https://orcid.org/0000-0001-6825-1453
Figures
Abstract
Antibodies to human papillomavirus (HPV) primarily recognize surface exposed residues on five loops of the major capsid protein (L1) that vary significantly among HPV types. We determined which loops were required for neutralization for 68 HPV16 specific human monoclonal antibodies (mAbs) cloned from participants who received an HPV vaccine and describe molecular features of those antibodies. Chimeric HPV16 pseudovirus (cpsV), each having one surface loop bearing multiple amino acid substitutions, were used to determine neutralization specificity. The HPV16-FG-loop was the loop most frequently required for neutralization (42 of 68, 61.8%), however, all surface loops were required for neutralization by multiple mAbs: HI (13, 19.1%), DE (15, 22.1%), EF (five, 7.4%), BC (four, 5.9%). Antibodies that required multiple loops were common (17, 25.0%). Three mAbs (4.4%) required sequences on the c-terminus of L1 and for another three mAbs the neutralization specificity could not be determined. Two types of mAbs appeared to be overrepresented: ten mAbs used immunoglobin heavy chain variable region 2–70 (IGHV2–70) with immunoglobin light chain variable region 1–40 (IGLV1–40), having characteristic mutations in complementarity determining region two (CDRL2) of the light chain. Cryogenic electron microscopy (Cryo-EM) revealed that two of these antibodies bound five Fabs per capsomer interacting with all five L1-surface loops. The other type of mAbs that appeared to be overrepresented were nine mAbs using IGHV4–34, six of which also used DH3-16*02 with conserved CDRH3 sequences. Cryo-EM for one of these mAbs, that required the FG-loop for neutralization, was shown to bind one Fab per capsomer at the apex, interacting with the DE- and FG-loops, with sequences of the Fab CDRH3 inserted between the DE- and FG-loops from two L1 proteins. These two types of mAbs were found in the four participants suggesting that these antibodies shared developmental pathways and bound to similar immunodominant epitopes on the virus.
Author summary
Vaccines that protect against infection with certain types of HPVs generate neutralizing antibodies against those specific types. To determine what those antibodies recognize we generated a panel of 68 HPV16 specific human monoclonal antibodies (mAbs), cloned from memory B cells or plasmablasts following HPV vaccination. Previous studies indicated that surface loops of the major capsid protein (L1) were required for neutralization. All five L1 loops were found to be used for neutralization by four or more antibodies, but the most frequent target of these antibodies was the FG loop followed by the HI and DE loops. Ten antibodies paired the heavy chain variable gene IGHV2–70 with the light chain variable gene IGLV1–40. Cryo-EM analysis of two IGHV2–70/IGLV1–40 mAbs showed five antibody binding fragments (Fabs) binding per L1 capsomer. Six other mAbs had a heavy chain composed of the variable region IGHV4–34 with the diversity gene D3-16*02 resulting in the sequence motif WSGYR in the CDRH3. A cryo-EM structure of one of these antibodies showed one Fab binding the capsomer apex with the WSGYR motif inserting between three loops. Both types of antibodies were found in all four study participants suggesting that such antibodies may be commonly elicited following HPV vaccination.
Citation: Carter JJ, Hurlburt NK, Scherer EM, Singh S, Rodarte JV, Smith RA, et al. (2025) Human monoclonal antibodies to HPV16 show evidence for common developmental pathways and public epitopes. PLoS Pathog 21(10): e1013086. https://doi.org/10.1371/journal.ppat.1013086
Editor: Zhen Lin, Tulane University School of Medicine, UNITED STATES OF AMERICA
Received: March 28, 2025; Accepted: October 11, 2025; Published: October 21, 2025
Copyright: © 2025 Carter et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: Atomic models have been deposited in the RCSB PDB with PDB IDs pdb_00009ml1 (D24.1M01), pdb_00009ml2 (A7M08), and pdb_00009ml3 (B25M05). Maps generated from the electron microscope data are deposited in the Electron Microscopy Data Bank under EMD-48344 (D24.1M01), EMD-48345 (A7M08), and EMD-48346 (B25M05).
Funding: This work was supported by National Institute of Allergy and Infectious Diseases (R01 AI038382 to DAG), National Cancer Institute (R35 CA209979 to DAG) and from a Pathogen Associated Malignancy Interactive Research Center (PAM IRC) Pilot from Fred Hutchinson Cancer Center (to MP). Cryo-EM data was supported by Pacific Northwest Center for Cryo-EM and National Institute of General Medical Sciences (R24GM154185) and Electron Microscopy Shared Resource, (RRID:SCR_022611) of the Fred Hutch/University of Washington/Seattle Children’s Cancer Consortium Cancer Center Support Grant from National Cancer Institute (P30 CA015704). A portion of this research was supported by a National Institutes of Health equipment grant (1S10OD028581-01). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: I have read the journal’s policy and the authors of this manuscript have the following competing interests. DG reports a relationship with Merck Sharp and Dohme Pharmaceuticals that includes: consulting or advisory and funding grants. Honoraria paid from Merck to EMS for lectures/seminars delivered. The other authors report no conflict of interest.
Introduction
Sexually transmitted human papillomavirus (HPV) infections are common and generally benign but occasionally result in malignant transformation of infected cells. HPVs classified as “high risk” infect mucosal epithelium and cause virtually all cervical cancers [1], a significant proportion of other anogenital cancers and head and neck cancers [2,3]. HPV16 is the HPV type most frequently found in cancers [4], accounting for approximately 50% of cervical cancers and greater than 90% of oropharyngeal cancers [3]. Current vaccines, when administered prior to the initiation of sexual activity, provide sustained protection from infection and disease [5,6]. Currently, two doses of vaccine are recommended [7] for children under 15 years old in the US, and the WHO has approved a single dose HPV vaccine [8–10]. We were interested in better characterizing the B cell response elicited by such an effective vaccine.
The HPV virion is composed of an approximately eight kb circular double stranded DNA genome, encapsidated by major (L1) and minor (L2) capsid proteins that assemble into a T = 7 icosahedral particle composed of 72 pentameric capsomers [11]. Twelve capsomers have five neighbors (pentavalent) and 60 capsomers have six neighbors (hexavalent). The C-terminus of each L1 connects capsomers by an extension, referred to as an arm, that bridges the intercapsomer canyon, forming disulfide bonds with the neighboring capsomer and returning to the parent capsomer [12]. Although the C-terminal arm has been shown to be immunogenic, [13–15] most antibodies target sequences encoded by five surface exposed loops (BC, DE, EF, FG and HI) [11,16–18] and because the amino acid sequences that form these loops vary by type, antibody responses tend to be type-specific or weakly cross-reactive [19,20]. Although sera are weakly cross-reactive, vaccines containing HPV16 and HPV18 have been shown to be cross-protective for closely related types HPV31 and HPV45 respectively [21–23].
With most anti-HPV antibodies recognizing the native conformation of L1, a combination of loop-swaps, site specific mutagenesis and cryogenic electron microscopy (cryo-EM) have been used to discover how antibodies bind virus and the residues on capsids important for recognition and neutralization [16–18,24–28]. Two types of antibodies have been repeatedly found [25,27,29]. The majority of antibodies described interact with residues on multiple loops, often including the FG loop. These antibodies which include the well-studied murine mAb H16.V5, bind 5:1 (5 Fabs per capsomer) in a manner described as top fringe binding [25–27]. The other type of antibody described multiple times, binds 1:1 (1 Fab per capsomer) at the apex, contacting residues on the DE and FG loops [25,27,29,30]. Antibodies recognizing sequences on a C-terminal arm of L1 bind in the canyon around hexavalent capsomers [13,14] but only one such antibody has been described. The majority of structural studies used mAbs derived from rabbits and mice [14,24,26,27,31,32]. Comparative studies using antibodies cloned from people will be important to corroborate those finding.
Characterization of antibody sequences induced by vaccination or following viral infection has led to a greater understanding of virus – host interactions, insights into the evolution of protective antibody responses and suggested new approaches for vaccine design [33–39]. Following vaccination or infection, antibodies from different people often share similar characteristics suggesting that, from the immense array of naïve B cells, a limited number of germline sequences are selected that follow similar pathways of affinity maturation and bind to shared, or public, epitopes. These recurrent antibodies may be neutralizing and, occasionally, broadly neutralizing [40].
Few studies have examined the molecular features of antibodies important for recognition and neutralization of HPVs in people [14,26,31,32,41]. Our studies on vaccine induced B-cell responses resulted in the molecular cloning of a large number of human mAbs from HPV16 specific memory B-cells and plasmablasts [42,43]. In this study we map L1 loops required for neutralization by 68 human mAbs, identify two types of mAbs that were found in all four study participants and describe the structural basis for their strain specific neutralization by cryo-EM. Our findings suggest that individuals develop antibodies with common features that bind similar, perhaps public, epitopes.
Results
Surface loops of HPV16 required for neutralization
The mAbs studied here (n = 68), were cloned from memory B cells or plasmablasts, isolated from four study participants who had received the quadrivalent Gardasil vaccine (4vHPV), comprised of types 6, 11, 16 and 18, at the recommended schedule of 0, two and six months with an additional off-label vaccine booster dose at month 24 [43]. Human mAbs (S1 Table cross-references previous mAb nomenclature) were initially tested for HPV16 binding and neutralizing activity. With few exceptions, it was found that binding antibodies were neutralizing (S1 Fig) with most having IC50s between one and ten pM (S1 Table and S2 Fig). No mAb tested cross-reacted with HPV18 in binding assays. To identify residues required for neutralization, cpsV were created in which each of the five surface loops and the variable region of the c-terminus of HPV16 L1 were substituted with sequences from other HPV types. The aim was to swap in loops from other HPV types that would be compatible with infectivity but different enough that HPV16 type specific epitopes would be disrupted. Substitutions were made using sequences from types HPV2 and HPV18 with 62.4% and 63.8% amino acid identity with HPV16 L1, respectively. Initial experiments using HPV31 sequences to replace HPV16 sequences (79.4% identical to HPV16) were not informative as most mAb could neutralize the modified psV. The cpsV with C-terminal substitutions from HPV18 L1 were not sufficiently infectious, therefore we used HPV35 L1 substitutions (80.8% identical to HPV16) which created fewer changes in this region. We chose to use the most infectious cpsV for each loop or region investigated (Fig 1A). Due to the length of the FG-loop, cpsVs were created for amino-terminal and carboxyl-terminal halves of this loop (FGa and FGb). We were not able to study all variable surface residues by this method because substitution of some residues (such as the C-terminal portion of the HI-loop) did not result in infectious psV. All cpsV reported in this study were infectious allowing us to directly assess sequences required for neutralization.
