key: cord-0334981-jphbh5a9
authors: Prasad, Vidya Mangala; Leaman, Daniel P.; Lovendahl, Klaus N.; Benhaim, Mark A.; Hodge, Edgar A.; Zwick, Michael B.; Lee, Kelly K.
title: Cryo-ET of HIV reveals Env positioning on Gag lattice and structural variation among Env trimers
date: 2021-09-01
journal: bioRxiv
DOI: 10.1101/2021.08.31.458345
sha: 143beeacfab8cbcf1b38e89a8002b7405de349e4
doc_id: 334981
cord_uid: jphbh5a9

HIV-1 Env mediates viral entry into host cells and is the sole target for neutralizing antibodies. However, Env structure and organization in its native virion context has eluded detailed characterization. Here we used cryo-electron tomography to analyze Env in mature and immature HIV-1 particles. Immature particles showed distinct Env positioning relative to the underlying Gag lattice, providing insights into long-standing questions about Env incorporation. A 9.1Å sub-tomogram averaged reconstruction of virion-bound Env in conjunction with structural mass spectrometry revealed unexpected features, including a variable central core of the gp41 subunit, heterogeneous glycosylation between protomers plus a flexible stalk that allows Env tilting and variable exposure of neutralizing epitopes. Together, our results provide an integrative understanding of HIV assembly and structural variation in Env antigen presentation.

heterogeneous glycosylation between protomers plus a flexible stalk that allows Env tilting and 24 variable exposure of neutralizing epitopes. Together, our results provide an integrative 25 understanding of HIV assembly and structural variation in Env antigen presentation. 26 27 Keywords: HIV Env glycoprotein, HIV assembly, Gag-Env interaction, Cryo-electron 28 tomography, sub-tomogram averaging, structural mass spectroscopy. 29 30

Human Immunodeficiency Virus-1 (HIV-1) continues to infect nearly two million people 32

worldwide each year, with no available vaccine against the virus (Pandey and Galvani 2019). The 33 envelope glycoprotein (Env) on the surface of HIV-1 is an essential viral entry machine that 34 mediates binding to host cell receptors and subsequent membrane fusion. As the sole target for 35 neutralizing antibodies, Env is also of singular importance for vaccine design efforts (van Gils and 36 Sanders 2013). Env is translated as a precursor gp160 protein, which trimerizes, and is cleaved 37 into gp120 and gp41 subunits that are primarily responsible for receptor binding and fusion, 38 respectively. The gp120 and gp41 subunits remain as non-covalently associated heterodimers with 39 

In immature particles, the Gag-MA layer is consistently arrayed underneath the inner 117 membrane ( Figure 1A , D and 2C), whereas in mature particles where the Gag polyprotein has been 118 proteolytically cleaved, the MA layer appears fragmented ( Figure 2G ), indicating that MA 119 rearranges during particle maturation. Notably, Env trimers in the mature particles are not 120 colocalized with the MA layer ( Figure 2G ). Given this ultrastructural reorganization following 121 maturation, it seems possible that disruption of membrane-associated Gag lattices may be a 122 prerequisite for Env trimers to gain necessary mobility to mediate membrane fusion ( A total of 32802 sub-volumes were used to reconstruct a C3-symmetric sub-tomogram 129 averaged structure of hVLP-Env with local resolution ranging between 5.7 -12 Å and a global 130 resolution of 9.1 Å ( Figure 1C , 3A, B and Figure S3 ) ( Table 1) . Classification of the sub-volumes 131 did not yield any other distinct conformational class, indicating that the hVLP-Env were in a 132 predominantly closed, pre-fusion state. The Env ectodomain connects to the membrane via a thin 133 tripod stalk ( Figure 1C and 3B). Though the TMD can be observed spanning the membrane layer 134 in raw tomograms ( Figure 1A ), no evidence of TMD is seen in the sub-tomogram averaged 135 structure ( Figure 3B ), similar to the Env structure from immature hVLPs. 136

