Broad CD8 T cell responses are required to overcome dominance of CTL escape in the latent HIV reservoir

Gareth Hardy, HIV i-Base

Results published on 15 January 2015 in Nature reveal that in chronic HIV infection the latent reservoir is almost entirely composed of viral sequences that have evolved escape mutations against cytotoxic T lymphocytes (CTL). [1]

The results published by Kai Deng et al from Johns Hopkins University School of Medicine, have implications for candidate cure protocols that seek to modulate host immune responses against reactivated latent virus. They also suggest that the latent reservoir may be subject to more turn-over than previously thought.

Reservoir elimination strategies might be made more effective by modulating immune responses to prevent viral spread to uninfected cells. CTLs exert strong selective pressure on HIV during acute infection, against which escape mutations rapidly evolve. It is not clear to what degree CTL escape variants constitute the latent reservoir, as infection progresses. If viruses induced from the latent reservoir are resistant to CTL responses, any protocol that relies on boosting CTLs against immunodominant epitopes is unlikely to be effective.

The researchers investigated whether individuals who start ART in chronic infection (at least 3 months after infection), when most of their HIV will have evolved CTL escape mutations, have CTLs that can recognise their own latent virus. They conducted deep-sequencing of gag proviral DNA in resting CD4 T cells to investigate the presence of CTL escape mutations, in 25 HIV-infected individuals, of whom 10 initiated ART during the acute phase (AP) of infection, defined as within 3 months of infection, and 15 initiated ART during the chronic phase (CP).

Deng et al found that proviral sequences from CP-patients were completely dominated by previously documented CTL escape mutations. In several CTL epitopes (SL9, RK9 and TW10) close to 100% of sequences contained CTL escape mutations. In contrast, very few CTL escape mutations were found in the proviral sequences of AP-patients. This striking difference between AP and CP-treated patients suggests that unless ART is started very early, the viral reservoir becomes dominated by variants resistant to common CTL responses.

ELISpot assays were conducted to assess the reactivity of patients’ CTLs to the mutant epitopes versus their wild types, in 7 CP-subjects. Their CTL responses to peptides containing escape sequences were very low, if detectable at all. In contrast, strong CTL responses were observed against wild type peptides. As most proviruses are not replication competent, it was important to determine whether the CTL escape mutations found in patients’ proviruses were capable of replicating and therefore able to contribute to viral rebound. Virus was grown from 9 CP-patients’ CD4 cells following T cell activation and found to include all the CTL escape mutations observed in proviruses. This suggests that the CTL escape variants that dominate the reservoir will be released and replicate if viral latency is reversed.

Deng at al next assessed whether CTLs from CP-patients retained the ability to kill cells infected with escape variants. Autologous CD4 T cells from 13 CP-subjects were infected with autologous, replication-competent virus derived from their own reservoirs and co-cultured with pre-stimulated CTLs. CTLs pre-stimulated with a gag peptide mix eliminated a median of 61% of infected cells from culture, compared with a median of 23% for CTLs that had not been pre-stimulated. Therefore, while patients’ CTLs could still eliminate infected cells despite the presence of CTL escape mutations, they required pre-stimulation with antigen in order to do so.

To characterise which CTL population contributed to killing activity, Deng et al compared the CTL population that targets wild type epitopes in which the reservoir has been found to contain escape mutations, versus the population that targets epitopes which are not mutated in the reservoir. For example in one CP-patient, the viral sequence in the CTL epitope SL9 was found to have escape mutations in close to 100% of that patient’s proviruses, whereas the CTL epitope WF9 had no escape variants observed in the proviruses. CTLs from two CP-patients were pre-stimulated with IL-2 or different synthetic peptides representing the wild-type forms of the relevant epitopes (eg SL9 or WF9) and subsequently co-cultured with autologous CD4 T cells that had been infected with autologous reservoir-derived viruses. The proportion of target CD4 T cells infected with HIV was significantly reduced by CTLs stimulated with single peptide epitopes to which no escape had occurred, in comparison to IL-2 treated cells (p <0.01). In contrast, CTLs stimulated with single peptide epitopes to which escape had occurred had no effect on the proportion of HIV infected target CD4 T cells. The greatest killing effect was observed with CTLs that had been stimulated with a mixture of gag peptides (p <0.001).

In order to assess whether CTLs that recognise unmutated viral epitopes are capable of inhibiting HIV replication and clearing infected cells, Deng et al generated a humanised mouse model using bone marrow from two CP-patients. Autologous virus was grown in the lab from the two patients viral reservoirs, which were used to infect the mice. Two weeks later, the mice were infused with autologous CD8 T cells from the donor patients that had been stimulated with viral peptides.

In control mice (where infused patient CD8 T cells had not been pre-stimulated with viral peptides) or mice that had received CD8 T cells that had been pre-stimulated with the wild type variant of the SL9 peptide (against which CTL escape had evolved), both plasma viral load and proviral DNA increased from day 14 to day 29 after infection. In contrast, mice infused with CD8 T cells that had been pre-stimulated with the wild type variant of the WF9 peptide (against which no CTL escape had evolved) experienced significantly lower levels of viral replication than controls or SL9-CD8 T cell-treated mice (p <0.05). Mice infused with CD8 T cells that had been pre-stimulated with a peptide mix experienced 100-1000 fold less viral replication than control or SL9-CD8 T cell-treated mice (p <0.05). Dramatically, two of three of the mice infused with CD8 T cell stimulated with the peptide-mix achieved undetectable levels of both plasma viral load and peripheral blood proviral DNA.

These data demonstrate that only CTL clones that target subdominant unmuated epitopes are effective against the CTL escape variants that dominate the viral reservoirs of patients who initiate ART during chronic infection. However, CTLs that target immunodominant epitopes, to which HIV will have evolved escape, are unlikely to have an effect on viral replication. These results have implications for immune-based viral eradication strategies, and suggest that reversal of viral latency should be supported by induction of subdominant CTL responses.

Another important implication of this data is that the viral reservoir may be less stable than thought. To be reconstituted with immune escape variants between acute and chronic HIV infection the reservoir must undergo replenishment. This observation alone should be grounds for reconsidering the assumed stability of the HIV reservoir.


  1. Kai Deng et al. Broad CTL response is required to clear latent HIV-1 due to dominance of escape mutations. Nature. 15 January 2015. Vol 517. 381-387.

Links to other websites are current at date of posting but not maintained.