HIV Vaccinology: 2021 Update

Abstract

HIV is a virus that remains a major health concern and results in an infection that has no cure even after over 30 years since its discovery. As such, HIV vaccine discovery continues to be an area of intensive research. In this review, we summarize the most recent HIV vaccine efficacy trials, clinical trials initiated within the last 3 years, and discuss prominent improvements that have been made in prophylactic HIV vaccine designs.

1. Introduction

Much has happened in the field of human immunodeficiency virus-1 (HIV) vaccine research in recent years. The field of vaccinology has advanced considerably in the last decade, giving rise to novel vaccines for human pathogens. HIV is a pathogen that has challenged vaccine researchers for over 30 years. Momentous HIV research milestones have been reached in the last decade, particularly in immunogen development. Yet, a working vaccine is still lacking. While public education, accessibility to antiretroviral therapy (ART), and pre-exposure prophylaxis (PrEP) have reduced spread of the virus especially in the developed world, HIV still lacks a complete cure and the ongoing HIV epidemic continues to heavily impact Sub-Saharan Africa. As of 2019, 38 million people globally were living with HIV, with 1.7 million new infections [1]. The development of an efficacious HIV vaccine continues to be a challenging but essential component for controlling the HIV epidemic.

Several aspects of HIV make vaccine development by traditional methods notoriously challenging. The vast majority of vaccines protect by eliciting efficient antibody responses [2]. On the surface of HIV, there is only one antigenic target, the Envelope (Env) trimeric viral spike protein, responsible for the fusion machinery of HIV-1. The full Env trimer consists of three non-covalently associated protomers formed by gp120 and transmembrane domain containing gp41 subunits, that are formed following post-translational cleavage of the gp160 polypeptide. The Env ectodomain has a median of 30 N-linked glycosylation sites per protomer [3], and this massive glycan shield limits accessibility to the protein surface. Genetic variability arising from high mutability of HIV, along with Env glycan cloaking makes it exceptionally challenging for the humoral immune system to generate antibodies that can neutralize the vast diversity of circulating HIV strains.

The sequence diversity of HIV Env results in stratification of isolates exhibiting varying degrees of sensitivity to neutralization referred to as tiers [4]. The HIV Env trimer is metastable, and exhibits ‘open’, ‘intermediate’, and ‘closed’ conformational states [5,6]. Tier 1A and tier 1B viruses are most sensitive to neutralization, and preferentially adopt the open and intermediate states. On the contrary, tier 2 (representing the majority of circulating HIV strains) and tier 3 viruses, are much more resistant to neutralization and favor the closed conformation [7]. This closed trimeric prefusion configuration of Env is heavily shielded by glycans, and is the primary target for antibody-based vaccines [8]. Presentation of the Env trimer and Env subunit components in alternate conformations, such as monomeric gp120 or open trimer conformations, shifts immunogenicity towards epitopes that are not visible on the closed conformation of Env. Env-based immunogens that fail to adopt the closed trimeric conformation drive preferential elicitation of non-neutralizing antibodies (nnAbs) [9–11] or antibodies that can only neutralize tier-1 viruses, irrespective of the virus tier from which the Env immunogen is derived. Although nnAbs and tier-1 neutralizing antibodies are easy to elicit, they are generally not protective against tier 2 circulating isolates, based on efficacy trials in humans to date (discussed in detail in sections below). In contrast, broadly neutralizing antibodies (bnAbs), isolated from some HIV infected patients, can neutralize a wide range of circulating tier 2 and 3 HIV strains [12]. BnAbs provide robust protection against viral challenge in animal models via passive immunity, and suppress viral load and rebound in both animals and humans [13,14].

While bnAbs have impressive potency, bnAbs among HIV⁺ individuals are uncommon, heavily mutated, acquire difficult somatic hypermutations such as insertions and deletions, may exhibit polyreactivity to autoantigens, and take longer to develop compared to neutralizing antibodies against common pathogens [12,15]. The elicitation of bnAbs by vaccination is an arduous task, and a substantial number of recent Env-directed immunogen design efforts focus on obtaining vaccine-elicited bnAb responses in a rational manner. In animal models, many of these stabilized soluble Env immunogens representing the closed conformation have demonstrated their ability to generate immunogen-matched tier 2 autologous neutralizing responses, and in some cases, weak heterologous tier 2 responses. Each of these topics is discussed in detail later in this review.

Many candidate HIV vaccines also focus on the elicitation of robust T cell immunity. A subset of HIV⁺ individuals are able to control the infection, maintaining low viral replication and sustained CD4 T cell counts without ART intervention due to strong CD8 T cell responses [16–18]. Using a rhesus cytomegalovirus (CMV) based CD8 T cell vaccine, ~50% of immunized non-human primates (NHPs) have reproducibly demonstrated robust viremic control following challenge with simian immunodeficiency virus (SIV) infection [19,20], highlighting the potential of T cell vaccines. This CMV vaccine was recently entered into a human phase 1 clinical trial (VIR-1111, NCT04725877). Protective virus-specific CD8 T cell responses most commonly target conserved regions within internal proteins such as the structural Gag protein [21–23], eliminating N-linked glycosylation of Env as a confounding factor in vaccine design. Further, unlike bnAb epitopes that have highly complex tertiary and quaternary structures, T cell epitopes are linear peptide sequences. Hence, T cell vaccines may be more effective in eliciting broadly protective responses. Additionally, because CD8 T cells target infected cells, T cell-focused vaccines are being developed as prospective therapeutic vaccines, as well as prophylactic vaccines. Novel T cell vaccines designed to elicit cross-clade conserved cytotoxic responses are also under development, several of which are being clinically evaluated for both therapeutic and preventative use. Approaches combining both bnAb and CD8 T cell vaccine modalities will also be exciting to explore, as both approaches have matured substantially in recent years [24].

Here, we summarize the HIV vaccine efficacy trials to date, recent phase 1 clinical trials (initiated within the last three years), and preclinical immunogens likely to progress to future trials.

2. History of HIV vaccine efficacy trials

In the long history of HIV vaccine development, seven candidate vaccines have proceeded to efficacy trials. The first of these were the recombinant gp120-based AIDSVAX trials, VAX003 [25] and VAX004 [26], aimed at eliciting neutralizing antibody responses (nAbs). VAX003 tested for vaccine efficacy in preventing blood borne transmission among injection drug users in Thailand, and VAX004 was designed to determine efficacy in the prevention of sexually transmitted HIV, primarily among men who have sex with men, and some heterosexual women in North America and the Netherlands. Each AIDSVAX® vaccine was a bivalent recombinant gp120 vaccine adjuvanted with aluminum hydroxide (alum) tailored for the specific regional transmitter founder virus subtypes for the two trials; AIDSVAX® B/E in the VAX003 being composed of gp120 from subtype B MN and subtype CRF01_AE A244, whereas AIDSVAX® B/B tested in VAX004 was composed of gp120 proteins from two subtype B strains, MN and GNE8. In both studies, the recipients generated antibody responses to the immunization-matched gp120 strain and neutralization of the tier 1 virus strain MN, but lacked broader neutralizing antibodies and lacked efficacy in reducing transmission [25,27,28].

Several HIV vaccine trials focused on eliciting T cell mediated cellular immunity. STEP (HVTN502) carried out across several global sites, and Phambili (HVTN503) taking place in South Africa, were phase 2b clinical studies testing the MRKAd5 trivalent vaccine developed by Merck. This was a replication-deficient Adenovirus serotype 5 (Ad5) vectored vaccine consisting of Gag, Pol and Nef genes. The STEP trial was prematurely terminated when interim analysis of the data demonstrated increased susceptibility to HIV-1 infection among those vaccinated relative to those receiving the placebo [29], which in turn led to the early termination of the Phambili trial. A long term follow up of both the STEP and Phambili participants confirmed an increased incidence of HIV-1 infection among those who were vaccinated [30,31]. This was thought have been related to increasing adenovirus-specific CD4 T cells at mucosal sites, which became HIV targets [32,33].

