Human immunodeficiency virus vaccines: Advances, challenges and future perspectives

Abstract

In the past 40 years, much effort has been made to develop a vaccine that can prevent human immunodeficiency virus (HIV) infection. But till today, this continues to be an unresolved challenge. The main obstacle in developing a vaccine is the ability of the virus to mutate at a swift rate, thereby evading the immune system. This has dramatically slowed the progress in the development of an HIV vaccine. Previous studies, like RV144, have demonstrated limited protection. Current trials, such as Imbokodo and Mosaico, have shown that the vaccines used were safe but did not demonstrate efficacy against the HIV-1 virus. With the advent of mRNA vaccines and broad neutralising antibodies, research has shown some progress in developing a promising candidate vaccine. Recently, the use of mosaic antigens and germline-targeting techniques has shed light on the development of a promising candidate vaccine. These techniques educate the immune system slowly. This review tracks the progress of HIV vaccine development to date and examines the most promising new strategies that could ultimately lead to the development of an efficacious vaccine.

Key Words: Human immunodeficiency virus-1 vaccine; Broadly neutralising antibodies; Germline-targeting; Mosaic antigens; mRNA vaccines; Prime-boost strategy; Immune evasion; Vaccine efficacy; Human immunodeficiency virus prevention

Core Tip: The article primarily focuses on the hurdles in developing a successful vaccine for human immunodeficiency virus-1, lessons learnt from some of the landmark trials, and the targets to be induced by a successful vaccine, and briefly touches on some current advances.

INTRODUCTION

Human immunodeficiency virus (HIV)-1 has remained a global health concern over the last four decades. Globally, an estimated 40.8 million people were living with HIV-1 in 2024, according to the United Nations Programme on HIV/acquired immune deficiency syndrome (UNAIDS) 2025 epidemiological estimates. Among these, around 77% were on antiretroviral therapy (ART)[1]. There has been substantial progress in treating HIV-1, and with modern antiretroviral regimens, the life expectancy of people living with HIV (PLHIV) on effective treatment is expected to approach that of the general population[2]. However, ART is a double-edged sword. It maintains viral suppression, increases CD4+ T-cell counts, halts disease progression and improves the quality of life of PLHIV on one hand. On the other hand, it also causes various adverse effects on the gastrointestinal, metabolic, dermatological and cardiovascular systems. These toxicities may compromise treatment adherence and diminish quality of life, particularly in ageing patients on lifelong therapy[3]. Additionally, current regimens do not eradicate infection. Therefore, there is a need for interventions that prevent the acquisition of infection. Accordingly, the development of a safe and effective HIV-1 vaccine remains a major focus of current research. Historically, vaccines have helped eradicate infections and reduce transmission of several infectious diseases. The quest for an HIV-1 vaccine has been going on for the past four decades, with only modest success to date.

HISTORY OF HIV-1 VACCINE DEVELOPMENT STRATEGIES

The pursuit of an effective HIV-1 vaccine has been ongoing for the last four decades. The inaugural human HIV-1 phase 1 vaccine trial in 1987 tested the recombinant vaccinia gp160 protein in HIV-negative volunteers and was found capable of inducing HIV-specific antibodies and cytotoxic T-lymphocytes[4]. VAX004 was the first large-scale trial initiated in 1998 to test the AIDSVAX B/B vaccine (two recombinant gp120 envelope proteins from subtype B isolates), but it did not demonstrate a protective effect[5]. HIV-specific CD8+ T-cell responses were higher among HIV-negative volunteers who subsequently acquired infection[6]. Subsequently, the VAX003, a phase 3 trial initiated in 1999, utilized the AIDSVAX B/E vaccine (gp120 Env proteins from clade B and clade E), which demonstrated no significant difference compared to the placebo[7]. In 2003, the Thai and the United States government jointly initiated RV144, a phase 3 trial to evaluate a novel HIV vaccine strategy commonly referred to as prime-boost, where the ALVAC-HIV (a canarypox vector vaccine coding for HIV-1 Env, Gag, and Pol) was the prime and AIDSVAX B/E (a gp120 subunit vaccine) was the boost. To date, this was one of the largest vaccine trials that enrolled 16000 participants showing a modest protection efficacy of 31.2% compared to the placebo group[8]. Overcoming the previous setbacks, the STEP (2004) and Phambili (2007) trials were designed as a phase IIb randomised double-blind trial, targeted at eliciting strong cellular immunity by using a modified adenovirus type 5 vector that delivered HIV-1 genes for the Gag, Pol, and Nef proteins from subtype B. However, both these trials were halted due to a lack of efficacy and an increased risk of infection[9,10]. HVTN 702 (Uhambo, 2016) was a large phase 2b/3 trial that used an RV144-like prime-boost regimen, where ALVAC-HIV (canarypox vector) served as the prime, and a bivalent gp120 protein subunit vaccine with MF59 adjuvant as the boost. This trial was halted in 2020 for non-efficacy reasons after an interim analysis[11]. The trials Imbokodo (2017) and Mosaico (2019) from Janssen tested an Ad26-based mosaic prime-boost regimen combined with soluble clade C gp140 (Imbokodo) or bivalent clade C/mosaic gp140 (Mosaico) Env protein with aluminium phosphate adjuvant. Both were halted due to futility after the interim analysis[12,13] (Figure 1).

Figure 1 This timeline illustrates over 25 years of human immunodeficiency virus vaccine development efforts. Early protein vaccines (VAX004, VAX003) failed. RV144 showed modest 31% protection-the only success so far. Adenovirus vectors (STEP, Phambili) were stopped early due to safety risk. Recent poxvirus and mosaic Ad269 (Adenovirus serotype 26) human immunodeficiency virus vaccine trials (HVTN 702, Imbokodo, Mosaico) have also failed. Now, researchers focus on bnAbs, mRNA, and nanoparticle vaccines.

