Drosophila versus Mycobacteria: A model for mycobacterial host–pathogen interactions

Animal models have played an essential role in understanding the host–pathogen interactions of pathogenic mycobacteria, including the Mycobacterium tuberculosis and emerging nontuberculous mycobacteria (NTM) species such as M. avium and M. abscessus. Drosophila melanogaster has become a well‐established model for the study of innate immunity and is increasingly being used as a tool to study host–pathogen interactions, in part due to its genetic tractability. The use of D. melanogaster has led to greater understanding of the role of the innate immune system in response to mycobacterial infection, including in vitro RNAi screens and in vivo studies. These studies have identified processes and host factors involved in mycobacterial infection, such as those required for cellular entry, those required to control or resist non‐pathogenic mycobacteria, or factors that become dysregulated as a result of mycobacterial infection. Developments in genetic tools for manipulating mycobacterial genomes will allow for more detailed studies into how specific host and pathogen factors interact with one another by using D. melanogaster; however, the full potential of this model has not yet been reached. Here we provide an overview of how D. melanogaster has been used to study mycobacterial infection and discuss the current gaps in our understanding.


| 601
MARSHALL And dIOnnE multiple pathogen detection pathways in D. melanogaster, defined several cytokine-like immunomodulatory signals, and explored the regulation and activities of different effector arms of the D. melanogaster immune response (Lemaitre & Hoffmann, 2007).
With increasing knowledge of D. melanogaster immunity, there has been growing interest in using the fruit fly as a model host for the exploration of host-pathogen interactions. As with other invertebrates, D. melanogaster lacks an adaptive immune response, allowing for study of the role of the innate immune response in host-pathogen interactions. Highly conserved components involved in innate immune signaling pathways are frequently targeted by pathogens during infection. D. melanogaster can serve as a host for a wide range of pathogens and is easily experimentally infected via injection, wounding, or feeding. Additionally, the ease of use, accessibility, low cost, and genetic tractability of D. melanogaster allows for a combination of both prokaryotic and eukaryotic genetic manipulation to probe host-pathogen interactions. Alongside in vivo studies, the use of D. melanogaster cell lines in high-throughput RNAi screens has led to the identification of cell-autonomous factors interacting with a range of pathogens, including mycobacteria (Cherry, 2008). (CF) or bronchiectasis. They are extremely challenging to treat, in part due to the high levels of intrinsic antibiotic resistance characteristic of these organisms (Shen et al., 2018).

| Mycobacteria
The unique properties of mycobacteria that make these species so successful also make them inherently difficult to study. While more closely related to Gram-positive bacteria, the membrane structure of mycobacteria resembles that of Gram-negative bacteria due to the presence of the mycomembrane (Touchette & Seeliger, 2017). This is a double bilayer outer membrane (Bansal-Mutalik & Nikaido, 2014), with the inner leaflet formed of long-chain mycolic acids covalently anchored to an arabinogalactan-peptidoglycan complex (Daffé et al., 2017). The structure of the outer leaflet is debated and possibly varies across mycobacterial species, with some studies suggesting the presence of mycolates, phospholipids, or triacylglycerols (Bansal-Mutalik & Nikaido, 2014;Chiaradia et al., 2017).
This highly impermeable membrane contributes to the high intrinsic resistance of mycobacteria to antibiotics and other environmental stresses (Jarlier & Nikaido, 1994;Kirschner et al., 1999;Schulze-Robbecke & Buchholtz, 1992;Taylor et al., 2000) as well as their slow growth. The impermeable nature of the mycomembrane contributes to mycobacterial survival in intracellular compartments; many of its components are also bioactive, as targets of recognition by host receptors, as virulence factors, or both (Alderwick et al., 2015;Bernut et al., 2014;Lavollay et al., 2011;Mahapatra et al., 2008;Squeglia et al., 2018).
The focus of this review is on studies of interaction between mycobacteria and D. melanogaster. Many such studies have focused on using NTM species such as M. marinum as a model for tuberculosis.
However, M. tuberculosis is a host-adapted pathogen: the biology of its interaction with humans has been shaped by coevolution, possibly limiting the tuberculosis-specific insights to be gained by study of interactions of NTM's with nonhuman hosts (Bryant et al., 2021). In contrast, NTM species are found widely in the environment and are often not adapted to specific hosts, suggesting that D. melanogaster should be used to study more general mechanisms of mycobacterial pathogenesis shared across many of these species. This approach, in combination with studies using other animal models used to study NTM such as zebrafish (Bernut et al., 2014), is providing important insights into the fundamental shared biology of mycobacterial disease. This is especially timely due to the increasing prevalence of NTM disease (Haworth et al., 2017). We would suggest that this is where the true value of these studies is to be found. Here we provide an overview of D. melanogaster as a model for mycobacterial infection, including both pathogenic species such as M. marinum and M. abscessus, and non-pathogenic species such as M. smegmatis.
We highlight findings into the role of the innate immune response in controlling mycobacterial infections, as well as factors found to be important in intracellular survival of mycobacteria identified from D. melanogaster studies. Finally, we discuss the broad potential D. melanogaster studies have in further understanding mycobacterial host-pathogen interactions.

