Assessing bnAb potency in the context of HIV-1 Env conformational plasticity

bnAbs, per definition, inhibit genetically divergent virus strains and retain high activity against Tier-2 viruses known to largely maintain a closed prefusion Env conformation [7,17,20,44,45]. While effects of modulating Env stability on neutralization activity have been described for some bnAbs [15,16,46,47], a refined understanding of how Env conformational plasticity affects neutralization efficacy across bnAb types remains important. To compare bnAb activity over the full spectrum of highly flexible to less flexible Envs, we used Tier-based neutralization data available in the CATNAP database (http://hiv.lanl.gov/catnap) [48] (Fig 1). We selected 21 prototypic bnAbs covering the five major bnAb epitopes for this analysis (CD4-binding site (CD4bs) bnAbs VRC01, NIH45-46, PGV04, 3BNC117, N6, N49P7 and VRC07-523; V2-glycan bnAbs PG9, PG16, PGDM1400, and PGT145; the gp120-gp41 interface bnAb PGT151; the V3-glycan bnAbs PGT121, PGT 128, PGT130, PGT135, BG18 and 2G12; and bnAbs against the membrane proximal external region (MPER) in gp41: 4E10, 10E8, and 2F5). Several weak-nAbs that represent Ab specificities prevalent at high titers in HIV-1 infection were chosen for comparison [49,50]. These included nAbs to the variable region 3 (V3-crown: 447-52D and 1–79), the CD4-induced epitope (CD4i: 17b) and agents directed to the CD4bs (nAbs 1F7 and b12, and the CD4 receptor analogue CD4-IgG2). Access to these epitopes requires either partial or full, spontaneous or receptor-induced conformational opening of the trimeric Env complex [9,13,51]. Geometric mean IC50 values stratified by neutralization Tier showed the expected pattern with weak-nAb activity being almost exclusively restricted to Tier 1 strains (Fig 1). This is in line with the more tightly closed Env prefusion conformation of higher tiered Envs disfavoring antibody attack by increased conformational shielding and/or altering entry kinetics [14–17,25,34].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Ab-specific variations in neutralization potency against HIV-1 strains assigned to different neutralization tiers.

All IC50 data available from the Los Alamos National Laboratory database (http://hiv.lanl.gov/catnap; as of October 2024) was downloaded for the antibodies listed (grouped by epitope as indicated) against all HIV-1 strains that had been attributed to a neutralization tier (tier 1A: most sensitive; tier 3: most resistant, see S1 Table for the neutralization data available per Ab and tier). A. Distribution of IC50 values for each antibody and for HIV-1 strains of different tiers (dots: median, boxplot: values within the interquartile range; whiskers indicating minimum and maximum values were omitted for clarity). B. Summary of differences between median potencies of Abs against Tier 1A and Tier 2 strains. Significant differences (p < 0.05, Mann-Whitney test) are indicated by a red star.

https://doi.org/10.1371/journal.ppat.1012825.g001

Interestingly, side-by-side comparison of the inhibitory activity of different bnAb classes against tiered viruses revealed, that bnAbs also benefit from some degree of Env trimer opening. A gradual decrease in potency with increase in Tier-level was evident for most bnAbs (Fig 1A). CD4bs and MPER directed bnAbs inhibited Tier 1A viruses most potently. V2-glycan bnAbs and the interface directed bnAb PGT151 showed a multi-phase activity pattern with Tier 1A and Tier 3 strains recording the lowest neutralization sensitivity and peak potency against Tier 1B or Tier 1B/2 classified strains. Patterns of V3-glycan bnAbs were diverse, but most of the tested V3-glycan bnAbs showed highest potency against either Tier 1A or Tier 1B viruses. While differences in median IC50 between Tier 2 and Tier 1A for bnAbs were in most cases within 0.5–1 orders of magnitude (Fig 1B), the differences for weak-nAbs remained the most pronounced (>1.5 orders of magnitude) leading in several cases to a switch from no activity to potent neutralization which is prototypical for mAbs against CD4i and V3-crown epitopes [11,16,52].

Harnessing perturbations in Env conformation to classify bnAb activity

The higher activity of most bnAbs against lower-tier viruses suggests a general effect of relaxing Env conformational stability on bnAb potency. We therefore hypothesized, that patterns of neutralizing activity based on Env stability could be used to classify nAbs. To explore this in a controlled setting, we utilized a panel of single-site gp120 mutants of the subtype B Tier 2 strain JR-CSF covering 125 of the 503 gp120 amino acid positions [38,53,54] (Figs 2 and S1). To select mutants with a generally enhanced neutralization sensitivity, JR-CSF mutant pseudoviruses were screened against plasma from people with chronic, untreated HIV-1 infection (N = 18; S2 Table), the CD4 receptor analogue CD4-IgG2 and a range of weak-nAbs (CD4bs: b6 and b12, V3-crown: 1–79, 447-52D).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Characterizing the generalized neutralization sensitivity of JR-CSF envelope point mutants.

The neutralization sensitivity of 126 JR-CSF envelope (Env) mutants was determined in a TZM-bl based pseudovirus inhibition assay. A. Each mutant virus was titrated with eleven plasma samples from HIV-1 chronically infected individuals (nine subtype B infected, two infected with non-B subtypes). All eleven plasmas neutralized JR-CSF wild-type virus at an inhibitory dilution (ID50) > 100, the minimal dilution probed. The distribution of ID50 ratios (mutant/wt) is shown for each mutant (center line: median; box limits extend from the 25th to 75th percentiles; whiskers indicate minimum and maximum values). A dotted line at ID50 ratio (mutant/wt) = 1 indicates equal neutralization sensitivity of the mutant Env compared to the wildtype reference. Thresholds used for categorizing mutants into highly (ID50 ratio (mutant/wt) >10; colored red) and moderately (10 > ID50 ratio (mutant/wt) >4; colored blue) sensitive phenotypes are indicated by dotted lines. B. IC50 values determined for each mutant against CD4 binding site (CD4-IgG2 and mAbs b6 and b12) and V3-crown (mAbs 1–79 and 447-52D) directed inhibitors were compared with median plasma ID50s as determined in A. Colorcode as in A. Titrations of each plasma and antibody on all viruses was done once, except for the JR-CSF wt reference for which plasma ID50s and nAb IC50s were derived by fitting a curve to pooled datapoints from n ≥4 titrations.

https://doi.org/10.1371/journal.ppat.1012825.g002

Consistent with the limited cross-neutralizing activity observed in HIV-1 infection, JR-CSF wild-type (wt) was neutralized by only a subset of plasmas (11/18) (S1A and S1B Fig). While the majority of gp120 mutations did not lead to a relevant change in neutralization sensitivity, 23 mutants displayed a high (>10-fold above wt) and 7 mutants a moderate (4–10 fold above wt) median increase against plasma (Fig 2A). Notably, these mutants also recorded high and modest neutralization sensitivity against plasma samples, that were inactive against JR-CSF wt (S1A and S1B Fig) and across plasmas from individuals with different subtype of infection (S1C Fig), demonstrating a general release of conformational shielding in these 30 mutants. This was supported by the inhibition data with weak-nAbs. Only few mAb-mutant combinations deviated from this overall pattern of increased sensitivity, suggesting that in these cases the mutation directly affected the mAb epitope. Collectively, the 30 neutralization sensitive mutants recorded with a >10-fold enhanced sensitivity to at least one V3-crown mAb (1–79, 447-52D) and one CD4bs directed inhibitor (b6, b12 or CD4IgG2) (Fig 2B and S3 Table).

