Introduction
Pulmonary tuberculosis (TB) is an often-lethal bacterial infection caused by
Mycobacterium tuberculosis (
Mtb), which is estimated to have infected one-third of the global population. Currently, over a million people die as a consequence of this infection annually [
1]. Increasing occurrence of multi-drug-resistant
Mtb strains is widespread, and in order to develop novel therapeutic strategies, a better understanding of tuberculosis is required [
2,
3].
During the pathogenesis of TB,
Mtb displays a complex interaction with the immune system of the host.
Mtb is phagocytized by macrophages, in which it prevents acidification and degradation of the phagosomal content [
4,
5]. In addition, there is evidence that
Mtb is able to escape from the phagosomes into the cytoplasm of the macrophage [
6,
7]. In these infected cells, the
Mtb bacteria create a niche, in which they can survive for long periods of time and replicate [
8,
9]. Pro-inflammatory signals from infected macrophages initiate the recruitment of other innate and adaptive immune cells to the primary infection site, leading to the formation of highly organized granulomatous lesions. In these granulomas,
Mtb can persist for many years, forming a latent infection by minimizing its metabolic and replicative activity. However,
Mtb can also be reactivated resulting in an active TB infection [
10].
Despite the fact that
Mtb exploits macrophages to persist inside its host, this cell type is indispensable for the host in order to keep
Mtb infection under control. Macrophages recognize invading pathogens at the first stage of infection and initiate an immune response. Recognition of pathogen-associated molecular patterns (PAMPs) and endogenous danger-associated molecular patterns (DAMPs) occurs through pattern recognition receptors (PRRs), of which the Toll-like receptors (TLR) are one of the major classes [
11,
12]. Myeloid differentiation factor 88 (MYD88) is a key adaptor protein in the TLR signalling pathway since it is used by all TLRs (except for TLR3) to initiate an inflammatory response [
13]. The C-terminal TIR domain of MYD88 enables interaction with TLRs or the interleukin-1 receptor (IL1R), and the N-terminal death domain enables the formation of a “Myddosome” signalling complex, consisting of IL-1 receptor associated kinases (IRAKs). The Myddosome plays a central role in inflammation and host defence by activating the mitogen-activated protein kinase (MAPK) signalling pathway and the nuclear factor-ĸB (NF-ĸB) transcription factor complex [
14‐
17]. MyD88-deficient mice show increased susceptibility to various pathogens, among them
Mtb [
18].
Zebrafish are naturally susceptible to mycobacterial infection, caused by
Mycobacterium marinum (
Mm), which is genetically closely related to
Mtb.
Mm induces a similar pathology to its human equivalent, including the formation of tuberculous granulomas [
19,
20]. The larval stage of the zebrafish enables detailed in vivo imaging and has been used extensively to study host-pathogen interactions during
Mm infection [
19,
21‐
23]. In zebrafish larvae, infected macrophages and neutrophils aggregate and form initial granulomas, which makes this model highly suitable to study the role of innate immune cells during the progression of mycobacterial infection [
24]. The observed granulomas at larval stage appear to be highly dynamic in nature, characterized by the active recruitment of macrophages during early
Mm infection and the reverse migration of infected macrophages from infected sites [
25].
In previous work, we have shown that the Myd88-signalling pathway has a protective role during
Mm infection in zebrafish larvae [
26‐
29]. Larvae from a
myd88 mutant line (
myd88-/-) or
myd88 knockdown larvae showed decreased induction of pro-inflammatory cytokines [
28,
30], lower production of reactive nitrogen species by neutrophils [
26] and attenuated initiation of autophagic defence [
27], resulting in increased rates of infection.
In the present study, we have used the
myd88-/- line to study the effect of Myd88 deficiency on granuloma morphology and subcellular localization of
Mm infection. To this end, we used our previously described tail fin injection model [
24,
31], in which the formation of a single granuloma can be monitored over time and imaged by a combination of confocal and electron microscopy. Our results show that
Mm infection in
myd88 mutant larvae results in an increased bacterial burden associated with strongly reduced recruitment of leukocytes to granulomas. The majority of
Mm was found to be located extracellularly in
myd88 mutants, and bacteria that were found inside cells were mostly observed as aggregates in compartments that were not acidified. These data indicate a specific role for Myd88-dependent signalling in the protection against
Mm infection.
Discussion
In this study, we provide a new insight in the early stages of granuloma development and ultrastructural morphology during
Mm infection in zebrafish larvae using both light and electron microscopy. We show that upon injection of bacteria in the tail fin, a localized infection develops in both
myd88+/+ and
myd88-/- larvae. In the
myd88-/- larvae, the infection developed more rapidly resulting in an increased infection rate at 4 dpi. This is consistent with our earlier observations of an increased bacterial burden in
myd88-/- larvae using a blood island infection model [
28]. In addition, using the tail fin infection model, we observed a clearly different phenotype of infection in the mutant compared to the wild-type larvae (Fig.
