Mitochondria and pathogen immunity: from killer to firestarter

Abstract

Serving as an innate defence mechanism, invading pathogens elicit a broad inflammatory response in cells. In this issue, Brokatzky et al (2019) report that pathogens can cause activation of BAX/BAK which permeabilises a limited number of mitochondria. Induction of DNA damage, or release of mtDNA, triggers STING‐dependent pro‐inflammatory cytokine expression and secretion, revealing an unexpected role for the mitochondrial apoptotic machinery in immune defence.

Diverse cellular stress signals converge on the mitochondria to induce cell death. Activation of the pro‐apoptotic BCL‐2 proteins BAX and BAK causes mitochondrial outer membrane permeabilisation (MOMP). This allows the release of mitochondrial intermembrane space proteins, such as cytochrome c, into the cytosol, whereupon they trigger the caspase protease activation and apoptosis (Campbell & Tait, 2018). Most often, widespread MOMP represents the point‐of‐no‐return for a cell since, irrespective of caspase activation, the cell will still die due to wholesale mitochondrial dysfunction. Recent work has shown that under conditions of sub‐lethal stress, a limited number of mitochondria can undergo permeabilisation, called minority MOMP. This leads to minimal caspase activation that, while not sufficient to kill the cell, is capable of causing DNA damage and promoting transformation (Ichim et al, 2015).

Paradoxically, while mitochondrial apoptosis is considered immunologically silent, the initiating process of MOMP is inherently pro‐inflammatory (Rongvaux et al, 2014; White et al, 2014; Giampazolias et al, 2017). Mitochondrial permeabilisation elicits inflammation in multiple ways. For instance, recent work shows that mtDNA is ejected from permeabilised mitochondria during cell death, activating cGAS‐STING‐dependent type I interferon production (McArthur et al, 2018; Riley et al, 2018). While dispensable for cell death, apoptotic caspases serve to dampen inflammation by targeting multiple cellular processes (McIlwain et al, 2013; Ning et al, 2019).

In this issue of The EMBO Journal, Brokatzky et al (2019) reveal an intriguing link between pathogen invasion, mitochondrial permeabilisation and immune defence. The authors hypothesised that sub‐lethal stress, sufficient to engage limited mitochondrial permeabilisation, may be pro‐inflammatory. Indeed, they found that treatment of cells with BCL‐2 targeting drugs, called BH3 mimetics, was able to stimulate the release of the pro‐inflammatory cytokines IL‐6, IL‐8, CXCL1 and FGF‐2 in the absence of cell death. Consistent with recent findings, this pro‐inflammatory phenotype was dependent on BAX/BAK and STING (McArthur et al, 2018; Riley et al, 2018). Accordingly, they found that depleting cells of mtDNA, using a mitochondrial‐targeted viral DNase, was able to reduce IL‐6 secretion following BH3‐mimetic treatment. Notably, BH3‐mimetic treatment could engage inflammation in a caspase‐proficient setting, indicating that in this setting, MOMP pro‐inflammatory signals dominate over caspase‐mediated anti‐inflammatory effects.

Next, the authors investigated whether pathogen infection could trigger the activation of the mitochondrial apoptotic machinery. A broad range of pathogens, including modified vaccinia virus Ankara (MVA), influenza virus A, the human‐pathogenic bacteria Chlamydia trachomatis and Salmonella Typhimurium, and parasite Toxoplasma gondii, all caused DNA damage without detectable caspase activation. Supporting engagement of minority MOMP, the DNA damage they observed was dependent on mitochondrial permeabilisation, since it was completely prevented by deletion of BAX/BAK or overexpression of the anti‐apoptotic protein BCL‐XL. Furthermore, CRISPR‐mediated deletion of CAD, a DNase activated following mitochondrial permeabilisation, also blocked DNA damage. Importantly, linking to earlier work—and directly relevant to pathogen control—minority MOMP triggered by all these infectious agents initiated cytokine production in a BAX/BAK‐dependent manner.

From this, the authors tested the potential physiological relevance of apoptotic mitochondrial signalling on C. trachomatis growth in cells. Bacterial growth was enhanced in cells lacking BAX/BAK, highlighting the importance for cells to produce pro‐inflammatory cytokines to limit bacterial replication and benefit the host. However, C. trachomatis growth did not appear to be affected by deletion of STING, CAD or APAF‐1—an essential adaptor molecular required for caspase activity. Intriguingly, S. Typhimurium replicated even more in APAF‐1‐deleted cells compared with BAX/BAK‐deleted cells.

This work provides a fascinating, physiological link between the pro‐inflammatory nature of mitochondrial apoptosis and the observation that a subset of mitochondria can undergo permeabilisation in response to sub‐lethal stressors (Fig 1). It also raises a number of interesting questions.

Figure 1. Limited mitochondrial permeabilisation drives inflammation in response to pathogen invasion.

Invading pathogens (e.g. protozoa, bacteria and viruses) trigger mitochondrial outer membrane permeabilisation (MOMP) and cytochrome c release in a limited number of mitochondria—minority MOMP. Infection with protozoa or bacteria triggers low levels of caspase activation, sufficient to activate CAD and induce DNA damage. Damaged DNA is able to activate STING to initiate inflammation. Alternatively, infection with viruses causes mtDNA release from permeabilised mitochondria, which is detected by cGAS initiating a STING‐dependent pro‐inflammatory response.

Firstly, how do pathogens trigger mitochondrial apoptosis? It is well understood that viruses and bacteria trigger immune‐stimulatory forms of cell death, including necroptosis and pyroptosis, but these are not thought to involve mitochondrial permeabilisation. Presumably, these pathogens can signal to activate BAX/BAK to mitochondrial membranes, and as the authors themselves propose, this may happen through pattern recognition receptors, such as cGAS‐STING and the NOD receptors. Alternatively, is there an as‐yet‐undefined mechanism for direct activation or depression of BAX/BAK during infection?

Downstream of mitochondrial permeabilisation, it is clear that different pathogens cause inflammatory cytokine secretion via different mechanisms. For example, depletion of mtDNA reduces levels of IL‐6 when infected with MVA, but not Salmonella, and deletion of CAD reduces IL‐6 secretion during Chlamydia and Salmonella infection, but not MVA. However, STING was required for IL‐6 production in response to all three agents, suggesting that STING can be activated by either mtDNA or CAD‐cleaved nuclear DNA, depending on the context. This is in line with a recent report that STING can act as a signalling hub to detect DNA damage independently of cGAS (Dunphy et al, 2018).

Secondly, why do only some mitochondria selectively permeabilise, leaving the vast majority of the mitochondrial network intact? Are some mitochondria altered in such a way that increases their propensity to undergo permeabilisation? Understanding the basis of this heterogeneity may help guide new approaches to enhance or inhibit cell death. Furthermore, previous studies have shown a link between minority MOMP and an increased propensity for cells to become transformed. As such, findings raised in this current study offer an intriguing link to observations that some infections, including Chlamydia, can give rise to cancer.

Finally, it will be exciting to see how mice that lack critical components of the mitochondrial apoptotic machinery cope with infection of these pathogens. Given that the expression of BAX/BAK decreases with age (Sarosiek et al, 2017), it is tempting to speculate that this may contribute to our decreased ability to fight infection as we get older.

The EMBO Journal (2019) 38: e102325