Hydrogen-deuterium exchange mass-spectroscopy (HDX-MS) analysis of hVLP-Env from 137 intact particles showed that key fiducial peptides which report on trimer integrity ( confirmed by good agreement seen in fitting the atomic structure of detergent extracted full-length 150 Env (PDB ID: 6ULC) into the sub-tomogram averaged map ( Figure 3A, B) . 151 152 Differences in gp41's terminal HR2 helix and stalk influence MPER surface exposure 153

Significant differences relative to available structures are evident, however, in the gp41 154 subunit of hVLP-Env. In nearly all known Env structures, the HR2 helix has a long, rod-like 155 conformation with Asp664 forming its distal tip (Kwon, Pancera et al. 2015, Ward and Wilson 156 2015) . In contrast, in hVLP-Env, Gln-653 forms the distal tip of the HR2 helix, making the helix 157 length 10 amino acids shorter ( Figure 3B ). The remaining residues in the C-terminal portion of 158 HR2 helix bend and form a thin stalk connecting the ectodomain to the membrane ( Figure 3B ). 159

Apart from hVLP-Env, this bend in HR2 helix has only been observed in the full-length, detergent-160 solubilized Env structure derived from strain 92UG037.8 (Pan, Peng et al. 2020 ). In the presence 161 of membrane, we observe that the thin stalks, corresponding to Env residues 654-664, form a tripod 162 that elevates the ectodomain ~10 Å above the membrane ( Figure 3B ). As a result, the bulk of the 163 MPER (Membrane-Proximal External Region) (residues 660-683), which is a desirable vaccine 164 target of some of the most broadly neutralizing HIV-1 antibodies isolated to date (Gray, Madiga 165 et al. 2009 , Rantalainen, Berndsen et al. 2020 , is primarily embedded within the membrane in our 166

The complete MPER peptide is unresolved in all reported structures of trimeric Env, owing 168 to its membrane proximal location and/or possible variations in its conformation. Current 169 knowledge of MPER structure is only derived from constructs without the Env ectodomain structures from different HIV strains shows that the general position of the bulk of the ectodomain 178 is ~10-12 Å above the interpreted membrane surface in all cases ( Figure S7 ). However, differences 179 arise in the position of HR2 helix and subsequent MPER sequence leading to variability in MPER 180 presentation and accessibility above the membrane amongst HIV-1 strains. 181 182 183 184 A flexible gp41 stalk allows Env orientational freedom and MPER epitope exposure 185

In our tomographic data, we observe tilting of hVLP-Env with respect to the membrane 186 ( Figure 4A) , which leads to a lower density level for the stalk in our sub-tomogram averaged map 187 ( Figure 4B ). Thus, the tripod stalk of Env is flexible in its native state, allowing differential 188 sampling of MPER and other membrane-proximal epitopes. Although tilting or lifting of Env from 189 the membrane on binding to MPER targeted antibody 10E8 (Lee, Ozorowski et al. 2016, 190 Rantalainen, Berndsen et al. 2020) and gp120-gp41 interface targeting antibody 35O22 (Huang, 191 Kang et al. 2014) has been suggested, our results provide the first direct evidence of Env tilting on 192 native membrane in its natural, unliganded state. 193

Sensitivity to MPER broadly neutralizing antibodies (bnAbs) has been shown to be 194 impacted by Env stability in its prefusion state (Kim, Leaman et al. 2014) , with increased 195 neutralization observed after cellular receptor engagement (Frey, Peng et al. 2008, Chakrabarti, 196 Walker et al. 2011). Based on the structural data discussed above, we envision that local structure 197 of HR2, height of the Env ectodomain above membrane, and propensity of Env tilting have 198 additional impact on Env sensitivity or resistance to MPER bnAbs. To test this hypothesis, we 199 conducted comparative neutralization assays using MPER bnAbs against HIV-1 strains whose Env 200 structures have been determined. As anticipated from the structural data, amongst the three viruses, 201

