Identification of a CD4-Binding-Site Antibody to HIV that Evolved Near-Pan Neutralization Breadth

Abstract

Detailed studies of the broadly neutralizing antibodies (bNAbs) that underlie the best available examples of the humoral immune response to HIV are providing important information for the development of therapies and prophylaxis for HIV-1 infection. Here, we report a CD4-binding site (CD4bs) antibody, named N6, that potently neutralized 98% of HIV-1 isolates, including 16 of 20 that were resistant to other members of its class. N6 evolved a mode of recognition such that its binding was not impacted by the loss of individual contacts across the immunoglobulin heavy chain. In addition, structural analysis revealed that the orientation of N6 permitted it to avoid steric clashes with glycans, which is a common mechanism of resistance. Thus, an HIV-1-specific bNAb can achieve potent, near-pan neutralization of HIV-1, making it an attractive candidate for use in therapy and prophylaxis.

Keywords: CD4-binding site; HIV; antibody; envelope; immunotherapy; neutralizing; prophylaxis; resistance; structure; vaccine.

Published by Elsevier Inc.

Figures

Figure 1. N6 is a CD4bs Antibody with Extraordinary Neutralization Breadth and Potency

(A) Neutralization potency and breadth of N6, in comparison to other bNAbs, against a 181-isolate Env-pseudovirus panel. (B) ELISA binding of N6 to the indicated proteins. (C) Neutralization profile of N6, in comparison to other VRC01-class antibodies, against 20 VRC01-resistant pseudoviruses. Values JRCSF alanine mutants by N6, in comparison with other bNAbs. CD4-Ig and 2G12 were used as positive and negative controls respectively. The numbering of JRCSF mutants is based on the HXBC2 sequence. Neutralization fold change was calculated as follows: IC50 of the JRCSF mutant / IC50 of the JRCSF wild-type. Values >5 are highlighted in yellow. See also Figure S1 and Table S1.

Figure 2. N6 Recognition of HIV-1 gp120 Has Features in Common with the VRC01 Class

(A) Crystal structures of N6 in complex with gp120 proteins from three different HIV-1 clades that are sensitive or resistant to VRC01. The heavy chain is colored light green and the light chain is colored light blue. (B) Epitopes of N6 on HIV-1 clade AE 93TH057, clade C DU172, and clade G X2088 are colored light blue on the gray gp120 surfaces. The initial binding site of CD4 on the outer domain of gp120 is highlighted in yellow. (C) Interactions between HIV-1 gp120 and the complementarity determining regions of N6. HIV-1 gp120 is shown in surface representation in a 90°-view from (A). Key residues, such as heavy chain Tyr 54, Arg71 and Trp100c, light chain Gln96, are shown in sticks. (D) Typical interactions between VRC01-class antibodies and HIV-1 gp120 are present with N6. Shown to the left are the conserved salt bridges between Arg71 and CD4-binding site Asp368, and to the right, the hydrogen bond between CDR H3 Trp and loop D Asn279. The color scheme in both (C) and (D) is the same as in (A). See also Table S4.

Figure 3. N6 Avoids Steric Clashes Common to Other Members of the VRC01 Class

(A) Antibody orientation relative to CD4 when structurally aligned on gp120. N6 is rotated more relative to gp120-bound CD4, compared to other VRC01-class antibodies. (B) Unique position of N6 CDR L3 among the VRC01 class. Superposition of structures of gp120-VRC01-class antibodies indicated a unique position of N6 CDR L3. The color scheme is the same as in Figure 2A. (C) When aligned on gp120, the binding axis of N6, defined as the line connecting Cα atoms of heavy chain Arg71 and light chain Glu/Gln96, is rotated toward loop D and tilted away from variable loop V5, as compared to other VRC01-class antibodies. (D) Greater interaction of N6 with HIV-1 loop D, as compared to VRC27. When compared to VRC27 isolated from the same patient, N6 binding surface area on gp120 loop D increased ~20% and that on V5 and outer-domain-exiting loop decreased ~25% due to the rotation and tilting of the antibody. (E) N6 interaction with gp120 V5. Conserved hydrogen bond between Glu/Gln96LC and Gly459gp120 becomes a water-mediated bond in N6. The CDR H2 Gly60GlyGly62 and the rotated light chain N terminus make more room on either to accommodate variations in loop V5. (F) Comparison of the N6 paratope or that of VRC01 with morphology of a hand. The light chain N terminus and bulky side chains of CDR H2 residues 60–62 acts like an index finger and thumb grabbing HIV-1 variable loop V5. The rotation and tilting of light chain as well as the Gly60GlyGly62 mutation in the CDR H2 shortened the index finger and removed the thumb, leaving more space to accommodate variations in V5. (G) Structural basis for N6 neutralization of clade G strain X2088. Rotation and tilting of N6 resolved potential clashes with loop E insertion, and the retreated light chain N terminus and CDR H2 Gly60GlyGly62 made space to accommodate bulky residues and glycosylation sites at base of V5. When modeled in VRC01, CDR L1 had a severe clash with the loop E insertion, and its light chain N terminus and CDR H2 Arg61 created a narrow path that could not accommodate the X2088 V5. See also Figure S3.

