Mycobacterium tuberculosis induces an atypical cell death mode to escape from infected macrophages

Abstract

Background: Macrophage cell death following infection with Mycobacterium tuberculosis plays a central role in tuberculosis disease pathogenesis. Certain attenuated strains induce extrinsic apoptosis of infected macrophages but virulent strains of M. tuberculosis suppress this host response. We previously reported that virulent M. tuberculosis induces cell death when bacillary load exceeds ∼20 per macrophage but the precise nature of this demise has not been defined.

Methodology/principal findings: We analyzed the characteristics of cell death in primary murine macrophages challenged with virulent or attenuated M. tuberculosis complex strains. We report that high intracellular bacillary burden causes rapid and primarily necrotic death via lysosomal permeabilization, releasing hydrolases that promote Bax/Bak-independent mitochondrial damage and necrosis. Cell death was independent of cathepsins B or L and notable for ultrastructural evidence of damage to lipid bilayers throughout host cells with depletion of several host phospholipid species. These events require viable bacteria that can respond to intracellular cues via the PhoPR sensor kinase system but are independent of the ESX1 system.

Conclusions/significance: Cell death caused by virulent M. tuberculosis is distinct from classical apoptosis, pyroptosis or pyronecrosis. Mycobacterial genes essential for cytotoxicity are regulated by the PhoPR two-component system. This atypical death mode provides a mechanism for viable bacilli to exit host macrophages for spreading infection and the eventual transition to extracellular persistence that characterizes advanced pulmonary tuberculosis.

Figure 1. Mtb induces a morphologically unique form of macrophage cell death.

(A) TEM images of uninfected macrophages (a) and macrophages infected with Mtb (MOI 25) for 3 h (b and c). Typical early changes (b) were peripheral nuclear condensation and partial nuclear protrusion from condensed areas (arrow [b]) while cells at the late phase of death had ruptured plasma membranes (arrow [c]) and advanced nuclear pyknosis. Micrographs are representative of three independent experiments. Bar  = 500 nm. (B) DAPI staining shows nuclear pyknosis and extrusion of chromosomal DNA (arrow) in a macrophage infected with Mtb. Bar  = 10 µm. (C) SEM images of uninfected (a) and Mtb-infected macrophages (b,c) showing extensive plasma membrane loss in the latter. The boxed area in b is shown with 3X zoom in c. Bar  = 10 µm.

Figure 2. Bax/Bak-independent mitochondrial injury in Mtb-infected macrophages.

(A) Macrophages infected with M. bovis BCG (MOI 25, 2 h and 3 h) were stained with anti-cytochrome c mAb and DAPI for fluorescence microscopy (X400). Micrographs are representative of two independent experiments. (B) Loss of ΔΨ_m in macrophages infected for 3 h with BCG or Mtb measured by flow cytometry with Mitotracker CMXRos. Histograms are representative of three independent experiments. (C) Macrophages were infected with Mtb (3 h) in the presence of absence of CsA (5 µM) and probed with Mitotracker dye CMXRos to measure ΔΨ_m. Histograms are representative of three independent experiments. (D) Macrophages lacking Bax and Bak were infected with Mtb and probed with Mitotracker Deep Red to measure ΔΨ_m. Histograms are representative of three independent experiments.

Figure 3. Mtb infection induces LMP independent of the ESX1 system.

(A) Cytosolic cathepsin B activity in macrophages were challenged with viable or heat-killed (HK) Mtb for 2 h and then cytosolic extracts were prepared for measurement of cathepsin B activity indicative of LMP. Results are expressed as the fold increase of cathepsin B activity relative to uninfected (control) cells, normalized by the fold increase in LDH compared to uninfected cells (*P<0.05; error bars, ± SD). (B) Cytosolic cathepsin B activity in macrophages challenged with RvΔRD1ΔespA or the parental strain H37Rv (*P<0.05). (C) Macrophages were challenged with RvΔRD1ΔespA or H37Rv (3 h) and probed with Mitotracker DeepRed developed mitochondrial injury (ΔΨ_m dissipation) comparable to parental H37Rv. Histograms are representative of three independent experiments. (D) RvΔRD1ΔespA induces macrophage cell death (PI positivity) comparable to H37Rv (18 h; *P<0.05 compared to uninfected cells).

