SARS-CoV-2 spike glycosylation affects function and neutralization sensitivity

The glycosylation of viral envelope proteins can play important roles in virus biology and immune evasion. The spike (S) glycoprotein of severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) includes 22 N-linked glycosylation sequons and 17 O-linked glycosites. Here, we investigated the effect of individual glycosylation sites on SARS-CoV-2 S function in pseudotyped virus infection assays and on sensitivity to monoclonal and polyclonal neutralizing antibodies. In most cases, removal of individual glycosylation sites decreased the infectiousness of the pseudotyped virus. For glycosylation mutants in the N-terminal domain (NTD) and the receptor binding domain (RBD), reduction in pseudotype infectivity was predicted by a commensurate reduction in the level of virion-incorporated spike protein. Notably, the presence of a glycan at position N343 within the RBD had diverse effects on neutralization by RBD-specific monoclonal antibodies (mAbs) cloned from convalescent individuals. The N343 glycan reduced overall sensitivity to polyclonal antibodies in plasma from COVID-19 convalescent individuals, suggesting a role for SARS-CoV-2 spike glycosylation in immune evasion. However, vaccination of convalescent individuals produced neutralizing activity that was resilient to the inhibitory effect of the N343 glycan.


Introduction
by antibodies (27,28). To comprehensively understand the role of spike glycans in viral 88 infectivity and antigenicity, we individually mutated each of the 22 N-linked glycosylation 89 sites in the spike protein as well as two O-linked sites (S323 and T325) in the RBD. We 90 found mutations introduced at many glycosylation sites in the NTD and RBD reduced 91 pseudotype infectivity, and the magnitude of this effect was predicted by the magnitude 92 of the loss of S incorporation into virions. Furthermore, while the S protein levels in 93 virions had little effect on neutralization sensitivity, the presence or absence of a glycan 94 on N343 in RBD governed the sensitivity to some monoclonal antibodies cloned from 95 convalescent individuals. Moreover, the glycan at N343 reduced neutralization 96 sensitivity to polyclonal antibodies from convalescent individuals, but this evasive effect 97 imparted by glycosylation was overcome by plasma antibodies from the same 98 individuals who were subsequently vaccinated. 99 purified S protein (S-6P-NanoLuc (29)) and recombinant p24 CA proteins as standards 111 on near-IR fluorescent western blotting (LiCor) and assuming 1500 capsid (CA) protein 112 subunits per mature HIV-1 particle (30, 31) (Fig. S1A). While the amount of S 113 expression plasmid used for transfection had little effect on the amount of virions 114 produced (Fig. 1A), the amount of S protein incorporated into virions correlated with the 115 amount expressed in cells. S incorporation into virions reached a plateau (~30 ng S/ml 116 in virions, or around 300 spike trimers per virion) when 0.125 μg or 0.25 μg of an S 117 expression plasmid was co-transfected with the HIV-1 proviral plasmid (Fig. 1A,1B,118 1C). 119 120 To assess the effect of the number of spikes on pseudotype infectiousness, the titers of 121 pseudotyped viruses were measured on 293T cells expressing ACE2 (293T/cl.22). Co-122 transfection with as little as 2 ng of S expression plasmid produced virions that induced 123 1000x the level of luciferase observed with 'bald virions', indicating that small amounts 124 of S protein (0.15 ng/ml or a mean of ~1-2 S trimers per virion) were sufficient to 125 mediate viral entry (Fig. S2,Fig. 1B and 1C). Nevertheless, pseudotype virion infectivity 126 increased with increasing spike numbers. Indeed, infectivity and the number of spikes 127 was approximately linearly correlated, in the range 1 spike to 300 spikes per virion (Fig. 128