Known resistance inducing mutations were confirmed for some bnAbs. However, in contrast to weak-nAbs, bnAbs showed no or comparatively moderate increases in neutralizing potency across the mutant panel regardless of their specificity, with only a few bnAb-mutant combinations reaching a 10-fold increase, most of which fell within the 30 defined mutants with generally enhanced neutralization sensitivity (2G12: 2/2; PGT135: 1/1; 2F5: 9/10; 4E10: 9/9; 10E8: 3/5) (Fig 3A). Notably, the most consistent increase in neutralizing activity against these mutants was observed for MPER bnAbs known to exert their neutralizing activity primarily after CD4 binding, when concomitant Env trimer opening increases MPER exposure [10,33,55]. Modest potency gains were also observed for 2G12 and for PGT135 which, similar to MPER bnAbs, has been found to harbor post-attachment neutralization activity [10]. The CD4bs bnAbs VRC01, VRC07-523, N6 and N49P7 recorded the strongest gain in potency upon mutation of the N197 glycosylation sequon in line with the reported shielding of the CD4bs by this glycan [56,57]. Otherwise, these CD4bs bnAbs showed no increase in neutralization potency >10-fold across the entire panel. Equally no gain in potency was observed for the V2-apex bnAb PGT145 and the V3-glycan bnAbs PGT121, PGT128, and PGT130.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. HIV-1 Env generalized neutralization sensitivity impacts differentially on the potency of broadly neutralizing antibodies (bnAbs).

The whole JR-CSF Env mutant panel was tested for neutralization sensitivity against bnAbs targeting four major epitopes (CD4bs: VRC01, VRC07-523, N6 and N49P7; V3-glycan: PGT121, PGT128, PGT130, PGT135; 2G12; V2-glycan: PGT145; MPER: 2F5, 4E10, 10E8). A. Graphs show the log IC50 ratio (mutant/wt) for each Env mutant. IC50 ratios corresponding to minimal and maximal concentrations of inhibitors probed are indicated with lines. Weak-nAb data from Fig 2B is shown for comparison. The order of Env mutants and shaded regions of interest are indicated on top. Titrations of most antibodies on all viruses was done once (n = 1), except for the JR-CSF wt reference (n ≥ 4). mAbs PGT128, PGT145, 10E8 and 2G12 were titrated twice (n = 2) on all mutants. B. Spearman correlation matrix and hierarchical clustering of neutralization fingerprints based on IC50 data in A and Fig 2. For correlation with PWH plasma the median reciprocal ID50 ratio (mut/wt) over 11 JR-CSF wt neutralizing plasma (Fig 2A) was used. All significant correlations (p-value ≤ 0.05) are indicated with a circle with bigger circles indicating stronger support (lower p-values).

https://doi.org/10.1371/journal.ppat.1012825.g003

Distinct, general patterns of activity of weak-nAbs, bnAbs, and patient plasma across the mutant panel were revealed by Spearman correlation analysis (Fig 3B). Antibody reactivity patterns fell into four categories defined by hierarchical clustering: cluster I characterized by strong positive correlation amongst plasma, weak-nAbs and bnAb PGT135, cluster II comprising MPER bnAbs, 2G12 and N6 with low but consistent positive correlation to cluster I, cluster III with V3-glycan bnAbs (PGT121, PGT128 and PGT130) as well as V2-glycan bnAb PGT145 showing negative correlations to cluster I and cluster IV containing CD4bs bnAbs VRC01, VRC07-523 and N49P7 for which no consistent correlations with other clusters were detected.

The similarity of MPER bnAbs and PGT135 with plasma and weak-nAb inhibition patterns underscores that the activity of these Abs benefits from structural destabilization of the trimer. PGT135 and 2G12 target overlapping epitopes including glycans at positions N295, N332 and N386 [58,59] suggesting that their simultaneous engagement by these two Abs is to some extent facilitated by enhanced conformational flexibility of the trimer. Conversely, a decreased activity of PGT145 against less stable trimers is expected as its quaternary V2-glycan epitope at the trimer apex depends on close vicinity of the gp120 protomers [60]. The data further suggested that PGT130, which like PGT121 and PGT128 binds the GDIR motif at the V3-base but differs in the dependency on glycans located in V1V2 (N137, N156) and V3 (N301, N332) [61,62], may also benefit from a more restrained closed conformation.

Collectively, the reactivity pattern highlighted qualitative differences of bnAbs, and suggests that the most potent and broad CD4bs bnAbs including VRC01, VRC07-523 and N49P7 show a comparatively high tolerance towards perturbation in Env conformational stability. Notably, however, we observed a modest increase in potency related to trimer opening for CD4bs bnAb N6, despite its already high potency.

Generalized increases in neutralization sensitivity link with de-stabilization of the closed prefusion Env conformation

To explore common antigenic features associated with a generalized neutralization sensitivity we probed five moderately and 22 highly neutralization sensitive Env mutants in a cell-based binding assay. Cell-expressed Envs were triggered by increasing doses of soluble CD4 (sCD4) and gradual opening of the trimer measured by the binding intensity of the prototypic CD4i mAb 17b that depends on trimer opening to access its epitope [17,63–67] (Fig 4A). JR-CSF wt and the N332A mutant, which showed no increased neutralization sensitivity, were added for comparison. N332A and wt Env failed to bind 17b in absence of sCD4 trigger while the majority of the sensitive mutants (20/27) bound 17b (Figs 4B, S2A and S2B). Neutralization sensitive mutants (7/27) that did not bind 17b in absence of sCD4 included mutants in residues 420–424 and 435 which are known to impact on 17b binding to monomeric gp120 [65]. In none of these cases 17b binding was however completely abolished since all neutralization sensitive mutants displayed a notable increase in 17b binding upon CD4 triggering. sCD4 dose escalation revealed that 24 out of 27 neutralization-sensitive mutants had a markedly increased propensity to adopt an open conformation, consistent with a decreased stability of the closed pre-fusion conformation in these mutants (Figs 4B and S2A and S2B). For the remaining three mutants (W112A, Y173A, and P470A) a more moderate effect was noted.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Characterizing the exposure of the co-receptor binding site by neutralization sensitive Env mutants.

