Features of Functional and Dysfunctional CD8⁺ T cells to Guide HIV Vaccine Development

Abstract

Purpose of Review

CD8⁺ T cell responses are a key component of the host immune response to HIV but vary significantly across individuals with distinct clinical outcomes. These differences help inform the qualitative features of HIV-specific CD8⁺ T cells that we should aim to induce by vaccination.

Recent Findings

We review previous and more recent findings on the features of dysfunctional and functional CD8⁺ T cell responses that develop in individuals with uncontrolled and controlled HIV infection, with particularly emphasis on proliferation, cytotoxic effector function, epitope specificity and responses in lymph nodes. We also discuss the implications of these findings for both prophylactic and therapeutic T cell vaccine development within the context of T cell vaccine trials.

Summary

The induction of HIV specific CD8⁺ T cell responses is an important goal of ongoing vaccine efforts. Emerging data on the key features of CD8⁺ T cell responses that distinguish individuals who spontaneously control from those with progressive disease continues to provide key guidance.

Introduction

CD8⁺ T cells play a pivotal role in the adaptive immune response to HIV infection. However, progressive HIV disease induces a dysfunctional T cell phenotype due to chronic antigen exposure. In rare instances, CD8⁺ T cells are able to durably control HIV infection due to a combination of unique functionality, specificity and anatomic location – findings which are being actively leveraged to generate new prophylactic and therapeutic T cell vaccines. In this review, we discuss recent advances in each of these key areas of HIV T cell immunology and vaccinology.

Features of Dysfunctional CD8⁺ T cells During Chronic HIV Infection

Functional HIV-specific CD8⁺ T cells have the capacity to proliferate, produce cytokines and release cytotoxic effector molecules, with proliferative capacity being the single strongest predictor of an effective response [1]. However, because of the persistent nature of HIV infection, the vast majority of HIV-specific CD8⁺ T cell responses become dysfunctional due to chronic antigen exposure, resulting in a loss of cytokine secretion, cytolytic activity and proliferative capacity [2]. Dysfunctional CD8⁺ T cells also demonstrate increased and sustained expression of inhibitory receptors. Among these is the canonical marker programmed cell death protein 1 (PD-1), which was classically shown to be upregulated on chronic antigen-exposed HIV-specific CD8⁺ T cells [3,4] and strongly correlated with measures of disease progression, such as viral load and low CD4⁺ T cell count. Subsequent studies have demonstrated that PD-1 functionally impairs T cells by upregulating the transcription factor BATF, which alone is sufficient to disrupt T cell proliferation and cytokine secretion [5].

In addition to PD-1, co-expression of other surface molecules, such as T cell immunoglobulin and immunoreceptor tyrosine-based inhibitory motif domain (TIGIT) and T cell immunoglobulin and mucin-domain containing protein-3 (TIM-3), have been linked to more severe CD8⁺ T cell dysfunction in both mouse models of chronic viral infection (i.e. lymphocytic choriomeningitis virus; LCMV) and HIV [6–9]. Compared to HIV-specific CD8⁺ T cells expressing PD-1 alone, those expressing a multitude of inhibitory receptors were shown to have even more reduced proliferation and secretion of cytokines IL-2 and IFN-γ [7]. In addition, CD8⁺ T cells expressing PD-1, TIGIT and TIM-3 have been shown to have altered glucose metabolism, which is part of an emerging set of observations regarding the critical role of glycolysis and metabolic plasticity in maintaining antiviral CD8⁺ T cell activity [10–12]. In fact, recent work has shown that restoration of HIV-specific CD8⁺ T cell function following blockade of PD-1 and TIGIT was enhanced by utilization of pro-glycolytic drugs in combination [13].

The loss of CD8⁺ T cell proliferative capacity is key feature of the dysfunctional cellular immune phenotype that emerges during chronic HIV infection. Prior work has demonstrated that decreased proliferation of HIV-specific CD8⁺ T cells was associated with increased necroptotic cell death and reversed by small molecule scavengers of mitochondrial reactive oxygen species [14]. This is consistent with numerous recent studies elucidating the contribution of altered mitochondrial function to HIV-specific CD8⁺ T cell dysfunction [15,16]. Longitudinal studies of individuals who spontaneously lose immune control of HIV (e.g. previous controllers) have further demonstrated that HIV-specific CD8⁺ T cells develop diminished proliferative capacity prior to increases in plasma viremia [17]. Importantly, this early stage of CD8⁺ T cell dysfunction was not characterized by increased surface expression of inhibitory molecules (e.g. PD-1, TIGIT, TIM-3). However, RNA sequencing analysis did identify transcription factor KLF2 as a putative regulator of early HIV-specific CD8⁺ T cell dysfunction , which has previously been shown to modulate CD8⁺ T cell proliferation [18].

