(C) PLOS One [1]. This unaltered content originally appeared in journals.plosone.org.
Licensed under Creative Commons Attribution (CC BY) license.
url:https://journals.plos.org/plosone/s/licenses-and-copyright

------------



Structurally related but genetically unrelated antibody lineages converge on an immunodominant HIV-1 Env neutralizing determinant following trimer immunization

['Safia S. Aljedani', 'Fred Hutchinson Cancer Research Center', 'Vaccine', 'Infectious Disease Division', 'Seattle', 'Washington', 'United States Of America', 'Tyler J. Liban', 'Karen Tran', 'The Scripps Research Institute']

Date: 2021-10

All D11A antibodies share the same heavy chain germline VH4_3T_S3452 and JH4*01 genes, as well as the same light chain germline VL6-2*01_S6633 and JL2*01 genes ( Fig 1E ). Antibodies isolated from animal D15, D15.SF6 and D15.SD7 share the same heavy and light chain germline genes: IGH1_2L, JH4*01, IGLV6-2*01 and JL2*01 ( Fig 1E ). D19-isolated antibodies, D19.PA8 and D19.PD8, use different heavy chains germline genes, VH3_2T_S2563, JH5-2*02 and VH3_4A_S7053, JH4*01, respectively. D19.PA8 shares the same light chain germline genes IGLV6-2*01 and JL2*01 as the D15-isolated mAbs described above, while D19.PD8 uses the light chain germline genes lib5lambda_1 and JL1*01. VD16.2C10 uses the VH4_5H_S2253, JH4*02 or JH5*02, lib4kappa_12 and JK2*01 germline genes. D20-isolated mAbs are clonally related and use the VH4.11_S9546, JH 5–1*01_S8786, lib4kappa_12 and JK2*01 or JK4*01 germline genes [ 26 ] ( Fig 1E ). The somatic hyper mutation (SHM) levels in VH and VL range from 4.2–17.5% and 3.1–11.1% at the residue (aa) level, respectively ( Fig 1E ) [ 27 ].

We used single memory B cell sorting to isolate mAbs from animals that showed the highest serum neutralization: 3 mAbs from group A animal D11 (D11A.F2, D11A.B5, and D11A.F9, already reported elsewhere [ 25 ]), 3 mAbs from group B animal D15 (D15.SF6 and D15.SD7) and animal D16 (VD16.2C10); and 5 mAbs from group C animal D19 (D19.PA8 and D19.PD8) and animal D20 (VD20.1C7, VD20.1F9 and VD20.5A4) using 16055 NFL trimer probes. Neutralization against 16055 pseudovirus was assessed with potencies ranging from 0.005–~4 μg/ml ( Fig 1C and 1D ).

(A) Overview of the immunization and sampling of the rhesus macaques. (B) Immunization strategies and groups. (C) Example of B-cell sorting with 16055 NFL probes to identify mAbs with potent tier 2 autologous neutralization. (D) Neutralization activity (IC 50 ) of mAbs isolated from each group. (E). Antibodies isolated from various immunization trials and germline lineages. # , *, † Clonal variants.

We have shown previously that the well-ordered, stabilized NFL Env trimer [ 19 ] elicited HIV-1 autologous tier 2 neutralizing Abs in NHP [ 25 , 26 ]. Here, we analyzed mAbs isolated from different immunization strategies with variants of the NFL Env-stabilized trimer in Chinese rhesus macaques (Macaca mulatta) to better understand the specificity of the elicited immune response. The animals were immunized at week 0, 4 and 12 ( Fig 1A ). Group A was immunized with 16055 NFL trimers conjugated to liposomes [ 25 ], group B was immunized with soluble 16055 NFL trimers with glycans at N276, N301, N360 and N463 deleted (degly 4 (Δ276, Δ301, Δ360, and Δ463)) and group C was immunized with soluble 16055 degly 4 trimers at week 0 and 4 and boosted with the 16055 NFL with glycans restored (wild type, WT) at week 12 ( Fig 1B ). Two weeks after the third immunization, samples were collected. Plasma neutralization assays indicated that animals in all 3 groups developed tier 2 16055 autologous titers ( S1 Fig ).

