| Abstract: | A vaccine which is effective against the HIV virus is considered to be the best solution to the ongoing global HIV/AIDS epidemic. In the past thirty years, numerous attempts to develop an effective vaccine have been made with little or no success, due, in large part, to the high mutability of the virus. More recent studies showed that a vaccine able to elicit broadly neutralizing antibodies (bnAbs), that is, antibodies that can neutralize a high fraction of global virus variants, has promise to protect against HIV. Such a vaccine has been proposed to involve at least three separate stages: First, activate the appropriate precursor B cells; second, shepherd affinity maturation along pathways toward bnAbs; and, third, polish the Ab response to bind with high affinity to diverse HIV envelopes (Env). This final stage may require immunization with a mixture of Envs. In this paper, we set up a framework based on theory and modeling to design optimal panels of antigens to use in such a mixture. The designed antigens are characterized experimentally and are shown to be stable and to be recognized by known HIV antibodies.Vaccines are the most important medical countermeasure for protecting entire populations against viruses, of which smallpox and measles vaccines are successful examples. In fact, a safe and effective HIV vaccine is considered to be the best way to end the global AIDS epidemic (1). However, how to produce a universal vaccine for highly antigenically variable viruses like HIV is a daunting and yet unsolved problem. The high variability of this virus allows it to elude the immune system, making the produced antibodies ineffective; that is, they are generally specific for a given strain of the virus but not for other strains resulting from mutation. In some cases, HIV-infected patients can elicit antibodies that can recognize and neutralize a broad range of different viral strains (2, 3). These broadly neutralizing antibodies (bnAbs) usually take a long time to appear naturally in infected patients and then only in a subset of such individuals.The reason that bnAbs can arise is that even highly variable pathogens have regions with a well-defined, relatively conserved structure, which is required for their function. In HIV, entry depends on the trimeric spike exposed on the external lipid membrane of the virion, a heterotrimer formed by the gp120 and gp41 glycoproteins produced by posttranslational cleavage of a gp160 precursor. This protein binds to the CD4 coreceptor on CD4 T lymphocytes during HIV infection, and it has some relatively conserved regions that can be used as a target for bnAbs. Indeed, many bnAbs target the CD4 binding site (CD4bs) (4–8). If naive B cells that can bind to one of these relatively conserved regions can be expanded upon exposure to different variants of the virus, antibodies could evolve to better recognize the conserved portions, while avoiding the variable ones. The resulting antibodies can acquire breadth in this way, thereby becoming bnAbs. A successful vaccine would contain immunogens that can guide the immune system to produce bnAbs, rather than strain-specific antibodies.In the past, numerous approaches for the development of an effective HIV vaccine have been tried. They include the use of cleverly chosen natural HIV proteins, the design of a consensus (9) or “center-of-tree” (10) antigens, and the creation of a mosaic protein from different HIV strains (11). All these methods used a single optimized antigen in the vaccine and were shown to be ineffective at eliciting bnAbs (12, 13). One possible reason for this is that, when exposed to a single antigen, the immune system will produce antibodies specific for that particular antigen, and neutralization escape variants can easily develop. A possible solution is to use more than one antigen in a vaccination protocol. This raises a number of questions: How many antigens are necessary? How different should they be from each other? And in what temporal order should they be administered? Answering such questions is far from trivial, in particular due to the limited mechanistic understanding of affinity maturation (AM) in vivo. Another problem is that bnAbs have an unusually high number of somatic mutations, not only in the complementarity-determining regions (CDRs) but also in the immunoglobulin framework regions (7, 14). Recent computational data on the flexibility of the antibody and the need for framework mutations in the simulated AM showed how important it is for a vaccination protocol to have a specific antigen that can prime a good antibody precursor B cell receptor (BCR) (15). Moreover, it has been shown that putative precursors of known classes of bnAbs are generally not able to neutralize HIV or recognize envelope (Env), often due to clashes of the antibody with the glycosylation shield that protects the HIV Env protein (8, 16–18). For example, VRC01-class bnAbs are known to introduce a deletion or a mutation to a flexible glycine in the CDRL1 loop to avoid the glycan at N276 (19, 20).The above discussion led to the proposal of a vaccination strategy consisting of three steps. First, a special purpose antigen is used to activate the correct naïve or precursor B cell (17, 21). Since this precursor will generally not bind to native HIV, as a second step, one or more antigens are used as intermediates to induce somatic mutations and to allow recognition of the native virus. In the third step, one or more antigens are used in a mixture or in sequence to increase the breadth of the antibody population (19, 22, 23). Implementations of the first and second steps have already been shown to be promising in experiments (16, 21, 24–27). However, much less is known about the third step. Some insights into this question can be obtained by in silico simulations of AM. Using coarse-grained models, it has been shown that, while administering a single mixture containing multiple antigens may induce too much frustration to lead to bnAbs formation, a sequential approach, in which antigens are administered one after another, seems to be more effective (23). It was also observed that the number of antigens required in a mixture is correlated with their sequence dissimilarity, and optimal breadth is obtained at an optimal number of antigens and dissimilarity (28). Given the coarse-grained nature of these studies, the actual antigen sequences to use in experiments cannot be obtained from them.In this work, we focus on the third step of the proposed vaccination protocol. In particular, we derive a set of empirical rules and protocols to select an optimal panel of antigens to maximize the breadth of the produced antibodies upon AM. To be able to do so, it is essential to understand, at an atomistic level of detail, the role of each antigen amino acid in the antibody/antigen interaction. This aspect will be presented in the next section based on an analysis of the available crystallographic structures of bnAbs bound to the gp160 Env glycoprotein. However, the structures do not provide information concerning HIV stability and function. For example, generating antigen sequences by introducing purely random mutations will likely lead to sequences that are lethal for the virus and/or are not representative of HIV in vivo. To overcome this problem, it is useful to consider the structural data together with a model of the gp160 fitness landscape (29), which is a measure of the ability of HIV to tolerate mutations in its gp160 sequence to escape immune pressure. Structural and fitness information together provide a classification of the antibody/antigen interface and indicate the residues to mutate and the amino acids that are more probable at those positions.While this analysis helps to reduce the number of antigen sequences to consider by highlighting the “hot spots” of antibody/antigen binding, it leaves open the question of how to select a combination of antigen sequences for use in a vaccine. Given rules of optimal sequence dissimilarity and optimal fitness according to the HIV landscape, a Pareto frontier approach will be described. It is able to select, from all possible panels of antigen sequences, the few that are predicted to best elicit antibodies with a broad activity spectrum. Experimental evidence of the viability of the designed antigens and of their immunogenic properties is presented in the final section. |
