, suggesting that these epitopes were not completely localized within loop 10 but required additional sequences located N terminal to the loop. Direct support for a role for loop 10 in fusion was provided by the demonstration that Env mutants containing an extra serine or threonine residue between the second and third positions of the loop were highly attenuated for infectivity and defective in fusion assays, despite wild-type levels of expression, processing, and receptor binding. Other mutations at positions 1 to 3 of loop 10 inhibited processing of the gPr80 precursor protein or led to increased shedding of SU, suggesting that loop 10 also affects Env folding and the stability of the interaction between SU and TM.

Entry of retroviruses into host cells requires fusion of the viral envelope with the cellular or endosomal membrane, a process mediated by the envelope glycoprotein complex (Env) (15,39). Env is organized in trimeric complexes of heterodimers consisting of surface (SU) and transmembrane (TM) subunits (17,27). Viral entry is initiated by binding of SU to specific cell surface receptors that, for murine leukemia virus (MuLV), are multitransmembrane proteins that function as transporters (1,5,6,9,10,19). The receptor binding activity of MuLV SU has been localized to the N-terminal 230 amino acids (2,14), and considerable information about this domain was derived from determination of its crystal structure (12). The receptor-binding domain (RBD) is linked via a prolinerich region (PRR) to the C-terminal domain (CTD) of SU, which is joined to TM by a labile disulfide bond (24,28,32,37). Membrane fusion mediated by retroviral Env proteins involves insertion of a hydrophobic, N-terminal region of TM into the target membrane, followed by rearrangement of TM heptad repeats into a six-helix coiled-coil bundle that brings the viral and target membranes into close proximity (38). Recent evidence suggests that following receptor binding, the N terminus of the RBD transmits a signal through the CTD that triggers the fusion activity of TM (3,4,(20)(21)(22). SU and TM in native Env complexes are joined by a metastable disulfide bond (28,33) that involves the CWLC sequence in the CTD that resembles the CXXC motif found in the active site of thiol-exchange enzymes (32). Although little is known about the structures and mechanisms by which the CTD transduces the signal initiated by receptor binding, it has been suggested that isomerization of this disulfide bond may be an essential step in this process (36,37).
We have recently identified two neutralization targets in SU, one in the receptor binding pocket of the RBD, and one involving a small region that includes the most C-terminal disulfide-bonded loop of the CTD (designated loop 10) (7). The two CTD-specific rat monoclonal antibodies (MAbs) used in that study, 35/56 (23) and 83A25 (11), have distinct strain specificities. Both of these MAbs neutralize MuLV infectivity by blocking a step prior to viral fusion but after receptor binding (7), and additional definition of these epitopes and characterization of the structure and function of this region were therefore expected to provide new insight into the functions of the CTD during viral infection. In the present study, residues within loop 10 that determined the specificities of these MAbs were identified, a requirement was demonstrated for additional residues outside of loop 10 for reactivity with these MAbs, and mutations at specific sites in loop 10 were identified that differentially affected MuLV Env processing, virion association, and viral fusion.
Sepharose Fast Flow (Amersham Pharmacia Biotech AB) according to the manufacturer's recommendations. Goat polyclonal sera raised to Rauscher MuLV SU and capsid (CA) were purchased from Quality Biotech, Camden, N.J. The antihemagglutinin (anti-HA) rat MAb 3F10 and the anti-myc murine MAb 9E10 were obtained from Roche. MAbs and sera were detected in immunological assays with the appropriate species-specific immunoglobulin G or sera conjugated to alkaline phosphatase or fluorescein isothiocyanate (Zymed).
Viruses and viral neutralization assays. NIH 3T3 cells chronically infected with MuLVs bearing Env substitution mutations were prepared by transfection of cDNA clones of these MuLVs into NIH 3T3 cells and passage of the cells until infection approached 90% as assayed by immunofluorescence (7,18). These cells were then used as a source of virions for immunofluorescence-based neutralization assays, which were performed as described previously (7,18). Percent neutralization was calculated as follows: [1 Ϫ (percent infection/percent infection of control wells)] ϫ 100. The 50% neutralization dose (ND 50 ) is defined as the concentration of MAb that reduced the number of infected cells by 50% relative to a control infection in the absence of antibodies.