ADA.CM hVLP, BaL-1 and BG505, ADA.CM was overall more resistant to MPER bnAbs ( Figure  202 4C and Table S2 ). Neutralization and mutational analysis of proximal residues in ADA.CM that 203 contribute to trimer stability ( Figure S8 ) showed modest individual effects but none were 204 individually responsible for ADA.CM's enhanced resistance to MPER bnAbs ( Figure S8 and 205 Table S3 ). Another prominent difference in the gp41 subunit of hVLP-Env is absence of complete 210 density for the central C-terminal HR1 helices (HR1-C) ( Figure 5A ). The HR1-C helices 211 comprised of amino acid residues 570-595 have been a hallmark of high resolution Env SOSIP 212 structures with the central trimeric bundle composed of a HR1-C helix from each of the gp41 213 domains. However, in our structure, only partial density for the HR1-C helix can be observed 214 corresponding to gp41 residues 570-577 and 582-595 ( Figure 5A ). Classification of Env sub-215 volumes with a tight cylindrical mask for the gp41 region did not yield any subset with higher 216 density for HR1-C helices in our data. Considering that our sub-tomogram averaged map has a 217 local resolution range for the ectodomain between 5.7-12 Å ( Figure S3 ), it is clear from 218

comparisons with other cryo-EM maps at similar resolutions, that the absence of complete HR1-219 C helix density in our structure is not an effect of map resolution ( Figure S9 ). 220

To ascertain whether the observed HR1-C density was affected by applied symmetry to the 221 map, we calculated an asymmetric structure of hVLP-Env at 10.7 Å resolution ( Figure S3 and 222 Table 1 ). The asymmetric structure ( Figure 5B ) appears overall similar to the three-fold 223 symmetrized map ( Figure 5A ). Partial density for the HR1-C helices is consistently present across 224 all three protomers in the asymmetric structure, with near-complete density for HR1-C observed 225 in one protomer ( Figure 5B) . analysis shows that the peptides in HR1-C region exist in a stable, ordered conformation in hVLP-231

Env even though they do not rigidly conform to Env's global three-fold structural symmetry. 232 233 An extensive and heterogenous glycan shield in hVLP-Env protects critical epitopes 234

The hVLP-Env structure displays distinguishable density features for N-linked glycans that 235 decorate the ectodomain surface. In contrast to most available structures where density near to 236 glycosylation sites often covers only the core GlcNAc sugars, in our unliganded hVLP-Env 237 structure, extended electron density for glycan moieties is observed across the protein. N276, N361 and N461 ( Figure 6B ). These glycans have also been shown to overlap with the 247 VRC01 epitope using molecular dynamic studies, but previously determined structures using 248

SOSIPs resolved only short glycan chains at these sites (except for N276) resulting in minimal 249 apparent glycan interactions with the bnAb (Stewart-Jones, Soto et al. 2016). In hVLP-Env, 250 extended density for these multiple glycans can be seen to clearly deter binding of CD4 epitope-251 directed nAbs ( Figure 6B ). These observations are corroborated in VRC01 based neutralization 252 assays which show that hVLP-Env is substantially more resistant than BaL-1 Env ( Figure 6B and 253 Table S2 ), whose structure lacks extended glycosylation around this site (Li, Li et al. 2020 Similarly, long-range extended densities for N-linked glycans are observed in the gp41 258 domain corresponding to residues N262, N448, N611 and N637. The close spacing of N262/N448 259 to N611 likely orders them such that they are visible in the averaged map ( Figure 6A , C). 260

Collectively, these glycans form a protective barrier over the fusion peptide (FP) in hVLP-Env 261 ( Figure 6C ). Env's FP is essential for viral cell entry but is surprisingly exposed to solution and Extensive glycan densities also persist in the asymmetric structure of hVLP-Env but 269 strikingly, we observe variations in densities between different protomers, indicating heterogeneity 270 in glycan occupancy. For example, the N355 glycan is present at all three protomers in the 271 asymmetric Env reconstruction but its density varies with respect to size and orientation at this site 272 ( Figure 6D ). Similarly, the conserved N88 glycan, which forms part of the epitope of bnAb 35O22 273 In mass spectrometry analysis of hVLP-Env, these glycan sites, along with others, exhibit presence 276 of a heterogenous population of complex sugars (Table S4) . Thus, though identified heterogeneity 277 in glycan processing of hVLP-Env (Table S4) with the membrane surface ( Figure 6E ). Nevertheless, hVLPs are effectively neutralized by 35O22 286 but show relative resistance to 3BC176 ( Figure 6E and Table S2) , presumably due at least in part 287 to 3BC176's lower binding capacity (Lee, Leaman et al. 2015) . Thus, a combination of 288 heterogenous presence of N88 glycan along with the ability of hVLP-Env to tilt on membrane can 289 account for differential access to these binding epitopes. 290 291

