1. Microbiology and Infectious Disease
  2. Structural Biology and Molecular Biophysics

Structure of HIV-1 gp41 with its membrane anchors targeted by neutralizing antibodies

  1. Christophe Caillat
  2. Delphine Guilligay
  3. Johana Torralba
  4. Nikolas Friedrich
  5. Jose L Nieva
  6. Alexandra Trkola
  7. Christophe J Chipot
  8. François L Dehez
  9. Winfried Weissenhorn  Is a corresponding author
  1. Univ. Grenoble Alpes, CEA, CNRS, Institut de Biologie Structurale (IBS), France
  2. Instituto Biofisika (CSIC, UPV/EHU) and Department of Biochemistry and Molecular Biology, University of the Basque Country (UPV/EHU), Spain
  3. Institute of Medical Virology, University of Zurich, Switzerland
  4. Laboratoire de Physique et Chimie Théoriques (LPCT), University of Lorraine, France
  5. Laboratoire International Associé, CNRS and University of Illinois at Urbana-Champaign, France
  6. Department of Physics, University of Illinois at Urbana-Champaign, United States
https://doi.org/10.7554/eLife.65005
Share this article
Cite this article
  1. Christophe Caillat
  2. Delphine Guilligay
  3. Johana Torralba
  4. Nikolas Friedrich
  5. Jose L Nieva
  6. Alexandra Trkola
  7. Christophe J Chipot
  8. François L Dehez
  9. Winfried Weissenhorn
(2021)
Structure of HIV-1 gp41 with its membrane anchors targeted by neutralizing antibodies
eLife 10:e65005.
https://doi.org/10.7554/eLife.65005
  2. Download BibTeX
  3. Download .RIS

Abstract

The HIV-1 gp120/gp41 trimer undergoes a series of conformational changes in order to catalyze gp41-induced fusion of viral and cellular membranes. Here, we present the crystal structure of gp41 locked in a fusion intermediate state by an MPER-specific neutralizing antibody. The structure illustrates the conformational plasticity of the six membrane anchors arranged asymmetrically with the fusion peptides and the transmembrane regions pointing into different directions. Hinge regions located adjacent to the fusion peptide and the transmembrane region facilitate the conformational flexibility that allows high-affinity binding of broadly neutralizing anti-MPER antibodies. Molecular dynamics simulation of the MPER Ab-stabilized gp41 conformation reveals a possible transition pathway into the final post-fusion conformation with the central fusion peptides forming a hydrophobic core with flanking transmembrane regions. This suggests that MPER-specific broadly neutralizing antibodies can block final steps of refolding of the fusion peptide and the transmembrane region, which is required for completing membrane fusion.

Introduction

Viral fusion proteins catalyze virus entry by fusing the viral membrane with membranes of the host cell, thereby establishing infection. The HIV-1 envelope glycoprotein (Env) is a prototypic class I fusion protein that shares common pathways in membrane fusion with class II and III viral membrane fusion proteins (Baquero et al., 2015; Harrison, 2008; Kielian and Rey, 2006; Weissenhorn et al., 2007). HIV-1 Env is expressed as a gp160 precursor glycoprotein that is cleaved into the fusion protein subunit gp41 and the receptor-binding subunit gp120 by host furin-like proteases. Gp41 anchors Env to the membrane and associates non-covalently with gp120, thereby forming a stable trimer of heterodimers, the metastable Env prefusion conformation (Chen, 2019; Wang et al., 2020). Orchestration of a series of conformational changes transforms energy-rich prefusion Env into the low-energy, highly stable gp41 post-fusion conformation, which provides the free energy to overcome the kinetic barriers associated with bringing two opposing membranes into close enough contact to facilitate membrane fusion (Harrison, 2008; Weissenhorn et al., 2007).

HIV-1 gp41 is composed of several functional segments that have been shown or suggested to extensively refold upon fusion activation: the N-terminal fusion peptide (FP), a fusion peptide proximal region (FPPR), the heptad repeat region 1 (HR1), a loop region followed by HR2, the membrane proximal external region (MPER), the transmembrane region (TMR), and a cytoplasmic domain. Structures of native Env trimers in complex with different broadly neutralizing antibodies revealed the conformation of the gp41 ectodomain lacking MPER in the native prefusion conformation (Julien et al., 2013; Kwon et al., 2015; Lyumkis et al., 2013; Pancera et al., 2014; Stewart-Jones et al., 2016; Ward and Wilson, 2017). Env interaction with CD4 results in opening of the closed prefusion trimer (Liu et al., 2008; Ozorowski et al., 2017), which includes the displacement of gp120 variable regions 1 and 2 (V1-V2) at the apex of the trimer and some changes in gp41 including extended and more compact HR1C helices, helical FP conformations (Wang et al., 2018) and accessibility of the MPER epitope for 10E8 interaction (Lee et al., 2016; Rantalainen et al., 2020). This is required for the formation of a stable ternary complex of Env-CD4 with the co-receptor (Chang et al., 2005; Dobrowsky et al., 2008; Shaik et al., 2019). Co-receptor-binding positions prefusion gp41 closer to the host-cell membrane (Chen, 2019) and induces a cascade of conformational changes in gp41. First, FP is repositioned by ~70 Å (Pancera et al., 2014) to interact with the target cell membrane, generating a 110 Å extended fusion-intermediate conformation (Frey et al., 2008; Lai et al., 2014) that bridges the viral and the host cell membrane (Weissenhorn et al., 1999). Subsequent refolding of HR2 onto HR1 leads to the formation of the six-helix bundle core structure (Caffrey et al., 1998; Chan et al., 1997; Weissenhorn et al., 1997), which pulls the viral membrane into close apposition to the host-cell membrane to set the stage for membrane fusion (Weissenhorn et al., 1997).

Membrane fusion generates a lipid intermediate hemifusion state, that is predicted to break and evolve to fusion pore opening (Chernomordik and Kozlov, 2005), which is regulated by six-helical bundle formation (Markosyan et al., 2003; Melikyan et al., 2000). Furthermore, residues within FPPR, FP, MPER, and TM have been as well implicated in fusion (Bellamy-McIntyre et al., 2007; Long et al., 2011; Peisajovich et al., 2000; Salzwedel et al., 1999; Shang et al., 2008) indicating that final steps in fusion are controlled by the conformational transitions of the membrane anchors into the final post-fusion conformation.

Here, we set out to understand the conformational transitions of the gp41 membrane anchors. We show that the presence of the membrane anchors increases thermostability. However, complex formation with a MPER-specific neutralizing nanobody stabilized an asymmetric conformation of the membrane anchors, which may represent a late fusion intermediate. We show that this conformation can be targeted by MPER bnAbs consistent with the possibility that MPER-specific nAbs can interfere all along the fusion process until a late stage. Starting from the asymmetric conformation, we used MD simulation based modelling to generate the final post-fusion conformation, which reveals a tight helical interaction of FP and TM in the membrane consistent with its high thermostability. In summary, our work elucidates the structural transitions of the membrane anchors that are essential for membrane fusion, which can be blocked by MPER-specific bnAbs up to a late stage in fusion.

Results

Gp41FP-TM interaction with 2H10

Two gp41 constructs, one containing residues 512 to 594 comprising FP, FPPR, and HR1 (N-terminal chain, chain N) and one coding for resides 629 to 716 including HR2, MPER and TM (C-terminal chain, chain C) (Figure 1A) were expressed separately, purified and assembled into the monodisperse trimeric complex gp41FP-TM (Figure 1—figure supplement 1A). Gp41FP-TM reveals a thermostability of ~93°C as measured by circular dichroism (Figure 1—figure supplement 2A) indicating that the presence of FP and TMR increases the thermostability by ~7°C compared to gp41 lacking FP and TM (Buzon et al., 2010). In order to facilitate crystallization, gp41FP-TM was complexed with the llama nanobody 2H10 (Lutje Hulsik et al., 2013) in β-OG buffer and purified by size exclusion chromatography (SEC) (Figure 1—figure supplement 1B). To determine the stoichiometry of binding, we performed isothermal titration calorimetry (ITC), which indicated that gp41FP-TM and 2H10 form a 3:1 complex with a KD of 2.1 ± 0.9 µM (Figure 1—figure supplement 2B). Interaction of gp41FP-TM with 2H10 was further confirmed by biolayer interferometry (BLI) analysis (Figure 1—figure supplement 2C).

Figure 1 with 7 supplements see all
Download asset Open asset
Crystal structure of gp41FP-TM in complex with 2H10.

(A) Schematic drawing of gp41 and expression constructs of gp41 chains N and C. Sequence numbering is based on the HIV-1-HBX2 envelope gp160 sequence. Color coding is as follows: FP, fusion peptide, red; FPPR, fusion peptide proximal region, orange; HR1, heptad repeat region 1, yellow; HR2, heptad repeat region 2, blue; MPER, membrane proximal external region, violet; TM, transmembrane region, beige; CC, cys loop region, light blue and cyt, cytoplasmic domain brown. Expression tags used are TrxA, thioredoxin fusion protein, His, His-tag, TEV, TEV protease cleavage sequence, Tag, chain N contains a Flag-tag (DYKDDDDK sequence) and chain C an N-terminal enterokinase cleavage site (DDDDK). (B) Ribbon presentation of gp41TM-FP in complex with 2H10. Color-coding of the different segments is as indicated in the gp41 scheme (A), the 2H10 nanobody is colored in green. (C) Ribbon presentation of gp41TM-FP including the core six-helical bundle trimer axis (black line) revealing the different orientations of FP and TM. (D) Close-up of the interaction of gp41FP-TM with 2H10. Residues in close enough contact to make polar interactions are shown as sticks. (E, F, G) Ribbon diagram of the individual protomers named chain A, B, and C. Residues within the FPPR and MPER hinge regions are indicated.

Crystal structure of gp41 in complex with 2H10

The structure of gp41FP-TM in complex with 2H10 was solved by molecular replacement to a resolution of 3.8 Å (Figure 1—figure supplement 3; Table 1). The asymmetric unit contained trimeric gp41FP-TM bound to one 2H10 nanobody as indicated by ITC (Figure 1—figure supplement 2B). The six-helix bundle structure composed of three N-terminal and three C-terminal chains is conserved from HR1 residue A541 to HR2 residue L661 in all three protomers, and identical to previous structures (Chan et al., 1997; Weissenhorn et al., 1997). However, TMR and FP do not follow the three-fold symmetry and their chains point into different directions (Figure 1B and C). 2H10 interacts with chain C-A (Figure 1B and D) and stabilizes a partially extended MPER conformation, including a kink at L669 that positions the rest of MPER and TM (N674 to V693) at a 45° angle with respect to the six-helix threefold symmetry axis. The corresponding N-terminal chain A (chain N-A) has its FP disordered and FPPR from G527 to A533 is flexible, while the remaining FPPR and HR1 form a continuous helix (Figure 1E). The protomer A chain C 2H10 epitope spans from residues Q658 to N671, which is involved in a series of polar contacts with 2H10 covering a total buried surface of 712.2 Å2. These include interactions of gp41FP-TM E662 to 2H10 Y37, S668 and the carbonyl of D664 to R56, K665 to E95, N671 and the carbonyl of A667 to R54, K655 to R97 and R93 contacts E95 to position it for interaction with K665 (Figure 1D). Notably, mutations of R56, R93, E95, and R97 have been shown to affect interaction with gp41 (Lutje Hulsik et al., 2013). Chain N-B of the second protomer forms a long continuous helix comprising FP, FPPR and HR1 from residues L518 to D589 with the first six residues of FP being disordered. Likewise, chain C-B folds into a continuous helix from M629 to A700 comprising HR2, MPER and TM (Figure 1F). Cα superimposition of chain C-B with MPER containing gp41 structures (Buzon et al., 2010; Shi et al., 2010) yields root mean-square deviations of 0.55 Å and 0.29 Å (Figure 1—figure supplement 4), indicating that the straight helical conformation is the preferred conformation in threefold symmetrical gp41. In the third protomer, chain N-C has a helical FP linked by flexible residues G531 to A533 to a short helix of FPPR that bends at A541 with respect to helical HR1. Its corresponding chain C-C contains helical HR2 and a flexible region from N671 to N674, which stabilizes a ~ 45° rotation of the remaining MPER-TM helix that extends to residue R707 (Figure 1G and Figure 1—figure supplement 3B). Thus, the structure reveals flexible regions within FPPR and MPER. FPPR flexibility is supported by strictly conserved G528 and G531, while MPER has no conserved glycine residues. However, the same kink within L661 to F673 has been observed in the MPER peptide structure (Sun et al., 2008), and in complex with bnAb 10E8 (Huang et al., 2012). The N-terminal FP residues 512 to 517 are disordered within the detergent micelle. Flexibility of this region in the absence of membrane is consistent with NMR peptide structures that propose a flexible coil structure of the N-terminal part of FP in solution followed by a helical region starting at L518 (Serrano et al., 2017) as observed here. Based on the flexible linkage of FP and TMR, we propose that both FPPR and MPER act as hinges during gp41 refolding leading to membrane fusion.

