A. HIV Cure Strategies Based on Targeted Killing of HIV-Infected Cells. We have proposed that different modes of targeted cytotoxic therapy are applicable for primary acute infection versus established chronic infection. For acute infection, relatively short-term treatment with a transiently active modality such as a recombinant immunotoxin (RIT) might provide sufficient complementation of HAART to achieve eradication of infectious HIV. For chronic infection, continuous suppression of persisting replication by destruction of newly activated reservoir cells can potentially be achieved with autologous T cells engineered to express a Chimeric Antigen Receptor (CAR). 1. Env-targeted recombinant immunotoxins (RITs). a. In collaboration with Dr. Ira Pastan, NCI, we showed the new VRC-07-PE it to be broadly active against all genetically diverse HIV-1 isolates tested. This RIT has enhanced potency compared to CD4-PE for isolates known to be relatively insensitive to neutralization by soluble CD4. b. We completed a collaborative NHP study with the VRC (Dr. Richard Koup, Dr. John Mascola) and MHRP (Dr. Diane Bolton) to test whether complementation of cART with an RIT can achieve a cure of rhesus macaques acutely infected with a SHIV. We found that when treatment was initiated at 3 days post-infection, the RIT was unable to sufficiently complement HAART to eliminate infection. The potential remains for this approach to achieve virus eradication when treatment is initiated at earlier times, e.g. for newborns. 2. Chimeric Antigen Receptors (CARs). Autologous CAR-T cells have potential long-term suppression of HIV, to achieve a functional cure. This requires the CARs to approach two critical properties: inescapable and non-immunogenic. We have designed bispecific CARs consisting of human CD4 linked to a second human moiety that binds to a distinct highly conserved element on HIV-1 Env. The second moiety enhances potency, and prevents the CD4 from acting as an entry receptor into CAR-expressing CD8+ T cells. Our favored CARs contain CD4 linked to the carbohydrate recognition domain (CDR) human mannose binding lectin, which binds to the high-mannose glycan patch universally displayed on all gp120 variants. The CD4-MBL CAR displays significantly enhanced potency compared to a monspecific CD4 CAR against genetically diverse HIV-1 isolates. The following advances were made in 2017 (including collaboration with DIR investigators and members of the BELIEVE Martin Delaney Collaboratory): a. Mechanistic features underlying the enhanced potency of bispecific CD4-based CARs. We propose that the bispecific CAR design might promote engagement of multiple CAR molecules to a single gp120. Mutational analyses indicate that the disulfide bond in the short extracellular hinge of our CARs promotes their dimerization, but is not required for high potency, suggesting importance of downstream CAR signaling events. Another mechanistic question is whether CAR-T cells can mediate killing based on Env introduced by the incoming virus particle, prior to synthesis of new Env. This would provide an important kinetic advantage against HIV spread. Preliminary experiments with HIV-1 pseudovirus infection support this model. b. Engineered trafficking of CAR-T cells to B cell follicles. In collaboration with Dr. Pamela Skinner (U. Minnesota) and Dr. Elizabeth Connick (U. Arizona), we have analyzed constructs containing the genes for both the CAR and CXCR5, the follicular homing receptor. Co-expression of CXCR5 does not impair the expression of suppressive activity of the CD4-MBL CAR. CXCR5 promotes directed migration of CAR-T cells to CXCL13 in an in vitro model, and enhanced accumulation into lymphoid tissues ex vivo. b. CAR-T cells in humanized mouse models. In collaboration with Dr. R. Brad Jones (GWU, BELIEVE Collaboratory), we are testing the CD4-MBL CAR in a novel humanized mouse system generated from T cells of HIV-infected subjects. Preliminary results reveal the importance of optimizing engraftment of the adoptively transferred cells, and testing their activity in animals with low viral loads, either at the acute stage of infection, or in combination with antiretroviral drugs. We have begun to establish within LVD the novel DRAG humanized mouse model, in which HIV-1 replication occurs extensively in Tfh cells, a phenomenon not observed in other humanized mouse models. This system will enable testing CAR-T cell function and directed trafficking mediated by CXCR5. c. CAR-T cells in nonhuman primate models. We generated an all-rhesus CD4-MBL CAR construct, containing a point mutation (I39N) in the CD4 component described by others to enhance CD4 binding to Env. This CAR gives very strong in vitro suppressive activity against different SIV and SHIV strains. Co-expression of CXCR5 does not impair the activity of the all-rhesus version. As an initial experiment (with Dr. Mario Roederer, VRC) SIV-infected rhesus macaques were injected with a mix of different CARs (rhesus CD4-MBL plus other scFv-based CARs developed in the Roederer lab). The major goal will be to learn whether there is selective presence/persistence/trafficking of any of these CARs in various tissues/compartments. Studies with BELIEVE Collaboratory investigators (P. Skinner, E. Connick, James Whitney, BIDMC/Harvard) have begun to examine CD4-MBL/CXCR5 in SIV-infected monkeys. B. Structure/Function Studies of HIV-1 Env-Receptor Interactions 1. Soluble chimeric fusion protein containing sCD4 linked to CCR5 tyrosine-sulfated N-terminal peptide. We previously described evidence that both moieties of the fusion protein interact with their corresponding binding sites on gp120. We are pursuing collaborative opportunities to use this construct for obtaining atomic level structural information that is lacking to date on a key element of the HIV entry mechanism, namely the gp120/coreceptor interaction. 2. Collaborative studies with Dr. Paolo Lusso, NIAID focused on the notion of molecular mimicry based on a short stretch of amino acids common to both HIV-1 gp120 (alpha-1 helix, highly conserved) and CD4. This has led to design of novel Env constructs containing newly introduced pair of Cys residues that form a new disulfide bond to generate an inter-domain lock, thereby stabilizing the prefusion conformation of the HIV-1 trimer (soluble and membrane-associated forms). The neo-disulfide bond completely blocked infectivity in pseudovirus assays. Compared to wild-type soluble trimers, the neo-disulfide-stabilized constructs form more compact trimeric structures with increased thermal stability. Importantly, they display desirable antigenic profiles, including enhanced binding to trimer-specific neutralizing mAbs, and greatly reduced/abolished binding to non-neutralizing mAbs as well as to CD4 and CD4-induced mAbs. These results contrast with other reported stabilized Env trimers, which did not fully prevent CD4 binding and associated conformational changes. Immunogenicity studies in rabbits revealed novel elicitation of tier-2 neutralizing antibodies against viruses with intact glycan shields, suggest potential value of these stabilized trimers for vaccine development.
Showing the most recent 10 out of 14 publications
