Evidence that the transmembrane domain proximal region of the human T-cell leukemia virus type 1 fusion glycoprotein gp21 has distinct roles in the prefusion and fusion-activated states.

To investigate the structural context of the fusion peptide region in human T-cell leukemia virus type 1 gp21, maltose-binding protein (MBP) was used as an N-terminal solubilization partner for the entire gp21 ectodomain (residues 313-445) and C-terminally truncated ectodomain fragments. The bacterial expression of the MBP/gp21 chimeras resulted in soluble trimers containing intramonomer disulfide bonds. Detergents blocked the proteolytic cleavage of fusion peptide residues in the MBP/gp21-(313-425) chimera, indicating that the fusion peptide is available for interaction with detergent despite the presence of an N-terminal MBP domain. Limited proteolysis experiments indicated that the transmembrane domain proximal sequence Thr(425)-Ala(439) protects fusion peptide residues from chymotrypsin. MBP/gp21 chimera stability therefore depends on a functional interaction between N-terminal and transmembrane domain proximal regions in a gp21 helical hairpin structure. In addition, thermal aggregation experiments indicated that the Thr(425)-Ser(436) sequence confers stability to the fusion peptide-containing MBP/gp21 chimeras. The functional role of the transmembrane domain proximal sequence was assessed by alanine-scanning mutagenesis of the full-length envelope glycoprotein, with 11 of 12 single alanine substitutions resulting in 1.5- to 4.5-fold enhancements in cell-cell fusion activity. By contrast, single alanine substitutions in MBP/gp21 did not significantly alter chimera stability, indicating that multiple residues within the transmembrane domain proximal region and the fusion peptide and adjacent glycine-rich segment contribute to stability, thereby mitigating the potential effects of the substitutions. The fusion-enhancing effects of the substitutions are therefore likely to be caused by alteration of the prefusion complex. Our observations suggest that the function of the transmembrane domain proximal sequence in the prefusion envelope glycoprotein is distinct from its role in stabilizing the fusion peptide region in the fusion-activated helical hairpin conformation of gp21.

Retroviral envelope glycoproteins (Env) 1 are synthesized as inactive precursors that are processed in the Golgi apparatus to yield a functional hetero-oligomeric complex. The Env complex is composed of a surface-exposed subunit (SU) that mediates attachment to cellular receptors, triggering the membrane fusion activity of the noncovalently associated transmembrane (TM) protein, leading to viral entry. Human T-cell leukemia virus type 1 (HTLV-1) is a retrovirus that is associated with various diseases including adult T-cell leukemia/lymphoma, HTLV-1-associated myelopathy/tropical spastic paraparesis, uveitis, and infectious dermatitis of children (1). HTLV-1 transmission occurs mainly via fusion between infected Env-expressing cells and receptor-bearing cells, because infection by cell-free HTLV-1 is inefficient in vitro and in vivo (2)(3)(4)(5). The receptor(s) recognized by the HTLV-1 SU-TM protein (gp46-gp21) complex are yet to be identified; however, the results of monoclonal antibody blocking studies and protein expression studies suggest that HTLV-1 fusion depends on multiple cellular factors (6 -11).
The ectodomain of HTLV-1 TM protein (gp21) contains an N-terminally located fusion peptide, an ϳ15-residue hydrophobic sequence that inserts into target cellular membranes and is critical for membrane fusion activity. The fusion peptide is connected through a glycine-rich segment to a coiled-coil forming oligomerization domain that forms the core of the gp21 trimer. The gp21 ectodomain is anchored to the viral envelope via an ϳ20-residue transmembrane domain (TMD) (Fig. 1). Our previous crystallographic study of a gp21 core fragment, gp21-(338 -425), revealed a trimeric helical hairpin structure comprising a central coiled-coil, a disulfide-bonded loop that stabilizes a chain reversal, and an antiparallel C-terminal segment packed on the outside of the coiled-coil (12). gp21-(338 -425) is similar in overall architecture to TM protein core fragments derived from other retroviruses (13)(14)(15)(16)(17), a filovirus (18,19), paramyxoviruses (20), and the low pH fusogenic conformation of influenza virus (orthomyxovirus) HA 2 (21,22). The three-dimensional structures of influenza virus HA 2 in the prefusion (23,24) and low pH fusogenic states (21,22) have illustrated the conformational changes associated with fusion activity and provide a paradigm for studying retroviral fusion. The structural changes accompanying HA 2 fusion-activation include helical extension at the N terminus of the central coiled-coil and refolding of the C-terminal portion of the ectodo-main such that TMD proximal sequences pack on the outside of the coiled-coil in a hairpin conformation. This helical hairpin architecture implies that the TMD is close to the N terminus of fusion-activated TM proteins. The currently available threedimensional structures of viral TM proteins lack the N-terminal fusion peptide region.
