Antibody Isotype Switching as a Mechanism to Counter HIV Neutralization Escape.

Neutralizing antibodies (nAbs) to highly variable viral pathogens show remarkable diversification during infection, resulting in an "arms race" between virus and host. Studies of nAb lineages have shown how somatic hypermutation (SHM) in immunoglobulin (Ig)-variable regions enables maturing antibodies to neutralize emerging viral escape variants. However, the Ig-constant region (which determines isotype) can also influence epitope recognition. Here, we use longitudinal deep sequencing of an HIV-directed nAb lineage, CAP88-CH06, and identify several co-circulating isotypes (IgG3, IgG1, IgA1, IgG2, and IgA2), some of which share identical variable regions. First, we show that IgG3 and IgA1 isotypes are better able to neutralize longitudinal autologous viruses and epitope mutants than can IgG1. Second, detrimental class-switch recombination (CSR) events that resulted in reduced neutralization can be rescued by further CSR, which we term "switch redemption." Thus, CSR represents an additional immunological mechanism to counter viral escape from HIV-specific antibody responses.

INTRODUCTION

Antibodies are key components of the adaptive immune response that act directly through Fab-mediated binding and neutralization. The role of somatic hypermutation (SHM) in the maturation of HIV-directed neutralizing antibody (nAb) lineages has been extremely well characterized, with the accumulation of mutations in variable regions conferring increased neutralization capacity (Bhiman et al., 2015; Doria-Rose et al., 2014; Liao et al., 2013; Soto et al., 2016; Wu et al., 2015). However, antibody diversification during infection is also achieved through class-switch recombination (CSR), which allows the expression of antibodies as one of five isotypes that differentially engage innate effector cells through binding to Fc receptors (Schroeder and Cavacini, 2010). Furthermore, isotypes can directly affect neutralization potency of engineered monoclonal antibodies (Kozel et al., 2007; Maurer et al., 2018; Richardson et al., 2019; Tudor et al., 2012), and the Fc region may contribute to viral escape (Horwitz et al., 2017). Recently, the identification of both immunoglobulin (Ig) G and IgA members of HIV-directed antibody lineages has further highlighted intra-lineage isotype variation (Jia et al., 2020). A key question, however, is whether CSR contributes to the maturation of natural antibody lineages during the course of viral infection.

RESULTS

We previously described an HIV-1 subtype-C-infected donor, CAP88, who developed potent neutralizing antibodies against the C3 region of the HIV envelope (Env), which drove rapid viral escape (Moore et al., 2009). The isolation of a C3-directed IgA1 monoclonal antibody (mAb), CAP88-CH06 (Gray et al., 2011), prompted us to explore the role of CSR in antibody-virus co-evolution in that donor. Longitudinal next-generation sequencing (NGS) of peripheral blood mononuclear cells (PBMCs) at multiple times identified clonal relatives of CAP88-CH06 (Figure 1A; Table S1), including multiple co-circulating isotypes comprising IgG3, IgG1, IgA1, IgG2, and IgA2 (Figure 1B). Average SHM increased until 46 weeks post infection (wpi), consistent with plasma neutralization titers, whereas, at 108 wpi, we detected only a single less mutated sequence, identical to one from 34 wpi and, thus, presumably persisting from an earlier time point (Figure 1A). SHM was greater in the heavy chain compared with the light chain (Figure 1A), suggesting that the heavy chain was the major contributor to neutralization, which we confirmed experimentally (Figure S2).

Figure 1.

The CAP88-CH06 Antibody Lineage Comprises Multiple Co-evolving Isotypes

(A) Longitudinal antibody NGS spanning the appearance and waning of C3-directed neutralizing antibodies in donor CAP88. Each box indicates the time point in weeks post-infection (wpi) when antibody NGS was performed and is gray-scaled based on the C3-specific plasma-neutralizing antibody titer (Moore et al., 2009). The time point at which the CAP88-CH06 monoclonal antibody was previously isolated (Gray et al., 2011) is indicated. The average levels of heavy and light chain SHM in clonally related sequences, compared with their germline genes (IGHV4-39 and IGLV3-21), are shown in the middle and bottom panels, respectively.

(B) Prevalence of CAP88-CH06 isotypes over 2 years of infection. Pie charts indicate the numbers of different isotypes observed at various time points, colored according to the isotype (gray denotes cases in which the IgG subtype could not be assigned because of similarities in the primer region).

(C) Levels of SHM over time, by isotype, within the heavy-chain variable regions for each clonally related sequence.

(D) Relative location of isotype-specific IGHC genes on chromosome 14, defining the direction in which CSR occurs. The numbers of sequences observed for each isotype is shown, nd, not done.

(E) Antibody-lineage reconstruction with sequences colored by isotype. Sequences inferred by Change-O, part of the immcantation portal, are indicated as translucent symbols, whereas filled symbols indicate nodes observed in NGS sequences. The number of mutations between nodes is shown as integers, and line lengths between nodes are proportional to those. Nodes indicating the LCA, the mature CAP88-CH06 sequence, and several heavy-chain sequences used for downstream analyses (H1–H9) are enlarged and labeled on the tree. H5-IgA1 was observed both at 34 and 108 wpi.

See also Figures S1 and S2 and Table S1.