A. Schematic illustrating the experimental sCD4-induced opening of cell surface expressed Env leading to exposure of the co-receptor binding site. For detection by flow-cytometry mAb 17b was used as a co-receptor mimic. B. Top panel: Opening of cell surface expressed Envs by increasing concentrations of sCD4 was monitored by staining with mAb 17b. Env mutants were divided into two groups showing enhanced 17b binding in absence of sCD4 (right) or not (left) compared to JR-CSF wt Env. The N332A mutant served as additional control. All titrations were performed once. Color code of mutant sensitivity as in Fig 2. Middle panel: magnification of top panel data indicating data points for 17b staining of individual Env mutants. Bottom panel: The bar graph depicts area under the curve values derived from the normalized MFI curves of the top panel graphs. C. Spearman correlation between 17b neutralization sensitivity of JR-CSF wildtype and mutant viruses and 17b binding according to B. Mutations directly affecting 17b binding [65] were not included (see also S2C Fig). D. Alanine-substitutions leading to moderate (blue) or high (red) generalized neutralization sensitivity of the JR-CSF Env were mapped onto the trimeric closed prefusion structure of the closely related JR-FL Env ectodomain (PDB: 5FYK; V1V2: yellow, V3: orange, β20-β21: brown, gp120: light grey, gp41: dark grey). Glycans on neutralization sensitive positions N156, N262 and N301 are depicted. The N197 glycan is not shown as JR-FL naturally lacks this PNGS.

https://doi.org/10.1371/journal.ppat.1012825.g004

To link 17b exposure to neutralization sensitivity we measured the neutralization potency of 17b against 19 Env mutants with high general sensitivity alongside controls (S2C Fig). Excluding mutants known to directly affect the 17b epitope (I420A, I423A, I424A, Y435A) [65,68], the remaining mutants all showed high 17b sensitivity. 17b neutralizing activity and binding were tightly correlated confirming a propensity for trimer opening as a common trait of generalized neutralization sensitivity (Figs 4C and S2C).

Mapping of the 30 JR-CSF Env mutations that confer generalized sensitivity on the crystal structure of the trimeric Env ectodomain from the closely related JR-FL strain (Fig 4D) highlights their positioning in domains relevant for conformational stability of the trimer [15,16,46]. The majority of mutations (26/30) were concentrated in and around V1V2, V3 and β20-β21 strands of the bridging sheet. The remaining four were mapped to C1 (W112A), C2 (N262A) and C5 (P470A and V506A). For three out of the 30 amino acids interactions mediated by the side chains were identified in the structure (D167: interprotomer salt bridges to T128 and R192; K305: intraprotomer salt bridge to E172; Y435: intraprotomer hydrogen-bonds to Y318 and T319), which are likely disrupted by mutation to alanine, providing an explanation for the observed destabilization of the closed conformation (S3A Fig). Glycans in gp120 contribute both to shielding and trimer stability [69]. JR-CSF wt contains 23 potential N-glycosylation sites (PNGS) within gp120. 17 of these PNGS were mutated in the full JR-CSF mutant panel (N = 126; Figs 2 and S1) and four of them (N156A, N197A, N262A and N301A) were associated with a generalized-sensitivity phenotype. As these residues are known to be important for Env stability [60,70,71], the gain in sensitivity upon loss of the glycosylation site may stem from both, a regained access to an epitope covered by this glycan or an increased propensity for trimer opening. Indeed, N197 has been shown to be critical for trimer integrity and neutralization resistance in several HIV-1 strains [34,72–74], as the associated glycan is positioned to protect both V3 and the CD4bs [56,57]. We also observed a dual role for the N301 PNGS in shielding and trimer stability. The decreased Env stability and increased neutralization sensitivity of N301A and T303A mutants was partially rescued in a T303I mutant, suggesting that the amino acid type in position 303 is also important (Figs 2 and 3 and S2D).

HIV-1 relies on the closed Env prefusion conformation to protect itself against antibody mediated neutralization [9,11,15,16]. Analysing group M Env sequences in the Los Alamos National Laboratory database (N = 6084, super-filtered web alignment (2021), including group M and CRFs, http://www.lanl.com/), we observed different degrees of conservation for the 30 positions associated with generalised sensitivity in JR-CSF Env. Notably, 19 of the positions were recorded with >90% preservation (S3B Fig), as would be expected for positions involved in maintaining the closed state.

Focusing on mutations that confer high neutralization sensitivity in the context of JR-CSF, we next sought to investigate their role in a different Env genetic background. We chose for this the subtype A Env BG505 T332N known for its exceptional trimer stability [7,45,56,75] and generated 20 variants carrying mutations inducing high sensitivity in JR-CSF. Indeed, the majority of BG505 mutants (13/20) showed an enhanced propensity to undergo CD4-induced conformational transitions, exposing the V3-crown to a higher degree than BG505 T332N wt following sCD4 triggering (S4 Fig). However, in contrast to JR-CSF, neutralization sensitivity was increased only to CD4-IgG2 by these mutations. Other weak-nAbs and PWH plasma did not become more potent (S5 Fig). Thus, mutations at these positions can entail different degrees of conformational destabilization and neutralization sensitivity depending on the virus strain.

Neutralization fingerprinting using the JR-CSF sensitivity Env mutant panel (SENSE-19)

Considering the functional conservation of the positions associated with generally increased neutralization sensitivity we reasoned that these JR-CSF mutants may allow standardized typing of nAbs with respect to their preference and tolerance of Env structures. To ease incorporation of generally sensitive virus mutants into future nAb screens, we sought to focus on a concise collection of mutants with maximal information gain. Hierarchical clustering of all 126 JR-CSF mutants according to their neutralization sensitivity revealed a cluster of 17 mutants with high generalized sensitivity (S6 Fig). From this group mutant F317A was excluded due to relatively low infectivity. We additionally included mutants T202A, N197K and S199A from a second cluster containing generally sensitive mutants. T202A is a naturally occurring mutation (S3B Fig). N197K and S199A mutations affect a PNGS with critical function in trimer stability and neutralization sensitivity [34,56,57,72–74]. We refer to this collection of 19 JR-CSF mutants as Sensitivity Env Mutant Panel (SENSE-19). As expected for mutations affecting trimer stability, we overserved a reduced infectivity among the SENSE-19 mutants, however the reduction was overall modest (median gp120/p24-normalized infectivity 0.47-fold compared to wt; range: 0.05- to 1.37-fold wt infectivity (S7A–S7C Fig)).