While substantial progress has been made towards understanding inhibitory molecule expression and metabolic features of T cell dysfunction in chronic HIV infection, studies of the epigenetic CD8⁺ T cell landscape have revealed additional features. In mouse models and primary HIV infection, antigen-specific CD8⁺ T cells undergo extensive changes in chromatin accessibility that are largely irreversible and therefore reflect a stable dysfunctional state [19,20]. This epigenetic “scarring” remains fixed even when the antigenic stimulus is removed, such as following anti-retroviral therapy (ART) [21], and greatly hinders functional memory CD8⁺ T cell differentiation resulting in compromised recall capacity [22]. This is consistent with previous reports of demethylation of the PD-1 locus in HIV-specific CD8⁺ T cells, which also remained following initiation of ART [23]. Another recent study showed that even during early HIV infection, individuals had substantial epigenetic changes in HIV-specific CD8+ T cells, similar to those seen in chronic HIV infection, which only partially shifted with ART initiation [24]. Collectively, these findings reveal a need for therapies that can counteract this extensive epigenetic remodeling, in addition to ones aimed at blocking inhibitory receptors, in order to effectively reverse T cell dysfunction.

Features of CD8⁺ T cell Responses During Spontaneous HIV Control

While most untreated individuals infected with HIV experience uncontrolled viral replication and develop progressive HIV-specific CD8⁺ T cell dysfunction, approximately 1 in 300 individuals control viral replication below the thresholds associated with transmission and disease progression [25]. These spontaneous controllers have provided insights into the features of CD8⁺ T cell responses that are able to control viral replication and provide guidance on the types of T cell responses that we should aim to achieve through vaccination for prevention of progressive HIV disease or therapeutic HIV cure strategies.

Studies of HIV-specific CD8⁺ T cells in spontaneous controllers have consistently revealed that they have highly advanced functional properties, which include the ability to secrete multiple cytokines, robustly proliferate in response to HIV antigen, upregulate cytotoxic effector molecules within lytic granules and suppress autologous viral replication [1,26–30]. These polyfunctional and proliferative characteristics are observed not only in controllers with robust effector HIV-specific CD8⁺ T cells, but also in those with low to absent effector CD8⁺ T cell responses of a largely central memory phenotype that rapidly expand upon exposure to ex vivo HIV antigen [31], putatively due to increased expression of transcription factor TCF-1 [32]. This potentially explains why some studies have suggested that spontaneous controllers lack functional and cytotoxic CD8⁺ T cell responses, when it is likely the lack of recent in vivo antigen exposure in these extraordinarily virally suppressed individuals which leads to poorly detected circulating HIV-specific CD8⁺ T cells in the absence of antigenic stimulation.

In addition to the functionality of HIV-specific CD8⁺ T cells, epitope specificity is another key determinant of spontaneous HIV control [33]. This was initially suggested by the observation that specific HLA class I alleles (e.g. HLA-B*57), which present a set of HIV epitopes distinct from other HLA alleles, were consistently enriched in cohorts of spontaneous HIV controllers [34–36]. This influence of HLA class I, with some alleles conferring protection and others conferring risk, was confirmed in several independent genome-wide association studies [37–41]. Moreover, these studies found that specific amino acid positions within the HLA-B peptide binding groove that facilitates epitope binding and presentation (amino acids 67, 70 and 97) were critical for defining protective (HLA-B*57, HLA-B*27, HLA-B*52, HLA-B*14) and risk HLA class I alleles [40,42]. A more recent GWAS, which analyzed a larger and more diverse multi-ancestry population, identified HLA-B residue 156 as also being independently implicated in HIV control, in addition to residues at positions 67 and 97 [43]. Collectively, these studies suggested that the distinct epitope presentation by HLA class I alleles for recognition by HIV-specific CD8⁺ T cells is a major genetic determinant of spontaneous HIV control.

