M. tuberculosis-Induced Necrosis of Infected Neutrophils Promotes Bacterial Growth Following Phagocytosis by Macrophages
With 10.4 million people worldwide suffering from and 1.8 million being killed by tuberculosis, Mycobacterium tuberculosis infection is the most common cause of death by infectious disease worldwide (World Health Organization, 2016). Increasing numbers of multidrug-resistant tuberculosis cases call for innovative host-directed therapies supplementing anti-microbial treatment. A massive influx of inflammatory cells into the lung and subsequent tissue destruction characterize active tuberculosis in humans (Ehlers and Schaible, 2012). Neutrophils (PMNs) represent the predominant mycobacteria-carrying cell population in bronchoalveolar lavage and sputum samples from patients with active tuberculosis (Eum et al., 2010). The role of PMNs in tuberculosis, i.e. protection or disease progression, is still controversial (Dorhoi and Kaufmann, 2015, Korbel et al., 2008, Dallenga and Schaible, 2016). We found that human PMNs fail to kill virulent mycobacteria due to M. tuberculosis infection-induced production of reactive oxygen species (ROS), which quickly send PMNs into necrotic cell death, thereby facilitating mycobacterial escape from elimination (Corleis et al., 2012). In contrast, an M. tuberculosis mutant lacking the virulence-associated genomic region of difference 1 (ΔRD1) encoding for virulence factors such as the early secreted antigenic target 6 (ESAT-6 also known as EsxA) (Young, 2003, Hsu et al., 2003) failed to drive PMN necrosis and was susceptible to killing by PMNs. Therefore, escape from killing by PMNs through necrosis induction requires virulence factors encoded on RD1 (Corleis et al., 2012). Massive accumulation of infected and dying PMNs, as observed in active tuberculosis lesions, requires efficient removal to protect the surrounding tissue from bioactive PMN molecules and subsequent pathological sequelae. Impaired dead cell removal leads to chronic inflammation, even under sterile conditions, and hinders resolution of inflammation (Mikolajczyk et al., 2014, Peng and Elkon, 2011, Fernandez-Boyanapalli et al., 2010). Therefore, dead cell removal is essential for resolving immunopathology. In tuberculosis, removal of apoptotic cells, termed efferocytosis, is envisaged as beneficial for host defense. Engulfment of M. tuberculosis-infected, apoptotic macrophages by non-infected ones has been shown to promote subsequent killing of the mycobacteria (Martin et al., 2012). Our observation that the type of PMN cell death, i.e., apoptosis versus necrosis, is determined by M. tuberculosis virulence, prompted us to analyze the fate of mycobacteria-infected PMNs in macrophages. Of note, in contrast to efferocytosis of apoptotic cells, little is known about necrotic cell removal during M. tuberculosis infection.
Here, we report that removal of necrotic PMN debris with virulent M. tuberculosis, but not apoptotic PMNs infected with attenuated ones, promoted mycobacterial survival and proliferation in human monocyte-derived macrophages (MDMs). Thus, necrotic cell death is a prerequisite for mycobacterial escape from growth control by efferocytes. Necrotic cell death required ESAT-6/EsxA and PMN-derived ROS. Our data demonstrate that removal of necrotic PMNs, as induced by virulent M. tuberculosis, is an ESAT-6/EsxA-dependent prerequisite for pathogen growth in macrophages. Consequently, inhibition of PMN necrosis identified ROS generation by PMNs as putative target for host-directed therapy.
