Granulocytes subsets and their divergent functions in host resistance to Mycobacterium tuberculosis – A ‘Tipping-point’ model of disease exacerbation

Abstract

Granulocytes are innate immune effector cells with essential functions in host resistance to bacterial infections. I will discuss emerging evidence that during Mycobacterium tuberculosis infection, counter-intuitively, eosinophils are host-protective while neutrophils are host-detrimental. Additionally, I will propose a ‘Tipping-point’ model in which neutrophils are an integral part of a feedforward loop driving tuberculosis disease exacerbation.

Introduction

Mycobacterium tuberculosis (Mtb) is the primary cause of tuberculosis (TB), which is one of the leading causes of infectious disease mortality worldwide [1]. Unlike many bacterial pathogens, Mtb is very slow-growing and establishes a chronic infection, often in the lungs. Many immune cell types accumulate at sites of Mtb infection. Granulocytes are abundant innate effector cells that can rapidly enter infected tissues, secrete inflammatory cytokines and lipid mediators, phagocytose pathogens, and excrete extracellular traps. Neutrophils, in particular, are fundamental mediators of host resistance to many bacterial infections. However, recent studies have revealed complex and divergent roles for different granulocyte subsets during Mtb infection. This article will highlight recent advances in our understanding that neutrophils are detrimental during Mtb infection and that eosinophils, typically associated with parasitic infections, are host-protective during TB. Finally, I will propose a ‘Tipping-point’ model in which neutrophils are central effector cells in a feed-forward loop of TB disease exacerbation.

Eosinophils

In acute bacterial infections, neutrophils are often the first granulocytes to respond, and the pulmonary influx of neutrophils is critical to contain acute bacterial pneumonia [2–5]. Likewise, during Mtb infection, which leads to chronic disease, neutrophils were for many years thought to be the first responders in the lung [6]. However, recent work from our group revealed that eosinophils, not neutrophils, are the first granulocytes to be recruited to the airways after Mtb infection in rhesus macaques and mice [7].

Eosinophils are granulocytes involved in a broad range of immune-mediated responses, including those to allergens, parasites, infections associated with type-2 immunity, fibrosis, wound healing, tissue repair, and remodeling [8–10]. They are characterized by large cytoplasmic granules, which store pre-formed bioactive molecules, including an extensive repertoire of cytokines, chemokines, lipid mediators, and cationic granule proteins, allowing them to tailor effector responses to different stimuli [11]. To quantitatively and qualitatively study granulocytes in the host response to Mtb in nonhuman primates (NHP), we developed a flow cytometric approach that distinguishes eosinophils from neutrophils based on the intracellular expression of eosinophil peroxidase (EPX) or myeloperoxidase (MPO) [7,12]. This method revealed that eosinophils are recruited to the airways as early as one week after Mtb and outnumbered neutrophils by two weeks [7]. Sampling of mouse lungs showed a similar enrichment of eosinophils, not neutrophils, in the lung vasculature of infected mice as early as four days after aerosol exposure, followed by recruitment into the lung parenchyma as early as one week after infection [7]. Early eosinophil recruitment, retention, and migration into the Mtb-infected lungs were not dependent on the classical eotaxin-CCR3 pathway often associated with eosinophil responses. Instead, recruitment required the eosinophil-intrinsic expression of GPR183, an oxysterol-sensing chemotactic receptor previously not considered for eosinophil migration after infection [7]. Oxysterols are a group of bioactive cholesterol derivatives downstream of cholesterol-25-hydroxylase (Ch25h), an enzyme that is transcriptionally regulated by innate immune stimuli, including type I interferons (IFN) and TLRs [13]. Eosinophil enrichment and migration into Mtb-infected lungs were abrogated in Ch25h^−/− gene-deficient mice indicating that CH25h derived oxysterols mediate early eosinophil recruitment to Mtb-infected lungs [7]. Importantly, Mtb-harboring, but not uninfected bystander alveolar macrophages, expressed higher levels of Ch25h early after Mtb infection [7,14]. These findings suggest that Mtb-infected alveolar macrophages increase Ch25h expression and oxysterol production early after infection to selectively recruit eosinophils, not neutrophils for cell-cell interactions. Indeed, an Mtb-infected lung explant model for dynamic imaging revealed direct cell-cell interactions between eosinophils and Mtb-infected alveolar macrophages [7]. The biological relevance and consequence of the cell-cell interactions between infected macrophages and eosinophils remain to be experimentally explored.

