Functional development of a V3-specific broadly neutralizing antibody isolated from a case of HIV superinfection

Stimulating broadly neutralizing antibodies (bnAbs) directly from germline remains a barrier for HIV vaccines. HIV superinfection elicits bnAbs more frequently than single infection, providing clues of how to elicit such responses. We used longitudinal antibody sequencing and structural studies to characterize bnAb development from a superinfection case. Mature bnAb QA013.2 bound both initial transmitted and superinfecting virus, but its inferred naïve bound only the superinfecting strain and was not neutralizing. QA013.2 requires residues spanning FWRH1-CDRH1 to attain breadth, which is uncommon for V3-specific bnAbs. A 4.15 Å cryo-EM structure of QA013.2 bound to heterologous native-like trimer showed recognition of V3 signatures (N301, N332, and GDIR). Antigenic profiling revealed that viral escape was achieved not only by changes in the structurally-defined epitope, but also by mutations in V1. These results highlight shared and distinct properties of QA013.2 relative to other V3-specific bnAbs in the setting of sequential, diverse antigenic variants.


43
Developing a vaccine that elicits broadly neutralizing antibodies (bnAbs) against HIV is 44 considered critical to achieving protection from the extensive diversity of circulating HIV 45 subtypes (Burton and Hangartner, 2016;Doria-Rose, 2010). The potency and protective 46 efficacy of these bnAbs has been demonstrated in passive immunization trials (Pegu et al.,  (Briney et al., 2016;Doria-Rose and Joyce, 2015). Such antibody lineage 59 reconstruction has been performed for a handful of HIV bnAbs, with particular focus on core 60 bnAb epitopes including the CD4-binding site, the variable loop 1/2 (V1/V2) apex, and the 6 detected in the sequence data through 765 dpii, suggesting that either we did not sample 162 sequences containing these mutations, or that these mutations accumulated between 765 dpii 163 and the time when QA103.2 was isolated at 2282 dpii. Thus, we can only reliably infer the 164 lineage through the early stages of development; the later stages of the bnAb's evolutionary 165 pathway are beyond the scope of the computationally inferred lineage (Fig. 1B,C). Despite this 166 lack of resolution, the lineage and longitudinal development that we were able to decipher offer 167 a means to dissect the determinants of antibody binding, potency, and breadth for QA013.2.

168
Here we defined neutralization breadth as the mediation of cross-clade heterologous 169 neutralization of Tier 2 HIV pseudoviruses as measured by the TZM-bl assay.

QA013.2 heavy chain maturation enables antigen binding and neutralization. 199
To determine whether key residues mediating breadth were represented in the inferred 200 intermediate sequences defined through 765 dpii, we selected these intermediates (Int6VH and 201 Int5VL ; Fig 1), as well as inferred naïve heavy and light chain sequences and paired them 202 together in various combinations ( Fig. 2A) to test their capacity to bind and neutralize HIV 203 relative to the mature QA013.2 bnAb. The resulting monoclonal antibodies (mAbs) were tested 204 for binding to autologous and heterologous Envelope protein (Env) by biolayer interferometry 205 (BLI). The inferred naïve had weak but detectable binding to the Env-gp120 from a clade A 206 autologous superinfecting virus variant from 765 dpii; there was even weaker binding to 207 heterologous clade A BG505.SOSIP.664 trimer (Fig. 2B,C). While these data were insufficient 208 for calculating reliable binding kinetics, they suggest that the naïve BCR of the QA013.2 bnAb 209 lineage is capable of recognizing the superinfecting virus. The inferred naïve antibody did not 210 detectably bind to the clade D virus from initial infection ( Fig. 2 -figure supplement 1).

212
By contrast, the mature QA013.2 bnAb exhibited strong binding kinetics to both the autologous 213 SI gp120 and heterologous BG505 Env trimer, with equilibrium dissociation constants of 94.9 214 nM and 3.2 nM, respectively. A mAb consisting of the mature heavy chain paired with the naïve 215 light chain (matVH 0VL) bound to autologous SI gp120, however the binding affinity was much 216 weaker relative to the mature bnAb and not suitable for calculating binding kinetics. This same 217 mAb bound to heterologous BG505 Env trimer (KD = 89.6 nM) with an affinity that was 218 approximately 30-fold less than that of the mature bnAb (Fig. 2B,C). The converse mAb pairing 219 (0VH matVL) did not result in any binding capacity to either tested antigen.