Analysis of heavy chain sequences revealed temporal variation of isotypes, with IgG1 predominant at 11 wpi, IgG3 at 17–38 wpi, and IgA1 at 46 and 108 wpi (Figure 1B). IgG2 and IgA2 were rarely observed, although both were detected at 11 wpi, and IgA2 was also present at 34 wpi. All isotypes showed increasing SHM reaching levels between 8% and 11% by 46 wpi (Figure 1C). A single sequence with no SHM in the V and J genes, but mutations in the D region through “N” additions/SHM (Figures S1A and S1B), was detected in the NGS data as either an IgG1 or IgG3 at 8 wpi and represents the last common ancestor (LCA) of this lineage. The relative position of the IGHC genes, responsible for encoding isotypes on chromosome 14, and the irreversible nature of gene excision during CSR enabled the determination of directionality for isotype switches within the lineage (Figure 1D). The existence of mature IgG3 lineage members (Figure 1E) indicated that the LCA must have occurred as an IgG3. A large proportion of the lineage continued to mature as IgG3; however, we also observed identical antibody-variable regions expressed as different isotypes (Figures 1E and S1C). One sequence (H1) must also have existed as IgG3 because it gave rise to IgG3 ancestors but was only observed as IgG1, IgG2, IgA1, and IgA2 in the NGS data (Figure 1E). The more mature H7 and H8 sequences were observed as both IgG1 and IgG3. The topology of the tree suggests that CSR events were largely sequential from IgM to IgG3, then IgG1 and, in some minor branches, to IgA1 (with one IgA1 branch ultimately resulting in CAP88-CH06). In addition, the accumulation of SHM between class switches varied, with IgG3 to IgG1 generally associated with few mutations (<5). In contrast, switches from IgG1 to IgA1 were associated with high levels of SHM (12–16), likely occurring within germinal centers (GCs) (Figure 1E). Overall, these data are consistent with previous reports of antibodies with identical variable regions but different isotypes in response to infections by other pathogens (Janda et al., 2016) and autoimmunity and, in healthy individuals, indicative of B cell clones having undergone multiple class-switch events (Petrova et al., 2018).

To assess the role of isotype in antibody-virus co-evolution, we selected nine heavy-chain sequences with varying levels of SHM (Figure 1E, H1–H9) and expressed these with the CAP88-CH06 light chain (Figures 2A and S1A). Identical antibodies that occurred as multiple isotypes were cloned and expressed in these forms. In addition, the LCA and CAP88-CH06 (which was isolated as IgA1) were expressed as IgG1, IgG3, and IgA1 for cross-isotype comparison. Antibodies were tested in an Env-pseudotyped neutralization assay against single-genome-derived, autologous Envs from viruses circulating during the first year of infection (Figures 2A and S3A). The LCA potently neutralized the transmitted/founder (T/F) virus from 5 wpi, similar to our previous observation of neutralization by an HIV-directed LCA (Doria-Rose et al., 2014). Furthermore, as expected, higher levels of SHM were associated with improved neutralization potency against the T/F virus, and an increased ability to neutralize viruses from later time points. However, no mAbs neutralized viruses from 57 wpi, consistent with the emergence of escape mutations (Figure S3B). These data confirmed that SHM contributed to antibody maturation, until viral escape precluded further neutralization.

Figure 2.

Potent Neutralization of Autologous Viruses by IgG3 and IgA1 Isotypes of CAP88-CH06

(A) Neutralization of the CAP88 T/F virus (5 wpi) and longitudinal viral Env clones isolated up to 57 wpi (when complete viral escape occurred) by the LCA, H1–H9, and the mature CAP88-CH06 mAbs, colored by isotype. Data are shown as a heatmap of neutralization titers (half-maximal inhibitor concentration [IC50]); no neutralization at 10 μg/mL (gray), neutralization between 1 and 10 μg/mL (yellow), 0.1 and 1 μg/mL (orange), and neutralization less than 0.1 μg/mL (red).

(B) Neutralization curves against autologous viruses for the LCA (left panel) and for CAP88-CH06 (right panel) expressed as different isotypes (IgG3, green; IgG1, blue; and IgA1, purple). The dotted line in the graphs represents IC50. Error bars represent standard deviations.

(C) Comparison of neutralization potency for all antibodies with identical variable regions but different isotypes against autologous viruses from 5 to 57 wpi. The mAbs are colored by isotype in order of CSR (IgG3, green; IgG1, blue; IgA1, purple; and IgA2, pink). Geometric mean of potency for each antibody is shown by a solid line. Statistically significant differences among isotypes were assessed with a Wilcoxon signed-rank test, denoted by p values, and had large effect sizes of 0.63, 0.63, 0.67, and 0.56 with Z scores of 2.6, 2.6, 2.5, and 2.9, respectively.

See also Figure S3.

Interestingly, we also observed differences in neutralization and binding of autologous viruses, based on the isotype of the antibody (Figures 2 and S4A). The LCA was more potent as an IgG3, whereas the mature CAP88-CH06 was more potent as an IgA1 (Figure 2B). In all instances, IgG3 and IgA1 showed better neutralization of autologous viruses, both in terms of potency and the number of viruses neutralized, compared with IgG1, with these differences reaching statistical significance for CAP88-CH06, H7, and H8 (Figure 2C). Similarly, CAP88-CH06 expressed as IgA1 and IgG3 showed better binding (KD = 2.01 and 1.54, respectively) than IgG1 (KD = 3.69) when tested against a gp120 that was derived from a 30-wpi virus with significantly different neutralization sensitivity (Figure S4A). This difference was not attributable solely to IgG3 antibody aggregation (Figure S4B). Because isotype differentially modulates effector function (Schroeder and Cavacini, 2010), we also showed that the LCA and CAP88-CH06 expressed as IgG3 mediated the increased antibody-dependent cellular cytotoxicity (ADCC) against the CAP88 T/F virus compared with IgG1 or IgA1 versions, consistent with other studies (Damelang et al., 2019) (Figure S4C). Overall, these data indicated that antibody isotype, in addition to affecting the Fc effector function, can improve the ability of antibodies to bind and neutralize emerging viral variants.