▼Removal of Infected PMNs Promotes Survival of M. tuberculosis
Human PMNs fail to kill wild-type (WT) M. tuberculosis due to mycobacteria-induced and ROS-mediated necrotic cell death. In contrast, after infection with attenuated ΔRD1 M. tuberculosis, PMNs undergo apoptosis similar to non-infected ones (Corleis et al., 2012). We hypothesize that the switch point, setting the course for PMN host cell death either toward necrosis or apoptosis, determines subsequent fate of mycobacteria in macrophages. Therefore, human PMNs were infected with either WT or ΔRD1 M. tuberculosis (MOI = 1) leading to ∼85% infected PMNs. Subsequently, we co-cultured infected PMNs with uninfected human MDMs and analyzed mycobacterial growth. To narrow down the RD1-encoded effector responsible for PMN necrosis, we included an M. tuberculosis strain lacking the RD1-encoded ESAT-6/EsxA gene. In parallel, we infected MDMs directly with WT M. tuberculosis and the respective mutants. WT M. tuberculosis as well as an ESAT-6/EsxA-reconstituted ΔESAT-6 strain were able to proliferate in directly infected MDMs that engulfed bacteria in the absence of PMNs (Figures 1A and 1B ). Confirming earlier studies by others, MDMs controlled the ΔRD1 mutant (Lewis et al., 2003) as well as the ΔESAT-6 mutant, although to a lesser degree and with variable outcomes. However, over a longer observation time of 7 days, MDMs controlled both mutants equally (Figure S1A). In co-cultures of infected PMNs and non-infected MDMs, WT M. tuberculosis, as well as an ESAT-6/EsxA-reconstituted ΔESAT-6 strain, were able to proliferate, whereas both the ΔRD1 and the ΔESAT-6 mutant failed to grow (Figures 1C and 1D). PMNs alone failed to reduce numbers of WT, ΔESAT-6, and the complemented M. tuberculosis strain, but slightly diminished numbers of ΔRD1 bacteria (Figures S1B and S1C; Corleis et al., 2012).
Taken together, we confirmed our previous results and those of others that PMNs and MDMs were able to control growth of the ΔRD1 and ΔESAT-6 mutants, but failed to control WT M. tuberculosis. Co-cultures of infected PMNs and non-infected MDMs controlled both the ΔRD1 and the ΔESAT-6 mutant. Still, WT M. tuberculosis maintained growth under these conditions. Therefore, a functional ESAT-6/EsxA secretion system (ESX-1) is crucial for intracellular growth of M. tuberculosis in the presence of both PMNs and MDMs.
▼Functional ESX-1 Is Essential for M. tuberculosis-Induced PMN Necrosis
Our results suggest that the type of PMN cell death, i.e. necrosis versus apoptosis, determines subsequent fate of M. tuberculosis in macrophages. Therefore, we analyzed the cell death of PMNs infected with WT, ΔRD1, or ΔESAT-6 M. tuberculosis. The majority of PMNs infected with WT M. tuberculosis appeared as Hemalaun-positive indefinable debris 16 hr after infection (Figure 2A). In contrast, ΔRD1 and ΔESAT-6 M. tuberculosis-infected PMNs (Figures 2B and 2C, arrows) remained mostly intact similar to uninfected ones (Figure 2D). Higher magnifications by light as well as electron microscopy revealed signs of necrosis in WT-infected PMNs, i.e., plasma membrane disintegration, nuclei dissolution, cell lysis, and cell ghosts 16 hr after infection (Figures 2E and 3A ). Due to PMN lysis, WT M. tuberculosis bacteria (Figures 2E, arrows; 3B, asterisks) were found extracellularly and only loosely associated with necrotic PMN material (Figure 2E, arrowheads), representing phagosomal membrane remnants (Figure 3B, arrowheads), granule content (Figure 3B, arrows), and granules (Figure 3B, “G”). In contrast, PMNs infected with ΔRD1 or ΔESAT-6 M. tuberculosis succumbed to apoptosis, as indicated by plasma membrane blebbing (Figures 2F and 2G, arrowheads) and chromatin condensation (Corleis et al., 2012). ΔRD1 and ΔESAT-6 M. tuberculosis (Figures 3D and 3F, asterisks) were observed enclosed in apoptotic bodies surrounded by intact plasma membrane and defined cell compartments such as granules (Figures 3D and 3F, “G”). Necrotic events were rarely observed 16 hr after ΔESAT-6 infection as indicated by disintegrated plasma membrane and cell ghosts (Figure 2G, arrowheads). Uninfected PMNs showed signs of apoptosis (Figures 2H and 3G, arrows).
We monitored kinetics of infection-triggered PMN cell death by time-lapse live-cell imaging. At 5 hr post infection (p.i.) with WT M. tuberculosis, PMNs showed leakage of intracellular material indicating necrotic cell death (Movie S1, arrows). These processes were not observed in ΔRD1- or ΔESAT-6-infected or uninfected PMNs. Instead, membrane blebbing and formation of membrane vesicles, indicating apoptosis, were observed (Movies S2, S3, and S4). Notably, at 12 hr p.i., most WT mycobacteria appeared extracellularly or loosely associated with PMN remnants, whereas ΔRD1 and ΔESAT-6 mutants remained enwrapped in PMNs.