221
Intermediate mAb Int6VH Int5VL, with 6% total SHM, represents the final inferred intermediates 222 from the deep sequencing data, and they were detected in the sampled NGS sequences at 99-223 100% nucleotide identity, further validating the sequence of these inferred intermediates (Supp.

224
File 2). We could not detect binding of the intermediate mAb to the gp120 of the autologous SI 225 virus, but there was clear evidence of binding to heterologous BG505 Env trimer (KD = 513 nM) 226 ( Fig. 2C), demonstrating HIV Env specificity. However, the binding affinity to heterologous 227 trimer was more than 100-fold lower when compared to the mature bnAb. Together these 228 studies demonstrate that VH is important for facilitating initial HIV recognition and binding. This 229 finding is supported by data collected for both autologous Env-gp120 and heterologous native-

254
We tested the same lineage mAbs for neutralization using a subset of HIV Env pseudoviruses that were neutralized by the mature QA013.2 bnAb, including the same clade A autologous Env detected between the intermediate mAb (Int6VHInt5VL) and the heterologous BG505 Env trimer, 259 it did not neutralize the same virus (Fig 2C,D) (Horns et al., 2019;Laserson et al., 2014). Indeed, we observed low sequence read depth 267 across all QA013 PBMC sample timepoints and antibody isotypes ( Table 1). When the mature VH was paired with the naïve light chain (matVH0VL), autologous and cross-270 clade Tier 2 heterologous neutralization was achieved, suggesting that the approximate 271 timeframe in which neutralization-mediating mutations were acquired in VH likely occurred 272 between 765 dpii (resolution of computationally inferred lineage) and 2282 dpii (isolation of 273 mature bnAb QA013.2). The naïve heavy chain paired with the mature light chain (0VHmatVL) did 274 not result in any detectable neutralization of the virus panel (Fig. 2D), suggesting that affinity 275 maturation of the heavy chain is critical for imparting both binding and neutralization. When

276
Int5VL was paired with the mature heavy chain (matVH Int5VL), there was an average four-fold 277 increase in neutralization potency for all viruses tested compared to matVHnaïveVL (Fig. 2D).

278
Cryo-EM structure of QA013.2 Fab bound to Env trimer demonstrates widespread 279 antibody contacts and shared structural characteristics with other V3-specific bnAbs. 280 To better identify which residues contribute to QA013.2's ability to bind and neutralize Env, we 281 performed single particle cryo-electron microscopy (cryo-EM) on the QA013.2 Fab and 282 BG505.SOSIP.664 trimer complex (Fig. 3A,B). The known structure of BG505.SOSIP.664 283 trimer (PDB ID: 5aco) was fit into the resulting 4.15Å resolution EM density map along with a 284 homology model for the variable domain of QA013.2 Fab (Fig. 3A,B). The map density

293
From the cryo-EM structure, it is apparent that QA013.2 interacts not only with the N332 glycan 294 as previously reported (Williams et al., 2018), but also with the glycan at site N301, and the 295 linear GDIR motif at the base of the V3 loop ( Fig. 3C-E). Key regions of the antibody paratope 296 include the CDRH1 and CDRH2 which are nearest the N301 glycan, as well as CDRH3 and 297 CDRL3, which are in close proximity to the N332 glycan. We observe that the CDR loops of the 298 heavy chain contain many polar and charged residues, which likely facilitate hydrogen bonding 299 interaction with the Env glycans and the charged residues in the GDIR motif ( Fig. 3 -figure   300 supplement 1A). The CDRH3 loop is distinctively long (21 residues) and appears to tilt away from the rest of the heavy chain towards the light chain, generating a cleft between CDRH2 and 302 CDRH3 and exposing an extended interface that has been observed for other V3-specific 303 bnAbs including 10-1074 and PGT121 ( Fig. 3 -