| DROSOPHIL A HOS T DEFEN CE
Humoral and cellular defence mechanisms comprise the innate immune response of D. melanogaster (Lemaitre & Hoffmann, 2007). An inducible humoral response leading to the secretion of antimicrobial peptides (AMPs) and production of reactive oxygen species (ROS) is mediated by two core microbe-detection pathways, termed IMD and Toll after their best-known components (Lemaitre et al., 1996). Cellular immunity is mediated via phagocytes that appear closest in their biology and function to mammalian macrophages. The cellular and humoral immune compartments communicate via several known cytokine signaling systems. Additionally, microorganisms can be sequestered by coagulation, which also contributes to wound healing.
Epithelial immunity ensures the integrity of the gut lumen with local production of AMPs and ROS (Lemaitre & Hoffmann, 2007). The role of each of these host defence systems will now be discussed in turn relating to their role in controlling mycobacterial infection.

| Humoral response
Drosophila melanogaster has two known signaling pathways that are directly activated by bacterial components. These are the Toll and IMD pathways, both of which are activated by bacterial peptidoglycan (Govind, 2008). The IMD pathway is generally responsive to peptidoglycan containing m-DAP, while the Toll pathway responds to peptidoglycan containing l-lysine. The structure of peptidoglycan in mycobacterial species contains meso-diaminopimelic acid (m-DAP) in the peptide stem, rather than lysine as is more common among Gram-positive bacteria, suggesting a role for the IMD pathway. These allow D. melanogaster to roughly differentiate between Gram-negative and Gram-positive bacteria (Buchon et al., 2014). The IMD pathway can also be activated by dmSTING, the D. melanogaster ortholog of STING, which acts as an intracellular detector of cyclic dinucleotides (Martin et al., 2018). Both IMD and Toll pathways act primarily via activation of NF-κB-related transcription factors (Relish for the IMD pathway, Dif and Dorsal for the Toll pathway) (Hoffmann & Reichhart, 2002;Tzou et al., 2002). These trigger the expression of a large suite of AMPs, with specific AMP expression controlled by specific pathway activation induced by different pathogens, however, wounding can also induce expression of AMPs. Some target AMP genes are preferentially induced by one pathway or the other.
For example, the AMP Drosomycin is primarily a Toll pathway target, whereas Diptericin is primarily an IMD pathway target (Buchon et al., 2014;Lemaitre et al., 1997). In addition to AMPs, a melanization system functions downstream of Toll pathway activation. This system relies on secreted prophenoloxidases to produce ROS and is the primary ROS-generating bactericidal system in D. melanogaster humoral immunity, responsible for clearance of bacterial species such as Staphylococcus aureus (Binggeli et al., 2014;Dudzic et al., 2019). Several puzzles remain. The apparent delay in recognition of mycobacteria by the humoral immune system relative to other bacteria suggests that the mycomembrane may shield the mycobacterial peptidoglycan from detection due to reduced efficiency in detection, though this has not yet been determined. It is also unclear why these mycobacteria appear to activate the Toll pathway rather than the IMD, despite mycobacterial peptidoglycan containing m-DAP, not llysine peptidoglycan typically associated with Toll-pathway activation (Buchon et al., 2014;Lavollay et al., 2011). An alternative mycobacterial microbe-associated molecular pattern (MAMP) such as a glucan, Glycopeptidolipid, or other factors in the mycomembrane may be recognized by receptors that trigger activation of the Toll pathway. Alternatively, it may be activated by the widespread tissue damage caused by the M. abscessus and M. marinum infections (Dionne et al., 2003;Oh et al., 2013b)-though this would not explain the requirement for the Toll pathway in the clearance of M. smegmatis. It is also possible that mycobacterial cyclic dinucleotides are recognized by dmSTING, and that there is some synergy between both Toll and IMD pathway signaling (Martin et al., 2018;Tanji et al., 2007). Further work will be required to understand the nature of these interactions.   (Philips et al., 2005(Philips et al., , 2008. Additionally, the ideal culture temperature of S2 cells at approximately 28°C closely matches the optimum growth temperature of many nontuberculous mycobacteria, such as M. marinum, which are typically found in environmental sources. However, S2 cells-and intact Drosophila-are limited in the study of M. tuberculosis and other obligate human pathogens, as the optimum growth temperature for those bacteria is closer to 37°C, which causes cytotoxic heat shock response in S2 cells (Koo et al., 2008).  (Chan et al., 2002). The combination of map24/map49-based GFP reporters, automated microscopy, and simple, highly-effective RNAi protocols enabled early genome-wide screens for host genes that affected intracellular growth of M. fortuitum in phagocytic Drosophila S2 cells (Philips et al., 2005). In total, 81 host factors were found to be required for M. fortuitum infection of S2 cells. 54 of these host factors were found to be required for gen-  (Philips et al., 2005). Further characterization was performed on one host factor specifically required for M. fortuitum uptake, a CD36 family scavenger named peste (Pes). Pes was found to be required for M. fortuitum infection of S2 cells, as well as entry of M. smegmatis and Listeria monocytogenes, but played no role in uptake of S. aureus or E. coli. Transformation of human embryonic kidney 293 (HEK293) cells to express Pes allowed them to be infected with M. fortuitum, to which they are normally resistant (Philips et al., 2005).