Reduced infectious virion yield conferred by SARS-CoV-2 spike glycosylation site 133 removal 134
To investigate the contribution of glycosylation to S protein function, we generated 22 135 spike substitution mutants, each containing a single Asn to Asp (N to D) substitution at 136 one of the 22 N-glycosylation sequons. Additionally, we generated a mutant with Ala 137 replacements at potential O-linked sites S323 and T325 in the receptor binding domain 138 (RBD) (13). None of these substitutions affected the levels of the S protein in 139 transfected cells, but some of them reduced S incorporation into virions ( Fig. 2A). 140 Specifically, N61D, N122D, N165D, N343D, and S323A T325A that fall within S1, that 141 includes the NTD and RBD, exhibited 10-fold or greater reductions in the levels of 142 virion-associated S protein, suggesting that these glycans affect S protein transport or 143 virion incorporation. Conversely, removal of the glycosylation sites in S2 had either no 144 or minor effects on S protein incorporation into virions ( Fig. 2A). Measurements of the 145 yield of infectious pseudotyped particles carrying each of the substitutions indicated that 146 several substitutions in the NTD and RBD markedly reduced infectivity. For example, 147 the N61D substitution reduced infectivity by almost 50-fold (Fig. 2B, Fig. S2) while 148 substitutions at glycosylation sites in the RBD, namely N331D, N343D, and S323A 149 T325A, resulted in 5-to 10-fold reduction in infectivity (Fig. 2B, Fig. S2). Western blot 150 analyses revealed that the amount of S protein in virions was directly correlated with 151 infectivity (Fig. 2B), suggesting a potential role for most glycans during synthesis or 152 folding of spike trimers, or their incorporation into virions, rather than in S protein 153 function after virion incorpration. In contrast, two substitutions (N1194D and N657D) did 154 not change the amount of S protein in virions, but substantially reduced infectivity (Fig.  155 2B, Fig. S2). This finding suggests a functional role for N1194 and N657 glycans in 156 spike conformation or stability on virions, or function during viral entry. Overall, we 157 conclude that a subset of glycosylation sites is important for the incorporation of S 158 protein into virions, while others are required for optimal particle infectivity. 159 160 Impact of spike density and RBD glycosylation on neutralization sensitivity 161 he RBD is the major target of neutralizing antibodies. To determine the effect of glycan 162 removal on sensitivity to neutralizing antibodies, we focused on glycosylation sites in the 163 RBD (N331 and N343), and one site adjacent to the RBD (N282). Because these 164 glycosylation sites affected the level of spike incorporation into virions, we first tested 165 the effect of SARS-CoV-2 spike density on neutralization sensitivity, as this parameter 166 could be a potential confounder. We harvested pseudotyped virions from cells 167 transfected with varying amounts (from 2 ng to 1 μg) of wild-type S expression plasmid 168 and tested their sensitivity to C144, a potent class 2 neutralizing human monoclonal 169 antibody cloned from a convalescent individual (34) (Fig. 3A). Notably, varying the 170 levels of WT S protein on virions over a wide range (Fig. 1B) had no discernable effect 171 on neutralization sensitivity to C144. 172 We next tested the neutralization sensitivity of the RBD glycosylation site mutants 173 bearing Asn to Asp mutation at N-linked sites (N331D and N343D) or proximal (N282D) 174 to the RBD or alanine substitutions at O-linked sites S323 and T325. To provide 175 matched control viruses with approximately similar numbers of S trimers and similar 176 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. ; levels of infectivity, neutralization of the glycosylation site mutants was compared to 177 virions generated with an appropriate level of the WT spike protein. We compared the 178 neutralization sensitivity of WT and mutant virions to C144 and another potent class 2 179 neutralizing antibody, C121, both of which target the ACE2 binding site. While the 180 N282D, S323/T325A, and N331D mutant pseudotypes exhibited neutralization 181 sensitivities that were similar to the WT pseudotype, the N343D substitution conferred 182 significantly increased neutralization sensitivity to both C144 and C121 (Fig. 3B). 183 Specifically, the half maximal inhibitory concentration (IC50) of C144 was reduced from 184 1.8 to 3.4 ng/ml (against the WT pseudotype) to 0.45 ng/ml (against the N343D 185 pseudotype), while the C121 IC50 was reduced from 2.3 to 2.9 ng/ml (against the WT 186 pseudotype) to 0.21 ng/ml (against the N343D pseudotype). 187 188 Positive and negative effects of RBD glycosylation on sensitivity to human 189

