We previously described caspase-independent macrophage cell death induced by Mtb when the intracellular load exceeds ~20 bacilli 
. In that report, infected macrophages were shown to exhibit apoptotic features of nuclear condensation and PS translocation but rapidly progressed to necrosis. In the present study we have further elucidated the mode and mechanism of Mtb-induced cell death. Infected macrophages show an atypical combination of apoptotic and necrotic ultrastructural features notable for widespread membrane damage. LMP is an early event that is followed by mitochondrial outer and then inner membrane permeabilization which is independent of Bax and Bak. Mitochondrial injury in the absence of these Bcl-2 family members indicates that Mtb-induced cell death does not operate through the classical intrinsic apoptosis pathway. Dissipation of ΔΨm, depletion of intracellular ATP and a drop in cytosolic pH are followed by widespread loss of membrane integrity observed by electron microscopy. This is accompanied by catabolism of several phospholipid species that are integral constituents of lipid bilayers. Factors required for pyroptosis and pyronecrosis (cathepsin B and caspase-1, respectively) are dispensable for Mtb cytotoxicity. The constellation of findings shown here and in our previous report is unique among infection-induced cell death modes.
The early appearance of LMP in the chain of events resulting in macrophage cell death, and the loss of cytotoxicity in bacilli incapable of causing LMP, implicate involvement of a lysosomal death pathway. Lysosomal cell death, which results from relocation of hydrolases into the cytosol, has been linked to a wide variety of pathological conditions including certain infections but not Mtb 
. Cells dying by this process may exhibit primarily apoptotic or primarily necrotic characteristics depending on the nature and extent of injury to the limiting membrane of the lysosome 
. The biochemical pathways mediating lysosomal death vary considerably in different cell types and with conditions of induction, including examples that are caspase-dependent or caspase-independent 
. Lysosomal cell death has been most frequently attributed to cytosolic translocation of cathepsins, particularly cathepsins B, L, S and D. We previously reported partial protection from Mtb cytotoxicity by chemical inhibitors of cathepsins B and L, but not cathepsin D 
. However, the effects were modest and limited by direct cytotoxicity of these agents. Subsequent experiments revealed that these agents also partially inhibit phagocytosis of Mtb (Fig. S10
) which would promote macrophage survival by decreasing intracellular bacillary load at a given MOI. Our current studies with BafA1, cathepsin B silencing and cathepsin L null macrophages () provide additional evidence that Mtb-induced cell death does not depend on cathepsins. Instead, the ultrastructural evidence of widespread membrane damage, the degradation of multiple phospholipid species and the evidence of Bax/Bak independent MOMP and subsequent loss of ΔΨm
are all consistent with membrane attack mediated by lysosomal lipases, possibly with a contribution of mycobacterial lipases and/or membrane-destabilizing detergent-like molecules. Bioenergetic collapse in this setting could be lethal but cell death appears inevitable regardless of mitochondrial function, given the catastrophic disruption of nuclear and plasma membranes. This model of Mtb-induced macrophage cell death is supported by the appearance of widespread lipase activity in infected macrophages revealed using a liposomal PED-6 probe and by the protective effects of CPZ which acts downstream of LMP. It is also consistent with previously reported effects of lysosomal lipases on mitochondrial membrane integrity 
. Our model of membrane injury may explain with the observations of Divangahi et al. 
that the fate of Mtb-infected macrophages depends in part on their capacity for membrane repair.