1d and e). The results of the present study suggest that the increased bacterial burden is a result of two compromised host-protective processes: the recruitment of leukocytes that can phagocytose bacterial aggregates and the acidification of phagosomes upon lysosomal fusion.
The number of leukocytes at the site of infection was significantly lower in the Myd88-deficient larvae. This lower number of leukocytes at the site of infection at 4 dpi was associated with a significantly lower number of TUNEL-positive cells at this time point. This suggests that the lower number of leukocytes at the site of infection is due to reduced recruitment of these cells in
myd88-/- larvae rather than to an increased level of cell death. However, alternative explanations are possible, such as alterations in haematopoiesis that may be Myd88-dependent, or death of immune cells cell taking place outside the site of infection or at earlier stages. On the other hand, since our infection model only induces a very localized infection at the tailfin of larvae, only a relatively small number of leukocytes interact with the pathogens, which makes systemic effects unlikely. In addition, the total number of leukocytes has been shown to be similar in Myd88-deficient and wild-type larvae at 3 and 5 dpf [
28,
36], so it seems most likely that the lower number of immune cell at the site of infection results from a reduced recruitment of leukocytes in the
myd88 mutant larvae.
In recent studies, a direct link has been demonstrated between reduced macrophage recruitment and increased susceptibility to
Mm infection using zebrafish with genetic or pharmacologically induced macrophage deficiencies. The reduced migration of macrophages in these models results in an impaired supply of macrophages, so apoptotic macrophages in the granuloma are not engulfed by recruited cells. This causes secondary necrosis, breakdown of these granuloma and consequently spread of bacteria and accelerated extracellular growth [
37‐
39].
However, at an early stage of infection (3 hpi), it has been shown that the recruitment of leukocytes towards an
Mm infection site is not dependent on Myd88-mediated signalling [
40]. Apparently, there is a difference in Myd88 dependency between initial and long-term leukocyte response to
Mm infection. The long-term recruitment is likely to be dependent on pro-inflammatory mediators including cytokines and leukotrienes and tissue remodelling factors like matrix metalloproteinases, and the genes encoding these factors are strongly induced during the formation of granulomas in zebrafish larvae [
28]. This induction has been shown to be dependent on Myd88 signalling [
28], which may explain the lower number of leukocytes at the infected site in
myd88-/- larvae observed in the present study. Interestingly, the ratio between the number of dead cells and leukocytes present at the site of infection is higher in the mutant larvae, suggesting a higher percentage of cell death in recruited leukocytes. This may be a result of a higher bacterial load in the mutant larvae, since earlier observations have indicated that the lifespan of leukocytes is negatively correlated with the
Mm load [
24].
Using electron microscopy, we showed that the majority of bacteria in Myd88-deficient larvae were located extracellularly, most likely as a result of the reduced number of phagocytes present at the site of infection. The increased extracellular growth of
Mm in
myd88-/- larvae found in this study is consistent with results obtained using comparable immune-compromised zebrafish models, using knockdown of the TNF and LTA4H expression [
41]. In addition, the
myd88 mutant larvae showed more compacted aggregations of
Mm at the site of infection than the wild types.
Performing ultrastructural analysis using TEM enabled us to quantitatively study the intracellular bacteria that resided in the immune cells of the mutant and the wild type. In previous studies, we showed that the largest fraction of intracellular
Mm in wild-type larvae were observed as electron-dense aggregates [
31], which are a result of efferocytosis [
24]. Efferocytosis, which is defined as reuptake of cell debris and bacterial content by phagocytes upon death of an infected cell, is a crucial innate immune response in the defence against mycobacterial infection [
42]. Interestingly, in the present study, we show that in mutant larvae, only a slight very small fraction was found as electron-dense aggregates and a larger fraction as aggregates without electron-dense content. We therefore conclude that the presence of electron-dense content, representing acidification of these compartments containing larger aggregates, resulting from fusion of the compartment with a lysosome, is highly Myd88-dependent. Restriction of bacterial burden is highly dependent upon acidification of
Mm containing compartments [
43], so the deficiency in this process most likely contributes to the increased bacterial growth in
myd88-/- larvae. Previously, we showed that DNA damage-regulated autophagy modulator 1 (Dram1) plays an important role in the lysosomal acidification of bacteria-containing compartments [
27]. It was also shown in this study that the induction of dram1 expression upon
Mm infection is Myd88-dependent. Therefore, we suggest that the reduced acidification of bacteria-containing compartments in the
myd88 mutant could be at least partly due to a compromised Myd88-Dram1 signalling pathway.
In summary, we have used a combination of light and electron microscopy applied to the tail fin Mm infection model in zebrafish larvae. We show that the inflammatory responses mediated by Myd88 affect the number of leukocytes present at the site of infection as well as the acidification of intracellular compartments. As a result, deficiency in Myd88-dependent signalling leads to an increased infection due to uncontrolled, mainly extracellular, mycobacterial growth.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.