Cross-competition binding analysis between the NHP neutralizing mAbs and known bNabs targeting different Env regions indicated that they all generally mapped to the V2 apical region of the 16055 NFL trimer while also displaying complete self- and cross-inhibition, assigning them to the same competition group ( S1B Fig ). Binding to 16055 gp120 constructs containing mutations in the V1/V2 loops confirmed specificity to the V2 region ( S1C Fig ). Epitope specificity was further mapped by neutralization sensitivity against a panel of 16055 pseudovirus mutants with residues along the 16055 V2 mini-loop (i.e., 182 VPLEEERKGN 187 ) mutated to alanine, or N187 mutated to glutamine ( S1D Fig ). The focused alanine scan confirmed dependence to the V2 hypervariable region as point mutants between residues V182 and K186C abrogated neutralization activity, while removal of the N187 glycan enhanced potency of the NHP mAbs ( S1D Fig ).

Since our high-resolution structures were solved with V2 peptide or V1V2 domain, we superimposed the above-described structures of mAb/V2b or V1V2 scaffold onto the structure of the 16055 NFL trimer (PDB ID: 5UM8) [ 19 ] by aligning the V2 or V1V2 region ( S3 Fig ). From this structural alignment, we would predict that some mAbs would have additional contacts with the trimer which were not observed in our structures, either because the residues were not present in the V2 peptide or V1V2 domains or because these residues were disordered or did not show interactions in the solved structure ( S3 Fig ). Interestingly, in the superposition, mAb VD20.5A4 did not show additional contacts to the 16055 NFL trimer.

In conclusion, the structural analyses support our previous alanine scanning results, which showed that Glu 185 , Glu 186 , Glu 186A , Arg 186B , and Lys 186C mutations resulted in decrease or loss of neutralizing activities of D11A.F2 [ 25 ]. Indeed, all mAbs interact with the above-mentioned V2 residues ( Fig 2 and S2 – S6 Tables ). We also observed additional interactions of all the mAbs with Val 182 , Pro 183 , Leu 184 , Gly 186D , and Asn 187 ( Fig 2G and S2 – S6 Tables ), with the light chains of D15.SD7 and D19.PA8 showing some contacts with the proximal N-acetylglucosamine (NAG) at residue Asn 187 ( Fig 2D and 2E ). We could not explain the slight difference in specificity at residues Pro 183 and Leu 184 described previously [ 26 ], which may highlight the importance of using both functional and structural analysis to provide a complete picture on the epitope and contact residues.

Similar to the D11A antibodies, D15.SD7, D19.PA8, and VD20.5A4 bind mostly the V2 hypervariable region. They bury ~ 824 Å 2 , ~ 615 Å 2 and ~522 Å 2 of the V1V2, respectively ( Fig 2D–2G ), of which ~744 Å 2 , ~603 Å 2 and ~ 480 Å 2 are in the hypervariable V2 region only.

Crystals of D15.SD7, D19.PA8 and VD20.5A4 with the scaffolded 16055 V1V2-1FD6 were obtained and diffracted X-rays to resolution of 2.8 Å, 2.0 Å, and 2.8 Å, respectively ( Figs 2D–2G and S1 , S4 – S6 Tables ). The V1V2 structure adopts the same conformation as seen in the 16055 NFL trimer (RMSD of 0.9 Å, 0.6 Å, and 0.8 Å over 44, 39, and 41 Cα atoms, respectively), confirming that these antibodies recognize an epitope elicited by the trimer and that likely no induced-fit conformational changes were induced by the antibodies binding to the scaffolded V1V2 compared to the trimer. Our structural analysis indicated that there are two copies in the asymmetric unit of D15.SD7/1FD6-V1V2 and D19.PA8/1FD6-V1V2 structures ( S4 and S5 Tables ). We observed clear density for three glycans in the gp120 V1V2 region at N156, N160 and N187 in one complex of D15.SD7/1FD6-V1V2, while the other complex in the asymmetric unit showed density for the N156 glycan only and thus chose the former for further analysis. Of note, D15.SD7 heavy chain showed some interactions with the 1FD6 scaffold ( S4 Table ), which we did not include in our analysis since they are not biologically relevant. Additionally, the 1FD6 scaffold was mostly disordered in the D19.PA8 and VD20.5A4 complex structures ( Fig 2 ). We also note that strand C was mostly disordered in the D19.PA8/1FD6-V1V2 complex.

