Increased KLRG1 and PD-1 expression on CD8 T lymphocytes in TB-IRIS

Introduction

During successful antiretroviral therapy (ART), a subgroup of HIV patients with a tuberculosis (TB) co-infection are at risk of developing a complication called paradoxical TB-associated immune reconstitution inflammatory syndrome (TB-IRIS) [1]. TB-IRIS is characterized by worsening symptoms of TB, despite an effective initial response to concurrent TB-treatment [2]. Marked by tissue-destructive inflammation and a wide array of symptoms, patients often require additional therapy which increases the cost of patient care [3]. Moreover, diagnosis of TB-IRIS still mainly relies on clinical examinations and is often difficult to distinguish from other complications. Thus, there is an urgent need for reliable laboratory markers to predict this syndrome, since the immune-pathogenesis of TB-IRIS is still not well understood [4].

TB-IRIS typically develops within the first 3 months after starting ART, with the majority of cases occurring before 1 month when CD4⁺ T cells are being replenished [5,6]. Known risk factors of TB-IRIS include a high TB-antigen burden, a short interval between TB treatment and ART and, most importantly, a low CD4⁺ T cell count prior to ART initiation [7–9]. It should be noted, however, that not all HIV-TB patients under similar conditions of immunosuppression develop TB-IRIS. One major characteristic of TB-IRIS is the occurrence of a cytokine storm during the peak of inflammation [10–13]. Thus, the idea that IRIS involves an atypical restoration of immune responses to TB has gained acceptance [5,14,15]. Whereas a dominant role of innate immune cells in the inflammatory cascade during TB-IRIS has become increasingly apparent [11,12,16,17], it still remains unclear which innate or adaptive factors prime the immune system to over-react before ART is administered.

A number of previous TB-IRIS studies have reported pre-ART anomalies in cells belonging to either the innate or the adaptive arm of the immune system. On one hand, increased frequencies of activated CD14⁺ monocytes have previously been reported as a predictor of TB-IRIS [18]. In addition, TB-IRIS patients have been reported to have higher toll-like receptor (TLR)-2 expression on monocytes [19], and a higher degranulation capacity of natural killer (NK) cells prior to starting ART [20]. On the other hand, TB-IRIS patients have been described to have higher pre-ART percentages of activated CD4⁺ T cells and CD8⁺ T cells [21,22]. Nonetheless, other studies did not observe such T cell activation [23,24], whereas we previously observed lower levels of HLA⁺CD38⁺ CD8⁺ T cells in TB-IRIS patients prior to ART [25].

As the predisposing role of these cells in TB-IRIS thus remains elusive, more research is required to fully understand the phenotypic characteristics of T cells, NK cells and monocytes before ART is administered. One of the hallmarks of long term HIV and TB infection is T cell exhaustion and senescence, linked to higher expression of PD-1, KLRG1 and interleukin-27 receptor (IL-27R) on CD4⁺ T cells, CD8⁺ T cells and NK cells [26–28]. Exhausted cells are known to be less functional and could allow the accumulation of antigenic stimuli prior to ART, thus priming the innate immune system for subsequent TB-IRIS development. We therefore hypothesised that higher levels of T cell exhaustion prior to ART would predispose HIV-TB patients with severe immunosuppression to develop TB-IRIS. In addition, we hypothesized that this could be reflected in the expression of TLR related proteins on monocytes [29]. Nested within a large prospective cohort of TB-IRIS patients, this case-control study describes higher pre-ART percentages of CD8^hi T cells and CD3^-CD8^lo lymphocytes with an exhausted phenotype in TB-IRIS patients, which persist during 3 months of ART. In contrast, no differences were observed for CD4⁺ T cells, nor the expression of TLR-related proteins on monocytes.

Methods

Results

Discussion

Decades of research into TB-IRIS have highlighted potential roles of the innate and the adaptive immune system in the inflammatory cascade that characterizes the disease [11,12,16,17]. Nonetheless, the pre-ART mechanisms that predispose the immune system to hyper-react upon ART initiation remain unclear. Despite being a well-documented risk factor, low pre-ART CD4 counts alone cannot account for all TB-IRIS cases. Since up to 25% of HIV-TB patients with severe immunosuppression develop TB-IRIS [2], other antecedent factors have to be at play in order for this complication to occur. We hypothesised that higher levels of T cell exhaustion prior to ART would predispose HIV-TB patients to develop TB-IRIS. Here, we describe higher levels of exhaustion in CD8^hi T cells and a subpopulation of NK cells, but not CD4⁺ T cells, which precede and persist after TB-IRIS. In severely immunocompromised individuals, the functionality of these cells may thus be an additional determining factor in TB-IRIS development.