monoclonal antibodies 190
To determine the effects of glycosylation more broadly on SARS-CoV-2 sensitivity to 191 neutralizing antibodies, we used pseudotyped viruses bearing spike proteins with 192 R683G substitution, which ablates the furin cleavage site. This substitution does not 193 affect S incorporation into virions (Fig. S1B) but enhances particle infectivity (7). The 194 glycosylation site mutations had a smaller effect on spike incorporation into virions in 195 the R683G context (Fig. S3A). Nevertheless, transfection of cells with 1 μg of N282D, 196 S323A T325A, and N331D S expression plasmids generated pseudotyped viruses with 197 S protein amounts and infectious properties comparable to those generated by 198 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 transfection of 30-100 ng of WT S expression plasmid ( Fig. S3A and S3B). Transfection 199 of cells with 1 μg of the N343D spike expression plasmid yielded pseudotyped virus 200 particles carrying a similar amount of S protein to those generated by cells transfected 201 with 10 ng of the R683G S protein expression plasmid (Fig. S3A), while the particle 202 infectivity was similar to those from cells transfected with 3 ng of the WT R683G 203 expression plasmid (Fig. S3B). To evaluate the effect of glycosylation site mutations on 204 sensitivity to neutralizing antibodies, we generated WT and mutant virion stocks bearing 205 similar amounts of spike protein and then assessed their susceptibility to neutralization 206 by a panel of RBD-specific human monoclonal antibodies of the various neutralizing 207 For two class 4 antibodies, C022 and C118, the N343D substitution affected the slope 241 of the neutralization curves and conferred partial resistance at high of antibody 242 concentrations (Fig. 3C). For example, C022 at 2000 ng/ml almost completely 243 neutralized the wild-type pseudotypes but only inhibited the N343D pseudotypes by 244 ~50%. A similar trend was seen for a second class 4 antibody, C118. 245

246
Overall, the N343D substitution had a range of positive and negative effects on 247 neutralization sensitivity that varied greatly dependent on the nature of the particular 248 monoclonal antibody tested. To better understand the molecular basis for the impact of 249 N343 glycan on antibody neutralization, we inspected the structures of some of the 250 aforementioned antibodies in complex with spike ( Fig. 4A, B, C). The glycosylation site 251 at N343 (shown in red in Fig. 4A, B, C) is distinct from the binding site of the class 1 252 antibody C098 -suggesting that the effects of the glycan on sensitivity to class 1 253 antibodies are mediated through effects of the glycan on RBD conformational dynamics 254 and epitope exposure (Fig. 4B). In contrast, for the class 2 antibodies C121 and C144, 255 N343 is proximal to the antibody bound to the neighboring spike subunit (Fig. 4C). Since 256 removal of the glycan increased sensitivity to these antibodies (Fig. 3C), it is likely that 257 the N343 glycan partly shields these class 2 antibody epitopes. For class 3 antibodies, 258 N343 protrudes towards the C135 binding site on the same S subunit (Fig. 4), 259 potentially explaining incomplete neutralization by this antibody (Fig. 3C). For the class 260 4 antibody C118, N343 is distal to the antibody binding site on spike (Fig. 4C), 261 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 suggesting that N343 glycan is not involved in direct antibody binding and likely exerts 262 changes in neutralization through effects on RBD conformational dynamics. 263

264
Other RBD substitutions (S323A/T325A and N331D) did not alter the neutralization 265 sensitivity to most monoclonal antibodies tested herein (Fig. S4, Fig. S5). However, the 266 N282D substitution, which lies outside the RBD, had marginal effect on sensitivity to the 267 class I antibodies C099, C936 and the class 3 antibody C581 (Fig. S6). Given that the N343D substitution exhibited different effects on sensitivity to neutralizing 272 monoclonal antibodies, we next asked whether this substitution affected neutralization 273 by polyclonal antibodies present in convalescent plasma. As with the monoclonal 274 antibodies, we compared the neutralization sensitivity of N343D pseudotyped particles 275 with that of WT pseudotyped particles containing same amount of WT spike protein or 276 showing the same infectivity as the N343D pseudotype, to convalescent plasma from 15 277 patients collected early in the COVID19 pandemic (at 1.3 months after infection) and 278 from the same individuals ~1 year later following subsequent vaccination (34) (36). The 279 N343D mutant pseudotypes were more sensitive than the WT pseudotypes to 280 convalescent plasma collected at 1.3 months after infection ( Fig. 5A and Fig. S7). 281 Indeed, the 50% neutralization titers (NT50) were a mean of 5.1-fold greater for the 282 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. ; https://doi.org/10.1101/2023.06.30.547241 doi: bioRxiv preprint N343D mutant as compared to WT pseudotypes (p=0.0022) (Fig. 5B). Notably, the 283 difference in neutralization sensitivity between N343D mutant (mean NT50=26179) and 284 WT S pseudotypes (mean NT50=20853) was negligible (p=0.2805) when plasmas 285 collected from the same individuals 1 year later and after subsequent vaccination were 286 tested. Of note, these subsequently collected plasma exhibited higher neutralization 287 potency than those collected shortly after infection (Fig. 5). We conclude that the glycan 288 at N343 confers protection against neutralization by antibodies generated shortly after 289 SARS-CoV-2 infection but that this effect is lost against antibodies from the same 290 individuals who are subsequently vaccinated. 291 292