A) Diagram of L1 protein with surface loops (green) and c-terminal variable region (orange) indicated. Numbers represent amino acid positions of the start and end of each region. Loop and c-terminal sequences are shown with amino acid substitutions (blue letters), inserts and deletions (-). Asterisks indicate no change from HPV16 sequence (black letters). B) The percent neutralization of 68 mAbs (on left) against psVs (listed on bottom) are represented in a heat map after manual clustering. Brackets indicate antibody clonotypes. C) Select antibodies were tested in a secondary neutralization screen against psVs listed on top. IC50 were calculated and the log fold change in IC50 of cpsV compared with psV16 are represented by colors on the scale (bottom). D) mAb were used to neutralize cpsV with C-terminal amino acid substitutions. Colors as in (C). E) The specificity for each of 68 mAbs, based on neutralization experiments represented above is summarized in a pie chart. Where multiple loops are listed both (or all three) were required for neutralization. F) The number of mAbs that required each surface loop (left axis), multiple loops or the c-terminus detected in either the primary (black bars) or secondary (gray bar) neutralization screens.
To determine which sequences of HPV16 L1 were required for neutralization by mAbs, assays were conducted with cpsV having substitutions on one of five surface loops (Fig 1A). To conserve reagents, a concentration of antibody 100-fold higher than the IC50 for HPV16 (psV16) with no amino acid substitutions was used for the initial screen (Fig 1B and S2 Table). HPV31 psV (psV31) was included to determine if mAbs were cross-reactive as HPV31, which is closely related to HPV16 but was not included in the 4vHPV vaccine. Twenty mAbs neutralized all cpsV by greater than 50%. These mAbs were serially diluted for a secondary screen to determine IC50s against each cpsV (Fig 1C). Six mAbs neutralized all cpsV with IC50s less than four-fold greater than for psV16. Those antibodies were subsequently tested in neutralizing assays against cpsV that had amino acid substitutions on the C-terminus (psV16-c-term) (Fig 1D). Neutralizing epitopes for each mAb were considered to include a loop if a mAb failed to neutralize a particular cpsV by more than 50% in the first screen or had an IC50 at least four-fold higher than psV16 in secondary screens. While most mAbs (63 of 68, 92.6.%) were type specific and failed to neutralize psV31 more than 50%, there were five mAbs (5.9%) that effectively neutralized (>88%) psV31 (Fig 1B and S2 Table). There were 15 different patterns of neutralization (Fig 1E). Each loop was required for neutralization by at least four mAb, 18 (26.5%) mAbs required two or more loops for neutralization and three (4.4%) mAbs were able to neutralize all cpsV tested (epitope undefined). The majority of neutralizing mAbs (43, 63.2%) required loop FG (FGa and/or FGb) as part of the neutralizing epitope (Fig 1F). The next most frequently required loops for neutralization were DE (15, 22.1%) and HI (13, 19.1%). The EF-loop was required for neutralization by five mAbs (7.4%), the BC-loop by four mAbs (5.9%) and three antibodies (4.4%) required the C-terminus. There were minor differences between mAbs from the four subjects but the numbers were too small determine if patterns were significantly different (S3 Fig). Thus, most HPV16 binding mAbs were potent neutralizing antibodies that bound residues on the FG-loop for neutralization, however, there were many different specificities, and every loop was required for neutralization by four or more mAb.
Clonotypes are defined here as antibodies from the same individual, with the same heavy and light chain variable regions (IGHV and IGLV), the same length CDRH3 and CDRL3 regions and greater than 70% amino acid sequence identity in the CDRH3, (bracketed in Fig 1B, 1C and 1D) were found to have the same pattern of neutralization in most instances.
Antibodies with HPV16-HPV31 cross-neutralization activity
Six mAbs were defined here as cross-reactive as they could prevent in vitro infection with both psV16 and psV31 (at least 50%). Four of these antibodies required the HPV16 FGb-loop for neutralization activity in the initial screen (Fig 2A), E25M11 was found to require FGb in the second screen (Fig 1C). One cross-reactive antibody (D24.1M07) was a member of a clonotype consisting of four antibodies that had variable levels of HPV31 cross-reactivity (Fig 2B and 2C). To determine which sequences accounted for the increased cross-reactivity, heavy-light chain-swap mAbs were generated using the cross-reactive mAb D24.1M07 and non-cross reactive mAb D25M11. The swapped mAbs neutralized psV16 equally well (Fig 2D) however, the swapped mAb having a D24.1M07 light chain neutralized psV31 approximately 10-fold more effectively (Fig 2D). In summary, all cross-reactivity mAbs required the FGb-loop for neutralization and, for one mAb, the cross-neutralization potential could be largely attributed to the light chain.
Antibody features associated with neutralization specificity
To determine if mAbs with shared genetic features corresponded to recognition of specific L1 loops, we collapsed our set of antibodies to include only 52 unique clonotypes derived from memory B cells and the antibody gene sequence that appeared to be most over-represented was IGkV1-39, with 15 clonotypes (19 mAbs, Fig 3A). However, these antibodies did not share a common neutralization pattern, paired with eleven different IGHVs (S4 Fig) and had few somatic mutations in common. We next focused on antibodies that employed the IGHV2–70 or IGHV3–34 heavy chain variable gene segments as they were the most frequently used IGHV in the collapsed set (7/52, 13.5%, for both, Fig 3A).
A) The antibody heavy (left) and light (right) chain variable region gene segments used by the mAbs are depicted in donut graphs with the names of each variable gene segment listed. B) Alignment of the variable region heavy chain (VH) protein sequence for mAb using IGHV2-70 that paired with IGLV1-40 with the reference IGHV2-70 gene sequence (top line) with Kabat numbering. Only residues that differed from the reference sequence are shown except for the CDRH3 region where all residues are shown. Dashes indicate a gap and empty cells indicate a residue the same as the reference. C) Alignment of the light chain variable region protein sequences (VL) with reference IGLV1-40 that were paired with IGHV2-70. Only residues that differed from the reference are shown. Dashes and empty cell are the same as above. D) The percent neutralization for 10 antibodies using IGHV2-70 paired with IGLV1-40 against psVs. E) Substitution mutations reverting putative somatic mutations to predicted germline were made to the CDRL2 region of A7M08, the light chains were co-expressed with the mature heavy chain and resulting antibodies tested for psV16 neutralization. The graph is a representative psV16 neutralization experiment using antibodies with CDRL2 substitution mutations indicated (triplicate tests). Error bars are the standard deviation of triplicate titrations.
All seven clonotypes that used IGHV2–70, paired with lambda light chain variable region IGLV1–40 whereas this IGLV paired with only one of the 45 other clonotypes (Fisher’s exact p < 0.0001). These antibodies tended to have similar length CDRH3 regions (Fig 3B) but diverse CDRH3 region sequences (Fig 3B). These mAb had few amino acid substitutions in common when compared to the reference IGHV2–70 sequence. Evidence for convergent evolution was seen in the IGLV1–40 CDRL2 regions (Fig 3C) where all seven unique clonotypes (10 mAbs) had characteristic CDRL2 mutations, resulting in substitution of amino acids G with V at position 51 (all but one) and N to F or Y at position 52.
Four IGHV2–70/IGLV1–40 mAbs required the BC-loop either in the primary (mAb A7M08) or secondary screen (mAbs A7M15, E25M04, D24.1M12 (S2 and S3 Tables)). The other six antibodies neutralized all cpsV in the primary screen but did not have a consistent pattern of neutralization in secondary screening (S2 and S3 Tables). Thus, while there were some similarities between these antibodies in neutralization assays, they did not have the same neutralization profile as might be expected based on the lack of sequence homology in the CDRH3 regions (Fig 3B).
To evaluate the importance of amino acid substitutions in the CDRL2, the FTY sequence of the mature A7M08 antibody was substituted with the sequence of the predicted germline IGLV1–40 gene segment (YGN). The resulting antibody, created by coexpression of the mature A7M08 heavy chain with the light chain having the predicted germline CDRL2, showed decreased neutralization potency with an increased IC50 for psV16 by more than three logs compared with mature A7M08 (Fig 3E and S3 Table). To measure the relative importance of each amino acid, germline residues were substituted with the mature sequence found in A7M08. Both the Y49F and G50T substitutions substantially improved neutralization activity compared with the germline sequence (IC50s = 4270 + /- 1028 and 908 + /- 230 pM respectively). The N51Y mutation showed the greatest improvement in neutralization compared to the predicted germline sequence with an IC50 of 198 + /- 78 pM (Fig 3E). The germline CDRL2 was substituted into six other IGLV1–40 antibodies, and coexpressed with their mature cognate heavy chains (S3 Table). Four of the antibodies (A7M15, D24.1M01, D24.1M02, E25M04) shared the pattern seen for A7M08 where substitution to the predicted germline (YGN) sequence significantly increased IC50s relative to the mature antibody. Although the pattern was the same for these four antibodies the magnitude of the change was not as great as for A7M08, ranging from 5.3-fold for D24.1M01 to 39.7-fold for E25M04. Unexpectedly, the IC50 for two antibodies (B25M02 and D25M03) were not significantly changed when the CDRL2 was changed to the presumed germline sequence. Similar results were obtained when these antibodies were tested in HPV16L1 binding assays. To investigate the importance of the CDRL2 for B25M02 neutralization, we created an antibody with the mature heavy chain paired with predicted germline IGLV1–40. Surprisingly, this antibody had potent neutralizing activity (IC50 30.7 pM + /- 8.2) however, substituting in just the mature CDRL2 (49HVF51) significantly improved neutralization (IC50 11.2 pM + /- 3.0) (S5 Fig). These data confirmed the importance of CDRL2 for the neutralization activity of IGHV2–70/IGLV1–40 antibodies.