HURDLES IN DEVELOPING A VACCINE FOR HIV-1

The Env (gp120) protein dominates HIV vaccine targets for neutralising antibodies. Developing HIV-1 Env vaccines is exceptionally difficult due to various reasons. The HIV-1 virus exhibits extraordinary genetic variability, allowing it to undergo mutation (3.4 × 10^-5 mutations per base in one replication cycle) and recombination, creating a rapidly shifting swarm of variants that makes neutralisation difficult[14]. The Env glycoprotein undergoes mutations due to the error-prone nature of reverse transcriptase[15]. The high conformational flexibility, low density of Env protein (approximately 7-14 spikes per virion), protection of the conserved regions of Env by a layer of glycan shield constituting about 50% of gp120 mass, the ability of Env to maintain its structural integrity while undergoing recurrent mutations, extraordinary genetic diversity of HIV-1 Env protein worldwide- all make Env protein a “difficult to neutralise” target[16-19]. This results in millions of Env variants in a single individual within 24 hours of replication[20]. HIV-1 is classified into main (M), outlier (O), non-M/non-O (N), and pending (P) groups with group M subdivided into nine subtypes/clades denoted by the letters A, B, C, D, F, G, H, J, and K. This results in amino acid variations within subtypes as high as 30% and up to 42% between subtypes[21]. In regions like Africa, around 10%-20% of the PLHIV are infected with two or more strains[22] making virus neutralization difficult. There is no animal model that recapitulates all salient features of HIV-1 infection in humans, limiting translational value[23]. Pharmaceutical companies generally show low investment in HIV-1 vaccine research and development due to high risks, scientific challenges, and uncertain financial returns compared to other pharmaceutical products[24]. Broad neutralising antibodies can neutralise a vast array of HIV-1 strains. These antibodies develop after years of infection due to somatic hypermutation in precursor cells. Producing these antibodies by traditional vaccination methods is difficult. Combinations of 2-3 broadly neutralising antibodies (bnAbs) can neutralise > 95% of strains, with ongoing trials testing them for prevention and cure alongside latency reversers[25-27]. Germline-targeting immunogens, such as eOD-GT8, act as a special bait to capture naive B cells that can eventually produce potent HIV-fighting antibodies called bnAbs. Phase 1 (IAVI G001) trial demonstrated that this approach is practical, with up to 97% of participants experiencing activation of these starter cells following vaccination[28]. Another issue is that immunogens that activate multiple bnAb precursors that target distinct epitopes on the Env protein may suffer competition[29]. The ability of the virus to remain dormant in CD4+ T-lymphocytes for years escaping from being recognized by immune system is another barrier. Virus-infected patients produce antibodies or activated CD4 T-cells which do not eradicate/neutralise the virus, questioning the benchmarks of protection against the virus. CD4 T-cells are required for eliciting a primary antibody response or a cytotoxic T-cell response. But the ability of the virus to downregulate CD4 T-cells makes them less reliable markers[30,31]. Several HIV-1 vaccine trials have been completed, with only one trial, the RV-144 demonstrating modest efficacy[8]. Immune correlates of protection in HIV vaccine trials vary across studies and are not universally specific or consistent, reflecting differences in vaccine designs, populations, and endpoints[32] (Figure 1 and Table 1).

Table 1 Some landmark human immunodeficiency virus-1 vaccine trials and lessons learnt.

Ref.	Trial (years)	Vaccine or strategy	Population	Lessons learnt
[7]	AIDSVAX B/E (1999-2003)	AIDSVAX B/E (gp120 protein, clade B/E)	2545 people who inject drugs in Bangkok, Thailand	The gp120 protein alone is insufficient, better immunogen design and/or vectors are needed
[5]	VAX004 (1998-2002)	AIDSVAX B/B (gp120 protein, clade B)	About 5400 MSM (men who have sex with men) and high-risk women in N. America/Europe	Protein-only approaches didn’t protect; highlighted the limits of the then available Env designs
[9]	STEP (2004-2007)	Modified adenovirus type 5 vector to deliver genes for gag/pol/nef (T-cell vaccine)	About 3000 MSM and high-risk women in the Americas/Australia	Vector pre-immunity and target biology matter. Avoid rAd5 strategies in at-risk subgroups
[10]	Phambili/HVTN 503 (2007)	Same rAd5 construct	Adults in South Africa (mostly heterosexual, clade C region)	Stopped early after STEP futility; no efficacy
[8]	RV144 (2003-2009)	Prime-boost: ALVAC-HIV (canarypox vector coding for HIV-1 Env, Gag, Pol) + AIDSVAX B/E (gp120 subunit)	16402 general-population adults in Thailand	Correlates: V1/V2-specific IgG linked to lower risk; high Env-IgA linked to higher risk
[11]	HVTN 702 (2016-2020)	RV144-like pox-protein regimen adapted for clade C	Men and women in South Africa	Vaccine components such as adjuvants, dosing schedules, and immune correlates of protection may need to be tailored to different viral subtypes and regional epidemic settings
[12]	HVTN 705 (Imbokodo) (2017-2021)	Ad26 mosaic prime + gp140 boost	2600+ young women in sub-Saharan Africa	No sufficient protection; trial ended. Elicit broadly neutralising activity and/or stronger functional antibody profiles
[13]	HVTN 706 (Mosaico) (2019-2023)	Ad26 mosaic prime + gp140 boost (modified)	About 3900 MSM and TG in America/Europe	No efficacy and trial discontinued. Results consistent with Imbokodo. The same vaccine platform cannot protect different populations

MSM: Men who have sex with men; rAd5: Recombinant adenovirus serotype 5; HIV: Human immunodeficiency virus; V1/V2: Variable loop 1/2; Ad26: Adenovirus serotype 26; gp: Glycoprotein; TG: Transgender.