A striking feature of the low pH-induced HA 2 conformational change is a loop-to-helix transition that extends the N terminus of the central coiled-coil by ϳ100 Å (21,25). This helical extension would translocate the fusion peptide from the HA 2 core to the tip of the helical hairpin rod for membrane insertion. The location of fusion peptides in prefusogenic retroviral Env is not known; however, in the helical hairpin they are also likely to be displayed at the coiled-coil N terminus. An early stage of membrane fusion is thought to involve the insertion of fusion peptides into the outer leaflet of the target membrane (26,27). Spectroscopic studies of the interactions between water-soluble model HA 2 fusion peptides and hydrated lipid bilayers indicated that membrane insertion correlates with a change in fusion peptide structure from random coil to ␣-helix (28 -30). At the pH of fusion residues 2-10 of the HA 2 fusion peptide insert obliquely into the outer bilayer leaflet as an ␣-helix with residues 11-14 mediating a turn at the membrane-solution interface and residues 15-18 re-entering the hydrophobic phase as a 3 10 helix (31). A similar mode of fusion peptide insertion was also observed for a trimeric N-terminal HA 2 fragment, where fusion peptide residues 1-10 enter the outer bilayer leaflet as monomeric ␣-helices at an oblique angle (32). Similar ␣-helical content, oblique mode of insertion, and insertion depths have been observed with retroviral fusion peptides suggesting a conserved mechanism of membrane insertion (28,29,(33)(34)(35). The monomeric insertion of fusion peptides into membranes may be facilitated by N-capping structures that terminate the HA 2 and gp21 central coiled-coils, directing N-terminal residues to extend away from the 3-fold symmetry axis (22,36). The oblique insertion of fusion peptides into membranes is considered to expand the center of the target bilayer, introducing negative curvature strain and instability to the site of insertion (37,38). Refolding of the ectodomain into a hairpin then presumably draws the TMDs and associated viral envelope toward the site of fusion peptide insertion in the target membrane. The refolding process is considered to provide free energy to help destabilize the closely apposed viral and target membranes for fusion (22, 39 -42).
To further understand the structural context of the HTLV-1 gp21 fusion peptide region, we biochemically characterized MBP/gp21 ectodomain chimeras containing the fusion peptide. The bacterially expressed chimeras form soluble trimers with intact intramonomer disulfide bonds. Limited proteolysis and thermal aggregation experiments indicated that the gp21 TMD proximal sequence (Thr 425 -Ala 439 ) protects fusion peptide residues from proteolysis and confers stability to the fusion peptide-containing MBP/gp21 chimeras. Alanine-scanning mutagenesis of the Thr 425 -Ser 436 sequence in HTLV-1 Env resulted in enhanced fusion activity in 11 of 12 alanine mutants. The enhanced fusogenicity of Env mutants is likely due to effects on prefusion Env because alanine substitutions did not significantly affect MBP/gp21 chimera stability. The function of the TMD proximal sequence in the prefusion Env complex may be distinct from its role in stabilizing the fusion peptide and adjacent glycine-rich region in the fusion-activated gp21 helical hairpin.

EXPERIMENTAL PROCEDURES
MBP/gp21 Escherichia coli Expression Vectors-A modified version of the MBP/gp21 expression vector pMBPLϪ/gp21-(338 -425) (43) was used for expression of fusion peptide-containing MBP/gp21 chimeras. The PvuII site that links MBP and gp21 moieties was replaced with a unique NotI site. PCR was used to generate a series of gp21 ectodomain fragments encoding the gp21 amino acids Ala 313 to Asn 421 , Thr 425 , Asn 430 , Leu 433 , Ser 436 , Ala 439 , or Thr 445 using pCELT.1 as the template (36). The forward primer, 5Ј-CCCCGCTCCGCGGCCGCGGTACCGG-TGGCGGTC, incorporates the N-terminal gp21 amino acid, Ala 313 (underlined), into a NotI restriction site (bold). Reverse primers incorporated a stop codon and a PstI site. PCR fragments were ligated into the modified pMBPLϪ/gp21-(338 -425) vector through NotI-PstI sites. In these vectors, MBP is fused to the gp21 N-terminal Ala 313 via an NAA linker.