IGHV2–70/IGLV1–40 mAbs bind to all variable loops
To determine if IGHV2–70/IGLV1–40 mAbs share common binding epitopes and to resolve any key binding features of this class of antibodies, we used cryo-EM to determine the structure of their antigen-binding fragment (Fab) to the HPV16 L1 capsomer. We first chose to look at the binding of D24.1M01, because its neutralization pattern was typical for IGHV2–70/IGLV1–40 mAbs, as it neutralized all cpsV in the first screen (Fig 3D). We expressed and purified a soluble HPV16 L1 capsomer and formed a complex with D24.1M01 Fab. Using this complex, we obtained a cryo-EM reconstruction to 3.0 Å (Figs 4A and S6 and S4 Table). The structure revealed five Fabs bound to the L1 capsomer at the interface between two L1 proteins (protomers) (Fig 4B). Each Fab interacts with all five variable loops, the DE- and EF-loops of protomer 1 (P1) and the BC-, FG-, and HI-loops of protomer 2 (P2) (Fig 4C). The Fab binds the two protomers with a total buried surface area (BSA) of ~1,745 Å2, with ~840 Å2 from the heavy chain and ~905 Å2 from the light chain (S7 Fig). All six CDRs are involved in binding as well as the framework 3 (FR3) of the light chain. On each L1 protomer, the total BSA is ~ 1,838 Å2, with ~434 Å2, ~ 81 Å2, ~ 346 Å2, ~ 592 Å2, and ~385 Å2 from the BC-, DE-, EF-, FG-, and HI-loops, respectively (S8 Fig).
A) Cryo-EM map of D24.1M01 Fab bound to the HPV16 L1 capsomer. L1 protomers are colored in shades black, white, and blue. The heavy and light chains of D24.1M01 are colored dark purple and light purple, respectively. B) A surface representation of the structure in two orientations. One protomer and one Fab is shown with a transparent surface with a cartoon representation. C) Cartoon representation of the D24.1M01 CDRs and the footprint on L1. Each L1 variable loop is shown and assigned to a protomer, P1 or P2. D) Cryo-EM map of A7M08 Fab bound to the HPV16 L1 capsomer. L1 protomers are colored in shades black, white, and blue. The heavy and light chains of D24.1M01 are colored dark orange and light orange, respectively. E) A surface representation of the structure in two orientations. One protomer and one Fab is shown with a transparent surface with a cartoon representation. F) Alignment of CDRs from D24.1M01 and A7M08. G) Structure of matured CDRL2 of both D24.1M01 and A7M08 showing the aliphatic surface on L1 that they bind to.
To better understand the role of maturation of the CDRL2 and the difference in binding of these IGHV2–70/IGLV1–40 mAbs, we also chose to determine the structure of A7M08 as it was the one mAb of this group that failed to neutralize cpsV with changes in the BC region (Fig 3D). We made a complex of A7M08 Fab bound to the HPV16 L1 capsomer and obtained a cryo-EM reconstruction to a resolution of 2.69 Å (Figs 4D, 4E, and S6 and S4 Table). The structure showed a nearly identical binding epitope as D24.1M01 (Figs 4F and S7). An alignment of the two structures, including the L1 capsomer and each Fab, has a root mean square deviation (RMSD) of 0.8 Å over 2050 Cα atoms. A7M08 Fab aligned to D24.1M01 Fab has an RMSD of 0.5 Å over 204 Cα atoms. The total BSA of A7M08 is ~ 1,736 Å2 with the heavy chain and light chain contributing ~794 Å2 and ~942 Å2, respectively, and have a nearly identical interacting profile as D24.1M01 (S8 Fig). The affinity matured CDRL2, 49FVY51 in D24.1M01 and 49FTY51 in A7M08 (Fig 4G), sit above a small aliphatic surface formed by the EF- and FG-loops of P1 and the HI-loop of P2. The hydroxyl group of the germline Y49 could clash with the aliphatic interface between the EF- and HI-loops. The polar N51 in the germline IGHV2–70 sequence could also clash with the aliphatic groove. The structure also showed a potential mechanism for the A7M08 Fab susceptibility to BC-loop mutations. IGHV2–70 encodes for a cysteine residue in the CDRH1. D24.1M01 contains a cysteine in the CDRH3 that forms a disulfide bond with the CDRH1 (S9A Fig) that could help to lock the CDRs into this orientation. A7M08 instead has a bulky tyrosine at the equivalent position (Y97) in the CDRH3 that would potentially make it more sensitive to mutations in the BC-loop. In conclusion, these structures show that IGHV2–70/IGLV1–40 mAbs bind to the HPV16 L1 with nearly identical epitopes and gives structural insight into the necessary affinity maturation of the CDRL2.
IGHV4–34 with D3-16*02 bind FG-loops
The other frequently used IGHV was IGHV4–34, found in seven clonotypes (nine mAbs). Alignment of heavy chain sequences employing IGHV4–34 revealed a homologous sequence: 100WGSYR100D, (numbering for B25M05) in the CDRH3, which is predicted to be encoded by diversity region D3-16*02, and was present in four of seven (57.1.5%) clonotypes using IGHV4–34 (Fig 5A) compared to four of 45 (8.9%) clonotypes using other IGHV (p = 0.002). Antibodies using IGHV4–34 required the FGb loop for neutralization (Fig 5B). There were five mAb (four clonotypes) that utilized D3-16*02 that did not utilize IGHV4–34 that also required FGb either in the primary or secondary screens (Fig 5B). Five of 11 antibodies utilizing D3-16*02 required the DE-loop in addition to the FG-loop for neutralization. Replacing the WGSYR sequence with five alanines in D25M07 resulted in a loss of neutralization activity (Fig 5C). When individual alanine substitutions were made, Y-to-A100C showed the greatest loss of neutralization potency with an IC50 increase (from 1.6 + /- 0.2 pM for the mature antibody to 50.8 + /- 12.4 pM for the single amino acid substitution). Substitutions at the other four positions increased the IC50 modestly in comparison (W100), 2.6 + /1 0.6; G100, 3.9 + /- 0.9; S100B, 2.2 + /- 0.4; R100D, 6.2 + /- 1.3 pM). Thus, the IGHV4–34 heavy chain was frequently used in mAbs that recognize an epitope in the FGb-loop for neutralization and tended to use D3-16*02 resulting in a conserved sequence in the CDRH3 region (WGSYR) that was important for neutralizing activity.
A) VH sequences alignment of mAbs using IGHV4-34 compared with the reference IGHV4-34*01 sequence. Dashes indicate a gap and empty cells indicate a residue the same as the reference. The parenthesis indicates mAb using D3-16*02. B) Neutralization results (percentages) for 10 mAbs in (A) and five mAbs that use D3-16*02 but have IGHV other than IGHV4-34 (below). C) Residues WGSYR in the CDRH3 region of D25M07 were substituted with alanines individually or with five alanine amino acids. Purified antibodies were serially diluted and tested for neutralization of psV16 along with unmutated D25M07.
Structure of IGHV4–34 mAb in complex with L1 by cryo-EM
To investigate the binding mechanism of the IGHV4–34 mAbs, we chose two antibodies that had WGSYR in the CDRL3 region but had different patterns of neutralization (B25M05 and E7M03) however we could not obtain high resolution data for E7M03. We made a complex of soluble HPV16 L1 capsomer mixed with B25M05 Fab and used cryo-EM to determine the structure to a resolution of 2.9 Å (Figs 6A, S4, and S6 and S4 Table). In contrast to the IGHV2–70/IGHV1–40 mAbs, only one B25M05 Fab bound to the capsomer (Fig 6B). The Fab bound to the top, interior of the capsomer at the interface between the DE-loops of P1 and P2 and the FG-loop of P1, with a total BSA of ~941 Å2. The binding site of B25M05 at the apex of the capsomer sterically occludes all other binding sites, limiting the binding to only one Fab per capsomer. On P1 the DE- and FG-loop are buried by ~295 and ~234 Å2, respectively, and the DE-loop of P2 is buried by ~412 Å2 (S8 Fig). B25M05 binds with a total BSA of ~839 Å2, with ~679 Å2 from the HC and ~150 Å2 from the LC (S7 Fig). The binding is mainly from the CDRH3, accounting for 72% of the total BSA, with some CDRH1 (8% of the total BSA) and CDRL1 (15% of the total BSA) (Fig 6C). The conserved WGSYR motif in the CDRH3 intercalates between the three variable loops (Fig 6D), displacing the DE-loop of P1, relative to the loop orientation in the non-bound protomers (S9B Fig). Y100C, which was shown to be most sensitive to alanine mutation, sits at the bottom of the pocket formed by the FG-loop of P1 and the DE-loop of P2. The FGb-loop of P1 makes substantial contact with 100BSYR100D residues of the CDRH3 and could explain why the IGHV4–34 mAbs are sensitive to mutations in the FGb-loops. Overall, this structure reveals a drastically different binding mechanism of the IGHV4–34 mAb to the HPV16 L1 capsomer and shows how the conserved WGSYR motif encoded by D3-16*02 in the CDRH3 is crucial for binding.
