INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The Env protein of ecotropic Moloney murine leukemia virus (Mo-MuLV) is a homotrimeric bifunctional protein responsible for binding to host receptor and fusion of the envelope with the cell plasma membrane (15, 28, 41, 44). Each monomer of the Env protein is synthesized as a gp85 precursor, which is posttranslationally cleaved by cellular proteases into gp70 (surface; SU) and p15E (transmembrane; TM) subunits (29) which are responsible for binding and fusion respectively. The core structure of Mo-MuLV Env is strikingly similar to those of other viral fusion proteins. Fusion proteins for Mo-MuLV (15), human immunodeficiency virus (5, 43), influenza virus (4), Ebola virus (42), and SV5 (a paramyxovirus) (2) contain a triple coiled-coil core surrounded by three C-terminal -helices running antiparallel to the central stem. The presence of common structural features suggests that different viral fusion proteins induce membrane merger by similar mechanisms. The mechanism of fusion has been most extensively delineated for the hemagglutinin (HA) of influenza virus, which thus serves as a prototypic fusion protein (17). Based on the similarity of the crystallographic structures of their TM subunits, it is expected that, analogous with HA, conformational changes in the TM subunits of Mo-MuLV Env lead to an extended coiled-coil stem region and insertion of the subunits' fusion peptides into the target membrane. In a manner not fully understood, this causes fusion between the Mo-MuLV envelope and the plasma membrane to which it is bound. Fusion of Mo-MuLV proceeds at neutral pH (32, 35). It is believed that the binding of the SU subunit to its specific receptor, which occurs at neutral pH, triggers the conformational changes in the Env protein, which allows fusion to proceed.

Mo-MuLV is different than most other enveloped viruses in that the fusogenic activity of its Env protein is controlled by the trimming of the protein's cytoplasmic tail (CT). When synthesized, the CT of the p15E subunit is 32 amino acid residues long (residues 601 to 632). At the time Mo-MuLV buds from the cell, the 16 C-terminal residues (referred to as the R peptide) of the CT are removed by a viral protease (10, 16, 36), greatly increasing the fusogenic ability of Env (18, 33, 35). The role of the length of the CT in Mo-MuLV Env-induced fusion has been established by expressing the fusion protein with full-length and truncated CTs in cells and testing their ability to form heterokaryons (i.e., syncytia) with target cells containing ecotropic virus receptors. Cells expressing Env with a CT truncated by 16 amino acid residues ("R-less" or Env 616*) have the highest syncytial potency (18, 33, 35). Deleting the CT altogether to residue 601 (Env 601*) leads to a somewhat reduced extent of syncytia formation, while further truncation that removes residues from the C-terminal portion of the membrane-spanning (MS) domain (e.g., Env 595*) completely abolishes polykaryon formation (Y. Rozenberg et al., submitted for publication). For syncytia to form, not only must fusion occur, but other events, such as major pore growth, cytoskeleton rearrangements, and movement of nuclei, must proceed as well. Therefore, additional experimental approaches are required to delineate the process of fusion pore formation and its enlargement.