Crystals of D11A.F2 and D11A.B5 were also obtained in complex with a 16055 V2 peptide, named here V2b peptide, 178 RLDIVPLEEERKGNSSKYRLINC 196 (numbering follows HXBc2 [ 33 ]), which diffracted X-rays to 2.8 Å and 2.0 Å resolution, respectively ( S1 – S3 Tables ). In both structures, the V2b peptide structure was fully resolved ( Fig 2B and 2C ) and adopted the same conformation as seen in the 16055 NFL trimer structure (RMSD of 1.1 Å and 0.8 Å over 17 and 16 Cα atoms, respectively) [ 18 ]. The high-resolution structures indicated that both D11A.F2 and D11A.B5 bind mainly to the 16055 V2 region, of which residues 185 EEER 186A appear unique to the 16055 strain ( Fig 2B and 2C ). The D11A.F2 antibody buries ~ 774 Å 2 of the V2b peptide, with ~701 Å 2 in the V2 region and ~73 Å 2 in Strand D [ 19 , 29 ] ( Figs 2B and S2 ). Similarly, D11A.B5 buries ~ 725 Å 2 of the V2b peptide, with ~674 Å 2 in the V2 region and ~51 Å 2 in Strand D ( Figs 2C and S3 ). We note that in both crystal structures, a shorter region of another V2b peptide appears to make additional interactions with D11A.F2 and D11A.B5 ( S2 Fig ), which we believe are crystallization artifacts. Since both the nsEM data and low-resolution crystal structure of D11A.F9 with 16055 NFL identified the hypervariable region V2 to be the epitope for D11A antibodies, we believe these additional contacts are not biologically relevant but the results of crystallization artifacts.

(A) nsEM 3D reconstruction with low resolution crystal structure of D11A.F9 Fab (Heavy chain, dark green; Light chain, light green) and 35022 scFv (gray) in complex with 16055 NFL (gp120, yellow; gp41, light brown) shown in two different views. ( B, C) Structures of D11A.F2 Fab (Heavy chain, sky blue; Light chain, cyan) and D11A.B5 Fab (Heavy chain, magenta; Light chain, light pink) bound to the V2b peptide (yellow). ( D) Structures of D15.SD7 (Heavy chain, blue: Light chain, light blue), (E) D19.PA8 (Heavy chain, orange; Light chain, light orange) and (F) VD20.5A4 (Heavy chain, raspberry; Light chain, light raspberry) Fabs in complex with the 16055 V1V2-1FD6 scaffold. ( B, C, D, E, F ) Interacting residues are shown in sticks and glycans in green. Pie charts summarize the buried surface area (BSA) of the V2b and V1V2-1FD6. ( G) Sequence of 16055 V1V2 highlighting the V2b peptide used for crystallization, the location of the V1, V2 and strands. Residues that contact the Mabs (within 5Å) are shown with asterisks underneath the sequence. N-linked glycosylation sites are shown in green.

The nsEM of 16055 NFL trimer in complex with D11A.F9 and 35022 Fab [ 30 ] confirmed that the D11A.F9 approached its epitope located at the apex of the HIV-1 trimer horizontally, or parallel to the viral membrane ( Fig 2A ), consistent with a previous study which showed that D11A.F9 bound V2 parallel to membrane [ 25 ]. D11A.F9 Fab crystals were obtained in complex with 16055 NFL trimer and 35022scFv [ 19 , 31 , 32 ], which diffracted X-ray to 6.5 Å. The low-resolution structure fitted well in the nsEM 3D reconstruction, confirming the horizontal angle of approach ( Fig 2A ).

To understand the molecular basis for the tier 2 autologous neutralization from the isolated mAbs, we used a combination of nsEM and X-ray crystallography. To increase our chances of obtaining structural information, we used variational crystallography [ 28 ] where antigen binding fragments (Fabs) from a select number of antibodies complexed with 16055 NFL trimer, a scaffolded 16055 V1V2-1FD6 [ 29 ] and a 16055 V2b peptide [ 25 ] were purified and used for crystallization. Below we described the high-resolution structures we obtained which are the focus of our analysis.

Polyclonal antibody response to a similar epitope

Since mAbs elicited from vaccination target the same V2 region, unique to 16055, but used diverse germline genes and their autologous neutralization potencies differed by almost a 1000-fold (IC 50 ranging from 0.005 μg/mL (VD20.5A4) to 3.68 μg/mL (D11A.B5)) (Fig 1D), we looked at differences and similarities of the paratope at the molecular level (Fig 3). We also analyzed the antibodies’ binding properties, including buried surface area (BSA), number of hydrogen bonds and salt bridges formed with the epitope, CDRH3 usage, electrostatics and angles of approach to decipher if some properties correlated with autologous neutralization potency (Figs 4–7).