The functional activities of insertion and deletion mutants in Env were quantitated by luciferase assays using luc-expressing virions pseudotyped with mutant Env proteins. These pseudotypes were produced by transfecting 293 cells with a mixture of three plasmids (a retroviral vector plasmid expressing luciferase, an expression plasmid for Friend FB29 MuLV Gag/Pol, and an expression vector for the Env protein) using Fugene reagent (Roche Biochemicals) essentially as described previously (7,35). Forty-eight hours posttransfection, the cell supernatant was harvested and clarified by filtration through a 0.45-m syringe filter and used in infectivity assays versus NIH 3T3 cells. The MuLV pseudotypes were normalized by Western blotting (for CA levels) and by enzyme-linked immunosorbent assay (for SU levels) on pellets of pseudotyped particles. Unless specifically stated, all pseudotypes produced from transfected 293 cells contained similar SU and CA concentrations, and therefore, any phenotypic differences observed in the functional assays accurately reflected differences in Env function.
Construction of loop 10 mutations. Mutations were introduced into the Friend MuLV clone 57 or Friend/AKR 393-426 chimeric Env (7) by QuikChange sitedirected mutagenesis (Stratagene). All mutations were initially introduced into the EcoRV-PstI fragments of the indicated SU sequence in pSP72 as described previously (7). After the presence of the desired mutation was verified by sequencing, the mutated fragments were used to replace the homologous region in a two-long terminal repeat MuLV genomic plasmid, and/or a pcDNA3/1 Zeo(Ϫ) (Invitrogen) expression plasmid for Env (7). During the construction of the NLV mutant, a variant containing an extraneous serine residue adjacent to the natural serine at position 405 in the loop 10 sequence was generated as a PCR artifact. The convention used in this paper is to consider the second serine the inserted residue and to refer to this mutant as NLV S405ϩS. Subsequently, variants of this mutant in which one or both of these serines was converted to threonine were generated by mutagenesis of the NLV S405ϩS Env.
Cell surface-expressed forms of loop 10-related protein fragments were constructed in pDisplay (Invitrogen), which provided a signal peptide and HA epitope N terminal to the loop 10 sequences and a myc epitope and transmembrane domain C terminal to the insert. Gene fragments incorporating flanking SfiI and SalI restriction sites and encoding AKR residues 397 to 420 (in the Friend MuLV-based numbering system used throughout this report), consisting of loop 10 and flanking residues, or residues 397 to 445, consisting of loop 10 and all SU sequence C terminal to it, were generated by PCR and cloned into the same restriction sites in pDisplay. Plasmids for the expression of full-length SU (Env truncated after R445) or SU truncated after E420 were constructed by generating gene fragments containing stop codons following the indicated codons and a HindIII site by PCR on the Friend/AKR 393-426 chimeric Env, and cloning sequenced EcoRV-HindIII fragments into the expression plasmid for the Friend/AKR 393-426 chimeric Env.
Metabolic labeling and immunoassays. MuLV proteins were radiolabeled by exposing cells to labeling medium (cysteine-free DMEM plus 10% dialyzed fetal bovine serum and 50 Ci/ml of [ 35 S]cysteine) for 12 to 14 h as previously described (29); for pulse-chase experiments, labeling medium was added for 0.5 h and then replaced with cold complete DMEM for the period described. Labeled cell culture supernatants were filtered through 0.45-m syringe filters to remove cellular debris before use. Cell-associated samples were collected by washing cells once with phosphate-buffered saline (PBS) and then lysing with radioimmunoprecipitation buffer (29). Radioimmunoprecipitation (29) and Western blotting assays (30) were performed as previously described.
Intersubunit disulfide bond isomerization was induced for wild-type (NLV) and mutant (S405TϩS) virions either by solubilization with 1% NP-40 or treatment with 2 M urea for 30 min at 30°C and inhibited by treating with 20 mM N-ethylmaleimide (NEM) for 1 h at room temperature as described previously (37). The effects of pretreatment with MAb 83A25 (20 g/ml for 1 h at 4°C) on isomerization of wild-type virions and the ability of the S405ϩT mutant to isomerize were examined.