A) Cryo-EM map of B25M05 Fab bound to the HPV16 L1 capsomer. L1 protomers are colored in shades black, white, and blue. The heavy and light chains of B25M05 are colored dark green and light green, respectively. B) A surface representation of the structure in two orientations. One protomer and the Fab is shown with a transparent surface with a cartoon representation. C) Top view of the binding interface of B25M05. The insert shows the three L1 loops involved in binding. D) Side view of the L1 capsomer cut out to show the binding pocket of the B25M05 CDRH3. Insert shows the WGSYR motif and how it binds in a pocket formed by the two protomers.
Discussion
As expected, most HPV16 mAbs were type specific and required sequences on one or more surface loops for neutralization activity. All surface-exposed loops on HPV16L1 were required for neutralization by one or more mAbs suggesting that much of the surface of HPV16 can be targeted by antibodies. In agreement with previous studies [18,23,41,44,45] we found that the FG-, HI- and DE-loops were frequently required for neutralization by mAbs. More than half of the antibodies studied here required FG for neutralization and DE- and HI-loops were required for neutralization by more than 19% of the mAbs tested. We could not mutate all surface-exposed residues without disrupting infectivity, therefore the methods used here likely underestimate the antibody diversity. The broad antibody diversity observed here helps to account for the nearly universal seroconversion to prophylactic HPV vaccination and the high proportion of neutralizing mAbs may explain the high efficacy of these vaccines [46]. Three human mAbs required sequences on the C-terminus for neutralization, analogous to a previously described mouse antibody [15,18].
HPV16-HPV31 cross-reactive serum antibodies are commonly induced by the bivalent and quadrivalent HPV vaccines, both of which contain HPV16 but not HPV31. Although the cross-reactive titers are several logs lower than type-specific titers both vaccines have shown protection against HPV31 and other closely related non-vaccine types [20,21,47]. Previous studies have demonstrated that sera from HPV vaccinees recognize cross-reactive neutralizing epitopes on the DE- and FG-loops [16,17,20,47]. The five HPV31 cross-reactive antibodies we identified all required the C-terminal portion of the FG-loop for neutralization (Fig 2). Interestingly, there were four mAb of one clonotype that had similar HPV16 neutralization titers but variable levels of HPV31 cross-reactivity (Fig 2). Swapping heavy and light chains among these mAbs revealed that the light chain conferred most of the psV31 cross-neutralization activity (Fig 2). Among the 52 clonotypes described here, the five cross-reactive antibodies had a higher number of putative amino acid changes in the heavy and light chains (12.70 + /- 7.61) compared to the non-cross-reactive antibodies (9.48 + /- 4.90; p = 0.048) and in the clonotypic family, the cross-reactive mAbs had more of somatic mutations in their heavy and light chain variable regions than the non-cross-reactive mAbs implying that cross-reactive memory B cells were derived from non-cross-reactive precursors. This finding is reminiscent of the development of broadly neutralizing antibodies to influenza and HIV [48,49] that evolve from more type-specific precursor B cell [48,49].
The genetic characteristics of antibodies that target specific epitopes have been studied for other viruses [37,39,40,50–52]. This is the first study to identify recurrent features of HPV specific mAbs that correlated with patterns of neutralization. We found that IGHV2–70 frequently paired with the IGLV1–40 and those antibodies had characteristic mutations in the CDRL2 region (Fig 3C). Pairing of unique heavy and light chains has been observed among recurring antibodies following viral infection and vaccination [37,53–55]. Recurrent Ebola virus antibodies used IGLV1–40 in combination with IGHV3–15 [55]. Those antibodies also had a characteristic pattern of mutations in which the CDRL2 serine was substituted with asparagine or threonine, a pattern not observed here. Here we found the CDRL2 of antibodies that paired IGHV2–70/ IGLV1–40 was important for HPV16 neutralization. Substitutions with the predicted germline sequence in A7M08 resulted in an antibody with greatly reduced neutralization capacity. Even a single amino acid change from the germline CDRL2 sequence to the residue found in the mature mAb, improved neutralization activity more than two logs and even a conservative substitution (YGN to FGN) improved neutralization over one log compared to germline. Similar results were observed for four of the other six IGHV2–70/IGLV1–40 mAbs as swapping the mature CDRL2 with the germline CDRL2, increased their IC50s, albeit to a lesser extent than for A7M08 (S3 Table). Although, the psV16 neutralization activity of two mAbs (B25M02, D25M03) were not similarly affected by substitutions in the CDRL2, we showed that in the context of the germline light chain, amino acid substitutions in the CDRL2 significantly improved neutralization activity (S5 Fig). These results support the importance of the IGLV1–40 CDRL2 for HPV16 recognition and neutralization, as implied by the apparent convergent evolution for the sequences in this region however, cryo-EM revealed that IGLV1–40 contacted L1 with residues in CDRL1, CDRL3 and framework regions in addition to CDRL2 residues (S6 Fig).
The structures of the two antibodies with paired IGHV2–70/IGLV1–40, bound to capsomers revealed nearly identical footprints (Fig 4). It was difficult to discern which residues were required for neutralization for antibodies in this set using cpsV with, all but A7M08, able to neutralize the entire panel of cpsV in the first neutralization screen (Fig 3D). This likely resulted from these antibodies having similar overlapping epitopes dispersed across the five loops and our assay was not sensitive enough to detect minor disruptions in neutralization potency when only one loop had amino acid substitutions. The footprint of D24.1M01 and A7M08 overlaps the published footprints for other HPV16 monoclonal antibodies including H16.V5 [25–27]. D24.1M01 and H16.V5 both footprint onto identical residues on all five loops (P55, A139, N181, P182, V267, K278, N285, I348, T358) [26]. These antibodies have been described as binding to the top fringe of capsomer to multiple loops, primarily DE, FG and HI [25]. The majority of papillomavirus binding antibodies described previously share this pattern of binding, thus this appears to be the immunodominant region of the virus and could be described as a public epitope, although, given the diversity of antibodies that bind here, this region might expose several overlapping epitopes.
The other set of antibodies focused on here were antibodies using IGHV4–34. There were ten mAbs using IGHV4–34 representing eight clonotypes that paired with seven different IGLV but surprisingly, five paired with the same diversity region gene segment (D3-16*2, same reading frame). Because diversity gene segments encode part of the CDRH3 region of it was unsurprising to find a conserved sequence (WGSYR) in this region. Using alanine scanning we showed that this conserved sequence, in particular the tyrosine was important for neutralization activity (Fig 5). Diversity gene segments [56] have been found to be important for broadly neutralizing influenza antibodies and hydrophobic antibody residues were found to be important for those interactions [56]. Although all mAbs with WGSYR in the CDRH3 required the C-terminal half of the FG-loop for efficient neutralization, they did not recognize an identical epitope, with four requiring the DE-loop for neutralization, and two having significant cross-reactivity with HPV31. The cyro-EM structure of B25M05 was strikingly different from A7M08 and D24.1M01. Instead of five Fabs per capsomer, B25M05 bound one Fab per capsomer at the pentameric apex. This structure has been observed with other papillomavirus binding antibodies suggesting that this is also an immunodominant public epitope. Although it might be expected that B25.M01 would be less potent than D24.1M01 or A7M08 we found that all three were quite potent (IC50s = 3.0, 14.2, 3.3 respectively). However, the stoichiometry of antibody-psV binding is just one factor influencing potency, with others being antibody avidity and the mechanism of neutralization. Further, our structural study is limited to Fab binding to soluble capsomer and does not include how the antibodies would interact on the full capsid or viral particle. D24.1M01 and A7M08 bind to the top, distal fringe of the capsomer and could occlude the binding site on an adjacent capsomer which would lower the total valency on the capsid.
Our ability to identify the loops required for neutralization was limited by several factors. Firstly, we could not make mutations to all residues on the surface loops. Using cpsV to characterize neutralization epitopes relies on particles folding into native conformations (an approach also taken by Bissett et. al. [17]) but limits us to cpsV that retain infectivity. Thus, the BC-loop has only two substitutions and one deletion across the 11-residue loop compared with 10 substitutions across the 14 residue FGb-loop. This certainly reduced the sensitivity for detection of BC specific mAbs compared with FGb specific mAbs. Additionally, the C-terminal portion of DE- and HI-loops could not be mutated without impairing infectivity. The mouse mAb H16.V5 was tested in this assay and FG-loop was confirmed to be required for neutralization, but this method did not detect the HI-loop that is a known binding site of this antibody [13,28,31]. Other limitations included the fact that all mAbs were not used in titration experiments that were likely more sensitive and only 17 human mAbs were tested using cpsV with C-terminal substitutions. Additionally, deciding if an antibody was dependent on any particular loop was based on arbitrary thresholds Categorizing the mAbs using these thresholds was useful for clustering antibodies with similar neutralization profiles together helping to determine if antibodies sharing neutralization profiles also shared molecular features but perhaps at the expense of a more sensitive analysis of neutralization. In spite of these limitations the loops most frequently recognized by this set of mAbs was similar to the loops required for neutralization by whole serum from vaccinees [18].
In summary, we have shown that all L1 surface loops are targeted by human antibodies with the FG loop containing a dominant neutralizing epitope. Although this set of antibodies was modest in size, we found two types of antibodies that appeared to be over-represented and bound to the virus in strikingly different fashions. Importantly, both types of antibodies were found in each of the four participants in this study, suggesting that these may be commonly induced, perhaps public epitopes.