Quantification of PMN necrosis by lactate dehydrogenase (LDH) activity in PMN supernatants revealed significantly higher values in PMN cultures 6 hr p.i. with WT M. tuberculosis than ΔRD1- and ΔESAT-6-infected or uninfected ones (Figure 3H). A ΔESAT-6 mutant complemented with the ESTA-6 gene induced similar levels of LDH activity in PMN cultures as the WT strain, further corroborating a functional ESX-1 as the essential factor for induction of PMN necrosis by M. tuberculosis (Figure S1D).
Taken together, our findings extend our previous results showing differential induction of PMN necrosis by WT and ΔRD1 M. tuberculosis, respectively (Corleis et al., 2012). We identified a functional ESX-1 as the prime RD1-encoded mechanism for PMN necrosis, with the loss of ESAT-6/EsxA being sufficient to impair this type VII secretion system (T7SS) function.
▼Differential Intracellular Trafficking of PMN-Shuttled M. tuberculosis Is Determined by ESX-1
Interference with phagosome maturation is a hallmark of mycobacterial virulence. M. tuberculosis associates with early phagosomes, but rarely with late endosomes/lysosomes (Weiss and Schaible, 2015). To reveal reasons why removal of infected PMNs by MDMs fails to limit growth of WT M. tuberculosis, we used LAMP-1 and Rab7 to monitor mycobacterial fate in MDMs after direct uptake or after passage through PMNs.
Mycobacteria and PMN material were found in common, as well as separated compartments in MDMs. WT M. tuberculosis associated with PMNs were less frequently found in LAMP-1- or Rab7-positive phagosomes, when compared with ΔRD1 or ΔESAT-6 mutants (Figures 4A and 4B ). At 24 hr p.i. we also found PMN-associated WT mycobacteria in LAMP-1-positive phagosomes, similar to ΔRD1 or ΔESAT-6 mutants. However, at 24 hr p.i., the majority of WT mycobacteria were separated from PMN material and those were not associated with LAMP-1. Numbers of Rab7-positive compartments containing both, M. tuberculosis and PMN material were highest for WT, lowest for ΔRD1, and in between for ΔESAT-6. In the absence of PMNs, association with LAMP-1- and Rab7 was similar for ΔESAT-6 and WT M. tuberculosis. Notably, frequencies of LAMP-1- as well as Rab7-positive phagosomes containing both, ΔESAT-6 and PMN material were similar to ΔRD1. These results demonstrate that interference by M. tuberculosis with phagosome biogenesis upon uptake of infected PMNs requires a functional ESX-1. However, without PMN shuttle, WT and ΔESAT-6 mycobacteria were less often associated with LAMP-1 or Rab7 than ΔRD1 ones mainly located in LAMP-1- and Rab7-positive phagosomes. Importantly, WT M. tuberculosis taken up alone or together with necrotic PMNs were hardly associated with LAMP-1 or Rab7.
Thus, intracellular WT M. tuberculosis, after necrosis induction in PMNs and removal by MDMs were devoid of phagosomal maturation markers. Early after infection, ΔESAT-6 and ΔRD1 mycobacteria, co-delivered with PMNs, primarily resided in mature phagosomes. In contrast, upon direct uptake of free bacteria, ESAT-6/EsxA was not required to interfere with transport to phagolysosomes. Thus, a functional ESX-1 is important for PMN necrosis and subsequent interference with trafficking to mature phagosomes in MDMs.
▼WT M. tuberculosis Taken up after PMN Necrosis Escapes Macrophage Defense Effectors
Phagolysosome fusion puts a halt on mycobacterial growth by acidification and amplifying anti-microbial effectors such as ROS by NADPH oxidase (Nox2) and cathelicidin (CAP18) (Liu et al., 2006). ΔRD1 mycobacteria with or without PMNs were more often observed in Nox2-and CAP18-positive compartments when compared with WT and ΔESAT-6 ones (Figures 4C and 4D). WT M. tuberculosis was significantly less often associated with Nox2 and CAP18 under all conditions. Importantly, upon co-uptake with PMNs, ΔESAT-6 M. tuberculosis was less often found in Nox2-positive phagosomes than ΔRD1 mutants, but still more often than WT ones. However, 24 hr after direct infection, frequencies of Nox2-associated ΔESAT-6 mycobacteria were similar to WT ones. This demonstrates that ESX-1-dependent necrosis induction of PMNs contributes to interference with NADPH oxidase integration into the phagosomal membrane.
