That the S and R forms of M. abscessus have different survival profiles in human monocytes was first reported by Byrd and Lyons (1999). While the S form survived poorly in human monocyte monolayers, the R form persisted. This result was later reproduced by independent studies, including our own (Howard et al., 2006; Nessar et al., 2011; Roux et al., 2016). Our observations of the sub-cellular events that exemplify infection with M. abscessus S or R forms (Figure 1) suggested that infection with the R form was reminiscent of an RGM infection, while that with the S form was more similar to an SGM infection (Roux et al., 2016).

To explain this dichotomy, we will first discuss what happens when the R variant encounters phagocytes. The R variant, which is highly aggregative in nature forms clumps in culture that are very hard to break using gentle techniques such as short bursts of sonication or passing the bacteria through a syringe needle (Bernut et al., 2015). It is thus very hard obtaining homogenous suspensions of the R variant and small clumps are practically impossible to avoid when infecting macrophage monolayers. As a consequence, macrophage cells engulf clumps of R bacteria resulting in large phagosomes containing numerous bacilli, quite different from the case for the S variant for which it is easier to prepare single cell suspensions (Roux et al., 2016). This effect of clumping of the M. abscessus R variant on infection of macrophages in vitro was the subject of a recent study (Brambilla et al., 2016) where phagocytosis of clumps of R bacteria resulted in rapid death of the infected J774 macrophages, while macrophages infected with the S variant stayed viable throughout the course of the infection. In our study, we observed similar events for infection of murine bone marrow-derived macrophages (BMDM). Under the transmission electron microscope, we observed large clumps of R variant bacteria lodged in phagocytic cups still on the exterior of the macrophage cells, either just prior to being phagocytosed or prohibiting their phagocytosis because of their sheer size (Roux et al., 2016). Once inside phagosomes, the R forms were rarely present as singular bacteria (loner phagosomes), but were found in groups of two or more bacilli per phagosome (social phagosomes) (Figure 1). These phagosomes contained lysosomal material probably because the close apposition between the mycobacterial cell walls and the phagosome membrane is interrupted in the social phagosome, which results in repression of the phagosome maturation block (de Chastellier et al., 2009). However, despite the presence of lysosomal material in these social phagosomes, the bacteria did not appear damaged, suggesting that they rescued themselves from the phagolysosomes (de Chastellier et al., 2009). In line with these results, R variant-containing THP1 macrophages were more acidified than those infected with the S variant and were autophagic and apoptotic, traits resembling infection with an RGM (Bohsali et al., 2010).

In contrast, the S variant behaved quite differently, where S forms were generally phagocytosed individually and also occurred most of the time in loner phagosomes (Roux et al., 2016). In these phagosomes, a tight apposition was maintained between the phagosome membrane and the bacterial cell wall all around. As a result, the phagosome maturation block was maintained and no lysosomal material was observed within these phagosomes. This result was supported by an absence of acidification observed in THP1 cells infected with the S form. Another striking feature of the S variant containing phagosomes was the presence of a large electron translucent zone (ETZ) surrounding the bacteria. The ETZ is a major part of the outer layer of the mycobacterial cell wall (Draper, 1974). On the contrary, the ETZ was barely visible in R variant containing phagosomes, as well as in phagosomes containing M. abscessus in which the gene encoding MmpL4b, a component of GPL synthesis and transport machinery, was deleted (Figure 1). In an independent study, we obtained similar results for M. bolletii, where the S variant exhibited a well-defined and large ETZ inside phagosomes, while an R variant carrying a single non-synonymous point mutation in the gene encoding MmpL4a, another determinant of GPL synthesis and transport, had almost no visible ETZ (Bernut et al., 2016). Strikingly, at later points of infection with only the S variant of M. abscessus we observed disruptions in the membranes of phagosomes containing bacteria, indicating that M. abscessus like the pathogenic SGM has the ability to induce phagosome-cytosol communications (Stamm et al., 2003; van der Wel et al., 2007; Simeone et al., 2012). Curiously, while it was shown that the ability of M. tuberculosis and M. marinum to escape the phagosome into the cytosol was strictly dependent on the presence of the type VII secretion system ESX1 (Simeone et al., 2012), M. abscessus genomes only encode ESX3 and ESX4 type VII secretion systems (Dumas et al., 2016), either of which have yet to be implicated in mycobacterial pathogenicity.