In this study, we have analyzed the structure of HIV particles displaying functional Env. 293

Although the ADA.CM Env trimers used are highly stable in a membrane environment, they 294 notably do not form well-ordered gp140 trimers in solution using the typically employed "SOSIP" 295 mutations (Leaman and Zwick 2013). This is true for the majority of Env sequences that require 296 extensive modifications in order to encourage formation of stable ectodomain trimers (Rawi, 297 that are not compatible with SOSIP modifications and thus have not been structurally characterized 299 so far. 300

In hVLP-Env, we find substantial evidence that key parts of gp41 including the HR1 central 301 helices and flexible stalk are not rigidly fixed relative to the rest of the trimer. Indeed, R.M.S.D. 302 calculations among static high-resolution structures of full-length Env and SOSIP structures show 303 higher deviation in the gp41 domain compared to gp120 domain (Table S5) , despite the strains 304 having greater than 75% sequence similarity (Table S6 ). Likewise, DEER spectroscopy studies of 305 SOSIP trimers have also shown a higher degree of variability in the gp41 domain (Stadtmueller, 306 Bridges et al. 2018), further substantiating the flexible nature of gp41 as inferred from our 307

Our data also demonstrate that even in membrane associated Env, the fusion peptide (FP) 309 is highly dynamic and exposed to solvent. Exposure of such a functionally critical and conserved This likely plays an important mechanistic role by affording fusion proteins the flexibility to bind 323 receptors and refold during membrane fusion. The flexible stalk also impacts the accessibility of 324 key epitopes that are sterically hindered on the membrane-facing side of the trimer ectodomain 325 such as the FP, gp120/gp41 interface, and membrane-associated MPER. Thus, our structural 326 analyses revealed previously uncharacterized Env features that impact presentation of prime 327 epitopes which can help explain differences in neutralization sensitivity across diverse HIV-1 328

Lastly, our structural analyses of Env in immature viral particles, shows that Env is situated 330 directly over two-fold symmetric contacts of the Gag lattice, providing the first structural evidence 331 of a direct physical Env-Gag interaction, helping to inform models for virion particle assembly subunits in hVLP-Env represented using glycans at N88 (orange circle) and N355 (purple oval). 447

Surface rendering of C3-symmetrized map on left followed by the three protomer interfaces from 448 the C1 symmetry hVLP-Env map. Scale bars equal 30 Å. 449 tomogram generation using standard procedures. Tilt series images were aligned using gold bead 497 markers and aligned tilt-series were used to generate a three-dimensional volume using weighted 498 back-projection method. The final tomograms were rotated, binned, and low pass-filtered for 499 visualization. Tilt-series with non-optimal alignments were discarded. In the end, 423 tilt-series 500 were selected for further processing. 501 502 Initial model generation: 503

Eight tilt-series were randomly selected. CTF estimation and correction for these were done using For C3-symmetrized hVLP-Env structure: 519

Tilt-series were imported into EMAN2's sub-tomogram averaging pipeline (Chen, Bell et al. 520 2019). 1k X 1k tomograms were generated within EMAN2 using default parameters. The binned 521 tomograms were then used for semi-automated particle picking in PEET software (Nicastro, 522 Schwartz et al. 2006 ). Consolidated particles from the semi-automated picking were further 523 curated manually using the 3dmod interface (Kremer, Mastronarde et al. 1996) to remove wrongly 524 positioned particle points such as those that were in the membrane or on the inside of VLPs. 2016) was used as initial model after low pass filtering to 60 Å. Sub-volumes were then re-532 extracted at 2X binning using the particle orientations from the 4X binned refinement output. Sub-533 tomogram refinement was repeated using the 2X binned data with a cylindrical mask covering 534 only the ectodomain and outer membrane layer. Particle orientations were locally refined within a 535 30º angular limit from the positions calculated in the 4Xbin refinement. Sub-volumes closer than 536 120 Å when measured between centers were removed. Sub-volumes with low cross-correlation 537 scores were also removed according to EMAN2's default settings. Remaining particles were then 538 re-extracted at the original, un-binned pixel size and re-refined starting from previously determined 539 orientations at 2X binning. Sub-tilt refinement in EMAN2 was then carried out using default 540 parameters following sub-tomogram refinement with un-binned sub-volumes (Chen, Bell et al. 541 2019). For sub-tilt refinement, a threshold mask was used that enclosed only the Env ectodomain 542 portion without any membrane density. The threshold mask was generated using the mask creation 543 tool in the Relion package (Scheres 2012) . The final masked ectodomain map contained 32802 544 sub-volumes with a calculated resolution of 9.13 Å at 0.143 FSC cut-off value. 545 546 For asymmetric hVLP-Env map: 547