HVTN505 was a phase 2b multi-site efficacy trial that took place in the United States and assessed the efficacy of an HIV vaccine candidate which consisted of a prime containing six DNA plasmids (Gag, Pol, Nef, and three gp145 Envs each from clades A, B and C). This was followed by a boost with four Ad5 vectors eah expressing a Gag-Pol fusion protein, and Env from clades A, B, and C [34]. The vaccine elicited CD4 and CD8 T cell responses to at least one antigen in the vaccine in ~66% of the participants, and all participants generated an IgG response to an Env component in the vaccine. However, interim analyses showed a lack in protective efficacy and the trial was terminated early.

The only efficacy trial to demonstrate any level of protection was the RV144 trial that took place in Thailand, chronologically initiated between the AIDSVAX and STEP/Phambili trials. The RV144 tested a heterologous prime-boost combination, priming with the canarypox vectored ALVAC-HIV vaccine expressing Env, Gag and Pro (vCP1521), followed by Env protein boost with the recombinant gp120 protein vaccine, AIDSVAX B/E, used in VAX004 [35]. The vaccine yielded a modest efficacy of 31% by one metric, although statistical conclusions for the study were controversial [36,37]. Among participants that acquired HIV, the vaccine had no impact on viral load or CD4 T cell counts [35]. Follow up immune correlates analyses observed V1V2-directed IgG nnAbs, and Env IgG3 responses were associated with reduced risk of infection [38–40]. Polyfunctional (expressing multiple cytokines) CD4 T cell subsets also correlated with reduced risk of infection [41]. These outcomes provided the foundation for the more recent HVTN702 trial.

3. Recent and ongoing efficacy trials

3.1. HVTN702

The seventh efficacy trial was the phase 2b/3 trial HVTN702, also known as the Uhambo trial (NCT02968849), conducted in South Africa. HVTN702 was a follow up of the RV144 trial, assessing the immune response to ALVAC-HIV primes followed by recombinant gp120 protein boost, with modifications to accommodate primary circulating subtypes in South Africa. The ALVAC-HIV prime (vCP2438) in HVTN702 is identical to vCP1521 used in the RV144 trials less the CRF01_AE subtype (92TH023) gp120 insert replaced by a regionally dominant subtype C gp120 insert (96ZM651). The recombinant protein boost contained two subtype C gp120s, from isolates TV1.C and 1086.C. Additionally, the alum adjuvant originally used in the RV144 trial was replaced with the oil-in-water emulsion adjuvant MF59, which elicits stronger responses than alum. Two related phase 1/2a trials, HVTN107 (NCT03284710) and HVTN120 (NCT03122223), comparing alum with MF59, and MF59 with AS01_B as adjuvants respectively, were completed late 2019 and mid 2020 (Table 1).

Table 1.

Recent efficacy trials and their related phase 1/2a trials

Immunogen	NCT Trial number	Other names	Phase	Adjuvant	Completion Date*
ALVAC-HIV (vCP2438), bivalent subtype C gp120	NCT03284710	HVTN107	1/2a	Alum and MF59 comparison	Dec 2019
ALVAC-HIV (vCP2438), bivalent subtype C gp120	NCT03122223	HVTN120	1/2a	MF59 vs AS01B	Jul 2020
ALVAC-HIV (vCP2438), bivalent subtype C gp120	NCT02968849	HVTN702/Uhambo	2b/3	MF59	Aug 2021; prematurely terminated
Ad26.Mos4.HIV, subtype C gp140 and/or mosaic gp140	NCT02935686	ASCENT/HVTN118/HPX2003	1/2a	Alum	Jan 2022
Ad26.Mos4.HIV, subtype C gp140	NCT03060629	HVTN705/Imbokodo	2	Alum	Jul 2022
Ad26.Mos4.HIV, bivalent subtype C gp140 and mosaic gp140	NCT03964415	HVTN706/Mosaico	3	Alum	Mar 2024

^*Actual or predicted completion date

HVTN702 was preceded by the phase 1/2 HVTN100 trial (NCT02404311) that confirmed the safety and immunogenicity of the vaccine [42]. The immunogenicity outcomes were largely analyzed against factors identified as correlates of protection in the RV144 trial. All participants developed IgG titers specific to all gp120 components in the vaccine and had >3-fold improvement in median IgG binding titers compared to RV144 participants against gp120 antigens matched for both the ALVAC-HIV and gp120 protein vaccine components. Total IgG3 titers and percent responders to the vaccine-matched gp120s were also greater among HVTN100 participants, and antibody responses to V1V2 were observed. Again, while high anti-gp120 titers were elicited, neutralizing responses were not detected against tier 2 viruses. Relative to RV144, a greater proportion of HVTN100 vaccinated individuals also generated vaccine-matched Env-specific CD4 T cell responses, and with higher magnitude based on cytokine expression. On the basis of high immunogenicity to the vaccine antigens, and acceptable immune responses related to correlates of protection interpreted from the RV144 trial, the regimen progressed to the HVTN702 efficacy trial in 2016. Unfortunately, interim analysis demonstrated that an equivalent proportion of vaccine and placebo recipients became infected with HIV. The vaccine was deemed not efficacious and the study was halted early in February of 2020 [43].

3.2. HVTN705 (Imbokodo) and HVTN706 (Mosaico)

HVTN705/Imbokodo (NCT03060629) and HVTN706/Mosaico (NCT03964415) are two currently ongoing efficacy trials. Both trials are designed to assess an adenovirus vector prime with recombinant Env protein booster immunizations. Imbokodo is an ongoing phase 2b clinical trial among heterosexual female participants in Sub-Saharan Africa initiated in 2017, whereas Mosaico is a more recently started large scale multi-center trial, currently recruiting participants in eight countries, including the USA.

Several adenoviruses were tested to circumvent issues linked to preexisting Ad5 seroprevalence, of which adenovirus serotype 26 (Ad26) was one of the candidates developed as an improved recombinant Ad vector candidate [44]. The HVTN705 and 706 studies use an Ad26-vectored tetravalent vaccine (Ad26.Mos4.HIV) which contains equal parts of two distinct mosaic Gag-Pol Ad26 vectors and two mosaic Env Ad26 later followed by Env gp140 protein boosts (monovalent in Imbokodo, bivalent in Mosaico). Mosaic antigens are computationally recombined HIV sequences to optimize coverage of HIV protein sequence diversity [45] and designed to elicit stronger and broader T cell responses than consensus or single native isolate sequence constructs [46]. Assessment of Ad26 mosaic vaccines in rhesus monkeys in challenge studies demonstrated some protection against repeated heterologous SHIV challenges [47], which could be improved by the addition of protein Env gp140 boosts [48]. Protection in monkeys was associated with Env binding antibodies [49].

The tetravalent Ad26.Mos4.HIV vaccine improves upon the trivalent Ad26.Mos.HIV vaccine that was tested in the phase 1/2a APPROACH study (NCT02315703) and NHPs [50], and elicited superior antibody and antigen-specific CD4 T cell responses in the phase 1/2a TRAVERSE study compared to the trivalent vaccine [51]. The Ad26 tetravalent vaccine-Env gp140 (subtype C, 97ZA012) regimen did not elicit nAbs other than against the MW965.26 tier 1 virus. This was expected, as elicitation of tier 2 nAbs generally requires native-like Env trimer antigens. The vaccine does not express an Env design with mutations known to be required for native-like Env trimer folding and stability (see sections below on immunogen design). It generated gp120 V1V2-directed IgG binding responses and autologous subtype C Env-specific IgG3 in 66% of those that received four doses of the tetravalent vaccine [51]. Tier 2 neutralizing antibodies were not reported. The primary study completion dates for HVTN705 and HVTN706 are projected to be in mid 2022 and early 2023, respectively. The phase 1/2a Ascent study (NCT02935686) will test a similar regimen (Table 1).