We next used SENSE-19 to establish fingerprints of larger numbers of weak-nAbs and bnAbs building on those included in the full JR-CSF mutant screen (Figs 2 and 3). SENSE-19 fingerprinting of 11 additional Abs (5 CD4bs Abs, 3 V2-glycan Abs, 1 V3-glycan Ab, 1 gp120-gp41 interface Ab and 1 CD4i Ab) showed a pattern consistent with observations made on the full panel. Weak-nAbs displayed the expected strong increase in neutralization potency while bnAbs recorded comparatively small gains in activity or resistance (Fig 5A). The fingerprints also revealed some differences in Env conformational intermediates sampled by the individual SENSE-19 mutants, as they restricted the potency of CD4bs, V2-glycan and V3-glycan bnAbs to different degree.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Neutralization fingerprinting of nAbs on a neutralization sensitive Env mutant panel (SENSE-19).

Pseudovirus neutralization data on TZM-bl cells. A. IC50 ratios (mutant/wt) of an extended panel of nAbs were displayed as a heatmap. Abs are ordered by epitope. All titrations were performed in one technical replicate. B. Spearman correlation matrix and hierarchical clustering of SENSE-19 neutralization fingerprints from Abs, bnDs and PWH plasma (median reciprocal ID50 ratio (mut/wt) over 11 JR-CSF wt neutralizing plasma (Fig 2A)). All significant correlations (p-value ≤0.05) are indicated with a circle with bigger circles indicating stronger support (lower p-values). C. Comparison of three Ab quality parameters: neutralization breadth, potency and SENSE-19 score. Breadth and potency were determined on a cross-subtype 40 virus panel.

https://doi.org/10.1371/journal.ppat.1012825.g005

We further probed the performance of monovalent, broadly neutralizing DARPin (bnD) based inhibitors against SENSE-19. We included the V3-crown specific bnDs (bnD.1, bnD.2, bnD.3 and bnD.4; [51,68]) and bnDs recognizing αV3C, an alpha-helix within the V3 C-terminal strand (bnD.8 and bnD.9; [68]). All six bnDs require Env opening for binding and exhibit dominant post-attachment neutralization activity [51,68]. In agreement, the bnDs yielded notably enhanced potency across SENSE-19 (Figs 5C and S7D). Interestingly, spearman correlation analysis and hierarchical clustering of SENSE-19 neutralization fingerprints distinguished a cluster of conformation-specific inhibitors recognizing CD4-induced epitopes comprising Ab 17b as well as the bnDs tested (Fig 5B). This cluster was separate from a second cluster comprising plasma and Abs with restricted activity against closed prefusion Env and some bnAbs with dominant post-attachment activity. While most CD4bs bnAbs clustered together, other bnAb types were spread across different clusters suggesting that the bnAb type alone does not allow to predict a bnAbs conformational dependence (Fig 5B).

To differentiate nAbs and bnDs independent of potential direct interactions with residues mutated in the panel, we introduced a SENSE-19 score, defined as the median log IC50 ratio (mutant/wt) over the panel. Abs with a SENSE-19 score close to zero showed highest tolerance against changes in Env conformational stability. Particularly low variations in potency were evident for CD4bs bnAbs VRC01, VRC07-523, 1NC9, and N49P7 suggesting that CD4bs bnAbs are particularly suited to achieve high tolerance to structural changes in Env. The interface bnAb PGT151 likewise recorded with low variation in potency across the SENSE-19 panel.

Classifying Abs and bnDs according to the SENSE-19 score together with neutralization potency and breadth, we observed four categories (Fig 5C). Abs with limited potency and breadth and strong dependence on trimer opening had SENSE-19 scores <−1.5. bnAbs and bnDs that benefit from trimer opening recorded a SENSE-19 score ranging from −1.5 to −0.2. bnAbs with high tolerance to structural changes displayed SENSE-19 scores between −0.2 to 0.2 and bnAbs depending on a closed trimer conformation scored >0.2. bnAbs 3BNC117, NIH45-46, PGV04, PGDM1400 and PGT145,recorded with an overall reduced potency on SENSE-19 (score >0.2) but showed a tendency for increased neutralization potency against Tier 1A strains compared to Tier 2 strains (Fig 1B) underlining that Env conformations sampled by SENSE-19 mutants are distinct differ from those of Tier-1A viruses.

While the SENSE-19 panel is already moderate in size, a smaller panel that yields the same information may be useful for larger screens. To probe if the number of mutants in the panel can be further reduced we conducted a random forest analysis and identified suitable panels consisting of 6 and 12 mutants, respectively (S7E Fig). The top scoring combination of 6 mutants and the top scoring combination of 12 mutants determined were then used to calculate corresponding SENSE-6 and SENSE-12 scores. Both panels classified nAbs and bnDs in a very similar manner compared to SENSE-19 (S7F Fig), confirming that use of SENSE-6 in high-throughput screens is possible. In the context of our study, we opted to continue with the larger SENSE-19 panel as this provides improved means to identify phenotypic features of distinct Env mutants and to characterize the respective Env conformational intermediates in their interaction with bnAbs. An additionally advantage of the larger SENSE-19 panel is that mutations directly affecting a bnAb epitope and leading to full resistance will not impact the readout strongly, while loss of 1 mutant out of 6 could be more critical.

Collectively, SENSE fingerprinting provides insight of bnAb tolerance to structural changes and may be used in conjunction with a conventional neutralization breadth screen to define post-attachment acting bnAbs and broadly neutralizing inhibitors.

Probing SENSE-19 mutants in the context of PBMC-based neutralization assays

An obstacle to the evaluation of neutralization activity remains the lack of standardizable, high-throughput assays to study HIV-1 inhibition in an environment similar to in vivo infection of primary CD4 T cells with replicating virus. Both the virus preparation and the variable infectivity of the target cells, peripheral blood mononuclear cells (PBMC) from random donors, are through-put limiting resources and can create a substantial assay-to-assay variability [76]. However, the use of PBMC-based neutralization in the evaluation of bnAbs has been reconsidered in recent years because the virus producer and target cells used may influence the neutralization efficacy recorded, at least for some Ab-virus combinations [76–83]. To verify whether SENSE-19 mutants record similar results in different assay systems, fingerprinting was performed in a PBMC based setup. We probed individual replication competent viruses in a conventional PBMC setup (Fig 6) and in a pooled virus library assay format similar to the approach used for Env mutational scanning (Fig 7) [84]. A requirement for both setups was to transfer SENSE-19 mutations into a full-length, replication-competent clone of JR-CSF (JR-CSF^rc). In the conventional PBMC setup, virus infectivity in the supernatant was monitored. The library approach depends on sequence-based detection of the infecting strains. To identify individual strains by Illumina sequencing, we included 11 mutants from the SENSE-19 panel that fall within a stretch of DNA with a length that can be covered by Illumina sequencing (S10 Fig). As controls JR-CSF wt and four glycosylation site mutants (N295A, N332A, T450A and N461A) were included.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. The generalized neutralization sensitivity phenotype is equivalent in different assay formats.

