Modelling the Interaction Between Synthetic Model Immunogens and the Induced B and T Cell Repertoires.

One of the most effective arms of the human immune system is the ability of very low concentrations of antibody proteins to bind to viruses, bacteria and toxins and "neutralise" their activity or ability to infect. In contrast to cellular immunity, which may cause tissue destruction and pathology, antibody-mediated immunity can be very passive, while completely preventing infection. How antibodies bind their targets varies enormously, ranging from unhelpful "blocking" antibodies or narrowly focussed neutralising antibodies, to highly protective "broadly neutralising" antibodies (bNAbs) that can neutralise a wide range of strains of the same pathogen. Such bNAbs are especially sought after in virus infections such as HIV, influenza and others where the virus mutates to evade immune responses that are too narrow or focussed. Antibodies arise when an "immunogen" (an immunogen is anything that induces an immune response, typically a foreign protein) is taken up by the immune system and shown to white blood cells - T and B cells - by specialised immune cells. In some cases the T and B cells bind the immunogens to receptors on their surface, triggering an immune response in which T cells "help" B cells to manufacture specific antibodies. The events around how the protein is processed into manageable pieces, shown to the T and B cells, and the pattern of chemical signals produced by the immune cells is highly complex, but eventually determines how broad the antibody response will be (its breadth). For infections like HIV and influenza, decades of research and clinical vaccine trials have had limited or no success. To take HIV as an example, we have an almost complete lack of understanding of how immunogens interact with the naive human B cell receptor (BCR) repertoire and the pathways required to induce bNAbs during an infection or after an immunisation. Animal models have failed as the naïve, germline encoded, B cell antibody receptor repertoires of non-human species are sufficiently different from those of humans to render design and selection of vaccine based on non-human species problematic. Additionally, bNAbs isolated from HIV-1-infected individuals have structural features that occur rarely or not at all in other mammals, such as unusually long loop-binding regions (CDRH3 loops) required to penetrate past glycans on the surface of the envelope spike that shield key neutralising epitopes (Mascola JR 2013). There is therefore a critical need to better understand, in human experimental medicine models of immune challenge, how immunogens and B/T cells interact in the development of protective bNAb anti-viral responses.

Our approach to resolving this impasse is to challenge the human immune system with rationally-designed model immunogens to determine the structural and other characteristics required to drive human B cell antibody responses towards neutralisation breadth. We have selected HIV as an experimental model as there is a reasonable understanding about the specificity and function of anti-HIV bNAbs, as well as an urgent need to identify novel immunisation approaches following decades of failed or poorly successful trials. There is also a huge database of safety using HIV proteins as immunogens, and the technological expertise to design and manufacture HIV viral proteins. Assays for HIV neutralising activity are also well established in our laboratories. Although focussed on HIV, our findings will be applicable to other viral infections.

The model immunogens we propose to use in these experimental medicine studies are unlikely to be suitable as vaccines, and any clinical development would require iterative cycles of design refinement and development based on immunological insights gleaned from these experimental investigations. Therefore, the focus is on in-depth characterisation of the elicited immune response to rationally-designed model immunogens that may inform the design process of actual vaccines. This experimental medicine approach is only now possible due to unprecedented progress in our abilities to study the human immune system and to obtain complete information on immune responses to vaccination, since performing research on the human immune system is now almost as easy as it has been in mice. The main focus of this study will be to determine which of the design strategies is able to prime human germline (naive) B cells and drive antibody responses towards induction of neutralising antibody breadth.

Our range of model immunogens will be based on the envelope (Env) glycoprotein of HIV-1, which is the only target of neutralising antibodies, and therefore the only virally-encoded immunogen relevant for induction of such antibodies by immunisation. To ensure reproducibility of results and the highest level of volunteer safety, all immunogens will be manufactured under cGMP, using techniques applied to vaccine immunogens.

Env has extensive amino acid variation, structural and conformational instability, and immunodominance of hypervariable regions (Kwong PD, 2011; Sattentau QJ, 2013). We will design soluble immunogens that closely mimic the native viral trimer in situ, but that incorporate design strategies that may alter the intrinsic viral immune evasion mechanisms (Sanders RW, 2013). Env is made up of three identical complexes (trimers) each of which contains two molecules, gp120 and gp41 that can be modified to make a soluble molecule called gp140, upon which our immunogens are based. We have developed model consensus gp140 Env trimers (consensus of all global strains) designed to prime B cell responses to common epitopes represented in all HIV-1 subtypes. We have utilised two design strategies to stabilise these in a native-like conformation: ConM SOSIP and ConS UFO. The ConM SOSIP trimer includes novel mutations that include the incorporation of a disulphide linkage between the gp120 and gp41 ectodomain (making up gp140) which prevents their disassociation into monomer subunits.

The ConS UFO includes a short flexible amino-acid linker to tether the gp120 and gp41 subunits together as an alternative strategy to prevent dissociation of the Env trimer. We wish to test both designs to determine the effect on B cell repertoire.

A critical adjunct to our consensus-based model design is to use a cocktail of three mosaic gp140 Env trimers designed to overcome the immunodominance of hypervariable regions of Env and to determine whether they will focus antibody responses towards conserved neutralisation epitopes. While designed using computer algorithms, these mosaics represent authentic Env structures that are fully functional and native in their conformation. Our novel designs aim to eliminate unwanted immunodominant antibody responses and focus B cells towards highly conserved supersites of vulnerability on Env, with particular emphasis on quaternary bNAb epitopes (Julien, JP, 2013; Kong L, 2013; Lyumkis D, 2013). Like the ConM SOSIP and ConS UFO trimers described above, the mosaic trimers have disulphide linkage which prevents disassociation of gp120 and gp41 into monomer subunits.

In the MOSAIC study groups, we will explore the use of the three mosaic immunogens used sequentially (in a series of different orders), or as a cocktail, to focus B cell responses towards conserved areas of Env. To amplify these responses we intend to give a final boosting immunization with both consensus immunogens (ConM and ConS).

The extent to which these different strategies may induce neutralising breadth, and the identification of the mechanisms and drivers involved, can only be determined empirically through human immunogen challenge studies.