Mutations able to mediate viral escape from CAP88-CH06 (Gray et al., 2011) were quantified by deep sequencing of the CAP88 plasma virus during the first year of infection (Table S2). Significant toggling occurred at residues 339, 343, 350, and 358, located within or proximal to the solvent-exposed alpha-2 helix of C3. The most common and sustained escape mutations were I339N (introducing a glycan sequon), E343K, E350K, and a single amino acid (aa) deletion (K358del). The E343K and E350K mutations peaked between 13 and 15 wpi and, thereafter, occurred with I339N, which became fixed in the population (Figure 3A). The combination of I339N and either E350K or K358del, with or without other escape mutations, accounted for more than 90% of the viruses from 38 wpi. Introduction of escape mutations into the CAP88 T/F virus was performed to assess their effect on neutralization by antibodies of different isotypes (Figure 3B). Similar to longitudinal viruses, increased SHM conferred improved neutralization of viruses bearing escape mutations, with early mAbs neutralizing fewer mutants than more-mature antibodies (Figure 3B; Table S3). Viral mutations were associated with varying levels of neutralization resistance, with E343K conferring resistance to early mAbs (<11wpi), and I339N and I339N+E343K introducing resistance to mAbs isolated up to 34 wpi. However, even the most mature mAbs failed to neutralize viruses containing the K358del and I339N+E350K mutations, accounting for the lack of antibody maturation after 46 wpi (Figure 1A).

Figure 3.

CAP88-CH06 as IgG1 Shows Reduced Neutralization of Viral Escape Variants Compared with IgG3 and IgA1

(A) Longitudinal kinetics of dominant CAP88 viral escape mutations, individually and in combination, from 5 to 57 wpi. Shown are the relative frequencies (percentage of autologous viruses) of each escape mutation over time (wpi) based on deep sequencing of the C3 region of the viral envelope (Table S2). The dotted line denotes the T/F virus. The combination of I339N and E350K or K358del, with or without other mutation(s), is denoted by I339N+E350K/K358del*.

(B) Neutralization titers against the T/F virus mutated to contain key escape mutations. Data are shown as a heatmap of neutralization titers (IC50); no neutralization at 10 μg/mL (gray), neutralization between 1 and 10 μg/mL (yellow), 0.1–1 μg/mL (orange), and less than 0.1 μg/mL (red). Viral escape mutations are ordered by degree of neutralization resistance.

(C) Neutralization curves of CAP88-CH06 expressed as different isotypes (IgG3, green; IgG1, blue; and IgA1, purple) against viruses bearing the E343K, I339N, and I339N+E343K mutations. The dotted line in the graphs represents IC50. Error bars represent standard deviations.

(D) Comparison of neutralization potency for CAP88-CH06, H7, and H8 that showed statistically significant differences in titers between isotypes against viral escape mutants. Shown are the IC50 (μg/mL) for each mAb against the CAP88 T/F and sensitive viral escape mutants. The mAbs are colored according to isotype (as above). Geometric mean of potency for each antibody is shown by a solid line. Statistically significant differences among isotypes were determined with a Wilcoxon signed-rank test, denoted by p values, and had large effect sizes of 0.6, 0.6, 0.63, and 0.51 with Z scores of 2.7, 2.7, 2.5, and 2.3, respectively.

See also Tables S2 and S3.

We also observed differential neutralization of escape variants between isotypes. For example, CAP88-CH06 expressed as IgG3 or IgA1 showed better neutralization of viruses containing E343K, I339N, and I339N+E343K than the matched IgG1 version did (Figure 3C). Overall, for CAP88-CH06, H7, and H8, the IgG1 versions were significantly less effective at neutralizing viral escape variants than otherwise identical antibodies (Figure 3D). These data show that isotype, in addition to SHM, influenced the neutralization of viruses containing escape mutations.

Our data suggest that isotype switching may have either beneficial or deleterious outcomes for antibody evolution. In this lineage, early switches from H1-IgG3 to IgG1, IgA1, or IgA2 resulted in “dead-end” sub-lineages that are unable to neutralize the early E343K viral escape variant that was neutralized by the LCA-IgG3 and H1-IgG3 (Figure 4). The selection of IgG3 clones better able to neutralize emerging viral variants fostered continued maturation of the lineage. However, again, detrimental switches from IgG3 to IgG1 (H7 and H8) resulted in reduced neutralization of viruses with escape mutations. In contrast, the emergence of IgA1 sub-lineages from IgG1 dead-end precursors (Figure 1E) suggests that switching to IgA1 restored the ability of antibodies to neutralize emerging viruses bearing escape mutations. This “switch redemption” is illustrated using CAP88-CH06, where an initial switch from an IgG3 to IgG1 resulted in reduced neutralization of the E343K and I339N mutants and loss of neutralization of I339N+E343K. Neutralization was restored by a further switch to an IgA1, which enhanced neutralization of E343K or I339N mutants and restored the ability of antibodies to neutralize viruses bearing I339N+E343K. This switch redemption thereby enabled further maturation of the IgA1 sub-lineage, with H9 showing increased SHM and enhanced neutralization potency against the same viruses (Figure 4).

Figure 4.

The Role of Class-Switch Recombination and Isotype in Antibody-Virus Co-evolution

Schematic diagram of the influence of isotype and SHM on antibody-virus co-evolution in donor CAP88. Curved horizontal arrows represent class-switch recombination (CSR) events and are colored to indicate isotypes being switched. Straight vertical arrows represent increases in SHM (also indicated schematically on antibodies as black lines within heavy-chain variable regions). IC50 neutralization titers for the T/F virus and selected escape mutations are shown above each antibody. Mutations are ranked in order of resistance, with E350K being the most resistant. Four examples (for H1, H7, H8, and CAP88-CH06) of detrimental switches are shown, where a switch from IgG3 to another isotype, usually IgG1, results in reduced neutralization. However, in CAP88-CH06, a further isotype switch from IgG1 to IgA1 rescued that loss of neutralization. See also Table S3.