The initial sub-volumes used for generation of the asymmetric map were the same as that 548 used for the C3 symmetrized map described above. Refinement strategies were also nearly 549 identical between the two structures except the asymmetric map was generated with C1 symmetry. A total of 1520 sub-volumes from only immature virions were used to generate C3 and C1 556 symmetrized maps of membrane bound Env using 2Xbinned data in EMAN2 (Chen, Bell et al. 557 2019) by similar procedures as described above. 558

In the unmasked maps, a third density layer was observed underneath the membrane 559 bilayer. Relaxing symmetry of the C3-density map to C1, showed a more defined organization in 560 the Gag layer. Hence, a short, cylindrical mask enclosing only the Gag layer was used for local 561 refinement of the Gag layer using C1 symmetry, starting from the final refined positions derived 562 from the C3-symmetrized immature Env map. This focused refinement gave rise to a 23 Å map of 563 the Gag protein layer (0.143 FSC cut-off). The refined Gag-CA density map was fitted back into 564 the Gag-CA layer of the relaxed to C1 symmetry full Env map for further analyses. 565 566 Sub-tomogram averaged structure of BG505-Env from VLPs: 567

Purified BG505-VLPs were mixed with 10nm gold beads (Aurion BSA Gold Tracer 10nm) 568 at a ratio of 15:1 (v/v). The mixture was applied to C-Flat grids and plunge frozen using a Vitrobot 569

Mark IV (FEI Co.) similar to the procedure used for hVLPs above. 570

Frozen grids were imaged using a 300 kV Titan Krios with a Gatan K2 direct electron 571 detector and GIF energy filter with slit width of 20 eV. Tilt-series were collected in a dose-572 symmetric tilting scheme from -54° to +54° with a step size of 3° using Leginon (Carragher, 573 Kisseberth et al. 2000) software. Tilt-series were collected in counting mode at a magnification of 574 53000X, corresponding to a pixel size of 2.58 Å per pixel. The total dose per tilt series was ~64-575 68 e -/Å 2 . A total of 49 tilt-series were collected. 576

Tilt-series were imported into EMAN2's sub-tomogram averaging pipeline (Chen, Bell et 577 al. 2019) . 1k X 1k tomograms were generated within EMAN2 using default parameters. Particle 578 points were picked manually in the e2spt_boxer.py interface (Chen, Bell et al. 2019 ). Further 579 refinement and processing steps were carried out similar to the hVLP-Env sub-tomogram 580 averaging procedures described above. Briefly, sub-volumes were initially extracted at 4xbinning 581 corresponding to a pixel size of 10.32 Å per pixel. Sub-tomogram refinement was carried using a 582 spherical mask including Env on surface and a part of the membrane. The initial model generated 583

for hVLP-Env in Relion (Scheres 2012, Bharat and Scheres 2016) was used as initial model for 584 BG505-Env also with low pass filtering to 60 Å. Sub-volumes were then re-extracted at 2X binning 585 using the particle orientations from the 4X binned refinement output. Sub-tomogram refinement 586 was repeated using the 2X binned data with a cylindrical mask covering only the ectodomain and 587 outer membrane layer. Particle orientations were locally refined, and duplicates were removed. 588

Sub-tilt refinement in EMAN2 was carried out using default parameters (Chen, Bell et al. 2019) . density. In regions where the unoccupied density was larger than the modeled glycan chains 604 available from the reference PDB structures, these extra densities were left un-modeled. 605