We have used fluorescence microscopy to monitor lipid continuity and electrophysiological measurements to record fusion pores in order to assess the effect of the CT of Env on membrane fusion and on the early stages of pore growth. We show that deleting the C-terminal R peptide from the full-length CT (yielding Env 616*) strongly promoted a very early step required for syncytia: formation of the fusion pore itself. Deletion of almost the entire CT (Env 601*) had little further effect on fusion pore formation or on the initial size or growth of the pore. Additional truncation of six C-terminal residues (up to G595) from the predicted MS domain completely abolished fusion. A chimera consisting of residues 1 to 599 of the Env of Mo-MuLV followed by a CT, consisting of the amphiphilic portion of melittin, induced fusion pores similar to those caused by Env 601* and Env 616*. We conclude that the CT itself is not needed for Env-induced fusion but that residues 596 to 601 are required (either directly or indirectly through affecting other portions of Env) for fusion.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Envelope constructs, cell culture, and Env expression. The envelope protein constructs were produced as described elsewhere (Rozenberg et al., submitted). An asterisk denotes that a stop codon follows the indicated residue number (Fig. 1). The expressed Env protein is denoted by its C-terminal residue number (e.g., 616* denotes the Env protein consisting of residues 1 to 616). MelR denotes a chimera between residues 1 and 599 of Mo-MuLV Env and a fragment of melittin (serving as a CT; Fig. 1). The "M" stands for methionine, a mutation inadvertently introduced during PCR mutagenesis. This position is normally occupied by isoleucine. The XC6 cell line (a highly fusion permissive subclone of XC rat sarcoma cells [Y. Rozenberg, unpublished observations]) and NIH 3T3 fibroblasts were grown within a humidified 5% CO₂ incubator in basal minimal essential medium or Dulbecco modified Eagle medium (DMEM), respectively, supplemented with 10% Cosmic Calf Serum (HyClone Laboratories, Logan, Utah), L-glutamine, and penicillin-streptomycin (GIBCO BRL, Gaithersburg, Md.). HEK 293T (referred to as 293T) cells were maintained in DMEM-Cosmic Serum supplemented with 0.5 mg of geneticin per ml. This cell line has excellent transfection efficiency and lacks ecotropic Env receptors, which precludes cell-cell fusion among themselves even though the Env protein is expressed on their surfaces. Different constructs of the Env protein of Mo-MuLV were transiently expressed in 293T cells by transfecting with the plasmid pHIT123 (37) by using calcium phosphate. A total of 15 µg of Env 632*, 601*, 595*, and MelR plasmids and 12 µg of Env 616* were used per each 6-cm culture dish. Cells were incubated with a DNA precipitate for 4 h at 37°C in the presence of 25 µg of chloroquine per ml. Relative levels of surface expression of Env constructs were assessed by flow cytometry 48 h after transfection by using an anti-gp70 rat monoclonal antibody, 83A25, against Env protein as described previously (18).

View larger version (14K): [in this window] [in a new window]   FIG. 1.   Mo-MuLV Env constructs. The asterisk indicates a stop codon terminating the CT sequence after the indicated residue number. MelR is a chimeric construct wherein the CT was derived from melittin, a membrane-active amphipathic peptide. The vertical dotted line shows the division of the MS domain and beginning of the CT. The precise location where the MS domain ends and the CT begins is not known. By standard considerations of hydrophobicity, the CT would begin at Arg601. However, the sequence Gly595-Pro596 might introduce a turn and initiate an amphiphilic -helix that runs from residues 598 to 616 and that functions as a unit (Rozenberg et al., submitted; see also below). In our descriptions, we retain the conventional division between the MS domain and the CT based on hydrophobicity but remain cognizant of a possible unity of function of the residue 595 to 616 region. Whereas residues 595 to 600 are part of the MS domain, based on hydrophobicity, it has been argued that this region is in fact part of the CT and is therefore also referred to as the "membrane-proximal region" (Rozenberg et al., submitted).

Labeling the cells with fluorescent dyes. The rationales for procedures used to fluorescently label cells have been described (9). In the current series of experiments, 293T cells expressing Env protein (defined as the "effector cells") were labeled with the cytoplasmic marker CalceinAM (CaAM; Molecular Probes, Eugene, Oreg.). Cells were lifted from a 6-cm culture dish with phosphate-buffered saline (PBS) supplemented with 0.5 mM EDTA, 0.5 mM EGTA, and 3 mg of glucose per ml and labeled with 2 to 4 µM CaAM according to the manufacturer's instructions. XC6 and NIH 3T3 (referred to as 3T3) fibroblasts (the "target cells") were colabeled with the cytoplasmic marker 7-amino-4-chloromethylcoumarin (CMAC; Molecular Probes, Inc.) and a lipophilic probe, either PKH-26 (Sigma Chemical Co., St. Louis, Mo.) or DiI (Molecular Probes). A confluent monolayer of target cells in a 10-cm culture dish (ca. 10⁷ cells) was washed twice with PBS, and incubated in OptiMEM (GIBCO BRL) containing 40 µM CMAC for 30 min at 37°C. Cells were then incubated with dye-free OptiMEM for 15 min, lifted by a brief exposure to a trypsin-EDTA solution, resuspended in DMEM containing 10% bovine serum, and washed twice with PBS. CMAC-loaded cells were subsequently labeled with 1 to 2 µl of a 1 mM stock solution of PKH26 in ethanol or with 1 to 2 µg of DiI. Membrane dyes were injected into PBS, vortexed, briefly sonicated, and immediately mixed with an equal volume of cell suspension. Double-dye-labeled cells were washed once in DMEM supplemented with 10% bovine serum and then twice in PBS.