Given the immense range of intra-host viral sequence diversity, it was initially hypothesized that targeting of epitopes with a high degree of sequence conservation could explain spontaneous control. However, no such association was observed when comparing the sequence entropy of epitopes targeted by spontaneous controllers and individuals with progressive disease [44], which may be attributable to the lack of clear association between residue conservation and mutational constraint that has been observed for HIV and other model proteins [45–47]. In a subsequent study, application of network theory to protein structure led to the development of a new method – structure-based network analysis – which outperformed sequence conservation in its ability to identify mutation constrained residues [48]. The structure-based network analysis approach was then applied to the HIV proteome and identified both mutation-constrained residues and CD8⁺ T cell epitopes (i.e. ‘highly networked’ epitopes), which were among the list of optimal A-list epitopes that have established immunogenicity [49]. Highly networked epitopes were preferentially targeted by the functional and proliferative CD8⁺ T cell responses in spontaneous controllers, although this was not observed in HIV progressors. These highly networked epitopes are presented by a diverse array of HLA class I alleles, but have been demonstrated to preferentially stabilize protective HLA class I alleles [50] as identified by GWAS [40], providing a molecular rationale for the observed genetic associations between HLA alleles and HIV control. A recent study of a specific highly networked epitope (QW9, Gag p24_176–184) demonstrated that cross-reactive HIV-specific CD8⁺ T cells of the wild-type epitope and a naturally arising variant (S3T) in spontaneous controllers were only present when the epitope was presented by a protective HLA class I allele (B*5701), but not a non-protective allele (B*5301) [51]. This suggests an additional potential advantage ascribed to protective HLA class I alleles that facilitates durable immune control. Importantly though, a distinct HIV-specific CD8⁺ T cell response that recognized the variant QW9 S3T epitope was present in both HLA class I allele backgrounds, as has previously been described [52], indicating that the QW9 S3T networked epitope mutation does not lead to immune escape.

Recent studies of the latent viral reservoir in spontaneous controllers have revealed an enrichment of genomes present in transcriptionally repressed sites [53], with emerging evidence suggesting that this is the result of ongoing immune selection pressure [54]. These data suggest that HIV-specific immune responses may be able to achieve a possible cure of HIV infection, which has been suggested in a small number of distinct individuals [55,56]. Among the barriers to this level of immune-mediated HIV clearance however, are follicular CD4⁺ T cells, which are a major cellular harbor of the HIV reservoir in both HIV progressors and controllers [57–59]. While CXCR5⁺ follicular CD8⁺ T cells have been shown to be important for control of viral infections within lymph nodes for a number of pathogens, including SIV [60–63], the immune-privileged status of the lymph node follicle has raised questions as to the mechanism and extent that CD8⁺ T cells can suppress HIV replication in lymphoid tissue [64,65].

Recent studies evaluating the lymph nodes of spontaneous controllers identified an enrichment of antiviral tissue-resident HIV-specific CD8⁺ T cells in lymphoid tissues [66], which were suggested to control viral replication within the follicle without demonstrable cytolytic activity [67]. However, similar to studies of HIV controllers with low to absent effector CD8⁺ T cell function [31], the minimal antigenic viral load in these study participants and the absence of ex vivo antigen stimulation, likely contributed to a primarily memory CD8⁺ T cell phenotype and observed low ex vivo cytolytic activity. Consequently, a recent study demonstrated that when HIV-specific CD8⁺ T cells derived from the lymph nodes of spontaneous controllers were stimulated with ex vivo antigen, they did in fact upregulate high levels of cytotoxic effector molecules perforin and granzyme B [68], indicating the clear capacity for cytolytic function. Moreover, cytotoxic CD8⁺ T cells could be identified near foci of active viral replication in lymph nodes and expression of perforin and granzyme B was directly correlated to their proximity to HIV-infected cells. Collectively, these findings suggest that HIV-specific CD8⁺ T cells are a key component of immune control at relevant tissue sites that harbor the latent reservoir such as lymph nodes and that cytotoxic effector function likely maintains control of ongoing viral replication.

Implications for HIV T cell Vaccine Development

Elucidating the features of successful and unsuccessful HIV-specific CD8⁺ T cell responses during natural HIV infection provides guidance for the development of efficacious prophylactic and therapeutic HIV T cell vaccines. These findings will be critical to move beyond the STEP and Phambili trials [69,70], which despite inducing robust HIV-specific CD8+ T cell responses in >75% of participants after vaccination with an adenovirus serotype 5 vector encoding full-length HIV Gag/Pol/Nef subtype B, failed to show evidence of protection or reduction in viral setpoint. These disappointing results led to a shift in effort towards the induction of HIV-specific broadly neutralizing antibody responses (bNab), but this has also been marked by challenges despite some recent successes inducing bNab precursors in humans [71]. In addition, insights from the Antibody Mediated Prevention (AMP) trial demonstrate that particularly high serum levels of bNabs will likely be needed to protect against HIV acquisition [72]. Consequently, there has been a growing interest in approaches that utilize T cell vaccines to bolster antibody-based approaches. A recent study demonstrated that non-human primates vaccinated with heterologous viral vectors (HVVs) expressing SIVmac239 Gag (shown to induce durable and functional CD8⁺ T cell responses) in concert with prototype native-like trimer BG505 SOSIP.664 (shown to produce high neutralizing antibody titers) resulted in the majority of dual vaccinated macaques being protected against infection at a lower nAb titer [73]. This suggested that vaccine-induced CD8⁺ T cells may be able to synergize with nAbs to lower the threshold needed for protection, providing support for additional investigation into combined B and T cell vaccination approaches.