Table 1
Crystallographic data collection and refinement statistics.
Data collectionGp41FP-TM*
Space groupC 2 2 21
Cell dimensions
a, b, c (Å)96.75, 101.41, 234.42
α, β, γ (°)90, 90, 90
Resolution (Å)48.38–3.8 (3.94–3.8) *
Unique reflexions11179 (631)*
Rmerge†0.23 (1.508)*
 Rp.i.m ‡0.081 (0.548)
 I / σI4.75 (1.74) *
 Completeness (%)78.01 (54.69) *
 Multiplicity9.1 (9.6) *
 CC (1/2)0.992 (0.628) *
Refinement
 Resolution (Å)48.38–3.8 (3.936–3.8)*
 No. reflections9154 (630)*
 Reflections used for Rfree§550 (51)*
 Rwork§ / Rfree**0.265/0.308
 No. atoms
 Protein4440
 Ligand/ion0
 Water0
 Wilson B (Å2)75.8
 Average B-factors (Å2)
 Overall91.76
 Protein91.76
 Ligand/ion
 Water
 R.m.s deviations
 Bond lengths (Å)0.003
 Bond angles (°)0.66
 Ramachandran Plot (%)
 Favored96.65
 Outliers0.37
 PDB ID7AEJ

  1. *Data collected from two crystals were used for structure determination.

    The statistics are for data that were truncated by STARANISO to remove poorly measured reflections affected by anisotropy. Rmerge, Rp.i.m and multiplicity are calculated on unmerged data prior to STARANISO truncation. For comparison, after STARANISO truncation, Rmerge in the resolution shell 3.97 Å - 3.85 Å is 0.787.

  2. † Parentheses refer to outer shell statistics.

    ‡ Rmerge = Σhkl Σi | Ihkl,i- < Ihkl > | / Σhkl ΣiIhkl,i, where Ihkl,i is the scaled intensity of the ith measurement of reflection h, k, l, and <Ihkl > is the average intensity for that reflection.

  3. § R p.i.m. = Σhkl √1/(n-1) Σi | Ihkl,i- < Ihkl > | / Σhkl ΣiIhkl,i,.

    Rwork = Σhkl | Fo - Fc | / Σhkl | Fo | x 100, where Fo and Fc are the observed and calculated structures factors.

  4. ** Rfree was calculated as for Rwork, but on a test set of 5% of the data excluded from refinement.

Table 1—source data 1

Env pseudoviruses.

https://cdn.elifesciences.org/articles/65005/elife-65005-table1-data1-v2.docx

MD simulation of gp41FP-TM in a lipid bilayer

In order to test whether the structure is influenced by the presence of the detergent, we probed its stability by MD simulation in a bilayer having the lipid composition of the HIV-1 envelope. This confirmed that the structure is stable in a membrane environment during a 1 µs simulation as only the flexibly linked FP of chain N-C moves within the bilayer during the simulation (Figure 1—figure supplement 5A). The tip of the 2H10 CDR3 dips into the bilayer (Figure 1—figure supplement 5B), hence confirming the membrane-anchoring role of W100 for neutralization (Lutje Hulsik et al., 2013).

Neutralization activity of 2H10 depends on membrane interaction

The structure suggests that 2H10 stabilized asymmetry within the membrane anchors. Crystal packing effects on the conformation of the membrane anchors can be excluded since the orientation of the membrane anchors is likely not influenced by interprotomer contacts (Figure 1—figure supplement 6). In order to test whether stabilizing the asymmetric conformation is a feature of neutralizing MPER Abs, we further evaluated 2H10 as a neutralizing nanobody. Previously, 2H10 showed only modest neutralization as a bi-head (bi-2H10), whereas neutralization depended on W100 located at the tip of CDR3 (Lutje Hulsik et al., 2013), a hall mark of MPER-specific bnAbs (Cerutti et al., 2017). In order to engineer breadth and potency of monovalent 2H10, we increased its potential membrane interaction capacity by changing CDR3 S100d to F (2H10-F) alone and in combination with additional basic residues S27R, S30K, and S74R (2H10-RKRF) within the putative 2H10 membrane-binding interface suggested by MD simulation (Figure 1—figure supplement 5C). Wild type 2H10 did not show significant neutralization against a panel of 10 clade B pseudo-viruses as reported previously (Lutje Hulsik et al., 2013), with the exception of some weak neutralization of NL4-3 and SF163P3. However, both 2H10-F and 2H10-RKRF show improved potency and breadth neutralizing six and eight pseudo-viruses, respectively, albeit with less potency than wild-type bi-2H10 and bnAb 2F5, the latter recognizing an overlapping epitope (Table 2). Notably, enhancing the membrane interaction surface did not increase the detection of non-specific membrane binding in vitro (Figure 1—figure supplement 7). This result confirms monovalent 2H10 as a modest anti-MPER Ab that neutralizes by engaging MPER and the membrane.

Table 2
Pseudovirus neutralization by 2H10, 2H10-F, 2H10-RKRF, and bi-2H10 in comparison to 2F5 and VRC01.

IC50s are indicated in µg/ml.

Tier2H10 wt2H10-F2H10-RKRFBi-2H102F5VRC01
NL4-3125.2018.689.151.840.160.20
MN-31>50.0030.389.361.390.030.06
BaL.261>50.0019.389.636.051.210.13
SF1621a>50.00>50.0025.196.141.220.39
SF162P3222.0413.146.761.321.960.24
SC422661.82>50.00>50.0027.933.791.000.27
JR-FL244.6516.936.951.490.970.11
JR-CSF2>50.0021.6610.852.851.240.37
QH0692.422>50.00>50.00>50.00>50.001.201.21
THRO4156.182>50.00>50.00>50.00>50.00>50.003.84

2H10 blocks fusion before the stage of lipid mixing

The efficacy of bi-2H10 and 2H10-RKRF for blocking membrane merging was further assessed in peptide-induced lipid-mixing assays (Figure 2). In the experimental setting (Apellániz et al., 2014b), a vesicle population is primed for fusion by addition of the N-MPER peptide containing the 2H10 epitope, which produces a fluorescence intensity spark at time 20 s (Figure 2A). Under these experimental conditions, incorporation of the peptide into the vesicles takes less than 10 s. After 120 s, the mixture is supplemented with target vesicles fluorescently labeled with N-NBD-PE/N-Rh-PE (indicated by the arrow in Figure 2A). The increase in NBD intensity as a function of time follows the mixing of the target vesicle lipids with those of the unlabeled vesicles (kinetic trace labeled ‘+N-MPER’), a process not observed when labeled target vesicles are injected in a cuvette containing unlabeled vesicles not primed with peptide (‘no peptide’ trace).

Figure 2 with 1 supplement see all
Download asset Open asset
Vesicle-vesicle fusion inhibition by 2H10-RKRF, bi-2H10, 2F5, and 10E8.

(A) Time course of the lipid-mixing assay using fusion-committed vesicles. At time 30 s (‘+N-MPER’), peptide (4 µM) was added to a stirring solution of unlabeled vesicles (90 µM lipid), and, after 120 s (indicated by the arrow), the mixture was supplemented with N-NBD-PE/N-Rh-PE-labeled vesicles (10 µM lipid). The increase in NBD fluorescence over time follows the dilution of the probes upon mixing of lipids of target and primed vesicles (+N MPER trace). NBD increase was substantially diminished in samples incubated with bi-2H10 (2 µg/ml) prior to the addition of the target vesicles (+bi-2H10, dotted trace), and totally absent if unlabeled vesicles were devoid of peptide (‘no peptide’ trace). (B) Kinetic traces of N-MPER-induced lipid-mixing comparing the blocking effects of 2H10-RKRF, bi-2H10, Fab 2F5 and Fab 10E8. (C) Absence of effects on lipid-mixing of 2H10-RKRF, bi-2H10 and Fab 2F5 when vesicles were primed for fusion with the C-MPER peptide, devoid of 2H10 and 2F5 epitope sequences. The positive control Fab 10E8 efficiently blocked the process. Antibody concentrations were 20 µg/ml in these assays. (D) Dose-response plots comparing the inhibitory capacities of 2H10-RKRF and bi-2H10 (purple and green traces, respectively). Levels of lipid-mixing 20 or 300 s after target vesicle injection were measured (initial rates D and final extents, E) and percentages of inhibition calculated as a function of the Ab concentration. The dotted line and empty symbols correspond to the effect of bi-2H10 when the concentration of the component 2H10 was plotted. The slashed vertical lines mark the 2H10-to-peptide ratios of 1:6 and 1:3. Plotted values are means ± SD of three independent experiments.

MPER antibodies have been shown to interact directly with membrane-bound peptide epitopes, thereby changing their insertion state (depth and oligomerization levels) and inhibiting their capacity to induce fusion (Lorizate et al., 2006; Sun et al., 2008). Accordingly, lipid mixing was strongly attenuated when the vesicles primed for fusion with N-MPER were incubated with bi-2H10 before addition of the target vesicles (Figure 2A,+N-MPER/+bi-2H10, dotted trace). Thus, the N-MPER-induced membrane perturbations, which can induce fusion with target membranes, were inhibited by incubation with bi-2H10. Comparison of the kinetics of the lipid-mixing blocking effect of 2H10-RKRF, bi-2H10 and Fabs 2F5 and 10E8 showed that the four antibodies inhibited both the initial rates and final extents of lipid mixing induced by N-MPER (Figure 2B). Using a control MPER peptide lacking the 2H10 and 2F5 epitopes for vesicle priming no inhibition of lipid mixing by 2H10-RKRF, bi-2H10 and 2F5 Fab was observed while 10E8 still blocked lipid mixing, (Figure 2C) corroborating that the inhibitory effects depend on epitope recognition. The use of the 10E8 Fab as an additional control further supports the blocking effect mediated by epitope recognition. Notably, Fab 10E8 inhibits lipid mixing in a concentration-dependent manner (Figure 2—figure supplement 1A). Furthermore, inhibition of lipid mixing depends on direct membrane interaction since increasing hydrophobicity within the 10E8 Fab-membrane interface increases the inhibitory effect and decreasing 10E8 hCDR3 hydrophobicity abrogates lipid mixing (Figure 2—figure supplement 1B and C), which is in agreement with neutralization determinants of 10E8 (Chen et al., 2014; Rujas et al., 2018). The complete 10E8 epitope encompasses C-terminal MPER residues plus N-terminal TM residues, the latter not included in the sequence covered by the N-MPER peptide. Consistent with this recognition pattern, 10E8 blocked C-MPER-induced fusion more efficiently than N-MPER-induced fusion, but was effective against both peptides. The capacity for blocking the lipid-mixing process also correlates with the neutralization potency of MPER bnAbs (Apellániz et al., 2014b). Following this trend, fusion inhibition levels estimated as a function of the antibody concentration confirmed the apparent higher potency exhibited by the bi-2H10 (Figure 2D). Lower concentrations of bi-2H10 compared to 2H10-RKRF were required to attain full blocking of the lipid-mixing process when measured 20 s (initial rates) or 300 s (final extents) after target-vesicle injection (Figure 2D and E). The higher inhibitory potency of bi-2H10 indicates an avidity effect, which was also evident when the concentration of the epitope-binding fragments was plotted (Figure 2D and E, empty squares and dotted line). Moreover, bi-2H10 appeared to block the process efficiently even at 2H10:N-MPER ratios below 1:3 (mol:mol), consistent with the involvement of peptide oligomers in the promotion of membrane fusion. Based on these data, we suggest that both 2H10-RKRF and bi-2H10 neutralize HIV-1 at the stage of lipid mixing.

GP41FP-TM interaction with MPER bnAbs

Although the 2H10 epitope overlaps with the 2F5 MPER epitope (Ofek et al., 2004), the 2F5-bound peptide structure (Ofek et al., 2004), cannot be superimposed without major clashes with adjacent gp41 protomers. In contrast, Cα superposition of the structures of 10E8 and LN01 in complex with MPER peptides demonstrated possible binding to gp41FP-TM chain C-C (Figure 3A and B). Furthermore, HCDR3 of both 10E8 and LN01 could make additional hydrophobic contacts with adjacent FP in this binding mode. To confirm 10E8 and LN01 interaction, we performed immunoprecipitation of gp41FP-TM with both bnAbs, which confirmed their interaction in vitro (Figure 3—figure supplement 1). We next validated binding by bio-layer interferometry (BLI) using gp41FP-TM as analyte. This revealed KDs of 0,2 nM for 10E8 and 34 nM for LN01 (Figure 3C and D). We conclude that bnAbs 10E8 and LN01 interact with gp41FP-TM with high affinity which may also stabilize an asymmetric gp41 conformation similar to the one observed in complex with 2H10 as suggested by the structural modeling (Figure 3A and B).

Figure 3 with 1 supplement see all
Download asset Open asset
Gp41FP-TM interaction with bnAbs LN01 and 10E8.

(A) Cα superposition of the MPER peptide structure in complex with LN01 (pdb 6snd) onto chain C-C of the gp41FP-TM-2H10 structure. The lower panel shows a close-up of the interaction oriented with respect to gp41 F673. (B) Cα superposition of the MPER peptide structure in complex with 10E8 (pdb 5iq7) onto the corresponding chain C-C of gp41FP-TM. The lower panel shows a close-up of the interaction in the same orientation as in A. (C) Bio-layer interferometry (BLI) binding of gp41FP-TM to 10E8 and (D) to LN01. 10E8 binding was fit to 1:1 model and for LN01 a steady state model was employed for fitting the data. For 10E8 binding, gp41FP-TM was used at concentrations from 0.2 to 25.6 nM and for LN01 binding gp41FP-TM concentrations ranged from 39 to 625 nM.

