Effects of Cyclic Permutation, Glycan Removal and Aspartate Mutagenesis on Conformation of HIV-1 Envelope

Abstract

It has been decades since Human Immunodeficiency Virus-1 (HIV-1) was discovered. The virus has created havoc worldwide. According to the UNAIDS 2014 Gap report, 35million people worldwide are infected with HIV-1 and yet there is no vaccine. The most abundant and exposed protein on the HIV-1 virion surface is Envelope (Env) glycoprotein. HIV-1 uses its Env glycoprotein to infect host cells. HIV-1 Env glycoprotein exists as a trimer of heterodimers of gp120 and gp41. gp120 is the surface exposed and gp41 the membrane anchored subunit. gp120, being surface exposed, is the primary target of neutralizing antibodies and therefore is an important candidate for immunogen design. The generation of an effective HIV-1 vaccine is a challenging job because the HIV-1 virus has acquired many immune evasive mechanisms. The HIV-1 virus has high sequence variability, because of which it is difficult to make a vaccine that can effectively act against all viral strains. HIV-1 gp120 is extensively glycosylated and thus much of the surface is concealed from the host immune system. It also has immunodominant loops that are highly exposed in shed gp120 and can drive the immune focus towards a non-neutralizing response. There are different strategies that have been employed in the search for an HIV-1 vaccine. Chapter 1, contains a brief introduction of HIV-1, the difficulties in eliciting neutralizing antibodies and the strategies employed so far in Env based immunogen design. In Chapter 2, we describe the use of cyclic permutation as a strategy to trimerize gp120 in the absence of gp41. Normally, the Env trimer present on the cell surface exists as a trimer of gp120 and gp41 heterodimers. The trimerization is largely mediated by gp41 and assisted by V1V2 loops on gp120 that form the apex of the trimer. The antibodies generated against gp41 are autoreactive and more often non-neutralizing. To avoid that, only gp120 was used as an immunogen and to trimerize gp120 in the absence of gp41, a cyclic permutant of gp120 was made. The original N and C termini were joined by a linker and new N and C termini were generated in the V1 loop region. The V1 loop is a flexible region in gp120 and can tolerate insertions at the chosen location between residues 143 and 144. The trimerization domain, human cartilage matrix protein (hCMP), was linked at the new N terminus of the V1 loop. In a separate construct, hCMP was also linked to the C terminus of gp120 as a negative control. We found that the resulting V1 cyclic permutants JRCSF-hCMP-V1cyc and JRFL-hCMP-V1cyc improved the binding of gp120 to broadly neutralizing antibodies like VRC01 and VRCPG04 (CD4 binding site antibodies) and reduced the binding to non-neutralizing antibodies like b6 and F105. These cyclic permutants showed ~ 10-20 fold increase in binding to the quaternary epitope specific broadly neutralizing antibodies PGT145, PG9, PG16, and PGDM 1400. The gp120 cyclic permutants showed an increased thermal stability by 4˚C compared to Wtgp120 and were shown to be trimeric by negative stain EM. On immunization in guinea the resulting sera could neutralize a subset of heterologous cross clade Tier-2 viruses. Some of the neutralization response was CD4 binding site directed. gp120 is a heavily glycosylated molecule. It contains ~25 glycosylation sites, ~14-15 in the outer domain, ~4 in the inner domain, ~7-8 in the V1V2 and V3 loops. Earlier it was believed that glycosylation is indispensable for viral infectivity as well as for proper folding of Env. In Chapter 3, we investigated the effect of removal of glycans from core gp120 on viral infectivity and conformation of the soluble protein. The glycans from gp120 were rationally removed by mutating to the second most frequent residue in a multiple sequence alignment. In natural infection, it takes many years to generate broadly neutralizing antibodies. During the course of development, the antibodies undergo extensive affinity maturation from their germline unmutated versions. It is observed that the glycosylated wild type gp120 does not bind to the germline reverted version of the broadly neutralizing, CD4 binding site antibody VRC01, and the binding progresses with the removal of glycans. We show that the removal of glycosylation from the outer domain of core gp120 in the JRFL molecular clone does not affect viral infectivity. The viral fitness of virus