304
Seven key residues in the Fab CDRH3 loop (L103 -I109) lie adjacent to the conserved GDIR 305 motif. Based on our sequencing data of the sampled repertoire of this subject, these amino 306 acids (excluding Y105 and D107) were likely introduced during affinity maturation of the heavy 307 chain between 765 dpii and 2282 dpii because they were not detected by NGS through 765 dpii 308 ( Fig. 1C), suggesting a potential role for these residues in facilitating antibody breadth and/or

Substitutions in FWRH1-CDRH1 are required for bnAb neutralization breadth. 321
With paratope data in hand, we had the opportunity to show which regions of QA013.2 were 322 playing a role in its functional activity by dissecting the regional and individual residue 323 contributions interpreted through the structure. We began by examining the functional 324 importance of VH mutations surrounding the CDR loops that were present in the mature 325 antibody but not in the ≤765 dpii inferred intermediates. We synthesized six additional variations 326 of QA013.2 VH, illustrated in Fig. 4A, that we based on Int6VH as the template and included all 327 CDR-localized and CDR-adjacent FWR residues present in the mature QA013.2 VH (Supp. File 328 3). All VH chimeras were paired with the mature light chain (matVL) and tested for neutralization 329 capacity (Fig. 4A). Mutations in individual CDRs, including adjacent FWR mutations, were not 330 sufficient to achieve potent neutralization comparable to the mature antibody when paired with 331 the mature light chain. We did observe weak autologous neutralization when mutations 332 surrounding FWRH3-CDRH3 were added to Int6VH. Mutations spanning both FWRH1-CDRH1 333 and FWRH3-CDRH3 of the mature VH together were required to obtain heterologous 334 neutralization, consistent with the cryo-EM map of key antigen contacts ( Fig. 3 and 4A).

335
Interestingly, the determinants of autologous and heterologous neutralization in this antibody 336 lineage were different, in which autologous virus neutralization required less SHM and was 337 facilitated by residues spanning CDRH2-FWRH3-CDRH3 (Fig. 4A). To determine the 338 importance of the framework residues and examine loss of function, we reverted all framework 339 mutations in the mature VH back to the inferred naïve residues (∆FWR). This reversion resulted 340 in no detectable binding or neutralization activity, demonstrating that residues in the heavy chain 341 FWRs are essential for QA013.2 binding and neutralization, as has been noted for other HIV 342 bnAbs (Fig. 4A,B) (Klein et al., 2013).

343
In a complementary approach, we used the mature VH as a template and reverted the 344 mutations spanning FWRH1-CDRH1 (∆CDR1VH) and CDRH3 (∆CDR3VH) back to naïve and as 345 above, paired these variants with the mature light chain (Fig 4A). We observed a 89-fold 346 increase in half maximal inhibitory concentration (IC50) neutralization of autologous 347 superinfecting virus and nearly complete ablation of all tested heterologous virus neutralization 348 when the six consecutive residues spanning FWRH1 and CDRH1 (∆CDR1VH) were reverted 349 back to naïve. Reversion of these six residues also led to weakened binding to heterologous 350 BG505.SOSIP.664 trimer (KD = 40.6 nM) relative to the mature bnAb ( Fig. 4A,B). When the and heterologous virus was maintained, but there was an impact on potency across all viruses 353 tested; although this antibody could still neutralize a heterologous clade C variant, it did not 354 neutralize SF162 at the highest antibody concentration tested ( Fig. 3 and 4A

358
To further interrogate the above hypothesis, we generated single point mutations in the mature 359 VH at residues of interest in FWRH1-CDRH1 and CDRH3 that reverted them to naïve ( Fig.   360 5A,B). These heavy chain point mutants were also paired with the mature light chain and tested 361 for binding to Env and neutralization. We found that four of the six residues spanning FWRH1-362 CDRH1 individually contributed to the mature bnAb's ability to neutralize autologous and 363 heterologous virus. Residues E26, G28, and I24 were most critical, with average increase in 364 IC50 of 191-, 27-, and 25-fold, respectively, across all pseudoviruses tested when these residues 365 were reverted to naïve (Fig. 5C,D). While reversion of individual CDRH3 residues did not 366 abolish neutralization, which correlates with the chimeric antibody results (Fig. 4), one residue 367 in particular (D106) demonstrated a much higher IC50 value when mutated back to naïve. D106 368 is adjacent to the GDIR motif in the structure. Reversion of this residue to naïve did not inhibit 369 clade A autologous virus neutralization, but did substantially impact heterologous neutralization,