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 3. Structural characterization of mAbs elicited from vaccination. Surface representations of (A) D11A.F2, (B) D11A.B5, (C) D15.SD7, (D) D19.PA8, and (E) VD20.5A4 Fabs. All mAbs are color coded as follows: CDR H1, chocolate; CDR H2, salmon; CDR H3, dark salmon; CDR L1, violet purple; CDR L2, deep purple and CDR L3, violet. The pie chart represents the relative contribution of each CDR loops to the total buried surface area of the paratope for each mAbs. (F) Sequence alignment of the mAbs to their germline genes with CDRs highlighted. Somatic hyper mutations (SHMs) are highlighted in red. Residues interacting with 16055 V2b peptide or 16055 V1V2 are shown as asterisks below the sequence (contact residues within 5 Å). https://doi.org/10.1371/journal.ppat.1009543.g003

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 4. MAbs binding properties and correlations with autologous neutralization potency. Correlation between autologous neutralization potency and (A) epitope surface area, (B) paratope surface area, (C) CDRH3 length, (D) relative contribution of the CDRH3 surface area in the paratope and (F) number of Hydrogen Bonds (HBs) and Salt Bridges (SBs). The lines indicate the fitted linear regression model with 95% confidence shown in shaded grey or color as indicated. The r2 and p values are displayed. (E) Graph indicates number of HBs and SBs between the epitope/paratope with the different mAbs. https://doi.org/10.1371/journal.ppat.1009543.g004

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 5. Electrostatics surface representation of 16055 NFL and different mAbs. Electrostatics surface representation of (A) 16055 NFL and (B) V1V2 and V2 region zooms (top and side views). Epitopes are highlighted by dotted lines and some residues in V2 are shown in stick and labeled. (C) Electrostatics surface representation of the mAbs paratope. Residues forming the paratope are shown in sticks. https://doi.org/10.1371/journal.ppat.1009543.g005

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 6. Vaccine-elicited mAbs target V2 region using a lateral approach with slightly different angles of approach. Side and top surface/cartoon representation of one gp120 (yellow)-gp41 (tan) 16055 NFL protomer with Fab bound: (A) D11A.B5 (pink), (B) D15.SD7 (blue), (C) D19.PA8 (orange) and (D) VD20.5A4 (raspberry) showing the angles of approach of each mAb. (E) Superimposition of gp120-gp41 16055 NFL protomer with D11A.B5 (pink), D15.SD7 (blue) and D19.PA8 (orange) onto VD20.5A4-bound gp120-gp41 protomer. (F) Summary of the mAbs angles of approach. https://doi.org/10.1371/journal.ppat.1009543.g006

PPT PowerPoint slide

PNG larger image

TIFF original image Download: Fig 7. NHP Autologous tier 2 neutralizing antibodies target a hole in the HIV-1 glycan shield. (A) Side and top view surface representation of 16055 NFL (PDB:5UM8) with gp120 shown in yellow, V2 region in grey, gp41 in wheat and glycans shown in green spheres or color-coded and labeled. Arrows indicate mAbs’ angle of approach. Epitopes targeted by the NHP mAbs are highlighted. (B) Side view and (C) Top view superpositions of the structures of D11A.B5, D15.SD7, D19.PA8 and VD20.5A4 onto the 16055 NFL trimer, showing how they access the glycan hole. Trimer and mAbs are shown in surface representation. Trimer is color coded as in (A) and mAbs as in Fig 3. (D) Effect of glycan removal surrounding the epitope on neutralization potency. Neutralization IC 50 values (μg/ml) shown with >10-fold differences highlighted in red. https://doi.org/10.1371/journal.ppat.1009543.g007

The total BSA of D11A.F2 is ~718 Å2, that of D11A.B5 is ~ 673 Å2, that of D15.SD7 is ~ 782 Å2, that of D19.PA8 is ~ 596 Å2 and ~ 557 Å2 of VD20.5A4 surface area is buried upon binding to its epitope (Fig 3A–3E). We did not observe a correlation between the BSA of the paratope or that of the epitope with neutralization potency (Fig 4A and 4B). Indeed, VD20.5A4 is the most potent mAb but showed the least amount of BSA upon binding its epitope, indicating that in this case, precise targeting with a smaller epitope footprint might be relevant to potency.