Nevertheless, in experiments with two separate eosinophil-deficient mouse strains, ΔdblGata (targeted deletion of a high-affinity GATA-binding site in the GATA-1 promoter) and transgenic B6 PHIL (cytocidal diphtheria toxin A is expressed under the eosinophil specific EPX promoter), mice exhibited increased susceptibility accompanied with higher bacterial burden [12]. However, two weeks after Mtb infection, when eosinophils were most abundant in the lung parenchyma of mice and although they directly interacted with Mtb and Mtb-infected cells, eosinophils themselves harbored exceedingly little intracellular bacteria compared to macrophages or neutrophils ([7]). Likewise, at later time-points eosinophils were not significantly infected with Mtb ([12]). Therefore, the exact mechanisms of eosinophil-mediated protection against Mtb infection and in bacterial control remain unknown but are unlikely to involve eosinophils as major cellular targets of Mtb replication and associated cell-autonomous antimicrobial functions [7]. Instead, eosinophils likely provide important protective cues to Mtb-infected cells and the Mtb-infected lung tissue environment, suggesting important, albeit indirect and/or extracellular mechanisms of bacterial control alongside possible direct effects on disease tolerance. Moreover, transcriptional profiling of the lungs of mice lacking eosinophils compared to control mice revealed lipid metabolism and neuro-immune interactions as possible pathways of immune modulation [12].

At later stages of disease in the lungs of mice, macaques, and individuals with active TB disease, eosinophils were an abundant granulocyte subset in TB lesions where their degranulation state inversely correlated with granuloma bacterial burden, validating their critical role in modulating the bacterial spread and host resistance across species [12]. Furthermore, data from three independent clinical cohort studies showed that circulating eosinophil numbers are decreased with increased clinical TB disease status and that decreases in circulating eosinophil numbers correlated with a shorter time to TB-related death [12]. In contrast, neutrophils displayed the opposite pattern in circulation, and indeed elevated neutrophil levels in the blood have been previously associated with active TB disease [12,15–17]. While much remains to be explored about the role of eosinophils in immunity to Mtb, the recent functional studies summarized here suggest a protective role of this granulocyte subset.

Recent observational studies of human and macaque TB granulomas flagged mast cells, a tissue-resident granule-bearing innate cell also typically associated with type-2 immunity, as potentially regulating tissue remodeling and granuloma formation during Mtb infection [18–21]. Intriguingly, both mast cells and eosinophils are more abundant in NHP or human TB granulomatous lesions compared to the lungs of Mtb-infected mice. Thus, mechanistic investigation of type-2-like immune responses involved in granuloma formation, fibrosis, and tissue remodeling is needed to understand better the role of both eosinophils and mast cells in tuberculosis. However, such studies must consider the absence of bonafide TB granulomas, cavity formation, or fibrotic responses in conventional mouse models of Mtb infection. Collectively, the studies discussed here highlight an existing knowledge gap in the function of type-2 associated innate effector cells, like eosinophils and mast cells, in the pathogenesis of TB and the clear need for further research and identification of relevant animal models.