Building a post-fusion conformation by MD simulation

In order to follow the final refolding of the membrane anchors, we modeled the post fusion conformation employing MD simulation. Assuming that the final post-fusion conformation shows a straight symmetric rod-like structure we constructed a model of gp41 from the protomer composed of the straight helical chains N-B and C-B (Figure 4—figure supplement 1A and B). This conformation is also present in the symmetric six-helix bundle structures containing either MPER (Shi et al., 2010) or FPPR and MPER (Buzon et al., 2010; Figure 1—figure supplement 3). In this model, FP and TM do not interact tightly (Figure 4—figure supplement 1B), which, however, does not explain the increased thermostability in the presence of FP and TM (Figure 1—figure supplement 2A). One-µs MD simulation of this model (Figure 4—figure supplement 1B) in solution, rearranges the membrane anchors such that they adopt a compact structure with trimeric FP interacting with adjacent TMs. Furthermore, the TMs kink at the conserved Gly positions 690 and 691, as observed previously (Pinto et al., 2019; Figure 4—figure supplement 1C). In order to recapitulate the stability of the model in the membrane, we performed an additional 1-µs MD simulation of the model (Figure 4—figure supplement 1C) in a bilayer resembling the HIV-1 lipid composition, which relaxed the TM to its straight conformation (Figure 4A). The final structural model reveals tight packing of trimeric FP flexibly linked to HR1 by FPPR G525 to G527 (Figure 4B). HR2-MPER and TMR form continuous helices with the TMRs packing against trimeric FP (Figure 4A and C), which spans one monolayer (Figure 4A). As conserved tryptophan residues within MPER have been previously implicated in fusion (Bellamy-McIntyre et al., 2007; Salzwedel et al., 1999), we analyzed their structural role in the post fusion model. This reveals that the indole ring of W666 is sandwiched between Leu669 and T536 and packs against L537. W670 makes a coiled-coil interaction with S534, while W672 is partially exposed and packs against L669 and T676. W678 binds into a hydrophobic pocket defined by I675, L679, I682 and adjacent FP/FPPR residues F522 and A526. W680 is partially exposed, but reaches into a pocket created by the flexible FPPR coil (Figure 4—figure supplement 2). We therefore propose that most of the tryptophan residues have structural roles in the post-fusion conformation, hence providing an explanation for their functional role in fusion (Salzwedel et al., 1999). The MPER epitopes recognized by 10E8 and LN01 are exposed in the post-fusion model, but antibody docking to this conformation produced major clashes, consistent with no expected binding to the final post fusion conformation.

Figure 4 with 2 supplements see all
Download asset Open asset
Interactions within the final post fusion conformation of gp41FP-TM modeled by MD.

(A) Model of gp41FP-TM (Figure 1—figure supplement 7C) after 1µs MD simulation in a bilayer. Phosphate groups of the phospholipids are shown as orange spheres to delineate the membrane boundaries. (B) Close-up on the MPER and FPPR flexible regions. (C) Close-up of the interaction of FP (residues 514–524) and TM (residues 681–692) viewed along the three-fold axis from the N-terminus indicating an intricate network of hydrophobic interactions (left panel) and from the side (right panel). Interacting side chains are labeled and shown as sticks.

Structural transitions of gp41

A number of Env SOSIP structures revealed the native conformation of gp41 (Figure 5A, B and C; Kumar et al., 2019; Pancera et al., 2014; Yuan et al., 2019). The gp41FP-TM crystal structure and the model of its post-fusion conformation provide further insight into the path of conformational changes that native gp41 must undergo to adopt its final lowest energy state conformation. The first major conformational changes in gp41 that take place upon receptor binding are extension of HR1 and FPPR into a triple stranded coiled coil with flexible linkers connecting FPPR to FP that may initially interact laterally with the external leaflet of the bilayer thereby projecting FP ~115 Å away from its starting position (Figure 5D). Notably, such an early intermediate fusion conformation structure has been reported for influenza hemagglutinin (HA) (Benton et al., 2020). This is likely followed by an extension and rearrangement of HR2 and MPER producing 11–15 nm long intermediates that connect the viral and cellular membranes (Ladinsky et al., 2020; Lai et al., 2014). Gp41 refolding into the six-helix bundle structure then produces flexibly linked asymmetric conformations of FPPR-FP and MPER-TM anchored in the cellular and viral membranes, respectively, as indicated by the gp41FP-TM structure. This intermediate conformation may bring viral and cellular membranes into close proximity (Figure 5E) or may act at the subsequent stage of hemifusion (Figure 5F). Further refolding and interaction of FP-FPPR and MPER-TM may generate the stable post fusion conformation (Figure 5G), a process that completes membrane fusion. Notably, in this model based on the crystal structure, FP spans only one leaflet of the lipid bilayer facilitating local lipid head group interactions with the N-terminus of FP.

Figure 5
Download asset Open asset
Conformational transitions of gp41 that lead to membrane apposition and membrane fusion.

(A) Representation of the different domains of gp41 with the residue numbers delimiting each domain as indicated. The same color code has been used in all the figures. (B) Ribbon presentation of the Env prefusion conformation (pdb 5fuu), gp41 is constrained by gp120 in its native conformation. The structure of native gp41 lacks the MPER and TM regions. MPER is spanning a distance of 1.5 nm (Li et al., 2020). (C) Ribbon of native gp41, one chain is colored according to the sheme in A and the other two chains are shown in gray. (D) Binding to cellular receptors CD4 and subsequently to CXCR4/CCR5 induces a series of conformational changes that eventually leads to the dissociation of gp120. During this process, HR1, FPPR and FP will form a long triple stranded coiled coil extending 11 nm toward the target cell membrane. In a first step, HR2 may keep its prefusion conformation in analogy to a similar intermediate, activated influenza virus HA structure (Benton et al., 2020). Alternatively, HR2 may dissociate and form a more extended conformation in agreement with locked gp41 structures bridging viral and cellular membranes that bridge distances of 11–15 nm (Ladinsky et al., 2020). (E) Bending of HR1 and HR2 will result in the six-helical bundle core structure bringing cellular and viral membranes into close apposition with the three FPs anchored in the cellular membrane and the three TMs anchored in the viral membrane, the gp41 conformation represented by the gp41FP-TM structure. This intermedaite gp41 conformation may have brought both membranes into close apposition or may have already induced hemifuison as indicated in F. (G) Further reolding of FPPR-FP and MPER-TM results in the final extremely stable post-fusion conformation. This suggests that rearrangment of the membrane anchors plays crucial roles in lipid mixing, breaking the hemifusion diaphragm to allow fusion pore opening. Boundaries of the lipid layers are shown with orange spheres representing the phosphate atomes of the lipids present in the MD simulation (snapshop taken after 1µs MD simuation).

Discussion

Membrane fusion is an essential step of infection for enveloped viruses such as HIV-1, and requires extensive conformational rearrangements of the Env prefusion conformation (Julien et al., 2013; Lyumkis et al., 2013; Pancera et al., 2014) into the final inactive post-fusion conformation (Harrison, 2008; Weissenhorn et al., 2007). The fusion model predicts that six-helix bundle formation apposes viral and cellular membranes with FP and TM inserted asymmetrically in the cellular membrane and the viral membrane, respectively (Weissenhorn et al., 1999). Here, we show that gp41 containing its membrane anchors can adopt this predicted conformation, which is facilitated by flexible hinges present in FPPR and MPER corroborating their essential roles in membrane fusion (Blumenthal et al., 2012; Chen, 2019; Harrison, 2008). The asymmetric conformation of the membrane anchors suggest further that bundle formation occurs before pore formation as suggested previously (Markosyan et al., 2003; Melikyan et al., 2000). The membrane-fusion model proposes further that final steps in fusion involves rearrangement and interaction of TM and FP (Weissenhorn et al., 1999), which is confirmed by the MD-simulation model of the post-fusion conformation. Furthermore, the length of the rod-like post-fusion structure of 13 nm lacking the C-C loop is consistent with the gp41 structure lacking FP and TM (Weissenhorn et al., 1996).

FP is helical in the gp41FP-TM-2H10 complex and the MD-based post-fusion conformation, in agreement with NMR-based helical FP peptide models (Li and Tamm, 2007; Serrano et al., 2014), although β-strand structures of FP have been implicated in fusion as well (Apellániz et al., 2014a). In comparison, in native Env conformations, FP adopts multiple dynamic conformations that are recognized by broadly neutralizing antibodies (Blattner et al., 2014; Kong et al., 2016; Kumar et al., 2019; Yuan et al., 2019). In the post-fusion conformation, FP spans one monolayer of the membrane, in contrast to suggested amphipathic helix-like interaction of FP with the outer layer of the target cell membrane, which may constitute a first contact with the cellular membrane (Han et al., 2001; Lai et al., 2012).

Furthermore, the coiled-coil interactions within FP and with TM in the post-fusion model explain the increased thermostability of gp41FP-TM compared to gp41 lacking FP and TM (Buzon et al., 2010). We propose that refolding of FP and TM can liberate additional free energy to catalyze final steps of fusion. Hence, replacement of TM by a phosphatidylinositol (PI) anchor inhibits membrane fusion (Salzwedel et al., 1993; Weiss and White, 1993), akin to the GPI-anchored HA inhibition of influenza virus membrane fusion at the stage of hemifusion (Kemble et al., 1994).

Combinatorial mutagenesis of the conserved tryptophan residues within MPER was shown to affect syncytium formation, although individual mutations had no effect (Muñoz-Barroso et al., 1999; Salzwedel et al., 1999). However, in another study a W666A mutation was shown to affect infection (Bellamy-McIntyre et al., 2007), which was further enhanced in combination with a I675A mutation albeit without effect on syncytium formation (Narasimhulu et al., 2018). Although it is not known at which stage of the fusion process the concerted action of the tryptophan residues is required, our structural model suggests that W666, W670, and W678 likely play a structural role in stabilizing the post-fusion conformation. Furthermore, combinatorial mutations to leucine of the TM region comprising residues 683–693 affect virus infectivity (Shang et al., 2008) in agreement with the proposed interaction of this part of the TM with FP in the post fusion model. In addition, mutation of the conserved arginine at position 693 reduces fusion efficiency (Long et al., 2011), which however, may be related to its role in stabilizing a tilted TM conformation in the bilayer during the fusion process (Chiliveri et al., 2018; Pinto et al., 2019). Our structural model of the post-fusion conformation is further in agreement with proposed interactions of FPPR and MPER, as well as FP and TM (Louis et al., 2016; Reuven et al., 2012), thereby corroborating their essential roles at late stages of membrane fusion.

Gp41FP-TM interaction with the 2H10 MPER-specific nanobody stabilizes the asymmetric conformation of the membrane anchors. In order to confirm that 2H10 is, indeed, a neutralizing MPER-specific nanobody, we engineered increased 2H10 membrane binding, which improved breadth and potency of 2H10 neutralization, in agreement with enlarged potency by increasing membrane-binding of 10E8 (Chen et al., 2014; Kwon et al., 2018; Rujas et al., 2018). This confirmed 2H10 as a modest anti-MPER neutralizing antibody, whose neutralization potency is not only enhanced by using bi-2H10 with increased avidity (Lutje Hulsik et al., 2013), but also by increasing potential membrane interaction of the 2H10 monohead. Consistent with its neutralization capacity, 2H10 inhibits membrane fusion at the stage of lipid mixing like 2F5 and other anti-MPER bnAbs (Apellániz et al., 2014b; Huarte et al., 2008). Moreover, gp41FP-TM interacts with MPER bnAbs 10E8 (Huang et al., 2012) and LN01 (Pinto et al., 2019) in agreement with docking both structures onto the kinked protomer C chain C MPER epitope. Notably, the kink in the MPER peptide in complex with 10E8 (Huang et al., 2012) is similar to the protomer C chain C kink and present in MPER peptide NMR structures (Serrano et al., 2014; Sun et al., 2008). Furthermore, Ala mutations in the kink (671-674) affect cell-cell fusion and lower virus infectivity (Sun et al., 2014) corroborating the physiological relevance of the kinked conformation. We therefore propose that 10E8 and LN01 binding to gp41FP-TM stabilizes similar asymmetry, as observed in the gp41FP-TM-2H10 structure by sampling the dynamic states of the membrane anchors.

Our data indicate that MPER antibodies can act all along the gp41 refolding pathway from blocking initial conformations of close to native Env (Carravilla et al., 2019; Lee et al., 2016; Rantalainen et al., 2020) up to a late fusion intermediate state that has already pulled viral and cellular membranes into close apposition. This opens a long temporal window of action for MPER bnAbs consistent with the findings that the half-life of neutralization of MPER bnAbs is up to 30 min post virus exposure to target cells (Shen et al., 2010; Williams et al., 2017). Furthermore, only one Ab per trimer may suffice to block final refolding of the membrane anchors required for fusion. Finally, the presence of dynamic linkers connecting the core of viral fusion proteins with their membrane anchors FP and TM may be a general feature of all viral membrane fusion proteins including the proposed flexible linkage of influenza virus HA FP to the central HA2 coiled coil (Benton et al., 2020).