Cell binding and fusion assays. Fluorescence-activated cell sorting (FACS) assays to detect MuLV binding were performed as previously described (7,40). Briefly, 293.mCAT cells were detached with trypsin-EDTA, and a total of 5 ϫ 10 5 cells were incubated with 1 ml of pseudotype MuLV particles at 4°C for 2 h with gentle agitation. Cells were washed with 1 ml of ice-cold 10% fetal bovine serum in PBS (FACS buffer) and resuspended in polyclonal goat anti-Rauscher MuLV SU serum diluted 1:250 in FACS buffer. Cells were incubated for an additional 1 h at 4°C, washed again, and incubated with fluorescein isothiocyanate-conjugated anti-goat immunoglobulin G (Zymed) at 5 g/ml in FACS buffer. After a 0.5-h incubation at 4°C, cells were washed again, fixed in 4% paraformaldehyde in PBS, and analyzed by flow cytometry using a FacsCalibur flow cytometer (Becton Dickinson) and associated software.
Syncytium formation by MuLV pseudotypes on XC cells was assayed as previously described (26). Briefly, XC cell monolayers were incubated with cell culture supernatants containing pseudotyped MuLV particles for 2.5 h of incubation at 37°C. After the cells were washed, fixed with methanol, and stained with cresyl violet, the extent of fusion was quantitated by counting the total number of nuclei present in syncytia containing more than four nuclei under the microscope.

RESULTS
Mapping determinants of the MAb 35/56 and 83A25 epitopes to polymorphic sites in loop 10. The region responsible for the differential reactivity of MAbs 35/56 and 83A25 for AKR and Friend MuLV clone 57 Env was previously mapped to AKR residues 393 to 426 (7). These sequences differ at six residues in this region, five of which are located in the first eight residues of loop 10 ( Fig. 1). In order to determine which of these polymorphisms was involved in the differential expression of these epitopes, we generated a series of Env mutants in which single or multiple Friend MuLV-specific residues were placed into the AKR sequence from this region, using the Fr/AKR(393-426) chimera. Plasmids containing wild-type or mutant proviral DNAs were transfected into NIH 3T3 cells, and the spread of infection was monitored by immunofluorescence. All of the substitution mutants were as infectious as the parental virus. The effects of the substitutions on the MAb 35/56 and 83A25 epitopes were then determined by assaying neutralization titers and immunofluorescence reactivities (Table 1).
Individually changing each residue in this region from the AKR-specific residue to the Friend MuLV-specific residue showed that the reactivity of both MAbs was most sensitive to  The 83A25 MAb has been reported to recognize two MuLV isolates, Moloney ecotropic and 4070A amphotropic viruses, which both contain an N at position 410 (8,11,16). Consistent with this, MAb 83A25 possessed weak reactivity in both neutralization and immunofluorescence assays for these two Env proteins (Table 1), similar to its level of reactivity with the D410N mutant of the Fr/AKR(393-426) chimera. As previously reported (31), MAb 35/56 showed no reactivity with these Env proteins, also consistent with the lack of reactivity of this MAb to the D410N mutant of the Fr/AKR chimera.
Sequences N terminal to loop 10 are required for expression of the MAb 35/56 and 83A25 epitopes. The studies described above map determinants for both epitopes to sites in loop 10 but do not rule out the involvement of additional residues, including positions outside of this region, in their expression. The loss of both epitopes upon reduction of disulfide bonds (7) might indicate a dependence on the disulfide at the base of loop 10 but could also be due to a requirement for disulfides at distal sites in SU. These possibilities were examined by studying the reactivities of these MAbs to several recombinant proteins expressing loop 10, with and without additional gp70derived sequences.