4. Virus vectored and DNA vaccines currently in early phase trials

One major advantage of nucleotide and virus vectored vaccines compared to protein subunit vaccines is that they can elicit both CD8 T cell and humoral immunity. Vaccines with the primary goal of eliciting CD8 T cell mediated responses have focused on the discovery and utilization of better viral vector constructs. Moreover, early studies including clinical trials such as RV144, demonstrated that a heterologous prime-boost approach, specifically a vectored/DNA vaccine prime followed by protein boost, enhanced the magnitude and quality of immune responses relative to a single format vaccination approach. As such, many groups directed efforts to improve upon heterologous prime-boost approaches by optimizing both the DNA or viral vectored prime and protein boost components, including the development of novel Ad vectors. The HVTN702, HVTN705 and HVTN706 trials discussed above are a few examples of this strategy. The following sections describe additional recently initiated early stage trials evaluating DNA and viral vectored vaccines (Table 2).

Table 2.

List of recent DNA and viral vectored vaccine clinical trials

Immunogen	NCT Trial number	Other names	Phase	Adjuvant	Completion Date*
Prime: Ad4-Env CN54, MVA-CN54, Boost: CN54 gp140	NCT03408262	Ad4HIV	1	With protein boost: MPLA	Feb 2020
p24CE1/2 DNA, p55gag DNA	NCT03181789	HVTN119	1	IL-12 via DNA	Apr 2020
ChAdOx1.tHIVconsv1, MVA.tHIVconsv3, MVA.tHIVconsv4	NCT04586673	HIV-CORE0052	1		Oct 2021
Prime: Polyvalent DNA with subtype A/B/C/AE gp120 + Gag Boost: Tetravalent gp120 protein from subtypes A/B/C/AE	NCT03409276	HVTN124	1	With protein boost: GLA-SE	Nov 2020
ChAdOx1.HTI and MVA.HTI	NCT04563377	HIV-CORE0051	1/2a		Oct 2021
ChAdOx1.tHIVconsv1, MVA.tHIVconsv3, MVA.tHIVconsv4	NCT04553016	HIV-CORE006	1		Jun 2022
Prime: Ad4-Env145NFL or Ad4-Env150KN Protein boost: 4571 gp140 trimer	NCT03878121	19-I-0069	1	With protein boost: Alum	Apr 2024
HCMV vectors expressing Gag, Rev/Tat/Nef, Pol, Env	NCT04725877	VIR-1111–2001	1a		Dec 2024

^*Actual or predicted completion date

4.1. Replication-incompetent virus vectored vaccines

Certain simian Ad vectors were found to be nearly as immunogenic as the human Ad5 vector, whereas the majority of other human adenoviruses with rarer seroprevalence tended to be less immunogenic [52]. The chimpanzee adenovirus (ChAd) isolate Y25 derived virus vector ChAdOx1 [53], is used as the vector for the development of several HIV vaccines in trials started this year (Table 2). A ChAdOx1 vectored SARS-CoV2 vaccine developed recently was generally well tolerated, elicited substantial T cell responses and nAbs, and demonstrated ~70% efficacy against COVID-19 [54,55].

There are three planned clinical studies utilizing a ChAdOx1/MVA (modified vaccinia virus Ankara) vectored vaccine combination regimen designed to elicit HIV-specific T cell responses. These vectors have chimeric T cell epitope inserts, referred to as tHIVconsvX variants (NCT04586673 and NCT4553016) or HIVACAT T-cell immunogen (HTI) (NCT04563377). The inserts consist of short linear segments of epitopes from the HIV proteome that are largely conserved and associated with viral control in HIV-1 infected individuals [56–59].

HTI is a rationally designed mosaic polypeptide sequence from Gag, Pol, Vif, and Nef proteins [57]. In mice and monkeys, CD4 and CD8 T cell responses to multiple regions within HTI were observed [57]. The HTI vaccine progressed to a phase 1 clinical trial (NCT04563377) to evaluate if it can elicit CD8 T cell responses to protective epitopes in humans.

The first generation HIVconvs immunogen was tested in humans in several delivery platforms, including as a DNA vaccine, ChAdV63 and MVA vectored vaccines (HIV-CORE 002, NCT01151319). The vaccine elicited CD8 T cell responses against multiple regions, and CD8 T cells from vaccinated individuals could inhibit autologous CD4 T cell infection by viruses from multiple clades to varying degrees [60]. The second generation tHIVconsvX constructs were modified to maximize inclusion of beneficial epitopes identified in HTI. The numbered tHIVconsvX constructs differ in the order in which the linear epitopes are stitched together, and denote two different mosaic variants that account for two most common epitope variants within the covered regions [58]. Two additional studies will evaluate these constructs. The first will compare responses in participants that only receive the ChAdOx1.tHIVconsv1 vaccine to those that receive ChAdOx1.tHIVconsv1 along with MVA.tHIVconsv3 and MVA.tHIVconsv4 (NCT04586673). In a second, larger study, participants will receive ChAdOx1.tHIVconsv1 as the initial vaccine dose followed by MVA.tHIVconsv3 and MVA.tHIVconsv4 at a later time point. These studies do not have an Env component and therefore will not rely on antibody responses for protective efficacy. These studies will be proof-of-concept studies for eliciting human T cell responses to conserved HIV epitopes with these vaccine formats.

4.2. Replication-competent virus vectored vaccines

There are two replication-competent Ad4 vectored vaccine studies for HIV: Ad4HIV (NCT03408262) taking place in the UK and 19-I-0069 (NCT03878121) in the US. A replication competent Ad4 vectored H5N1 influenza vaccine in humans [61], and an Ad4 vectored Env vaccine derived from subtype C strain 1086 in mice [62] elicited vaccine-specific T cell responses, and antibody titers were observed in response to Ad4 vaccine immunization followed by boost immunization with inactivated H5N1 vaccine in humans [61] or Env gp140 in rabbits [62]. The Ad4 study will evaluate multiple different combinations of an orally administered Ad4 vectored CN54 Env (Ad4-Env CN54) prime, followed by boosting with replication-incompetent MVA vectored CN54 Env (MVA-CN54) and/or recombinant protein CN54 gp140, to assess which boosting option can generate most optimal protective antibody responses. These vaccines do not encode HIV Env with stabilizing mutations known to express as native trimers (see sections below).

The 19-I-0069 study assesses the safety and immunogenicity of the priming of two different Ad4 vectored vaccines, Ad4-Env145NFL and Ad4-Env150KN, followed by boost with soluble protein Env trimer 4571 (described in section 5.1). The Env150KN construct encodes an Env that has a mutated cleavage site [62], whereas Ad4-Env145 NFL encodes a native-flexibly linked (NFL) Env trimer. NFL trimers are not cleavage dependent but harbor a long linker between gp120 and gp41 that has been demonstrated to allow for quaternary folding that reflects the structure of native cleaved Env trimers [63]. NFL Env trimers can elicit tier 2 nAb responses in animal models, consistent with other Env immunogens with proper native Env trimer structure (see section 5.1).

4.3. DNA Vaccines

The HVTN119 study (NCT03181789) features a Gag-directed vaccine designed to elicit protective CD8 T cell responses. Like the rationale for developing conserved and mosaic inserts in the ChAdOx vaccines, the p24CE vaccine was designed such that only epitopes derived from cross-clade conserved elements (CE) in p24^gag are incorporated in the DNA vaccine. Mice and NHPs immunized with the p24CE DNA vaccine were more likely to develop CD8 T cell responses and overall T cell responses to CE epitopes relative to those immunized with full length p55^gag DNA [64,65]. This suggests that the full length p55^gag has immunodominant epitopes distracting responses away from conserved CE epitopes. The prime-boost regimen involving a p24CE DNA prime followed by p24CE+p55^gag DNA boost was defined in a subsequent NHP study[66], where a dual p24CE construct (p24CE1 and p24CE2), each expressing two common sequence variants of the CE epitopes, was made. The study also incorporates an IL-12-expressing DNA plasmid as the adjuvant, which has been suggested to improve T cell responses in a dose dependent manner in some human DNA vaccination studies [67–69], but not in others [70]. As in the HTI and tHIVconsX construct immunization studies, the primary question to be answered in a human phase 1 trial would be whether this vaccination strategy can generate T cell responses to conserved, broad epitopes in humans.