Limited Proteolysis of MBP/gp21 Chimeras-MBP/gp21 trimers at 0.5 mg/ml in 10 mM Tris (pH 8.5), 30 mM sodium chloride, and 0.1 mM EDTA (250 g of protein) were proteolyzed with sequencing grade chymotrypsin (Roche Molecular Biochemicals) using a 1:150 ratio of protease to protein (w/w). Proteolysis was performed for 1, 5, 10, 30, and 60 min at 37°C. Aliquots were taken at each time point and quenched by the addition of 0.1% trifluoroacetic acid (v/v) for mass spectrometry (25 g of protein) or quenched by boiling for 5 min in SDS-PAGE sample buffer containing 1% ␤-mercaptoethanol for SDS-PAGE in 10 -17% gradient gels (5 g of protein). For detergent binding assays, MBP/ gp21-(313-425) trimer (0.5 mg/ml) was preincubated for 16 h at room FIG. 1. Schematic representation of MBP/gp21. MBP is linked via an NAA linker to the gp21 N-terminal residue, Ala 313 . The fusion peptide residues, Ala 313 -Leu 324 , which are predicted to insert into membranes (28,29) are in bold, and the adjacent glycine-rich segment, Ala 325 -Ser 337 , is underlined. The crystal structure of a monomer of the gp21 helical hairpin core region, Met 338 -Asn 421 , shows an N-terminal ␣-helix (Leu 340 -Leu 385 ) that participates in the formation of the central coiled-coil in the MBP/gp21 trimer, the disulfide-bonded loop (Cys 393 -Cys 400 ; Cys 401 is unpaired), and the anti-parallel C-terminal segment (Cys 401 -Asn 421 ); the Arg 422 -Thr 425 segment is disordered (Protein Data Bank accession number 1MG1, Ref. 12). TMD proximal residues, Arg 422 -Thr 445 , which are absent from the gp21 crystal structure are shown in single letter code. MBP/gp21 chimeras were terminated at Thr 425 , Asn 430 , Leu 433 , Ser 436 , Ala 439 , or Thr 445 (underlined and indicated by residue numbers). The transmembrane domain (TMD, residues 446 -464) and cytoplasmic tail (C-tail, residues 465-488) are also shown. The three major chymotrypsin sites are indicated by an arrow and labeled Chy. Residues that were substituted with alanine in HTLV-1 Env are indicated by asterisks.
Assessment of Thermostability of MBP/gp21 Chimeras-The purified MBP/gp21 trimers were exchanged into a low ionic strength buffer (50 mM sodium chloride, 50 mM glycine, pH 8.3) using Superdex 200 (PC3.2/30) gel filtration. The thermostability of MBP/gp21 chimeras was monitored by a thermal aggregation assay. Briefly, MBP/gp21 trimers were heat-treated (temperature range: 37-52°C, 5 min) and then cooled on ice prior to analytical Superdex 200 (PC 3.2/30) gel filtration in the same buffer. The maximum temperature at which Ͼ95% trimeric structure was maintained (T MAX.TRI ), and the minimum temperature required for conversion of Ͼ95% of trimer to soluble aggregate (T MIN.AGG ) was identified for each chimera.
Mass Spectrometry-Purified MBP/gp21 trimers or chymotrypsintreated MBP/gp21 samples were desalted and concentrated by precipitation in methanol/chloroform and infused directly into a PE Sciex API IIIϩ mass spectrometer in 15% (v/v) acetic acid, 50% (v/v) acetonitrile using a nanoelectrospray ion source (44). The ion spectrum was visualized with Tune 2.5-FPU software and deconvoluted using the Hypermass facility in MacSpec 3.3 (PE Sciex). The redox state of cysteine residues was identified by mass spectrometry analysis of chimeras treated with the alkylating agent 4-vinylpyridine (4-VP; Sigma) as described previously (43).
Cell Lines and Viruses-293T and HeLa cells were maintained in Dulbecco's modification of minimal essential medium with 10% fetal calf serum and transfected using Fugene-6 according to the manufacturer's specifications (Roche Molecular Biochemicals). The recombinant vaccinia virus vTF7.3, which drives expression of bacteriophage T7 polymerase, was obtained from T. M. Fuerst and B. Moss (45).
Antibodies-mAb C8, directed to the HIV-1 gp41 cytoplasmic domain, was obtained from G. Lewis (46), and mAb 46, directed against HTLV-1 gp46, was a gift from David Tribe (The University of Melbourne, Victoria, Australia). Immunoglobulin G was purified from the plasma of an HTLV-1-infected individual (anti-HTLV) using protein A-Sepharose (Amersham Biosciences).
Western Blotting-At 24-h posttransfection, Env-expressing 293T cells were lysed for 10 min on ice in phosphate-buffered saline containing 1% Triton X-100, 0.02% sodium azide, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 5 g/ml aprotinin, 5 g/ml leupeptin, and 1 mM dithiothreitol. Lysates were clarified by centrifugation at 10,000 ϫ g at 4°C prior to SDS-PAGE in 12% polyacrylamide gels under reducing conditions. Proteins were transferred to nitrocellulose prior to Western blotting with mAb C8. Immunoblots were developed using the chemiluminescence blotting substrate procedure (Roche Molecular Biochemicals).