The loop 10 region was expressed in the absence of other SU sequences as a fusion protein in which this loop and a few adjoining residues were flanked by HA and myc epitope tags, and it was expressed as a cell surface protein by linking to a C-terminal transmembrane domain (construct 1 in Fig. 2A). After immunoprecipitation of lysates of cells expressing this construct with an antibody to the myc epitope, a band of the expected size was detected in Western blots by probing with an antibody to the HA epitope tag (Fig. 2B, left panel). The mobility of this band was slightly higher when the sample was analyzed under nonreducing conditions than under reducing conditions, suggesting the presence of the disulfide bond at the base of loop 10. When the anti-myc immunoprecipitate was probed with the 83A25 and 35/56 MAbs, no reactivity was observed (Fig. 2B, right panel), consistent with the lack of immunofluorescent staining of these cells when probed with these MAbs (not shown). These results suggested that loop 10 and the immediate adjoining sequences expressed in this fusion protein were not sufficient for expression of these epitopes. A similar lack of reactivity was observed for a related protein that also contained the additional SU sequences C terminal to loop 10 (not shown), suggesting that residues N terminal to position 397 were required for expression of these epitopes.
Cells expressing either secreted full-length SU or a secreted SU fragment that terminated at position 420, four residues C terminal to loop 10 (constructs 2 and 3 in Fig. 2A), were both positive in immunofluorescent staining assays with the two loop 10-dependent MAbs. The fluorescence intensity of cells expressing the truncated molecule was however lower than that of cells expressing full-length SU (not shown). When the reactivity of these proteins was examined quantitatively by radioimmunoprecipitation, striking differences were observed in the percentage of molecules that expressed these epitopes (Fig. . Two sequential immunoprecipitations performed with the MAbs followed by immunoprecipitation with a polyclonal antiserum showed that essentially all of the full-length SU molecules were recognized by both MAbs, as evidenced by the lack of material in the sample subsequently precipitated by the polyclonal serum (Fig. 2C, upper panel). However, only a small percentage of the truncated molecule was recognized by MAb 35/56 or 83A25; the majority of this protein was found in the final immunoprecipitate with the polyclonal serum (lower panel). This indicated that whereas the additional N-terminal sequences were sufficient for the formation of the loop 10dependent epitopes, these epitopes were formed very inefficiently in the absence of SU sequences C terminal to loop 10, presumably due to aberrant folding. A control MAb, 59C9, directed against a disulfide-dependent conformational epitope located in the RBD of SU (7), precipitated the great majority of molecules for both the full-length and truncated molecules, indicating that the N-terminal domain of the truncated molecule folded correctly and that misfolding of this protein was limited to the CTD. Effects of loop 10 mutations on Env function. The potent neutralizing activity and fusion-blocking activity of MAbs 35/56 and 83A25 suggested a role for loop 10 and/or adjacent regions in a postbinding function of Env. In an attempt to confirm such a role for this domain and to localize the residues involved, mutations in loop 10 were generated and their effects on infectivity determined using luciferase-encoding viral pseudotypes ( Table 2). The general requirement of this loop for Env function was first examined by replacing the 10 central residues of the 12-residue loop of the Fr/AKR(393-426) chimera (positions 405 to 414) with a glycine-alanine-glycine (GAG) tripeptide (⌬ loop 10). This mutant was completely defective in the MuLV/luciferase pseudotype assay (Fig. 3A). An analysis of the levels of SU incorporated into virions by radioimmunoprecipitation of labeled virions showed that this lack of infectivity of ⌬ loop 10 was due to the absence of SU from viral particles (Fig. 3B). The basis of this defect was examined by a pulse-chase analysis (Fig. 3C). After a 30-min pulse, the Env protein of wild-type Fr/AKR(393-426) existed mostly in the form of the gPr80 precursor, with some mature SU also present, while after the 1-h chase period, almost all of the gPr80 was converted to SU, which remained cell associated. In contrast to this, cells producing ⌬ loop 10 Env contained only unprocessed Env precursor even at the latest time points. The absence of any intermediate gPr90 product typically formed upon transport into the Golgi bodies (18) suggested that this mutant Env was defective in transport from the endoplasmic reticulum, presumably due to improper folding.