DISCUSSION

In conclusion, we have shown that isotype switching, in addition to SHM, contributes to antibody-virus co-evolution by conferring differential neutralization of emerging viral variants. Although several studies of individual mAbs have shown that isotype can affect neutralizing activity (Jia et al., 2020; Kozel et al., 2007; Maurer et al., 2018; Richardson et al., 2019; Tudor et al., 2012), this is the first, to our knowledge, to systematically evaluate the contribution of CSR to the evolution of a nAb lineage. The combined role of both SHM and CSR in adjusting B cell specificity has been well described in the intestinal mucosa, where continuous exposure to commensal microbiota results in iterative “editing” of IgM/IgA specificities in response to the changing microbiota (Benckert et al., 2011; Ellebrecht et al., 2018; He et al., 2007; Horns et al., 2016; Kawamoto et al., 2014; Kitaura et al., 2017; Lindner et al., 2015; Magri et al., 2017; Planer et al., 2016). This is analogous to what we have shown here, with continuous exposure to a changing viral quasispecies similarly driving Env-reactive B cells to “adjust” their variable and constant domains through SHM and CSR.

The CAP88-CH06 antibody lineage was sequenced from PBMCs and sequences were thus most likely from plasma and memory B cells, although their specific origin could not be determined. The identification of IgA1 and IgA2, enriched at mucosal surfaces, raises the possibility that these antibodies may be derived from gut-homing memory B cells (He et al., 2007). However, although IgA1/2 exist predominantly as dimers at mucosal surfaces, these antibodies are mostly monomeric in plasma (80%–99%), enabling us to directly compare the neutralization of IgA1 monomers with IgG1/IgG3 in this study (Schroeder and Cavacini, 2010; Woof and Mestecky, 2005).

CSR can occur both sequentially from IgM through IgG, then to IgA, or directly from IgM to IgA through the deletion of intermediate genes. Much of the CAP88 lineage persists as IgG3; however, we also observed sequential CSR in sub-lineages from the LCA-IgG3 to IgG1, followed by IgA1. Our data are consistent with both with reports of preferential switching to IgG3 during early HIV-1 infection (Kardava and Moir, 2019; Yates et al., 2011) and with previous studies of CSR frequency in which IgG3 B cells most frequently switch to IgG1/IgG2 (~87%) and IgG1 to IgA1/IgG2 (92%) (Horns et al., 2016; Jackson et al., 2014). Our sequencing has certainly not fully captured the total level of isotype variation within this lineage, as indicated by the fact that we did not detect the H1-IgG3 transcript. Despite that, we show that class switching has important functional consequences for lineage maturation.

It has become increasingly evident that SHM can have either advantageous or disadvantageous effects on neutralization (Pappas et al., 2014; Sacks et al., 2019). Here, we show, likewise, that CSR can have variable outcomes for neutralization. We provide evidence that disadvantageous class switches can be redeemed by a further stochastic switch resulting in antibodies with increased neutralization, which we term “switch redemption.” The persistence and progeny of isotypes with enhanced neutralization indicates that, within germinal centers, these B cells were preferentially selected for (Victora and Nussenzweig, 2012). This concept of switch redemption is, in some respects, similar to that of “clonal redemption” of auto-reactive nAbs by SHM (Finney and Kelsoe, 2018; Reed et al., 2016; Sabouri et al., 2014).

The effect of Fc on epitope recognition is influenced by structural differences among antibody isotypes (Giuntini et al., 2016; Janda et al., 2016; Schroeder and Cavacini, 2010). IgG3 has the longest hinge (up to 62 aa), which can enhance neutralization (Richardson et al., 2019; Vidarsson et al., 2014). Although IgA1 has a similar hinge length to IgG1 (~17–18 aa), it typically has a more open conformation, is highly glycosylated, and has a distinct CH1 that has been linked to improved viral neutralization (Maurer et al., 2018; Miranda et al., 2007; Tudor et al., 2012; Woof and Mestecky, 2005; Woof and Russell, 2011). Together these features likely provide a mechanism for the enhanced neutralization of IgG3 and IgA1 versions of CAP88-CH06 lineage antibodies compared with IgG1. However, many HIV antibodies isolated as IgG1 have high levels of neutralization breadth and potency (Kwong and Mascola, 2018), and IgA and IgG3 have been associated with divergent outcomes in vaccine trials and transmission experiments (Astronomo et al., 2016; Kulkarni and Ruprecht, 2017; Neidich et al., 2019). Thus, the influence of isotype on neutralization almost certainly varies between and within lineages.

Nonetheless, our data indicate that previous co-evolution studies, which have used IgG1-expressed antibodies, regardless of the native isotype, may not completely capture the diversity of antibody maturation pathways. Furthermore, because isotype diversity correlates with neutralization breadth (Richardson et al., 2018), these data have translational implications for vaccines to HIV and other pathogens, highlighting the need for adjuvants that drive both SHM and isotype diversity (Avramidis et al., 2002; Knudsen et al., 2016). Overall, these data add a new layer of complexity to our understanding of nAb maturation.

Limitations of Study

Here, we measured the relationship between antibody and antigen by functional neutralization assays and, in a smaller subset, by binding to soluble antigens. Follow-up studies will need to provide a mechanistic explanation for the varying effect of isotype on binding and neutralization in the context of otherwise identical antibodies. In addition, such studies will need to interrogate the relationship between binding, neutralization, and B cell receptor activation, a more sensitive proxy of B cell selection in germinal centers. These data will further define the role of switch redemption in antibody lineage maturation.

STAR★METHODS

RESOURCE AVAILABILITY

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Penny Moore (pennym@nicd.ac.za).