These data show that intracellular WT M. tuberculosis, after uptake via necrotic PMNs, did not associate with the microbicidal effectors NOX2 and CAP18. In contrast, uptake of ΔRD1 mutants enwrapped in apoptotic PMN targeted mycobacteria to bactericidal compartments, a prerequisite for elimination of bacteria. Importantly, enhanced association of ΔESAT-6 with Nox2 when compared with WT mycobacteria shows that a functional ESX-1 and previous PMN necrosis is required to limit exposure to ROS.
▼Only Partial Association of WT M. tuberculosis with PMN Debris in Early MDM Phagosomes
To monitor the intracellular kinetic of mycobacteria engulfed by macrophages after previous PMN infection, we used live-cell imaging for MDMs that were infected by mycobacteria via PMNs. The majority of engulfed WT mycobacteria appeared to be separated from PMN debris (Movie S5) as indicated by absence of overlapping fluorescent signals. In contrast, ΔRD1 and ΔESAT-6 M. tuberculosis mainly co-localized with PMN material (Movies S6 and S7, arrow). An apoptotic PMN associated with ΔRD1 M. tuberculosis was taken up by a macrophage (Movie S6, arrowhead), which led to disruption of the apoptotic PMNs into multiple small remnants. PMN material and the mycobacterium resided in separate compartments. Similarly, an uninfected apoptotic PMN taken up by an MDM ended up in multiple smaller efferosomes (Movie S7, center bottom, arrowhead). Two MDMs shared one apoptotic PMN infected with ΔESAT-6 M. tuberculosis (Movie S7, arrow).
We gained further insights into differential localization of M. tuberculosis and PMN remnants in MDMs by ultrastructure analysis using transmission electron microscopy. To identify phagolysosome fusion events, macrophages were pulsed with 5 nm gold particles for 2 hr followed by a 2 hr chase prior to addition of infected PMNs. WT-infected PMNs within phagosomes showed necrotic appearance with irregular debris and ill-defined electron-dense and -light structures (Figures 5A, 5B , and S3A–S3C). Necrotic PMN debris (#) was engulfed together with electron-dense organelles, most likely PMN granules (G), surrounded by loose phagosome membranes. WT mycobacteria (asterisks) resided in tight phagosomes often as single bacteria (Figure 5A, arrowheads) while occasionally associated with necrotic PMN material (Figure 5A, arrows). Magnification revealed membrane structures resembling the PMN phagosomal membrane (Figure 5B, arrowheads) partly associated with bacteria (asterisks). Absence of gold particles indicates that WT M. tuberculosis and PMN material failed to fuse with lysosomes. In contrast, apoptotic ΔRD1- and ΔESAT-6-infected PMNs within efferosomes had distinctively structured, membrane-enclosed cell organelle remnants (Figures 5C–5F, S3D, and S3F, #). Gold particles within efferosome (circles) demonstrate efferosome-lysosome fusion. Figure 5G depicts a macrophage that efferocytosed a remnant of an apoptotic PMN (#) without mycobacteria, containing a disintegrated nucleus (N). PMN organelles were still preserved and distinguishable. In contrast to necrotic PMN debris of infected cells, apoptotic vesicles from uninfected PMNs resided in tight efferosomes (Figure 5H, arrows) with preserved PMN plasma membrane (black arrowheads) despite phagolysosome fusion, as indicated by the presence of gold particles (circles). Similar results were found when infected PMNs and uninfected MDMs were co-cultured from 2 hr p.i. (Figure S4). After 2 hr of co-culture, MDMs neither took up nor tethered to infected PMNs regardless of the mycobacterial strain used (data not shown). After 6 hr of co-culture, few necrotic WT M. tuberculosis-infected PMNs were observed, which were phagocytosed by MDMs (Figures S4A and S4B), while ΔRD1 and ΔESAT-6-infected PMNs were not observed (Figures S4C and S4D). At these time points, PMNs still appeared healthy and most likely not exposing any find-me or eat-me signals. Indeed, these findings corroborate our LDH activity measurement data, which showed healthy PMNs at 2 hr p.i. and, for ΔRD1- and ΔESAT-6-infected ones, at 6 hr p.i. Semi-quantitative analysis of live-cell imaging data revealed that MDMs started to tether to ΔRD1 and ΔESAT-6 M. tuberculosis-infected PMNs at around 12 hr p.i. on average. Only 15% of the WT M. tuberculosis-infected PMNs succumbed to necrosis at 6 hr p.i., which was also reflected in the electron microscopy (EM) data. Importantly, at 16 hr p.i. and co-culturing, cells looked similar to those in Figures 5 and S3,, with WT M. tuberculosis establishing close MDM phagosomal membrane contact and ΔRD1 and ΔESAT-6 M. tuberculosis being enwrapped in apoptotic PMN material within MDM efferosomes (Figures S4E–S4J).