A previous phase 1 clinical trial DP6–001 (NCT00061243) in the early 2000s assessed a vaccine consisting of three doses of a hexavalent (pentavalent gp120 from multiple subtypes and monovalent subtype C Gag) DNA prime followed by one or two pentavalent gp120 protein boosts [71]. In this study, while high anti-gp120 titers were observed, T cell responses measured by IFNg production were low, except in the high DNA prime dose group. Overall, there were low neutralizing serum titers to the 5 autologous viruses used in the vaccine. With the expansion of available curated Env sequences, the original DP6–001 formulation was improved upon by screening additional Envs in rabbits and selecting gp120s that induced the greatest breadth, albeit at extremely low titers (1:10) [72]. Four of the original five gp120 sequences in the DP6–001 vaccine were replaced. The newly formulated vaccine induced improved neutralizing titers and V1V2 biding Ab titers relative to the original DP6–001 regimen in rabbits. In the same study, bioinformatics analysis suggested that a tetravalent formulation consisting of their newer subtype A, B, C, and AE gp120 constructs were sufficient to cover consensus sequences of all HIV-1 subtypes. The HVTN124 (NCT03409576) is designed to assess the updated tetravalent gp120-based vaccine formulation.

5. Innovation in Env-based immunogen design

As the above efficacy trials have shown, previous Env protein vaccine candidates have been unsuccessful in eliciting nAbs against tier-2 clinical isolates. Assessment of past trials strongly indicates that tier 1 neutralizing activity and Fc-mediated nnAb effector functions do not consistently correlate with protection in humans. More broadly reactive antibodies against tier 2 viruses is likely to be necessary for robust protection by antibodies. In the last decade, a plethora of bnAbs targeting distinct epitopes on Env have been isolated from chronically infected HIV-1 patients, several of which demonstrate extraordinary breadth and potency [12].

Elicitation of bnAbs by vaccination remains a significant challenge. Longitudinal and antibody development pathway prediction studies have a large number of mutations are required before most bnAb clonal lineages develop breadth [12]. Some bnAbs and their developmental intermediates can be polyreactive [73–76] which may lead to negative selection. Further, when the human B cell repertoire was directly probed for the presence of naïve precursors to specific bnAb-classes such as those in the VRC01-class and BG18-class (described in further detail in section 5.2), it was found that bnAb precursor naïve B cells are present in most people, but they are rare [77–79]. Hence, the activation of the proper bnAb precursor B cells by vaccination is a major hurdle to overcome. The current generation of immunogen designs take these various factors into account, to varying degrees, in attempting to elicit bnAbs, discussed below.

5.1. Env trimer immunogens

Presenting the Env molecule in conformations other than the native, prefusion form often drives non-neutralizing responses directed towards relatively glycan-free surfaces, which are not broadly neutralizing sites [9,10]. Many early recombinant gp140 trimers incorporated a non-cleavable mutation in the native furin cleavage site to prevent dissociation of gp120, but these trimers do not elicit potent tier 2 neutralizing antibodies [80–82]. It was later described that in uncleaved trimers, the gp120 subunits adopted open conformations, exposing non-neutralizing epitopes similar to those on monomeric gp120 subunit proteins[83,84]. As a result, intensive efforts led to the design of soluble Env trimers that stably adopt the prefusion, closed conformation.

One prefusion trimer that garnered excitement was the SOSIP.664 trimer. This trimer retains the furin cleavage site, but incorporates an engineered disulfide bond between the gp120 and gp41 subunits (SOS) to prevent subunit dissociation, along with a I559P helix-breaking mutation in gp41 heptad repeat 1 (HR1) that helps maintain the trimer in the prefusion conformation [85–89]. The SOSIP.664 trimer based on the sequence of a subtype A virus BG505 [90] was particularly stable [91] and demonstrated an excellent antigenicity profile – binding bnAbs while having reduced or minimal binding to nnAbs [92]. The stable BG505 SOSIP.664 trimer allowed for the first set of high-resolution structures to be determined for the complete pre-fusion Env ectodomain[87–89], which was later shown to be nearly identical to a wildtype Env trimer [93]. The SOSIP stabilizing mutations were also shown to be compatible with several other Env variants [6,94–96], and results in a glycosylation profile similar to membrane-expressed Envs [97–100]. Importantly, BG505 SOSIP.664 was the first Env protein vaccine to elicit robust tier 2 neutralizing antibodies in animal models (rabbits and NHPs) [101], promoting the development of a myriad of prefusion stabilized soluble native-like Env trimers. Several preclinical animal models using various SOSIP-type recombinant trimers have reproducibly elicited tier 2 neutralizing antibody responses [101–105]. Prefusion Env trimer immunogens continue to undergo further development to tackle issues related to in vivo immunogen stability and non-bnAb class immunodominant responses, as will be discussed below. Several Env trimer immunogen candidates are currently in phase 1 clinical trials for initial immunogenicity assessment in humans.

Multiple phase 1 clinical trials are testing soluble prefusion Env ectodomain trimers (Table 3), including the original BG505 SOSIP.664 trimer (IAVI W001, NCT03699241). The formulation of the vaccine in the IAVI W001 trial uses the MPLA/QS-21 adjuvant AS01_B, which has been used in the development of the RTS,S malaria vaccine [106] and the recent very successful herpes zoster vaccine [107]. A second clinical trial involving the BG505 SOSIP.664 trimer proposes to test the immunogen formulated with various TLR agonists (HVTN137, NCT04177355). Numerous HIV Env vaccination preclinical studies discussed in this section have been performed with ISCOM-type adjuvants containing saponin, which has not yet been licensed for use in humans. However, Novavax recently reported a successful COVID-19 vaccine using their ISCOM-type adjuvant, Matrix-M1, which was safe and well tolerated [108].

Table 3.

Ongoing and planned clinical trials to assess soluble Env trimer protein based vaccines

Immunogen	NCT Trial Number	Other Trial names	Phase	Adjuvant	Completion Date*
C.6980 gp145 trimer	NCT03382418	HVTN122	1	Alum	Sep 2019
BG505 SOSIP.664 gp140 trimer	NCT03699241	IAVI W001	1	AS01B	May 2020
4571 gp140 Trimer	NCT03783130	19-I-0031	1	Alum	Aug 2020
BG505 SOSIP.664 gp140 trimer	NCT04177355	HVTN137	1	Comparison of 3M-052-AF, CpG 1080, GLA-LSQ, and Alum	May 2022
ConM SOSIP.v7 gp140 trimer	NCT03961438	ACTHIVE-001	1	MPLA Liposome	Nov 2022

^*Actual or predicted completion date

Early SOSIP.664 immunization studies in animals indicated that SOSIP-type native-like trimers still underwent some amount of degradation or exhibited conformational breathing[6], based on the observation that one of the immunodominant sites being targeted by Env antibodies after SOSIP immunizations, was the V3 loop [101,103,109]. This epitope is concealed on the prefusion Env trimer; the tip of the V3 loop normally becomes exposed upon conformational opening of the Env trimer following binding to CD4 [110–112]. Certain V3 loop Abs are capable of neutralizing tier 1 viruses and can demonstrate breadth against open tier 1 trimers, but V3 loop Abs are not potent against tier 2 HIV isolates [4,113] where Env is less prone to transiently expose the coreceptor binding site [5]. To ameliorate this issue, the next round of trimer designs opted to further stabilize the prefusion conformation of the trimer [102,114–116].