Fluorescence video microscopy. The fluorescence of CaAM-labeled 293T cells was monitored by using a standard fluorescein filter set for an Axiovert 100A microscope (Carl Zeiss, Inc., Thornwood, N.Y.). The fluorescence of CMAC-PKH26-colabeled target cells was monitored by standard DAPI (4',6'-diamidino-2-phenylindole) and rhodamine filter sets, respectively. The amount of fluorescent dyes used to label the cells and the spectral characteristics of filter cubes was selected to minimize a bleed-through of fluorescence from one dye when using a filter set of another. Fluorescence images were monitored with an intensified (KS1380; Video Scope, Washington, D.C.) charge-coupled device video camera (Dage 72, Michigan City, Ind.) and recorded onto S-VHS videotape. Images were digitized by a video frame-grabber (Meteor; Matrox Electronic Systems, Dorval, Quebec, Canada) and a PC-based computer. Images acquired for each fluorescent dye were pseudocolored according to their emission wavelength (red, green, or blue) and superimposed on each other by using locally written imaging software.

Cell-cell binding and fusion. Plastic culture dishes (35 mm) were precoated with poly-L-lysine (M_r, 70,000 to 150,000; Sigma) according to manufacturer's instructions to ensure that the cells attached to the dishes. An approximately equal number of effector and target cells were mixed, transferred into the culture dishes, and allowed to bind to each other for 1 h at 4°C in PBS. When fusion was quantified, the density of effector and target cells was made sufficiently low so that cell pairs still formed but larger aggregates did not. After coincubation at 4°C, cells were washed once to remove unattached cells, and fusion was induced by raising the temperature to 37°C for controlled times.

Electrophysiological measurement of fusion pore formation. Fusion of an Env-expressing 293T cell to a target XC6 cell was monitored electrically in the whole-cell patch clamp configuration as an increase in electrical capacitance of the patched cell membrane due to the addition of the fused cell membrane (30, 31, 39). The full capacitance (i.e., area) of the target cell membrane is revealed only when the fusion pore connecting the effector and target cells is large. Conductances of small and intermediate fusion pores were calculated from the increment in whole-cell admittance as G_P = (Y₀² + Y₉₀²)/Y₀, where Y₀ and Y₉₀ are the increases in the in-phase and 90° out-of-phase components of electrical admittance, respectively (34). For electrophysiological experiments, target and effector cells were mixed, adhered to a poly-L-lysine-coated coverglass in the cold, and stored on ice prior to the experiment. Patch clamp experiments were conducted at 37°C. A 293T cell (labeled with CaAM to allow easy identification) was patched and lifted from the coverglass after a high-resistance seal formed between the patch pipette and the cell. The 293T cell was then brought into contact with a solitary XC6 cell (either unlabeled or labeled with PKH26). At the moment of physical contact between the two cells, the time was set equal to zero.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

An assay for cell-cell fusion: tracking the movement of three dyes. Fusion was monitored between cells expressing Env proteins (defined as effector) and cells containing receptors for ecotropic Mo-MuLV Env (defined as target). Human 293T cells, transiently transfected with a plasmid containing the desired Mo-MuLV Env construct, do not contain receptors recognized by Mo-MuLV Env and served as the effector cells. XC6 cells and 3T3 fibroblasts were used as target cells. All the Env constructs were expressed efficiently on surfaces of 293T cells as judged by the presence of Env epitopes detected by flow cytometry (Table 1). The expression of Env 601* was consistently lower but still within about a factor of 2 of the other constructs. Attempts to increase the expression level of Env 601* by using a greater amount of plasmid DNA for transfection were unsuccessful (not shown). Env 595* and Env 616* were expressed at higher densities than the full-length Env 632*.

