Scientists are developing on-target HIV vaccines to avoid unhelpful immune responses

Several studies presented at the 26th International AIDS Conference (AIDS 2026) in Rio de Janeiro last month dealt with vaccines designed to counteract a phenomenon that has limited their efficacy in the past – immunodominance.

Immunodominance is the immune system's tendency to focus on the most conspicuous parts of a virus. With HIV, those parts act like decoys: they vary freely from one virus to the next, so antibodies against them work on few other strains.

Effective antibodies, on the other hand, are able to target epitopes that are parts of essential viral mechanisms and are therefore hard to change (‘conserved’). These are often hidden from the surveillance of the immune system, either by only being exposed during brief moments of change, or through camouflage – in the case of HIV by a cloak of sugar molecules (glycans).

Two studies presented at AIDS 2026 demonstrated different ways of designing vaccines to overcome the problem of immunodominance.

In the first case, researchers used a candidate vaccine consisting of the HIV Envelope DNA (env) or protein (Env) but with an immunodominant component removed. This is a part of the HIV envelope called the V1 loop.

Dr Genoveffa Franchini of the US National Cancer Institute told the conference that her team vaccinated 24 monkeys, 12 with a vaccine containing the full env DNA sequence followed by the Env protein, and 12 of env/Env with the V1 loop removed. By removing this component it enables the antibodies which are involved in an immune response to gain better access to another component of Env, the V2 helix.

Seventeen years ago, antibody sensitivity to the V2 loop was one of the factors most strongly associated with the modest efficacy of 31% seen in the RV144 vaccine trial in Thailand  – the only human HIV vaccine efficacy trial ever to produce a significant result.

Franchini observed that 11 years later, Uhambo, a trial using almost the same vaccine in South Africa, was stopped due to lack of efficacy. In fact, in this trial, there were slightly more infections seen in vaccine than in placebo recipients, though this difference was not statistically significant. Later investigations showed that there was actually a stronger overall immune response to HIV in Uhambo than in RV144.

So why didn’t it work? Adjuvants are chemicals that set up an inflammatory reaction to a vaccine that magnifies the immune response to it, and the one difference between the two candidate vaccines was that RV144 used alum (aluminium hydroxide) as an adjuvant and Uhambo used MF59, an oil-based compound. The overall inflammatory reaction in Uhambo, stimulated by MF59, was stronger but not specific, whereas the ‘gentler’ reaction to the RV144 vaccine induced by alum included a specific reaction to the V2 loop.

Franchini’s vaccine – which did use alum as adjuvant – consisted firstly of two shots of env DNA, lacking the sequence that codes for the V1 loop, four weeks apart. Then at weeks 8 and 12, a monkey-adapted version of the very same vaccine that was used in RV144 – ALVAC – was given. This is a vector vaccine, with the env DNA (including the V1 loop) enclosed in a harmless canarypox virus engineered to carry the env gene, which infects cells and makes Env protein but cannot reproduce. Finally, also at week 12, a boost of the HIV gp120 Env protein was also given, again with its V1 loop removed.

They then tried to infect the 24 monkeys, plus another 11 not given any vaccine, with a classic rectal challenge of 11 weekly doses of a monkey-adapted HIV virus (SHIV) that was resistant to neutralisation by typical antibodies. At week 11, 85% of the monkeys given the V1-deleted vaccine remained uninfected compared with 30% of the monkeys given the vaccine containing V1 – a relative efficacy of 80%.

This is not the first time such efficacy has been observed for this type of vaccine. In 2025, a study challenging female macaques with vaginal SIV doses found that a similar vaccine, containing a different alum-based adjuvant, was 79% effective.

Further studies have found that the V1-deleted vaccine works by calming the immune reaction to HIV in mucosal cells, not by inflaming it. As an initial line of defence, it increases the production of a cytokine called IL-17 and of nucleoside adenosine, both of which act to make mucosal cells less receptive to HIV. If HIV does penetrate the mucosa, antibodies against the V2 loop signal the immune system to produce natural killer (NK) cells which attack HIV-infected cells directly. The antibodies also summon the large cells called macrophages to clear away cells that have already died, without triggering further inflammation.