A central component to these T cell vaccine efforts, as revealed by work on spontaneous controllers, will be the development of immunogens that induce functional HIV-specific CD8⁺ T cell responses that are focused on mutationally constrained sites. Vaccines that incorporate full-length proteins, such as the one used in the STEP trial, are likely to be suboptimal as evidenced by the observed sieve effect where breakthrough infections in vaccine recipients have significant genetic distance at mutable HLA-restricted epitope sites specifically within vaccine-encoded Gag, Pol and Nef [74]. Newer immune-focusing vaccine candidates include those that specifically incorporate multivalent mosaic antigens [75,76], highly sequence conserved viral regions [77–79] or epitopes that have been statistically associated with spontaneous controllers [80]. These T cell-based vaccines have been largely tested in therapeutic vaccine trials with mixed results. The BCN02 study utilized a conserved HIV T cell immunogen [77] delivered by Modified Vaccinia Ankara (MVA) virus vaccination, followed by three doses of latency reversal agent Romidepsin [81,82] and then a second MVA boost, before ART cessation [83]. Interestingly, ~23% of individuals, who had previously been vaccinated in the BCN01 study [84] prior to enrollment in BCN02, showed sustained suppression of plasma viremia up to 32 weeks. In comparison, the AELIX002 study utilized the HIVACAT T cell immunogen (HTI) delivered by a combination of DNA, MVA and ChAdOx.1 vector [85] to perform a randomized, placebo-controlled study in early ART treated individuals to evaluate the safety, immunogenicity and therapeutic effect of this vaccine regimen on viral rebound [86]. While there was no significant difference in the percentage of participants in each study group that remained off ART, there was an apparent difference in placebo and HTI-vaccine recipients who lacked protective HLA class I alleles (although some of the alleles classified as protective, e.g. B*15, have not previously been delineated as such by GWAS) [37,40,43]. Nonetheless, this study did demonstrate significant correlations between CD8⁺ GzmB⁺ HTI-specific CD8⁺ T cell responses and both time off ART and viral load at the end of the acute treatment interruption, further illustrating the importance of cytotoxic CD8⁺ T cell responses, as was observed with spontaneous controllers.

Vaccine studies involving highly networked epitope immunogens remain in development, although an ex vivo DC-based priming model demonstrated a relative enhancement of CD8⁺ T cell response induction in comparison to both full-length Gag and conserved antigens [87]. Additionally, while highly networked epitopes are derived from the optimal A-list [49], they have quite limited overlap with other epitope-based immunogens [88], indicating that they will target distinct regions of the viral proteome. The key next step is determining the optimal vaccine modality to elicit highly functional and de novo CD8⁺ T cell responses, particularly in cure settings given concerns regarding the irreversible dysfunctional features of pre-existing responses [21]. A number of recent studies using neoantigen epitope-based vaccines in novel formats, such as mRNA [89,90] and heterologous simian adenovirus/self-amplifying RNA immunization [91] provide reason for optimism. Inducing tissue-resident HIV-specific CD8⁺ T cell memory responses at relevant mucosal sites will also be another important part of future vaccine efforts [92].

Conclusion

Studies of HIV-specific CD8⁺ T cells in progressive and controlled HIV infection have provided critical insight into their function and dysfunction, specificity and features in both blood and lymph nodes that modulate immune-mediated outcomes to HIV infection. With these findings in hand, we now collectively move forward to realize the potential and power of CD8⁺ T cells to limit HIV acquisition and curtail pathogenesis through continued T cell vaccine development.

Key Points.

• Progressive HIV disease leads to a dysfunctional HIV-specific CD8⁺ T cell phenotype that is increasingly observed as being irreversible even with fully suppressive treatment.

• Successful T cell-mediated control of HIV infection is mediated by a highly functional and cytolytic response directed towards mutationally constrained epitopes in blood and anatomical sites that harbor foci of ongoing viral replication.

• Prophylactic and therapeutic HIV T cell vaccines are being developed to induce responses similar to those observed in spontaneous controllers.