The final model of hVLP-Env, as generated above, was used for rigid body fitting into the 606 other sub-tomogram averaged maps of asymmetric hVLP-Env and immature hVLP-Env. All rigid 607 body fitting procedures were carried out using UCSF Chimera (Pettersen, Goddard et al. 2004 ADA.CM.V4 hVLPs were prepared for neutralization assays as described above except omitting 617 the AT-2 inactivation step. Pseudotyped viruses were similarly produced by co-transfection of 618 HEK 293T cells using pSG3ΔEnv backbone plasmid and Env-complementation plasmid. Serial 619 dilutions of antibodies were added to virus and the mixture was incubated for 1 h at 37°C prior to 620 addition to TZM-bl target cells. DEAE-dextran was added to wells to a final concentration of 10 621 µg/ml. After incubating for 72 h at 37°C, cells were lysed, Bright-Glo luciferase reagent (Promega) 622 was added, and luminescence was measured using a Synergy H1 microplate reader (Bio-Tek). vortexing for 30seconds on ice. The bead/protein mixture was then moved to a 0.22µm cellulose 655 acetate centrifuge filter tube (Spin-X, Corning) and spun for 30s at 13,000xG, 0°C. The flow-656 through was transferred to a thin-wall PCR tube and frozen in liquid nitrogen. Samples were 657 thawed on ice and passed over a custom packed pepsin column (2.1 × 50 mm) kept at 15°C with a 658 flow of 0.1% trifluoroacetic acid (TFA), 2% acetonitrile (ACN) at 200 μL/min for 5 minutes. 659

Digested peptides were collected on a Waters XSelect CSH C18 XP VanGuard Cartridge, 660 130Å, 2.5 µm, 2.1 mm X 5 mm before separation on a Waters ACQUITY UPLC Peptide CSH 661 C18 Column, 130Å, 1.7 µm, 1 mm X 100 mm using a gradient of 7 to 14% B over 1 minute, 14 662 to 30% B over 12.5 minutes, and 30 to 50% B over 1 minute, followed by washing with three rapid 663 gradients between 95 and 5% B (A: 0.1% formic acid, 0.025% trifluoroacetic acid, 2% acetonitrile; 664 B: 0.1% formic acid in 100% acetonitrile). The liquid chromatography system was coupled to a 665

Waters Synapt G2-Si Q-TOF with ion mobility enabled. Source and de-solvation temperatures 666 were 70°C and 130°C respectively. The StepWave ion guide settings were tuned to prevent non-667 uniform gas phase proton exchange in the source (Guttman, Wales et al. 2016) . Env on hVLPs was quantified by SDS-PAGE using BG505 SOSIPs as a standard. HDX-MS 677 reactions were initiated by diluting 20 uL of either hVLPs or BG505 SOSIPs with 180 uL 678 deuteration buffer (10 mM Phosphate, 150 mM NaCl, 85% D2O (Cambridge Isotope 679 2µg/mL in the deuterated buffer to serve as a fully deuterated control. The back-exchange level 703 ranged from 12-16% across experiments; deuterium uptake was not corrected. 704

Totally deuterated (TD) samples were prepared by collecting purified peptide eluent following 705 reverse phase LC separation of a pepsin digested undeuterated sample. Following evaporation of 706 the LC elution buffer the peptides were resuspended in HDX PBS pH 7.50 Buffer, deuterated in 707 deuteration Buffer for 1 hours at 65°C, and quenched and frozen as described above. 708

For peptide identification, un-deuterated peptides were collected from the LC system, dried 709 by speed-vac, and resuspended in 5% acetonitrie, 0.1% formic acid for re-injection on an Orbitrap 710 Fusion for MS/MS using EThcD fragmentation. Data was analyzed using Byonic (Protein Metrics) 711

and manually compared to undeuterated sample data. 712 713 Peptic digest analysis: 714