Discussion 293
While SARS-CoV-2 spike pseudotyped viruses have been widely used as a surrogate to 294 study S-mediated viral entry (28) little is known about the varying effect of S level on 295 virions on particle infectivity and neutralization sensitivity. Previously it was reported that 296 approximately 8 HIV-1 trimer-receptor interactions are required for HIV-1 to infect a 297 target cell (38), a result that is broadly consistent with our finding that a small number 298 (minimally an average of 1-2 spike per virion) is sufficient for detection of SARS-CoV-2 299 pseudotype infection. To achieve neutralization, monoclonal antibodies must encounter 300 prefusion spikes and reduce the number of functional spike trimers below the threshold 301 required for infection. We found that an increased level of S on pseudotyped virions is 302 associated with increased infectivity but had little effect on neutralization sensitivity to 303 monoclonal antibodies. This result suggests that monoclonal antibodies are in excess 304 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. ; https://doi.org/10.1101/2023.06.30.547241 doi: bioRxiv preprint over functional spikes in a typical neutralization assay, and their measured potency is 305 limited by their affinity, rather than by the amount of functional spike protein in a 306 pseudotype neutralization assay. 307

308
In this study, we found that glycosylation at several sites in the spike NTD and RBD, 309 including N61, N122, N165, N331, N343 and S323 T325 are required for full 310 pseudotyped virus infectivity. Substitution of these glycosylation sites resulted in a 311 reduction in pseudotype infectivity of about 10-fold or more and correlated with a 312 reduction in spike incorporation into virions. While it is unclear precisely how 313 glycosylation on these sites drives S incorporation into virions, glycans could in principle 314 affect protein stability and trafficking through the secretory pathway (39). We note, 315 however, that the glycosylation site mutations did not affect steady state levels of S in 316 transfected cells. Compared with other glycosylation sites in the S1 domain, these sites 317 are highly conserved among sarbecovirus S proteins (40). Likewise, the conservation of 318 these glycosylation sites is observed among the major SARS-CoV-2 variant lineages, 319 including Alpha, Beta, Gamma, Delta and Omicron (41). Some studies, largely 320 employing molecular dynamics simulations, have suggested glycans on spike proteins 321 affect the conformational dynamics of the spike's RBD "up" and "down" states (10) (42). 322 In particular, the glycan at N343 stabilizes RBD states in a process termed "glycan 323 gating" (22,43,44). Our findings show that the N-glycan at N343 affected incorporation 324 into virions, suggesting that the conformational state, might affect trimer assembly or 325 stability, transit through the secretory pathway or incorporation into pseudotyped virions. 326 327 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. ; https://doi.org/10. 1101/2023 The N343 glycan also affected neutralization by RBD-targeted monoclonal antibodies, 328 and the effect was largely dependent on the class to which antibodies belong. The N343 329 glycan is positioned on the RBD distal to the ACE2 and class 1 antibody (C098) binding 330 site, but increased sensitivity to class 1 antibodies (C098, C099, C936), which bind only 331 to "up" RBDs (24), suggesting that this effect is mediated through the impact of the 332 glycan on RBD conformation. Conversely, the N343 glycan reduced sensitivity to class 333 2 antibodies, including C121 and C144, both of which can bind both 'up' and 'down' 334 RBDs. Cryo-EM structures (24) (Fig. 4) show that N343 is located proximally to the 335 C121 and C144 antibody bound to the neighboring subunit, in a manner that might 336 interfere with antibody binding. Overall, our results are consistent with a model in which 337 N343 glycan affects sensitivity to class 1 and class 2 antibodies by affecting the RBD 338 conformational dynamics (43) and also potentially by sterically hindering class 2 339 antibody binding into neighboring RBDs in the "down" conformation ( Fig. 4). 340