Neutrophils

The most studied granulocyte subset during Mtb infection is the neutrophil, and their complex role in host resistance to Mtb infection has recently been extensively reviewed [17,22–29]. It is now clear that increased neutrophilic responses, including those by granulocytic myeloid-derived suppressor cells (MDSC), are associated with loss of bacterial control in Mtb-infected susceptible mouse models and individuals with active TB [12,15,16,30–39]. Accordingly, recent research on the role of neutrophils and neutrophilic MDSCs during Mtb infection was focused on their disease-promoting properties. Neutrophils are the most common Mtb-containing phagocytes in the airways of active TB patients [40] yet appear unable to effectively restrict intracellular bacterial growth, providing a niche for Mtb to survive [32,36,41–44]. In addition, neutrophils release proinflammatory cytokines, reactive oxygen species, proteases, and neutrophil extracellular DNA traps (NET) that can further promote tissue damage and disease exacerbation during Mtb infection. These neutrophil effector functions likely serve different roles contextualized at different stages and time points of disease and tissue localization within granulomas. For example, neutrophils inside granulomas of Mtb-infected cynomolgus macaques have been reported to express IL-10 [45], an immunomodulatory cytokine produced by MDSCs and low-density neutrophils in human TB [15]. A recent study in mice functionally examined whether immunosuppressive properties of neutrophils or neutrophilic MDSCs, including IL-10 production, may contribute to the pro-bacterial role of neutrophils [41]. Through a series of elegant genetic and bone marrow chimeric approaches, it was shown, however, that neither IL-10 production nor T cell suppressive activity was required for neutrophil-driven disease in susceptible Nos2^−/− mice and C3HeB/FeJ mice, a model widely used to mimic humanTB-like necrotic lung pathology [41]. These experiments suggest that in the absence of Nos2, the disease-promoting properties of neutrophils are less immune-modulatory or suppressive but may directly facilitate bacterial growth by creating a replicative niche [41]. In contrast, even in scenarios and Mtb models where pathogen burdens are unchanged, increased neutrophil responses have been implicated to negatively affect weight loss and survival [37,38,46]. Thus, neutrophils may not necessarily serve as a replicative niche for Mtb growth but could also be involved in the breakdown of disease tolerance.

In addition to detrimental neutrophil responses, dysregulated type I IFN responses emerged as significant contributors to TB disease and pathogenesis in infected mice and humans [47–52]. Since excessive type I IFN responses and neutrophilic inflammation are both associated with active disease in patients and susceptible mouse models [33,51–54], the question remains whether and how type I IFNs and neutrophils synergize to exacerbate Mtb-driven disease. Another recent study [55] revealed that type I IFN worsens disease in susceptible mice by directly inducing neutrophil-mediated lung inflammation and NETosis, a process commonly associated with neutrophil cell death and tissue damage [56]. In this study, GM-CSF blockade in WT Mtb-infected mice increased bacterial loads, type I IFN levels, neutrophil accumulation, and NETosis [55]. GM-CSF blockade in type I IFN receptor (IFNAR)-deficient mice induced less neutrophil accumulation, bacterial burdens, and NETosis, indicating IFNAR has a role in promoting detrimental neutrophil responses and disease exacerbation during GM-CSF deficiency [55]. However, conditional depletion of IFNAR in neutrophils reduced pulmonary neutrophil accumulation and lowered bacterial burden while the degree of NETosis itself remained unaltered [55].

Interestingly, elevated levels of NETosis were also found in the lungs of Mtb-infected C3HeB/FeJ mice, an animal strain known to display exacerbated Mtb-induced tissue necrosis associated with elevated type I IFN responses [55]. These findings suggest that type I IFNs may function through other cell types that extrinsically promote NETosis and raise additional questions about how type I IFNs may directly or indirectly modulate neutrophil responses. NETs themselves can trigger type I IFN production through toll-like receptor 9 (TLR9) signaling and cGAS-STING-dependent activation in settings of autoimmunity [57–60]. Thus, in addition to serving as a potential replicative niche, the detrimental role of neutrophils in host resistance may also be linked to the adverse effects of type I IFNs or the breakdown of disease tolerance. Much remains to be explored about the relative contribution of neutrophils as a cellular target for Mtb replication versus their diverse effector functions and possible roles in lowering disease tolerance.