Biosynthetic Labeling and Immunoprecipitation-At 16-h posttransfection, 293T cells were incubated for 30 min at 37°C in cysteine-and methionine-deficient medium (ICN, Costa Mesa, CA) and then labeled with 120 Ci of [ 35 S]cysteine (ICN) per well for 24 h. Immunoprecipi-tations with mAbs C8 and 46 were performed as described previously (48).
Assessment of Cell Surface Env Expression Using an Antibody Binding Assay-Cell surface expression of Env mutants was determined using a modified method (36). Radioiodinated anti-HTLV was precleared with 10 7 untransfected 293T cells for 2 h at room temperature before addition to transfected 293T cells at 48-h posttransfection. The transfected cells were incubated with 125 I-labeled anti-HTLV in complete medium for 4 h at 37°C. The cells were then washed four times (5 min/wash) with phosphate-buffered saline containing 10 mg/ml of bovine serum albumin that had been prewarmed to 37°C before counting in a Packard Auto-Gamma counter.
Luciferase Reporter Assay for Cell-Cell Fusion-Cell-cell fusion activity of Env mutants was determined using a modified method (36). 293T (effector) cells were cotransfected with pTMluc and wild-type or mutated pCELT.1 vectors. In parallel, HeLa target cells were infected with vTF7.3 at a multiplicity of infection of 1 plaque-forming unit per cell. At 16-h postinfection, HeLa cells were resuspended in phosphatebuffered saline containing 50 M EDTA, washed twice in complete medium, and cocultured with transfected 293T cells for a further 12 h in the presence of 1 g/ml of actinomycin D and 40 g/ml of cytosine arabinoside at 37°C. Cells were then lysed and assayed for luciferase activity using the Promega (Madison, WI) luciferase assay system.

E. coli Expression and Purification of Trimeric MBP/gp21
Chimeras Containing the Fusion Peptide-Previously we found that expression in E. coli of a chimera comprising MBP linked to the gp21-(338 -425) ectodomain fragment lacking the fusion peptide resulted in high yields of soluble trimer suitable for x-ray crystallographic studies (12,43). We therefore examined the utility of the MBP expression system for production of gp21 ectodomain fragments containing the fusion peptide and the adjacent glycine-rich segment (Fig. 1). A series of MBP/gp21 chimeras were generated comprising MBP linked via NAA to the gp21 N-terminal residue, Ala 313 . The chimeras were terminated at Thr 425 , Asn 430 , Leu 433 , Ser 436 , Ala 439 , or Thr 445 ; Thr 445 is predicted to be the last residue of the gp21 ectodomain (Fig. 1). The chimeras were expressed in E. coli and MBP/gp21 trimers purified by amylose-agarose affinity chromatography and Superdex 200 gel filtration. The purified chimeras coelute with trimeric MBP/gp21-(338 -425) in analytical Superdex 200 gel filtration experiments ( Fig. 2A). Therefore, the tendency of hydrophobic fusion peptides to cause precipitation or aggregation of the TM protein ectodomain, as has been observed for other viral TM proteins (49), is mitigated by the N-terminal MBP moiety.
The gp21 ectodomain contains a conserved disulfide-bonded loop formed by Cys 393 -Cys 400 , whereas Cys 401 is unpaired ( Fig.  1) (12). The disulfide-bonded loop is associated with a chainreversing structure at the base of the helical hairpin and is a key determinant of Env fusogenicity (36). We therefore assessed the redox state of MBP/gp21 chimeras by treatment with the alkylating agent 4-VP followed by electrospray mass spectrometry to detect the covalent modification of free sulfhydryls (43). The molecular mass of 4-VP-treated chimeras increased by ϳ105 Da (4-VP, M r 105) and is consistent with the presence of a single free cysteine (Fig. 2B). The molecular mass of chimeras increased by ϳ315 Da when pretreated with the reducing agent dithiothreitol prior to alkylation, confirming the presence of three cysteine residues per monomer (data not shown).