Materials and methods

Key resources table
Reagent type
(species) or
resource		DesignationSource or
reference		IdentifiersAdditional
information
Recombinant DNA reagentChain N pETM20 (plasmid)This studypETM20, PEPcore facility-EMBL
Recombinant DNA reagentChain C pETM11
(plasmid)		This studypETM11, PEPcore facility-EMBL
Recombinant DNA reagent2H10 pAX51
(plasmid)		Lutje Hulsik et al., 2013PMID:23505368
Recombinant DNA reagent2H10-F pAX51
(plasmid)		This studyS100d to F mutation
Recombinant DNA reagent2H10-RKRF pAX51
(plasmid)		This studyS27R, S30K, S74R and S100d to F mutations
Strain, strain background (Escherichia coli)BL21(DE3)ThermoFisherCat# EC0114
Strain, strain background (Escherichia coli)C41(DE3)LucigenCat#60442
Cell line
(Homo sapiens)		TZM-blNIH AIDS Reagent ProgramCat#ARP-8129
RRID:CVCL_B478
Cell line
(Homo sapiens)		HEK293TATTCCat# CRL-11268
RRID:CVCL_1926
Antibody
(human monoclonal)		α-gp41, LN01Pinto et al., 2019PMID:31653484See Materials and methods
Antibody
(human monoclonal)		α-gp41, 10E8Huang et al., 2012PMID:23151583
RRID:AB_2491067		See Materials and methods
Antibody
(human monoclonal)		α-gp41, 2F5Muster et al., 1993PMID:7692082See Materials and methods
Antibody
(human monoclonal)		α-gp140, VRC01Zhou et al., 2010PMID:20616231
RRID:AB_2491019		See Materials and methods
Antibody
(llama nanobody)		α-gp41, 2H10Lutje Hulsik et al., 2013PMID:23505368See Materials and methods
Antibody
(llama nanobody)		α-gp41, bi-2H10Lutje Hulsik et al., 2013PMID:23505368See Materials and methods
Antibody
(llama nanobody)		α-gp41, 2H10-FThis studySee Materials and methods
Antibody
(llama nanobody)		α-gp41, 2H10-RKRFThis studySee Materials and methods
Chemical compound, drugChelating sepharose
FF		GE HealthcareCat# 17057501
Chemical compound, drugQ sepharose FFGE HealthcareCat# 17051010
Chemical compound, drugQ sepharose FFGE HealthcareCat# 17051010
Chemical compound, drugn-Octyl-β-D-glucosid n-Octyl-β-D-GlucopyranosideAnatraceCat#O311
Chemical compound, drugCHAPS (3-[(3-cholamidopropyl) diméthylammonio]−1-propanesulfonate)EuromedexCat#1083E
Chemical compound, drug1-palmitoyl-2-oleoylphosphatidylcholine
(POPC)		Avanti Polar LipidsCat#850457P
Chemical compound, drug1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine
(POPE)		Avanti Polar LipidsCat#850757P
Chemical compound, drug1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-L
-serine
(POPS)		Avanti Polar LipidsCat#840034C
Chemical compound, drugsphingomyelinAvanti Polar LipidsCat#860062C
Chemical compound, drugcholesterolAvanti Polar LipidsCat#700000P
Chemical compound, drugN-(7-nitro-benz-2-oxa-1,3-diazol-4-yl)phosphatidylethanolamine (N-NBD-PE)Molecular ProbesCat#N360
Chemical compound, drugN-(lissamine Rhodamine B sulfonyl) phosphatidylethanolamine (N-Rh-PE)Molecular ProbesCat#L1392
Chemical compound, drugBiscinchoninic Acid microassayPierceCat# 23235
Chemical compound, drugBio-Rad Protein Assay Dye Reagent Concentrate
BioradCat# 5000006
Software, algorithmXDSKabsch, 2010PMID:20124693
Software, algorithmPhaserMcCoy et al., 2007PMID:19461840
Software, algorithmCOOTEmsley et al., 2010PMID:20383002
Software, algorithmPHENIXAdams et al., 2010PMID:20124702
Software, algorithmSBGridMorin et al., 2013https://sbgrid.org/
Software, algorithmPymolWarren DeLanohttp://www.pymol.org
Software, algorithmAimlessEvans and Murshudov, 2013PMID:23793146
Software, algorithmSTARANISOTickle et al., 2018http://staraniso.globalphasing.org/cgi-bin/staraniso.cgi
Software, algorithmCHARMM-GUIJo et al., 2008http://www.charmm-gui.org
Software, algorithmNAMD (Version 2.13)Phillips et al., 2005https://www.ks.uiuc.edu/Research/namd/
Software, algorithmPrism 8GraphPadhttps://www.graphpad.com/scientific-software/prism/
Software, algorithmForteBio analysis software version 11.1.0.25ForteBiohttps://www.fortebio.com
Software, algorithmMicroCal Origin software (origin 7)Malvern Panalytical
(MicroCal)		https://www.malvernpanalytical.com
Peptide, recombinant proteinNEQELLELDKWASLW NWFNITNWLWYIK (N-MPER)This studySee Materials and methods
Peptide, recombinant proteinKKK-NWFDITNWLWYIKLFIMIVGGLV-KK (C-MPER),This studySee Materials and methods
Commercial assay or kitBiscinchoninic Acid microassayPierceCat# A53225
Commercial assay or kitBright-Glo Luciferase Assay System
Streptavidin (SA) biosensors		PromegaCat# E2610
Commercial assay or kitStreptavidin (SA) biosensorsForteBioCat#18–5019
Commercial assay or kitEZ-LinkNHS-PEG4-Biotinylation KitThermoFisherCat#21455

Cell lines

Request a detailed protocol

HEK 293 T cells were obtained from the American Type Culture Collection (ATCC CRL-11268).

TZM-bl cells were obtained through the NIH HIV Reagent Program, Division of AIDS, NIAID, NIH: TZM-bl Cells, ARP-8129, contributed by Dr. John C. Kappes and Dr. Xiaoyun Wu. HEK 293T and TZM-bl cells were maintained in Dulbecco’s modified Eagle’s medium supplemented with 10% heat-inactivated fetal bovine serum, 100 units/ml of Penicillin and 0.1 mg/ml of Streptomycin (all medium components from Gibco, Thermo Fisher Scientific) at 37°C, 5% CO2, and 80% relative humidity. Both cell lines are regularly checked for absence of Mycoplasma (Mycoplasmacheck test by Eurofins Genomics (formerly GATC), Germany). No specific cell line authentication was performed.

HIV-1 primary viruses

Request a detailed protocol

Env-pseudotyped viruses were prepared by co-transfection of HEK 293 T cells with plasmids encoding the respective env genes and the luciferase reporter HIV vector pNLluc-AM as described (Rusert et al., 2009). A full list of Env pseudotyped viruses generated with corresponding gene bank entry, subtype and Tier information is provided in Table 1—source data 1.

GP41 expression and purification

Request a detailed protocol

DNA fragments encoding HIV-1 Env glycoprotein amino acids 512–594 (N-terminal chain, chain N) and residues 629 to 716 (C-terminal chain, chain C) were cloned into vectors pETM20 and pETM11 (PEPcore facility-EMBL), respectively. Chain N contains an N-terminal Flag-tag (DYKDDDDK sequence) and chain C an N-terminal enterokinase cleavage site (DDDDK) and two C-terminal arginine residues (Figure 1A). The sequence of chain N is GAMDYKDDDDK-512-AVGIGALLLGFLGAAGSTMGAASMTLTVQARQLLSGIVQQQNNLLRAIEAQQHLLQLTVWGIKQLQARILAVERYLKDQQLLG-594 and of chain C is GAMDDDDK-629-MEWDREINNYTSLIHSLIEESQNQQEKNEQELLELDKWASLWNWFNITNWLWYIKLFIMIVGGLVGLRIVFAVLSVVNRVRQGYSPLS-716-RR after TEV protease cleavage of the constructs. The N-MPER peptide region used in the fusion assay is underlined and the C-MPER peptide region is shown in bold.

Proteins were expressed separately in E. coli strain C41(DE3)(Lucigen). Bacteria were grown at 37°C to an OD600nm of 0,9. Cultures were induced with 1 mM IPTG at 37°C for 3 hr for gp41 chain N and at 25°C for 20 hr for gp41 chain C. Cells were lysed separately by sonication in buffer A containing 20 mM Tris pH 8, 100 mM NaCl and 1% CHAPS (3-[(3-cholamidopropyl) diméthylammonio]−1-propanesulfonate (Euromedex)). The supernatant was cleared by centrifugation at 53,000 g for 30 min. Gp41 chain N supernatant was loaded on a Ni2+-sepharose column, washed successively with buffer A containing 1M NaCl and 1M KCl, then buffer A containing 50 mM imidazole. Gp41 chain N was eluted in buffer A containing 500 mM imidazole and further purified on a Mono Q column in buffer A and eluted with buffer B (buffer A with 0.5 M NaCl). Gp41 chain C was purified employing the same protocol as for gp41 chain N. Gp41 chain C was subsequently cleaved with TEV (Tobacco Etch Virus) protease for 2 hr at 20°C and then overnight at 4°C at a gp41 chain C:TEV ratio of 5:1 (w:w). After buffer exchange with a mono Q column using buffer A and elution in buffer B, uncleaved gp41 chain C was removed by a second Ni2+-sepharose column in buffer A. TEV-cleaved gp41 chain C and un-cleaved gp41 chain N were then mixed in a molar ratio 4:1 and incubated overnight. To remove the excess of gp41 chain C, the gp41 complex was further purified on a 3rd Ni2+-sepharose column in buffer A, washed with buffer A containing 50 mM imidazole and eluted with buffer A containing 500 mM imidazole. Subsequently the gp41 chain N TrxA-His-tag was removed by TEV digestion for 2 hr at 20°C and overnight at 4°C at a gp41:TEV ratio of 5:1 (w:w). Followed by another buffer exchange with a mono Q column (buffers A and B), uncleaved complex and the TrxA-His-tag fusion were removed by a 4th Ni2+-sepharose column. The final gp41FP-TM complex was concentrated and loaded onto a Superdex 200 size exclusion column (SEC) in buffer C containing 20 mM Tris pH 8.0, 100 mM NaCl and 1% n-octyl β-D-glucopyranoside (Anatrace).

Nanobody 2H10 expression

Request a detailed protocol

2H10 encoding DNA was cloned into the vector pAX51 (Lutje Hulsik et al., 2013) and expressed in the E. coli BL21(DE3) strain (ThermoFisher). Bacteria were grown at 37°C to an OD600nm of 0.7 and induced with 1 mM IPTG at 20°C for 20 hr. After harvesting by centrifugation, bacteria were resuspended in lysis buffer containing 20 mM Hepes pH 7.5 and 100 mM NaCl. Bacteria were lysed by sonication and centrifuged at 48,000 g for 30 min. Cleared supernatant was loaded onto a Protein A sepharose column, washed with lysis buffer and eluted with 0.1 M glycine pH 2.9. Eluted fractions were immediately mixed with 1/5 vol of 1M Tris pH 9.0. 2H10 was then further purified by SEC on a superdex 75 column in PBS buffer. Genes corresponding to mutants of 2H10, 2H10-F (S100d) and 2H10-RKRF (S27R, S30K, S74R, and S100d) were synthesized (Biomatik) and the expressed mutant proteins were purified as described for the wild type. The 2H10 bi-head was purified as described (Lutje Hulsik et al., 2013).

Circular dichroism

Request a detailed protocol

CD measurements were performed using a JASCO Spectropolarimeter equipped with a thermoelectric temperature controller. Spectra of gp41-TM were recorded at 20°C in 1 nm steps from 190 to 260 nm in a buffer containing PBS supplemented with 1% n-octyl β-D-glucopyranoside. For thermal denaturation experiments, the ellipticity was recorded at 222 nm with 1°C steps from 20° to 95°C with an increment of 80°C h−1, and an averaging time of 30 s/step. For data analysis, raw ellipticity values recorded at 222 nm were converted to mean residue ellipticity.

Isothermal titration calorimetry (ITC)

Request a detailed protocol

The stoichiometry and binding constants of 2H10 binding to gp41 FP-TM was measured by ITC200 (MicroCal Inc). All samples used in the ITC experiments were purified by SEC in a buffer containing 20 mM Tris pH 8.0, 100 mM NaCl and 1% n-octyl β-D glucopyranoside and used without further concentration. Samples and were equilibrated at 25°C before the start of the experiment. The ITC measurements were performed at 25°C by making 20 2 μl injections of 267 µM 2H10 to 0.2 ml of 19.5 µM gp41FP-TM. Curve fitting was performed with MicroCal Origin software. Three experiments were performed, with an average stoichiometry N = 1.1 +/- 0.2. 2H10 binds to gp41FP-TM with a KD of 2.1 µM ± 0.9.

Bio-layer interferometry binding analysis

Request a detailed protocol

Binding measurements between antibodies (10E8 IgG, LN01 IgG, and 2H10) were carried out on an Octet Red instrument (ForteBio). For the determination of the binding between antibodies and gp41FP-TM, 10E8 IgG or LN01 IgG or 2H10 were labeled with biotin (EZ-Link NHS-PEG4-Biotin) and bound to Streptavidin (SA) biosensors (ForteBio). The biosensors loaded with the antibodies were equilibrated in the kinetic buffer (20 mM Tris pH 8.0, 100 mM NaCl and 1% n-octyl β-D glucopyranoside) for 200–500 s prior to measuring association with different concentrations of gp41FP-TM for 100–200 s at 25°C. Data were analyzed using the ForteBio analysis software version 11.1.0.25 (ForteBio). For 10E8, the kinetic parameters were calculated using a global fit 1:1 model and 2:1 model. For the determination of the binding of LN01 IgG and 2H10, KDs were estimated by steady state analysis. All bio-layer interferometry experiments were conducted at least three times.

Immunoprecipitation of gp41FP-TM by bnAbs 10E8 and LN01

Request a detailed protocol

A total of 220 µg of Gp41FP-TM were incubated alone or with 50 µg of 10E8 or LN01 antibodies for 10 hr at 20°C in buffer C. The complex was loaded on Protein A sepharose affinity resin and incubated for 1 hr. The resin was subsequently washed three times with buffer C and eluted with SDS gel loading buffer and boiling at 95°C for 5 min. Samples were separated on a 15% SDS-PAGE and stained with Coomassie brilliant blue.