Given the obvious link between low CD4 counts prior to ART and TB-IRIS, we first assessed the exhausted phenotype of CD4⁺ T cells in TB-IRIS patients. Contrary to our hypothesis but in line with previous studies [23,24], we could not observe any significant differences in CD4⁺ T cell phenotypes that singled out TB-IRIS patients from the HIV⁺ control groups. It is therefore unlikely that CD4⁺ T cell exhaustion specifically contributed to TB-IRIS development in our cohort. Corresponding to our hypothesis, however, the frequency of memory CD8^hi T cells which co-expressed KLRG1 and PD-1 was much higher in TB-IRIS patients before and after starting ART. This finding corresponds to a previous report of higher frequencies of PD-1⁺CD8⁺ T cells in non-pathogen-specific IRIS, although the same was observed for CD4⁺ T cells [33]. Importantly, the exhaustion levels in TB-IRIS patients did not normalize to the level of HIV-negative controls by 3 months of ART, and trended to be remain higher after 9 months, suggesting a more long-lasting effect on the immune system. High levels of immune exhaustion have typically been observed after prolonged periods of antigenic stimulation, in particular with infections such as TB [27] and HIV [34]. It is therefore plausible that a high antigen load at an earlier stage may have driven CD8⁺ specific exhaustion, perhaps even leading to an anergic state of these cells before ART was administered. Of note, TB-IRIS patients in our study showed modestly higher CRP levels prior to ART, and a slightly shorter treatment interval, which could indicate higher antigen levels. However, these differences did not reach statistical significance compared to HIV⁺TB⁺ controls.

Interestingly, our findings on CD8^hi T cells were mirrored by innate lymphocytes as well. Indeed, we observed higher pre-ART frequencies of KLRG1⁺CD8^lo lymphocytes that did not express CD3. CD3^-CD8^lo lymphocytes have previously been reported to mainly consist of NK cells [32]. In addition to CD8^hi T cells, TB-IRIS patients in our cohort thus also experienced exhaustion in a subset of the NK cell compartment prior to ART. Both CD8⁺ T cells and NK cells drive immune responses against intracellular pathogens such as HIV and common co-infections such as TB and cytomegalovirus (CMV), either directly through cytotoxicity or indirectly by interferon-gamma (IFN-γ) mediated stimulation of monocytes/macrophages. A decrease in their function as a consequence of exhaustion could thus hypothetically result in the accumulation of intracellular antigenic signals prior to ART. Although these results advocate for an additional role of the innate immune system in TB-IRIS, we could not observe differential expression of TLR2, TLR4, IL1RL1 or TRAILR on monocytes in TB-IRIS patients. While our results thus cannot confirm an active participation of monocytes in TB-IRIS development prior to ART, they do not exclude a role for monocytes during TB-IRIS, as suggested by previous studies [35,36].

Taken together, the immune-exhaustion observed here could be a consequence of increased pre-ART antigen loads, which have been associated with TB-IRIS before [30]. However, as we did not observe exhaustion on CD4⁺ T cells, our results suggest a specific role for lymphocytes that express CD8 in the early predisposition to TB-IRIS. One of the more attractive theories to date states that a lack of CD4⁺ T cell help in severely immunocompromised individuals could lead to priming of the innate immune system by an increased antigen load. Initiation of ART and the subsequent rise in IFN-γ could thus trigger an excessive response to these accumulated antigenic signals [5]. Nonetheless, even with severe immunosuppression, not all patients develop TB-IRIS. Building on this theory, we now hypothesise that TB-IRIS might find its origins in a form of “immune paralysis” prior to ART. Infections such as HIV, TB and CMV are highly co-endemic, result in long-term infection and are known to cause CD8⁺ T cell exhaustion. Given time, continuous stimulation by any of these pathogens could push CD8⁺ T cells and NK cells past normal activation, into a stage of dysfunctional exhaustion. When CD4 counts reach dangerous levels, the resulting lack of cytotoxicity and/or IFN-γ signalling would prime innate cells with ever increasing antigenic signals, to be unleashed upon ART. HIV-TB patients who do not experience TB-IRIS may not yet have reached an exaggerated level of exhaustion due to less pronounced or shorter exposure to these pathogens.

Due to the retrospective nature of our study, one limitation was the availability of frozen PBMCs. Moreover, the unpredictability of TB-IRIS resulted in a limited number of samples collected during the TB-IRIS event of every patient even further. As such, frozen PBMCs taken at follow-up time points where not available for each single patient included in this study. We thus cannot provide insight in the role of immune exhaustion during the ongoing TB-IRIS inflammation. Nonetheless, 8 out of the 13 patients in each group did have samples available at time points that span a timeframe both before and after TB-IRIS development. This allows for a broad overview of the process leading up to IRIS, as well as the recovery after. While the resulting study population may have been underpowered to detect certain effects (e.g. TLR2 expression on monocytes), the effects observed on immune-exhaustion are thus likely stronger than observed here.

In conclusion, we report high frequencies of exhausted (KLRG1⁺PD-1⁺) CD8^hi T cells and (KLRG1⁺) CD3^-CD8^lo cells which precede TB-IRIS and, in the case of CD8^hi T cells, persist during ART. In contrast, we observed no evidence of altered CD4⁺ T cell or monocyte phenotypes prior to ART. Our findings thus suggest that higher levels of exhaustion in cells specialised in intracellular immunity predispose HIV-TB patients to TB-IRIS. Although a role for the innate immune system in TB-IRIS should not be excluded, the potential causal link with cytotoxic cells prior to ART merits further investigation. Resolving the immune responses leading to TB-IRIS pathogenesis could guide a targeted approach to accurately predict TB-IRIS in patients at risk.

Supporting information

Data Availability

All relevant data are within the manuscript and its Supporting Information files.

Funding Statement

This study was supported by EC FP6 Specific Targeted Research Project (STREP) grant LSHP-CT-2007-037659-TBIRIS. The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.