and inflammation ( 79 ). A well-characterized path-
way that selectively removes damaged mitochon-
dria involves the ubiquitin kinase PINK1 and the
ubiquitin ligase Parkin (Fig. 3). Like the tran-
scription factor ATFS-1—which is rescued from
degradation by mitochondrial stress, leading to
thelossofmembranepotential—PINK1 is nor-
mally imported into mitochondria, clipped by the
protease PARL embedded in the inner mito-
chondrial membrane, and degraded by the N-end
rule for proteosomal proteolysis ( 80 ). Damage
to mitochondria that prevents protein import
prevents PINK1 degradation, allowing it to ac-
cumulate bound to the outer mitochondrial mem-
brane ( 81 ). There, it phosphorylates ubiquitin that
has been previously attached to outer mitochon-
drial membrane proteins that face the cytosol,
which serve as receptors for cytosolic Parkin.
Parkin binding to phospho-ubiquitin on mitochon-
dria activates Parkin ubiquitin ligase activity, yield-
ing more ubiquitin chains on outer-membrane
proteins for PINK1 to phosphorylate. This yields
a powerful amplification loop ( 80 ). The ubiquitin
chains on mitochondria recruit autophagy recep-
tors such as NDP52 that can directly recruit the
Ulk1 kinase complex to induce autophagosome
biogenesis ( 82 ). Importantly, the loss of either
PINK1 or Parkin induces early onset Parkinson’s
disease. The fact that damaged mitochondria are
selectively degraded suggests that mitochondrial
quality control may be important for protecting
dopaminergic neurons that are lost in Parkinson’s
disease. This model is compatible with a variety
of other evidence that mitochondrial defects are
involved in Parkinson’sdisease( 83 ).
Mice that lack either PINK1 or Parkin do not
display parkinsonism-related phenotypes unless
stressed from mtDNA mutation accumulation
( 84 ). Under such conditions, mice display an in-
crease in serum amounts of mtDNA and the
cytokines IL-6 and IFNbmediated by the cGAS
STING DNA–sensing pathway ( 85 ). Because evi-
dence is accumulating that inflammation may
lead to Parkinson’sdisease( 86 ), a key finding is
that blocking the inflammation in Parkin-null mice
through loss of STING activity prevents neuro-
degeneration ( 85 ). These results lead to the model
that PINK1/Parkin–mediated mitophagy is im-
portant to prevent mtDNA from activating the
innate immunity that can lead to neuron loss. Be-
cause cGAMP can transfer between cells ( 87 – 89 ),
it will be important to understand what cell types
release the mtDNA and cGAS activation and which
cell types express STING and secrete the cytokines,
ultimately causing neuron death. After the in-
flammatory stress of chronic exposure to lipo-
polysaccharide, a classical PAMP, mice in which
Parkin was deleted display neurodegeneration
( 90 ). Mice in which Parkin or PINK1 was deleted
also display an increased immune response to a
mitochondrial antigen ( 91 ). It is thought that Parkin-
mediated autophagy mitigates self-reactivity, and
interestingly bacterial infection of mice in which
PINK1 was deleted leads to impaired dopaminer-
gicneuronfunction( 92 ).
STING’s primordial function in metazoans
appears to induce autophagy and not interferon
production ( 93 ). Autophagy that leads to lyso-
somal destruction of bacteria can be considered
another branch of innate immunity and shares
some common molecular mechanisms with mito-
phagy, such as using the autophagy receptors
Optineurin and NDP52 and the kinase Tbk1 ( 94 ).
Although Tbk1 is also required for STING acti-
vation of interferon regulatory factor–3(IRF3)to
drive IFNbexpression, STING appeared in meta-
zoan evolution as an autophagy inducer before
acquiring the Tbk1 interaction domain. cGAS also
appeared with the earliest metazoans, such as anem-
ones ( 58 ). However, the DNA-binding domain
of cGAS was not apparent until vertebrates, when
IFN defense pathways appeared during evolution.
It remains unclear what may activate insect and
more primitive cGAS homologs. Similarly, OAS
appeared in sponges, whereas RNase L did not
appear until much later in vertebrate evolution
(in birds and mammals) ( 95 ). PINK1 and Parkin
appeared somewhat later in evolution than cGAS
and STING, not appearing in sponges or anem-
ones but in nematodes, insects, and vertebrates.
BothC. elegansandDrosophila,lackingParkin,
display increased expression of innate immune
genes ( 73 , 96 , 97 ), supporting the proposed link
between mitophagy and DAMP removal. Thus,
this mitophagy pathway may be a relatively late
evolutionarily patch to mitigate innate immune
signaling by mtDNA.Mitochondrial induced
apoptosis—Another innate
immune mechanism
Mitochondrial cytochrome c, the Ox/Phos elec-
tron carrier located between the outer and inner
membranes, could be considered the ultimate
DAMP. It is released into the cytosol by Bax and
Bak to induce apoptosis. Once in the cytosol, cyto-
chrome c binds to monomeric apoptotic pepti-
dase activating factor 1 (APAF-1), triggering it
to form the heptameric apoptosome, a complex
similar to the NLRP3 inflammasome (Fig. 4B).
The assembled apoptosome recruits caspase 9,
where induced proximity of this protease effects
self-cleavage and proteolytic activation and sub-
sequent activation of caspase 3 and apoptosis.