Peptic digest products were collected from the HDX chromatography system, dried by speed-715 vaccuum, and resuspended in aqueous buffer for nanoscale liquid chromatography (nanoLC-MS) 716 using a 90-minute linear gradient from 2-40% acetonitrile. Products were analyzed on an Orbitrap 717 Fusion mass spectrometer (ThermoFisher Scientific) using a high-energy collisional dissociation 718 (HCD) product-dependent electron-transfer/high-energy collision dissociation (EThcD) method 719 with a targeted mass list method with a targeted mass list using HexNAc, HexHexNAc, and 720

Hex2HexNAc m/z (204, 366, and 528 respectively) to trigger EThcD. 721

Glycopeptide data were visualized and processed by Byonic™ (Version 3.8, Protein Metrics Inc.) 722 using a 10 ppm precursor and 10 ppm fragment mass tolerance. Glycopeptides were searched using 723 the N-glycan 309 mammalian database in Protein Metrics PMI-Suite and scored based on the 724

Laboratories)) to a final pH = 7.45. Samples were deuterated for 5 seconds, 60 seconds, 30 minutes

or 3 hours before being diluted 1:1 with ice cold quench buffer (8 M urea

FA)) to a final pH of 2.5. Quenched samples were 682 digested with 30 ug/mL of porcine pepsin (Worthington Labs) under quench conditions for 5 683 minutes on ice. Labeled peptides were purified by high speed centrifugation at 0°C (2 minutes at 684 25,000 rcf) and immediately flash frozen in liquid nitrogen. BG505 SOSIP samples were handled 685 identically to ensure consistent labeling and back exchange. Frozen samples were stored at -80°C 686 until analysis

LC system kept at 0°C using a 500 uL sample loop. Samples were trapped on a Waters ACQUITY 689

UPLC CSH C18 VanGuard 130Å, 1.7 µm, 2.1 mm by 5 mm trap column for 7 minutes with a flow 690 of solvent A

Peptides were resolved over a Waters ACQUITY UPLC CSH C18 130Å, 1.7 µm, 1 x 100 692 mm column using a 20 minute linear gradient of 3% to 50% solvent B (Solvent B: 100%

1% FA) and analyzed using Waters Synapt G2-Si Q-TOF as described above

Following each injection, the sample loop and trap were washed as described above

Deuterium uptake analysis was performed with HD-Examiner (Sierra Analytics) and HX-696

For solution digestion experiments data was extracted using 697

2013) with binomial fitting 698 and bimodal deconvolution

Back-exchange was measured by including bradykinin peptide at 702 assignment of correct c-and z-fragment ions. The true-positive entities were further validated by 725 the presence of glycan oxonium ions m/z at 204 (HexNAc ions)

Supplemental Information: Supplemental information can be found online at ***

Conformational dynamics of a neurotransmitter:sodium symporter in a lipid bilayer

Analysis of HIV-1 Matrix-Envelope Cytoplasmic Tail Interactions

739 "Visualization of the HIV-1 Env glycan shield across scales

Resolving macromolecular structures from electron 742 cryo-tomography data using subtomogram averaging in RELION

Differential processing of HIV 747 envelope glycans on the virus and soluble recombinant trimer

Leginon: an automated system for acquisition of images from vitreous ice 750 specimens

Direct antibody access to the HIV-1 membrane-proximal external region 753 positively correlates with neutralization sensitivity

A complete data 757 processing workflow for cryo-ET and subtomogram averaging

Features and development of 760 Coot

False EX1 signatures caused by 762 sample carryover during HX MS analyses

Maturation of the HIV-1 core by a non-765 diffusional phase transition

A fusion-767 intermediate state of HIV-1 gp41 targeted by broadly neutralizing antibodies

Structure of the membrane proximal 771 external region of HIV-1 envelope glycoprotein

Broad neutralization of human immunodeficiency 775 virus type 1 mediated by plasma antibodies against the gp41 membrane proximal external 776 region

Sequencing of Membrane Remodeling Leading to Influenza Virus Fusion

Antibody potency relates to the ability to recognize the closed, pre-fusion form of HIV 782 Env

CD4-induced activation in a soluble HIV-1 Env trimer

Solution structure, 787 conformational dynamics, and CD4-induced activation in full-length, glycosylated, monomeric 788 HIV gp120

Tuning a High Transmission Ion Guide to Prevent Gas-Phase Proton Exchange During H/D 791 Exchange MS Analysis