The gp21 TMD Proximal Residues, Gly 426 to Ala 439 , Protect the Fusion Peptide Region of MBP/gp21 Chimeras against Limited Proteolysis with Chymotrypsin-The purified chimeras, with the exception of MBP/gp21-(313-425), were stable following storage for 4 weeks at 4°C. SDS-PAGE revealed proteolysis of the shortest chimera, MBP/gp21-(313-425), suggesting that the TMD proximal residues Gly 426 -Thr 445 are required for chimera stability. We therefore assessed the role of gp21 TMD proximal residues in chimera stability by subjecting freshly purified MBP/gp21 trimers to limited proteolysis with chymotrypsin. Chymotrypsin cleaves on the C-terminal side of the amino acids Tyr, Phe, and Trp and to a lesser extent Leu, Met, Ala, Asp, and Glu. The gp21 ectodomain contains 55 chymotrypsin targets with 22 target residues in the N-and C-terminal regions for which there is no available three-dimensional structure (Fig. 1). MBP/gp21 trimers (250 g, 0.5 mg/ml) were digested at 37°C with limiting amounts of chymotrypsin (1:150 ratio of protease/protein (w/w)) for 1 to 60 min, and the digestion patterns were analyzed by SDS-PAGE. Fig. 3A reveals that protease resistance correlated with chimera length. The shortest construct, MBP/gp21-(313-425), was the most sensitive to chymotrypsin with Ͼ 50% cleavage occurring after 1 min and almost 100% cleavage after 5 min. Incremental increases in chimera stability were seen with gp21 C-terminal extensions. Approximately 50% of MBP/gp21-(313-430) was cleaved after 10 min and almost 100% cleaved after 60 min. MBP/gp21-(313-433) and MBP/gp21-(313-436) required 60 min for ϳ50% cleavage, whereas only ϳ10% of MBP/gp21-(313-439) was cleaved after 60 min.
Electrospray mass spectrometry of the proteolyzed chimeras identified three major sites of chymotrypsin cleavage within the gp21 moiety. The mass spectrometry profile of MBP/gp21-(313-430) after digestion with chymotrypsin for 30 min illustrates the pattern of proteolysis (Fig. 3B) 11,983) was detected at earlier time points. The Cterminally truncated chimera MBP/gp21-(313-427) was not detected, therefore it follows that cleavage at Trp 319 (Fig. 1, Chy1) is required for subsequent cleavage at Leu 324 and Trp 427 (Fig. 1,  Chy2). Mass spectrometry analysis of the other chimeras at various time points revealed the same sites of chymotrypsin cleavage, giving rise to the protease-resistant gp21 core, gp21-(325-427) or gp21-(325-425) from the shortest chimera, MBP/gp21-(313-425) (see supplementary Table I). In the context of a MBP/ gp21 chimera containing the fusion peptide, the gp21 TMD proximal residues 431-439 are required for protection of the gp21 fusion peptide region from proteolysis. Simultaneously, fusion peptide residues 313-319 are required for the protection of the TMD proximal residues 427-439 against proteolysis.
The gp21 TMD Proximal Residues, Gly 426 to Ser 436 , Confer Thermostability to MBP/gp21 Chimeras Containing the Fusion Peptide-To confirm the role of the TMD proximal sequence in stabilization of fusion peptide-containing MBP/gp21 chimeras, we devised a thermostability assay based on the temperaturedependent conversion of MBP/gp21 trimers to high molecular weight aggregates as monitored by Superdex 200 gel filtration. The thermal aggregation assay distinguishes gp21 unfolding from MBP domain unfolding; the monomeric structure of MBP is not affected by treatment at 60°C, 5 min, and limited proteolysis of heat-induced (50°C, 5 min) MBP/gp21-(313-436) aggregate releases intact monomeric MBP, whereas the gp21 fragment is degraded (data not shown). MBP/gp21 trimers (ϳ100 g, 2 mg/ml in 50 mM sodium chloride, 50 mM glycine, pH 8.3) were subjected to heat treatment over a temperature range of 37-50°C for 5 min. The Superdex 200 profiles of the MBP/gp21 chimeras, before and after heat treatment at 46°C (Fig. 4A), illustrate incremental increases in resistance to thermal aggregation with extension of the gp21 chimeras from Thr 425 to Asn 430 , Leu 433 , and Ser 436 . Chimeras terminated at Ser 436 , Ala 439 (Fig. 4A), and Thr 445 (data not shown) were the most stable with comparable thermostability values. This trend in chimera thermostability was consistent over the temperature range 37-52°C (data not shown) and is reflected in the chimera T MAX.TRI (maximum temperature at which Ͼ95% trimeric structure was maintained) and T MIN.AGG (the minimum temperature required to convert Ͼ 95% of trimer to soluble aggregate) (Fig. 4B). These results indicate that the presence of the fusion peptide and glycine-rich segment (residues 313-337) has a destabilizing influence on the shorter MBP/gp21 chimeras. However, inclusion of TMD proximal residues stabilizes the MBP/gp21-(313-436) and MBP/gp21-(313-439) chimeras such that the thermostabilities approach that of the original core-domain chimera, MBP/gp21-(338 -425). These findings reveal an overall positive correlation between chimera length, thermostability, and protease resistance, the most stable chimera being MBP/gp21-(313-439).