341
The N343 glycan is proximal to the binding sites of class 3 antibodies, and the N343D 342 substitution had a range of effects on sensitivity to class 3 antibodies. Three clonally 343 related antibodies, C032, C080 and C952, were unaffected by the N343D substitution, 344 while substantial but opposing effects were seen for two others, C135 and C581. In the 345 case of C135, removal of the glycan reduced the fraction of virions that resisted 346 neutralization at high antibody concentration. A possible explanation for this 347 phenomenon is that the N343 site is heterogeneously glycosylated, and some 348 subfraction of the glycans occlude the C135 binding site. For C581, removal of the 349 N343 glycan had the opposite effect, reducing neutralization sensitivity. In this case it 350 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. ; https://doi.org/10.1101/2023.06.30.547241 doi: bioRxiv preprint seems likely C581 mimics the properties of two previously described cross-reactive 351 class 3 antibodies, S309 and SW186, for which the N343 glycan constitutes part of the 352 antibody binding site (25) (26). Finally, for the two class 4 antibodies, the glycan at N343 353 affected the character of the neutralization curve. Since in these cases the antibody 354 binding site is on the opposite face of the RBD to that of the glycan, it is likely that these 355 effects are mediated through alteration of spike conformational dynamics and exposure 356 of the class 4 epitope, which is shielded in the RBD 'down' conformation. 357

358
The N343 glycan reduced neutralization by convalescent plasma collected at 1.3 359 months after infection, echoing the effects on sensitivity to the C121, C144 (class 2) and 360 C135 (class 3) antibodies. Conversely, neutralization by convalescent plasma from the 361 same individuals (collected at 12 months following subsequent vaccination) was 362 unaffected by the N343 glycan. That neutralizing antibodies are relatively sensitive to 363 glycan-mediated protection early after infection may be a reflection of the fact that the 364 initial neutralizing response is based to a large extent on class 2 antibodies that are 365 easily escaped (29, 34). Subsequent increases in antibody affinity and neutralizing 366 potency and a shift in the RBD epitopes that are recognized, following months of affinity 367 maturation and boosting by vaccination (7,35,36,45) results in neutralizing antibodies 368 that are mostly unaffected by the N343 glycan. In sum, while the SARS-CoV-2 N343 369 glycan affects both spike conformation and neutralization sensitivity shortly after 370 infection, antibody evolution confers sufficient potency and breadth to combat glycan-371 mediated immune evasion. 372 373 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made Medium (DMEM) supplemented with 10% fetal bovine serum (Sigma F8067) and 419 gentamycin (Gibco). All cell lines used in this study were monitored periodically to 420 ensure the absence of retroviral contamination and mycoplasma. 421

438
. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made

Western blot analysis 439
Cell lysates were separated on NuPage Novex 4-12% Bis-Tris Mini Gels (Invitrogen). 440 Proteins were blotted onto nitrocellulose membranes. Thereafter, the blots were probed 441 with primary antibodies and followed by secondary antibodies conjugated to IRDye 442 800CW or IRDye 680. Fluorescent signals were detected and quantitated using an 443 Odyssey scanner (LI-COR Biosciences). 444

Neutralization assays 450
To measure neutralizing activity of monoclonal antibodies, serial dilutions of antibodies 451 beginning at 3 μg/ml were four-fold serially diluted in 96-well plates over seven dilutions. 452 To determine the neutralizing activity in convalescent plasma, the initial dilution started 453 at a 1:30 (for plasma at 1.3 months) or a 1:150 (for plasma at 12 months). Thereafter, 454 SARS-CoV-2 spike pseudotyped viruses were incubated with monoclonal antibodies or 455 the convalescent plasma for 1 hr at 37°C in 96-well plates. The mixture was then added 456 to 293T/ACE2cl.22 target cells seeded one day prior to infection so the final starting 457 dilutions were 1.5 μg/ml for monoclonal antibodies and 1:60 or 1:300 for plasma. Cells 458 were then harvested 48 hours post infection for NanoLuc luciferase assays. 459 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made plates with pseudotyped viruses. The S1 incorporated into viruses (on right y axis) was 478 determined by quantitative Western Blot, using purified recombinant S-6P-NanoLuc 479 proteins as standard, representative of three independent experiments. The mean and 480 range of two technical replicates are shown. 481

