Autoreactivity and Exceptional CDR Plasticity (but Not Unusual Polyspecificity) Hinder Elicitation of the Anti-HIV Antibody 4E10

Broadly neutralizing antibodies (bNAbs) specific for conserved epitopes on the HIV-1 envelope (Env) are believed to be essential for protection against multiple HIV-1 clades. However, vaccines capable of stimulating the production of bNAbs remain a major challenge. Given that polyreactivity and autoreactivity are considered important characteristics of anti-HIV bNAbs, we designed an HIV vaccine incorporating the molecular adjuvants BAFF (B cell activating factor) and APRIL (a proliferation-inducing ligand) with the potential to facilitate the maturation of polyreactive and autoreactive B cells as well as to enhance the affinity and/or avidity of Env-specific antibodies. We designed recombinant DNA plasmids encoding soluble multitrimers of BAFF and APRIL using surfactant protein D as a scaffold, and we vaccinated mice with these molecular adjuvants using DNA and DNAprotein vaccination strategies. We found that immunization of mice with a DNA vaccine encoding BAFF or APRIL multitrimers, together with interleukin 12 (IL-12) and membrane-bound HIV-1 Env gp140, induced neutralizing antibodies against tier 1 and tier 2 (vaccine strain) viruses. The APRIL-containing vaccine was particularly effective at generating tier 2 neutralizing antibodies following a protein boost. These BAFF and APRIL effects coincided with an enhanced germinal center (GC) reaction, increased anti-gp120 antibody-secreting cells, and increased anti-gp120 functional avidity. Notably, BAFF and APRIL did not cause indiscriminate B cell expansion or an increase in total IgG. We propose that BAFF and APRIL multitrimers are promising molecular adjuvants for vaccines designed to induce bNAbs against HIV-1. IMPORTANCERecent identification of antibodies that neutralize most HIV-1 strains has revived hopes and efforts to create novel vaccines that can effectively stimulate HIV-1 neutralizing antibodies. However, the multiple immune evasion properties of HIV have hampered these efforts. These include the instability of the gp120 trimer, the inaccessibility of the conserved sequences, highly variable protein sequences, and the loss of HIV-1-specific antibody-producing cells during development. We have shown previously that tumor necrosis factor (TNF) superfamily ligands, including BAFF and APRIL, can be multitrimerized using the lung protein SP-D (surfactant protein D), enhancing immune responses. Here we show that DNA or DNA-protein vaccines encoding BAFF or APRIL multitrimers, IL-12p70, and membrane-bound HIV-1 Env gp140 induced tier 1 and tier 2 neutralizing antibodies in a mouse model. BAFF and APRIL enhanced the immune reaction, improved antibody binding, and increased the numbers of anti-HIV-1 antibody-secreting cells. Adaptation of this vaccine design may prove useful in designing preventive HIV-1 vaccines for humans.

mentioning

confidence: 99%

DNA Vaccine Molecular Adjuvants SP-D-BAFF and SP-D-APRIL Enhance Anti-gp120 Immune Response and Increase HIV-1 Neutralizing Antibody Titers

Gupta

Clark

Termini

et al. 2015

“…1A to C). In other words, this structure may not represent a fully "ligand-free" state of 4E10, but, unluckily, recapitulates the ligand-bound state, resolving the potential contradiction.The CDR-H3 movement in our ligand-free Fv structure (3) also revealed a potential phosphate-binding site, explaining 4E10's membrane-binding properties (3,6,10,11). We found this result compelling, because we have argued that the ligand-bound state has no obvious feature accounting for phospholipid binding (3, 6).…”

mentioning

confidence: 59%

The Broadly Neutralizing, Anti-HIV Antibody 4E10: an Open and Shut Case?

