Early clearance versus control: what is the meaning of a negative tuberculin skin test or interferon-gamma release assay following exposure to Mycobacterium tuberculosis?

I. Early recognition of Mycobacterium tuberculosis

The airway is a complex immune organ and likely critical in determining the outcome of Mtb exposure. It exhibits anatomic and functional heterogeneity and contains a diverse array of mechanisms to prevent pulmonary infection, including mucociliary clearance, secreted antibodies, and antimicrobial proteins such as defensins²⁸. Although most research has focused on the alveolus as the initial site of infection, Mtb has ample opportunity to interact with the entire respiratory tract, like other aerosolized particulates. However, how these interactions result in clinically relevant outcomes such as infection and progression to disease is poorly understood. Reuschl et al. used polarized human lung epithelial cells in conjunction with myeloid cells to model these early interactions. Following Mtb infection, they mapped global transcriptomic changes in host cells. Interestingly, they found that myeloid cells could license epithelial cells, through interleukin-1 beta (IL-1β) and type I interferon, resulting in enhanced mycobacterial control³⁰. Additionally, following Mtb infection in the mouse, cathelicidins and related antimicrobial proteins produced by lung macrophages and epithelial cells were required for early clearance of infection³¹. Mounting evidence suggests a role of humoral immunity, including antibodies, and antibody-responsive innate immune cells bearing Fc-receptors, in protective immunity to Mtb^32–35. While studying patients already infected with Mtb, Lu et al. explored 70 different antibody features of total serum IgG from patients with LTBI compared with active TB. The authors’ data suggested an innate antibody Fc effector profile which led to the restriction of bacterial survival within macrophages³³. When evaluating HCs of patients with active TB, Chin et al. observed a different IgA antibody V-gene/D-segment repertoire in those without LTBI compared with those with LTBI, suggesting that a specific type of secretory IgA may promote mucosal protection from Mtb infection³⁶. A recently published genome-wide association study of persistently TST-negative HIV-positive HCs investigated mechanisms of resistance to infection²². Here, HIV-positive contacts who resisted Mtb infection were postulated to have robust innate immunity. Sobota et al. found a significant association between TST negativity with a single-nucleotide polymorphism (SNP) at 5q31.1, which is located between SLC25A48, a mitochondrial amino acid transporter, and IL-9, a cytokine²². IL-9 has been implicated in bronchial hyperreactivity and as a growth factor for mast cells and T cells²². In an effort to search for biomarkers of stages of Mtb infection, Bark et al. evaluated HCs prospectively for differences in serum proteins over two years using mass spectrometry. They compared changes in host proteins in those who were persistently TST- and IGRA-negative, had LTBI, or had TB¹⁵. Tissue integrity proteins, such as keratins, hornerin, lumican, and a component of the extracellular matrix, were more highly expressed in the serum of HCs who did not convert their TST and IGRA. It is interesting to speculate whether these proteins contribute to an essential early barrier against infection. Taken together, these data suggest a distinct role for the respiratory tract as the first line of defense against Mtb infection.

II. Cell-intrinsic mechanisms of host resistance

Mtb virulence is a critical determinant of disease outcome following exposure. Although virulence must be broadly defined, one aspect is certainly the ability of the microbe to circumvent host immunity, such that the outcome reflects this complex interaction. It has been observed that the acquisition of adaptive immunity following infection with Mtb is delayed compared with other infections^37,38. As a result, it has been postulated that Mtb has developed immune-evasive mechanisms to orchestrate this delay, allowing unrestricted replication in the lung. In fact, while using a low-dose aerosol model of TB in non-human primates, Gideon et al. asked whether there are differences in the immune response upon initial infection that determine the severity of the outcome³⁹. Through longitudinal blood transcriptome analysis, these data suggested that a highly orchestrated innate and adaptive immune response was crucial for containment of the bacilli.