To identify which region of the loop was responsible for this processing defect, contiguous three-residue sequences in loop 10 were substituted with GAG (Fig. 3A). Whereas the GAG 407-409, 410-412, and 413-15 substitution mutants all induced levels of luciferase activity similar to those induced by the control Env, the GAG 404-406 mutant possessed no activity. The defect of the GAG 404-406 mutant resembled that of the ⌬ loop 10 mutant in that no SU protein was associated with virions ( Fig. 3B) and lysates of transfected cells contained only the unprocessed Env precursor (Fig. 3C). This suggested that the loss of residues 404, 405, and/or 406 accounted for the defectiveness of ⌬ loop 10. To further pinpoint the contribution of these residues to the defectiveness of these mutants, individual substitutions were prepared at each of the first three positions of the loop. 404G is found in other natural sequences, and 404L and 406A are present in the Friend MuLV Env, and thus, it was presumed that these substitutions would retain function. I404A, S405A, and T406G mutants were generated and the resulting pseudovirions analyzed for Env incorporation levels and functional activity ( Table 2 and Fig. 3). The I404A mutant possessed wildtype levels of infectivity and a normal level of SU in virions. The T406G mutant also possessed the normal amount of SU but was slightly attenuated for infectivity, with luciferase levels three-to eightfold below that of the wild-type control. In contrast, the S405A Env was strongly attenuated (Ͼ200-fold less luciferase activity than the control) (Fig. 3A). There was a significantly decreased level of S405A SU present in virions (Fig. 3B), which presumably accounted for the attenuation. While this mutant did undergo proteolytic processing with only slightly lower than normal efficiency, there was little accumulation of mature SU in the cell lysates (Fig. 3C). Additional experiments showed that the processed SU was released from cells much more rapidly than wild-type SU (not shown), consistent with increased shedding of SU due to destabilization of the SU-TM interaction. These results suggested that residue S405 played a role in the stable association between SU and TM and that additive effects of changes at residues 405 and 406 accounted for the defectiveness of the GAG 404-406 mutant.
Effects of loop 10 mutations on epitope expression. The effects of the mutations in loop 10 described above on the expression of the epitopes recognized by MAb 35/56 and 83A25 were examined by immunofluorescence assays on transfected cells and, for infectious mutants, by neutralization assays ( Table 2). The ⌬ loop 10 mutant and the first three GAG substitutions, encompassing residues 404 to 412, were not reactive with either MAb 35/56 or 83A25. This was consistent with the contributions of residues in the region of positions 404 to 411 to these epitopes ( Table 1). The GAG(413-415) substitution mutant retained reactivity with both MAbs and was considerably more sensitive to neutralization by these two MAbs, with ND 50 s 35-to 45-fold lower than that for the wildtype Env. The sensitivity of this mutant to neutralization by the RBD-specific MAb 59C9 was similar to that of wild-type Env, suggesting that the substitution of these residues did not cause a general hypersensitivity to neutralization but rather induced an increased affinity or accessibility specifically for MAbs a Concentration of MAb that reduced infectivity by 50% (g/ml). Ͼ Ͼ10, no neutralization was observed. b MAb reactivity as determined by immunofluorescence (IF). Symbols: ϩ, strong fluorescence; ϩ/Ϫ, weak fluorescence; Ϫ, no fluorescence.
83A25 and 35/56. The S405A mutation possessed the same (for MAb 35/56) or slightly higher (for MAb 83A25) neutralization sensitivity as the wild type, indicating that despite the functional defects of this mutation, it did not affect the structure or accessibility of these epitopes. However, in contrast to the natural I404L polymorphism, the I404A mutation resulted in a significant decrease (three-to eightfold) in neutralization sensitivity to both MAbs, and the T406G substitution resulted in a complete loss of neutralization by both MAbs, again in contrast to the natural T406A polymorphism that had only a small effect on neutralization (Table 1). These results indicated subtle contributions of these positions to the formation of these epitopes.

Effects of insertions in the N-terminal region of loop 10 on Env function.