Franchini told the conference that similar results, as yet unpublished, have been seen in a V1-deleted vaccine that uses mRNA inside virus-like particles rather than DNA. This vaccine was 86% effective in protecting monkeys from 11 intrarectal SHIV challenges.

“Engagement of multiple arms of immunity at systemic and mucosal sites is necessary to protect against SIV and SHIV acquisition,” Franchini commented.

She warned that there were several unknowns still to be investigated. Firstly, not all HIV strains may be susceptible to anti-V2 antibodies. Secondly, the duration of protection is unknown.

Nonetheless, a human study with 96 volunteers, the CLEAR study, is scheduled to start next year. Its primary endpoint is safety, but the secondary line of investigation is to see whether V1 deletion also gives a signal of effectiveness in humans.

Franchini commented that while the vaccine model in her study will hopefully offer localised mucosal immunity, it does not induce the more systemic and longer-lasting broadly neutralising antibodies.

A second study, presented in the same session at AIDS 2026, also tried to ‘correct’ immunodominance, but with the aim of changing an unhelpful immune reaction to one part of the HIV Env protein – the gp41 protein which forms the base of HIV’s ‘spikes’ – into a bnAb-like response to another part, namely the glycan ‘shield’ that protects another part of the Env protein, the V3 loop.

Dr Maria Tarquis Medina of the Wistar Institute in Philadelphia said that researchers have previously noticed that some of the unhelpful antibodies to gp41 shared amino acid sequences with more potent and useful broadly neutralising antibodies (bnAbs) to the V3 glycan.

In particular, they found a gp41 antibody in monkeys numbered Ab1718 that showed remarkable sequence similarity to PGT121, a potent bnAb that has already been used in a number of prevention and treatment studies in humans.

In experiments on mice, they first engineered the animals’ immune systems to produce Ab1718 – and only that – and then, by immunising them with an innovative immunogen displaying the V3 loop protein on a virus-like particle, succeeded in turning it into an antibody much more likely to bind to the V3 glycan.

Although this is very early preclinical science, it shows that it may be possible to persuade existing antibodies to develop a more bnAb-like profile, thereby avoiding the need to seek out rare precursor cells that have the genetic capability to produce them and then shepherd them into producing bnAbs by the process of ‘germline targeting’.

Redirecting antibodies to a more useful target

In a third presentation, Dr Wilton Williams of Duke University in North Carolina presented the latest in a series of trials of candidate HIV vaccines in monkeys.

The immunogen used represents an attempt at a different kind of shortcut to inducing the production of effective bnAbs by vaccines. The vaccine was a complex one, involving three doses of a genetically-altered piece of HIV RNA that coded for a section of HIV Env protein called CH505, then three vaccinations of the genetically unaltered RNA coding for the CH505 Env protein, and finally the CH505 env protein itself, delivered via a lipid nanoparticle.

The novel aspect of this vaccine is that although CH505 may sound like a bnAb, it is in fact the identifier of a person: someone found to have a clade C HIV virus (the most common subtype in Africa), which gave rise to a cluster of bnAbs closely related to one called CH235, which attaches to the CD4 binding site on the HIV Env protein. If this particular Env protein can reliably induce the production of the CH235 bnAb in other people, then it would be a shortcut to a bnAb-based vaccine.

This particular trial was the first to use, not the genetically-tweaked CH505 Env protein itself, but instead a sequence of RNA that induced the body to make its own CH505 env protein. It was found that the RNA vaccine (plus the CH505 booster) induced “CH235-like bnAb precursors” in 100% of the monkeys.

Nine monkeys were challenged with 10 intrarectal doses of SHIV that carried the CH505 Env protein. Three were completely protected and the vaccine offered partial protection to most of the others: it took only two challenges for more than half of the control monkeys to be infected, and all were after seven challenges, whereas it took nine challenges to infect over half of the monkeys receiving the vaccine.

Williams told the conference that the lipid nanoparticle used as a booster acted as an adjuvant in itself, broadening the generation of different related bnAbs. But it would not be so broad in its effect to be able to work against the wide variety of strains in people, so further trials were needed to elucidate how to generate the ideal mix of bnAbs.