To confirm the EM results, we combined 3D confocal laser scanning microscopy with EM for analysis of the very same MDMs by using a dish coordinate system (Figure S5). Thereby, we confirmed the distinct morphological features of phagosomes depending on whether they contained PMN material infected with WT, ΔESAT-6, or ΔRD1 mycobacteria, as well as sequestrated compartments containing either PMN remnants or mycobacteria (Figures 4 and 5).
Taken together, WT M. tuberculosis were mainly localized within tight early-stage MDM phagosomes as free bacteria only occasionally associated with necrotic PMN remnants. In contrast, efferosomes containing ΔRD1 and ΔESAT-6 M. tuberculosis were associated with or enwrapped in apoptotic PMN remnants, showing well-preserved organelles and membranes, and readily fused with lysosomes.
▼Necrotic Cell Death of PMNs after WT M. tuberculosis Infection Is ROS Dependent
To assess whether PMN infection with virulent M. tuberculosis, associated with specific necrotic processes, subsequently determines the fate of M. tuberculosis in MDMs, we used specific inhibitors for different types of necrotic cell death: ROS-dependent cell death (myeloperoxidase [MPO] inhibitor ABAH), RIP1K-dependent necroptosis (necrostatin-1), iron-dependent ferroptosis (ferrostatin), and PI3K-dependent autophagy (3-methyladenine). Inhibition of MPO-derived ROS resulted in a significant decrease in LDH release, as a correlate of necrosis, compared with untreated, infected PMNs (Figure 6A). In contrast, inhibiting necroptosis or ferroptosis reduced LDH activity only slightly (Figures 6B and 6C). Inhibition of autophagy (Figure 6D), apoptosis (by pan-caspase inhibitor zVAD) and other key molecules of the necroptosis pathway (RIP3K, MLKL) did not alter necrotic PMN cell death upon WT M. tuberculosis infection (data not shown). Inhibitor combinations did not further reduce LDH release, excluding a synergistic effect between different necrosis pathways (data not shown). Of note, inhibitors did not directly affect mycobacterial viability. Thus, necrotic PMN cell death upon WT M. tuberculosis infection is ROS dependent and does not represent programmed necrosis.
▼Necrotic Cell Death of PMNs as Prerequisite for Mycobacterial Growth in Macrophages Requires Functional ESX-1
Growth of WT but not ΔESAT-6 M. tuberculosis in MDMs, when engulfed together with PMNs (see Figure 1B), prompted us to study the role of necrotic PMN cell death for mycobacterial growth in MDMs. Macrophages were either co-cultured with infected PMNs or directly infected with WT, ΔRD1, or ΔESAT-6 M. tuberculosis. LDH activity in MDM supernatants increased strongly between days 1 and 2 after direct infection with WT M. tuberculosis (Figure 7A). In contrast, the ΔRD1 mutant caused only a modest LDH release comparable with those levels found in uninfected MDM cultures. The ΔESAT-6 mutant induced less necrosis in MDMs than WT M. tuberculosis, although still more than the ΔRD1 mutant suggesting a partial role for ESX-1 in MDM necrosis. In co-cultures of infected PMNs and uninfected MDMs, LDH activity increased over 2 days, but was significantly higher in WT M. tuberculosis compared with ΔRD1- or ΔESAT-6 infected co-cultures (Figure 7B). In general, LDH release was higher in co-cultures than in MDM mono-cultures most likely due to the presence of PMNs succumbing to secondary necrosis under these conditions. Importantly, in contrast to directly infected MDMs, LDH levels in supernatants of co-cultures infected with ΔESAT-6 or ΔRD1 mutants were comparable. Similar observations were obtained by flow cytometric analysis of annexin-V- and live/dead cell stain-labeled cells (data not shown). Of note, the ESAT-6/EsxA-complemented ΔESAT-6 strain caused necrosis at similarly high levels as WT M. tuberculosis, both in MDMs alone as well as in co-cultures of infected PMNs and uninfected MDMs (Figures S1E and S1F). Differential growth of WT versus ΔRD1 and ΔESAT-6 strains correlated with LDH concentrations triggered by these strains (compare Figure 1 with Figures 7A and 7B).