The Detergents n-Octanoyl Sucrose and CYMAL-3 Protect the Fusion Peptide Region against Limited Proteolysis with Chymotrypsin-Previous studies have shown that synthetic peptide analogs of the HIV-1 gp41 and influenza virus HA 2 fusion peptides can insert into detergent micelles (28,31,50). We therefore tested the ability of the fusion peptide in MBP/gp21-(313-425) to interact with detergents by determining whether various detergents can protect the fusion peptide residues Trp 319 and Leu 324 from chymotrypsin proteolysis. SDS-PAGE indicated that chymotrypsin treatment for 5 min at room temperature of MBP/gp21-(313-425) led to substantial amounts of MBP/gp21 FRAG and gp21 CORE (Fig. 5, Chy). A 16-h preincubation of MBP/gp21-(313-425) with the detergents C 12 E 8 , C 12 E 9, and Deoxy BigChap at critical micellar concentrations, prior to chymotrypsin treatment, had no effect on the digestion profile (Fig. 5, D1, D2, and D3). In contrast, preincubation with n-octanoylsucrose and CYMAL-3 led to the protection of MBP/ gp21-(313-425) from chymotrypsin cleavage (Fig. 5, D7 and D8), the protected MBP/gp21 protein migrating as a single ϳ53-kDa band corresponding to untreated MBP/gp21-(313-425) (Fig. 5, Unt). Other detergents, n-decyl-␤-D-maltoside, CY-MAL-5, and n-nonyl-␤-D-glucoside were partially protective (Fig. 5, D4, D5, and D6), although n-nonyl-␤-D-glucoside promoted the degradation of MBP/gp21 FRAG and gp21 CORE . The detergents did not affect chymotrypsin activity as assessed using the chromogenic substrate Suc-Ala-Ala-Pro-Phe-pNA under identical digestion conditions (data not shown). We also tested the effects of detergents on the proteolysis of more stable MBP/gp21 chimeras, which require an incubation temperature of 37°C for significant cleavage. However, digestion at 37°C in the presence of detergents promoted the degradation of MBP and gp21 domains, probably caused by the denaturing effects of heat plus detergent. These results indicate that the fusion peptide in MBP/gp21-(313-425) can interact with n-octanoyl sucrose and CYMAL-3 preventing proteolysis at fusion peptide residues Trp 319 and Leu 324 .
Alanine Substitutions in the gp21 TMD Proximal Ectodomain Sequence Thr 425 -Ser 436 Result in Enhanced Cell-Cell Fusion Activity of HTLV-1 Env-The biochemical characterization of fusion peptide-containing MBP/gp21 chimeras indicated that residues Thr 425 -Ser 436 were important for stability. To determine whether these residues were also important for Env function, we performed alanine-scanning mutagenesis of the Thr 425 -Ser 436 sequence in full-length HTLV-1 Env and tested the effects of the mutations on gp62 precursor processing, gp46-gp21 association, cell surface expression, and cell-cell fusion activity.
The HTLV-1 Env precursor (gp62) is cleaved in the Golgi apparatus to yield gp46 and gp21, which remain noncovalently associated (51). The effect of the Ala substitutions on gp62 synthesis and processing to gp21 in transfected 293T cells was assessed by Western blotting with mAb C8, which is directed to an epitope tag joined to the gp21 C terminus. The 12 point mutants were expressed and processed to yield gp21 at levels comparable with wild type (Fig. 6A) indicating intracellular translocation of cleavage-competent Env structures. We next determined whether the gp21 mutants had retained the ability to anchor the SU (gp46). Transfected 293T cells were metabolically labeled with [ 35 S]Cys, and HTLV-1 Env proteins were immunoprecipitated from cell lysates and clarified culture supernatants with mAb 46, which is directed to gp46. We also used the control mAb C8 to help distinguish gp46 in cell lysate immunoprecipitations. Both gp62 and gp46 were immunoprecipitated by mAb 46 from lysates of wild-type and mutant Env-expressing cells, whereas mAb C8 immunoprecipitated gp62 but not gp46 (Fig. 6B). Only gp46 was obtained from corresponding clarified culture supernatants (Fig. 6C). Similar levels of cell-associated and shed gp46 were observed for wildtype and mutant Env indicating that the mutations had not significantly affected the gp46-anchoring ability of gp21. We verified that the Env mutants were expressed at the cell surface using a surface binding assay employing 125 I-labeled anti-HTLV. The levels of cell surface expression of five mutant Env glycoproteins (T425A, G426A, G428A, L429A, and S436A) were comparable with wild-type surface expression (Fig. 7). Seven of the mutants (W427A, N430A, W431A, D432A, L433A, G434A, and L435A) exhibited slightly elevated cell surface expression at less than 1.4 times the wild-type level. These results are consistent with the normal maturation of Env mutants.