ADVANCES IN HIV-1 VACCINE DEVELOPMENT

Broad neutralising antibodies

With the lack of efficacy observed in the HVTN 702, HVTN 705, and HVTN 706 studies the focus has now moved towards finding new vaccine strategies that induce broad neutralising antibodies[33]. Several strategies aim to induce bnAbs against HIV-1 by activating rare naive B-cell precursors and guiding their maturation through somatic hypermutation. BnAb can bind and neutralise conserved regions on the viral Env glycoprotein and are currently being developed through reverse vaccinology. Naive B cell antibodies (germline versions) typically bind HIV Env protein very weakly or not at all due to shielding and mismatch, while bnAbs from mutated B cells bind strongly. But these bnAbs develop only after years of infection in an individual due to non-frequent germline B-cell priming and the need for extensive antibody somatic hypermutation. The initial infecting virus stimulates weak antibody responses, but as HIV mutates rapidly, it creates diverse Env variants that progressively activate rare B cells, guiding their maturation toward bnAb development over the years. These bnAbs develop from a single clonal lineage and constitute only a minor component of the overall HIV-1-specific antibody response[34-37]. Induction of bnAbs is challenging due to various reasons. BnAbs have unusual traits, such as the frequent presence of long, heavy-chain complementarity-determining region 3 (HCDR3s) and extensive somatic hypermutation. Polyreactivity and autoreactivity with similar antigens are other features that make induction difficult. Precursors of B cells capable of producing such bnAbs are usually deleted in the bone marrow owing to autoreactivity. Overcoming these difficulties, bnAbs are produced in the following way. Germline-targeting immunogens, such as eOD-GT8, are designed to engage these unresponsive B-cell receptors and effectively prime them. Following activation, B-cells enter germinal centers where they undergo somatic hypermutation. Following this sequential immunisation with progressively affinity-optimised Env trimers favour the acquisition of otherwise improbable mutations. Through this stepwise process, rare “jackpot” lineages can mature into bnAbs capable of neutralising diverse HIV-1 strains[28,37]. The various types of broad neutralising antibodies include the CD4 binding site-targeted bnAbs, HCDR3-binder CD4 binding site bnAbs and V3 glycan-targeted bnAbs. The CD4-binding site targeted bnAbs are the critical targets of HIV-1 vaccine development efforts. CD4 binding site targeted bnAbs are classified into CD4 mimic bnAbs and HCDR3-binder bnAbs. VRC01-class antibodies are a type of CD4-mimic bnAbs. VRC01 bnAbs are one of the most potent bnAbs known, with their precursors being more common than other bnAbs precursors[28,37,38]. BnAbs antibodies can be engineered to increase the half-life, and combining different classes can increase the breadth of coverage. BnAbs also target latent reservoirs, prolong viral suppression, and reduce viremia in infected individuals. bnAbs have the potential to serve as the blueprint for next-generation HIV-1 vaccine design. Currently, bispecific and trispecific antibodies containing a combination of various bnAbs groups have been developed to increase the breadth and potency of the immune response. The newer generation of these antibodies is much more effective than the first-generation ones. They can neutralise a wide variety of HIV strains at lower concentrations, making them a significant advancement in the prevention and treatment of HIV[37,39,40]. RIO (NCT04319367) is a phase II, randomized (1:1), double-blind, placebo-controlled trial evaluating dual long-acting bnAbs (3BNC117-LS and 10-1074-LS) administered after 40 weeks of ART in individuals who initiated ART during primary HIV infection. The results revealed that dual long-acting bnAbs were safe and significantly improved viral control in PLHIV off ART compared with placebo. About 65% of participants receiving bnAbs remained virally suppressed through week 20 compared to the 9% in the placebo group. Several participants maintained suppression for up to 72 weeks. In some people the virus became resistant, meaning this approach is promising but not yet a permanent cure[41]. A phase 2 trial (NCT05729568) demonstrated that twice-yearly subcutaneous administration of lenacapavir along with teropavimab (GS-5423, TAB) and zinlirvimab (GS-2872, ZAB), abbreviated as LTZ, maintained HIV-1 RNA levels by < 50 copies/mL in 89% of virologically suppressed participants through week 52 compared to stable oral ART (96% at week 26). The regimen was well tolerated with a median increase in CD4+ T-cell count (+32 cells/μL)[42]. Somatic hypermutation and bnAbs lineage maturation face major hurdles due to HIV Env protein’s evasion of host immune system. B cells need high somatic hypermutation (> 30% nucleotide changes, 15% amino acid), but these excessive mutations can result in apoptosis due to autoreactivity checkpoints and further the unusual features disfavour survival. Central (bone marrow) and peripheral tolerance delete self-reactive intermediates; regulatory T cells suppress; high somatic mutations downregulate B-cell receptor signalling. In germinal centers, off-target B cells outcompete bnAbs precursors, and without sequential affinity gradients, lineages tend to delete[25,43,44] (Figure 2).

Figure 2 From elite neutralisers, B-cells producing broad neutralising antibodies are isolated, and antibody genes are extracted and cloned into expression vectors for production. Applications include: (1) Passive immunisation via bnAb infusions for immediate viral control; (2) Germline targeting vaccines designed to activate bnAb precursor B-cells; and (3) Gene therapy delivering bnAb genes for long-term endogenous production.