370
with an average 40-fold increase in IC50 across all tested pseudoviruses (Fig. 5E). However, the 371 D106V point mutant only resulted in a modest decrease in binding to heterologous 372 BG505.SOSIP.664 trimer relative to the mature bnAb (Fig. 5F). These data suggest that while 373 D106 is not critical for Env binding, this residue does contribute to heterologous neutralization, 374 likely by interacting directly with the linear GDIR motif. Taken together, it is clear that although 375 the antibody paratope of QA013.2 includes the CDRH3, where interaction with the N332 glycan 376 and GDIR motif occurs, the QA013.2 paratope is also comprised of critical residues in FWRH1 377 and CDRH1.

QA013.2 light chain confers higher bnAb potency. 418
Though less impactful than the heavy chain, our aforementioned data suggested that light chain 419 development indeed contributes to QA013.2 neutralization potency (Fig 2). To identify which 420 regions of the light chain were essential for these contributions, we reverted regions of SHM antibody binding to BG505.SOSIP.664 heterologous Env (KD = 1.3 µM) relative to the mature 424 bnAb by more than 400-fold. This effect was less pronounced when residues in the CDRL3 425 were reverted to naïve (Fig. 6A,B). These light chain variants also demonstrated reduced 426 neutralization potency relative to the mature bnAb. Reversion of residues present in FWRL2-

427
FWRL3 resulted in an average 19-fold increase in IC50 (range 6-42 fold) and reversion of 428 residues in CDRL3 causing a 14-fold increase in average IC50 (range 1-41 fold) across all 429 pseudoviruses tested when compared to the mature bnAb IC50 values (Fig. 6A). Together, 430 these data demonstrate that residues in FWRL2-FWRL3 and CDRL3, which are situated along 431 the cleft between heavy and light chains and near the N332 glycan, contribute to the mature 432 bnAb's binding affinity and neutralization potency, with residues in FWRL2-FWRL3 appearing to 433 have a larger influence on bnAb binding. Notably, the residues spanning FWRL2-FWRL3 do not 434 interact directly with the N332 glycan on Env. However, the residues in FWRL3 may influence 435 the conformation of residues in the adjacent CDRL3 loop, which interacts with the N332 glycan.

436
Furthermore, the residues in FWRL2 interact with FWRH4 of the Fab heavy chain (residues 437 117-120) which follows the CDRH3 loop (Fig. 6C). Attempts to identify the light chain residues 438 in FWRL2-FWRL3 responsible for mediating this interaction were inconclusive, as reversion of 439 the residues located in FWRL2 and CDRL2-FWRL3, separately, did not influence the IC50 440 values of the tested pseudoviruses, indicating that the cumulative effect of these mutations was 441 beneficial (Fig. 6A).    (Fig. 7). In addition to the more 470 traditional signatures of Env escape from V3-specific bnAbs, including mutations that disrupt the 471 N301 and N332 potential N-linked glycosylation sites (PNGS) or alter the GDIR motif (Dingens 472 et al., 2019), we also observed viral escape via mutations in the V1 loop (Fig. 7A,B). In addition 473 to the two PNGS, enrichment of viral escape mutations was strongest at sites D325, R327, and 474 H330; validation of these data using pseudoviruses with individual mutations in TZM-bl 475 neutralization assays showed that mutations to these sites can result in >32-fold change in IC50 476 relative to the wildtype BG505.T332N (Fig. 7B,C). However, not all mutations to these sites

481
Viral escape in the V1 loop occurred at two sites that are adjacent in the BG505 structure; D141 482 and R151 (HXB2 numbering), both of which caused a modest ~3-fold change in IC50 relative to 483 the wildtype virus (Fig. 7B,C). While the V1 loop of Env does not appear to be in direct contact 484 with QA013.2 Fab in the complex structure, the backbone Cα atoms of the Fab CDRH3 loop 485 and Env V1 loop are within 8Å of each other (Fig. 7D), suggesting there could be amino acid 486 side-chain interactions between the two loops that can affect neutralization.