Neutralization assay

Request a detailed protocol

The neutralization activity of the 2H10 variants and mAbs (IgG) 2F5 (Muster et al., 1993) and VRC01 (Zhou et al., 2010) was evaluated using TZM-bl cells and Env pseudotyped viruses as described (Rusert et al., 2009). Briefly, serial dilutions of inhibitor were prepared in cell culture medium (DMEM with 10% heat-inactivated FBS, 100 U/ml penicillin and 100 μg/ml streptomycin (all from Gibco)) and added at a 1:1 vol ratio to pseudovirus suspension in 384 well plates (aiming for 500,000–5,000,000 relative light units (RLU) per well in the absence of inhibitors). After one-hour incubation at 37°C, 30 µl of virus-inhibitor mixture was transferred to TZM-bl cells in 384 well plates (6000 cells/well in 30 µl cell culture medium supplemented with 20 µg/ml DEAE-Dextran seeded the previous day). The plates were further incubated for 48 hr at 37°C before readout of luciferase reporter gene expression on a Perkin Elmer EnVision Multilabel Reader using the Bright-Glo Luciferase Assay System (Promega).

The inhibitor concentration (referring to the mix with cells, virus, and inhibitor) causing 50% reduction in luciferase signal with respect to a reference well without inhibitor (inhibitory concentration IC50) was calculated by fitting a non-linear regression curve (variable slope) to data from two independent experiments using Prism (GraphPad Software). If 50% inhibition was not achieved at the highest inhibitor concentration tested, a greater than value was recorded. To control for unspecific effects, all inhibitors were tested for activity against MuLV envelope pseudotyped virus.

2H10 membrane interaction

Request a detailed protocol

Lipids, 1-palmitoyl-2-oleoylphosphatidylcholine (POPC), 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine (POPE), 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-L-serine (POPS), Sphingomyelin, and cholesterol (Avanti Polar Lipids) were dissolved in chloroform and mixed in a 1.2∶2.6∶1.0∶2.4∶5.5 ratio (w/w) as reported previously (Chen et al., 2014) and close to the HIV-1 envelope lipid composition (Brügger et al., 2006). After solvent evaporation, the lipid film was further dried under vacuum, and resuspended in PBS at a final concentration of 1 mg/ml. The liposome suspension was extruded through a 0.2 µm polycarbonate filter membrane. Fifty µg of 2H10, 50 µg of 2H10-RKRF and 100 µg 10E8 were incubated at room temperature without and with 50 µl of liposomes for 1.5 hr. The total volume of proteoliposome mixtures of 75 µl was then mixed with 150 µl of 70% sucrose generating the bottom fraction of the gradient, which was successively overlaid by 100 µl of 25% sucrose, 10% sucrose, and 5% sucrose PBS solutions. After centrifugation at 40,000 rpm in a Beckman Optima TL Ultracentrifuge using a SW55TI rotor for 3.5 hr at 10°C, four 75 µl fractions and the bottom input fraction (225 µl) were collected, separated by SDS-PAGE and protein bands were detected by Instant Blue staining.

Fusion assay

Request a detailed protocol

The peptides used in the fusion inhibition experiments, NEQELLELDKWASLW NWFNITNWLWYIK (N-MPER) and KKK-NWFDITNWLWYIKLFIMIVGGLV-KK (C-MPER), were synthesized in C-terminal carboxamide form by solid-phase methods using Fmoc chemistry, purified by reverse phase HPLC, and characterized by matrix-assisted time-of-flight (MALDI-TOF) mass spectrometry (purity >95%). Peptides were routinely dissolved in dimethylsulfoxide (DMSO, spectroscopy grade) and their concentration determined by the Bicinchoninic Acid microassay (Pierce, Rockford, IL, USA).

Large unilamellar vesicles (LUV) were prepared following the extrusion method of Hope et al., 1985. 1-palmitoyl-2-oleoylphosphatidylcholine (POPC) and cholesterol (Chol) (Avanti Polar Lipids, Birmingham, AL, USA) were mixed in chloroform at a 2:1 mol:mol ratio and dried under a N2 stream. Traces of organic solvent were removed by vacuum pumping. Subsequently, the dried lipid films were dispersed in 5 mM Hepes and 100 mM NaCl (pH 7.4) buffer, and subjected to 10 freeze-thaw cycles prior to extrusion 10 times through two stacked polycarbonate membranes (Nuclepore, Inc, Pleasanton, CA, USA). Lipid mixing with fusion-committed vesicles was monitored based on the resonance energy transfer assay described by Struck et al., 1981, with the modifications introduced by Apellániz et al., 2014b. The assay is based on the dilution of co-mixed N-(7-nitro-benz-2-oxa-1,3-diazol-4-yl)phosphatidylethanolamine (N-NBD-PE) and N-(lissamine Rhodamine B sulfonyl) phosphatidylethanolamine (N-Rh-PE) (Molecular Probes, Eugene, OR, USA), whereby dilution due to membrane mixing results in increased N-NBD-PE fluorescence. Vesicles containing 0.6 mol % of each probe (target vesicles) were added at 1:9 ratio to unlabeled vesicles (MPER peptide-primed vesicles). The final lipid concentration in the mixture was 100 µM. The increase in NBD emission upon mixing of target-labeled and primed-unlabeled lipid bilayers was monitored at 530 nm with the excitation wavelength set at 465 nm. A cutoff filter at 515 nm was used between the sample and the emission monochromator to avoid scattering interferences. The fluorescence scale was calibrated such that the zero level corresponded to the initial residual fluorescence of the labeled vesicles and the 100% value to complete mixing of all the lipids in the system (i.e. the fluorescence intensity of vesicles containing 0.06 mol % of each probe). Fusion inhibition was performed with bi-2H10, 2H10-RKRF and 2F5 Fabs at concentrations of 10 µg/ml and 20 µg/ml as indicated. Fabs 2F5 and 10E8 were used as positive controls in experiments of Ab-induced lipid-mixing inhibition. Potencies of bi-2H10 and 2H10-RKRF Abs were compared in dose-response assays performed at concentrations in the 0–20 µg/ml range.

Crystallization, data collection, and structure determination

Request a detailed protocol

For crystallization, 1 mg of gp41FP-TM was mixed with 1.5 mg of 2H10. The complex was purified by SEC on a Superdex 200 column in a buffer containing 100 mM NaCl, 20 mM Tris pH 8,0 and 1% n-octyl β-D-glucopyranoside and concentrated to 7–10 mg/ml. Crystal screening was performed at the EMBL High-Throughput Crystallization Laboratory (HTX lab, Grenoble) in 96-well sitting drop vapor diffusion plates (Greiner). Following manual refinement of crystallization conditions, crystals of gp41FP-TM in complex with 2H10 were grown by mixing 1 µl of protein with 1 µl of reservoir buffer containing 0.1 M sodium citrate pH 6.0, 0.2 M ammonium sulfate, 20% polyethylene glycol 2000 and 0.1 M NaCl at 20°C (293 K) in hanging drop vapor diffusion plates. Before data collection, crystals were flash frozen at 100K in reservoir solution supplemented with 1% n-octyl β-D-glucopyranoside and 25% ethylene glycol for cryo-protection.

Data were collected on the ESRF beamline ID30b at a wavelength of 0.9730 Å. Data were processed with the program XDS (Kabsch, 2010). The data from two crystals were merged with Aimless (Evans and Murshudov, 2013). The data set displayed strong anisotropy in its diffraction limits and was submitted to the STARANISO Web server (Tickle et al., 2018). The merged STARANISO protocol produced a best-resolution limit of 3.28 Å and a worst-resolution limit of 4.74 Å at a cutoff level of 1.2 for the local Imean / σ(Imean) (note that STARANISO does not employ ellipsoidal truncations coincident with the crystal axes). The gp41FP-TM-2H10 crystals belong to space group C 2 2 21 and the structure was solved by molecular replacement using the program Phaser (McCoy et al., 2007) and pdb entries 1env and 4b50. The model was rebuilt using COOT (Emsley et al., 2010) and refined using Phenix (Adams et al., 2010). Data up to 3.28 Å were initially used for model building but were finally truncated to 3.8 Å. Statistics for data reduction and structure refinement are presented in Table 1.

One copy of gp41FP-TM in complex with 2H10 are present in the asymmetric unit. Numbering of the nanobody 2H10 was performed according to Kabat. The gp41FP-TM-2H10 complex was refined to 3.8 Å data with an R/Rfree of 26.6/30.8%. 99.6% of the residues are within the most favored and allowed regions of a Ramachandran plot (Evans and Murshudov, 2013). Some of the crystallographic software used were compiled by SBGrid (Morin et al., 2013). Atomic coordinates and structure factors of the reported crystal structures have been deposited in the Protein Data Bank (https://www.rcsb.org; PDB: 7AEJ).

Figure generation

Request a detailed protocol

Molecular graphics figures were generated with PyMOL (W. Delano; The PyMOL Molecular Graphics System, Version 1.8 Schrödinger, LLC, http://www.pymol.org). The symmetry axis shown in Figure 1C was calculated with the program Galaxy/SymD using the core of gp41 HR1 residues Ala541 to Leu587 (Tai et al., 2014).

Molecular dynamics (MD) simulation

Molecular assays

Request a detailed protocol

Starting from the crystal structure determined herein, we built two molecular assays. The first is based on the crystal structure of gp41FP-TMin complex with 2H10. The disordered FP and TM residues were modeled as helices (chain A FP 512–526 and TM 694–707, chain B FP 512–517 and TM 701–707, chain C FP 512–517). The second assay is based on a symmetric gp41 model which was generated from protomer B by applying a three-fold symmetry operation. Missing FP residues 512–517 and TM residues 701–709 were modeled as helices (Figure 4—figure supplement 1B), and TM residues were modeled based on Env gp41 TM structures (pdb entry 6SNE and 6B3U). All residues were taken in their standard protonation state.

The first assay included a fully hydrated membrane composed of 190 cholesterol, 40 1-palmitoyl-2-oleoyl-glycero-3-phosphocholine (POPC), 88 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-ethanolamine (POPE), 36 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho-L-serine (POPS) and 56 N-stearoyl sphingomyelin, present in the HIV-1 lipid envelope (Brügger et al., 2006), using the CHARMM-GUI interface (Jo et al., 2008; Wu et al., 2014). The resulting molecular assembly consisted of about 178,000 atoms in a rhomboidal cell of 106 × 106×169 ų. The second computational assay featured a water bath of 91 × 91×114 ų, representing a total of 95,700 atoms. Both assays were electrically neutral, with a NaCl concentration set to 150 mM.

Molecular dynamics

Request a detailed protocol

All simulations were performed using the NAMD 2.14 program (Phillips et al., 2005). Proteins, cholesterol, lipids and ions were described using the CHARMM force field (Best et al., 2012; Klauda et al., 2010; MacKerell et al., 1998) and the TIP3P model (Jorgensen et al., 1983) was used for water. MD trajectories were generated in the isobaric-isothermal ensemble at a temperature of 300 K and a pressure of 1 atm. Pressure and temperature were kept constant using the Langevin thermostat and the Langevin piston method (Feller et al., 1995), respectively. Long-range electrostatic interactions were evaluated by the particle-mesh Ewald (PME) algorithm (Darden et al., 1993). Hydrogen mass repartitioning (Hopkins et al., 2015) was employed for all simulations, allowing for using a time step of 4 fs. Integration was performed with a time step of 8 and 4 fs for long- and short-range interactions, respectively, employing the r-RESPA multiple time-stepping algorithm (Tuckerman et al., 1992). The SHAKE/RATTLE (Andersen, 1983; Ryckaert et al., 1977) was used to constrain covalent bonds involving hydrogen atoms to their experimental lengths, and the SETTLE algorithm (Miyamoto and Kollman, 1992) was utilized for water.

The computational assay formed by gp41 in an aqueous environment was simulated for a period of 1 µs, following a thermalization of 40 ns. For the gp41FP-TM/2H10 complex, the lipid bilayer was first thermalized during 200 ns using soft harmonic restraints on every dihedral angle of the protein backbones, allowing the complex to align optimally with its membrane environment. Following the equilibration step, a production run of 1 µs was performed.

The final structural model of hydrated gp41 was also embedded in the HIV-1-like envelope membrane employed for the gp41FP-TM/2H10 crystal structure complex. The same 200 ns equilibration protocol was applied followed by a production run of 1 µs.

Data availability

Diffraction data have been deposited in PDB under the accession code 7AEJ. All data generated or analysed during this study are included in the manuscript and supporting files. Source data files have been provided for Table 2.

The following data sets were generated
    1. Caillat C
    2. Guilligay D
    3. Weissenhorn W
    (2020) RCSB Protein Data Bank
    ID 7AEJ. Crystal structure of asymmetric HIV-1 gp41 containing all membrane anchors.
    https://www.rcsb.org/structure/7AEJ

References

    1. Emsley P
    2. Lohkamp B
    3. Scott WG
    4. Cowtan K
    (2010) Features and development of coot
    Acta Crystallographica. Section D, Biological Crystallography 66:486–501.
    https://doi.org/10.1107/S0907444910007493
    1. Evans PR
    2. Murshudov GN
    (2013) How good are my data and what is the resolution?
    Acta Crystallographica Section D Biological Crystallography 69:1204–1214.
    https://doi.org/10.1107/S0907444913000061
    1. Harrison SC
    (2008) Viral membrane fusion
    Nature Structural & Molecular Biology 15:690–698.
    https://doi.org/10.1038/nsmb.1456
  90. Book
    1. Tickle IJ
    2. Flensburg C
    3. Keller P
    4. Paciorek W
    5. Sharff A
    6. Vonrhein C
    7. Bricogne G
    (2018)
    STARANISO
    Cambridge: Global Phasing Ltd.