Materials and methods
Cell culture
Cells (293TT generously provided by John Schiller, National Cancer Institute, USA) were grown in DMEM (Thermo Fisher Scientific, Waltham MA) with 10% Bovine serum albumin (Corning, Corning, NY)) and supplements as previously described [57]. Pseudovirus were produced and purified as described by Buck et al [57] except that four T225 cm2 plates (Corning) were transfected, each with 57 μg DNAs (pSHELL plasmids encoding HPV L1 and L2 and pYSEAP plasmids) with 171μL TransIT-293 reagent (Mirus Bio LLC, Madison WI). All psV were purified on Optiprep gradients (Millipore Sigma Burlington, MA). Monoclonal antibody cloning and expression were previously described [58]
Mutagenesis of psV and mAb
Amino acid sequences of HPV16 L1 targeted for substitution were identified by alignment of HPV16L1 (accession AJ313179), HPV18L1 (accession WAB54476.1), HPV35 (accession WAB53999.1) and HPV2 L1 (accession QLM04846). Residues constituting loops were as defined by Chen et al [11]. P16sheLL and p31sheLL (accession EU127831) were obtained from Addgene (Cambridge, MA) and all mutagenesis was performed on the p16sheLL plasmid, using oligonucleotide primers (Integrated DNA Technology Inc (IDT), San Jose, CA) and In-Fusion cloning technology (Clontech Inc., Mountain View, CA). Sequence changes were verified by Sanger sequencing (Genewiz, South Plainfield, NJ). Initially all loops were substituted with HPV18 sequence but not all constructs produced infectious cpsV. Non-infectious cpsV were remade using HPV2 sequences, or for the c-terminal psV HPV35 sequences, and if those cpsV were non-infectious, the process was repeated substituting only a portion of each loop with HPV18 (loops DE, EF, FGa and HI) or HPV2 (BC and FGb) sequences until infectious cpsV were obtained with mutations on each of the five surface loops. Mutation of antibodies in the AbVec plasmids [59] were made using similar methods. Mutations to CDRL2 of IGLV1–40 in multiple antibodies and the CDRH3 of IGHV3–34 of D25M07 were created using the methods described above or using the Q5 Site-Directed Mutagenesis Kit (New England BioLabs, Ipswich, MA) with primers and sequencing performed as above. Designation of CDR regions and numbering was adopted from IGMT [60]. The entire IGLV1–40 sequence (accession M94116 with modifications to the CDRL3 region to match B25M02) was codon modified and synthesized (IDT) and cloned into Abvec-hIglambda (accession FJ517647.1) kindly provided by Patrick Wilson’s lab (University of Chicago).
Neutralization assays
To determine the optimal concentration of each cpsV, 3X105 293TT cells were plated in the wells of a 96 well flat bottom tissue culture plate (Corning) in 100 μl of medium. Approximately four hours later cpsV were serially diluted across a 96 well polypropylene plate (USA scientific, Orlando, FL) excluding the outer wells. After incubation at room temperature for one hour, 100 μl of diluted cpsV was used to infect the 293TT cells (adding media only to the last row) and transferred to a 37°C incubator (5% CO2) for three days. Supernatants (30 μl per well) were carefully removed and transferred to a 96 well ELISA plate (Fisher Scientific International, Hampton, NH). Secreted alkaline phosphatase was detected using phosphatase 104 substrate (Millipore Sigma) dissolved in 0.1M Carbonate buffer (pH9.6) with 10 mM MgCl2 (100 μl per well). After a 30-minute incubation in the dark, plates were read on a Synergy H4 plate reader (BioTek, Winooski, VT) at 405nm. Dilutions of cpsV that resulted in an absorbance of approximately 1.5 were determined to be the optimal for maximizing signal/noise ratios.
Primary mapping experiments were performed using a concentration of mAb 100-fold higher than the IC50 determined using psV16. Pseudovirus were diluted to the optimal concentration and mixed with diluted mAb in triplicate. The remainder of the assay was conducted as above. Background alkaline phosphatase was determined for each plate and was the average of six uninfected wells. Percent neutralization was calculated for each cpsV [1-((cpsV + mAb) – background)/ ((cpsV.-.mAb)-background)]. Repeat testing of 10 antibodies was highly reproducible with a median coefficient of variations (CVs) of 0.56% within test and 2.9% between tests.
Human mAbs that neutralized all cpsV more that 50% were retested in a secondary screen using one row of a 96 well plate for each mAb/psV combination. Antibodies were serially diluted 1:3 across a plate starting at a dilution 100-fold higher than the IC50 for psV16, mixed with cpsV with subsequent steps conducted as above. To compute IC50 values the percent neutralization values for each well were fit to a four-parameter logistic regression equation (4-PL) using GraphPad Prism7.0 (GraphPad Software Inc, San Diego, CA).
Neutralization assays using mutated mAbs began at 100 nM with 3-fold serial dilutions. Antibodies that did not titrate down to background were retested starting at a lower concentration (10 nM). Antibodies that failed to neutralize at 100 nM (<50% neutralization) were classified as non-neutralizing. All antibodies were tested a minimum of three times and the median CV for repeated testing was 14.9% within test and 20.5% between tests. Most experiments used purified mAb with known concentrations but for the neutralization experiments testing the c-terminal mutated psV some of the antibodies were cell culture supernatants. For those experiments antibody dilutions were used to calculate the IC50 rather than antibody concentrations.
Statistical analysis
Clonally related antibodies (clonotypes) were defined here as being derived from the same subject, having the same heavy and light chain variable regions (as assigned by IMGT-V Quest), with CDRH3 regions having the same length and greater than 70% amino acid sequence identity. All statistical analyses and heat maps were conducted on GraphPad Prism. Fishers’ exact tests were used to determine if mAb molecular features were associated with patterns of neutralization or other molecular features. Mathematic calculations were conducted on EXCEL 365 (Microsoft corporation, Redmond WA, USA). Kabat numbering: http://www.abysis.org/abysis/.
Sequence alignments
Reference amino acid sequences for IGHV4–34*01 (accession P06331.2), IGLV1–40 (accession M94116 and IGHV2–70 (accession P01814) were downloaded from IGMT https://www.imgt.org/IMGTindex/antibodies.php. Alignments were generated using ClustalW: https://www.genome.jp/tools-bin/clustalw.
Cloning and deletion construct for HPV16-L1
The full-length HPV16-L1 clone in pGEX-2T was used as parent clones for generating the deletion mutant. The L1 protein was expressed fused to GST protein tag in pET28b vector. We used a pair of forward and reverse primers to amplify the clone using the standard PCR method, except for the deleted regions. The regions deleted were N-terminal 9 residues, C-terminal 31 residues, and helix 4 (residues 404– 436), known to be required for HPV16 L1 virus particle assembly (15). The PCR fragment with NdeI and BamHI restriction sites incorporated into both primers was digested with the same enzyme for ligation. DNA sequencing was used to confirm the deletion mutant.
HPV16L1 protein purification
The expression and purification of HPV16L1 protein was done as described previously [61]. Briefly, the E. coli (pLysS) cell culture was grown to the O.D. of 0.3 at 37 C, induced with 0.5 mM IPTG, and the protein was allowed to overexpress overnight at room temperature. The cells were pelleted and resuspended in lysis buffer having 50 mM Tris-HCl, pH 8.0, 0.2 M NaCl, 1 mM dithiothreitol (DTT), 1 mM EDTA, and 10 mM phenylmethylsulfonyl fluoride. The cells were lysed by a sonication cycle of 30sec on and 30sec off for 30 min.
Further, 5mM MgCl2 and 1mM ATP was added to the lysate and incubated for 1hr at room temperature on rocker. The lysate was centrifuged for 1hr at 10000 rpm and the supernatant was passed through GST bead column by gravity flow. The GST beads were washed with lysis buffer to remove the contaminating proteins and the L1 protein was eluted with lysis buffer having 10mM glutathione. The eluted L1 protein was treated overnight at 4°C with thrombin enzyme (Sigma; T6634) to remove the GST tag. The cleaved L1 protein further passed through GST beads to remove the GST protein and to get a pure L1 protein. The eluted L1 protein was purified by HiLoad 16/600 Superdex 200 pg column.
Expression and purification of Fabs
Fabs were expressed in 500ml of HEK293-EBNA cells (RRID:CVCL_HF20) at a density of 1 x 106 cells/mL and transfected with 125 μg each of His-tagged heavy chains and light chains using 1 mg of PEI (1:4 DNA-to-PEI ratio). Cultural supernatant was harvested 6 days after transfection. Fabs were purified using Ni-NTA resin (Takara) and eluted using 5mM HEPES, pH 7.5, 310 mM NaCl, and 250mM imidazole. The fabs were further purified using HiLoad 16/600 Superdex 200 pg column.
Cryo-EM of complexes
The HPV16L1-D24.1M01 Fab complex was prepared by mixing HPV16L1 with Fabs in a protomer:Fab ratio of 1:1.5 and then incubated at room temperature for 2 h. The incubated complex was further purified using HiLoad 16/600 Superdex 200 pg column. Aliquots of the purified complexes were loaded and vitrified on Ultrathin Carbon Film on Lacey Carbon Support Film (400 mesh, Cu), (Tedpella) in an FEI Mark IV Vitroblot. 2,332 movies were acquired using Titan Krios microscope (ThermoFisher) at a voltage of 300 keV equipped with Falcon 3EC direct electron detector (ThermoFisher) at counting mode at 96,000x (1.054 Å pixel size). Automated data collection was done using EPU (ThermoFisher) varied the defocus range between −0.8 to −2.2 μm. All movies were collected using an exposure time of approximately 5 s for a total dose of 50 e − /Å2 fractionated over 50 frames.
The HPV16L1-A7M08Fab complex was prepared the same way as the D24.1M01 complex. Sample was loaded at a concentration of 0.25 mg/mL with 0.04% Dodecyl-β-D-maltoside (DDM) onto UltrAuFoil R2/2 on 200 mesh (Ted Pella) using a Vitrobot Mark IV (ThermoFIsher). Grids were blotted for 6 s with a blot force of 2 at 4 °C and 100% humidity. Data was collected on a 200 kV Glacios microscope (ThermoFisher) with a K3 direct electron detector (Gatan) at a magnification of 36,000x (1.122 Å per pixel). Initial analysis showed orientation bias at 0° so a total of 2,372 movies were collected, 560 at 0°, 1,237 at 30°, and 575 at 45° tilt using SerialEM [62] for automated collection with a defocus range of -1.5 to -2.2 μm with exposures over 6 s.