487
Interestingly, while the glycan at N332 is essential for bnAb neutralization and disrupting this 488 motif results in viral escape, we observed that it is not required to achieve antibody binding -as 489 demonstrated by the mature bnAb binding to the gp120s from the autologous clade D and 490 BG505.W6.C1 viruses, both of which lack the PNGS at site 332 ( Fig. 7 -figure supplement 1).

491
It is likely that this binding, while measurable, is too weak to confer neutralization, confirming the 492 importance of the N332 glycan.

522
HIV bnAbs, was not found to be polyreactive (Fig. 8 -figure supplement 1). We present data 523 detailing the lineage development and structure-function relationships of QA013.2 using 524 antibody variable region deep sequencing and single particle cryo-EM respectively. The high-525 resolution structure of QA013.2 Fab bound to the heterologous native-like Env trimer revealed 526 specific regions of the bnAb paratope that interact with HIV Env, which guided targeted mutational approaches to identify the antibody determinants of potency and breadth. From our 528 structural data and functional investigations, we demonstrate that while QA013.2 shares traits 529 with other V3-specific bnAbs, such as targeting conserved glycans and the linear GDIR motif at 530 the base of V3, it also relies less on the CDRH3 loop, which often plays a pivotal role in 531 achieving bnAb neutralization breadth. Rather, residues spanning FWRH1-CDRH1 region were 532 most critical for QA013.2 neutralization breadth. Thus, our results highlight universal aspects of 533 the core V3 epitope on Env as well as the multitude of ways in which bnAbs evolve to target it.

535
Biolayer interferometry data collected on the QA013.2 bnAb lineage showed that in contrast to 536 many other V3-specific bnAbs, the inferred naïve antibody was capable of binding autologous 537 gp120 monomer from the Env protein of the infecting HIV strain, in this case, the superinfecting 538 strain. While we were unable to quantify the binding affinity of the inferred naïve antibody, it has 539 been estimated that affinities in the range of ≤ 1 µM are needed to initiate naïve BCR activation,    includes key framework residues in both the heavy (E23-I25) and light chains (FWRL2 and 559 FWRL3) (Fig. 8). In this case, we observed that without the acquired mutations in FWRL2-bnAb affinity and antibody-antigen complex formation. The importance of framework 562 substitutions to HIV bnAbs has been documented, in which they often provide flexibility that 563 facilitates enhanced binding affinity and in some cases, interact directly with antigen (Klein et 564 al., 2013). Additional involvement of framework regions in QA013.2 may result from a "tilted" 565 angle of approach adopted by the heavy chain of the Fab. A similar configuration was observed 566 in 10-1074 and PGT121 bnAbs (Mouquet et al., 2012), In this case, the CDRH3 loop is tilted 567 away from the rest of the heavy chain, towards the light chain resulting in a large portion of the 568 CDRH3 loop being exposed at the antigen interface of the N332 glycan. We propose that this 569 may facilitate additional interactions with this core glycan. The tilting also brings FWRH4 on the 570 opposing side of the CDRH3 loop into close proximity with the light chain. Despite the minimal 571 affinity maturation in this region, FWRH4 (residues 117-120) was found to interact with residues 572 in the light chain FWRL2, likely stabilizing the conformations of the CDRH3 and CDRL3 loops, 573 which in turn, contributes to the overall binding affinity of this bnAb. In contrast to other V3-574 targeting bnAbs, QA013.2 required residues outside of the CDRH3 region in FWRH1-CDRH1 to 575 achieve cross-clade neutralization breadth. Based on our targeted mutagenesis studies, the 576 CDRH3 as a whole did not appear to mediate heterologous neutralization breadth. Only D106 577 alone was critical for cross-clade neutralization. The CDRH3 of the mature bnAb contains many 578 hydrophobic residues, which could be implicated in Van der Waals interactions with the surface 579 of Env or possibly in shielding water molecules from interacting with the charged residues of the 580 GDIR motif, thus enabling D106 to interact with this linear motif instead. Alternatively, the 581 charged D106 residue may be important for electrostatic interactions with the polar GDIR motif.