Finally, the cell-cell fusion activity of wild-type and mutated HTLV-1 Env glycoproteins was determined using a luciferase reporter assay employing HeLa cells as fusion targets. Fig. 7 shows that following a 12-h coculture between Env-expressing 293T cells and vTF7.3-infected HeLa targets 11 of the 12 mutants exhibited enhanced fusion activity. The increases in fusion activity ranged from 150 -200% for G426A, W427A, G434A, and S436A; 300 -350% for T425A, G428A, L429A, N430A, and L433A and 400 -450% for W431A and L435A. Only one mutant, D432A, exhibited an ϳ25% reduction in fusion activity. These results indicate that Ala substitutions in the TMD proximal sequence did not affect Env maturation but led to significant enhancement of fusion activity for 11 of 12 mutants.
Alanine Substitutions Do Not Affect MBP/gp21-(313-439) Chimera Stability-To determine whether the enhanced fusion activities associated with Ala substitutions were related to changes in gp21 helical hairpin stability, we introduced T425A, G428A, W431A, D432A, and L433A substitutions into the MBP/gp21-(313-439) chimera. As observed for the wild-type MBP/gp21-(313-439) chimera, the alanine mutants acquired trimeric structures (Fig. 8A) with an intact disulfide bond in each monomer (data not shown). Thermal aggregation analysis indicated that single alanine mutations had no significant effect on chimera stability. The gel filtration profiles of MBP/ gp21-(313-439) mutants treated at 48°C were comparable with 48°C-treated wild-type chimera (Fig. 8A). The stability of wild-type and mutant chimeras was also comparable over the temperature range 37-51°C (data not shown). All MBP/gp21-(313-439) chimeras had a T MAX.TRI of 45°C, and 4 of 5 mutated chimeras showed a wild-type T MIN.AGG of 50°C; T425A exhibited a T MIN.AGG of 51°C (Fig. 8B). The comparable stability of wild-type and mutated chimeras was also reflected in limited proteolysis experiments using chymotrypsin (data not shown). These results indicate that multiple residues within the TMD proximal and fusion peptide/glycine-rich regions contribute to MBP/gp21-(313-439) stability, thereby overriding the potential effects of single residue substitutions. The enhanced fusion activities of HTLV-1 Env mutants are likely to be a result of subtle alterations to the prefusion Env complex. DISCUSSION Previously, we used MBP as an N-terminal solubilization partner to aid the bacterial expression, crystallization, and structure determination of a trimeric HTLV-1 gp21-(338 -425) ectodomain fragment lacking the fusion peptide region (residues 313-337) and TMD proximal sequence (residues 422-445) (12,43). These studies indicated that the gp21 fragment had acquired a helical hairpin conformation resembling the low pH-induced fusogenic form of influenza virus HA 2 (12,21,22). We now show that the MBP expression system also confers solubility to fusion peptide-containing gp21 ectodomain constructs, enabling purification as soluble trimeric MBP/gp21 chimeras. Therefore, the MBP expression system provides a new means of expressing an entire viral TM protein ectodomain in addition to a previously described method employing the highly polar FLAG peptide as an N-terminal solubilization partner for influenza virus HA 2 (49).
Thermal aggregation experiments illustrated the destabilizing influence of the fusion peptide region on MBP/gp21, because the MBP/gp21-(313-425) chimera was significantly less stable than MBP/gp21-(338 -425). The results of studies with hydrated model fusion peptides indicate that they are in a disordered, high entropy state, but adopt ordered ␣-helical structures when inserted in detergent micelles or lipid bilayers (29 -31, 50). The instability of the fusion peptide-containing chimera, MBP/gp21-(313-425), may be due in part to exposure of fusion peptide residues to aqueous solvent leading to disorder in this region. This idea is supported by the observation that fusion peptide residues Trp 319 and Leu 324 in MBP/gp21-(313-425) were rapidly hydrolyzed by chymotrypsin. However, the Trp 319 and Leu 324 sites were not hydrolyzed in the presence of the detergents n-octanoyl sucrose and CYMAL-3, indicating that the fusion peptide within MBP/gp21-(313-425) becomes inaccessible to chymotrypsin when it is in a detergent-bound state.