Virion associated Spike
Virion associated Spike (ng/ml) . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. ; https://doi.org/10.1101/2023.06.30.547241 doi: bioRxiv preprint

(C) Characterization of virions pseudotyped with spike expression plasmids as in (B). 482 virus infectiousness (on y axis) was plotted against S trimers per virion (on x axis), 483
which was determined by quantitative Western blot, using purified recombinant S-6P-484 NanoLuc proteins as standard. Representative of three independent experiments. 485 486 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 (RLUs) following infection of 293T/ACE2.cl22 cells in 96-well plates with pseudotyped 497 viruses. The S1 incorporated into viruses (on x axis) was determined by quantitative 498 Western blot. Glycosylation site mutants are shown in black and wild-type spike is 499 shown in red. The mean and range of two technical replicates are plotted. 500 501 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 neutralizing monoclonal antibody C144. Infectivity was quantified by measuring 509 NanoLuc luciferase levels (RLU). The mean and range of two technical replicates are 510 shown.

511
(B) Quantification of glycosylation site mutant N282D, S323A T325A, N331D, and 512 N343D or wild-type spike pseudotyped virus infection in the presence of the indicated 513 concentrations of the class 2 neutralizing monoclonal antibody C144 (left) or C121 514 (right). Infectivity was quantified by measuring NanoLuc luciferase levels (RLU). The 515 mean and range of two technical replicates are shown. including class 1 (C098, C099, and C936), class 2 (C121 and C144), class 3 (C032, 520 C080, C135, C581, and C952), and class 4 (C022 and C118) antibodies. As controls, 521 glycosylation intact spike expression plasmid (WT in the R683G background) was 522 transfected at two doses, 10 ng or 3 ng, and the neutralization sensitivity of the resulting 523 viruses were assessed in parallel. The mean and range of two technical replicates are 524 shown. The C135 neutralization graph is depieced with both a linear and logarithmic Y-525 axis to more clearly show effects of glycosylation on the completeness of neutralization 526 at high antibody concentrations 527 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. Representative of two independent experiments. 561 (B) Western blot analysis of virions pelleted through 20% sucrose from 100 μl 562 supernatant harvested at 48 hours after transfection with 0.008 μg, 0.024 μg, 0.073 μg, 563 0.22 μg, or 0.67 μg of wild-type spike expression (furin uncleavable R683G background) 564 along with envelope-deficient HIV-1 proviral plasmid expressing NanoLuc luciferase. 565 The blot was probed with anti-p24 antibody using recombinant HIV p24 protein as 566 standard ( (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023  (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023 NanoLuc luciferase activity (RLU). Virus generated in the absence of S (0 μg), shown in 576 thin black line, was used as a background control. 293T/ACE2.cl22, as target cells, 577 were infected with the indicated volumes of pseudotyped viruses in 96-well plates and 578 harvested 48 hours post infection for NanoLuc luciferase assay. The mean and range 579 deviation from two technical replicates are shown. 580 581 . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023  including class 1 (C098, C099, and C936), class 2 (C121 and C144), class 3 (C032, 602 C080, C135, C581, and C952), and class 4 (C022 and C118) antibodies. As controls, 603 glycosylation intact spike expression plasmid (WT in the furin uncleavable R683G 604 background) was transfected at two doses, 10 ng or 3 ng, and the resulting viruses 605 were assessed for neutralization sensitivity in parallel. The mean and range of two 606 technical replicates are shown. . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. class 1 (C098, C099, and C936), class 2 (C121 and C144), class 3 (C032, C080, C135, 625 C581, and C952), and class 4 (C022 and C118). As controls, glycosylation intact spike 626 (WT in the furin uncleavable R683G background) was transfected at two doses, 10 ng 627 or 3 ng, and the resulting viruses were assessed for neutralization sensitivity in parallel.

628
The mean and range of two technical replicates are shown. . CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. CC-BY 4.0 International license available under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprint this version posted June 30, 2023. ;https://doi.org/10.1101https://doi.org/10. /2023