Based on these results, we asked whether reduction of necrotic cell death in WT M. tuberculosis-infected PMNs can influence subsequent mycobacterial growth in MDMs. Since we identified MPO as a target for prevention of WT M. tuberculosis-induced PMN necrosis (see Figure 6A), ABAH was used to inhibit PMN necrotic cell death. ABAH reduced mycobacterial loads in PMNs slightly and, importantly, only at 16 hr p.i. (Figure 7C). For co-cultures, ABAH was removed from PMNs at 6 hr p.i., before uninfected MDMs were added. At this time point, ABAH had no effect on mycobacterial numbers in PMNs (Figure 7C). Ultimately, inhibition of PMN necrotic cell death armed MDMs to control growth of WT M. tuberculosis (Figure 7D).
Taken together, these findings corroborated the essentiality of a functional ESX-1 T7SS for induction of host cell necrosis as the prime outcome of WT M. tuberculosis infection in both MDMs and PMNs. Thus, ESAT-6/EsxA is a key factor for a functional ESX-1, which represents a prerequisite for ROS-mediated necrosis of PMNs. Importantly, PMN necrosis is indispensable for mycobacterial growth upon subsequent removal by macrophages. Therefore, PMN necrosis represents an intriguing target for host-directed therapy. Indeed, in our setting we were able to restore growth control of M. tuberculosis by macrophages by preventing previous PMN necrosis.
Here, we demonstrate that M. tuberculosis escapes from anti-bacterial effectors and thrives in macrophages that phagocytosed infected PMNs, which was dependent on previous PMN necrosis. However, mutants deficient in a functional ESX-1 T7SS failed to induce PMN necrosis and were unable to grow in macrophages upon uptake enwrapped in PMNs. More importantly, subsequent mycobacterial growth was prevented by inhibiting necrotic PMN death. Thus, neutrophils represent an intriguing target to restore growth control of M. tuberculosis by host-directed therapy.
PMNs are central for inflammatory host responses to M. tuberculosis during active disease and are associated with exacerbated pathogenesis (Dallenga and Schaible, 2016). They represent the predominant cell population and carry the main mycobacterial load in bronchoalveolar lavages and sputa from patients with active tuberculosis (Eum et al., 2010). Neutrophilic transcriptome signatures have been associated with active tuberculosis and were suggested as biomarkers for poor immune control of M. tuberculosis (Repasy et al., 2015, Berry et al., 2010). This notion is further corroborated by mouse models, which are highly susceptible to M. tuberculosis, such as the C3HeB/FeJ and I/St strains, in which PMNs contribute to pulmonary pathology (Yeremeev et al., 2015, Harper et al., 2012). Reducing PMN numbers in these mice either by Ibuprofen or anti-PMN antibody treatment ameliorated inflammation, lowered mycobacterial numbers (Vilaplana et al., 2013), and enhanced T cell responses (Yeremeev et al., 2015). Increased PMN numbers upon experimental tuberculosis in C57BL/6 mice with atg5-deficient myeloid cells exacerbated pathology and disease progression when compared with WT mice (Kimmey et al., 2015). Furthermore, anti-tumor necrosis factor alpha treatment of M. tuberculosis-infected guinea pigs increased PMN to macrophage ratios and, subsequently, mycobacterial loads in these animals (Ly and McMurray, 2009). Taken together, PMNs failed to control M. tuberculosis despite their highly efficient bactericidal armamentarium against different infectious agents. This is in line with our observations that human PMNs were unable to control M. tuberculosis in vitro. Instead, PMNs quickly succumbed to necrotic cell death. Notably, attenuated mycobacteria, such as the ΔRD1 mutant and the vaccine strain M. bovis BCG, failed to induce necrosis in PMNs (Corleis et al., 2012). The RD1 region encodes for certain virulence-associated gene products, including the ESAT-6/EsxA-CFP10 complex and its respective secretion apparatus components ESX1-5 (Houben et al., 2014, Parkash et al., 2009). Here we show that a mutant lacking only ESAT-6/EsxA was much less efficient in driving PMNs into necrosis compared with WT M. tuberculosis. This result indicates (1) a central function of ESAT-6/EsxA for a functional ESX-1 secretion