A ‘Tipping-point’ model of TB disease exacerbation

In several gene-deficient mouse strains rendered susceptible to Mtb infection for seemingly unrelated reasons, ranging from defects in adaptive immunity to molecules involved in innate pattern recognition, specific neutrophil depletion decreases bacterial burden and extends the host survival [31,35,36,39,41,55,61–63]. Thus, the TB disease-amplifying properties of neutrophils can be largely uncoupled from the primary mechanisms underlying the initial loss of host control [22,30,41]. Here, I propose a ‘Tipping-point’ model of TB disease exacerbation where increased neutrophil activation is an integral part of a disease-promoting feedforward loop that occurs as a secondary event, regardless of the initial cause for the loss of bacterial control or disease tolerance (Figure 1). In other words, neutrophils can hasten the host’s demise during tuberculosis, even if they were not the problem in the first place.

Figure 1: A ‘Tipping-point’ model of TB disease exacerbation.

Primary loss of host resistance or disease tolerance results in bacterial growth that precipitates a critical transition (tipping-point) towards a secondary, self-propagating disease exacerbation cascade (DEC). Central features of this program are neutrophilic granulocytes, dysregulated innate inflammatory and cell death pathways, wasting disease, lung tissue damage, and highly amplified bacterial replication.

Many complex systems exhibit non-linear behavior, such that minor perturbations can be absorbed up to a certain point when small changes in certain variables can lead to abrupt shifts in the system’s overall behavior [64–66]. These shifts are often associated with critical transitions or tipping-points, beyond which the system emerges in a state of self-propelled, accelerated catastrophic change [64,67,68]. The tipping-point may be determined by the underlying biological properties of the system as well as by external factors. For example, during Mtb infection, the ‘tipping-point’ may be precipitated by an initial breakdown of bacterial control or a breakdown of disease tolerance, resulting from multiple, diverse failures of host resistance. Regardless of the underlying cause, the primary loss of host control represents a critical juncture, the tipping-point, after which a secondary, self-propagating TB disease exacerbation cascade (DEC) escalates bacterial replication and lung damage, likely facilitating transmission (Figure 1).

While the ability of the host to contain bacterial replication and maintain disease tolerance is influenced by many factors, including bacterial strain and drug sensitivity, host genetics, innate and adaptive immunity, sex, age, nutritional status, stress, and microbiome composition, co-infections and co-morbidities, the subsequent DEC entails a more restricted set of inflammatory pathways. Due to the feed-forward nature, the DEC can become the dominant disease-amplifying driver, outweighing the contribution from the original cause that led to the initial increase in bacterial replication or breakdown of disease tolerance. In this model, neutrophilic granulocytes play a central role (Figure 1). Additional features commonly observed during the DEC of tuberculosis are dysregulated innate inflammation (i.e., TNF-a, IL-1, and type I IFN axes [52,54,69–71]), necrotic cell death and tissue damage [72,73], all of which may further contribute to runaway bacterial replication and wasting disease. While the DEC is clearly dominated by neutrophils, a possible role for eosinophils in the DEC remains to be explored. Currently, possible contributions of eosinophils to this model are likely linked to their protective role in host resistance in mice and may involve prevention and delay of the ‘Tipping-Point’.

This model highlights the difficulty separating cause from consequence (the infamous “chicken-egg” dilemma) in experimental settings of end-stage Mtb-driven lung disease. For example, unrestrained pulmonary bacterial growth may lead to increased inflammation and tissue damage, which may recruit Mtb-permissive phagocytes, etc. These self-propagating feedforward loops are characteristic features of ‘tipping-point’ models beyond the critical transition [64,74]. Accordingly, many DEC core features may influence or directly regulate each other, as exemplified above by the regulation of neutrophils NETosis through type I IFNs [55]. The ‘Tipping-point’ model of TB disease exacerbation may provide a framework to study existing networks between neutrophilic granulocytes, innate cytokines like type I IFNs, IL-1 and TNF-α, necrotic cell death, tissue damage, runaway bacterial replication, and wasting disease as interconnected features of TB disease beyond the tipping-point.