The HPV16L1-B25M05Fab complex was prepared in the same way as the D24.1M01 complex. Sample was loaded at a concentration of 0.6 mg/mL with 0.05% DDM onto UltrAuFoil R2/2 on 200 mesh (Ted Pella) using a Vitrobot Mark IV (ThermoFIsher). Grids were blotted for 4 s with a blot force of 2 at 4 °C and 100% humidity. A total of 1,079 movies were collected without tilt on a 200 kV Glacios microscope (ThermoFisher) with a K3 direct electron detector (Gatan) at a magnification of 36,000x (1.122 Å per pixel) using SerialEM [62] for automated collection with a defocus range of -1.8 to -2.2 μm with 100 exposures over 6 s.
Cryo-EM Data processing
For the D24.1M01 data, alignment and CTF estimation of aligned micrographs were carried out with Motioncor2 [63]. Processed micrographs were imported into cryoSPARC (v3.1) [64]. Particles were selected using blob picker and extracted with a box size of 220 px. Following two rounds of 2D classification (150 classes and 100 classes), there were 258,311 particles. Two ab-initio reconstructions were generated and following heterogeneous refinement with C5 symmetry, one showed much higher resolution (3.55 Å and 5.35 Å). The higher resolution reconstruction was further processed using Local CTF refinement and non-uniform refinement with 188,743 particles. The final 3D density map resolution was estimated based on the gold-standard FSC (GSFSC) curve with a cut-off of 0.143 (GSFSC0.143) to 3.0 Å. Information for the 3D reconstruction can be found in S4 Table and S6 Fig.
For the A7M08 data, motion correction was performed using WARP and further processed using cryoSPARC (v4.4) [64]. Micrographs were curated down to a total of 1,584 micrographs, 449 at 0°, 937 at 30°, and 198 at 45°. Template picker was used to pick particles using the map of the D24.1M01 with HPV16 L1 to generate templates and particles were extracted using a box size of 320 px. Following three rounds of 2D classification (150 classes, 100 classes, and 50 classes, there were 989,215 particles. A single ab-initio reconstruction was generated and 3D classing was used to further select particles. From this, 301,713 particles were used to generate the final reconstruction to a GSFSC0.143 resolution of 2.69 Å. Information for the 3D reconstruction can be found in S4 Table and S6 Fig.
For the B25M05 data, motion correction was performed using WARP and processed using cryoSPARC (V4.4) [64]. Micrographs were curated down to 753 and particles were picked using blob pickers and extracted with a box size of 320 px. A single ab-initio reconstruction was generated but refinement failed to give a high-resolution map. Particles from the best homogeneous refinement were symmetry expanded with C5 symmetry and 3D classified into 5 classes. The best refined map was moved forward and duplicate particles were removed. The final reconstruction was generated using 83,711 particles with C1 symmetry to a final reconstruction GSFSC0.143 resolution of 2.90 Å. Information for the 3D reconstruction can be found in S4 Table and S6 Fig.
Model building and refinement
For all structures, the model of the soluble HPV16 L1 capsomer from PDB ID pdb_00005w1o [65] was fit into the sub-particle reconstruction density maps using ChimeraX [66]. Alphafold2 [67] was used to generate a model of the Fv domain of D24.1M01, A7M08, and B25M05. For D24.1M01 and A7M08 bound structures, the model was fit into the reconstruction using ChimeraX at one binding site. The models were then corrected and adjusted manually by Isolde [68] in ChimeraX and COOT [69] and automatically refined using real_space_refinement in Phenix [70]. Once the D24.1M01 Fv or A7M08 Fv and HPV16 L1 was well fit at one binding site the others were generated using symmetry to place the remaining Fvs and L1 protomers. For B25M05 bound structure, no symmetry was used in the final 3D reconstruction of the map so all HPV16 L1 protomers were refined individually. The model statistics is summarized in S5 Table. PISA server (www.ebi.ac.uk/pdbe/pisa) was used to analyze the HPV16 L1-Fab interaction and calculate the BSA of the interface.
Highlights
- Most human mAbs recognized L1 surface loops but three of 68 recognized sequences on the C-terminal arm of L1Some antibodies induced by HPV vaccination follow shared developmental pathways.
- Human monoclonal antibodies using IGHV2–70/IGLV1–40 were found in all participants and bound at a stoichiometry of five Fabs per capsomer.
- Human monoclonal antibodies using the diversity gene segment D3-16*02 were found in all participants and one Fab was shown to bind with a stoichiometry of one Fab per capsomer.
Supporting information
S1 Table. Cross reference for antibodies with previous nomenclature.
Antibodies in this manuscript (column 1) are cross referenced to identifiers used in Scherer et. al. 2018 [22] (column 2). The first letter in the current ID represents the study subject, the following number is the month at which the sample was collected. If the number is a decimal it was collected one week after the vaccine dose. The letter M or P indicates the source was a memory B cells or plasmablast and last two digits were added to create unique identifier. The neutralization titer (IC50) units are picomolar (pM).
https://doi.org/10.1371/journal.ppat.1013086.s001
(PDF)
S2 Table. Screen for determining which loops were required for neutralization.
These are the same data shown in Fig 1B however in this figure the percent neutralization is shown. Antibodies were tested in triplicate in neutralization assays against six cpsV, psV16 and psv31 using a concentration of antibody of 100-fold higher than their IC50 versus psV16. Each cpsV had amino acid substitutions on one of the five surface loops of HPV16 as indicated in Fig 1A. The value is percent neutralization: setting psV with no antibody equal to zero and no psV equal to 100%. Clustering was performed manually keep similar neutralization profiles and clonotypes together. Clonotypes are indicated by shading in column 2 and 3.
https://doi.org/10.1371/journal.ppat.1013086.s002
(PDF)
S3 Table. Measuring the importance of the CDRL2 region for neutralization of psV16.
The germline CDRL2 sequence was cloned into the light chains for each mAb listed. The modified light chains were coexpressed with their cognate heavy chain and used in neutralization assays with psV16. The fold differences are compared with results using the mature light chain for each (see sequence Fig 3C).
https://doi.org/10.1371/journal.ppat.1013086.s003
(PDF)
S4 Table. Cryo-EM data collection and refinement statistics.
Data collection information for the cryo-EM data and statistics from refinement of the models. MolProbity [71] and EMRinger [72] were used for analysis of the quality of the final model.
https://doi.org/10.1371/journal.ppat.1013086.s004
(PDF)
S1 Fig. Most HPV16 binding antibodies neutralize.
148 hmAb, including the 68 mAb included in this study were tested for HPV16 neutralization and binding. Antibodies that failed to neutralize were given an arbitrary IC50 of 100 nM. There were four mAb (circled) that had significant binding but failed to neutralize. The p value is from linear regression, indicating that the slope of the line is statistically significantly not equal to zero.
https://doi.org/10.1371/journal.ppat.1013086.s005
(PDF)
S2 Fig. Human mAbs potently neutralize HPV16 in vitro.
A histogram showing the distribution of HPV16 IC50 neutralization values for the 68 hmAbs used in this study.
https://doi.org/10.1371/journal.ppat.1013086.s006
(PDF)
S3 Fig. Number of mAbs that required each HPV16L1 surface loop for each subject.
The number of mAb that required a particular loop (indicated on the horizontal axis), multiple loops or the c-terminus in either the primary or secondary tests is shown for each of four subjects (different colored bars).
https://doi.org/10.1371/journal.ppat.1013086.s007
(PDF)
S4 Fig. Neutralization profile for antibodies using the IGKV1–39 light chain.
List of the percent neutralization for each mAb (left column) versus each cpsV (indicated on top) in the first screen (A) and the IGHVs for all mAb that used IGKV1–39. B) Results in fold difference in titer compared with psV16 for mAbs vs each cpsV (top). The fold difference titer vs cpsV with C-terminal amino acid substitutions is on the right, tested separately. A value of 10 was assigned when a mAb failed to neutralize at least 50%. These are the same data as in Fig 1 and S2 Table.
https://doi.org/10.1371/journal.ppat.1013086.s008
(PDF)
S5 Fig. Importance of CDRL2 for B25M02 neutralization.
B25M02 light chain with just the CDRL2 mutated – see sequences – panel A. B25M02 heavy chain was coexpressed with either the fully mature B25M02 light chain, fully germline IGLVl1-40 light chain or the germline light chain with the mature (HVF) CDRL2. Each data point (B) is an IC50 vs psV16, determined from a titration curve. *** = p < 0.001. ns = not significant. GL = germline seq.
https://doi.org/10.1371/journal.ppat.1013086.s009
(PDF)
S6 Fig. Data processing of cryo-EM data.
Cryo-EM reconstructions of D24.1M01 Fab bound to HPV16L1 (left), A7M08 Fab bound to HPV16L1 (middle), and B25M05 Fab bound to HPV16L1 (right). Representative micrographs, representative 2D class averages, final resolved map colored based on local resolution estimation, and FSC plot with resolution estimation based on GSFSC0.143, and view direction distribution are shown. A quality of fit image for each Fab at the binding interface with HPV16 L1 is shown. Only the cryo-EM reconstruction within 3 Å of the protein is shown for clarity.
https://doi.org/10.1371/journal.ppat.1013086.s010
(PDF)
S7 Fig. BSA of mAbs D24.1M01, A7M08, and B25M05 binding to HPV16 L1.
Buried surface area (BSA) of each mAb amino acid upon binding with HPV16 L1. Values were determined using the PDBePISA server [73]. Interactions with protomer 1 are outlined in gray and interactions with protomer 2 are outlined in black.
https://doi.org/10.1371/journal.ppat.1013086.s011
(PDF)
S8 Fig. BSA of HPV16 L1.
Buried surface area (BSA) of the amino acids of HPV L1 upon binding by D24.1M01 or B25M05. Values were determined using the PDBePISA server [73]. Heavy and light chain interactions are differentiated in dark and light colors. Amino acids from protomer 1 are solid colors and amino acids from protomer 2 have a checkered pattern.
https://doi.org/10.1371/journal.ppat.1013086.s012
(PDF)
S9 Fig. Importance of disulfide bond linking D24.1M01 CDRH1 and CDRH3 and movement of DE-loop in B25M05 binding.