This is similar to the related NLRP3 inflamma-
some activation caspase 1,whichprocessesIL-1bto
promote inflammation. Many viruses express pro-
teins that block apoptosis ( 98 ), indicating that
apoptosis is a form of innate immunity. Cyto-
chrome c is conserved in bacteria, and apoptosis
may have evolved from an innate immune bacte-
rial sensor to induce cell death after mitochon-
drial damage. However, cytochrome c activation
of APAF-1 does not induce inflammation but
prevents it through elimination of mitochondrial
DAMPs and other cell debris to avoid inflammation
through the sterile death process of apoptosis ( 99 ).Future perspectives
Mitochondria are central to our energy supply
and metabolism. Understanding the particular
challenges for mitochondria that stem from their
endosymbiont origin is advancing rapidly and
includes how gene expression is coordinating be-tween two genomes, how the integrity of the
mitochondrial genome is maintained, and how
inflammation caused by loss of mitochondrial
integrity is avoided. How some of these chal-
lenges are mitigated in yeast is becoming clear,
but how these pathways and others work in the
variety of different cell types in humans—in
vivo, with very different energy and metabolic
demands—remains to be elucidated. Disruptions
to mitochondrial homeostasis caused by excessive
damage or insufficient repair lead to a wide array
of human diseases, many with muscle and/or
neuron phenotypes, and understanding how fail-
ures in mitochondrial maintenance lead to human
diseases is in its infancy. There are undoubtedly
more medical consequences to the risks asso-
ciated with endosymbiosis than are currently
appreciated. mtDNA mutations and inflammation
increase during normal human aging. Whether
mitochondrial dysfunction contributes to inflam-
maging would be important to know. The rapid
progress on these topics in the past few years will
continue to unveil how the delicate balance be-
tween host and endosymbiont is maintained.REFERENCES AND NOTES- J. D. Woodson, J. Chory, Coordination of gene expression
between organellar and nuclear genomes.Nat. Rev. Genet. 9 ,
383 – 395 (2008). doi:10.1038/nrg2348; pmid: 18368053 - R. S. Isaac, E. McShane, L. S. Churchman, The multiple levels
of mitonuclear coregulation.Annu. Rev. Genet. 52 , 511– 533
(2018). doi:10.1146/annurev-genet-120417-031709;
pmid: 30230928 - A. P. West, G. S. Shadel, Mitochondrial DNA in innate immune
responses and inflammatory pathology.Nat. Rev. Immunol. 17 ,
363 – 375 (2017). doi:10.1038/nri.2017.21; pmid: 28393922 - S. E. Weinberg, L. A. Sena, N. S. Chandel, Mitochondria in the
regulation of innate and adaptive immunity.Immunity 42 ,
406 – 417 (2015). doi:10.1016/j.immuni.2015.02.002;
pmid: 25786173 - J. Nunnari, A. Suomalainen, Mitochondria: In sickness and in
health.Cell 148 ,1145–1159 (2012). doi:10.1016/
j.cell.2012.02.035; pmid: 22424226 - M. S. Sharpleyet al., Heteroplasmy of mouse mtDNA is
genetically unstable and results in altered behavior and
cognition.Cell 151 , 333–343 (2012). doi:10.1016/
j.cell.2012.09.004; pmid: 23063123 - J. B. Stewartet al., Strong purifying selection in transmission
of mammalian mitochondrial DNA.PLOS Biol. 6 , e10 (2008).
doi:10.1371/journal.pbio.0060010; pmid: 18232733 - V. I. Floroset al., Segregation of mitochondrial DNA
heteroplasmy through a developmental genetic bottleneck in
human embryos.Nat. Cell Biol. 20 ,144–151 (2018).
doi:10.1038/s41556-017-0017-8; pmid: 29335530 - T. Lieber, S. P. Jeedigunta, J. M. Palozzi, R. Lehmann,
T. R. Hurd, Mitochondrial fragmentation drives selective
removal of deleterious mtDNA in the germline.Nature 570 ,
380 – 384 (2019). doi: 10 .1038/s41586-019-1213-4;
pmid: 31092924 - O. G. Berg, C. G. Kurland, Why mitochondrial genes are most
often found in nuclei.Mol. Biol. Evol. 17 , 951–961 (2000).
doi:10.1093/oxfordjournals.molbev.a026376; pmid: 10833202 - K. L. Adams, J. D. Palmer, Evolution of mitochondrial gene
content: Gene loss and transfer to the nucleus.
Mol. Phylogenet. Evol. 29 , 380–395 (2003).
doi:10.1016/S1055-7903(03)00194-5; pmid: 14615181 - D. U. Mick, T. D. Fox, P. Rehling, Inventory control: Cytochrome
c oxidase assembly regulates mitochondrial translation.
Nat. Rev. Mol. Cell Biol. 12 ,14–20 (2011). doi:10.1038/
nrm3029; pmid: 21179059 - M. T. Couvillion, I. C. Soto, G. Shipkovenska, L. S. Churchman,
Synchronized mitochondrial and cytosolic translation
programs.Nature 533 , 499–503 (2016). doi:10.1038/
nature18015; pmid: 27225121 - R. Richter-Dennerleinet al., Mitochondrial protein synthesis
adapts to influx of nuclear-encoded protein.Cell 167 , 471–483.
e10 (2016). doi:10.1016/j.cell.2016.09.003; pmid: 27693358
Youle,Science 365 , eaaw9855 (2019) 16 August 2019 5of7
RESEARCH | REVIEW