JR-CSF wt virus and Env mutants were tested for neutralization sensitivity in three assay formats differing in the virus/cell combination: JR-CSF pseudovirus/TZM-bl, JR-CSF^rc/TZM-bl, JR-CSF^rc/PBMC. A. Heatmap comparing IC50 values from the three assay types. Titrations in the pseudovirus assay were performed once, except for the JR-CSF wt reference (n ≥ 4). Titrations in the JR-CSF^rc/TZM-bl assay were set up in triplicates. For the JR-CSF^rc/PBMC assay results are shown from one of two independent experiments set-up in triplicate for each mAb respectively. B. Spearman correlation of shifts in neutralization sensitivity (log IC50 ratio (mutant/wt)) between the JR-CSF^rc/PBMC and JR-CSF pseudovirus/TZM-bl assays (corresponding to data shown in S8A Fig).

https://doi.org/10.1371/journal.ppat.1012825.g006

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Validation of a virus mini-library neutralization assay for mAb fingerprinting with sequencing-based readout.

The virus mini-library comprised 15 JR-CSF^rc mutant viruses including 11 of high generalized neutralization sensitivity from the SENSE-19 panel and 4 PNGS mutants (controls). Wt JR-CSF^rc was also included as reference. A. Schematic explaining the assay set-up and readout by Illumina sequencing. B. Comparison of results from the virus mini-library neutralization assay with a separately conducted PBMC based assay format testing JR-CSF^rc mutants individually against four different antibodies. Results indicating resistance and sensitivity to a nAb are colored red and blue respectively. Left panels for each Ab: Enrichment and depletion of mutant viruses in the mini-library in presence (25 µg/ml) versus absence of Ab are depicted as Z-scores (error bars: STD), with Z-score (Env variant y)_{replicate z} = (arF(Env variant y)_{replicate z with Ab} – mean arF(Env variant y) _{without Ab})/standard deviation arF(Env variant y) _{without Ab}, wherein arF(Env variant y) is the relative read frequency of Env variant y after adjustment according to the proportion of total JR-CSF reads versus total of reads including the internal control recovered (for details see materials and methods). A positive Z-score indicates enrichment and therefore resistance to the Ab, a negative Z-score indicates depletion and therefore sensitivity to the Ab (n = 3). The dotted lines indicate the threshold at which the enrichment/depletion in presence versus absence of antibody exceeds two times the standard deviation of read frequencies in absence of antibody (Z-scores = 2 and −2). Right panels for each Ab: Resistance of individually tested mutants in the PBMC-based assay was attributed if the IC50 was above 25 µg/ml (dotted line).

https://doi.org/10.1371/journal.ppat.1012825.g007

In a first step, JR-CSF^rc mutants were generated and probed in two conventional neutralization setups, on PBMC and on TZM-bl cells. The results were compared to the setup with JR-CSF mutants as pseudoviruses on TZM-bl cells (Fig 6). Overall, we observed a similar pattern for the three assay systems with much stronger variations in potency across the mutant panel for weak-nAbs compared to bnAbs (Figs 6, S8 and S9). Only modest differences between the PBMC based assay and the two TZM-bl assays were apparent. Notably, while IC50s for most Abs correlated between the three assays (S9 Fig), Abs tended to be less potent in the JR-CSF^rc/TZM-bl setup (Figs 6A and S8B). Moreover, IC50 ratios (mut/wt) correlated for most Abs between the JR-CSF/TZM-bl and JR-CSF^rc/PBMC formats (Fig 6B). Correlations tended to be less strong for bnAbs (in particular PGT128) which likely is a result of the bnAbs low variability in potency paired with the inherent variability of PBMC-based assays.

To probe the JR-CSF mutants in the library approach (Fig 7) individually propagated virus stocks were pooled into a library and used to infect stimulated PBMC in presence and absence of nAbs. Unlike in conventional neutralization assays, where nAb activity is probed over a wide concentration range, a single, high-dosed concentration of mAb was used in this library neutralization assay (25 µg/ml). Twelve hours post infection, DNA was isolated from the PBMC cultures and the Env region chosen for monitoring was PCR amplified from reverse transcribed non-integrated viral DNA using barcoded primers. A DNA fragment derived from a plant-gene flanked by the same primer binding sites used for Env-enrichment was added to each sample as an internal control. The amplified DNA from antibody treated and non-treated samples together with the combined internal control was sequenced by Illumina Mi-seq, and individual reads were attributed to the internal control, different mutants or the wt Env using the previously established pipeline VirVarSeq [85]. The frequency of observed reads of each individual Env mutant were used to calculate Z-scores. A positive Z-score indicates relative enrichment of the Env in presence versus absence of a nAb and thus resistance, a negative Z-score indicates sensitivity (Fig 7). We probed the feasibility of this assay format with three weak-nAbs (b6, 1–79 and 17b) and bnAb VRC01. The library setup yielded results in agreement with the TZM-bl pseudovirus assays (Fig 6A) and separate inhibition assays conducted with individual replication competent Env variants on PBMCs (Fig 7B). Resistant mutants were enriched in the library in the presence of Abs 1–79, b6 and 17b while most SENSE-19 mutants were depleted or remained at similar level as when no Ab was present (Fig 7B). In contrast, all mutants were sensitive to VRC01 and by consequence none of them was enriched. We therefore conclude that the translation of the SENSE-19 screen to a library-based screen on PBMC is feasible and has potential for screens where a qualitative readout is useful. This approach may be particularly useful, when a large number of mutant and naturally evolved Envs are combined into a screening library. Overall, the comparison of PBMC and TZM-bl assay systems highlighted that screening with SENSE-19 mutants can be integrated in the assay format of choice and provides information on the tolerance of inhibitors to trimer opening.

Ethics statement

All analyzed plasma samples (S2 Table) were derived from specimens stored in the biobanks of the Swiss HIV Cohort study (SHCS) and the Zurich Primary HIV Infection Study (ZPHI). The Swiss HIV Cohort Study (SHCS) is a prospective, nationwide, longitudinal, non-interventional, observational, clinic-based cohort with semi-annual visits and blood collections, enrolling all HIV-infected adults living in Switzerland [105]. The SHCS is registered under the Swiss National Science longitudinal platform: http://www.snf.ch/en/funding/programmes/longitudinal-studies/Pages/default.aspx#Currently%20supported%20longitudinal%20studies. Detailed information on the study is openly available on http://www.shcs.ch.