A) Cartoon diagram of the overlayed D24.1M01 and A7M08 structures. D24.1M01 has a disulfide bond linking the CDRH1 and CDRH3 but in A7M08 the CDRH3 contains a tyrosine. B) Cartoon diagram showing the movement of the DE-loop between unbound and B25M05 bound L1 protomers. Only the DE-loop undergoes rearrangement the FG-loop is unchanged upon binding.
https://doi.org/10.1371/journal.ppat.1013086.s013
(PDF)
Acknowledgments
We thank Duncan Ralf and Eric Matsen at the Fred Hutchinson Cancer Center for helpful discussions. Harry Scott provided assistance at the Pacific Northwest Center for Cryo-EM (PNCC).
References
- 1. Bosch FX, Lorincz A, Muñoz N, Meijer CJLM, Shah KV. The causal relation between human papillomavirus and cervical cancer. J Clin Pathol. 2002;55(4):244–65. pmid:11919208
- 2. de Martel C, Plummer M, Vignat J, Franceschi S. Worldwide burden of cancer attributable to HPV by site, country and HPV type. Int J Cancer. 2017;141(4):664–70. pmid:28369882
- 3. Gillison ML, Chaturvedi AK, Anderson WF, Fakhry C. Epidemiology of Human Papillomavirus-Positive Head and Neck Squamous Cell Carcinoma. J Clin Oncol. 2015;33(29):3235–42. pmid:26351338
- 4. Clifford GM, Smith JS, Plummer M, Muñoz N, Franceschi S. Human papillomavirus types in invasive cervical cancer worldwide: a meta-analysis. Br J Cancer. 2003;88(1):63–73. pmid:12556961
- 5. Porras C, Tsang SH, Herrero R, Guillén D, Darragh TM, Stoler MH, et al. Efficacy of the bivalent HPV vaccine against HPV 16/18-associated precancer: long-term follow-up results from the Costa Rica Vaccine Trial. Lancet Oncol. 2020;21(12):1643–52. pmid:33271093
- 6. Harper DM, DeMars LR. HPV vaccines - A review of the first decade. Gynecol Oncol. 2017;146(1):196–204. pmid:28442134
- 7.
World Health Organization. Human papillomavirus vaccines: WHO position paper. Geneva, Switzerland: World Health Organization. 2014.
- 8. Anderer S. WHO Approves Another HPV Vaccine for Single-Dose Use. JAMA. 2024. pmid:39514220
- 9. Whitworth HS, Mounier-Jack S, Choi EM, Gallagher KE, Howard N, Kelly H, et al. Efficacy and immunogenicity of a single dose of human papillomavirus vaccine compared to multidose vaccination regimens or no vaccination: An updated systematic review of evidence from clinical trials. Vaccine X. 2024;19:100486. pmid:38873638
- 10. Markowitz LE, Drolet M, Lewis RM, Lemieux-Mellouki P, Pérez N, Jit M, et al. Human papillomavirus vaccine effectiveness by number of doses: Updated systematic review of data from national immunization programs. Vaccine. 2022;40(37):5413–32. pmid:35965239
- 11. Chen XS, Garcea RL, Goldberg I, Casini G, Harrison SC. Structure of small virus-like particles assembled from the L1 protein of human papillomavirus 16. Mol Cell. 2000;5(3):557–67. pmid:10882140
- 12. Modis Y, Trus BL, Harrison SC. Atomic model of the papillomavirus capsid. EMBO J. 2002;21(18):4754–62. pmid:12234916
- 13. Carter JJ, Wipf GC, Benki SF, Christensen ND, Galloway DA. Identification of a human papillomavirus type 16-specific epitope on the C-terminal arm of the major capsid protein L1. J Virol. 2003;77(21):11625–32. pmid:14557648
- 14. Guan J, Bywaters SM, Brendle SA, Lee H, Ashley RE, Christensen ND, et al. The U4 Antibody Epitope on Human Papillomavirus 16 Identified by Cryo-electron Microscopy. J Virol. 2015;89(23):12108–17. pmid:26401038
- 15. Li Z, Yan X, Yu H, Wang D, Song S, Li Y, et al. The C-Terminal Arm of the Human Papillomavirus Major Capsid Protein Is Immunogenic and Involved in Virus-Host Interaction. Structure. 2016;24(6):874–85. pmid:27276427
- 16. Fleury MJJ, Touzé A, Maurel M-C, Moreau T, Coursaget P. Identification of neutralizing conformational epitopes on the human papillomavirus type 31 major capsid protein and functional implications. Protein Sci. 2009;18(7):1425–38. pmid:19533761
- 17. Bissett SL, Godi A, Beddows S. The DE and FG loops of the HPV major capsid protein contribute to the epitopes of vaccine-induced cross-neutralising antibodies. Sci Rep. 2016;6:39730. pmid:28004837
- 18. Carter JJ, Wipf GC, Madeleine MM, Schwartz SM, Koutsky LA, Galloway DA. Identification of human papillomavirus type 16 L1 surface loops required for neutralization by human sera. J Virol. 2006;80(10):4664–72. pmid:16641259
- 19. Godi A, Bissett SL, Miller E, Beddows S. Relationship between Humoral Immune Responses against HPV16, HPV18, HPV31 and HPV45 in 12-15 Year Old Girls Receiving Cervarix® or Gardasil® Vaccine. PLoS One. 2015;10(10):e0140926. pmid:26495976
- 20. Harari A, Chen Z, Rodríguez AC, Hildesheim A, Porras C, Herrero R, et al. Cross-protection of the Bivalent Human Papillomavirus (HPV) Vaccine Against Variants of Genetically Related High-Risk HPV Infections. J Infect Dis. 2016;213(6):939–47. pmid:26518044
- 21. Wheeler CM, Kjaer SK, Sigurdsson K, Iversen OE, Hernandez-Avila M, Perez G, et al. The impact of quadrivalent human papillomavirus (HPV; types 6, 11, 16, and 18) L1 virus-like particle vaccine on infection and disease due to oncogenic nonvaccine HPV types in sexually active women aged 16-26 years. J Infect Dis. 2009;199(7):936–44.
- 22. Brown DR, Kjaer SK, Sigurdsson K, Iversen OE, Hernandez-Avila M, Wheeler CM. The impact of quadrivalent human papillomavirus (HPV; types 6, 11, 16, and 18) L1 virus-like particle vaccine on infection and disease due to oncogenic nonvaccine HPV types in generally HPV-naive women aged 16-26 years. J Infect Dis. 2009;199(7):926–35.
- 23. Bissett SL, Godi A, Jit M, Beddows S. Seropositivity to non-vaccine incorporated genotypes induced by the bivalent and quadrivalent HPV vaccines: A systematic review and meta-analysis. Vaccine. 2017;35(32):3922–9. pmid:28633892
- 24. Huang W, He M, Ning T, Nie J, Zhang F, Zheng Q, et al. Structural characterization of a neutralizing mAb H16.001, a potent candidate for a common potency assay for various HPV16 VLPs. NPJ Vaccines. 2020;5:89. pmid:33042588
- 25. He M, Chi X, Zha Z, Li Y, Chen J, Huang Y, et al. Structural basis for the shared neutralization mechanism of three classes of human papillomavirus type 58 antibodies with disparate modes of binding. J Virol. 2021;95(7):e01587-20. pmid:33472937
- 26. Guan J, Bywaters SM, Brendle SA, Ashley RE, Makhov AM, Conway JF, et al. High-Resolution Structure Analysis of Antibody V5 and U4 Conformational Epitopes on Human Papillomavirus 16. Viruses. 2017;9(12):374. pmid:29211035
- 27. Li Z, Wang D, Gu Y, Song S, He M, Shi J, et al. Crystal Structures of Two Immune Complexes Identify Determinants for Viral Infectivity and Type-Specific Neutralization of Human Papillomavirus. mBio. 2017;8(5):e00787-17. pmid:28951471
- 28. Christensen ND, Cladel NM, Reed CA, Budgeon LR, Embers ME, Skulsky DM, et al. Hybrid papillomavirus L1 molecules assemble into virus-like particles that reconstitute conformational epitopes and induce neutralizing antibodies to distinct HPV types. Virology. 2001;291(2):324–34. pmid:11878901
- 29. Zhao Q, Potter CS, Carragher B, Lander G, Sworen J, Towne V, et al. Characterization of virus-like particles in GARDASIL® by cryo transmission electron microscopy. Hum Vaccin Immunother. 2014;10(3):734–9. pmid:24299977
- 30. Liu X, Chen J, Wang Z, Wang D, He M, Qian C, et al. Neutralization sites of human papillomavirus-6 relate to virus attachment and entry phase in viral infection. Emerg Microbes Infect. 2019;8(1):1721–33. pmid:31769733
- 31. Lee H, Brendle SA, Bywaters SM, Guan J, Ashley RE, Yoder JD, et al. A cryo-electron microscopy study identifies the complete H16.V5 epitope and reveals global conformational changes initiated by binding of the neutralizing antibody fragment. J Virol. 2015;89(2):1428–38. pmid:25392224
- 32. Guan J, Bywaters SM, Brendle SA, Lee H, Ashley RE, Makhov AM, et al. Structural comparison of four different antibodies interacting with human papillomavirus 16 and mechanisms of neutralization. Virology. 2015;483:253–63. pmid:25996608
- 33. Li Z, Song S, He M, Wang D, Shi J, Liu X, et al. Rational design of a triple-type human papillomavirus vaccine by compromising viral-type specificity. Nat Commun. 2018;9(1):5360. pmid:30560935
- 34. Liljeroos L, Malito E, Ferlenghi I, Bottomley MJ. Structural and Computational Biology in the Design of Immunogenic Vaccine Antigens. J Immunol Res. 2015;2015:156241. pmid:26526043
- 35. Pantaleo G, Correia B, Fenwick C, Joo VS, Perez L. Antibodies to combat viral infections: development strategies and progress. Nat Rev Drug Discov. 2022;21(9):676–96. pmid:35725925
- 36. Barnes CO, West AP Jr, Huey-Tubman KE, Hoffmann MAG, Sharaf NG, Hoffman PR, et al. Structures of human antibodies bound to SARS-CoV-2 spike reveal common epitopes and recurrent features of antibodies. bioRxiv. 2020. pmid:32577645
- 37.