There is a stretch of three Ser/Thr residues at positions 405 to 407 near the N terminus of AKR loop 10. In the course of producing the NLV variant (Table 1), an extraneous serine residue was serendipitously introduced adjacent to Ser405, resulting in the NLV S405ϩS insertion mutant. This mutant Env was partially attenuated in the luc transduction assay, with a fourfold reduction in activity ( Fig. 3A and Table  3). Pseudoviral particles generated with this Env mutant contained a normal amount of SU (Fig. 3B), indicating that the decreased activity was not due to defects in expression, processing, or association of SU with particles. In contrast to the NLV Env, the NLV S405ϩS mutant was not recognized by MAb 83A25 (Table 3), consistent with the sensitivity of this epitope to other changes near the N terminus of loop 10.
The basis for the defectiveness of this mutant was further examined by measuring the extent of binding of virions pseudotyped with Friend MuLV clone 57, NLV, or NLV S405ϩS Env to 293.mCAT-1 cells by flow cytometry (Fig. 4A, left  panels). Equivalent concentrations of the control and S405ϩS mutant particles displayed roughly equal fluorescent intensities, indicating no apparent disruption in the binding activity of this mutant Env to its receptor. On the other hand, the S405ϩS mutant was severely defective for fusion of XC cells relative to the Friend MuLV wild-type and NLV controls (Fig. 4A, right  panels); no syncytia were observed in wells that received the S405ϩS mutant, while control viral pseudotypes yielded an average of 72 (Ϯ10) (for Friend MuLV clone 57) and 81 (Ϯ11) (for NLV) fused nuclei per grid area. The more apparent defectiveness of this mutant in fusion assays than in infection assays presumably reflects the fact that the level of fusogenicity required for efficient syncytium formation is greater than that required for viral infection, perhaps because syncytium formation requires the fusion of multiple membranes (virions and several cells), whereas infection reflects a single fusion event. These results demonstrate that this insertion in loop 10 blocked a postbinding step required for fusion.
The importance of this region in loop 10 to function was further highlighted by the analysis of several modifications of the NLV S405ϩS insertion mutant. The inserted Ser residue was converted to Thr (S405ϩT), the original Ser at position 405 was converted to Thr (S405TϩS), and both serines were converted to Thr (S405TϩT). A deletion mutant in which S405 was deleted was also generated. All of these mutants lost reactivity with MAb 83A25, further confirming the importance of positions 405 and 406 to the MAb 83A25 epitope. Insertion of Thr instead of Ser slightly increased the level of attenuation, while converting the original Ser to Thr resulted in a highly defective Env, and the mutant with Thr at both positions was completely inactive in the luciferase assay ( Table 3). All of the insertion mutants were also defective in syncytium assays (not shown). Binding assays to cells expressing murine cationic amino acid transporter (MCAT) indicated that even the highly defective mutants retained normal binding activity, and an analysis of the protein content of these virions by Western blotting indicated that the defective Env proteins all underwent normal processing and were incorporated at normal levels into viral particles (Fig. 4B, shown for the most defective S405TϩT mutant). In contrast to these effects for the insertion mutants, the Ser deletion mutant exhibited essentially normal function. Thus, while S405 could be deleted without affecting function, substitution by a threonine together with insertion of a serine or threonine at the adjacent position strongly affected function at a postbinding step, with the level of activity decreasing with increasing size of the substituted and inserted amino acids.
Neither mutation of loop 10 nor its binding by neutralizing MAb prevented denaturation-induced isomerization of the intersubunit disulfide bond. SU and TM are present in virions as a disulfide-bonded heterodimer that dissociates into its subunits by isomerization of the intersubunit disulfide bond upon solubilization of virions or treatment with denaturants (24,32,33,36,37). This isomerization is blocked by prior treatment with the thiol alkylating agent NEM (33) and involves a cysteine present in the CWLC motif at positions 312 to 315 of the CTD domain (32). The possibility that the functional effects seen upon antibody binding or mutagenesis of loop 10 were related to this disulfide bond rearrangement was investigated for the Friend MuLV NLV Env and its fusion-defective derivative bearing the S405TϩS mutation (Fig. 5). As expected, solubilization with NP-40 induced isomerization of essentially all of the NLV SU to free SU, and this isomerization was blocked by pretreatment with NEM. A similar effect was seen upon denaturation with urea, although in this case isomerization was less complete. Prebinding of MAb 83A25 to NLV virions at neutralizing concentrations did not inhibit dissociation to SU under either condition, indicating that binding of this MAb did not interfere with isomerization of the disulfide bond. Similarly, for the defective S405TϩS Env mutant, the SU-TM disulfide-bonded complex was also present after treatment with NEM, and the dissociation of this complex to free SU occurred as efficiently following treatment with either NP-40 or urea for this mutant as for the parental NLV Env. These results did not support a role for interference with isomerization of the intersubunit disulfide bond for either the neutralizing activity of 83A25 or the lack of function of the S405TϩS Env mutant.