Analysis of overlapped and noisy 793 hydrogen/deuterium exchange mass spectra

Optimization of Feasibility Stage for Hydrogen/Deuterium 795 Exchange Mass Spectrometry

797 "Crystal structures of the trimeric human immunodeficiency virus type 1 matrix protein: 798 implications for membrane association and assembly

Broad and potent HIV-1 neutralization by a human antibody that binds the gp41-gp120 805 interface

Structures and distributions of SARS-CoV-2 spike proteins on intact 809 virions

Antibody to gp41 MPER alters functional 811 properties of HIV-1 Env without complete neutralization

Antibody mechanics on a membrane-bound HIV segment essential for GP41-targeted viral 815 neutralization

Fusion peptide of HIV-1 as a site of vulnerability 821 to neutralizing antibody

Computer visualization of three-823 dimensional image data using IMOD

Crystal structure, conformational fixation and entry-related interactions of 832 mature ligand-free HIV-1 Env

Increased functional stability and homogeneity of viral 834 envelope spikes through directed evolution

Antibodies to a conformational epitope on gp41 neutralize HIV-1 by destabilizing the 839 Env spike

Cryo-EM structure of a native, fully 841 glycosylated, cleaved HIV-1 envelope trimer

Subnanometer structures 844 of HIV-1 envelope trimers on aldrithiol-2-inactivated virus particles

The EMBL-EBI search and sequence 848 analysis tools APIs in 2019

Minimizing carry-over in an online pepsin digestion system used for the H/D 851 exchange mass spectrometric analysis of an IgG1 monoclonal antibody

Automated electron microscope tomography using robust prediction 854 of specimen movements

Automated tilt series alignment and tomographic 856 reconstruction in IMOD

Engineering HIV envelope 859 protein to activate germline B cell receptors of broadly neutralizing anti-CD4 binding site 860 antibodies

The 862 molecular architecture of axonemes revealed by cryoelectron tomography

Env Bound With the Fab of Antibody PG16

The global burden of HIV and prospects for control

UCSF Chimera--a visualization system for exploratory research and analysis

UCSF ChimeraX: Structure visualization for researchers, educators, 873 and developers

Structural basis of transmembrane coupling of 876 the HIV-1 envelope glycoprotein

HIV-1 Envelope and MPER Antibody Structures in Lipid 881 Assemblies

Automated Design by Structure-Based Stabilization and Consensus Repair to Achieve 885

Prefusion-Closed Envelope Trimers in a Wide Variety of HIV Strains

Inactivation of human immunodeficiency virus type 1 infectivity with 889 preservation of conformational and functional integrity of virion surface proteins

Clustering and mobility of HIV-1 Env at 892 viral assembly sites predict its propensity to induce cell-cell fusion

RELION: implementation of a Bayesian approach to cryo-EM structure 894 determination

NIH Image to ImageJ: 25 years of 896 image analysis

Structure of the immature HIV-1 capsid in intact virus particles at 8.8 Å resolution

Conformations of HIV-1 SOSIP Envelopes that Show Similarities with Envelopes on Native 903 Virions

Dense Array of Spikes on HIV-1 Virion Particles

The role of matrix in HIV-1 envelope glycoprotein 914 incorporation

Biochemical evidence of a 916 role for matrix trimerization in HIV-1 envelope glycoprotein incorporation

HIV-1 virus-like particles bearing 919 pure env trimers expose neutralizing epitopes but occlude nonneutralizing epitopes

Broadly neutralizing antibodies against HIV-1: 922 templates for a vaccine

Epitope-Independent Purification of Native-Like 925

Envelope Trimers from Diverse HIV-1 Isolates

Topological 928 analysis of the gp41 MPER on lipid bilayers relevant to the metastable HIV-1 envelope prefusion 929 state

Insights into the trimeric HIV-1 envelope glycoprotein 931 structure

Antibody neutralization and escape by HIV-1

Coupling of human immunodeficiency virus type 1 fusion to virion maturation: a novel role of 938 the gp41 cytoplasmic tail

CTF 940 determination and correction for low dose tomographic tilt series

942 "MotionCor2: anisotropic correction of beam-induced motion for improved cryo-electron 943 microscopy