In the conditions of our experiments, only viable Mtb are capable of killing macrophages. This indicates that one or more bacterial gene products are essential for the LMP that results in cell death. Evidence supporting this conclusion includes the capacity of macrophages to survive with a high intracellular load of M. smegmatis
as we reported 
or M. leprae
as reported by others 
. While the ESX1 secretion system and ESAT-6 have been implicated in membrane injury and LMP, and are clearly associated with Mtb virulence, we found that they are dispensable for macrophage cytolysis at high MOI. Having eliminated a requirement for the ESX1 system we next tested RvΔphoPR
based on the in vivo attenuation of this strain 
, the role of the PhoPR two-component system in altering bacterial gene expression upon phagocytosis, and the recent report that fibroblasts permit bacillary replication and survive a high intracellular burden of the PhoPR-deficient strain SO2 
. We found that that the RvΔphoPR
mutant was significantly impaired in its ability to induce LMP, mitochondrial injury and macrophage cell death after challenge at MOI 25. The attenuation of cytotoxicity could not be attributed to any failure of bacilli to enter macrophages. Indeed, the intracellular burden of RvΔphoPR
was greater than that of the parental strain H37Rv after challenge at similar MOI. These results indicate that one or more essential mediator(s) of proximal events in macrophage cytotoxicity lie within the PhoPR regulon. Which Mtb gene(s) are responsible for promoting LMP and cytolysis from among the 54 so far known to be regulated by PhoPR remains to be determined. It is worth noting that 7 genes of this regulon are predicted to be involved in lipid metabolism 
. While the preponderance of data suggest that host lysosomal lipases are the executioners of Mtb-infected cells, our studies do not exclude a role for mycobacterial lipases or other bacterial products such as lipid carrier proteins or lipids with detergent properties that could disrupt membranes to cause LMP and/or facilitate lipolytic membrane attack.
We delivered attenuated mycobacteria (BCG, RvΔRD1ΔespA
) in higher intracellular numbers than could be achieved by their limited capacity for replication in macrophages after low MOI challenge. Our in vivo data (Fig. S2
) unequivocally support the relevance of high bacillary load to pulmonary TB disease. High MOI challenge in vitro makes it possible to distinguish between reduced cytopathicity based on the loss of microbial mediators of host cell death versus reduced capacity for bacteria introduced at low MOI to grow to a number capable of killing the host cell. Our experiments model events that occur when macrophages engulf a small number of virulent bacilli that subsequently kill the host cell after a period of intracellular replication. This concept was directly tested in our published MOI/cell death dose-response study 
and in detailed experiments by Park et al. 
where macrophages were challenged at MOI 5 with several Mtb strains differing in virulence. The most virulent strains grew rapidly over 6 days and killed their host cells after reaching an intracellular burden similar to the cytoxic threshold we found with direct high MOI challenge. Park et al. also showed that pretreatment with IFN-γ suppressed Mtb replication and preserved macrophage viability at day 6 post-infection. This confirmed that cytotoxicity depends on bacterial burden and is not an intrinsic property of certain virulent strains unrelated to intracellular bacillary load.
Lung macrophages provide an environment required by the facultative intracellular pathogen Mtb to establish infection in a new host 
. Successive rounds of intracellular bacillary replication followed by host cell lysis, invasion of naïve macrophages and resumption of intracellular replication are essential steps for the exponential increase in lung bacillary burden following aerosol infection. This cycle is interrupted by the expression of adaptive immunity. Once IFN-γ is present in sufficient quantity in the lung it activates macrophages to limit Mtb replication and thereby survive with a sublethal load of persistent bacilli. The mode and rate of infection-induced macrophage cell death may be determining factors in the pathogenesis TB. The known anti-apoptotic functions of virulent Mtb could result in a more indolent disease like leprosy without a means for bacilli to eventually exit the macrophage. The death mode that we describe in macrophages infected with virulent Mtb has several features consistent with its proposed role in TB pathogenesis: 1, it is induced only after intracellular bacilli have accumulated to a high number; 2, bacilli are released to the extracellular space without confinement in apoptotic vesicles; 3, macrophage cell death is not accompanied by a significant loss of bacillary viability; 4, the pro-inflammatory nature of macrophage necrosis might aid in the recruitment of naïve phagocytes to host subsequent rounds of infection; and 5, macrophage death and unchecked lipolysis might help shape a nutrient milieu in necrotic TB lesions supporting extracellular bacterial survival and disease transmission.