To determine whether HCs who resist infection or have LTBI differ in their ability to produce innate cytokines in response to Mtb infection, Mahan et al. employed a whole blood enzyme-linked immunosorbent assay. Here, the responses to a panel of innate receptor ligands indicated no differences between HCs with or without LTBI²⁵. Although these data imply similar innate immune capabilities between those who remain TST-negative and those with LTBI, the authors state the need for more comprehensive study of innate protective mechanisms. Correspondingly, a subsequent study in HCs from the same district indicated that initial recognition of Mtb by distinct innate receptors contributes to controlling infection. Hall et al. performed comparative candidate immune gene SNP analysis in HCs with or without LTBI. They found SNPs in NOD1, NOD2, SLC6A3, STAT1, IL12RB1, IL12RB2, and TLR4 that associated with a persistently negative TST²⁴. As these are intracellular and extracellular sensors of bacteria, this suggests the need for recognition of infection and activation of the host cell in resistance. Interestingly, NOD2 signaling is increased post-BCG vaccination for up to one year through a phenomenon called trained immunity^40–42. Trained immunity is defined as resistance to reinfection and is thought to reflect persistent changes in innate pathways that can lead to memory. Further information from Manzanillo et al. supports intracellular sensing; phagosome acidification of Mtb can be triggered through another cytosolic surveillance pathway recognizing microbial cyclic dinucleotides and DNA⁴³. These data indicate that intracellular sensing of Mtb in the host cell is a crucial trigger to controlling the early infection.

A common feature of innate sensing of extracellular and intracellular Mtb is host cell activation, resulting in microbial killing. Host cells can directly kill Mtb by the production of antimicrobial proteins, reactive oxygen species, and acidification of the phagosome^44,45. Acidification leads to autophagy, which in turn may help prevent or limit infection. Horne et al. found that a human SNP in ULK1 was associated with infection. CRISPR/Cas9 gene editing to delete ULK1 revealed that its deficiency in macrophages resulted in augmented Mtb growth, diminished tumor necrosis factor-alpha (TNF-α) production, and diminished autophagy¹⁷. Other genes associated with the macrophage phagosome, such as SLC11A1 or natural resistance-associated macrophage protein 1 (NRAMP1), have been linked to the development of TB previously⁴⁶. Although its full function remains to be elucidated, three studies have identified phagosome-associated NRAMP1 expression as a protein associated with resistance to infection with Mtb. A seminal study of HCs in Uganda²⁷ used a genome-wide linkage analysis to reveal three regions associated with early clearance. Whereas the two most significant regions did not contain characterized genes, a third contained NRAMP1. Also, an NRAMP1 polymorphism in the 3′ untranslated region (UTR) was more common in Venezuelan hospital workers who were persistently TST-negative than those with LTBI²³. Additionally, the transcriptome of host cell monocytes derived from persistently TST-negative HCs compared with those with LTBI has been explicitly interrogated by Seshadri et al.²¹. Transcripts associated with histone deacetylase (HDAC) inhibition were enriched among persistently TST-negative HCs. Using chemical inhibition of HDACs in monocytes in vitro, the authors observed a role for HDACs in the early immune response to Mtb infection. As HDACs are associated with closed chromatin, this study would support a role for epigenetic programming of host cells in the control of Mtb infection.