Figure 4. Mtb cytotoxicity does not depend on cathepsins B or L.

(A) Macrophages pretreated with BafA1 or buffer control where challenged with Mtb (18 h) and then cell death was quantified by PI staining and microscopy (*P<0.05). (B) Cell death 18 h after Mtb challenge was measured by PI staining in wildtype macrophages treated with control siRNA (empty bar), wildtype macrophages treated with cathepsin B siRNA (striped bar), cathepsin L null macrophages treated with control siRNA (speckled bar) and cathepsin L null macrophages treated with cathepsin B siRNA (filled bar).

Figure 5. Mtb does not kill macrophages by pyroptosis.

(A) Caspase-1 null (Casp 1 KO) and wildtype (WT) macrophages on the same NOD genetic background were infected with Mtb (18 h). Cell death was measured by PI incorporation (*P<0.05). (B) C57BL/6 macrophages were infected with S. typhimurium (Stm) or Mtb, both at MOI 25 for 4 h. Cell death was measured by neutral red dye uptake (*P<0.05 compared with uninfected or Mtb-infected macrophages. (C) TEM image of S. typhimurium-infected macrophage (MOI 25, 2 h) showing evidence of osmotic lysis with preserved nuclear morphology. Bar  = 2 µm.

Figure 6. Increased lipase activity in Mtb-infected macrophages.

(A) Macrophages loaded with liposomal PED-6 were challenged with Mtb (3 h) and then examined by fluorescence microscopy. Bar  = 20 µm. (B) Macrophages loaded with liposomal PED-6 were cultured in control medium, infected with Mtb, or treated with 1 µmol PMA. Cell lysates were prepared and fluorescence intensity measured by fluorometry was calculated as fold increase relative to uninfected cells and normalized for autofluorescence. (*p<0.05; error bars ± SD). (C) Macrophages were infected with Mtb or maintained in control medium for 6 h and then lysates were prepared for HPLC/MS analysis of total lipid content. Data are expressed as fold change of infected cells compared to uninfected cells. DG, diacylglycerol; MG, monoacylglycerol; PI, phosphatidylinositol; SM, sphingomyelin; PE, phosphatidylethanolamine; PC, phosphatidylcholine; PS, phosphatidylserine.

Figure 7. Lysosomal lipases participate in Mtb-induced cytolysis.

(A) Macrophages were infected with Mtb at the indicated MOI for 150 min and then cytosolic pH was measured by using the dual emission fluorescent dye SNARF-1 (*P<0.05). (B) Macrophages loaded with liposomal PED-6 were pretreated with control buffer, CPZ or orlistat and then infected with Mtb. Fluorescence intensity in cell lysates was calculated as fold increase relative to uninfected cells and normalized for autofluorescence (*P<0.05). (C) Macrophages with or without CPZ pretreatment were infected with Mtb (3 h) and probed with Mitotracker Deep Red. Histograms are representative of three independent experiments. (D) Macrophages were treated with CPZ, orlistat or control buffer prior to Mtb infection (18 h). Cell death was measured by PI incorporation (*P<0.05).

Figure 8. The PhoPR two component system is required for LMP, lipase activation, mitochondrial injury and death of Mtb-infected macrophages.

(A) Macrophages were challenged with RvΔphoPR or parental H37Rv and LMP was measured by cytosolic translocation of cathepsin B 2 h later. Results are expressed as mean fold increase of cytosolic cathepsin B activity ± SEM. * P<0.05. (B) Mitochondrial injury (ΔΨ_m dissipation) was measured in macrophages challenged with H37Rv as compared to RvΔRD1ΔespA (left panel) or RvΔphoPR (right panel) using Mitotracker Deep Red. Histograms are representative of three independent experiments. (C) Cell death assessed by PI incorporation was measured in uninfected macrophages and macrophages challenged with H37Rv, RvΔRD1ΔespA, or RvΔphoPR (18 h). A representative experiment of three performed is depicted. * P<0.05 comparing cells infected with H37Rv or RvΔRD1ΔespA to uninfected cells or cells infected with RvΔphoPR. Error bars, ± SD.