View larger version (73K): [in this window] [in a new window]   FIG. 2.   Relative fusion activities of Env constructs monitored by a three-fluorescent-dye assay. Effector (293T) cells expressing various Env constructs (as indicated in labels to the left) were labeled with CaAM (green fluorescence). Target 3T3 (left panels) and XC6 (right panels) cells were colabeled with CMAC and PKH26 that have blue and red fluorescence, respectively. Cells were allowed to form contacts for 1 h at 4°C and then were incubated at 37°C for either 60 min (3T3 target cells) or 10 min (XC6 target cells). Fluorescence images for each dye were acquired, pseudocolored, and superimposed on each other. The resulting colors are as follows: green for effector cells, purplish for double-labeled target cells, and white or near white for the fused cells (shown by arrows). The occasional white spots seen for 632* and 595* resulted from a spatial overlap between unfused effector and target cells that occurs when part of one cell lays on the top of another.

Env-mediated fusion is more efficacious with XC6 cells than with NIH 3T3 cells as the target. We compared the ability of two cell lines, NIH 3T3 and XC6 cells, to fuse to human 293T expressing Env constructs. 293T cells expressing an Env (that supported fusion) had to be incubated with 3T3 cells at 37°C for at least 25 min for the fluorescent dye to redistribute; with a 60-min incubation, the dye redistribution was complete. In contrast, with XC6 cells as targets, fusion occurred as early as 2 to 3 min after an increase of the temperature to 37°C. (These observations dictated the choice of the time points in Fig. 2.) These results are consistent with the previously observed higher Env-induced syncytium-forming activity for XC cells than for 3T3 fibroblasts as targets (18, 21). Fusion was so efficient that even when temperatures were kept relatively low, between 16 and 23°C, some Env 616*-expressing 293T cells fused to XC6 cells within ca. 1 h. Thus, the fusion step alone could account for the higher level of observed syncytium formation with XC cells; there need not be any differences in the further steps required for syncytia to form, such as extensive pore enlargement.

Env 616* and Env 601*, but not Env 595* and Env 632*, were capable of inducing substantial fusion. In order to quantify the extent of fusion for the Env proteins with CTs truncated to differing extents, cells were plated at lower densities so that cell pairs, rather than aggregates, were preferentially formed. The fraction of effector-target cell pairs that were stained with all three fluorescent dyes provided the extent of fusion. The quantitative level of fusion for the various forms of Env (Fig. 3) agreed with the representative images shown (Fig. 2). After a 1-h incubation of effector cells with 3T3 target cells at 37°C, the full-length CT (Env 632*) did not result in fusion (Fig. 3A). In contrast, virtually all cells expressing Env 616* fused to their bound target cells. Truncation of the CT to 601 (i.e., Env 601*) resulted in only a minor reduction of fusion activity, whereas a further truncation of six more amino acid residues (Env 595*) abolished fusion. The chimera consisting of residues 1 to 599 of Env, followed by the amphiphilic portion of the peptide melittin (Env MelR), supported efficient fusion.

View larger version (26K): [in this window] [in a new window]   FIG. 3.   Extent of fusion between 293T cells expressing Env protein and either 3T3 fibroblasts (A) and XC6 cells (B). Fusion between effector and target cells was quantified based on fluorescent dye redistribution. (A) After the prebinding of effector and 3T3 cells at 4°C, cells were incubated at 37°C for 1 h, and fusion was assessed by determining the fraction of cell pairs stained with all three fluorescent probes. (B) Extent and the time dependence of Env-induced fusion to XC6 cells. Effector cells were prebound to target cells at 4°C, and the temperature was subsequently raised to 37°C for either 10 min (striped bars) or 40 min (open bars). Error bars represent the standard error of the mean.