582
From the cryo-EM structure, the FWRH1-CDRH1 region is adjacent to the conserved N301 583 glycan, highlighting this interaction as a possible driver of neutralization breadth. It's possible 584 that the N332 glycan required for neutralization may have served as the initial anchor for the 585 QA013.2 lineage and subsequent affinity maturation to accommodate and interact with N301 586 promoted neutralization breadth as has been previously reported for sialic acid-bearing glycans

625
Collectively, our study reveals several features of V3-specific bnAb development in a case of 626 HIV superinfection. These include paratope differences for autologous and heterologous 627 neutralization as well as non-overlapping determinants of neutralization breadth and potency such results. In addition, our data highlight features of the conformation-dependent V3 epitope 630 that could be valuable for immunogen design, including short V1 regions and accommodation of 631 the conserved glycan at N301 as a potential driver of neutralization breadth. Designing 632 immunogens that lack nearby obstructive Env features may influence nAb accommodation of 633 N301 and drive subsequent breadth as has been shown previously for a VRC01-class bnAb and 634 the N276 glycan (Umotoy et al., 2019, p. 01). In summary, our findings suggest a strategy that

682
Following next-generation sequencing, sequence reads were pre-processed into amplicons using FLASH, primers were trimmed using cutadapt, and the FASTX-toolkit was used to remove 684 sequence reads containing low-confidence base calls (N's) as previously described (Simonich 685 et al., 2019;Vigdorovich et al., 2016). Sequence reads from each timepoint were merged 686 together into an aggregate dataset that was deduplicated and annotated using partis 687 (https://github.com/psathyrella/partis) with default settings. Sequences that were out of frame or 688 contained internal stop codons were removed, while singletons, or sequences that were 689 observed only once in the sampled repertoire, were included in an attempt to retain 690 undersampled or rare sequences. Sequencing statistics for this study can be found in Table 1.

692
Clonal family clustering was performed on both individual timepoint and aggregated datasets using both partis seeded and unseeded clustering methods (Ralph and Matsen, 2016). The of interest. The 'seeds' here refer to the previously-isolated QA013.2 VH and VL sequences 696 (Williams et al., 2018). Unseeded clustering, on the other hand, clusters all sequences in the 697 repertoire into clonal families. Since this is more computationally demanding, some of the very 698 large datasets were subsampled down to 50-150k sequences. Three random subsamples were 699 analyzed and compared in order to measure statistical uncertainties. No subsampling was done 700 for the seeded analyses.

702
Because the light chain has much lower diversity than the heavy chain, and because B cells   customized settings to ensure that likelihood and posterior estimated sample sizes were >100 for the heavy and light chain lineages. For both VH and VL lineage inference, MCMC iterations 734 were 100,000 with a thinning frequency of 100 iterations and 10,000 burn-in samples.
Linearham generates summary graphics following phylogenetic inference that display only the edges that satisfy the given posterior probability threshold, and only the nodes that contact 737 edges above the threshold. These graphics detail the relative confidence in unique inferred 738 sequences and amino acid substitutions ( Fig. 1 and Fig. 1 -figure supplement 2). Inferred 739 intermediate sequences present in the most probable lineage paths were selected for study 740 (see Figure 1).

742
To determine whether the computationally-inferred naïve and intermediate ancestor sequences

743
for VH and VL lineages were observed in the sampled NGS sequences, we ran the following 744 script: https://github.com/matsengrp/cft/blob/master/bin/blast.py . A local BLAST database was 745 created for the two seeded clonal families and queried for sampled NGS sequences that had 746 high nucleotide sequence identity to each of the VH and VL lineage members using the 'blastn' 747 command (Biopython package). E-value of 0.001 was used; other settings were default. Blastn 748 "hits" for each lineage member were sorted according to their percent nucleotide identity and 749 alignment length. Percent nucleotide identity of the top blastn "hit" for each lineage member is 750 listed in Figure 1.

752
Antibody heavy and light chain sequences were synthesized as "fragmentGENES" by Genewiz harvested after 3-6 days. Supernatants were collected following centrifugation of cells at 1,900 761 rpm for 10 minutes. Antibody supernatants were passed three times over columns packed with 762 immobilized Protein G resin (Thermo Fisher, catalog #20397) as previously described (Scherer 763 et al., 2014).