The 4571 trimer, also referred to as DS-SOSIP, is a stabilized BG505 SOSIP.664-derived gp140 trimer currently in a phase 1 trial (NCT03783130). The addition of a 201C-433C intramolecular disulfide bond in gp120 reduced V3 loop exposure and V3 loop mAb binding in vitro, even in the presence of soluble CD4 [117]. However, immunization of guinea pigs with the 4571 trimer and further stabilized variants of DS-SOSIP showed that majority of the DS-SOSIP variants elicited similar levels of V3 directed responses as the original BG505 SOSIP.664 trimer. Neutralization to the autologous BG505 virus trended towards improvement among the DS-SOSIP variants relative to the original BG505 SOSIP.664 [118]. Heterologous tier 2 neutralization was not observed, consistent with most SOSIP-type Env trimer immunization studies. A different Env trimer design have instead incorporated CD4 binding site (CD4bs)-knockout mutations [114] to prevent trimer opening upon engaging CD4 present on CD4 T cells.

A second concern observed among Env trimer immunized animals, including SOSIP-type trimers, has been that immunodominant autologous neutralizing responses commonly occur to strain-specific ‘glycan holes’. Immunization with a single Env trimer, or sequentially/co-immunizing with a mixture of Env trimers derived from different viruses, commonly elicits responses to strain-specific glycan holes [105,119–122]. Most N-linked glycosylation sites are conserved across all HIV isolates, but certain viruses exhibit strain-specific loss of glycosylation sites. The missing glycan opens up a more easily targetable protein epitope, and much of the tier 2 autologous responses to Env trimers have been observed to be directed to these strain-specific glycan holes [102,104,105,123,124]. To counter the issue of isolate-specific immunodominance, an HIV-1 group M consensus Env sequence derived SOSIP.664 trimer was developed [125]. This ConM SOSIP.v7 trimer is under phase 1 clinical trial (NCT03961438). In addition to its consensus Env sequence, the ConM SOSIP.v7 trimer also incorporates several mutations to enhance trimer stability. Preclinical studies were carried out in rabbit and rhesus monkeys, with the latter being immunized with a ferritin nanoparticle multimerized version of the Env trimer immunogen. In both studies, animals did not develop tier 2 neutralizing antibodies [125]. Reduced V3 loop directed serum Ab responses as well as reduction in heterologous tier 1 virus neutralization were observed, suggestive of stabilizing mutations incorporated in the trimer design successfully limiting the exposure of some of these undesired epitopes. Because the BG505 Env trimer has well characterized glycan holes, comparisons between ACTHIVE-001, IAVI W001, and 19-I-0031 studies will inform the impact of glycan hole driven immunodominance in human B cell responses.

The development of cleavage-independent stabilized prefusion trimers has also been accomplished. NFL trimers [63,126,127] (used in NCT03408262) and uncleaved prefusion-optimized (UFO) trimers [128] incorporate a flexible linker between the gp120 and gp41 subunits in place of the furin cleavage site. Like SOSIP, these designs also have a proline mutation in gp41 HR1. These trimers also have demonstrated the capability to elicit tier 2 autologous neutralizing responses [102,129,130].

A C.6980 gp145 trimer [131] is the sole Env trimer protein-only vaccine currently in phase I trial (NCT03382418) that is not designed with prefusion Env stabilizing mutations. This trimer design was based on a subtype C Env sequence from viral isolate CO6980v0c22 to account for subtype C strains dominating Sub-Saharan Africa. It remains uncertain whether this trimer truly adopts the prefusion conformation, as its antigenicity profile shows binding to 17b, a V3 mAb, while the binding of quaternary-specific bnAbs could not be tested due to the fact the parent virus strain is resistant to neutralization by prefusion epitope specific antibodies such as PGT145 [132] and PGT151 [133,134]. A substantial proportion of the C.6980 Env trimer exhibited non-native conformations when analyzed by electron microscopy [131]. Assessment of immunogenicity in rabbits only demonstrated neutralization of tier 1 isolates; no tier 2 neutralization was observed [131].

The immunogenicity of near-native Env trimers in humans is unknown, but should be reported in the near future. Of note, none of these initial Env trimer clinical trials are expected to elicit heterologous (broad) tier 2 responses. The primary scientific goal of these trials is to answer whether the tier 2 autologous nAb responses observed in NHPs, rabbits and guinea pigs can be recapitulated in humans. It will also be of interest to elucidate potential issues related to epitope response hierarchies. For example, the Env trimers used in these trials do not integrate strategies to reduce responses to the base of the trimer. The Env trimer base is a non-neutralizing set of neoepitopes arising from the truncation of the trimer at the ectodomain, and it has been shown that the base is highly immunodominant in animal models [9,104,109,124,135,136].

5.2. BnAb germline-targeting immunogens

One strategy to overcome immunodominance is to target the activation of specific bnAb-class naïve B cells in the priming immunization. This is a challenging task. In addition to bnAb precursor B cells being rare, many inferred germline (iGL) variants of bnAbs typically do not bind wildtype Env immunogens [137–139]. Germline-targeting (GT) immunogens [140] are specifically engineered such that certain classes of iGL bnAbs can bind with much higher affinity than they would to a wildtype gp120 or Env trimer protein. In this manner, the goal of GT priming is to initiate activation and proliferation of specific precursor bnAb-class B cells. Subsequent immunizations then would use a series of immunogens that sequentially become more representative of native Env trimer to shepherd maturation of the antibody response to full HIV bnAb development [141]. Two GT immunogens are currently in phase 1 trial (Table 4), along with several additional promising GT immunogens in preclinical development, discussed below.

Table 4.

Phase 1 clinical trials testing epitope- and lineage-targeting vaccines

Immunogen	NCT Trial Number	Other Trial names	Phase	Adjuvant	Completion Date*
eOD-GT8 60mer	NCT03547245	IAVI G001	1	AS01B	Dec 2020
MPER-656 liposomes	NCT03934541	HVTN133	1	Alum	Jun 2021
BG505 SOSIP GT1.1 gp140 trimer	NCT04224701	IAVI C101	1	Unspecified	Jul 2022
CH505TF gp120	NCT04607408	HVTN135	1	GLA-SE	Oct 2022
Longitudinally derived CH505TF gp120 variants	NCT03220724	HVTN115	1	GLA-SE	Jul 2023

^*Actual or predicted completion date

GT immunogens to the CD4bs have particularly garnered strong interest, because highly potent CD4bs bnAbs, referred to as the VRC01-class of bnAbs, have been isolated in several independent donors. Equally important, VRC01-class bnAbs have a very well-defined set of B cell receptor (BCR) characteristics. Namely, VRC01-class bnAbs are defined by the use of a IGHV1–2 heavy chain V gene and a 5-amino acid residue light chain complementarity determining region (CDR) 3 [142–144]. VRC01-class bnAb targeting immunogen. eOD-GT8 is a minimalistic CD4bs presenting immunogen consisting of an engineered outer domain (eOD) of gp120. The eOD-GT variants were designed to have extremely high affinity to multiple VRC01-class iGL antibodies [77,138]. The eOD-GT8 immunogen has been successfully used as a probe to isolate authentic VRC01-class naïve peripheral B cells from healthy donors [77,78] (unlike the earlier iterative design, eOD-GT6, with lower affinity to iGL antibodies than eOD-GT8) [77], demonstrating that eOD-GT8 can selectively engage VRC01-class naïve B cells in a large pool of irrelevant B cells in humans. Immunoglobulin (Ig)-humanized mice [145] and various VRC01-class precursor B cell receptor (BCR) knock-in (KI) mice [146–154] were all capable of generating VRC01-class B cell responses upon immunization with eOD-GT8 60mer and related eOD-GT variants. VRC01-class responses elicited by eOD-GT 60mers in adoptive transfer mouse models that recapitulate VRC01-class naïve B cell precursor frequencies and affinities represented in humans [149,153], was an indicator supporting that eOD-GT8 60mer is likely to be successful in activating VRC01-class precursor B cells in humans. One recent study using BCR KI mice demonstrated that VRC01 precursors could be matured to breadth by eOD-GT8 60mer priming followed by sequential shepherding [155].