system and (2) a crucial role for ESX-1 in PMN necrotic cell death. Other RD1-encoded virulence factors that act in concert with a functional ESX-1 (Fortune et al., 2005, Augenstreich et al., 2017, Lou et al., 2017) may also contribute to necrosis of PMNs, as indicated by the total failure of RD1-deficient M. tuberculosis to induce necrosis. After uptake of infected PMNs by macrophages, ESAT-6/EsxA appears to be a central factor in ESX-1-mediated intracellular mycobacterial survival, growth, and induction of macrophage cell necrosis. Upon efferocytosis of apoptotic PMNs infected with ΔRD1 or ΔESAT-6 mutants, mycobacteria ended up in a double- or triple-membrane compartment representing a potentially non-surmountable barrier for mycobacteria lacking a functional ESX-1 secretion system. Disruption of the PMN phagosomal and plasma membrane before, during, or after efferocytosis likely requires virulence factors and mechanisms exerted by ESX-1 to interfere with or lyse host membranes. Although membranolytic properties of ESAT-6/EsxA have been solely made responsible for disruption of phagosomal membranes and escape into macrophage cytoplasm (van der Wel et al., 2007, Aguilo et al., 2013, Jonge et al., 2007, De Leon et al., 2012), a recent study showed that recombinant ESAT-6/EsxA has no intrinsic membrane-lytic activities (Conrad et al., 2017). Instead, all previously described lytic properties of rESAT-6/EsxA are attributed to the detergent ASB-14, which was used in those preparations. However, the same publication describes that a single point mutation in the ESAT-6/EsxA gene suffices to abrogate ESX-1-mediated cell lysis and compromise M. tuberculosis virulence as shown previously by others (Frigui et al., 2008, Brodin et al., 2005, Brodin et al., 2006). Importantly, one of the above-mentioned studies found that host cell lysis is mediated through M. tuberculosis contact-dependent membrane disruption (Conrad et al., 2017). Host cell necrosis frees M. tuberculosis, leaving the bacteria only loosely associated with host cell material, which more likely facilitates direct contact with phagosomal membranes of the subsequent host cell. In contrast, upon efferocytosis of apoptotic PMNs infected with attenuated mycobacteria, multiple membrane wrappings likely prohibit close contact between the mycobacterium and the phagosomal membrane. In conclusion, ESX-1-driven PMN necrosis is a prerequisite for early direct contact between M. tuberculosis and the phagosomal membrane within macrophages. This promotes mycobacterial escape from lysosomal effectors, intracellular growth, and induction of necrosis (Welin et al., 2011, Simeone et al., 2012, Mishra et al., 2010, Pathak et al., 2007, van der Wel et al., 2007, Simeone et al., 2015). The underlying mechanism of how M. tuberculosis escapes from phagolysosomes, either by arresting phagosomal maturation or escape into the host cell cytosol, remains to be elucidated. Our data demonstrate that WT M. tuberculosis, engulfed after PMN necrosis by MDMs, is less frequently associated with markers for late endosomal/lysosomal compartments, whereas ESAT-6/EsxA- or RD1-deficient ones engulfed in association with apoptotic cells predominantly end up in phagolysosomes. This is interesting, since necrotic PMNs most likely contain danger-associated molecular patterns (DAMPs), which potentially activate macrophages, whereas apoptotic cells keep these signals enwrapped to avoid phagocyte activation. Activation of murine macrophages has been shown to promote maturation of mycobacterial phagosomes (Axelrod et al., 2008, Weiss and Schaible, 2015). Notably, WT M. tuberculosis did not associate with phagosomes positive for the NADPH oxidase subunit Nox2, indicating mycobacterial escape from anti-microbial effectors. Despite our observation that WT M. tuberculosis cells were decorated with PMN material, co-uptake with PMN debris, including CAP18, failed to mark virulent WT M. tuberculosis for killing by macrophages, but promoted mycobacterial growth in these cells. Concomitantly delivered apoptotic PMN material, including granule-associated effectors such as anti-microbial peptides, however, may contribute to growth inhibition of ΔESAT-6 and ΔRD1 mutants in macrophages.