Lineage and germline vaccine strategies

Env immunogens are developed from longitudinal analyses of bnAbs from infected individuals. In the lineage vaccine strategy, these Env immunogens (priming immunogen) show affinity for a single unmutated common ancestor of the bnAbs’ lineage. Following this, vaccination with Env proteins that are capable of inducing somatic hypermutation is administered. These strategies have been tested in knock-in mice and non-human primates, where Env immunogens that bind to precursor B-cells of either the CD4-binding site or V3-glycan bnAbs lineage were found to be capable of inducing a bnAbs response[45]. HVTN 300 (NCT04864072) is an ongoing phase 1 trial evaluating stabilised CH505 TF chTrimer immunogens (native-like Env trimers) with 3M-052-AF (novel synthetic toll-like receptor 7/8 agonist adjuvant) and alum adjuvant to prime CH103-like bnAb B-cell lineages targeting the CD4-binding site[46].

The germline vaccine strategy involves priming naive B cells using engineered Env immunogens (e.g., eOD-GT8) that bind to diverse, unmutated, common ancestors within a specific bnAbs class (e.g., VRC01-class), followed by boosts with native-like trimers or lineage-specific immunogens[43,45]. Ray et al[47] demonstrated that targeting the membrane-proximal external region (MPER) epitope of the gp41 subunit can drive germinal-centre selection and maturation of B-cell precursors toward bnAbs lineages. Wang et al[48] demonstrated that mRNA-LNP delivery of the soluble, self-assembling nanoparticle immunogen eOD-GT8 successfully activated and evolved VRC01-class precursor B cells toward bnAbs lineages in humanised mice. IAVI G003 (NCT05414786) is a phase 1, open-label trial testing the safety of the germline-targeting eOD-GT8 60mer mRNA vaccine (mRNA-1644, 100 μg dose) administered at weeks 0 and 8 in HIV-uninfected adults. The study evaluates a two-dose vaccine regimen developed to stimulate VRC01-class immune cells in African participants. Participants are screened for eligibility up to 56 days before giving first dose and after receiving two doses they are followed up for 6 months. The main goal of the study is safety evaluation[49]. IAVI G002 (NCT05001373) is a phase 1, randomised, first-in-human, open-label trial evaluating a sequential combo of eOD-GT8 60mer (mRNA-1644, 100 μg) for priming, followed by core-g28v2 60mer (mRNA-1644v2-Core, 100 μg) for boosting to stimulate the VRC01-class and V3-glycan bnAb precursors in 56 HIV-uninfected adults. The study includes four arms: EOD-GT8 alone, eOD-GT8 followed by core-g28v2, core-g28v2 followed by eOD-GT8 and core-g28v2 alone based on the hypothesis that sequential germline-targeting via mRNA guides bnAb B-cell maturation[50].

Epitope-focused design

In this strategy, immunogens that target the neutralising epitope on HIV-1 Env protein are presented, masking the non- neutralising epitopes. Structurally characterised bnab epitopes on HIV-1 include CD4-binding site at the V2 trimer apex, V3-glycan supersite and the MPER[51]. A phase 1 trial, IAVI G001, tested eOD-GT8 nanoparticles targeting the CD4bs epitope, with results showing 97% of recipients activated VRC01-class bnAbs precursors[28]. HVTN 139 (NCT05182125) is an ongoing phase 1 trial evaluating the safety and immunogenicity of chimpanzee-derived adenovirus vectors (AdC6-HIVgp140, AdC7-HIVgp140) expressing clade C gp140, boosted by CH505TF gp120/GLA-SE protein in healthy, HIV-uninfected adults. It aims to determine the safety and tolerability of these vector and protein combinations[52].

Immunisation with SOSIP trimers

The next advancement is the development of SOSIP trimers[53]. SOSIP trimers are engineered HIV-1 proteins that resemble the native Env protein, essential for eliciting bnAbs[54]. Using computer software, the I55P9 point mutation was incorporated into the pre-fusion form of gp41, followed by linking their gp120 and gp49 with a covalent disulfide bond[53]. This resulted in the development of a stable SOSIP timer resembling the Env protein[54]. SOSIP timers obtained from different HIV-1 strains can be combined to increase the breadth and potency of the immune response[55]. Animal studies have shown that this SOSIP trimer can induce broad neutralising antibodies[56]. BG505 SOSIP.664 was one of the first SOSIP trimers used[57]. SOSIP trimers have been developed from multiple strains across different clades[55]. Studies have shown that antibody responses often target exposed glycan-hole regions on these trimers, such as those identified in BG505 SOSIP.664[56]. Rabbits immunised with SOSIP trimer (JR-FL_SOSIP) utilising the DNA prime and protein boost technique produced neutralising antibodies to sensitive strains[58]. ACTHIVE-001 was a phase 1 trial that tested ConM SOSIP.v7 Env trimers with MPLA as adjuvant in 24 HIV-uninfected adults. The vaccine has proved to be safe and immunogenic, inducing strong autologous tier-1/2 neutralizing antibodies targeting V1V2, CD4bs, and base epitopes in all participants after multiple doses. Notably, females showed superior antibody responses compared to males, highlighting sex-based differences in immunogenicity[59]. Currently, SOSIP trimers have been used in combination with various other methods of vaccine delivery, including mRNA and nanoparticle technology, to maximise efficacy[60]. As with advantages, there are some issues with the use of SOSIP trimers. Animal studies have shown that low pH or a DNA component, if present in adjuvants, can result in destabilisation of SOSIP trimers[61]. SOSIP trimers use the non-native tPA (tissue plasminogen activator) signal peptide (SP) coding sequence primarily for superior endoplasmic translocation efficiency. As native SP has a role in Env folding, presentation of Env and overall glycosylation, these factors might be altered in SOSIP trimer[62]. Exposure of the gp41 trimer base in the SOSIP trimer, which is usually masked by the viral membrane, is another limitation resulting in off-target bnAbs directed to these sites[63]. Another debated feature is that the SOS bond itself may introduce non-native, structural changes[61]. Thus, the SOSIP trimer has proved to be a key design in presenting the Env spike in a form that closely resembles the native protein.