A phase 1 study wherein healthy participants were immunized with eOD-GT8 60mer adjuvanted with AS01_B (NCT03547245 / IAVI G001) was initiated in 2018 in the US, to assess whether eOD-GT8 could prime VRC01-class naïve B cells in humans. The preliminary outcome of the study presented at the HIVR4P meeting (Feb 2021) showed robust priming of VRC01-class naïve B cells in all vaccine recipients, demonstrated by an increase in the frequency of eOD-GT8 binding B cells, a large proportion of which harbored IGHV1–2 utilizing BCRs paired with 5-AA LCs [156]. This is the first clinical trial that confirmed bnAb-class precursor B cells can be primed and expanded in healthy humans, and validated that rationally designed GT-immunogens can target a specific set of rare B cells.

A second GT immunogen in phase 1 trial is the BG505 SOSIP GT1.1 trimer (NCT04224701), a slightly modified version of BG505 SOSIP GT1. The BG505 SOSIP GT1 trimer is a BG505 SOSIP.664 based Env trimer that incorporates GT mutations to target two distinct bnAb epitopes, the Env CD4bs and the V1V2 apex [160]. Affinity binding assessment showed enhanced binding of iGL CD4bs antibodies to the GT trimer and a moderate improvement in binding of iGL V1V2 apex antibodies. Germline VRC01 heavy chain KI mice (VRC01^gH) that were immunized with BG505 SOSIP GT1 generated CD4bs-directed antibodies that were not observed in mice immunized with the regular BG505 SOSIP trimer. In this study, the CDbs-specific B cells were not directly sequenced to confirm that they were indeed VRC01-class B cells. The affinity (K_D) of iGL VRC01 to BG505 SOSIP GT1 is much weaker than to eOD-GT8. It is important to note that authentic human derived VRC01-class naïve B cells have much lower affinity to the GT immunogens than the iGL mAbs used in the design [77,78]. For example, iGL CD4bs mAbs have an average K_D in the range of 1–10 nM to eOD-GT8 monomer, but authentic VRC01-class naïve B cell derived mAbs bind eOD-GT8 with a K_D in the 1–10 μM range. As germline affinity has been implicated to be an important factor in rare B cell activation [149], the comparison of the ability of BG505 SOSIP GT1.1 versus eOD-GT8 to activate target precursor B cells in humans will enlighten GT immunogen affinity threshold requirements.

426c core is another VRC01-class GT immunogen based on the gp120 sequence of a clade C isolate 426c. It consists of both the inner and outer domains of 426c gp120, with V1V2V3 along with 3 N-linked glycosylation sites at Env N276, N460 and N463 removed [161]. This immunogen in both 4–7mer and 24mer formats were also shown to be capable of activating VRC01 B cells in VRC01^gH mice [162], and is planned for evaluation in future clinical studies [152].

The three GT immunogens discussed above all target the Env CD4bs, but promising designs have been published for other broadly neutralizing epitopes. One class of highly potent bnAbs bind in the high-mannose patch of the Env trimer, many of which recognize a highly conserved N-linked high-mannose glycan at position N332 at the base of V3 [132,163–166]. This N332-glycan centric epitope is commonly called the N332 supersite, or the V3-glycan bnAb epitope. N332 bnAbs utilize a long heavy chain CDR 3 (HCDR3) loop, typically 20 amino acid residues in length or greater. The BG505–10MUT trimer [167] was designed to bind lineage reverted variants of PGT121, a highly potent N332 bnAb [132,168]. BG505–10MUT successfully primed PGT121^gHL KI mice, where a shepherding sequential immunization approach was able to drive breath inducing SHMs in PGT121^gHL B cells [169]. This was the first successful demonstration of bnAb development after immunization [167,169]. Following the discovery of BG18, an N332 bnAb similar to PGT121 but without indels [163], a series of iteratively designed BG505–10MUT immunogens were developed to bind a diverse array of BG18-like germline antibodies, yielding the N332-GT Env trimer series of immunogens [79]. Of these, N332-GT2 Env trimer multimerized as an 8mer was able to activate BG18 precursor B cells in mice in a low precursor frequency adoptive transfer mouse model using BG18^gH cells [79]. N332-GT1 and N332-GT2 were successfully used to isolate BG18-like naïve precursors from healthy human donors as well[79], signifying that this GT immunogen, like eOD-GT8, would be a promising clinical candidate. However, the naïve precursor frequency of BG18-like B cells was found to be lower than that of VRC01-class B cells [78,79], thus the priming of N332 bnAb-class B cells in humans is expected to be more challenging than VRC01-class bnAbs.

An Env trimer immunogen called RC1 [170] that targets the N332 site was designed from a BG505 11MUTB base [167], of the BG505–10MUT series of Env trimer immunogens. In this design, the N156 glycan near the N332 site was removed to open up V1-proximal regions around the N332 epitope. Additional modifications were made to generate RC1–4fill, which added N-linked glycans outside of the N332 epitope to reduce other glycan-hole specific responses. Wildtype mice, rabbits and NHPs immunized with RC1 and RC1–4fill trimers (trimers in mice, AP205 VLP presented multimers in rabbits and NHPs) elicited responses targeting the N301 and N332 glycans and the underlying GDIR peptide epitope [168,171,172] of RC1 trimer [170]. Nevertheless, most of the antibodies possessed HCDR3 of average lengths, unlike N332 supersite bnAbs, and thus it remains unknown whether responses to RC1 can be shepherded to develop neutralizing breadth.

5.3. Epitope-focusing immunogens

GT immunogens have the advantage of potentially exhibiting very high affinity to specific classes of bnAb precursors, but may be restrictive in the set of precursor B cells they can activate. On the other hand, epitope-focusing immunogens are designed to target a more diverse repertoire of naïve precursor B cells to a given bnAb epitope. The N156 glycan deletion RC1 immunogen discussed in the above section, for example, was introduced to broaden the range of targeted B cells. This strategy in theory would allow for the priming immunogen to activate a variety of clonal lineages that can bind a given bnAb epitope. All key broadly neutralizing epitopes on Env are currently being targeted by either GT or epitope-focusing strategies, and this section covers epitope-focusing immunogens targeting regions outside of the CD4bs and N332 site.

The membrane proximal external region (MPER) of Env is one of the earliest identified bnAb epitopes. Unlike other bnAb epitopes that consist of a tertiary or even quaternary surfaces for recognition, MPER directed antibodies recognize highly conserved, non-glycan dependent, simple linear peptide segments consisting of the C-terminal end of the Env trimer ectodomain [173–180]. Nonetheless, it is a challenging epitope in vaccine design. One reason is its proximity to the lipid membrane; as a result, both the epitope itself and MPER-directed bnAbs are hydrophobic. MPER bnAbs often engage the phospholipid bilayer along with the MPER epitope, driving polyreactivity exhibited by several MPER bnAbs [73,177,178]. As such, it has been proposed that elicitation of MPER bnAbs may require overcoming tolerance mechanisms [181,182]. One MPER-focused vaccine is the MPER-656 liposome, which consists of gp41_656–680 peptide anchored on liposomes via an added hydrophobic peptide sequence [183]. Immunogenicity studies in guinea pigs and rhesus monkeys demonstrated that while MPER-656 liposomes were able to elicit MPER-specific responses, they were unable to elicit 2F5-like (an MPER bnAb) responses alone; a recombinant subtype B JRFL gp140 protein containing MPER and MPER-656 liposomes had to be used together in a prime-boost regimen in order to elicit 2F5-type responses [184]. In a follow up NHP study, a much more aggressive immunization regimen administering multiple immunogens (a combination of recombinant vaccinia virus expressing JRFL gp140, recombinant JRFL gp140 protein, and MPER-656 liposomes mixed with several TLR agonists) still could not break tolerance during affinity maturation, leading to selection against 2F5-like features that allow lipid recognition [185]. HVTN133 is an ongoing phase 1 clinical trial (NCT03934541) that looks to evaluate MPER-656 liposomes in humans.