Host cell necrosis and immune cell apoptosis at the sites of mycobacterial infection result in accumulation of necrotic and apoptotic materials. Their removal displays a prerequisite for infection control and tissue regeneration (Headland and Norling, 2015, Szondy et al., 2014, Martin et al., 2012). Efferocytosis of apoptotic cells is a highly regulated event. In contrast, necrotic cells release heterogeneous and rather undefined cellular material into the extracellular space. Despite recent identification of “find-me” and “eat-me” signals and DAMPs (Zelenay et al., 2012, Peter et al., 2010, Yamasaki et al., 2008), removal of necrotic cells in the context of infection has not yet been studied. (As the term efferocytosis has been used to describe removal of apoptotic cells, we suggest necrophorocytosis (from the Greek necrophoros meaning “undertaker”) for phagocytosis of necrotic cell material.) Here, we demonstrate that M. tuberculosis-infected necrotic PMNs were efficiently phagocytosed by macrophages. M. tuberculosis was able to grow inside these macrophages and, more importantly, subsequently also promoted necrosis of these cells. Among others (Hartman and Kornfeld, 2011, Lee et al., 2006), Martin et al. (2012)) described efferocytosis as an anti-mycobacterial effector mechanism. They found that efferocytosis of apoptotic, M. tuberculosis-infected macrophages by non-infected ones resulted in mycobacterial killing. This notion is corroborated by our observation that macrophages controlled growth of WT M. tuberculosis after efferocytosis of infected apoptotic PMNs, which were treated with the MPO inhibitor to block necrosis. The scenario, in which efferocytosis of infected apoptotic macrophages results in a beneficial outcome for the host, is in stark contrast to the scenario in which PMNs undergo necrosis and mycobacteria enter macrophages as extracellular bacteria, only partially associated with PMN debris. The latter situation is advantageous for the pathogen and sets off a vicious cycle of cellular necrosis between phagocytes such as PMNs and macrophages. Such a situation represents very likely the conditions in active human tuberculosis lesions. We hypothesize that this circle of necrotic events leads to pathological sequelae in patients with active tuberculosis as characterized by tissue damage and caseous necrotic debris. This lung damage leads to contagious aerosols coughed up for subsequent infection of the next host (Dallenga and Schaible, 2016). We propose that interruption of this vicious circle by blocking necrosis represents an anchor point for host-directed therapy. PMN necrosis, as a result of M. tuberculosis infection, was strongly dependent on ROS. NETosis has also been described to be dependent on MPO-generated ROS (Bjornsdottir et al., 2015). Moreover, PMNs from chronic granulomatous disease patients who have deficits in ROS production fail to undergo NETosis (Romao et al., 2015), and ROS-dependent NETosis occurs after infection with M. tuberculosis (Ramos-Kichik et al., 2009). Occurrence of DAPI-positive, extracellular DNA, and morphology changes during PMN cell death (see Movie S1) suggest that the type of necrosis caused by M. tuberculosis is similar to NETosis. On the contrary, we did not observe NET-like structures stainable by Sytox Green (data not shown). Instead, the whole M. tuberculosis-infected PMNs rapidly became positive for Sytox Green indicating plasma membrane rupture. In addition, preliminary data indicate that M. tuberculosis-induced PMN necrosis is accompanied by mitochondrial damage (our unpublished data). ROS and mitochondrial damage have been linked to cell death (Corda et al., 2001, Choi et al., 2009, Goossens et al., 1999) and NETosis (Dan Dunn et al., 2015).
In summary, an ESAT-6/EsxA-dependent functional ESX-1 is the prime requisite for M. tuberculosis to escape from concerted anti-bacterial effectors of PMNs and macrophages. ROS-mediated PMN necrosis triggered by M. tuberculosis is essential to survive subsequent removal by macrophages, and to replicate and induce necrosis in these successive host cells. Importantly, inhibition of MPO blocked necrosis of infected PMNs and, ultimately, restored control of mycobacterial growth by macrophages that had removed infected PMNs. Therefore, M. tuberculosis-induced, ROS-mediated PMN necrosis represents a potential target for host-directed therapy to improve host defense and support anti-mycobacterial treatment.