Decision letter

  1. Axel T Brunger
    Reviewing Editor; Stanford University, United States
  2. Volker Dötsch
    Senior Editor; Goethe University, Germany

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

This manuscript describes an X-ray structure at 3.8 A resolution of the gp41 subunit of the HIV envelope protein complex (Env) using a construct that spans the membrane-interacting regions, the fusion peptide (FP) and the trans-membrane (TM) region, respectively located at its N- and C-termini and for which no structural information was available. The structure reveals an asymmetric trimer with the core portion folded as in previous structures, but with the N- and C-terminal ends diverging from 3-fold symmetry. The regions visible in the present structure are important functional elements, and in this context the paper provides new insights into the function of gp41.

Decision letter after peer review:

Thank you for submitting your article "Structure of HIV-1 gp41 with its membrane anchors targeted by neutralizing antibodies" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by Axel Brunger as Reviewing Editor and John Kuriyan as the Senior Editor. The reviewers have opted to remain anonymous.

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

We would like to draw your attention to changes in our policy on revisions we have made in response to COVID-19 (https://elifesciences.org/articles/57162).

Summary:

This manuscript describes an X-ray structure of the gp41 subunit of the HIV envelope protein complex (Env) using a construct that spans the membrane-interacting regions, the fusion peptide (FP) and the trans-membrane (TM) region, respectively located at its N- and C-termini and for which no structural information was available. The post-fusion form of gp41 is a trimer of hairpins with a 6-helix bundle core constituted by a central parallel α-helical coiled coil made by an N-terminal α helix, spanning a region termed "heptad repeat 1" (HR1), and an antiparallel C-terminal helix running along the grooves of the coiled coil, spanning a region termed HR2. In this work, to obtain crystals the authors produced two fragments of gp41 that include the HR1 and HR2 regions respectively, an N-terminal fragment with roughly the N-terminal half (70 residues), and a C-terminal fragment of ∼90 residues. The cytosolic tail of gp41, spanning the very C-terminal 150 residues, is absent in the construct. To obtain crystals, they produced the two fragments separately and mixed them to obtain the trimeric complex, termed gp41FP-TM, which crystallized in the presence of a single-chain camelid antibody, 2H10, which binds at the MPPR to make a complex at a stoichiometry 1:3 (i.e, one 2H10 per gp41FP-TM trimer). The resulting structure at 3.8 Å resolution shows an asymmetric trimer with the core portion folded as in previous structures, but with the N- and C-terminal ends diverging from 3-fold symmetry. The regions visible in the present structure are important functional elements, and in this context the paper provides new insights into the function of gp41. Also, the model of the potential interactions of the resulting complex with membranes is validated by mutagenesis of the 2H10 nanobody, which increases in functionality when modifying residues predicted to be directly in contact with the membrane (either by making them more hydrophobic, or positively charged in order to interact ionically with the phosphate head groups of the lipids). The paper also reports a fusion assay involving pre-activated liposomes with peptides spanning the MPER, to show that the antibody blocks fusion, and the antibody mutants with higher affinity (including a bivalent version of 2H10), do so more efficiently. However, the molecular dynamics simulations and models are highly speculative.

Essential revisions:

The overall resolution of the crystal structure is 3.8 Å with a data completeness of 78.01%, but the highest resolution shell has a completeness of 54.69% and a Rmerge of 150.8% (after STARANISO truncation, Rmerge in the resolution shell 3.97 Å – 3.85 Å is 78.7%). It is hard to tell the quality of the final map used for model building, in particular, for the fusion peptides and transmembrane segments. How much do the noisy high resolution reflections contribute to the refinement and map calculation? In addition, at this resolution, is there enough side chain density for one to be confident about the sequence register? In any case, it might be useful to show some representative density figures and B factors of the final model in the supplemental material to give reader a better sense how reliable the various regions of the model really are.

In Figure 1—figure supplement 1 and Figure 1—figure supplement 2, it is a bit odd that the larger chain C peptide looks smeary, but the smaller chain N is much sharper (sometimes looks like more than one band). Have these peptides been checked by mass spec? Since it is difficult to read the protein sequence from a 3.8 Å map, it would be important to confirm that the heterogeneity of the ends in density was not coming for the samples.

It is unclear whether the asymmetrical structure of gp41 could accommodate the bivalent 2H10 (bi-2H10) or not. If not, the authors might want to speculate why bi-2H10 is much more potent than the 2H10 nanobody.

p.5, Ln 58 – "packing effects… can be excluded'. While crystal contacts may not readily explain the asymmetric conformation of the FP and TM domains, it is possible that the observed conformation may have been stabilized by (i) the BOG detergent. Also would the symmetric 6HB conformation clash with the crystal lattice, and if so, the crystal packing does rule out a symmetric conformation? Please comment.

L132. Related to the previous point, it is not obvious that it is the antibody that induces the bent conformation. It is more likely that the detergent-solubilized protein maintains an equilibrium between multiple conformations, and that the nanobody selects one of them. Such a sampling of conformations is in line with gp41's function, as it needs to sample asymmetric conformations, in particular of this membrane interacting regions, to drive membrane fusion. There is a longstanding controversy in biochemistry between induced fit and conformational selection, but there are ways to distinguish them (see, for instance, https://doi.org/10.1016/j.bpc.2014.03.003). The point is not that the authors explore this suggestion with their construct, but rather that they consider the terminology used, as the end effect is the same. Just saying that the antibody stabilized an asymmetric form of the protein would be enough, without saying that it "induces it".

Lines 171-173. The bent on chain A-C at the epitope appears to be stabilized by 2H10 binding (with the caveat mentioned above about induced fit/conformational selection). But chain A-C has the shortest ordered density for its TM helix in the crystals. And the actual orientation of the TM helices of the two other chains, which are longer, may be stabilized by crystal packing. The authors indeed point to packing interactions made by chain C-B. But the same Figure suggests that chain C-C, which has the longest ordered TM helix, appears to be stabilized in the observed conformation by packing against the TM helix of chain C-B. This raises the question of the biological significance of the actual docking that the authors make on membranes, which they say is validated by MD.

p.6 In the lipid mixing experiments, bi-2H10 shows more potent inhibition than 2F5 but 2F5 is more potent than bi-2H10 in Table 1. This suggests that the lipid mixing assay may not represent the HIV gp41 mediated fusion reaction. Please clarify.

Do the mutations in 2H10-F and 2H10-RKRF that enhance membrane interactions and neutralization also increase autoreactivity or polyreactivity (toward Hep2 cells, lipids, etc.)?

Lines 112 and around: the N and C chains were purified separately. Do they behave as monodisperse monomers? Or does the N chain make trimers and the C chain behaves as monomer? Was SEC done previous to mixing the two to assemble the mono-disperse complex displayed in Figure 1, supplement 1B?

Lines 114-115: please compare the thermostability of both constructs in parallel in the same experiment, to make this statement stronger.

L131. It is not clear that the TM region and FP in the three chains in the trimer point in opposite directions, as stated. Figure 1A appears to indicate that they all point toward the right of the Figure.

Lines 131-140: It is not mentioned in this description what surface area is buried by the nanobody in the complex, and whether or not all three CDRs are involved in the interaction. These data should be provided, even if at 3.8 Å resolution the atomic model may not be very accurate. Does it contact only chain C-A, or are there contacts with the other chains stabilizing the interaction? The authors should be more explicit here.

Lines 161-168. The relevance of the molecular dynamics is questionable since there are many assumptions being made. The MD simulations of the protein in a lipid bilayer mimicking the composition of the HIV envelope are described twice in the manuscript (here and then from lines 233 on). The overall flow seems a little strange, although the present paragraph is to anticipate that the CDR3 of 2H10 is predicted to also interact with the membrane. In the Methods section, it is explained that to run the MD calculations the regions that were not visible were completed as α-helical extensions. But apparently only for FP residues 512 to 518 and TM from residues 700 to 709, as stated in line 532 in the Methods section. But according to Figure 1C, chain A ends at residue 693, so this means that the segment 694 to 700 was also built for this chain? Just as an α helix extending from 693? And what about the FP for chain A, which according to Figure 1C only starts at position 527? Was the FP not used for this segment? It is stated in lines 155-158 that the FP may not be helical upstream of residue 518. At any rate, as modeled in Figure 1, Figure supplement 4A, the two FP helices modeled end at the middle of the membrane, in the aliphatic moiety. The charged N-termini for the two chains modeled will be very unstable there, so there is no wonder that the MD moves them to reach the phosphate headgroups of the lipids across the membrane. This model goes against the current paradigm that the fusion peptides only interact with the outer leaflet of the target membrane, and does not cross it. It is questionable how the model has been docked into the membrane. Figure 1 Figure supplement 4B indeed suggests that the two FP helices traverse the membrane to interact with the lipid headgroups of the inner leaflet. Please soften the statements regarding the molecular dynamics simulations and model and emphasize that it they are highly speculative.

Lines 233-266. Related to the previous point, the authors should compare by also docking the FP on the membrane not as in the symmetrized X-ray structure, which is derived from the protein having this flexible non-polar region within a detergent micelle, but with the FP anchored only on the outer leaflet, as it would initially insert into target membranes in an extended intermediate conformation. In this respect, the illustration of Figure 5D is highly mis-leading, as the peptides are not expected to insert into the target membrane in the proposed way, with the charged N-termini within the aliphatic moiety of the membrane. Figure 5 indeed shows that lipid head groups move to the center of the membrane in order to compensate for the charge at the N-termini, which is highly unlikely. This Figure should be adapted to reflect this.

Lines 220-222. Did the authors check that 2F5 does not bind the gp41FP-TM construct? It could select for a different conformation than does 2H10 and still bind. Or does it have to bind its epitope exclusively when in the context of pre-fusion gp160? Because this antibody was used in the fusion experiments, it should be possible to test its binding properties. And the reverse question: Do 10E8 or LN01 inhibit fusion in the system described in the previous paragraph? Since the peptide used should include their epitopes, they could also have a fusion blocking effect.

https://doi.org/10.7554/eLife.65005.sa1

Author response

Essential revisions:

The overall resolution of the crystal structure is 3.8 Å with a data completeness of 78.01%, but the highest resolution shell has a completeness of 54.69% and a Rmerge of 150.8% (after STARANISO truncation, Rmerge in the resolution shell 3.97 Å – 3.85 Å is 78.7%). It is hard to tell the quality of the final map used for model building, in particular, for the fusion peptides and transmembrane segments. How much do the noisy high resolution reflections contribute to the refinement and map calculation? In addition, at this resolution, is there enough side chain density for one to be confident about the sequence register? In any case, it might be useful to show some representative density figures and B factors of the final model in the supplemental material to give reader a better sense how reliable the various regions of the model really are.

Molecular replacement using the coordinates of the gp41 core (1env) and of 2H10 (4b50) resulted in only one solution placing the gp41 core unambiguously in the electron density. Datasets of several crystals have been analysed and the determination of the space/point group by XDS, pointless (and other software) always gave as solution the point group C 2 2 2. Space groups of lower symmetry have also been tested. Zanuda was used for refinement based checking of the space group (P1, P 1 21 1, C 1 2 1 and C 2 2 21). Changing the screw axis in C 2 2 21 was also tested. However, the only space group that gave a clear solution by molecular replacement was C 2 2 21 with a TFZ score of 39.9.

We added 2Fo-Fc composite omit maps of two close-up regions, the HR1-HR2 core and the kinked MPER conformation of chain C to show the sequence register in Figure 1—figure supplement 3. Notably, the same kinked MPER conformation was reported in the original 10E8-MPER peptide structure (4G6F). The transmembrane domains and FP regions were subsequently built in the 3.8 Å resolution map. Although, the map provides much less precision in the position of the atoms, we are confident in the chain traces. For example, chain B forms a continuous helix of HR2-MPER-TM and its structure is in agreement with NMR and crystal structures of MPER-TM (6SNE and 6B3U). Furthermore, chain B is in the same conformation as the protomers present in gp41 structures including MPER (3k9a) and FPPR-MPER (2x7r) as shown in Figure 1—figure supplement 4.

In Figure 1—figure supplement 1 and Figure 1—figure supplement 2, it is a bit odd that the larger chain C peptide looks smeary, but the smaller chain N is much sharper (sometimes looks like more than one band). Have these peptides been checked by mass spec? Since it is difficult to read the protein sequence from a 3.8 Å map, it would be important to confirm that the heterogeneity of the ends in density was not coming for the samples.

Chain C looks always smeary on SDS-PAGE. We have analyzed the complex by mass spectrometry to confirm the integrity of the chains, Author response image 1. The calculated MW of chain N is 10134.75 Da and of chain C is 11688.32 Da. The mass detected by MALDI-TOF for chain N is 10132 Da, and for chain C 11686 Da, in agreement with the calculated MW. We detect only 2/3 minor peaks, which might be degradation/contamination products and do not explain the abnormal SDS-PAGE migration of chain C.

Author response image 1
Download asset Open asset
MALDI-TOF analysis of gp41FP-TM.

The raw spectrum shows the single and doubly charged ion of chains N and C.

It is unclear whether the asymmetrical structure of gp41 could accommodate the bivalent 2H10 (bi-2H10) or not. If not, the authors might want to speculate why bi-2H10 is much more potent than the 2H10 nanobody.