The V1V2 apex bnAb epitope is a quaternary epitope formed at the Env trimer three-fold axis, and encompasses the V1V2 region from multiple gp120 protomers, also referred to as the trimer apex. As such, V1V2 bnAbs have a stringent Env trimer binding dependency [91,186–189] and are distinct from the RV144 elicited V1V2 nnAbs that bind monomeric V1V2 constructs. Due to its complex epitope specificity, apex bnAb-targeting immunogens have focused on recapitulating the quaternary structure of the Env trimer [190–193]. Several Env isolates have been shown to be neutralized by germline reverted apex bnAbs [190,193]. SOSIP trimers derived from V1V2 iGL bnAb responsive Env sequences [190], or chimeric BG505 SOSIP trimers transplanted with V1V2 regions of those particular Envs [193] are potential immunogen design strategies to elicit V1V2 apex bnAbs. One interesting apex targeting immunogen is an SIV Env SOSIP.664 trimer named MT145K [192]. This immunogen was generated based on the observation that several chimpanzee infecting SIV isolates were cross-neutralized by apex binding bnAbs, whereas they failed to be neutralized by bnAbs targeting other broadly neutralizing epitopes [194]. The MT145K SIV Env trimer retains the epitope specificty for the majority of known HIV V1V2 apex bnAbs and binds iGL variants of several V1V2 bnAbs with measurable affinity, without showing antigenicity at other bnAb epitopes. Theoretically, the only shared epitope between the SIV and HIV Env trimers is the V1V2 apex and therefore, MT145K primed V1V2 targeting B cells can be selectively boosted by subsequent immunizations with HIV Env trimers. In CH01^gH (apex-class bnAb) KI mice, two immunizations with the MT145K SOSIP followed by a boost consisting of a trivalent cocktail of iGL V1V2 bnAb binding Env trimers elicited heterologous serum neutralizing antibody titers in a subset of immunized animals [192]. Still, because all V1V2 bnAbs have long HCDR3s including those that have greater than 30 amino acid residues [132,186,195], apex-class bnAb precursors are expected to be rare among mature naïve human B cells. Thus, the impact of precursor frequencies will be a key question that needs to be answered for the viability of this bnAb epitope for HIV vaccine development.

The fusion peptide (FP) of Env is an epitope that has garnered strong interest for several reasons. Like MPER, the FP has high sequence conservation between cross-clade HIV isolates and is a relatively simple epitope primarily consisting of a linear peptide sequence [196]. Several FP binding bnAbs have long HCDR3s [93,197], but length is not a requirement for FP binding or breadth [196]. Of note, a FP targeting vaccine strategy employing several sequential immunizations of FP of varying lengths conjugated to a keyhole limpet hemocyanin (KLH) carrier protein, followed by boosting with BG505 SOSIP.664 trimers is the only vaccination regimen to date that has been able to reproducibly elicit tier 2 heterologous neutralizing antibodies in wildtype mice, guinea pigs and rhesus monkeys [198–200]. While FP mAbs with breadth were isolated, serum antibody titers in immunized monkeys did not have neutralizing breadth against a broad panel of tier 2 strains [198,199]. Thus, FP-specific memory B cells were generated by the immunization, but additional work is needed to drive high neutralizing antibody titers. A human clinical trial evaluating these immunogens is likely to be initiated in 2021 (VRC019 trial) [201].

5.4. Lineage-targeting immunogens

Longitudinal tracking of HIV infected patients from early infection to development broad serum neutralization breadth has made it possible to track viral evolution and the maturation of bnAb lineages from their most likely unmutated common ancestors (UCA) [75,76,164,165,188,195,202–204]. One strategy in designing priming and sequential immunogens is to derive them from Env sequences that likely started specific bnAb lineages in patients [205]. For example, the CH103 and CH235 are two cooperating CD4bs bnAb lineages identified from a longitudinal study of an African patient infected with a clade C virus [75,76,204]. In the studies, the UCA of the two bnAb lineages and the CH505 transmitter/founder (T/F) virus that likely started the antibody lineages were identified. Unlike most iGL antibodies that cannot bind heterologous wildtype Env proteins, the UCA of CH103 and to a weaker extent, CH235, bound CH505 T/F gp120, suggesting that CH505 T/F Env may be suitable as an immunogen capable of initiating CH103-lineage type bnAbs. In NHPs repeatedly immunized with the CH505 T/F gp120, or sequential immunization with CH505 T/F gp120 along with CH505 lineage gp120s derived from additional later time points, did not result in tier 2 neutralizing antibody tiers to CH505 TF, nor B cells with CH103-like HC-LC pairs possibly owing to a lack of sufficient LC V gene homologs in NHPs [206]. A second study that used a CH505T/F gp120:BG505 gp41 chimeric SOSIP Env trimer did not elicit CD4bs tier 2 neutralizing responses in NHPs or rabbits [207]. A third study used a ferritin nanoparticle approach to generate a CH505 trimer-based immunogen with 8 Env trimers per nanoparticle. In NHPs, this approach elicited high tiers of Abs capable of neutralizing a glycan modified CH505 virus (GnT-I^−/−, expressing all high-mannose glycans) or G458Y mutated CH505 [208].

There are currently two phase 1 studies evaluating recombinant CH505 T/F gp120 protein as immunogens. HVTN115 (NVT03220724) is designed to test a vaccination regimen similar to a previous NHP study by Wilton et al. [206], comparing repeated immunizations of a single CH505 T/F gp120 immunogen to sequential immunization of CH505 T/F gp120 variants derived from progressive infection time points. A second study, HVTN135 (NCT04607408) proposes to evaluate safety and immunogenicity of CH505 T/F gp120 in infants, possibly based on the observation that infants infected with HIV may generate bnAb responses more often and quicker than adults [209–211]. However, both of these clinical trials use monomeric recombinant gp120 constructs as the immunogen.

6. Delivery strategies

Multiple immunogen candidates continue to be improved, particularly those based on near-native Env trimers. Yet in most cases, animal studies indicate that a good immunogen design alone does not guarantee elicitation of cross-neutralizing responses, or even autologous tier 2 Abs. As such, optimized immunization strategies may be required, using insights from immunology and vaccinology. In this section, we outline immunization strategies that may be apt for development in the administration of upcoming HIV vaccine immunogens in humans.

6.1. Long-lasting antigen retention strategies

In order for B cells to accrue a high number of SHMs, it is imperative that the immunogen and immunization strategy employed induce a robust germinal center (GC) response. Slow but sustained antigen delivery mimics the kinetics following an infection; slow delivery immunization strategies lead to prolonged presentation of antigens by antigen presenting cells (APCs), which enhances GC responses [212]. In several proof-of-concept mouse and NHP studies, the use of non-mechanical osmotic pumps (OP) that steadily release immunogens over time outperformed equivalent doses of conventional bolus injections in terms of the magnitude of the GC response [102,104,109,213]. In NHPs, OP immunized animals also had increased Env-specific GC T follicular helper (T_FH) cells, BG505 Env binding titers, and elicited robust autologous neutralization titers [102,104]. A second slow delivery strategy is dose escalation (DE). In this strategy, one full vaccine dose is divided across multiple injections, gradually increasing the quantity of antigen given per successive injection [213]. Like animals immunized via OPs, NHPs immunized by DE had superior GC responses, Env specific serum binding, and autologous tier 2 neutralization compared to conventional immunization [104]. Notably, NHPs immunized by both slow delivery methods developed antibody responses targeting a wider range of distinct epitopes, whereas conventional immunization resulted in limited Ab epitope diversity, largely targeting the immunodominant trimer base and a glycan hole [104]. Thus, slow or extended antigen delivery strategies are promising for driving improved neutralizing antibody responses.