Mo-MuLV Env-mediated fusion to XC6 cells is fast. In order to determine the fusion kinetics at early times after creating the conditions that permit fusion, Env-expressing 293T cells and XC6 cells were preincubated at 4°C for 1 to 2 h, and then the temperature was quickly stepped from 4 to 37°C. The lag times from raising the temperature to the onset of calcein redistribution from 293T to XC6 cells were monitored by video microscopy and plotted as cumulative distributions (Fig. 4A). These distributions provide the kinetics of fusion pores that have formed and grown large enough to allow the small molecule calcein (M_r, ~600) to pass through them. We used this temperature-raising method to continuously monitor the movement of dye spread between individual cell pairs for as long as 4 to 5 min. As cells that would have fused later were excluded, the distributions of waiting times to pore formation were truncated.

View larger version (29K): [in this window] [in a new window]   FIG. 4.   Kinetics of Env-mediated fusion pore formation as measured by the onset of calcein redistribution (A) and electrical measurements of fusion pores (B). (A) Lag times from a temperature jump from 4 to 37°C (time = 0) until the onset of calcein transfer into a target cell were plotted as cumulative distributions. Env 632* and 595* Env proteins failed to form a pore 10 min after the temperature was raised. Env 616* induced fusion pores after a shorter delay than MelR, which, in turn, induced fusion with less of a delay than did Env 601*. (B) Lag times were measured electrically (at 37°C) as the time intervals between establishing physical contact (by manipulating an effector cell into proximity of a target cell) and fusion pore formation (detected by capacitance measurements). The lag times measured electrically for Env 601* () and 616* () were similar to their lag times measured by dye spread measurements (see panel A). Only cells that fused were used to obtain the cumulative distributions in panels A and B. Thus, the fraction fused in these distributions plateau at long times after acidification to the value of 1.

Initial conductance and enlargement of small fusion pore formed by Env. Fusion pore behavior was analyzed for Env 616*, 601*, and MelR constructs by means of capacitance measurements in the whole-cell patch clamp configuration. Small fusion pores are the earliest detectable events in fusion. They establish cytoplasmic continuity and membrane merger. As readily seen from the representative traces (Fig. 5), all three constructs generated similar pores at their early stage of growth, regardless of the length (Env 616* versus Env 601*) and the sequence (Env 616* versus Env MelR) of the CT.

View larger version (15K): [in this window] [in a new window]   FIG. 5.   Representative traces of fusion pores formed between 293T expressing Env 616*, Env 601*, and Env MelR and XC6 cells. All experiments were carried out at 37°C. After the patch clamping of an effector cell, a whole-cell configuration was established, and the cell was lifted from its supporting coverglass and brought in contact with a solitary XC6 cell. Fusion pore formation was detected shortly thereafter by characteristic changes in the electrically recorded cell admittance. Fusion pore conductance was calculated off-line. (See Materials and Methods for details.)

View larger version (33K): [in this window] [in a new window]   FIG. 6.   Formation and enlargement of fusion pores formed by Env constructs. Fusion pores induced between Env-expressing 293T cells and XC6 cells were detected in a whole-cell patch-clamp mode by electrical capacitance measurements, and the average pore conductance profiles were constructed. (A) Average initial pores formed by Env 616*, Env 601*, and Env MelR. The error bars are not shown for MelR for visual clarity. (B) Enlargement of an average fusion pore within the first 100 s of opening. In both panels A and B, Env 616*-induced pore conductances (n = 17) are shown by solid circles, those formed by Env 601* (n = 10) are shown by open circles, and those formed by Env MelR (n = 11) are shown by triangles. Error bars are the standard errors of the mean. The greater average pore conductance and unusually large error bars for Env 601* were caused by fusion pores that enlarged very quickly in 2 of the 10 experiments. When these two experiments were disregarded, the initial conductance profiles and subsequent enlargement were identical for Env 601* and Env 616* (not shown).