Materials Availability

This study did not generate new unique materials.

Data and Code Availability

Unprocessed MiSeq antibody sequencing data for the CAP88-CH06 lineage, and viral C3 deep sequencing, have been deposited for public access into the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA), under the following Bio-Project accession numbers: SRA:PRJNA556126 and SRA:PRJNA557574. The heavy and light chain antibody sequences used to generate the LCA and CAP88-CH06 as well as heavy chains of H1 – H9 have been deposited into GenBank, and their accession numbers are listed in the Key Resources Table. Viral envelope sequences are also available in GenBank, accession numbers provided in the Key Resources Table.

KEY RESOURCES TABLE

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Palivizumab	Medimmune	Synagis; RRID: AB_2459638
A32	NIH AIDS Reagent Program	11438
CAP88-CH06	NICD	n/a
KC57-FITC p24 antibody	Beckman-Coulter	6604665
Bacterial and Virus Strains
Escherichia coli JM109	Inqaba	T3003
Escherichia coli XL 10-Gold Ultracompetent Cells	Agilent Technologies	200315
Biological Samples
CAP88-derived PBMC samples	CAPRISA (van Loggerenberg et al., 2008)	CAPRISA (https://www.caprisa.org)
CAP88-derived Plasma samples	CAPRISA (van Loggerenberg et al., 2008)	CAPRISA (https://www.caprisa.org)
Pooled HIV Clade C IgG	NIH AIDS Reagent Program	3957
HIV negative donor PBMC	NICD	N/A
Chemicals, Peptides, and Recombinant Proteins
Superscript III Reverse Transcriptase	Invitrogen	18080093
PfuUltra II Fusion HS DNA polymerase	Agilent	600380
X-tremeGene 9 DNA Transfection Reagent	Roche	06365809001
Ampure XP Beads	Beckman-Coulter	A63880
SPIRselect Magnetic Beads	Beckman-Coulter	B23317
FreeStyle 293 Expression Medium	LTC Tech SA	11625019
Protein A-Agarose	BioVision	6526-100
Immobilized Protein G	ThermoFischer Scientific	20397
CaptureSelect™ IgA Affinity Matrix	ThermoFischer Scientific	194288010
PhiX Control V3	Illumina	FC-110-3001
Critical Commercial Assays
QiAamp Viral RNA kit	QIAGEN	52904
AllPrep DNA/RNA mini kit	QIAGEN	80204
Nextera XT Indexing Kit V2 Set B (Antibody NGS)	Illumina	FC-131-2002
Nextera XT Indexing Kit V2 Set A (Viral NGS)	Illumina	FC-131-2001
Big Dye™ Terminator v3.1 Cycle Sequencing Kit	Applied Biosystems	4337456
MiSeq Reagent Kit V3	Illumina	MS-102-3003
Qubit dsDNA HS assay	ThermoFischer Scientific	Q32854
Agilent High Sensitivity DNA Kit	Agilent Technologies	5067-4626
Deposited Data
Raw Antibody MiSeq Data	This paper (SRA)	SRA:PRJNA556126
Lineage Antibody Sequences	This paper (GenBank)	GenBank:MN228641 - GenBank:MN228657
Raw C3 Envelope MiSeq Data	This paper (SRA)	SRA:PRJNA557574
Single Genome Amplification Env Sequences	This paper (GenBank)	GenBank:KU198436.1, GenBank:MK205449.1, GenBank:MK205458.1, GenBank:MK205486.1, GenBank:MK205487.1, GenBank:MK205490.1, GenBank:MK205492.1, GenBank:MK205495.1, GenBank:MK205508.1, GenBank:MK205514.1L, GenBank:MK205520.1, GenBank:MK205531.1, GenBank:MK205564.1, GenBank:MK205575.1, GenBank:MK206069.1, GenBank:MK206074.1, GenBank:MK206078.1, GenBank:MK206112.1 and GenBank:MK206156.1
Experimental Models: Cell Lines
HEK293F	LTC Tech SA	11625019
HEK293T	Dr George Shaw	University of Alabama
CEM.NKRCCR5	NIH AIDS Research Reagent Program	4376
TZM-bl	NIH AIDS Research Reagent Program	8129
Oligonucleotides
EnvM primer: 5′-TAGCCCTTCCAGTCCCCCCTTTTCTTTTA-3′	Gao et al., 1996	N/A
EnvA primer: 5′-GGCTTAGGCATCTCCTATGGCAGGAAGAA-3′	Gao et al., 1996	N/A
IGHV4-39 forward primer: 5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCAGSTGCAGCTGCAGGAGTCGG-3′	Integrated DNA Technologies	N/A
IGLV3-21 forward primer: 5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGTCCTATGTGCTGACTCAGCCACCC-3′	Integrated DNA Technologies	N/A
IgG reverse primer: 5′-GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGCTTGACCAGGCAGCCCAGGGC-3′	Integrated DNA Technologies	N/A
IgA reverse primer: 5′-GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGAAGAAGCCCTGGACCAGGCA-3′	Integrated DNA Technologies	N/A
Lambda reverse primer: 5′-GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGCACCAGTGTGGCCTTGTTGGCTTG-3′	Integrated DNA Technologies	N/A
ENV_C3_cDNA_R: 5′-GCCTTGCCACACGCTCAGGCNNNNNNNNNNNNNNNTGTGTTGTAAYTTCTAGRTC-3′	Integrated DNA Technologies	N/A
ENV_C2_Fwd,5′-TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGNNNNNNNGCTGGTTATGCGATTCTAAAGTG-3′	Integrated DNA Technologies	N/A
Univ_i7_Rev,5′-GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGCCTTGCCACACGCTCACGC-3′	Integrated DNA Technologies	N/A
Random Hexamers	Integrated DNA Technologies	51-0118-01
Recombinant DNA
pCDNA 3.1D-TOPO	Invitrogen	K4900-01
pSG3ΔEnv	NIH AIDS Reagent	11051
Software and Algorithms
Sequencher v5.4.6	Genecodes	http://www.genecodes.com/
Clustal X v1.83	Larkin et al., 2007	http://www.clustal.org/
BioEdit v7.2.5	Hall, 1999	https://bioedit.software.informer.com/7.2/
SONAR	Schramm et al., 2016	https://github.com/scharch/SONAR and https://hub.docker.com/r/scharch/sonar/
PEAR v0.9.6	Zhang et al., 2014	https://cme.h-its.org/exelixis/web/software/pear/
USEARCH v9.0.2132	Edgar, 2010	http://www.drive5.com/usearch
R v3.5.3		https://www.r-project.org/
R coin package	Hothorn et al., 2008	https://cran.r-project.org/web/packages/coin/index.html
R igraph package	Csardi and Nepusz, 2006	https://igraph.org/r/
R alakazam package	Stern et al., 2014	https://cran.r-project.org/web/packages/alakazam/index.html
Immcantation portal	Gupta et al., 2015	https://immcantation.readthedocs.io/en/stable
IMGT High V-Quest	Alamyar et al., 2012	http://www.imgt.org/HighV-QUEST/home.action
NGS Data Processing Pipeline	HIV Diversity Group, UCT	https://github.com/HIVDiversity/NGS_processing_pipeline
MotifBinner2	HIV Diversity Group, UCT	https://github.com/HIVDiversity/MotifBinner2
MAFFT v7.427	Katoh and Standley, 2013	https://mafft.cbrc.jp/alignment/software/
Python scripts for viral aa frequencies	HIV Diversity Group, UCT	https://github.com/HIVDiversity/NGS_analysis_pipeline/blob/master/calc_multi_site_freq.py