lacking glycans in core gp120 was assessed in HUTR5 cells. It was found that the infectivity of glycan deficient virus is higher than wild type, suggesting that the primary role of glycosylation is not to stabilize Env conformation, but rather to shield the molecule from the humoral immune system. We show that upon selective removal of glycosylation from the outer domain of the trimeric cyclic permutant of JRFL described in (chapter 2), the conformational integrity of the cyclic permutant is not perturbed. The glycan deficient molecules bound to the trimer specific, quaternary epitope targeting antibodies like PGDM 1400 similarly to the wild type molecule. It also bound mature VRC01 with an affinity similar to the WT. The glycan deficient molecules bound weakly to germline reverted VRC01 but the wild type molecule did not bind at all. Such glycan-deficient molecules can be used in immunizations to target germline B cells and can potentially broaden the antibody response. A high resolution crystal structure of the Env trimer in the absence of any stabilizing ligands is needed for an effective immunogen design. All the existing crystal and EM structures contain one or more ligands. In most structures of the soluble gp140 ectodomain of HIV-1, no density is visible for residues 512-517 of fusion peptide and 547-568 of the N-heptad region of gp41. In Chapter 4, we have used an aspartate scanning mutagenesis strategy to probe the residue burial of these gp41 regions for which the electron density is missing in most structures. We mutated these residues individually into aspartate in the mammalian cell surface expression plasmid JRFLgp160dCT and the effect of mutations was assessed by FACS. Cell surface expression of Env was monitored using 2G12 binding. The conformational integrity was probed by binding of conformational specific CD4 binding site neutralizing antibody b12 and non-neutralizing antibody b6. The b12/b6 ratio differentiates the effect of different mutations on native trimer integrity compared to the wild-type. V513D, Q562D, L576D, I580D were the mutations, highly affecting the expression and trimer integrity. L576 and I580 are both buried in the native Env. However, Q562 is not, so this is not consistent with the existing structures. Asp substitutions at several other residues in this region also appear to affect the native trimer integrity, suggesting that the region is unlikely to be disordered in native Env on the virion surface. The NHR and CHR regions of gp41form a six-helix bundle post fusion of viral and cellular membranes. In this structure, the three CHR wrap around the NHR trimer. We previously showed the V570D mutation, that destabilizes the six-helix bundle also prevents gp120 shedding without significantly perturbing the conformation of native Env. We, therefore, examined the effects of Asp mutants that apparently have native like conformation on CD4 induced, gp120 shedding. R557D, A558D, A561D, Q563D, R564D, M565D were all able to prevent gp120 shedding, but, N553D, N554D, Q560D could not prevent gp120 shedding. Several residues preventing gp120 shedding, namely R557D, A558D, A561D, Q563D, R564D, M565D are exposed in the native trimer, but surprisingly R557D and R564D are exposed in the post fusion form as well (PDB 1AIK). The mutations L555D, I559D, Q562D, L566D, T569D, L576D and I580D are all buried in the post fusion structure. Pseudoviral infectivity studies showed that virus loses all its infectivity in the presence of these mutants, presumably because they disrupt the six helix bundle formation. Residues preventing CD4 shedding are able to lock the trimer in a pre-fusion context on the cell surface. However, further studies need to be done in the context of a soluble trimeric protein, which can be used as an immunogen. The native trimer is metastable. In Chapter 5, we have explored the utility of priming with gp160 DNA and boosting with trimeric gp120 or gp140 in prime boost immunizations in rabbits. We have used JRFLgp160dCT-V570D (dCT is cytoplasmic tail truncated), JRFLgp160dCT-Wt, and JRFL gp160dCT-SEKS (non-cleavable gp160) as DNA prime and boosted with JRFL-hCMP-V1cys or with ZAFV gp140 protein. In another set of studies, animals were subjected to a protein prime with variants of the gp120 outer domain and boosted with JRFL-hCMP-V1cyc or ZAFV g140. However, only Tier 1 neutralization was observed. Further work is needed to optimize these immunogens in order to elicit broadly neutralizing antibodies.