The extended 2H10 epitope is only present in one chain, indicating that a bi-head is unlikely to interact with the two chains within the same trimer in the conformation of the crystal structure. We have previously suggested that bi-2H10 is more potent in the neutralization assay (Lutje-Hulsik et al. 2013, PLoS Pathogens; doi: 10.1371/journal.ppat.1003202) because of the avidity effect by binding two epitopes at the same time. SPR measurement presented in Lutje-Hulsik et al. revealed binding to a fusion intermediate conformation of gp41 (gp41int) with a KD of 29 nM for wt 2H10 and 2.5 nM for bi-2H10. The optimal length of the linker of bi-2H10 is 15 amino acids, which would also allow binding to different trimers at the fusion site.

We have added the following sentence in the discussion:

“This, thus, confirmed 2H10 as a modest anti-MPER neutralizing antibody, whose neutralization potency is not only enhanced by using bi-2H10 with increased avidity (Lutje Hulsik et al., 2013), but also by increasing potential membrane interaction of the 2H10 monohead.”

p.5, Ln 58 – "packing effects… can be excluded'. While crystal contacts may not readily explain the asymmetric conformation of the FP and TM domains, it is possible that the observed conformation may have been stabilized by (i) the BOG detergent.

The asymmetric conformation of the membrane anchors are not explained by crystal packing (see our response below). We also think that its stabilization by β-OG is unlikely, since the TM is slightly increased in the gp41FP-TM construct (in β-OG buffer) in comparison to gp41 without FP and TM (Buzon et al. PLoS Pathogen). This increase of the TM is in agreement with the MD model of the post-fusion conformation (without 2H10), which reveals coiled-coil interactions within the trimeric FP, and lateral interactions between the FP and the TM domains.

Also would the symmetric 6HB conformation clash with the crystal lattice, and if so, the crystal packing does rule out a symmetric conformation? Please comment.

The electron density is in agreement with the asymmetric model presented here and superimposing the 6HB “core structure” would not generate clashes. However, a model of the 6HB core including MPER-TM and FP, such as the post-fusion model presented in Figure 4 would produce clashes between the C-terminal residues of the TM domains of two symmetry-related molecules.

L132. Related to the previous point, it is not obvious that it is the antibody that induces the bent conformation. It is more likely that the detergent-solubilized protein maintains an equilibrium between multiple conformations, and that the nanobody selects one of them. Such a sampling of conformations is in line with gp41's function, as it needs to sample asymmetric conformations, in particular of this membrane interacting regions, to drive membrane fusion.

We agree with the reviewers that 2H10 might sample one conformation out of an equilibrium between multiple conformations, although a thermostability of ~ 93°C is indicative of some limitations in the conformational flexibility.

There is a longstanding controversy in biochemistry between induced fit and conformational selection, but there are ways to distinguish them (see, for instance, https://doi.org/10.1016/j.bpc.2014.03.003). The point is not that the authors explore this suggestion with their construct, but rather that they consider the terminology used, as the end effect is the same. Just saying that the antibody stabilized an asymmetric form of the protein would be enough, without saying that it "induces it".

We agree with the reviewers, and we have changed this accordingly throughout the text.

Lines 171-173. The bent on chain A-C at the epitope appears to be stabilized by 2H10 binding (with the caveat mentioned above about induced fit/conformational selection). But chain A-C has the shortest ordered density for its TM helix in the crystals. And the actual orientation of the TM helices of the two other chains, which are longer, may be stabilized by crystal packing. The authors indeed point to packing interactions made by chain C-B. But the same Figure suggests that chain C-C, which has the longest ordered TM helix, appears to be stabilized in the observed conformation by packing against the TM helix of chain C-B. This raises the question of the biological significance of the actual docking that the authors make on membranes, which they say is validated by MD.

There are no defined crystal contacts stabilizing the TM orientations (Figure 1—figure supplement 6). We apologize for the orientation shown in the previous Figure supplement 5, suggesting a defined crystal contact between C-terminal chains of the B protomer.

Crystal packing is, thus, unusual and we suggest that the disordered regions of the C-termini, such as chain B residues 701 to 716 might stabilize interlayer contacts, which are less regular, and hence, not defined by electron density at this resolution. We acknowledge that this is quite unusual for crystal packing.

An error in space group setting can be excluded as out of all space group settings tested as stated above, only C2221 provided a clear solution.

p.6 In the lipid mixing experiments, bi-2H10 shows more potent inhibition than 2F5 but 2F5 is more potent than bi-2H10 in Table 1. This suggests that the lipid mixing assay may not represent the HIV gp41 mediated fusion reaction. Please clarify.

Antibody 2F5 was included as a positive control in virus neutralization and fusion blocking assays because the linear epitope recognized by 2H10 overlapped with that recognized by 2F5. However, we note that, whereas the IgG was used in the neutralization assays (Table 2), the Fab version of 2F5 was used in the fusion inhibition assay (Figure 2). Thus, we infer that avidity effects may account for the slightly better performance of the 2F5 IgG in comparison with bi-2H10 in neutralization, an effect not observed in fusion inhibition. We now specify more clearly the different positive controls used in both assays (see Materials and methods).

Regarding the relevance of the lipid-mixing assay for the gp41-induced fusion, we agree with the reviewers that peptide-induced fusion in vitro can only recapitulate certain aspects of the process. However, recognition of the MPER peptides inserted in the membrane and blocking of their restructuring activity is relevant for the mechanism of molecular recognition of MPER Abs for the following reasons:

First, the capacity for blocking the lipid-mixing process correlates with the neutralization potency of the MPER Abs (ref: Apellaniz et al., 2014b). To illustrate this point, please see also Author response image 2, which compares 10E8 Fab variants with regard to neutralization and lipid-mixing inhibition. Since both activities of 10E8 and its variants correlate, we believe that these observations warrant the use of the assay to compare different 2H10 variants.

Author response image 2
Download asset Open asset
Neutralization potency of 10E8 Fab variants (cell entry inhibition measured against HIV-1 virions pseudotyped with JR-CSF Env) and their ability to block C-MPER-induced lipid-mixing.

The top diagram (Figure 2—figure supplement 1B) shows the mutated residues and their positions with respect to the putative membrane binding.

Second, MPER antibodies interact directly with membrane-bound peptide epitopes, change their insertion state (depth level and oligomerization state) and inhibit membrane restructuring (permeabilization and fusion).

Thus, to address this question, we extended the Results section accordingly – line 196:

“In the experimental setting (Apellaniz et al., 2014b) a vesicle population is primed for fusion by addition of the N-MPER peptide containing the 2H10 epitope, which produces a fluorescence intensity spark at time 20 s (Figure 2A). […] MPER antibodies have been shown to interact directly with membrane-bound peptide epitopes, thereby changing their insertion state (depth and oligomerization levels) and inhibiting their capacity to induce fusion…”.

Line 216:

“The use of the 10E8 Fab as an additional control further supports the blocking effect mediated by epitope recognition. […] Following this trend, fusion inhibition levels estimated as a function of the antibody concentration confirmed the apparent higher potency exhibited by the bi-2H10 (Figure 2D).”

Do the mutations in 2H10-F and 2H10-RKRF that enhance membrane interactions and neutralization also increase autoreactivity or polyreactivity (toward Hep2 cells, lipids, etc.)?

We have added new data presented in Figure 1—figure supplement 7. Using a liposome flotation assay, we show that 2H10 and 2H10-RKFR do not reveal detectable membrane interaction in vitro whereas bnAb 10E8 reveals some membrane binding in this assay.

Lines 112 and around: the N and C chains were purified separately. Do they behave as monodisperse monomers? Or does the N chain make trimers and the C chain behaves as monomer? Was SEC done previous to mixing the two to assemble the mono-disperse complex displayed in Figure 1, supplement 1B?

We described the assembly and purification protocol in the methods section. Both chains were purified separately. They were not separated by SEC prior to complex formation. We have purified chain-C prior to analyzing it by SEC, which showed a monodisperse peak eluting at 13.5 ml (between the elution peak of 150 kDa and 75 kDa marker proteins) from a superdex 200 SEC column (Figure 1, Lutje-Hulsik et al. PLoS Pathogens 2013). In an unrelated experiment, the chain-N fusion protein was separated by SEC, revealing a monodisperse peak eluting at 12 ml from a superdex 200 SEC column.

Lines 114-115: please compare the thermostability of both constructs in parallel in the same experiment, to make this statement stronger.

We agree with the reviewers that a side-by-side comparison could strengthen the conclusion. Since the original experiments have been performed, our JASCO CD spectrophotometer broke down and has not yet been replaced. Due to restrictions imposed by the ongoing health crisis, we did not have access to a comparable instrument in a near-by facility. We had, however, previously performed side-by side CD experiments with a shorter gp41FP-TM construct in a buffer containing DDM, which also showed a slight increase of the Tm compared to gp41 without FP and TM. See Author response image 3.

Author response image 3
Download asset Open asset
Gp41(512-711) used in this experiment, composed of residues 512-581 and 628-711 (including FP-FPPR and MPER-TM), reveals a Tm of 90°C (green curve, buffer DDM) and gp41(528-683) (including FPPR and MPER) has a Tm of 88°C (black curve, no detergent).

Please note that gp41FP-TM used in the current study is composed of residues 512-594 with a 13 aa longer N-terminal coiled coil and a 5 residue C-terminal extension to residue 716. The two curves are shown in comparison to the gp41 core structure composed of residues 541-665 (blue curve, no detergent).

L131. It is not clear that the TM region and FP in the three chains in the trimer point in opposite directions, as stated. Figure 1A appears to indicate that they all point toward the right of the Figure.

We have changed “opposite directions” to “different directions”. We also added Figure 1C, which includes the trimer axis of the symmetric 6-helical bundle core structure, and demonstrates further that FP and TM point in different directions.

Lines 131-140: It is not mentioned in this description what surface area is buried by the nanobody in the complex, and whether or not all three CDRs are involved in the interaction. These data should be provided, even if at 3.8 Å resolution the atomic model may not be very accurate. Does it contact only chain C-A, or are there contacts with the other chains stabilizing the interaction? The authors should be more explicit here.

We have included the buried surface area and added on line 137:

“The protomer A chain C 2H10 epitope spans from residues Q658 to N671, which is involved in a series of polar contacts with 2H10 covering a total buried surface of 712.2 Å2.”

Lines 161-168. The relevance of the molecular dynamics is questionable since there are many assumptions being made. The MD simulations of the protein in a lipid bilayer mimicking the composition of the HIV envelope are described twice in the manuscript (here and then from lines 233 on). The overall flow seems a little strange, although the present paragraph is to anticipate that the CDR3 of 2H10 is predicted to also interact with the membrane. In the Methods section, it is explained that to run the MD calculations the regions that were not visible were completed as α-helical extensions. But apparently only for FP residues 512 to 518 and TM from residues 700 to 709, as stated in line 532 in the Methods section. But according to Figure 1C, chain A ends at residue 693, so this means that the segment 694 to 700 was also built for this chain? Just as an α helix extending from 693? And what about the FP for chain A, which according to Figure 1C only starts at position 527? Was the FP not used for this segment?

We used MD simulations for two different computational experiments. First, we used MD to analyze the crystal structure of the complex placed in a lipid environment with the aim to determine whether the membrane anchors, which are surrounded by detergent in the crystal structure are stable in a lipid environment. This is shown in Figure 1—figure supplement 5. The computational experiment was performed with the gp41FP-TM-2H10 crystal structure model, which has the missing FP and TM residues modeled as helices in all 3 protomers.

We clarified this in the Methods section line 594:

“Starting from the crystal structure determined herein, we built two molecular assays. The first is based on the crystal structure of gp41FP-TM in complex with 2H10. […] All residues were taken in their standard protonation state.”

It is stated in lines 155-158 that the FP may not be helical upstream of residue 518.

This is correct. The visible N-termini of N-terminal chains C and B start at residue 518. The remaining six residues are disordered in the crystal structure. In the MD simulations of gp41FP-TM-2H10 in a lipid bilayer (Figure 1—figure supplement 5), and in the MD simulation of the post-fusion conformation model in the lipid bilayer (Figure 4), most modeled FP residues (512 to 517) stay helical in the membrane environment. In the MD post-fusion model, only residue A512 frays in one protomer, and residues A512 and V513 unfold in the other two protomers. In the gp41FP-TM-2H10 simulation in the bilayer, only residue A512 unfolds in all three protomers, suggesting that the N-terminal region of FP (residues 513/514 to 517) may, indeed, also adopt a helical conformation within a lipid bilayer.

At any rate, as modeled in Figure 1.

Figure 1 shows the model of the crystal structure; it only features those residues that are accounted for by electron density.

Figure supplement 4A, the two FP helices modeled end at the middle of the membrane, in the aliphatic moiety. The charged N-termini for the two chains modeled will be very unstable there, so there is no wonder that the MD moves them to reach the phosphate headgroups of the lipids across the membrane.

The structure shown in this figure supplement (now Figure 1—figure supplement 5) is based on the crystal structure. Only missing FP and TM residues have been modeled. We added the figure to emphasize that the crystal conformation is, indeed, stable within a lipid bilayer during MD simulations, which revealed only small movements of TM and FP, including occasional lipid head groups moving up to contact form noncovalent interactions with the N-terminus of FP. We reason that local thinning of the membrane could be important for the fusion process, as reported previously for SNARE catalyzed fusion (Smirnova et al. PNAS, 116, 2019). Nevertheless, we have toned down our statement on FP orientation in the manuscript.

This model goes against the current paradigm that the fusion peptides only interact with the outer leaflet of the target membrane, and does not cross it.