mRNA-based vaccines

This technique uses the virus’s genetic material to produce a protein of interest, which stimulates the virus’s immune system[64]. In one such approach, mRNA is coupled with nanoparticles, which can protect it from degradation by RNases and help with delivery. With the success of the coronavirus disease 2019 vaccines, the development of mRNA vaccines for HIV-1 has generated hope. There are two types of mRNA platforms in the development of HIV-1 vaccines. Conventional non-replicating mRNA that encodes Env, Gag, or immunogenic trimers, such as SOSIP gp140, is preferred for its enhanced stability and resemblance to native folding[65]. The other one is the saRNA, which encodes an RNA-dependent RNA polymerase derived from alphaviruses. This polymerase can recognise and amplify specific elements in the saRNA within the host cells[66].

Two AGS-004 mRNA studies (NCT00672191 and NCT00833781) evaluated dendritic cell-based immunotherapy in which autologous dendritic cells are transfected with RNA encoding HIV antigens (including Gag, Nef, Rev, Vpr) from the patient’s own virus and then returned to the participant, with both studies failing to stimulate an antiviral response. Viral rebound occurred in approximately 4 weeks in both[67,68]. The HIVACATTriMIX mRNA study (NCT02888756) tested iHIVARNA01, which is an mRNA-based therapeutic vaccine combining a conserved T-cell immunogen with TriMix (mRNAs for CD40 L, constitutively active TLR4, and CD70) against TriMix alone and placebo in chronic HIV-1-infected patients on stable ART. The Phase IIa trial was terminated after an interim analysis because it did not demonstrate significant immunogenicity without any significant increase in interferon gamma Tcell responses measured by ELISPOT[65,69]. A phase I dose-escalating clinical trial using a naked mRNA (iHIVARNA) combined with TriMix (CD40 L, CD70, and constitutively active TLR4) and a novel HIV-1 immunogen encoding Gag, Pol, Vif, and Nef was completed, showing only a moderate Tcell immune response[70]. Actively conducted trials include Moderna’s HVTN 302 mRNA vaccine (NCT05217641), which evaluates three BG505derived Env trimer immunogens (BG505 MD39.3, BG505 MD39.3 gp151, and BG505 MD39.3 gp151 CD4KO) delivered as mRNA vaccines to assess safety and immune responses, including the induction of autologous neutralizing antibodies against HIV strains similar to the vaccine immunogen. BG505 MD39.3 is a stabilized soluble SOSIPbased trimer derived from a clade A transmitted/founder virus resembling a native-like Env spike. The gp151 constructs encode a membraneanchored trimer including gp120 and gp41 ectodomain, and gp151 CD4KO incorporates a mutation in the CD4binding site to reduce CD4 interaction and potentially focus immune responses on other conserved epitopes[71]. Several antigen formats, delivered via mRNA vaccines for HIV-1, have been explored, including membrane-anchored Env glycoproteins, soluble Env trimers, and virus like particle. Recent preclinical and early-clinical data suggest that membrane-anchored Env formats elicit superior B cell activation (including precursor bnAb B cells), resulting in greater neutralizing-antibody induction than soluble trimer formats. For example, in a phase 1 trial, approximately 80% of participants receiving mRNA-encoded membrane-anchored trimers developed autologous tier-2 neutralising antibodies compared with approximately 4% receiving soluble trimer[72]. When DNA prime was combined with an mRNA liquid nanoparticle boost, both T-cell and antibody responses in animal models were stimulated[73]. A novel advancement, NanoVac, utilises short carbon nanotubes to co-deliver mRNA and HIV-1 glycoproteins with high efficiency and minimal toxicity. Its optimised surface chemistry enhances antigen presentation and supports both intramuscular and intranasal delivery. In humanized mouse models of HIV1 infection NanoVac induced strong immune responses, with 33% viral clearance in HIV-infected humanised mice. Moreover, NanoVac maintains mRNA stability for up to three months under refrigeration, easing storage and distribution challenges[74]. Realistic timelines for mRNA-based HIV vaccines remain long, uncertain, and iterative[33,65]. Current phase 1 germline-targeting and Env-mRNA trials represent only the initial steps toward bnAb induction[75,76]. Phase 2 and 3 studies will require multi-dose sequential regimens that demonstrate the elicitation of heterologous tier 2 bnAbs in humans, a goal that has not yet been achieved[33,77]. mRNA platforms fasten immunogen iteration[65], still, the fundamental challenge of driving rare germline bnAb precursors through extensive somatic hypermutation by overcoming B-cell tolerance remains[75].

The essential advantages of mRNA vaccines include the rapid design flexibility, low cost, safety over DNA and viral vector vaccines, and potent immune activation[64,65]. Lipid nanoparticles deliver mRNA encoding the HIV-1 Env or Gag protein, which act as antigens, triggering an immune response by the body[64,65]. The main challenges include the short-lived nature of antibody responses[65]. The glycan shield of the Env protein helps in evasion of the immune response, making vaccine development difficult[16]. The difficulty in delivering mRNA due to its unstable nature limits the widespread application[64]. Carefully determined immunisation regimens are needed to elicit broad neutralising antibodies[37].