Lastly, macrophages capable of non-inflammatory apoptosis and efferocytosis have inherently improved mycobacterial control by limiting intercellular spread⁴⁷. Recent work in animal models of natural transmission supports the hypothesis that augmented apoptosis can result in better control of Mtb infection. In this regard, cows that demonstrate resistance to Mycobacterium bovis infection have persistent TST negativity following herd exposure^48–50. To explore these mechanisms in cattle, Wu et al. generated an sp110-TALEN-mediated knock-in cow strain. The gene Ipr1/sp110 had been previously demonstrated to be associated with innate immunity to TB in mice⁵¹. Sp110 is a nuclear body that permits apoptosis over inflammatory cell death in infected macrophages which decreases the viability of Mtb and limits intercellular spread⁵². In a breakthrough finding, transgenic cattle where murine sp110 is selectively expressed in macrophages were more resistant to M. bovis infection when a cow with active TB was placed in a herd⁵³. Macrophages from transgenic cattle were better able to “restrain” intracellular Mtb and preferentially died by apoptosis over necrosis in vitro. Although nitric oxide (NO) is traditionally thought to play a direct role in TB control, it is also shown to limit granulocytic inflammation and tissue damage in the context of Mtb infection, in part through modification of the inflammasome⁵⁴. Subsequently, it is interesting that Mtb-infected macrophages from TST-negative cattle were recently shown to produce more NO⁵⁰. Therefore, NO, a molecule made in abundance in human lung epithelial cells, has additional functions that may contribute to preventing infection with Mtb. These studies advocate that non-inflammatory cell death and innate antimicrobial responses derived from genetics or trained epigenetics of the host cell can contribute to controlling Mtb infection.

III. Unconventional lymphocytes coordinate early lung inflammation to limit Mycobacterium tuberculosis infection

Longitudinal case control studies (Table 1 and Table 2) would indicate that infection by Mtb can be controlled without an adaptive T-cell response^{18,20,25,55–58}. In addition to macrophages, lymphocytes, especially T cells, could orchestrate this response to Mtb infection. In addition to antigen-specific HLA-I and HLA-II restricted cells that can be reflected in a TST and IGRA, a variety of cells are capable of recognizing the Mtb-infected cell that may not be reflected in these assays. First, it is possible that a TST or IGRA test fails to reflect lung-resident memory T cells that do not circulate systemically^59,60. Specifically, it has been argued that classically restricted tissue-resident memory T cells may recognize Mtb-infected cells and confer protection before the acquisition of a peripheral adaptive immune response^61–63. Second, unconventional T cells, including CD1-restricted cells, MR1-restricted mucosal-associated invariant T (MAIT) cells, HLA-E/Qa-1-restricted cells, and γδ T cells, can also recognize Mtb infection and are poised at mucosal sites such as the lung⁵⁹. Each of these cell populations can also kill infected host cells through cytotoxic granules and produce cytokines in the context of Mtb. Mouse models of TB have been used to decipher whether these subsets have a protective role in the antimicrobial immune response. Using mice deficient in the antigen-presentation element, researchers have observed a role for MR1-restricted MAIT cells and Qa-1-restricted cells in the early response to Mtb infection^64–66. In parallel experiments, group I CD1-restricted T cells specific to Mtb-derived lipids conferred protection in human CD1-transgenic mice^67,68. Mouse models of TB do not suggest a non-redundant role for natural killer T cells or γδ T cells in a chronic TB setting, although they do show changes in frequency and phenotype in association with human Mtb infection^69,70. Moreover, group-I CD1-, HLA-E-, and MR1-restricted T cells specific to Mtb antigen have been detected in individuals with active or LTBI infection^68,71–75. Collectively, these studies indicate a strong possibility that lung-resident T-cell populations play a role in the early immune response to Mtb in humans.

Lastly, natural transmission models of Mtb infection in guinea pigs have provided us with an important example of sterilizing adaptive immunity. Guinea pigs demonstrate a range of susceptibility to Mtb infection in natural transmission experiments^76,77. A unique set-up of these experiments includes the direct exposure of guinea pigs to air from patients. They reflect many features of human immunity such as variable progression to disease. Additionally, some guinea pigs displayed the early clearance phenotype in that they became infected and then reverted to a negative TST. Sterilizing immunity was supported by the observation that the animals did not have granulomas or cultivable mycobacteria, and conversion was prevented by irradiating the air. Furthermore, administration of dexamethasone did not result in TB. As the prevalence of sterilizing immunity in humans is uncertain, future lessons from model systems may provide a paradigm of immune memory that may be crafted into future vaccine strategy or immune therapeutics.