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Fusion pores formed by retroviral Env protein are similar to those created by other viral proteins. This is the first study that has electrically characterized fusion pores created by a retroviral fusion protein. The nascent fusion pores formed by Env protein of Mo-MuLV have an average diameter of about 7 nm (conductance, ca. 2 nS). These sizes are similar to those of pores formed by baculovirus gp64 (30) and Semliki Forest virus E1/E2 fusion protein (20), but they are appreciably larger than the 3-nm (initial conductance, ~0.5 nS) pores induced by influenza virus HA (23, 39, 40, 46). A larger pore size may indicate that more Env trimers than HA trimers participate in pore formation. Unlike fusion pores formed by influenza virus HA, pores induced by Env protein did not flicker. Rather, they remained open and steadily enlarged. Thus, although HA and Env of Mo-MuLV have similar core structures, as determined crystallographically (4, 15), the properties of their fusion pores are different. Perhaps the lack of flickering for pores formed by Env of Mo-MuLV, by baculovirus gp64 protein, and by Semliki Forest virus E1/E2 protein is related to their larger initial size.

Binding is not a rate-limiting step in Mo-MuLV Env fusion. It had been known that the processed Mo-MuLV Env protein (Env 616*) promoted syncytium formation more effectively with the XC6 cell line than with murine 3T3 fibroblasts and that Env 632* caused syncytium formation with XC cells but not with 3T3 cells (18, 21). We have now demonstrated that fusion pores themselves occur to a greater extent and much more quickly with XC6 cells as target and that there was significant pore formation between Env 632*-expressing cells and XC6 cells (Fig. 3B) but not to 3T3 fibroblasts (Fig. 3A). These correlations indicate that the previously observed differences in syncytium formation occur at the point of membrane fusion rather than at the subsequent steps required for syncytia to form.

Residues 602 to 616 are not required for membrane fusion. For some viral Env proteins, such as for Env of human immunodeficiency virus type 1, syncytia may not form despite the occurrence of fusion (13). We have now shown that the previously observed blockage of appearance of syncytia by R peptide (the C-terminal 16 amino acid residues 617 to 632) of Mo-MuLV (18, 35, 36) is due to prevention of fusion pore formation: deletion of the R peptide greatly promotes Mo-MuLV Env-mediated fusion. The presence of an R peptide can inhibit fusion induced by other viral Env proteins as well, but this is not universally the case. A chimera between the ectodomain and the TM domain of simian immunodeficiency virus gp160 protein and the full-length CT of Mo-MuLV Env protein did not promote fusion, whereas fusion occurred for the chimera lacking the R peptide (but containing the remaining 16 residues of the CT of Mo-MuLV Env) (45). In contrast, a chimera between the ectodomain of human T-cell leukemia virus type I and the TM domain and full-length CT (which contains the R peptide) from Friend murine leukemia virus supported fusion (12).

Possible roles of the CT in controlling Env-mediated membrane fusion. Hemifusion is the merger of contacting leaflets of two membranes, while distal, inner leaflets remain distinct and form a bilayer known as a hemifusion diaphragm that continues to separate aqueous compartments. Hemifusion is thought to be a key intermediate of fusion. As evidence for this view, a number of mutations of fusion proteins from several different viruses have been shown to result in lipid dye spread without mixing of aqueous dye when the mutant proteins were expressed on cell surfaces (1, 6-8, 19, 22, 25, 26, 31). But no mutant of Mo-MuLV Env created so far has been shown to produce hemifusion. If viral fusion proteins, in general, induce hemifusion as an intermediate, it is likely that Mo-MuLV Env does so as well. Assuming this is the case, how may the CT of Mo-MuLV Env be involved in regulating fusion?