The Zurich Primary HIV Infection study (ZPHI) is an ongoing, observational, non-randomized, single center cohort founded in 2002 that specifically enrolls patients with documented acute or recent primary HIV-1 infection (www.clinicaltrials.gov; ID NCT00537966) [106].

The SHCS and the ZPHI have been approved by the ethics committee of the participating institutions (Kantonale Ethikkommission Bern, Ethikkommission des Kantons St. Gallen, Comité départemental d’éthique des spécialités médicales et de médicine communautaire et de premier recours, Kantonale Ethikkommission Zürich, Repubblica e Cantone Ticino - Comitato Ethico Cantonale, Commission cantonale d’éthique de la recherche sur l’être humain, Ethikkommission beider Basel for the SHCS and Kantonale Ethikkommission Zürich for the ZPHI) and written informed consent had been obtained from all participants.

Antibodies and reagents

Reagents were kindly provided by following groups: Soluble CD4 (sCD4) comprising all four immunoglobulin domains [92,93] and CD4IgG2 [94] by W. Olson (Progenics Pharmaceuticals Inc., Tarrytown, New York, USA); monoclonal antibodies (mAbs) VRC01, PGV04 [39] by J. Mascola (NIAID, NIH, Bethesda, USA); mAbs b12 [95], b6 [96], PG9 and PG16 [54], PGT121, PGT128, PGT130, PGT135 and PGT145 [38] and PGDM1400 [97] by D. Burton (The Scripps Research Institute, La Jolla, USA); mAbs 2G12 [98], 2F5 [99], 4E10 [100,101] and 1F7 [102] by H. Katinger (Polymun, Vienna, Austria); mAbs 1–79, 3BNC117 and NIH45-46 [103] by M. Nussenzweig (Rockefeller University, New York, NY, USA); 447-52D [104] by S. Zolla-Pazner (Icahn School of Medicine at Mount Sinai, New York, NY, USA); 10E8 [37], 48D and 17b via the NIH AIDS Research and Reference Reagent Program, Division of AIDS, NIAID, NIH. Soluble CD4 containing the two amino-terminal immunoglobulin domains (sCD4-183) was expressed in E. coli and purified as described [25]. An overview is provided in S3 Table.

Cells

293-T cells were obtained from the American Type Culture Collection. The TZM-bl cell line was obtained from NIH ARP. Both cell lines were cultivated in DMEM medium supplemented with 10% fetal calf serum (FCS) and 100 µg/ml streptomycin and 100 U/ml penicillin (P/S) (referred to as cell culture medium hereafter). All cell culture reagents were obtained from Gibco (Thermo Scientific, Waltham, USA).

PBMC were purified from buffy coats from anonymous blood donations from healthy individuals obtained by the Zurich Blood Transfusion Service (http://www.zhbsd.ch/) under a protocol approved by the local ethics committee. PBMC from three individual donors were CD8+ T-cell depleted using Rosette Sep cocktail (StemCell Technologies Inc.), pooled, split into three parts and incubated for three days with either 5 µg/ml PHA, 0.5 µg/ml PHA or anti-CD3 mAb OKT3 in RPMI 1640 medium (10% FCS, 10 U/ml IL-2, glutamine and 1% penicillin-streptomycin) [107].

Viruses

The plasmid collection of JR-CSF env alanine mutants [38,53,54] and was a kind gift from D. Burton (The Scripps Research Institute, La Jolla, USA). The BG505.W6M.ENV.C2 [108] encoding source plasmid was obtained from the NIH AIDS Research and Reference Reagent Program. BG505 Env mutants and additional JR-CSF Env mutants were generated using the QuikChange II XL Site-Directed Mutagenesis Kit (Agilent, Santa Clara, USA) according to the manufacturer’s instructions. A BG505 Env mutant with V1V2 deletion was generated as described previously [11]. Pseudotyped virus production was performed by co-transfection of 293T cells with Env expression plasmid and HIV vector pNLluc-AM carrying a luciferase reporter cassette (provided by A. Marozsan and J. P. Moore) as described previously [81]. Information on the composition of a 40 virus panel used to assess mAb breadth and potency can be found in S4 Table.

To generate replication competent mutant viruses, the JR-CSF Env encoding fragments containing the desired alanine substitutions were ligated in-frame into a replication competent JR-CSF infectious molecular clone (pYK-JRCSF obtained through the NIH ARP, catalog no. 2708 [109]) using InFusion technology (Takara Bio Inc., USA) according to instructions by the manufacturer. Replication competent virus was produced by transfection of 293T cells with pYK-JRCSF based vectors analogous to the production of pseudotyped virus [81].

Neutralization assay using Env-pseudotyped virus

The neutralization sensitivity of Env-pseudotyped viruses against mAbs and plasma from people living with HIV-1 (PWH) was evaluated on TZM-bl reporter cells in 384 well plates. Virus input was diluted with cell culture medium aiming for a luciferase reporter readout of 500,000–1,000,000 relative light units (RLU) per well measured on an EnVision luminometer (Perkin Elmer, Waltham, USA) in the absence of inhibitors. The antibody concentration (IC50) or reciprocal plasma titer (ID50) causing 50% reduction in viral infectivity were calculated by fitting the data to sigmoid dose–response curves (variable slope) using Prism (GraphPad Software). If 50% inhibition was not achieved at the highest or lowest concentration, a greater than or less than value was recorded.

Two different assay setups were established. To verify their comparability, PG128, PGT145, 10E8 and 2G12 were tested in both setups over the whole JR-CSF panel (N = 126) and no systemic discrepancies were identified.

In the first set up, serial dilutions of inhibitors and PWH plasma in cell culture medium supplemented with 50mM HEPES were prepared robotically in 384 well plates (60 µl per well) (Corning, Corning, USA). This was followed by adding 20 µl per well of pseudovirus suspension. After 1–2 hours incubation at 37 °C, 20 µl TZM-bl cell suspension at 150’000 cells per ml density in cell culture medium supplemented with 60 µg/ml DEAE-Dextran were pipetted robotically to each well and the plates were incubated for 48 hours at 37 °C before proceeding to the readout of luciferase signal on an EnVision luminometer (Perkin Elmer, Waltham, USA) as described [81].