Strong

Finton

2016

Self Cite

R ujas et al. (1) recently studied the recognition mechanism of the anti-HIV antibody 4E10. We completely agree with their overall conclusion, that 4E10 "must recognize an antigenic structure more complex than just the linear ␣-helical epitope," which echoes our own analysis (2): "4E10 may make contacts to elements of Env outside of the linear 4E10 epitope." Their ligand-free Fab structure showed "that the CDR-H3 loop does not undergo significant conformational changes during recognition of the epitope," consistent with thermodynamics that "ruled out major conformational changes of CDR-H3."However, our structure of the ligand-free 4E10 Fv (3) had revealed a dramatic restructuring of CDR-H3, compared to Fv or Fab structures bound to HIV peptides (4-7). The difference was accounted for by the "low pH required to crystallize the Fv construct" and/or by "different structural stabilities" (1). Crystallization conditions can affect structure, and our ligand-free Fv crystallization conditions were nonphysiological, as noted by others (8). However, these conditions are not "extreme" in terms of protein crystallization overall. Solution thermostabilities of Fvs are also typically lower than those of Fabs (9), partly because Fvs lack the interchain disulfide that stabilizes Fabs. Since we had carefully shown that the 4E10 Fv retains the structure, binding, and neutralization properties of the 4E10 Fab (6), these explanations may not fully account for the apparent discrepancy. Rojas et al. state that the binding thermodynamics also argue against major alterations of CDR-H3 conformation. We counter that the conformational switch may not proceed through a full order-disorder transition, with a large entropic effect, but may switch between two relatively ordered states, with smaller entropic effects, much like a door swinging between completely open and completely shut. Also, the thermodynamics, performed on an isolated envelope peptide, have a surprisingly low entropic signature, considering that the peptide would be expected to go through an order-disorder transition during binding.What then might account for the different structural results, Fv versus Fab? Crystallization affects structure not only through solution conditions, but also through the requirement to pack molecules together in the crystal lattice. In the ligand-free Fab structure, the two Fabs in the asymmetric unit both make crystal contacts through CDR-H3. These contacts partially mimic interactions with HIV ligands (Fig. 1A to C). In other words, this structure may not represent a fully "ligand-free" state of 4E10, but, unluckily, recapitulates the ligand-bound state, resolving the potential contradiction.The CDR-H3 movement in our ligand-free Fv structure (3) also revealed a potential phosphate-binding site, explaining 4E10's membrane-binding properties (3,6,10,11). We found this result compelling, because we have argued that the ligand-bound state has no obvious feature accounting for phospholipid binding (3, 6). Rujas et al. (1) alternately interpret the...

“…For example, we acknowledge that the discrepancy between the crystal structures of the unbound form of the Fab and Fv constructs does not by itself constitute sufficient evidence to completely rule out the existence of large conformational changes in the CDR-H3 loop (2). In addition, neither the crystal structure of the unbound Fv nor the "shut-open" mechanism precludes the existence of a distinct unbound form of the antibody competent in binding to the helical epitope.…”

mentioning

confidence: 92%

“…In that study, we determined the crystal structure of the unbound form of the Fab region of 4E10, as well as that of nonneutralizing mutants with the peptide epitope bound (1). In particular, the conformation of the critical CDR-H3 loop in the crystal structure of the unbound Fab differed greatly from that previously observed for the Fv version of the same antibody (2). Collectively, our data support the existence of a preformed crevice for binding of the helical membrane-proximal external region (MPER) epitope.…”

mentioning

confidence: 95%

Reply to “The Broadly Neutralizing, Anti-HIV Antibody 4E10: an Open and Shut Case?”

Rujas

Gulzar

Morante

et al. 2016

W e thank Drs. Strong and Finton for their interest in our work involving the broadly neutralizing anti-HIV-1 antibody 4E10 (1). In that study, we determined the crystal structure of the unbound form of the Fab region of 4E10, as well as that of nonneutralizing mutants with the peptide epitope bound (1). In particular, the conformation of the critical CDR-H3 loop in the crystal structure of the unbound Fab differed greatly from that previously observed for the Fv version of the same antibody (2). Collectively, our data support the existence of a preformed crevice for binding of the helical membrane-proximal external region (MPER) epitope.In the accompanying letter (3), Strong and Finton concur with the main tenet of our article, i.e., that 4E10 must "recognize an antigenic structure more complex than just the linear ␣-helical epitope." However, in the same letter these authors disagree with our interpretation of the structural data corresponding to the new unbound form of the Fab fragment and with our criticism of the structure of the unbound form of the Fv version (2). Strong and Finton argue that the low pH of the crystallization solution and the use of the shorter Fv construct are not compelling arguments to explain the conformational discrepancies with respect to the Fab. They argue that the Fv construct of 4E10 "retains the structure, binding, and neutralization properties of the Fab" (4). In addition, they explain that the thermodynamic signature of the binding (low entropic component) would not be consistent with an order-disorder transition during binding of the epitope peptide in solution. Moreover, based on a qualitative crystal contact analysis, they argue that our structure of the Fab in the unbound state does not reflect the "fully-free state of 4E10, but, unluckily, recapitulates the ligand-bound state, resolving the potential contradiction." The closed conformation observed in the Fv construct may generate a lipid-binding site representing an alternative conformation to the epitope-binding structure, implying that a conformational transition must occur between both forms for effective engagement of the gp41 antigen (2, 4).We find the arguments expressed in the letter by Strong and Finton reasonable, but only to some extent. For example, we acknowledge that the discrepancy between the crystal structures of the unbound form of the Fab and Fv constructs does not by itself constitute sufficient evidence to completely rule out the existence of large conformational changes in the CDR-H3 loop (2). In addition, neither the crystal structure of the unbound Fv nor the "shut-open" mechanism precludes the existence of a distinct unbound form of the antibody competent in binding to the helical epitope. In the framework of the model described in Strong and Finton's letter, the structure described in our paper would correspond to the "open-unbound" antibody.Nevertheless, a fair critical assessment of the unbound structures should take into account the entire data set available in our study (1) and the structures...