Robbiani DF, Gaebler C, Muecksch F, Lorenzi JCC, Wang Z, Cho A, et al. Convergent Antibody Responses to SARS-CoV-2 Infection in Convalescent Individuals. bioRxiv. 2020. https://doi.org/10.1101/2020.05.13.092619 32511384
- 38. Galson JD, Pollard AJ, Trück J, Kelly DF. Studying the antibody repertoire after vaccination: practical applications. Trends Immunol. 2014;35(7):319–31. pmid:24856924
- 39. Schmidt AG, Therkelsen MD, Stewart S, Kepler TB, Liao H-X, Moody MA, et al. Viral receptor-binding site antibodies with diverse germline origins. Cell. 2015;161(5):1026–34. pmid:25959776
- 40. Graham BS. Advances in antiviral vaccine development. Immunol Rev. 2013;255(1):230–42. pmid:23947359
- 41. Xia L, Xian Y, Wang D, Chen Y, Huang X, Bi X, et al. A human monoclonal antibody against HPV16 recognizes an immunodominant and neutralizing epitope partially overlapping with that of H16.V5. Sci Rep. 2016;6:19042. pmid:26750243
- 42. Scherer EM, Smith RA, Gallego DF, Carter JJ, Wipf GC, Hoyos M, et al. A Single Human Papillomavirus Vaccine Dose Improves B Cell Memory in Previously Infected Subjects. EBioMedicine. 2016;10:55–64. pmid:27423190
- 43. Scherer EM, Smith RA, Carter JJ, Wipf GC, Gallego DF, Stern M, et al. Analysis of Memory B-Cell Responses Reveals Suboptimal Dosing Schedule of a Licensed Vaccine. J Infect Dis. 2018;217(4):572–80. pmid:29186468
- 44. Wang X, Wang Z, Christensen ND, Dillner J. Mapping of human serum-reactive epitopes in virus-like particles of human papillomavirus types 16 and 11. Virology. 2003;311(1):213–21. pmid:12832218
- 45. Carpentier GS, Fleury MJJ, Touzé A, Sadeyen J-R, Tourne S, Sizaret P-Y, et al. Mutations on the FG surface loop of human papillomavirus type 16 major capsid protein affect recognition by both type-specific neutralizing antibodies and cross-reactive antibodies. J Med Virol. 2005;77(4):558–65. pmid:16254978
- 46. Schiller JT, Castellsagué X, Garland SM. A review of clinical trials of human papillomavirus prophylactic vaccines. Vaccine. 2012;30 Suppl 5(0 5):F123-38. pmid:23199956
- 47. Barzon L, Squarzon L, Masiero S, Pacenti M, Marcati G, Mantelli B, et al. Neutralizing and cross-neutralizing antibody titres induced by bivalent and quadrivalent human papillomavirus vaccines in the target population of organized vaccination programmes. Vaccine. 2014;32(41):5357–62. pmid:25045814
- 48. Verkoczy L, Kelsoe G, Moody MA, Haynes BF. Role of immune mechanisms in induction of HIV-1 broadly neutralizing antibodies. Curr Opin Immunol. 2011;23(3):383–90. pmid:21524897
- 49. Gerhard W, Yewdell J, Frankel ME, Webster R. Antigenic structure of influenza virus haemagglutinin defined by hybridoma antibodies. Nature. 1981;290(5808):713–7. pmid:6163993
- 50. Chen F, Tzarum N, Wilson IA, Law M. VH1-69 antiviral broadly neutralizing antibodies: genetics, structures, and relevance to rational vaccine design. Curr Opin Virol. 2019;34:149–59. pmid:30884330
- 51. Corti D, Bianchi S, Vanzetta F, Minola A, Perez L, Agatic G, et al. Cross-neutralization of four paramyxoviruses by a human monoclonal antibody. Nature. 2013;501(7467):439–43. pmid:23955151
- 52. Yu L, Guan Y. Immunologic Basis for Long HCDR3s in Broadly Neutralizing Antibodies Against HIV-1. Front Immunol. 2014;5:250. pmid:24917864
- 53. Scheid JF, Mouquet H, Ueberheide B, Diskin R, Klein F, Oliveira TYK, et al. Sequence and structural convergence of broad and potent HIV antibodies that mimic CD4 binding. Science. 2011;333(6049):1633–7. pmid:21764753
- 54. Robbiani DF, Bozzacco L, Keeffe JR, Khouri R, Olsen PC, Gazumyan A, et al. Recurrent Potent Human Neutralizing Antibodies to Zika Virus in Brazil and Mexico. Cell. 2017;169(4):597-609.e11. pmid:28475892
- 55. Ehrhardt SA, Zehner M, Krähling V, Cohen-Dvashi H, Kreer C, Elad N, et al. Polyclonal and convergent antibody response to Ebola virus vaccine rVSV-ZEBOV. Nat Med. 2019;25(10):1589–600. pmid:31591605
- 56. Wu NC, Yamayoshi S, Ito M, Uraki R, Kawaoka Y, Wilson IA. Recurring and Adaptable Binding Motifs in Broadly Neutralizing Antibodies to Influenza Virus Are Encoded on the D3-9 Segment of the Ig Gene. Cell Host Microbe. 2018;24(4):569-578.e4. pmid:30308159
- 57. Buck CB, Pastrana DV, Lowy DR, Schiller JT. Generation of HPV pseudovirions using transfection and their use in neutralization assays. Methods Mol Med. 2005;119:445–62. pmid:16350417
- 58. Scherer EM, Smith RA, Simonich CA, Niyonzima N, Carter JJ, Galloway DA. Characteristics of memory B cells elicited by a highly efficacious HPV vaccine in subjects with no pre-existing immunity. PLoS Pathog. 2014;10(10):e1004461. pmid:25330199
- 59. Smith K, Garman L, Wrammert J, Zheng N-Y, Capra JD, Ahmed R, et al. Rapid generation of fully human monoclonal antibodies specific to a vaccinating antigen. Nat Protoc. 2009;4(3):372–84. pmid:19247287
- 60. Lefranc M-P, Pommié C, Ruiz M, Giudicelli V, Foulquier E, Truong L, et al. IMGT unique numbering for immunoglobulin and T cell receptor variable domains and Ig superfamily V-like domains. Dev Comp Immunol. 2003;27(1):55–77. pmid:12477501
- 61. Bang HB, Lee YH, Lee YJ, Jeong KJ. High-Level Production of Human Papillomavirus (HPV) Type 16 L1 in Escherichia coli. J Microbiol Biotechnol. 2016;26(2):356–63. pmid:26608168
- 62. Mastronarde DN. Automated electron microscope tomography using robust prediction of specimen movements. J Struct Biol. 2005;152(1):36–51. pmid:16182563
- 63. Zheng SQ, Palovcak E, Armache J-P, Verba KA, Cheng Y, Agard DA. MotionCor2: anisotropic correction of beam-induced motion for improved cryo-electron microscopy. Nat Methods. 2017;14(4):331–2. pmid:28250466
- 64. Punjani A, Rubinstein JL, Fleet DJ, Brubaker MA. cryoSPARC: algorithms for rapid unsupervised cryo-EM structure determination. Nat Methods. 2017;14(3):290–6. pmid:28165473
- 65. Dasgupta J, Bienkowska-Haba M, Ortega ME, Patel HD, Bodevin S, Spillmann D, et al. Structural basis of oligosaccharide receptor recognition by human papillomavirus. J Biol Chem. 2011;286(4):2617–24. pmid:21115492
- 66. Meng EC, Goddard TD, Pettersen EF, Couch GS, Pearson ZJ, Morris JH, et al. UCSF ChimeraX: Tools for structure building and analysis. Protein Sci. 2023;32(11):e4792. pmid:37774136
- 67. Jumper J, Evans R, Pritzel A, Green T, Figurnov M, Ronneberger O, et al. Highly accurate protein structure prediction with AlphaFold. Nature. 2021;596(7873):583–9. pmid:34265844
- 68. Croll TI. ISOLDE: a physically realistic environment for model building into low-resolution electron-density maps. Acta Crystallogr D Struct Biol. 2018;74(Pt 6):519–30. pmid:29872003
- 69. Emsley P, Lohkamp B, Scott WG, Cowtan K. Features and development of Coot. Acta Crystallogr D Biol Crystallogr. 2010;66(Pt 4):486–501. pmid:20383002
- 70. Liebschner D, Afonine PV, Baker ML, Bunkóczi G, Chen VB, Croll TI, et al. Macromolecular structure determination using X-rays, neutrons and electrons: recent developments in Phenix. Acta Crystallogr D Struct Biol. 2019;75(Pt 10):861–77. pmid:31588918
- 71. Williams CJ, Headd JJ, Moriarty NW, Prisant MG, Videau LL, Deis LN, et al. MolProbity: More and better reference data for improved all-atom structure validation. Protein Sci. 2018;27(1):293–315. pmid:29067766
- 72. Barad BA, Echols N, Wang RY-R, Cheng Y, DiMaio F, Adams PD, et al. EMRinger: side chain-directed model and map validation for 3D cryo-electron microscopy. Nat Methods. 2015;12(10):943–6. pmid:26280328
- 73. Krissinel E, Henrick K. Inference of macromolecular assemblies from crystalline state. J Mol Biol. 2007;372(3):774–97. pmid:17681537