In an alternative assay format serial dilutions of inhibitors in cell culture medium (40 µl per well) were added to 40 µl of virus suspension in 384 well plates. After 1–2 hours incubation at 37 °C 50 µl of the virus-inhibitor mixture were added to pre-seeded TZM-bl cell cells (384 well plates, 6000 cells/well in 30µl culture medium supplemented with 26.6 µg/ml DEAE-Dextran) and the plates were incubated for 48 hours at 37 °C before readout of the luciferase signal on an EnVision luminometer (Perkin Elmer, Waltham, USA) as described [81].

Neutralization data of bnD.1, bnD.2, bnD.3 and mAb 447-52D on the full JR-CSF mutant panel was reproduced from Friedrich et al [51]. Neutralization data of bnD.4, bnD.8, bnD.9 as well as mAbs PGT128 and 17b on strongly neutralization sensitive Env mutants was reproduced from Glögl et al [68].

Neutralization assay using replication competent HIV

In addition to the pseudovirus neutralization assay performed on TZM-bl cells, replication competent virus was used to infect TZM-bl cells or PBMC in presence and absence of an inhibitor. To determine relative infectivity the individual replication competent virus stocks were first titrated on TZM-bl cells (seeded the previous day in a 384 well plate at 6000 cells/well density in 30 µl culture medium supplemented with 26.6 µg/ml DEAE-Dextran). The plate was incubated for 48 hours at 37 °C before readout of the luciferase signal analogous to the procedure when using pseudovirus detailed above. As replication competent viruses are cytolytic for TZM-bl cells at high concentrations, this allowed to determine the dilution of the viral stock at which the TZM-bl cells survive for at least 3 days while ensuring high signal over background during readout. This virus dilution was not only used for inhibition assays of replication competent virus on TZM-bl cells, but also as input for the PBMC-based neutralization assay in 384 well plates.

For inhibition assays on TZM-bl cells, replication competent virus was titrated with inhibitor in triplicates, pre-incubated for 1 h at 37 °C and 30 µl/well of virus-inhibitor mix were then transferred onto TZM-bl cells seeded the previous day (384 well plate, 6000 cells/well in 30 µl culture medium supplemented with 26.6 µg/ml DEAE-Dextran). The plate was incubated for 48 hours at 37 °C before readout of the luciferase signal analogous to the procedure when using pseudovirus detailed above.

For inhibition assays on PBMC, replication competent virus was titrated with inhibitor in triplicate, pre-incubated for 1h at 37 °C and 30 µl/well of virus-inhibitor mix were then transferred onto PBMC (seeded at 1.5 × 10⁶ cells/ml density in 30 µl/well RPMI 1640 with glutamine supplemented with 10% FCS, 100 U/ml IL-2, and 1% penicillin-streptomycin). After incubation for 7 days at 37 °C, plates were spun down at 450 g for 2 min and 30 µl/well of supernatant were transferred onto TZM-bl cells (seeded the previous day in a 384 well plate at 6000 cells/well density in 30 µl culture medium supplemented with 26.6 µg/ml DEAE-Dextran). After a further 24–72 hours of incubation infection was quantified by luciferase reporter readout as described for the pseudovirus assay to assess virus inhibition by antibody relative to wells without antibody. Additional wells which received the double amount of virus input in the PBMC-step were used as control to ensure that cytolysis was not affecting the assay.

Sequencing based evaluation of neutralization assay with JR-CSF point mutant mini-library

To prepare the mini-library of replication competent JR-CSF Env point mutant viruses, individually prepared stocks of replication competent mutant viruses were mixed at equal volume. For the neutralization assay with this virus library PBMC were seeded at 3 × 10⁶ cells/ml in 250 µl RPMI 1640 with glutamine supplemented with 10% FCS, 100 U/ml IL-2, and 1% penicillin-streptomycin on 24-well culture plates. 125 µl of the virus library were preincubated either with or without 125 µl inhibitory agent diluted in cell culture medium aiming for a final concentration of 25 µg/ml after transfer of the mix onto PBMC. Each condition was set up in triplicates. The mixture of virus library and inhibitor was pre-incubated for 1h at 37 °C and then transferred onto the seeded PBMC. 12 hours post-infection the cells were spun down at 16’000 g for 10 min and washed with PBS. The reverse-transcribed non-integrated viral cDNA was then extracted from PBMC using QIAprep Spin Miniprep Kit (Qiagen, Aarhus, Denmark). Samples were handled with frequent glove and pipette tip changing while any solution used was aliquoted to prevent cross contamination [110,111]. As an internal control, 1 µl at 1 pg/ml of a custom synthesized 544 base-pair sized dsDNA fragment (Twist Bioscience, South San Francisco, California, USA) containing 500 base-pairs of a plant-gene (EMB2768 – Farinopsis Salesoviana) flanked on each side by one of the two primer binding sites used for amplification of Env-fragments (described below) was added to each sample. In a first PCR (see S5 Table) amplification of Env fragments covering all alanine substitutions in the virus mini-library (and the internal DNA control fragment) was done with Env specific forward and reverse primers of three different lengths containing adapter sequences at the 5’ end.

Forward primers: CTTTCCCTACACGACGCTCTTCCGATCT(N)_6,7,8AACCATAATAGTACAGCTGAATGAATC.

Reverse primers: GGAGTTCAGACGTGTGCTCTTCCGATCT(N)_6,7,8TCTGAAGATCTCGATCTCACTCT.

The differing length of the primers artificially increases diversity of the otherwise highly similar amplicons for efficient MiSeq sequencing. In addition, 5 random nucleotides between the adapter-region and binding region of the primers were added to facilitate cluster generation during illumina amplification of reads. The forward and reverse primers of different lengths were used at 3.33 µM concentration each, to result in 10 µM forward and reverse primer mixtures.

The DNA was then purified from the PCR reactions using AMPure XP beads (Backman Coulter Life Sciences, Indianapolis, USA) at a 1:1 ratio (vol/vol) as recommended in the instructions by the manufacturer. After the clean-up, a 2nd PCR was performed (see S5 Table) to barcode the individual samples using Index primers (TruSeq HT Kits, Illumina Inc., San Diego, California, USA) able to bind the adapter-region established during the previous PCR.

After PCR dsDNA concentration in samples was measured with the QuantiFluor system (Promega). The samples were subsequently diluted to 4 nM concentration and twenty-four samples were mixed together in equal parts. The Ilumina sequencing used a MiSeq Reagent Kits v2 (500 cycles) and was executed based on manufacturer’s instructions.

For evaluation individual reads were attributed to the internal control, different Env mutants or the wt Env using VirVarSeq [85]. Then, the relative frequencies of each Env variant (rF(variant y)_{rep z}) computed by VirVarSeq (controlled for position-specific read quality) in samples treated or not with antibody were transformed into adjusted relative frequencies (arF (variant y)_{rep z}) taking into account the number of reads corresponding to the internal control.

rep z: replicate 1, 2 or 3.

n (JRCSF reads): total number of reads aligned to JR-CSF reference by VirVarSeq.

n (EMB2768 reads): total number of reads aligned to plant reference by VirVarSeq.