Nano-vaccines for HIV-1

Nanoparticles utilise proteins or lipids in the form of tiny particles to deliver antigens to the immune system. These particles can enhance the immune response by presenting genetic material in the form of natural infection. Nanotechnology enables precise antigen delivery, enhances antigen presentation, and promotes potent immune system activation by mimicking the structural and functional properties of pathogens[78,79]. Nanovaccine platforms offer improved stability, controlled antigen release, and enhanced presentation to antigen-presenting cells[80]. Nanoparticle vaccines can be designed to present conserved regions of the HIV-1 Env protein, which are less prone to mutation[79]. This strategy aims to elicit bnAbs that can neutralise a wide range of HIV-1 strains[79]. Broadly, nanovaccines are divided into non-viral and viral platforms[78,80]. The non-viral systems include inorganic, polymeric, and lipidic nanoparticles[78]. Among inorganic systems, gold nanorods and silica-coated calcium phosphate nanoparticles have demonstrated the ability to stimulate Env-specific antibodies and T-cell responses in experimental HIV-1 immunisation models[81-83]. Polymeric nanoparticles such as poly lactic-co-glycolic acid (PLGA) and poly-methyl methacrylate can be tailored for mucosal targeting, and PLGA-Eudragit microparticles have been shown to elicit strong mucosal IgA and IgG responses following intra-colorectal immunisation[84,85]. Lipid nanoparticles encapsulating Env mRNA have enabled efficient intracellular delivery and rapid antigen expression, activating both humoral and cellular immune arms[78,79]. Self-assembling protein nanoparticles, such as ferritin, E2p, and I53-50 scaffolds, allow for the controlled multivalent display of HIV-1 Env trimers, like BG505 SOSIP, thereby improving immunogenicity[78,79]. Ferritin nanoparticles displaying eight Env trimers per particle induced higher binding-antibody titres and improved germinal-centre B-cell responses compared with soluble Env trimers[86]. These advances highlight nanotechnology’s expanding role in next-generation HIV-1 vaccine design. In a preclinical study, a self-assembling protein nanoparticle (1c-SApNP) conjugated to BG505 uncleaved prefusion-optimised (UFO) was used to induce bnAbs in mice and rabbits. These 1cSApNPs presenting the nativelike BG505 UFO trimers induced improved antibody responses in mice and rabbits compared with soluble trimers[87]. An eOD-GT8 60-mer nanoparticle was designed, which was capable of priming VRC01-class HIV-1-specific B cells for bNAbs production in mice[88]. Recently the same group revealed that eOD-GT5 60-mer induced a CD4 T-cell response in mice, thereby emphasizing the highly immunogenic nature of nanoparticle vaccine[89]. Another important advantage is that these nanoparticles can be combined with adjuvants and such adjuvants can increase the production of antibodies. In a preclinical study, ConM SOSIP trimers were evaluated as free proteins and presented on ferritin nanoparticles with different adjuvants (squalene emulsion, ISCOMATRIX, GLALSQ, and MPLA liposomes) in rabbits. Compared with soluble trimers, the trimers presented on the ferritin nanoparticles enhanced neutralising antibody responses, with the strength of this effect depending on the adjuvant, with ISCOMATRIX and squalene emulsion generally providing stronger responses than GLA-LSQ[90]. In NHPs the use of an imidazoquinoline adjuvant 3M-052 when paired with PLGA nanoparticle-based vaccines induced sustained tier 1 nAb production[91].

CRISPR/CASP9 technique

The CRISPR/Cas9 technique, in the context of HIV-1, does not refer to vaccines but rather to curative gene-therapy approaches designed to eliminate or disable the virus in people already suppressed on ART[92]. Rather than preventing infection, these methods aim to cut out the integrated HIV DNA (the provirus) or disrupt essential host or viral factors[92]. As of 2025, there are no CRISPR-based HIV vaccines in clinical trials. Preclinical studies have shown strong potential. Using multiple guide RNAs targeting conserved HIV regions, CRISPR/Cas9 can excise the entire proviral genome in cell cultures and suppress viral replication in humanised mouse models[93]. Clinical results have been far more limited. The first human trial, EBT-101 (NCT05144386), used an AAV9-delivered CRISPR therapy in ART-suppressed individuals. The treatment was generally safe and well tolerated, but in participants who stopped ART, HIV rebounded within weeks, indicating that only a small fraction of the viral reservoir was edited[94]. No phase 2 trial has yet been announced.