Together, these observations point to different infection programs followed by M. abscessus S and R variants. The S variant is readily phagocytosed by macrophages without severely impacting the survival of the macrophages, while the R variant, phagocytosis of which is detrimental to macrophage viability, prefers an extracellular lifestyle typified by cording as a major immune evasive mechanism whereby phagocytosis is physically impeded. We recently also showed that intracellular M. abscessus shares the ability with M. tuberculosis to use the abundant host lipid triacylglycerol (TAG) found in foamy macrophages as a rich carbon nutrient (Viljoen et al., 2016), a characteristic which is believed to contribute to the ability of M. tuberculosis to cause a latent infection (Peyron et al., 2008). Indeed, evidence exist that M. abscessus, like M. tuberculosis, can cause asymptomatic infection lasting for years before a full-fledged acute infection emerges (Moore and Frerichs, 1953; Cullen et al., 2000). One could, therefore, speculate that the S variant may act as a reservoir of live bacilli during this latent period of infection, and once mutations in the GPL locus occur allowing the emergence of R bacilli, a much more aggressive lifestyle is adopted by the bacteria, characterized by acute disease. Indeed, clinical evidence points out toward the S variant being the invasive form probably causing initial infection, while the R form which later emerges causes more severe forms of the disease (Catherinot et al., 2009).

In two recent reports, we detail our discovery of novel unrelated non-toxic chemotypes that efficiently inhibit both extracellular and intracellular M. abscessus populations through inhibition of highly essential mycolic acid transport (Figure 1). Mycolic acids are extremely large fatty acids consisting of a long β-hydroxy fatty acid chain (C₆₀₋₉₀) with a shorter α-alkyl branch (C₂₄₋₂₆) and lend to the mycobacterial cell wall its renowned hydrophobicity and impermeability to extraneous compounds like antibiotics. These essential fatty acids are biosynthesized for the larger part within the mycobacterial cytosol through the joint actions of fatty acid synthase I (FasI), the FasII complex, an acyl-AMP ligase and a polyketide synthase, the last step resulting in the transacylation of trehalose with the α-alkyl β-ketoacyl mycolic acid to produce trehalose monomycolate (TMM) (Quémard, 2016). After an acetylation step of TMM, the molecule is transferred to the periplasmic space via the essential mycolic acid transporter, MmpL3 (Yamaryo-Botte et al., 2015; Xu et al., 2017). Once inside the periplasm, TMM is probably deacetylated by an unidentified enzyme. In the mycomembrane, the antigen 85 enzyme complex uses TMM as substrate to transfer the mycolic acids onto arabinogalactan as well as to produce trehalose dimycolate (TDM), also known as cord factor (Belisle et al., 1997).

A surprisingly large number of recent chemical hits against M. tuberculosis have been assigned to target MmpL3 activity leading to the view that this protein could represent the Achilles' heel of mycobacteria (Nataraj et al., 2015). We also serendipitously identified a novel piperidinol derivative potently inhibiting MmpL3 in M. abscessus (compound 1, Figure 1), when we performed a cross screen of a library of compounds with known anti-M. tuberculosis activity against M. abscessus (Dupont et al., 2016). More recently, we evaluated a structurally unrelated chemotype class, the indole-2-carboxamides (exemplified by compounds 2 and 3, Figure 1) previously shown as potent anti-tubercular compounds (Lun et al., 2013; Onajole et al., 2013; Stec et al., 2016), for their activity against M. abscessus and found that they too potently inhibited mycolic acid transport (Kozikowski et al., 2017). Sequencing analysis of the mmpL3 gene in spontaneous resistant mutants selected on either of the compound classes revealed a common Ala309Pro substitution in transmembrane domain 5. Over-expression of MmpL3 carrying the Ala309Pro mutation in M. abscessus wild-type bacteria conferred high level resistance to both the piperidinol-based compound and indole-2-carboxamides, confirming their target. Modeling of the three-dimensional structure of MmpL3 revealed a large cavity formed by transmembrane helices 5, 7, 8, 9, and 10 where the piperidinol-based compound could potentially bind (Dupont et al., 2016). Importantly, both the piperidinol-based and indole-2-carboxamide compounds showed good activity against M. abscessus both extracellularly (Table 1) and intracellularly in macrophages and improved the survival of zebrafish embryos infected with the R strain in the case of the piperidinol-based compound (Dupont et al., 2016). Not only were these compounds found to be non-toxic toward the HepG2 human cell line (compound 1) (Ballell et al., 2013) and Vero cells (compounds 2 and 3) (Kozikowski et al., 2017), but they were also active at lower concentrations than the drugs that are currently being used to treat M. abscessus in the clinical setting (Table 1). In addition, both chemotypes were equally efficient against the S and R variants of a wide panel of clinical strains of all three M. abscessus subspecies isolated from CF and non-CF patients. The fact that these non-toxic compounds worked efficiently inside and outside macrophages, highlights their potential for development into a new class of antibiotics active against M. abscessus. From a medicinal chemistry perspective and for future drug development, these compounds are easy to prepare and early studies on the indole-2-carboxamides indicated that they show reasonably good ADME properties despite their high lipophilicity (Onajole et al., 2013; Stec et al., 2016; Kozikowski et al., 2017).

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.