Based on the adjusted relative frequencies of reads, Z scores were calculated for individual Env variants of each experimental replicate as follows:

Wherein the mean adjusted relative frequency of an Env variant in absence of antibody is:

The internal control allows the comparison of read frequencies from Env variants in presence and absence of Ab and is particularly important in case the majority of viruses in the mini-library is sensitive to the Ab.

Flow cytometric analysis of antibody binding to cell surface-expressed HIV-1 Env

10⁵ 293T cells per well were seeded in 1 ml cell culture medium in 12-well tissue culture plates and incubated at 37 °C. 24 hours later, cells in each well were transfected with a total of 1 μg DNA (Env-expression plasmid and pCMV-rev expression helper plasmid in 4:1 ratio) mixed with 3 μg 25 kDa linear PEI (Polysciences Inc., Warrington, USA) in 200 μl 150 mM NaCl. After settling the DNA-PEI complexes by a short spin (3 min, RT, 300 g) the cells were incubated 36 hours at 37 °C. All subsequent steps were carried out at room temperature. Cells were harvested, wells transfected with the same Env-plasmid pooled, and distributed into 96-well round bottom tissue culture plates as individual staining reactions (cells from one 12-well were split into four staining reactions). For each staining reaction cells were washed once with 200 μl FACS buffer (DPBS (Gibco) with 2% FBS, and 2 mM EDTA) and stained for 20 min in 15 μl of FACS buffer with 10 μg/ml of primary antibody in the presence or absence of sCD4-183. After washing twice with 200 μl FACS buffer the cells were incubated in 30 μl of FACS buffer for 20 min with 1:1000 diluted APC-conjugated F(ab’)₂ fragment goat anti-human IgG (Jackson ImmunoResearch, West Grove, USA). Following two further washes with FACS buffer the cells were resuspended in 100 μl FACS buffer with 0.1 µg/ml propidium iodide (BD Pharmingen). Flow cytometry analysis was performed on a FACSVerse system (BD biosciences, San Jose, USA). Arithmetic mean of APC fluorescence intensity was calculated for Propidium Iodide negative single cell population as a measure of primary antibody binding to the live cells in each staining reaction.

Env mutants were devided into two groups showing enhanced 17b binding in absence of sCD4 or not compared to JR-CSF wt Env (threshold 10% of normalized maximal staining in presence of sCD4). To determine the area under the normalized MFI curves on a log-x-scale, all sCD4 concentrations (in uM), at which binding was measured, were enhanced by “0.01”, such that MFI measurements in the absence of sCD4 could be included.

Entry capacity

The entry capacity of Env variants was assessed by measuring the infectivity of Env pseudoviruses on TZM-bl cells in the presence of 10 µg/ml diethylaminoethyl-dextran (DEAE), normalized to the gp120 and p24 content of the pseudovirus stocks. Pseudovirus stocks using all SENSE-19 JR-CSF Env mutants and JR-CSF wt Env were generated as described above on the same day using the same stock of HIV-1 backbone plasmid NL-lucAM, polyethylenimine (PEI) max transfection reagent and 293 T cells. Cell culture supernatants containing pseudovirus were harvested three days later and aliquots were frozen at −80 °C. Infectivity, gp120 and p24 content were measured in parallel for all Env variants and measurements were performed in biological duplicates. For the infectivity measurements, pseudovirus serial dilutions were added to TZM-bl cells and luciferase activity (relative light units (RLU)) was measured three days post-infection using Luciferase assay reagent (Promega, USA). In parallel, aliquots of pseudovirus stocks were inactivated with 1% Empigen final concentration and serially diluted in TBS/1% Empigen. Gp120 (in-house produced affinity-purified JR-FLgp120 protein) and p24 (AG 6054, Aalto Bioreagents, Ireland) standards were prepared accordingly. Gp120 and p24 content of virus lysates was measured by ELISA as previously described [112]. Briefly, gp120 was captured by sheep anti-gp120 antibody (D7324, Aalto Bioreagents, Ireland) coated the day before to the surface of high-binding 384-well ELISA plates (corning). Biotinylated human anti-gp120 2G12 mAb was used as primary detection antibody, alkaline-phosphatase-conjugated streptavidin as secondary detection reagent. For the p24 ELISA setup, p24 was captured on sheep anti-p24 antibody (D7320, Aalto Bioreagents, Ireland) coated 384-well plates and detected using alkaline-phosphatase-coupled antibody BC1071-AP (Aalto Bioreagents, Ireland). Gp120 and p24 content were then interpolated from standard curves of recombinant gp120 and p24 respectively or determined relative to the gp120 and p24 content of JR-CSF wt Env pseudovirus. The obtained infectivity (RLU/µL) values for each stock were normalized to the gp120/p24 concentrations of the respective stock.

Bioinformatics and software

To evaluate whether SENSE-19 could be reduced to a smaller number of mutants, we assessed the variable importance (mean decrease accuracy) of each mutant in a random forest predicting the conformational preference of each antibody (open, close, none as in Fig 5C), as determined by SENSE-19. We then selected the top 6 and top 12 mutants based on visual inflection points in the variable importance curve (S7E Fig). We computed SENSE-6 and SENSE-12 scores on the 6-mutant and 12-mutant panels respectively, similarly as the SENSE-19 score (i.e., log value of the IC50 medians). Using the same thresholds as for SENSE-19 (−0.2 and 0.2), we then verified whether all antibodies were classified in the same categories as predicted by SENSE-19. Analyses were conducted in R version 4.3.1 with the package randomForest.

To analyze Ilumina sequencing data from the JR-CSF library, VirVarSeq [85] was used to calculate the frequency of wildtype and mutant Envs as well as the internal control in presence or absence of an inhibitor. Sequences were trimmed using SeqTK [113] to remove residual, random nucleotides introduced to support Miseq amplification of reads.

Figs 1, 2, 3C, 4C, 5, 6, 7B, and S1, S2C, S3B, S5, S7, S8 were produced using the R packages ggplot2 [114], heatmaply [115], tidyverse [116] and Hmisc [117]. Figs 3A, 4B and S2D and S4B were produced in GraphPad Prism 10 (GraphPad Software, Boston, Massachusetts, USA). FACS-generated data was processed using FlowJo^tm Software v10.9 (BD Life Sciences). Fig 4D was designed using PyMOL v2.5 (https://pymol.org/2/). Biorender (Subscription model, Canada) was utilized to make Figs 4A and 7A. Affinity Designer (SerifEurope, UK) was used to finalize all figures.