A third innovation for delivery is the use of phosphoserine (pSer) linkers conjugated to the protein antigen of choice. It was demonstrated that conjugated pSer moieties on protein immunogens results in strong adsorption of the immunogen onto aluminum hydroxide via ligand exchange between the phosphate groups and hydroxyls on alum [214]. This resulted in a true alum mediated depot mechanism, with antigen being retained up to two weeks at the injection site. Further, the pSer-Ag:alum complex promoted co-uptake of alum and immunogen by B cells, and enhanced the upregulation of genes important for interactions with cognate CD4 T cells [214]. An important feature that distinguishes this strategy from the other slow delivery strategies is that the presented protein orientation can be controlled by the location of the pSer-linker. Access to the immunodominant base of the Env trimer was secluded by placing the pSer-linker at the C-terminus of Env, which led to substantially reduced base directed responses in rabbits immunized with pSer-Env [214]. As alum has been long approved for clinical use, and the pSer-linker can be used with any protein immunogen by the introduction of a free cysteine, this approach is extremely in the development of various vaccine candidates. Immunogenicity of both pSer-conjugated Env trimer and several other immunogens are currently being assessed in NHPs.

6.2. RNA Vaccines

The shepherding strategy to drive bnAb development by immunization involves multiple different immunogens. There are numerous protein immunogens currently in development, and each proposed prime-boost schedule will involve a set of distinct immunogens that needs to be produced in accordance with GMP standards. Not only is this costly, but it also involves numerous approval and production steps for each protein vaccine to be tested in clinical trials. Synthetically produced nucleic acid vaccines bypass these issues, by utilizing a standardized manufacturing pipeline. As a testament to the power of RNA vaccines, both the Moderna RNA-1273 [159] and Pfizer BNT162b2[158] SARS-CoV2 vaccines were developed and progressed to efficacy trials at record speeds and exhibited ~95% efficacy. RNA based HIV vaccines like DNA and viral vectored vaccines, have the advantage of being able to elicit robust cellular responses [215,216] in addition to humoral responses. Although no RNA vaccines in the HIV field have progressed to clinical trials, several Env based RNA vaccines have shown to generate robust antigen-specific responses [157,216–218]. Further exploration of RNA vaccines expressing state-of-the-art immunogen constructs are expected to be forthcoming.

7. Conclusions

An HIV vaccine may still be years away, but many noteworthy recent advances have been made that have generated renewed optimism. Various vaccine candidates continue to be created and evaluated, particularly by employing novel immunogen design approaches, delivery methods, and adjuvant formulation.

Development of SOSIP-type stabilized soluble Env trimers provided huge momentum in preclinical animal studies. Compared with the previous generation of Env-based vaccines such as monovalent gp120 or uncleaved gp140 or gp160 trimers, near-native Env trimer immunogens more routinely elicit autologous tier 2 neutralizing antibody responses in animal models. Still, immunodominance to autologous neutralizing epitopes or non-neutralizing epitopes is a key concern, and weak tier 2 breadth has only sporadically been induced in animal immunizations. It remains unknown which epitopes will be targeted in humans. The results of the upcoming Env trimer protein immunization trials will strongly inform the direction of future vaccine designs and delivery methods, whether it be increased focus on GT and epitope focusing immunogens, even greater trimer stabilization, rerouting immunodominance, or any combination of these and additional factors. Phase 1 clinical trials for the early assessment of GT immunogens designed to elicit bnAb responses are ongoing, full results from those trials will be awaited with high anticipation. The preliminary report from the eOD-GT8 60mer clinical trial is extremely encouraging, and appears to herald a new era of GT vaccine designs.

The licensure of new adjuvants has also led to improved formulation of HIV vaccines in preclinical studies. Up until relatively recently, only a few adjuvants had been licensed, with a vast majority of currently licensed vaccines using alum. Several novel adjuvants had been developed and licensed in the last ~15 years, particularly multicomponent vaccines such as those including TLR agonists and saponins that induce stronger innate responses, and thereby enhanced adjuvanticity compared to alum, as reviewed by Del Giudice and colleagues [219]. As noted from several phase 1 studies discussed in this review, current studies, especially in preclinical settings, are increasingly shifting towards the use of diverse adjuvants in their formulations to maximize protein vaccine-induced immune responses.

A substantial number of clinical studies focus on vaccines aimed at generating cellular immunity. Importantly, progress has been made in viral vector designs to elicit strong CD8 T cell responses while minimizing cross reactivity to the vector. As with B cell responses and Ig gene diversity, breadth and magnitude of T cell responses to candidate immunogens elicited in animal models may not translate to humans due to interspecies diversity in MHC genes. Whether improvements introduced in the various consensus and mosaic vaccine candidates will elicit potent T cell responses in humans can only be answered through human trials. However, most published vaccines intended to elicit both cellular and Env-specific humoral responses have used a gp120 insert in the vectored vaccine component, followed by monomeric gp120 or non-prefusion gp140 or gp160 trimers as the Env protein boost. This is still the case with most current phase 1 vaccines discussed in this review. With a greater variety of superior Env trimer immunogens available, exploration of vaccines that incorporate and combine the best available candidates from both bnAb and T cell strategies is highly relevant [24].

As the field presses forward, there will undoubtedly be an increasing number of vaccine candidates progressing into large scale clinical trials. All HIV vaccine efficacy trials since RV144 have not been able to demonstrate protective efficacy. Despite the RV144 trial having been carried out over a decade ago, the criteria for assessing immunogenicity of vaccines in the recent efficacy trials relied heavily on correlates of protection derived from the RV144 trial. While the results of the Mosaico and Imbokodo trials will not be available for several years, the disappointing outcomes of HVTN702 are an indication that the field should revisit whether benchmarks set by the RV144 trial will indeed translate to substantial protective efficacy in humans. The fact that most vaccine candidates are unable to generate robust tier 2 virus neutralization and breadth has been overlooked as a go/no-go criteria for phase 3 trials. Further, many bnAbs with excellent tier 2 pseudovirus neutralization breadth and potency demonstrate reduced neutralization capacity to replication competent primary cell line derived viruses [220–223], suggesting that protective capacity by antibodies may need to be assessed much more critically.

On the other hand, the success of eOD-GT8 60mer in activating naïve VRC01-class precursor B cells provides a lesson that employing strict animal models, and preclinical assessment of B cells in humans, can be predictive of immunization outcomes. Multiple independent models were performed to yield consistent results to confirm the presence of target precursor immune cells in humans, and that those cells can be specifically targeted by the immunogen of interest. eOD-GT8 can also elicit affinity matured antibodies in stringent animal models. The human phase1 clinical trial, although preliminary, remains the only vaccine to demonstrate activation of bnAb-class B cells by vaccination in humans. Overall, lessons learned suggest that more rigorous sets of HIV vaccine preclinical assessment criteria may need to be implemented in order to make a more accurate prediction of whether an HIV vaccine candidate is likely to succeed in meeting clinical trial goals.

Abbreviations

bnAb

Broadly neutralizing antibody

CD4bs

CD4 binding site

CE

Conserved elements

ChAd

Chimpanzee adenovirus

DE

Dose escalation

Env

HIV Envelope glycoprotein

eOD

Engineered outer domain

FP

Fusion peptide

gH

Germline heavy chain

gHL

Germline heavy and light chain

GnT-1

Glucose N-acetyltransferase-1

Gp

Glycoprotein

GT

Germline-targeting

HTI

HIVACAT T-cell immunogen

iGL

Inferred germline

MPER

Membrane proximal external region

nAb

Neutralizing antibody

NFL

Native-flexibly linked

nnAb

Non-neutralizing antibody

OP

Osmotic pump

pSer

Phosphoserine

T/F

Transmitter/founder

UCA

Unmutated common ancestor

UFO

Uncleaved prefusion optimized