Custom data analysis pipelines used for viral NGS analysis are available on GitHub (links provided in the Key Resources Table).

EXPERIMENTAL MODELS AND SUBJECT DETAILS

Human Subjects

CAP88, a CAPRISA 002 cohort participant, was infected with an HIV-1 subtype C virus in 2005 (Gray et al., 2007; van Loggerenberg et al., 2008). Stored plasma and peripheral blood mononuclear cells (PBMCs) from 5 - 108 weeks post-infection were used for virus gp160 envelope cloning and for antibody studies. The CAPRISA 002 study was approved by the ethics committees of the University of KwaZulu-Natal (E013/04), the University of Cape Town (025/2004) and the University of the Witwatersrand (MM040202) and included informed consent from donor CAP88.

Cell Lines

HEK293F suspension cells were cultured in 293Freestyle media (GIBCO) and grown in a shaking incubator at 37°C, 5% CO2, 70% humidity at 125rpm. HEK293T cells were obtained from Dr. George Shaw (University of Alabama, Birmingham, AL) and were used for the expression of proviral IMC plasmid DNA for ADCC assays. These adherent cell lines were cultured at 37°C, 5% CO2, in DMEM containing 10% heat-inactivated fetal bovine serum (GIBCO) and supplemented with 50 μg/ml gentamicin (Sigma). Cells were disrupted at confluence with 0.25% trypsin in 1 m MEDTA (Sigma) every 48–72 hours. CEM-NKRCCR5 were obtained from the AIDS Reagent Program (Division of AIDS, NIAID, NIH) and used as target cells in the ADCC assay. These cells were cultured at 37°C, 5% CO2 in RPMI containing 10% heat-inactivated fetal bovine serum (GIBCO) and 1% Penicillin Streptomycin (GIBCO). TZM-bl cells (JC53-bl (clone 13)) cells were obtained from AIDS Reagent Program (Division of AIDS, NIAID, NIH) and used for the neutralization assays (Montefiori, 2005).

METHOD DETAILS

CAP88-CH06 lineage next generation sequencing

Total RNA was extracted from cryopreserved PBMC at 7 different time-points (8, 11, 17, 34, 38, 46 and 108 wpi) using the AllPrep DNA/RNA mini kit (QIAGEN). Reverse transcription was carried out using Random Hexamers (Integrated DNA Technologies) and Superscript III RT enzyme (Invitrogen). Lineage specific primers bound leader regions of IGHV4-39 and IGLV3-21 and included the Illumina MiSeq barcodes to allow sequencing on the MiSeq (primers listed in the Key Resources Table). Samples from each time point were amplified as seven replicates for both the heavy and light chains, to ensure adequate coverage and to minimize PCR bias. PCR conditions were based on previous experiments (Scheepers et al., 2015) but modified by increasing the annealing temperature to 65°C and reducing cycles to 25. Nextera XT unique dual indexing combinations selected from Illumina Indexing Kit V2 Set B were added to the pooled MiSeq amplicon libraries. All products were checked on an Agilent Bioanalyser High Sensitivity DNA kit (Diagnostech) and Qubit dsDNA HS assay (ThermoFischer Scientific) and cleaned-up using 0.75X Ampure Beads (Beckman-Coulter). A final concentration of 4.5 pM denatured DNA library with 10% PhiX control (Illumina) was run on the Illumina MiSeq, using the MiSeq reagent kit (version 3) with 2 × 300 paired-end reads.