DISCUSSION
This study localized determinants of two potent neutralization epitopes to loop 10 in the CTD of MuLV SU and demonstrated effects for this loop on both processing and function of Env. The earlier demonstration that loop 10-specific MAbs inhibited fusion but not receptor binding (7) suggested a role for this region in viral fusion. This conclusion was supported in the present study by the demonstration of attenuated infectivity and severe defects in syncytium-inducing activity of mutants bearing serine or threonine insertions near the N terminus of loop 10, despite normal processing and receptor binding activity of these mutants (Fig. 4). These results suggested that the insertion mutations in the loop interfered with the same function blocked by binding of the loop 10-directed neutralizing MAbs. A role for the first three positions of loop 10 in regulating maturation of the Env precursor was shown by the defective processing of the GAG 404-406 substitution mutant, and an additional role for the second position of the loop in the stable association between SU and TM was suggested by the increased shedding and decreased incorporation of SU into virions for the S405A mutant ( Fig. 3B and C).
The in the complete loss of reactivity with both MAbs, while conversion of Asp410 to Asn resulted in complete loss of reactivity with MAb 35/56 and a significant reduction in the affinity and neutralizing activity of MAb 83A25. An Asn at position 410 introduces a signal for N-linked glycosylation (13,18), and the presence of this glycan may contribute to the inhibition of binding of these MAbs. Earlier studies have reported that MAb 35/56 was highly type specific for endogenous MuLV isolates represented by the AKR strain (23,25,31), while MAb 83A25 was more broadly reactive with MuLVs, including ecotropic Moloney MuLV and amphotropic 4070A MuLV (8,11,16). Asp410 is present in the Akv provirus locus and in endogenous MuLVs produced from expression of this locus, while Asn410 is in the consensus sequence for exogenous MuLV Env proteins. The weak reactivity of MAb 83A25 with the Asn410 substitution mutant in the AKR-derived loop 10 was consistent with its relatively low reactivity with Moloney MuLV and 4070A Env proteins in both immunofluorescence and neutralization assays (Table 1).
Additional positions in the N-terminal half of the loop also contributed to the reactivity of these epitopes but in more subtle ways. The reactivity of both MAbs was completely lost upon replacement of the Thr406 by Gly, while substituting an Ala at this position or mutating the Ile at position 404 to Ala resulted in relatively small increases in ND 50 s for these MAbs (Tables 1 and 2). Mutations of other polymorphic residues in this region (Ser 397 to Asn, Ile404 to Leu, and Ile408 to Val) had little or no effect on either MAb when introduced individually but had a significant impact on the activity of MAb 35/56, but not 83A25, when combined (the NLV substitution in Table  1). Adding the Thr406-to-Ala substitution to the other three changes resulted in complete loss of MAb 35/56 reactivity and strongly attenuated the neutralization activity of MAb 83A25. Thus, the greater strain specificity of MAb 35/56 over 83A25 can be accounted for by its higher sensitivity to the Asn at position 410, along with the greater sensitivity of this MAb to combinations of several other substitutions in the N-terminal region of loop 10.