The mAbs use all six complementary determining regions (CDRs) to bind their epitope, except for VD20.5A4 which does not use the CDRL2, and D15.SD7 which does not use the CDRH2. D11A, D15.SD7 and D19.PA8 mAbs also use part of the framework regions although these account for less than 6% of the total BSA (Fig 3). Finally, both heavy and light chains are similarly involved in the interactions, except for D15.SD7 and VD20.5A4 which use primarily the heavy chain (57% and 78% of the total BSA of the paratope, respectively), with the CDRH3 accounting for 56% and 50% of the total BSA of the paratope and 98% and 64% of the heavy chain BSA, respectively (Fig 3C, 3E and 3F). The CDRH3 length varies from 11 to 20 residues but no correlation with potency was observed although D15.SD7 used primarily its 20-amino-acid CDRH3 to interact with its epitope (Fig 4C). We then assessed the correlation between potency and the relative contribution of the CDRH3 over the paratope (BSA from the CDRH3 over the total paratope BSA), and determined that there was a trend to significance correlation (Fig 4D). Indeed, it is interesting that the two mAbs that used most of their CDRH3 (in the context of our analysis, which only takes into account the V1V2 region and not the whole 16055 NFL trimer) proved to be the most potent autologous neutralizing mAbs.

All the mAbs used both germline and affinity matured V-gene residues in the interactions with their epitope, and within a clonally related family, some of the interacting residues differ, however it is unclear what the difference or role in the affinity maturation is regarding the overall potency (Fig 3F). We note that the D11A mAbs have an intradisulfide bond in the CDRH3, which appears to rigidify the loop causing it to be less involved in the interactions. Such disulfide bonds have been observed before in mAbs isolated in humans with HIV and HCV infections [34,35]. In these studies, the disulfide bonds were thought to be responsible for the antibodies’ neutralization potencies by stabilizing the affinity matured antibodies.

We next assessed the number of hydrogen bonds (HBs) and salt bridges (SBs) formed in each paratope/epitope interaction (Fig 4E). D11A.F2 and D11A.B5 form 8 and 10 HBs with the V2b peptide, 6 and 8 of which interact directly with the hypervariable V2 region, respectively. In addition, D11A.F2 and D11A.B5 form 16 and 12 SBs with the V2b peptide, 12 and 8 of which interact with the hypervariable V2 region, respectively. D15.SD7 and D19.PA8 form 10 and 7 HBs with their epitope, all of which interact with the hypervariable V2 region. Moreover, D15.SD7 and D19.PA8 form 10 SBs with their epitope, all of them with the hypervariable V2 region. VD20.5A4 forms 7 HBs (6 with the hypervariable V2) and 3 SBs with the hypervariable V2 region (Fig 4E). In conclusion, the number of HBs and SBs between the paratope/epitope did not correlate with the mAbs autologous neutralization potency (Fig 4F).

To further understand the differences in the potency, we looked at the electrostatics of the epitope and paratopes (Fig 5). While the epitope is overall positively charged (Fig 5A and 5B), the paratopes showed different electrostatics [36], with VD20.5A4 being strongly negatively charged towards the center of its paratope (Fig 5C). It appears that the paratope electrostatic of VD20.5A4 is more compatible with the overall positively charged epitope, which could explain its increased potency.

Finally, to understand the various mAbs’ angles of approach to their epitope on the 16055 NFL trimer, we used the superposition mentioned above of the bound structures of D11A.B5, D15.SD7, D19.PA8, and VD20.5A4 on 16055 NFL trimer by aligning the V2b region of each structure to the NFL trimer (PDB:5UM8) [19] and calculated their angles of approach from a side and top view (Fig 6). Our analysis suggests that D11A.B5, D15.SD7 and D19.PA8 mAbs approaches the V2 region with a similar angle (122–126°) from the side (lateral) (Fig 6A–6C, 6E and 6F) while VD20.5A4 approaches the 16055 NFL trimer slightly from above (~114°) and rotated compared to the other mAbs (Fig 6D–6F). While all mAbs approach their epitope with the same angle as seen from a top view, D19.PA8 is tilted 5° from the others (Fig 6C, 6E and 6F).

In conclusion, our structural analysis suggests that the difference in potency between the vaccine-elicited mAbs that bind the same epitope is likely due to differences in electrostatics in the paratope and angle of approach of the mAbs. Additionally, the nature of the CDRH3 interaction with the epitope also plays a role in the mAbs potency.

[END]

[1] Url: https://journals.plos.org/plospathogens/article?id=10.1371/journal.ppat.1009543

(C) Plos One. "Accelerating the publication of peer-reviewed science."
Licensed under Creative Commons Attribution (CC BY 4.0)
URL: https://creativecommons.org/licenses/by/4.0/


via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.