Passive immunisation

In the past years the therapeutic potential of bnAbs has been demonstrated in humanized mice, non-human primates, and humans. Initially, the first generation bnAbs such as 2G12, 2F5, and 4E10 reduced viral load in HIV infected patients only to be deferred by the development of escape mutants and emergence of resistance[95-97]. After ART interruption in PLHIV, administration of bnAbs delayed viral rebound, with observed delays ranging from approximately 3 to 24 weeks across different studies, depending on the antibody used and participant characteristics[98,99]. However, the undesirable features, including autoreactivity exhibited by 2F5 and 4E10 bnAbs, whilst 2G12 demonstrates low potency with resistance, limiting their usage[100,101]. Even though we have moved to second-generation bnAbs after the development of single-cell antibody cloning techniques in 2009[102], the first generation remains the foundation for future therapies. The second generation bnAbs with improved breadth and potency target highly conserved regions as discussed before including the CD4 binding site (CD4bs), the V1V2 glycan region, the V3-glycan region. Others include the gp120-gp41 interface, the gp120 silent face, the gp41 MPER, and the gp41 fusion domain[103-106]. Early phase 1 trials in PLHIV with bnAbs targeted either the CD4-binding site (3BNC117 and VRC01) or the V3-glycan region (10-1074 and PGT121). In bnAbs-sensitive virus-infected PLHIV, a single infusion of 3BNC117, VRC01, PGT121, or 10-1074 produced a considerable reduction in viral copies from baseline, and viremia remained suppressed up to 28 days, only to be followed by viral rebound due to escape mutants[107-111]. In patients with ART interruption with reduced viral levels, the efficacy of bnAbs is enhanced. After ART discontinuation, VRC01 or 3BNC117 monotherapy demonstrated effective viremia control and delayed viral rebound for 4 and 10 weeks. Rebound viruses arose from a single clone either due to waning immunity or due to the emergence of resistance. About 30% of 3BNC117 participants remained virally suppressed. 3BNC117 exerts significant selective pressure on emerging viruses during ATI in humans[112-115]. Talking about combination bnAbs therapy, several studies showed that a combination of bnAbs achieved durable viral suppression in animal models. After ART interruption, the combination of 3BNC117 and 10-1074 delayed viral rebound to a median of about 21 weeks in participants whose virus was sensitive to both antibodies in HIV-positive individuals. In untreated viremic patients without pre-existing bnAbs resistance, the same combination suppressed viral replication for up to approximately 3 months[116,117]. Provided adequate concentration is maintained, a combination of 3BNC117 and 10-1074 offers some degree of viral suppression in patients with bNAb-sensitive viruses. To conclude, optimal bNAb combinations and concentrations are necessary to achieve durable viral suppression in HIV-infected individuals.

Role of artificial intelligence

Artificial intelligence and computational tools accelerate HIV vaccine development by predicting epitopes, designing immunogens, analysing trial data, and optimising trials for speed and equity.

Epitope prediction: Ragon institute/Massachusetts institute of technology (MUNIS, 2025) applies deep learning to over 650000 HLA-bound peptides to predict CD8+ T cell epitopes accurately thereby spotting new HIV and Epstein-Barr virus targets to support T cell vaccines. It outperforms predictors such as NetMHCpan-4.1 in identifying likely presented and immunogenic epitopes and has been validated on influenza, HIV, and Epstein-Barr virus datasets, identifying both known and novel epitopes to support Tcell vaccine design efforts[118]. NetMHCpan-4.1, along with transformer models, predicts peptide-MHC matches for over 13000 alleles, including glycan effects on stable Env trimer regions[119].

Immunogen design: Protein language models process HIV sequences and mutations to create bnAbs mimics, thereby predicting the binding strength and further improving designs obtained from elite controllers, using genomic and population data for local variants[120]. Artificial intelligence algorithms are trained to rapidly select Env mutants capable of boosting bnAb lineages to heterologous breadth and potency. The path to a successful vaccine lies in learning to engineer the immune system to stimulate and mature rare neutralising antibodies that infrequently occur in PLHIV. Formulation of a multivalent immunogen is the next step to be used in a trial. Currently, work is going on to design Env immunogens with many bnAb triggering sites on the same immunogen[118-121].

ETHICAL AND SCIENTIFIC CONSTRAINTS FOR HIV TRIALS

Phase 1 HIV-1 trials mainly assess safety and immune priming and as a result face fewer ethical concerns. Moving forward to phase 3 efficacy trials is difficult because highly effective prevention tools like oral pre-exposure prophylaxis (PrEP) and long-acting lenacapavir are now the standard of care, making placebo groups ethically sensitive. Further UNAIDS guidance requires providing full prevention packages to all participants making vaccine efficacy distinction difficult[122]. Vaccinating PLHIV who are virologically suppressed on ART could significantly accelerate both preventive and therapeutic HIV vaccine development. PLHIV offer a unique “in vivo model” where vaccine immunogens can be tested for their ability to boost pre-existing HIV-specific B and T-cell responses, including bnAb lineages, with much smaller sample sizes and without relying on incident infection as an endpoint. Vaccination paired with analytic treatment interruption could test whether enhanced immune responses delay viral rebound[123]. Other ethical issues include vaccine-induced seropositivity leading to lifelong stigma or discrimination, emphasizing the importance of strong informed consent and clear plans for post-trial access, especially in low-resource settings[124]. Widespread use of PrEP reduces the number of people who are at risk of acquiring infection, resulting in enrollment of a large number of participants to detect vaccine effects. Long follow-up periods can result in attrition[125]. The lack of definitive immune correlates of protection further complicates trial design[32]. Reduced industry investment following recent vaccine failures has made sustained community engagement essential to support participation and acceptance of next-generation vaccine strategies[126].