While these studies localized determinants of both epitopes to loop 10 and accounted for the type specificities of these MAbs, they did not fully define these epitopes. A role for residues outside of loop 10 in the expression of these epitopes was indicated by the lack of immunoreactivity of a fusion protein expressing the isolated loop 10 with either of these MAbs (Fig. 2B), and the localization of these residues to Nterminal sequences was suggested by the efficient reactivity of these MAbs with a fraction of a secreted SU truncation mutant that terminated shortly after loop 10 (Fig. 2C). The absence of these epitopes on the majority of the truncated molecules may be due to an alternate disulfide bond pattern for those molecules. All of the truncated molecules were efficiently precipitated by 59C9, a MAb that reacts with a disulfide bond-dependent epitope located in the N-terminal receptor-binding domain, indicating that the folding defects of this protein were limited to the CTD, and secreted full-length SU was completely precipitated by MAbs 35/56 and 83A25, indicating that improper folding of the CTD was not due to the absence of the TM domain. This suggested that the presence of SU sequences C terminal to loop 10 facilitated proper folding of the CTD.
The critical role of loop 10 in Env function is consistent with a previous mutational analysis of Moloney MuLV Env based on five-residue insertions throughout SU and TM, including seven mutations within loop 10 (34). Mutants containing inserts after the first, second, and third residues of the loop, in the same region as the serine/threonine insertions described in this study, were expressed on the cell surface but were noninfectious, while mutants with insertions in the middle of the loop, after residues 408, 411, and 413, remained infectious. An insertion at the C-terminal end of loop 10 abrogated cell surface expression of Env, presumably because of misfolding. Cell surface binding activity was assayed for two of the N-terminal insertions and was not detected. However, binding was measured by flow cytometry using MAb 83A25, and the results described above suggest that the observed result was due to the loss of the MAb 83A25 epitope, rather than an inability of these mutants to bind to the receptor. How does modification of loop 10 by antibody binding or mutagenesis lead to the inhibition of fusion? Denaturationinduced isomerization of the intersubunit disulfide bond between SU and TM was not inhibited either by MAb 83A25 binding to wild-type Env or by a Thr insertion after position 405 that blocked fusion activity (Fig. 5). This suggested that loop 10 function is either independent of this isomerization or that it occurs subsequent to the rearrangement. However, it is also possible that the normal stimulus for isomerization may depend on loop 10 function but that the denaturation conditions used to experimentally induce isomerization bypass this requirement.
Previous studies demonstrated an interaction between sites near the N terminus of the RBD and the CTD of SU that mediate postattachment steps required for infection by MuLV (3,4,21). For virions bearing amphotropic/ecotropic chimeric Env proteins, an incompatibility was detected between the amphotropic RBD and a region of the ecotropic CTD that encompassed disulfide-bonded loops 8 and 9 of SU, but not loop 10 (20). This incompatibility was attributed to a defect in a putative interaction between the amphotropic RBD and the ecotropic CTD, suggesting a role for loops 8 and 9, but not loop 10, in these interactions. There are several possible ways to reconcile these results with a role for loop 10 in a postattachment step required for fusion. One possibility is that the key CTD structure is a complex of loops 8, 9, and 10 that together interacts with the RBD, mediating a conformational change of Env following receptor binding that is required for membrane fusion. Such a complex may contain the remaining determinants for the epitopes of MAbs 35/56 and 83A25. It is also possible that loop 10 mediates a distinct step required for fusion in addition to the interaction of loops 8 and 9 with the RBD, such as an interaction with the TM subunit that leads to the exposure of the fusion peptide of TM. In either scenario, either the Ser/Thr insertions near the N terminus of loop 10 or the binding of MAbs to this region presumably interferes with the necessary interaction and thereby inhibits fusion. It is also possible that loop 10 may not be directly involved in fusion, but by virtue of its close location to a fusion-critical structure (e.g., loops 8 and 9), changing the size and/or shape of this loop or binding of an antibody to this region indirectly disrupts the function of the proximal structure.
Areas for further study that can help distinguish these possibilities include additional identification of determinants VOL. 79, 2005 C-TERMINAL LOOP OF MuLV SU PROTEIN 7875 within and outside of loop 10 that mediate its effects on fusion, including the identification of regions in SU and/or TM that interact with this region, and more complete determination of the structures of the MAb 35/56 and 83A25 epitopes. A powerful way of obtaining such information would be by determining the crystal structure of complexes between SU and loop 10-directed MAbs. Isolation and characterization of mutants that are resistant to loop 10-directed MAbs and functional revertants of defective loop 10 mutants would also provide useful insights into these questions.