| In vitro cell culture screening
Following entry into cells, many pathogenic mycobacteria can resist intracellular killing by inhibiting phagosome maturation and fusion with lysosomes (Clemens & Horwitz, 1995). So, somewhat unexpectedly, an RNAi screen in D. melanogaster S2 cells revealed lysosomal β-hexosaminidase to play a role in controlling M. marinum infection (Koo et al., 2008). M. marinum was shown to be internalized by S2 cells within 2 hr of inoculation, then showed steady intracellular growth for 3-4 days before lysis of the S2 cells. Next, S2 cells were treated with dsRNA directed against a range of D. melanogaster genes and infected with M. marinum to identify genes that, when knocked down, led to increased M. marinum growth. RNAi knockdown of HEXO2 in S2 cells, a homolog of mammalian lysosomal enzyme β-hexosaminidase β-subunit, led to a clear increase in intracellular M. marinum (Koo et al., 2008).
This finding was replicated in murine bone marrow-derived macrophages (BMDMs), with intracellular growth of M. marinum being significantly increased in HexB-deficient mice, the HEXO2 homolog, compared to wild-type mice (Koo et al., 2008). It was further found that β-hexosaminidase was secreted by BMDMs in response to M. marinum, and not as a general phagocytic response, and this secretion led to extracellular killing of M. marinum at the cell surface (Koo et al., 2008). Interestingly, Tay-Sachs disease, an inherited neurodegenerative disorder, is caused by genetic loss of the α-subunit of hexosaminidase; instead of producing the αβ-heterodimer, Tay-Sachs disease patients exhibit increased production of the β-subunit (the HEXO2 homolog) (Koo et al., 2008;Okada & O'Brien, 1969;Utsumi et al., 2002).
These patients display increased resistance to tuberculosis (Koo et al., 2008;Spyropoulos, 1988). This is consistent with the findings from Koo et al. that RNAi knockdown of HEXO1, the D. melanogaster homolog of the α-subunit, had no effect on M. marinum infection of S2 cells. The authors also noted that the bactericidal effect of β-hexosaminidase was not seen against L. monocytogenes and Salmonella typhimurium, suggesting a specific mycobactericidal action, possibly due to the far slower replication rate of mycobacteria or to specific interactions with a mycobacterial factor, such as a glycopeptidolipid within the mycomembrane (Koo et al., 2008).
Drosophila melanogaster S2 cells have also contributed to understanding the modulation of phagosomes, and the role of the endosomal sorting complex required for transport (ESCRT) systems in restricting intracellular growth of mycobacteria. S2 cells were inoculated with M. smegmatis, which constitutively expressed GFP, and treated with dsRNA targeting D. melanogaster host factors in order to identify genes normally responsible for the restriction of M. smegmatis within S2 cells by measuring intracellular growth (Philips et al., 2008). The genes identified that led to an increase in the percentage of M. smegmatis infected S2 cells included Rab7, involved in vesicle maturation and a marker of late endosomes (Philips et al., 2008;Rink et al., 2005;Via et al., 1997), dVps28, a component of the D. melanogaster ESCRT I complex (Philips et al., 2008;Sevrioukov et al., 2005), and CG8055, a homolog of CHMP4B/SNF7, an ESCRT III complex protein (Philips et al., 2008). Other components of ESCRT complexes were then targeted with dsRNAs. Vps4, Vps20, Vos24, Vps25, Vps36, and dTsg101 knockdown all led to deficiencies in ESCRT function, shown by the impact on ubiquitin trafficking, and also led to increased intracellular growth of M. smegmatis. The dsRNAs that led to the most disruption to ESCRT function (those targeting Vps4, Vps28, dTsg101, and CG8055) were also found to alter the phagosome environment of the  (Koo et al., 2008;Mehra et al., 2013;Mittal et al., 2018;Philips et al., 2005). As a result of ESCRT inhibition by EsxH, M. tuberculosis was found to undermine the recognition of infected macrophages by CD4+ Tcells, potentially due to a role of ESCRT in antigen presentation (Portal-Celhay et al., 2016).  (Kimmey et al., 2015). Many studies have found that virulent mycobacteria, predominantly M. tuberculosis, actively impair autophagic function and phagosome maturation with secreted virulence factors such as ESAT6/EsxA, EsxG, and EsxH (Mehra et al., 2013;Mittal et al., 2018;Romagnoli et al., 2012). It seems increasingly likely that individual autophagy genes do not significantly contribute to protection from virulent mycobacteria, and that many of these genes have roles beyond autophagy that act to control mycobacterial infection. A notable example using a Atg7 is another autophagy factor investigated for its role in mycobacterial infection using D. melanogaster models. In vivo experiments using Atg7 mutant D. melanogaster also showed decreased survival when infected with M. marinum compared with wild-type flies, and higher bacterial loads after 3 days (Kim et al., 2012(Kim et al., , 2017.