Another important aspect of ‘Tipping-point’ models is identifying so-called ‘early warning’ signs that can predict when a given system is at risk of undergoing a critical transition, such as the DEC [67]. Understanding and predicting essential shifts of complex systems can be accomplished by identifying universal indicators that signal when a system approaches a tipping-point [68]. For example, whole blood transcriptional profiling across many clinical cohorts in TB has revealed an interferon-inducible, neutrophil-driven transcriptional signature that distinguishes patients with active TB disease from those infected but healthy [33,49,50,75]. Moreover, in an adolescent cohort study of TB contacts, an interferon-driven transcriptional signature preceded the onset and diagnosis of active disease by more than a year [76,77]. As such, the interferon-inducible neutrophil-driven transcriptional signature can precede disease diagnosis and could function as an ‘early warning’ sign or leading indicator of the critical transition towards active TB disease.

The suggested ‘Tipping-point’ model of TB disease exacerbation may provide a helpful theoretical framework for contextualizing neutrophilic inflammation that worsens the outcome of TB disease. Findings that bacterial burden or susceptibility of Mtb-infected mice deficient in Ifng^−/− [31], Card9^−/− [35], Acod1^−/− (Irg1) [39], miR-223^−/− [61], Atg5 [32], Gcnt1^−/− [37], Nos2^−/− [41], C3HeB/FeJ also known as Kramnik mice [55], as well as late-stage disease in wild-type mice [78], can all be lowered or reversed with neutrophil depletion support the tipping-point model of TB disease exacerbation. In the case of gene deficiency, the primary cause of bacterial replication or breakdown of disease tolerance is the absence of relevant genes, and the transition towards the DEC through the tipping-point ultimately reveals the increased susceptibility of the system. Notably, despite initially diverse genetic settings, all characterized by ineffective bacterial control and/or disease tolerance, the transition into runaway disease exacerbation remains shared, and neutrophils dominate. Thus, phenotypes in mice may require careful evaluation of whether a mechanism in question may underly the primary susceptibility or operate at the secondary stage of runaway disease exacerbation.

The proposed model may also have practical implications for host-directed therapy (HDT) approaches. HDT strategies aim to modulate antimicrobial effector mechanisms, limit tissue damage and promote disease tolerance rather than target bacteria themselves [79]. Thus, HDTs may be beneficial in TB disease caused by antibiotic drug-resistant strains of Mtb. Therefore, targeting pathways that can be involved in the DEC, such as cell death pathways and type I IFNs [28,52,80], may prove more feasible than identifying and pursuing underlying immune defects that caused initial susceptibility in the context of multigenic and multifactor predispositions for TB within patient populations. One could also speculate that augmenting protective eosinophilic responses that contribute to increased disease tolerance, bacterial control of infected cells, and lung tissue homeostasis may delay the onset of DEC and thus may be a worthwhile HDT approach. Taken together, however, directly modulating the detrimental neutrophilic response likely represents the most promising granulocyte targeting HDT strategy in light of the proposed ‘Tipping-point’ model.

While tipping-point models can help predict the behavior of complex systems, they are also subject to uncertainties and limitations. It, therefore, remains challenging to accurately capture all the relevant factors contributing to a system’s behavior and predict how these factors interplay over time. Nevertheless, identifying and characterizing the components of runaway disease exacerbation and delineating the interaction networks of detrimental granulocytes, tissue damage, cell death, and inflammatory lipids and cytokines will be essential to uncover novel intervention points and druggable targets that aid in the development of treatment strategies targeting advanced TB disease.