CAP88-CH06 lineage analysis

Paired-end MiSeq reads were merged into full-length reads for each time point using PEAR (Zhang et al., 2014). Paired sequences were then de-replicated using USEARCH (Edgar, 2010) resulting in unique reads. The SONAR bioinformatics pipeline (Schramm et al., 2016) was used to identify CAP88-CH06 clonally related reads. In brief, germline V and J genes were assigned, the CDR3 regions were identified, and identity to CAP88-CH06 for each of the reads was calculated. Clonally related sequences were selected based on germline gene usage and CDR3 identity ≥ 90% to CAP88-CH06. To account for sequencing and/or PCR error, all singletons were removed and the remaining sequences were clustered at 99% identity using USEARCH, and the most abundant representative was selected from clusters with three or more transcripts for downstream analyses. Tools from the immcantation portal were used to reconstruct the antibody lineage and create the phylogenetic tree (Gupta et al., 2015). Briefly, clonally related heavy chain sequences were submitted to IMGT High V-quest (Alamyar et al., 2012; Li et al., 2013) for analysis and the resulting summary files were then parsed into a Change-O database. The LCA of the lineage was used as the germline defined sequence for the reconstruction. Within R, the Alakazam package (Stern et al., 2014) was used to read in the Change-O database and reconstruct the lineage creating clones and introducing inferred sequences using buildPhylipLineage. The resulting tree was plotted using igraph (Csardi and Nepusz, 2006). The edge lengths of the tree were further modified to be proportional to the number of mutations between nodes and all nodes with no mutations from the parent node were deleted using Inkscape v 0.91.

CAP88-CH06 lineage antibody production

Selected clonally related heavy and light chain sequences were ordered from GenScript. All heavy chain clonally related sequences were paired with the mature CAP88-CH06 light chain. The LCA heavy chain was paired with the LCA light chain (LCA). These heavy and light chain pairs were sub-cloned into IgG1, IgG3, IgA1 or IgA2 backbones, co-transfected and expressed in HEK293F cells (NIH AIDS Reagent Program), confirmed to be Mycoplasma free, grown in FreeStyle 293 Expression Medium (GIBCO) at 37°C, 5% CO2, 70% humidity and 125 rpm. Cultures were harvested after seven days by centrifugation at 4000 x g, and supernatants were filtered through a 0.22 μm filter. Filtered supernatant of IgG1, IgG3, IgA1 and IgA2 antibodies were purified by chromatography using protein A-Agarose (BioVision), immobilized protein G (ThermoFischer Scientific) and CaptureSelect™ IgA affinity matrix (ThermoFischer Scientific), respectively.

Size exclusion chromatography

The different isotypes of CAP88-CH06 and LCA were analyzed using size exclusion chromatography (SEC) on a Superdex 200 HiLoad 16/600 pg column (GE healthcare). The mAbs were run on the column equilibrated with 1x phosphate buffered saline (PBS). All mAbs were run at a concentration of 1 mg/mL and volume of 1mL, with the exception of the CAP88-CH06-IgG3 in which only 500uL was run at 1mg/mL. The calibration and operations of the system were set as per manufacturer’s instructions. Graphs were produced in GraphPad Prism version 8.0.

Single genome envelope amplification and sequencing

HIV-1 RNA was purified from plasma using the QiAamp Viral RNA kit (QIAGEN), and reverse transcribed to cDNA using Superscript III Reverse Transcriptase (Invitrogen). The env genes were amplified from single cDNA copies (Keele et al., 2008) and amplicons were directly sequenced using the ABI PRISM Big Dye Terminator Cycle Sequencing Ready Reaction kit (Applied Biosystems) and resolved on an Applied Biosystems 3500xL Genetic Analyzer. The full-length env sequences were assembled and edited using Sequencher v.5.4.6 software (Gene Codes Corporation). Multiple sequence alignments were performed using Clustal X v1.83 and edited with BioEdit v7.2.5.

Envelope cloning and production of env-pseudotyped viruses

Selected viral amplicons were cloned into the expression vector pcDNA™ 3.1D/V5-His-TOPO (Invitrogen) by re-amplification of SGA first-round products using PfuUltra II Fusion HS DNA polymerase (Agilent) with the EnvM and EnvA primers (Gao et al., 1996). Env-pseudotyped viruses were obtained by co-transfecting the Env plasmid with pSG3Δenv (NIH AIDS Reagent) using FuGENE transfection reagent (Roche) as previously described (Gray et al., 2007).

Neutralization assays

Neutralization was measured as a reduction in luciferase gene expression after a single round infection of TZM-bl cells (NIH AIDS Reagent Program), confirmed to be Mycoplasma free, with Env-pseudotyped viruses (Montefiori, 2005). Titers were calculated as the inhibitory concentration (IC50) or reciprocal plasma/serum dilution (ID50) causing 50% reduction of relative light units (RLU) with respect to the virus control wells (untreated virus). Reported neutralization titers are the average of three or more titers for any given virus/antibody combination.

Antibody-gp120 binding assay using biolayer interferometry

The binding affinities of CAP88-CH06 expressed as IgG3, IgG1 and IgA1 against a his-tagged gp120 core derived from the CAP88_3100_030wpi_29 virus were measured by biolayer interferometry (BLI) using the Octet Red system (Pall FortéBio). Ni-NTA biosensors were initially hydrated in 1x kinetics buffer (1x PBS, 0.002% Tween, 0.01% bovine serum albumin) and loaded with 10 μg/mL gp120 for 60 s. Biosensors were then transferred into wells containing 1x kinetics buffer to achieve a baseline before being transferred into wells containing 2-fold serial dilutions of the antibodies at starting dilutions of 500 nM. This association phase was subsequently followed by a dissociation step with 1x kinetics buffer. Analysis was performed using Octet software with a 1:1 fit model. Experiments were repeated in duplicates or triplicates, and values were averaged.