| In vivo studies
Atg7 encodes the sole E1-like enzyme in D. melanogaster which is responsible for the activation of Atg8 (LC3) and Atg12. These are ubiquitin-like protein conjugation systems that are vital for the formation and closure of the membrane structures of autophagosomes (Juhász et al., 2007;Nagy et al., 2014). In mammals, these ATG factors play essential roles in LC3-associated phagocytosis (LAP), a noncanonical autophagy pathway linking pathogen recognition receptors and phagosomal maturation. The role of LAP in mycobacterial clearance is unclear, though a recent study using mice found that M. tuberculosis produces CpsA to inhibit recruitment of NADPH oxidase 2 (NOX2) to the phagosome and evade LAP-mediated killing (Köster et al., 2017).

Drosophila melanogaster larvae were used for in vivo infections
showing that autophagy is required for the action of several antimycobacterial drugs. Kim et al. showed that isoniazid (INH) and pyrazinamide (PZA) induced autophagy in macrophages during in vitro M. tuberculosis infection, however chloramphenicol and tetracycline did not (Kim et al., 2012). INH and PZA are both pro-drugs, converted into their active forms by the action of mycobacteria (Kim et al., 2012;Rozwarski et al., 1998;Zhang & Mitchison, 2003). Induction of autophagy by INH and PZA was shown to require the production of cellular and mitochondrial ROS by the host cell initially induced by the production of bacterial ROS following treatment with INH and PZA (Kim et al., 2012). These findings were supported by an in vivo study, which showed increased autophagy marker Atg8a in dissected fat body of larvae administered amikacin or rifampicin following M. marinum infection. GFP-tagged Atg8a also showed colocalization with lysosomes in the antibiotic-treated larvae, suggesting that treatment with these antimycobacterial drugs was promoting the degradation of Atg8a labeled vesicles (Kim et al., 2012).  (Kim et al., 2012).

This study continued by infecting adult
In addition to roles in antimycobacterial defence, there may be a relationship between autophagy-related factors and systemic pathology induced by mycobacterial infection. Autophagic mechanisms are controlled by endocrine signals that regulate systemic metabolism, in particular, the insulin signaling pathway (O'Farrell et al., 2013;Scott et al., 2004). M. marinum infection impairs systemic insulin signaling, leading to systemic metabolic pathophysiology; animals in which the insulin-inhibited transcription factor FOXO is mutated survive longer after infection and exhibit reduced metabolic pathology (Dionne et al., 2006). The loss of insulin signaling and potential changes in autophagy are almost certainly connected, though the precise nature of this connection is not yet clear.

| MYCOBAC TERIAL VIRULEN CE FAC TO R S
Drosophila melanogaster as a host model has led to many developments in understanding host defence factors important in the control of mycobacterial infection, however, its potential to study important mycobacterial virulence factors has not been exploited.

Recent developments in CRISPR interference (CRISPRi) now allows
for targeted gene silencing in mycobacteria, including species such as M. tuberculosis and M. marinum (Choudhary et al., 2015;Meijers et al., 2020;Rock et al., 2017). This system will allow for specific gene repression in these pathogenic mycobacteria, and the D. melanogaster host model will allow for broad screening of these strains to identify key virulence factors and genes essential to the intracellular survival of these mycobacteria. The genetic tractability of D. melanogaster then provides further opportunities to probe deeper into the specific host-pathogen interactions between the mycobacteria and the innate immune response.

| CON CLUDING REMARK S
Drosophila melanogaster is widely recognized as a powerful tool in studying host-pathogen interactions and has already contributed to a better understanding of how the host innate immune system responds to infection with mycobacteria. Initial studies have also made use of D. melanogaster as a model for mycobacterial infection to screen new antimycobacterial drug compounds (Oh et al., 2013a).

ACK N OWLED G M ENTS
This work was supported by a studentship from the Medical Research Council. The authors also thank members of the Dionne lab who contributed to this work with their valuable discussions.

CO N FLI C T O F I NTE R E S T
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.