The current models of FP-lipid bilayer interaction are entirely based on isolated FP peptide structures and membrane interaction studies with FP peptides. These studies indicate indeed lateral immersion of FP into one lipid bilayer, which may well constitute a first encounter of FP with the lipid environment. However, the crystal structure shows that FP adopts a continuous helix with FPPR and HR1 in protomer B, and forms a helical extension in protomer C. As suggested in the manuscript, protomer B likely constitutes the protomer closest to the final post-fusion conformation, which indicates that FP crosses one leaflet of the bilayer within the complete gp41 structure.

We toned down the statement, and included lateral FP interaction in the model shown in Figure 5.

It is questionable how the model has been docked into the membrane. Figure 1 Figure supplement 4B indeed suggests that the two FP helices traverse the membrane to interact with the lipid headgroups of the inner leaflet. Please soften the statements regarding the molecular dynamics simulations and model and emphasize that it they are highly speculative.

The structure was docked into a bilayer by using TM residues K683 and R707 defining the membrane boundaries.

Lines 233-266. Related to the previous point, the authors should compare by also docking the FP on the membrane not as in the symmetrized X-ray structure, which is derived from the protein having this flexible non-polar region within a detergent micelle, but with the FP anchored only on the outer leaflet, as it would initially insert into target membranes in an extended intermediate conformation. In this respect, the illustration of Figure 5D is highly mis-leading, as the peptides are not expected to insert into the target membrane in the proposed way, with the charged N-termini within the aliphatic moiety of the membrane.

We would like to point out that Figure 5D is a model and no complete structure of this proposed intermediate conformation of gp41 is available. It has been modeled based on a recent intermediate conformation of influenza HA (Benton et al., Nature 583, 2020), which also suggests a flexible linkage of the central coiled coil to FP. We have added the model proposed by the reviewers that suggests that FP first interacts laterally with the outer leaflet of the host cell membrane. FP may then only interact with each other in a later conformation such as the post-fusion conformation MD model presented here. Notably, the conformation of a triple stranded FP coiled coil is robust within a lipid bilayer upon MD simulation, thereby corroborating that FP can span one leaflet of the bilayer.

Figure 5 indeed shows that lipid head groups move to the center of the membrane in order to compensate for the charge at the N-termini, which is highly unlikely. This Figure should be adapted to reflect this.

The models of Figures 5E and F correspond to the crystal structure (with helical extensions of FP and TM) and Figure 5G shows the MD post fusion model. We have added the following sentence to the description on page 9, line 303:

“Notably, in this model based on the crystal structure, FP spans only one leaflet of the lipid bilayer facilitating local lipid head group interactions with the N-terminus of FP.”

Lines 220-222. Did the authors check that 2F5 does not bind the gp41FP-TM construct? It could select for a different conformation than does 2H10 and still bind.

We did not test 2F5 binding to gp41FP-TM. We agree with the reviewers that it could select a different extended conformation than 2H10.

Or does it have to bind its epitope exclusively when in the context of pre-fusion gp160?

The initial target of MPER bnAbs is still a matter of debate. Only structural information on 10E8 interaction with an “open” Env conformation is available. Other studies indicate that MPER bnAbs interact with an extended fusion-intermediate conformation that bridges viral and host cell membranes (Frey et al. Proc Natl Acad Sci USA 105, 2008). It was indeed surprising to detect high-affinity interactions with the gp41 conformation presented here.

Because this antibody was used in the fusion experiments, it should be possible to test its binding properties. And the reverse question: Do 10E8 or LN01 inhibit fusion in the system described in the previous paragraph? Since the peptide used should include their epitopes, they could also have a fusion blocking effect.

We have included the 10E8 Fab in the fusion assay as described above, showing its inhibitory effect.

https://doi.org/10.7554/eLife.65005.sa2

Article and author information

Author details

  1. Christophe Caillat

    Univ. Grenoble Alpes, CEA, CNRS, Institut de Biologie Structurale (IBS), Grenoble, France
    Contribution
    Conceptualization, Formal analysis, Investigation, Visualization, Methodology, Writing - original draft
    Contributed equally with
    Delphine Guilligay
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2959-1871

  2. Delphine Guilligay

    Univ. Grenoble Alpes, CEA, CNRS, Institut de Biologie Structurale (IBS), Grenoble, France
    Contribution
    Investigation, Visualization, Methodology, Writing - original draft
    Contributed equally with
    Christophe Caillat
    Competing interests
    No competing interests declared

  3. Johana Torralba

    Instituto Biofisika (CSIC, UPV/EHU) and Department of Biochemistry and Molecular Biology, University of the Basque Country (UPV/EHU), Bilbao, Spain
    Contribution
    Investigation, Visualization
    Competing interests
    No competing interests declared

  4. Nikolas Friedrich

    Institute of Medical Virology, University of Zurich, Zurich, Switzerland
    Contribution
    Investigation, Visualization, Methodology, Writing - original draft
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0694-657X

  5. Jose L Nieva

    Instituto Biofisika (CSIC, UPV/EHU) and Department of Biochemistry and Molecular Biology, University of the Basque Country (UPV/EHU), Bilbao, Spain
    Contribution
    Conceptualization, Supervision, Funding acquisition, Methodology, Writing - original draft
    Competing interests
    No competing interests declared

  6. Alexandra Trkola

    Institute of Medical Virology, University of Zurich, Zurich, Switzerland
    Contribution
    Conceptualization, Supervision, Funding acquisition
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1013-876X

  7. Christophe J Chipot

    1. Laboratoire de Physique et Chimie Théoriques (LPCT), University of Lorraine, Vandoeuvre-lès-Nancy, France
    2. Laboratoire International Associé, CNRS and University of Illinois at Urbana-Champaign, Vandoeuvre-lès-Nancy, France
    3. Department of Physics, University of Illinois at Urbana-Champaign, Urbana, United States
    Contribution
    Conceptualization, Formal analysis, Investigation, Visualization, Methodology, Writing - original draft
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9122-1698

  8. François L Dehez

    1. Laboratoire de Physique et Chimie Théoriques (LPCT), University of Lorraine, Vandoeuvre-lès-Nancy, France
    2. Laboratoire International Associé, CNRS and University of Illinois at Urbana-Champaign, Vandoeuvre-lès-Nancy, France
    Contribution
    Conceptualization, Formal analysis, Supervision, Funding acquisition, Investigation, Visualization, Methodology, Writing - original draft
    Competing interests
    No competing interests declared

  9. Winfried Weissenhorn

    Univ. Grenoble Alpes, CEA, CNRS, Institut de Biologie Structurale (IBS), Grenoble, France
    Contribution
    Conceptualization, Formal analysis, Supervision, Funding acquisition, Methodology, Writing - original draft, Project administration, Writing - review and editing
    For correspondence
    winfried.weissenhorn@ibs.fr
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-5532-4959

Funding

H2020 Health (681137)

  • Winfried Weissenhorn

Agence Nationale de la Recherche (ANR-17-EURE-0003)

  • Winfried Weissenhorn

Ministerio de Economía, Industria y Competitividad, Gobierno de España (BIO2015-64421-R)

  • Jose L Nieva

Ministerio de Ciencia y Tecnología (RTI2018-095624-B-C21)

  • Jose L Nieva

French Infrastructure for Integrated Structural Biology (ANR-10-INBS-05–02)

  • Winfried Weissenhorn

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

WW acknowledges support from the Institut Universitaire de France (IUF), from the European Union's Horizon 2020 research and innovation program under grant agreement No. 681137, H2020 EAVI and the platforms of the Grenoble Instruct-ERIC center (IBS and ISBG; UMS 3518 CNRS-CEA-UGA-EMBL) within the Grenoble Partnership for Structural Biology (PSB). Platform access was supported by FRISBI (ANR-10-INBS-05–02) and GRAL, a project of the University Grenoble Alpes graduate school (Ecoles Universitaires de Recherche) CBH-EUR-GS (ANR-17-EURE-0003). IBS is part of the CEA DRF Interdisciplinary Research Institute of Grenoble (IRIG). JLN acknowledges funding from Spanish MCIU (RTI2018-095624-B-C21; MCIU/AEI/FEDER, UE) and Basque Government (IT1196-19). We thank Miriam Hock and Serafima Guseva for previous contributions to the project, Davide Corti for providing bnAb LN01, the ESRF-EMBL Joint Structural Biology Group for access and support at the ESRF beam lines, J Marquez (EMBL) from the HTX crystallization facility and C Mas and J-B Reiser for assistance on ISBG platforms.

Senior Editor

  1. Volker Dötsch, Goethe University, Germany

Reviewing Editor

  1. Axel T Brunger, Stanford University, United States

Publication history

  1. Received: November 18, 2020
  2. Accepted: April 18, 2021
  3. Accepted Manuscript published: April 19, 2021 (version 1)
  4. Version of Record published: April 29, 2021 (version 2)

Copyright

© 2021, Caillat et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 1,746
    Page views
  • 238
    Downloads
  • 9
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Christophe Caillat
  2. Delphine Guilligay
  3. Johana Torralba
  4. Nikolas Friedrich
  5. Jose L Nieva
  6. Alexandra Trkola
  7. Christophe J Chipot
  8. François L Dehez
  9. Winfried Weissenhorn
(2021)
Structure of HIV-1 gp41 with its membrane anchors targeted by neutralizing antibodies
eLife 10:e65005.
https://doi.org/10.7554/eLife.65005

Categories and tags

Research organism

Further reading

    1. Microbiology and Infectious Disease

    Phage tRNAs evade tRNA-targeting host defenses through anticodon loop mutations

    Daan F van den Berg, Baltus A van der Steen ... Stan JJ Brouns
    Short Report

    tRNAs in bacteriophage genomes are widespread across bacterial host genera, but their exact function has remained unclear for more than 50 years. Several hypotheses have been proposed, and the most widely accepted one is codon compensation, which suggests that phages encode tRNAs that supplement codons that are less frequently used by the host. Here, we combine several observations and propose a new hypothesis that phage-encoded tRNAs counteract the tRNA-depleting strategies of the host using enzymes such as VapC, PrrC, Colicin D, and Colicin E5 to defend from viral infection. Based on mutational patterns of anticodon loops of tRNAs encoded by phages, we predict that these tRNAs are insensitive to host tRNAses. For phage-encoded tRNAs targeted in the anticodon itself, we observe that phages typically avoid encoding these tRNAs. Further supporting the hypothesis that phage tRNAs are selected to be insensitive to host anticodon nucleases. Altogether our results support the hypothesis that phage-encoded tRNAs have evolved to be insensitive to host anticodon nucleases.

    1. Immunology and Inflammation
    2. Microbiology and Infectious Disease

    Cellular and molecular dynamics in the lungs of neonatal and juvenile mice in response to E. coli

    Sharon A McGrath-Morrow, Jarrett Venezia ... Alan L Scott
    Research Article

    Bacterial pneumonia in neonates can cause significant morbidity and mortality when compared to other childhood age groups. To understand the immune mechanisms that underlie these age-related differences, we employed a mouse model of E. coli pneumonia to determine the dynamic cellular and molecular differences in immune responsiveness between neonates (PND 3-5) and juveniles (PND 12-18), at 24, 48, and 72 hours. Cytokine gene expression from whole lung extracts was also quantified at these time points, using qRT-PCR. E. coli challenge resulted in rapid and significant increases in neutrophils, monocytes, and γδT cells, along with significant decreases in dendritic cells and alveolar macrophages in the lungs of both neonates and juveniles. E. coli challenged juvenile lung had significant increases in interstitial macrophages and recruited monocytes that were not observed in neonatal lungs. Expression of IFNg-responsive genes was positively correlated with the levels and dynamics of MHCII-expressing innate cells in neonatal and juvenile lungs. Several facets of immune responsiveness in the wild-type neonates were recapitulated in juvenile MHCII-/- juveniles. Employing a pre-clinical model of E. coli pneumonia, we identified significant differences in the early cellular and molecular dynamics in the lungs that likely contribute to the elevated susceptibility of neonates to bacterial pneumonia and could represent targets for intervention to improve respiratory outcomes and survivability of neonates.

    1. Microbiology and Infectious Disease

    The Plasmodium falciparum apicoplast cysteine desulfurase provides sulfur for both iron-sulfur cluster assembly and tRNA modification

    Russell P Swift, Rubayet Elahi ... Sean T Prigge
    Research Article Updated

    Iron-sulfur clusters (FeS) are ancient and ubiquitous protein cofactors that play fundamental roles in many aspects of cell biology. These cofactors cannot be scavenged or trafficked within a cell and thus must be synthesized in any subcellular compartment where they are required. We examined the FeS synthesis proteins found in the relict plastid organelle, called the apicoplast, of the human malaria parasite Plasmodium falciparum. Using a chemical bypass method, we deleted four of the FeS pathway proteins involved in sulfur acquisition and cluster assembly and demonstrated that they are all essential for parasite survival. However, the effect that these deletions had on the apicoplast organelle differed. Deletion of the cysteine desulfurase SufS led to disruption of the apicoplast organelle and loss of the organellar genome, whereas the other deletions did not affect organelle maintenance. Ultimately, we discovered that the requirement of SufS for organelle maintenance is not driven by its role in FeS biosynthesis, but rather, by its function in generating sulfur for use by MnmA, a tRNA modifying enzyme that we localized to the apicoplast. Complementation of MnmA and SufS activity with a bacterial MnmA and its cognate cysteine desulfurase strongly suggests that the parasite SufS provides sulfur for both FeS biosynthesis and tRNA modification in the apicoplast. The dual role of parasite SufS is likely to be found in other plastid-containing organisms and highlights the central role of this enzyme in plastid biology.