Figure 4. N6 Tolerates Mutations in the CD4 BLP and V5

(A) Neutralization of N6 against three N6-resistant and the Z258 autologous pseudoviruses. Neutralization of these pseudoviruses with reverse mutations in loop D, CD4 BLP, and V5 region are also shown. Sequence variations of gp120, as compared to reference sequences, are listed in bold red. Reverse mutations are highlighted in bold and underlined. (B) Details of the structural interaction between N6 and HIV-1 gp120 loop D. (C) Details of the structural interaction between N6 and HIV-1 gp120 CD4-binding loop C. (D) Details of the structural interaction between N6 and HIV-1 gp120 V5. N6 components are shown in ribbon with key residues highlighted in sticks. HIV-1 loop D, CD4-binding loop, and V5 are shown in sticks. Color scheme is the same as Figure 2A. See also Figure S4 and Table S5.

Figure 5. Features across the Length of the Heavy Chain Contribute to the Breadth and Potency of N6

(A) Alanine scanning of N6, VRC01, and VRC27. Residues with a high buried surface area were selected for alanine scanning. Each alanine mutant was tested for neutralization with six VRC01-sensitive viruses. See Table S6. Neutralization fold change was calculated as follows: IC50 of antibody mutant / IC50 of antibody wild-type. Residues that resulted in fold change values >50 are highlighted in red, values between 20 and 50 are highlighted in yellow, and values between 5 and 20 are highlighted in green. (B) Neutralization of cross-complemented antibodies, including the heavy and light chains of the N6, VRC01, VRC27, and 12A21 antibodies. IC50 values <0.1 μg/mL are highlighted in red, values between 0.1 and 1 μg/mL are highlighted in orange, values between 1 and 10 μg/mL are highlighted in yellow, and values between 10 and 50 μg/mL are highlighted in green. Neutralization fold change was calculated as follows: IC50 of original antibody / IC50 of the antibody combination. Median IC50 is calculated based on VRC01-resistant viruses only. A value of 50 was assigned to the resistant viruses with an IC50 >50. Breadth was calculated as the number of viruses sensitive to the combination antibody / total number of tested VRC01-resistant viruses. (C) Neutralization of six VRC27-resistant and two VRC27-sensitive viruses by VRC27 mutants with substitutions of N6 heavy chain framework (FR) or CDRs. (D) Neutralization IC50 of six VRC27-resistant viruses by N6 mutants with substitutions of VRC27 FR or CDRs. Neutralization fold change was calculated as follows: IC50 of N6 mutant / IC50 of N6. See also Figure S5 and Tables S6 and S7.

Figure 6. N6 Evolved from an Early Intermediate to Circumvent Mechanisms of Resistance to the VRC01 Class

(A) Paired phylogenetic tree of the N6 lineage. The curated heavy chain transcripts of the lineage-related members were identified using the sequence identity in CDR H3 to that of N6, VRC27, F8, or N17. The curated light chain transcripts of the lineage-related members were identified from next-generation sequencing transcripts originated from IGKV1-33*01 and contained the five-amino-acid CDR L3 signature of the VRC01-class antibodies. (B) Neutralization by N6 intermediate antibodies against 20 VRC01-resistant pseudoviruses. (C) Mapping of I2 to I3 and I3 to I4 mutations onto the N6 heavy chain. It is of note that most I2 to I3 mutations are close to the antibody:antigen interface, whereas the I3 to I4 mutations are distant from the interface. See also Figure S6 and Table S8.

Figure 7. Changes in V5 Induced Evolution of N6 Intermediates

(A) Model showing key I2 to I3 mutations that increased I3 breadth by reducing CDR H2 clashes with the HIV-1 V5. (B) Model showing key I2 to I3 mutations that increased I3 breadth by reducing CDR H2 clashes with the HIV-1 outer-domain-exiting loop. (C) Model showing key I2 to I3 mutations that increased I3 breadth by improving the VH domain core packing. (D) Model showing key I2 to I3 mutations that increased I3 breadth by removing heavy chain and light chain clashes. (E) Model showing key I2 to I3 mutations that increased I3 breadth by optimizing the elbow region. (F) Contact regions of I2 to I3 mutations on HIV-1 gp120. Regions within 8.5 Å of the I2 to I3 mutation are colored red on the cartoon representation of gp120. (G) Sequence comparison of I2, I3, and I4 sensitive HIV-1 strains that are resistant to VRC01. Residues within 8.5 Å of bound N6 antibody were shown in Weblogo representation (upper panel), with the degree of conservation represented by the height of the residue. Sequence alignment (lower panel) with the loop E insertion, bulky residues, and glycosylation sites at base of V5 highlighted.