Infected cell ADCC assay

The HIV-1 reporter virus used in the ADCC assays was a replication-competent infectious molecular clone (IMC) designed to encode the T/F virus (CAP88_2000_005wpi_17, subtype C) env gene in cis within an isogenic backbone (NL-LucR.T2A-ENV.ecto) that also expresses the Renilla luciferase reporter gene, and preserves all viral open reading frames (Edmonds et al., 2010; Li et al., 2005). Reporter virus stocks were generated by transfection of HEK293T cells (NIH AIDS Reagent Program) with proviral IMC plasmid DNA, and titered in CEM.NKRCCR5 cells (NIH AIDS Reagent Program) for infectivity by p24 staining (Beckman-Coulter). All cell lines were confirmed to be Mycoplasma negative. The infected cell assay was used to measure ADCC activity as previously described (Pollara et al., 2014). Briefly, a CEM.NKRCCR5 cell line (NIH AIDS Reagent Program) was used as targets for ADCC luciferase assays after infection with the HIV-1 IMCs listed above. The target cell line was infected with IMC using titered stock that generated more than 50% infected target cells after 72 hours of infection. The target cells were incubated with 5-fold serially diluted mAbs starting at 50 μg/mL. Cryopreserved peripheral blood mononuclear cells (PBMC) obtained from a HIV seronegative donor with a high-affinity 158V/V Fcg receptor IIIa phenotype were used as source of effector cells. After thawing, the cryopreserved PBMCs were rested overnight and used at an effector to target ratio of 30:1. The effector cells, target cells, and Ab dilutions were plated in white 96-well half area plates and incubated for 6 hours at 37°C in 5% CO2. The final readout was the luminescence intensity (in relative light units) generated by the presence of residual intact target cells that have not been lysed by the effector population in the presence of any ADCC-mediating mAb. The percentage of killing was calculated using the formula:

In this analysis, the RLU of the target plus effector wells represents non-antibody background. The RSV-specific mAb Palivizumab (Medimmune; Synagis) and A32 (NIH AIDS Reagent Program) (to show correct conformation of the envelope) were used as negative controls and a polyclonal mixture of IgG from Clade C HIV infected individuals from the NIH AIDS Reagent Program. Data are represented as the area under the curve (AUC) of percentage specific killing over the serially diluted antibodies.

Viral deep sequencing library preparation

Preparation of viral amplicons for deep sequencing (RNA extraction, cDNA synthesis and subsequent amplification), was carried out as described previously (Jabara et al., 2011), with the following modifications: A minimum of 5,000 HIV-1 RNA copies were isolated from longitudinal plasma samples spanning two years of infection, using the QIAamp Viral RNA kit (QIAGEN). The cDNA synthesis primer (ENV_C3_cDNA_R) was designed to bind to the C3 region of the HIV-1 envelope gene (HXB2 gp160 amino acid position 368-374) and included a randomly assigned 15-mer tag (Primer ID method), to uniquely label each cDNA molecule and assist with phasing over the first cycles of the MiSeq sequencing run. This tag was preceded by a universal primer binding site to allow out-nested PCR amplification of cDNA templates (primers are listed in Key Resources Table). The amplification primer (forward: ENV_C2_Fwd and reverse: Univ_i7_Rev) was designed to bind to the end of the C2 region (HXB2 gp160 amino acid position 221-228) and contained a template-specific binding region, preceded by seven randomly assigned bases to assist with phasing during MiSeq (primers provided in Key Resources Table). In addition, PCR primers contained 5′ overhangs, introducing binding sites for the Nextera XT indexing primers (Illumina, CA). This allowed the amplification of the C2 to C3 region of the envelope spanning HXB2 gp160 amino acid positions 228 to 368. Nested PCR reactions were carried out using the Nextera XT indexing kit (version 2, Set A). After indexing, samples were purified using SPIRselect magnetic beads (Beckman Coulter, CA) with a ‘sample volume to bead ratio’ of 1:0.65, for the removal of < 300bp fragments. Samples were quantified using the Qubit dsDNA HS assay (ThermoFisher) and pooled in equimolar concentrations. The pooled library was cleaned again using the SPIRselect magnetic beads to ensure primer removal before the final library was submitted for sequencing on an Illumina MiSeq (version 3), using 2 × 300 paired-end chemistry.

Viral deep sequencing data analysis

Custom data analysis pipelines used for viral NGS analysis are available on GitHub (links provided in the Key Resources Table). Raw reads were processed using an NGS_processing_pipeline to automate all steps. Briefly, this pipeline processes through the following steps: raw reads were length and quality filtered, merged using PEAR (Zhang et al., 2014), and processed according to the Primer ID method (Jabara et al., 2011), including the removal of sequences resulting from offspring bins (Zhou et al., 2015), using MotifBinner2. Consensus sequences were further processed to remove sequences with degenerate bases, large deletions (> 50bp), any non-HIV contaminating sequences. Finally, the resulting sequences were codon aligned to HXB2 using MAFFT v7.427(Katoh and Standley, 2013). The resulting codon alignment was then manually curated to correct any alignment errors. The entire haplotype (positions 339, 343, 351 and 358) was considered when calculating the amino acid frequencies for each escape variant genotype.

QUANTIFICATION AND STATISTICAL ANALYSIS

Differences in potency between isotypes for the LCA, H1, H7, H8 and CAP88-CH06 were assessed using the Wilcoxon Signed-Rank Test (R coin package) and effect sizes were calculated based on the absolute value of Wilcoxon Signed-Rank Test z-scores divided by the square root of the number of viruses tested for both antibodies: effect size = abs(z-score)/sqrt(# viruses). The number of viruses per comparison for autologous viruses were as follows: n = 6, n = 4, n = 16, n = 26, n = 18 for the LCA, H1, H7, H8 and CAP88-CH06, respectively. The number of viruses per comparison for escape mutations were as follows: n = 8, n = 6, n = 16, n = 20, n = 18 for the LCA, H1, H7, H8 and CAP88-CH06, respectively. All statistical tests were calculated using R, version 3.5.3.