ONGOING MOMENTUM

After the discontinuation of Imbokodo and Mosaico studies due to lack of efficacy[12,13], there has been a shift in the landscape of HIV vaccine trials. Studies like IAVI G001 (NCT03547245) are the landmark proof-of-concept studies for germline targeting immunization. It demonstrated that bnAb precursor activation in humans is feasible[127]. The IAVI G003 (NCT05414786) study tests the safety/immunogenicity of two doses (100 μg) of eOD-GT8 60mer mRNA (mRNA-1644) to induce VRC01-class B-cells[49]. The results might instil confidence in mRNA-only germline targeting techniques. The HVTN 302 (NCT05217641) study evaluated the safety and immunogenicity of mRNA-encoded envelope trimers. Membrane-anchored trimers elicited tier 2 neutralizing antibodies in 80% of vaccinees compared to the mere 4% vaccinees who received soluble trimer vaccine[72]. Going further, the HVTN 300 (NCT04864072) evaluates whether bnAbs of a predictable B-cell lineage can be safely induced using carefully chosen immunogens[128]. Further studies like ACTHIVE-001 demonstrated that females exhibited higher Env-specific antibody responses than males. It highlighted the importance of considering sex-based difference in HIV vaccine immunogenicity[59]. The HVTN 137A (NCT04177355) demonstrated the first clinical evaluation of a good manufacturing practice produced BG505 SOSIP.664 soluble native-like trimer along with a novel toll-like-receptor 7/8 signalling adjuvant. It induced autologous tier-2 nAbs in 5/17 (about 30%) participants. The regimen used here appeared safe[129]. The HVTN-139 (NCT05182125) study is unique in that it pairs a chimpanzee adenoviral vector as a prime with a CH505 transmitted/founder gp120 protein[130]. In another phase 1, first-in-human trial, the HIV-1 prefusion-stabilized Env candidate trimer 4571 proved to be safe, well-tolerated, and immunogenic in healthy adults, inducing specific antibody responses. When administered with alum, it promises to be a stable approach for eliciting targeted immune responses[131]. Another novel study IAVI C114 (NCT06617091) uses a gorilla adenovirus vector to induce T-cell responses against highly networked, mutationally constrained HIV epitopes to limit viral escape[132] (Table 2).

Table 2 Post-2022 human immunodeficiency virus-1 vaccine trials.

Ref.	Clinical trial	Vaccine concept	Platform	Vaccine description	Status
[127]	NCT03547245 (IAVIG001)	Germline targeting (CD4bs)	Protein nanoparticle + adjuvant (Phase 1)	eOD-GT8 60mer nanoparticle + AS01B (liposome-based adjuvant containing MPLA + QS-21)	Completed
[50]	NCT05001373 (IAVI G002)	Sequential germline targeting	mRNA-LNP (Phase 1)	eOD-GT8 60mer mRNA + core-g28 version 2 60mer mRNA nanoparticle	Active, not recruiting
[49]	NCT05414786 (IAVI G003)	Germline targeting (CD4bs)	mRNA-LNP (Phase 1)	eOD-GT8 60mer mRNA alone	Recruiting
[71]	HVTN 302 NCT05217641	Stabilized BG505 MD39.3 Env trimers	mRNA-LNP (Phase 1)	BG505 MD39.3, BG505 MD39.3 gp151, and BG505 MD39.3 gp151 CD4KO mRNA (3 groups)	Active, not recruiting
[46]	NCT04864072 (HVTN 300)	B-cell lineage (CD4bs)	Protein + adjuvant (Phase 1)	CH505 transmitted/founder chimeric trimer + 3M-052 aqueous formulation with aluminum hydroxide	Active, not recruiting
[59]	NCT03961438 (ACTHIVE-001)	Native-like trimer	Protein + adjuvant (Phase 1)	Consensus M SOSIP version 7 gp140 + MPLA liposomes(as adjuvant)	Completed
[129]	HVTN 137A (NCT04177355)	Native-like trimer	Protein + adjuvant (Phase 1)	BG505 SOSIP.664 + 3M-052 aqueous formulation with aluminium hydroxide	Completed
[52]	NCT05182125 (HVTN 139)	B-cell lineage priming	Adenoviral vector + protein (Phase 1)	Adenovirus chimpanzee serotype 6/7 expressing HIV gp140 envelope glycoprotein+ CH505 transmitted/founder gp120 glycoprotein adjuvanted with glucopyranosyl lipid A stable emulsion	Recruiting
[131]	NCT03783130 (Trimer 4571)	Prefusion-stabilized trimer	Protein + adjuvant (Phase 1)	BG505 DS-SOSIP.664 (Trimer 4571) + alum	Completed
[132]	IAVI C114 (NCT06617091)	T -cell-inducing preventive/therapeutic HIV vaccine	Non-replicating Gorilla Adenovirus (Phase 1)	Gorilla adenovirus vectored HIV networked epitopes 1 encode highly networked, mutationally constrained HIV epitopes from stable structural regions (less prone to mutate)	Not yet recruiting

HIV: Human immunodeficiency virus; CD4bs: CD4 binding site; eOD-GT8: Engineered outer domain gp120 germline-targeting version 8; QS-21: Quillaja saponaria saponin-21; mRNA-LNP: Messenger RNA-lipid nanoparticle; MD39.3: Membrane distal variant 39.3; CD4KO: Cluster of differentiation 4 knockout; MPLA: Monophosphoryl lipid A.

CONCLUSION

HIV-1 vaccine development has faced decades of setbacks due to the virus’s extreme genetic variability and extensive glycan shielding. Developments in germline-targeting immunogens such as the eOD-GT8, which activates up to 97% of VRC01-class naïve B-cell precursors together with stabilised SOSIP Env trimers and mRNA delivery platforms (for example, HVTN 302), support a stepwise vaccination strategy. AI-based tools like MUNIS and NetMHCpan can improve the prediction of CD8 T-cell targets. Computer-guided refinement of bnAb lineages from elite controllers may help optimise early-phase trial design. In this PrEP era, the progression to phase 3 efficacy trials is also accelerated, possibly by testing vaccines in PLHIV. Overall, artificial intelligence-guided vaccines which target multiple neutralisable regions of HIV such as the CD4 binding site, V3-glycan region, and MPER, including bispecific designs maintained with booster doses, hold promise for the future. This should be followed by sustained long-term engagement from industry supported by robust public-private partnerships.