Inclusion of the TMD proximal sequence Gly 426 -Ala 439 also confers stability to fusion peptide-containing MBP/gp21 chimeras. Incremental increases in chimera thermostability and resistance to chymotrypsin cleavage were observed with extension of the gp21 ectodomain C terminus from Thr 425 to Asn 430 , Leu 433 , Ser 436 , or Ala 439 . Furthermore, initial cleavage at Trp 319 within the fusion peptide is required for cleavage at the TMD proximal residue Trp 427 , indicating that an intact gp21 N-terminal region confers stability to the TMD proximal region. Inclusion of TMD proximal residues beyond Thr 425 may enable contacts to form with the exterior of the gp21 N-cap that terminates the coiled-coil and with residues of the glycine-rich segment thereby imposing structural order to this region and improving the overall stability of chimeras. Consistent with this idea was the finding that the protease-resistant core, gp21-(325-427), retained an intact glycine-rich segment (325-337) and a portion of the TMD proximal sequence (419 -427) despite  (12,43), and ovalbumin (43 kDa) are marked. B, summary of thermal aggregation-gel filtration data. The maximum temperature at which Ͼ95% trimeric structure was maintained (T MAX.TRI ) and the minimum temperature required for conversion of Ͼ95% trimers to misfolded soluble aggregates (T MIN.AGG ) are listed for each chimera. the presence of potential internal chymotrypsin sites (Ala 325 , Met 326 , Ala 328 , Ala 331 , Leu 419 , Glu 420 , and Leu 424 ). In functional terms, interactions between N and C termini in the gp21 helical hairpin may result in a stable structure at the membrane-inserted end of the hairpin, its formation contributing free energy to help drive membrane fusion (22). Alternatively, the TMD proximal sequence may bind to another region of gp21 contributing to helical hairpin stability via an allosteric mechanism.
In contrast to the glycine-rich segment, residues Trp 319 and Leu 324 of the fusion peptide were targets for chymotrypsin even in the most stable chimera, MBP/gp21-(313-439), after extended incubations with protease. This observation is consistent with a theoretical requirement that the fusion peptide in the fusion-activated gp21 helical hairpin does not mediate contacts in order that it is free to insert into a target membrane.
Our results indicate that the glycine-rich segment is stabilized by contacts with TMD proximal residues in the helical hairpin; however, transient flexibility in the glycine-rich segment may be favorable at the early pre-hairpin stages of gp21 refolding induced by SU receptor binding. A flexible glycine-rich linker may decouple unstable transiently hydrated fusion peptides from the N-cap of the coiled-coil, thereby maintaining a stable core while allowing the fusion peptide to attain a favorable membrane-inserted conformation. This scenario is supported by the NMR structure of the detergent-associated HIV-1 gp41 N-terminal region, where fusion peptide residues 8 -14 adopt ␣-helical structure and are linked through a flexible segment (residues 15-23) to the coiled-coil (29).
Alanine-scanning mutagenesis of the Thr 425 -Ser 436 sequence in full-length HTLV-1 Env led unexpectedly to an ϳ1.5to 4.5-fold increase in fusion activity for 11 of 12 mutants. By contrast, introduction of T425A, G428A, W431A, D432A, and L433A mutations into MBP/gp21-(313-439) did not significantly affect thermostability nor protease sensitivity, indicating that multiple residues within the TMD proximal sequence (Thr 425 -Ser 436 ) and fusion peptide/glycine-rich regions contribute to protein stability thereby overriding the potential effects of single residue substitutions in the helical hairpin chimera. The enhancements in fusion caused by single substitutions may be caused by subtle changes in the prefusion Env complex, because they are not explained by alterations to the stability of the helical hairpin. These residues may participate in labile contacts that maintain the SU-TM complex in a prefusogenic state. Ala substitutions in the TMD proximal region may destabilize prefusogenic Env, lowering the fusion-activation threshold resulting in increased fusogenicity. Alternatively or additionally, the TMD proximal sequence may constitute a metastable structural element within the prefusion complex, and Ala substitutions may enhance the kinetics of HTLV-1 Env refolding into a fusion-active structure. However, in the hairpin conformation when N and C termini become juxtaposed, Thr 425 -Ser 436 residues function to stabilize the fusion peptide/ glycine-rich region. The TMD proximal Thr 425 -Ser 436 region therefore appears to have distinct roles in prefusion and fusionactivated gp21 structures. Fusion-enhancing effects are also obtained with mutation of Leu 368 in the central coiled-coil and Glu 419 in the C-terminal ␣-helix (47), indicating that the TMD proximal segment is not the sole determinant for maintenance of Env in a prefusion state.
The observed dual role of gp21 TMD proximal residues may have a functional precedent in influenza virus HA 2 . In the prefusogenic HA 2 trimer, the TMD proximal residue (Arg 163 ) forms an intersubunit salt bridge with Glu 131 (23). The ablation of this intersubunit contact by an R163I mutation is sufficient to increase the pH of fusion by 0.4 units (23, 52), effectively lowering the fusion-activation threshold of HA 2 . Arg 163 and other TMD proximal residues (Asp 164 , Arg 170 , Gln 172 , and Lys 174 ) that mediate intersubunit contacts in the prefusion structure (23) are relocated to an N-cap proximal location in the fusion-activated HA 2 helical hairpin (22). Contacts between the N-cap and TMD proximal residues (Lys 174 -Leu 178 ) result in the formation of a stable structure at the tip of the rodshaped helical hairpin (22). Such interactions between N-terminal and TMD proximal regions may be a conserved mechanism for conferring stability to the membrane-interactive end of viral helical hairpins.