765
Antibody chimeras were synthesized as fragmentGENES as shown in Fig. 4A, while VH 766 mutagenesis (Fig. 5) was performed using phosphorylated primers (25 µM) that introduced the 767 mutation of interest into the variable heavy chain, using mature QA013.2 VH DNA in the IgG1 768 plasmid as a template (2.5 ng/µL). PCR-mediated mutagenesis was performed using Phusion

769
Hi Fidelity polymerase (Thermo Fisher Scientific) with the following thermocycler settings: 770 denature at 98˚C for 1 min, 30 cycles of 98˚C for 30 sec, annealing at 55-70˚C for 30 sec, 771 extension at 72˚C for 5 min, and final extension at 72˚C for 10 min. PCR products were digested 772 with DpnI (New England BioLabs) overnight at 37˚C and purified using Qiagen PCR Purification 773 kit, with products eluted in ddH20. Mutants were ligated at room temperature for at least two 774 hours prior to transformation into One Shot™ TOP10 chemically competent cells (Thermo 775 Fisher Scientific). Mutant antibody chains were sequence-confirmed prior to transfection and 776 antibody purification as described above.

834
All data processing steps were performed within the Relion (version 3.0.7) software setup 835 (Scheres, 2012). Frame alignment and dose-weighting were performed using MotionCor2 836 (Zheng et al., 2017), and CTF estimation was done using CTFFIND4 (Rohou and Grigorieff, 837 2015). Further processing was continued with Relion. A total of 1,082,742 particles were picked 838 using the LOG-based automatic particle picking routine. The particles were initially extracted at 839 4x binning with a pixel size of 10.8Å/pixel. The binned particle stack was subjected to two 840 rounds of unsupervised 2D classification, out of which classes that clearly showed SOSIPs with 841 three Fabs bound were selected. A total of 113,470 particles from these classes were used for 842 3D refinement. The previously published negative stain reconstruction of BG505.SOSIP.664 complexed to QA013.2 Fab (EMD-7471) was used as an initial model (Williams et al., 2018).

844
The initial model was low-pass filtered to 60 Å and refinement was performed on the full particle 845 stack with C3 symmetry imposed. Subsequent map sharpening and post processing in Relion 846 resulted in a 4.186 Å structure using the "gold-standard" FSC cutoff of 0.143. The unbinned 847 particle stack was further subjected to CTF-refinement at per micrograph level which improved 848 the map to its final resolution of 4.15 Å (Fig. 3 -figure supplement 2).

850
The atomic model of BG505.SOSIP.664 trimer (PDB ID: 5ACO) (Lee et al., 2015) was rigid 851 body fitted into the 4.18 Å resolution map using UCSF Chimera (Pettersen et al., 2004). No  DNA libraries (Haddox et al., 2018) were incubated with the QA013.2 bnAb at ≥ IC95 878 concentration for one hour. One million SupT1.CCR5 cells were then infected with these mutant 879 libraries in the presence of 100 µg/mL DEAE-dextran. In parallel to QA013.2 antibody selection, 880 each mutant virus library was also infected into 1x10 6 SupT1.CCR5 cells without antibody 881 selection to serve as the non-selected control. Non-integrated viral DNA was isolated from each 882 pool of infected cells at 12 hours post infection using a Qiagen Miniprep kit. The env gene was 883 amplified from each sample (selected and non-selected) using a barcoded subamplicon 884 sequencing approach that generates seven tiling subamplicons across env as previously 885 described (Haddox et al., 2018). These pools of amplicons were then deep sequenced on an 886 Illumina HiSeq using 250 bp paired-end reads.

888
Following deep sequencing, differential selection values were calculated as described in Doud

897
To validate the mutational antigenic profiling data, TZM-bl neutralization assays were performed 898 using BG505.T332N pseudoviruses bearing single point mutants representing residues that 899 were enriched for viral escape at individual sites across Env. TZM-bl neutralization assays were 900 performed as described above in at least two independent assays, each performed in technical

976
Partis software was run on the aggregated IgG NGS dataset from QA013 using the 'seeded'

1000
Sequence file includes all VH and VL chimeras (presented in Figures 4 and 6), and VH mutants 1001 (presented in Figure 5) generated to interrogate QA013.2 functional evolution in this study.