reSeArcH Article
response to osmotic changes: matrix contraction leads to expanded
cristae and matrix swelling causes the collapse of the intracristae com-
partment. Genetic ablation of MITOK caused both a widening of the
intracristae space (Extended Data Fig. 8c) and a lower oligomerization
of OPA1 (an additional biochemical readout for cristae remodelling, for
which higher multimerization correlates with tighter cristae^32 ). Even
in these cases, valinomycin partially recovered normal morphology of
the cristae (Extended Data Fig. 8c, d).
We then considered how the mitoKATP channel affects organelle
adaptations to energy stress. We treated wild-type and MITOK-
knockout cells with the glycolysis inhibitor 2-deoxyglucose, which
rapidly decreases global cellular metabolism (Extended Data Fig. 8e, f).
First, we tested how the mitochondrial morphology changes in
response to ATP depletion. Wild-type HeLa cells rapidly underwent
fragmentation of the mitochondrial network (Fig. 4f). By contrast, met-
abolic inhibition in MITOK-knockout cells caused no evident change
of the overall mitochondrial morphology (Fig. 4f), which indicates that
mitochondrial morphology adapts promptly to the energetic state of
the cells through a mitoKATP-dependent mechanism. Then, we moni-
tored the production of reactive oxygen species (ROS) during metabolic
stress and/or pharmacological modulation of the mitoKATP channel.
Extended Data Figure 8g indicates that (i) loss of MITOK increases
ROS production, notwithstanding the decreased oxygen consumption
rate (which provides further support for the idea that this represents
latent mitochondrial dysfunction); (ii) diazoxide increases ROS in
wild-type but not in MITOK-knockout cells (which supports the idea
of the mitoKATP channel as a regulator of redox state); (iii) metabolic
stress can increase ROS production in control cells; and (iv) ROS levels
marginally increase when mitoKATP is absent, thus indicating that mito-
chondria K+ homeostasis impinges on the regulation of redox balance
during metabolic stress. Overall, our data indicate that the mitoKATP
channel regulates mitochondrial adaptations to cellular stress, possibly
through the regulation of matrix volume (Fig. 4g). In addition, as pre-
viously suggested^33 , the loss of MITOK increases cell death triggered
by oxidative stress (Extended Data Fig. 8h), which is consistent with
cristae widening^34.MITOK is required for pharmacological preconditioning
Finally, we generated Mitok-knockout mice through the specific deletion
of exon 4, which contains most of the coding sequence. Overall, these
mice show no overt phenotype (being born at the expected Mendelian
ratio, with a similar aspect and weight gain) until at least four months
of age. To demonstrate the lack of mitoKATP activity, we measured
organelle K+ fluxes using^86 Rb+ as surrogate^18. Energized mitochon-
dria isolated from wild-type livers showed ATP- and diazoxide-
sensitive K+ uptake (Fig. 5a). By contrast, neither ATP nor diazoxide
were able to alter K+ fluxes when MITOK was absent (Fig. 5b, c).
Finally, we performed ex vivo ischaemia–reperfusion experiments in
wild-type and Mitok-knockout mice and evaluated the cardioprotec-
tive effect triggered by pharmacological preconditioning induced by
diazoxide. As shown in Fig. 5d, the hearts of untreated Mitok-knockout
mice are slightly more sensitive to the ischaemia–reperfusion protocol,
which provides further confirmation of the cytoprotective role of
MITOK. As previously shown^10 ,^12 ,^35 ,^36 , pharmacological precondi-
tioning with diazoxide efficiently protects the heart from reperfu-
sion damage. Most importantly, the effects of this pharmacologicaleHeLa WTHeLa MITOKKO no. 1HeLa MITOKKO no. 2Control +mitoKATPMITOK MITOKMITOK MITOKK+MITOSUR MITOSURMITOSUR MITOSURAAAATPPAAAATPP AAAATPPAAAAATPPHigh ATPmitoKATP closedLow ATPmitoKATP openedRegulation of mitochondrial volumeMITOK MITOKMITOK MITOKMITOSUR MITOSURMITOSUR MITOSURATPATPATPATPgArea (%)020406080(^100) HeLa WT HeLa MITOK KO no. 1
HeLa MITOK KO no. 2
t = 0 t = 15 t = 60
Elongated
Intermediate
Fragmented
Area (%)
0
20
40
60
80
100
Area (%)
0
20
40
60
80
100
t = 0 t = 15 t = 60
t = 0 t = 15 t = 60
0 min 15 min 60 min
WT
KO no. 1
KO no. 2
f
a
Absorbance (at 520 nm)
0.19
0.20
0.21
0.22
0.23
0.24
No ATP
2 mM ATP
2 mM ATP +
50 50 μμM diazoxideM diazoxide
30 s
b c
Swelling rate
0
0.2
0.4
0.6
0.8
1.0
1.2
1.4 HeLa WT
HeLa MITOK KO no. 1
HeLa MITOK KO no. 2 HeLa WT
HeLa MITOK KO no. 1
HeLa MITOK KO no. 2
No ATP ATP ATP +
diazoxide
t=250 sect=250 sec
0
100
200
300
400
OCR (pmol min
–1
)
HeLa MITOK KO no. 1
HeLa WT
HeLa MITOK KO no. 2
10 min
Oligo
FCCP Ant A
Rot
d
mitoKATP–+ –+ –+
Ψ
m
ashes
0
1
2
3
4
5
Δ
Fig. 4 | Loss of MITOK impairs mitochondrial structure and function.
a, Swelling traces of wild-type mitochondria. Three independent
experiments with similar results. b, Mitochondrial swelling rates in
K+-based medium. n = 2 independent experiments, P ≤ 0.009 using
two-way ANOVA with Holm–Sidak correction. KO, knockout; WT,
wild type. c, Quantification of ΔΨm flashes (number of depolarizations per
cell per ten minutes). n > 10 independent experiments, P ≤ 0.001 using
two-way ANOVA with Holm–Sidak correction. d, Oxygen consumption
rate (OCR) measurements. n = 5 biological replicates, representative of
3 independent experiments. Ant A, antimycin A; FCCP, carbonyl
cyanide-4-(trifluoromethoxy)phenylhydrazone; oligo, oligomycin; rot,
rotenone. e, Transmission electron microscopy images of mitochondrial
ultrastructure, representative of two independent preparations. f, Analysis
of mitochondrial morphology during energy stress. Box plots indicate the
percentage of organelle area occupied by elongated (cyan), intermediate
(grey) or fragmented (magenta) mitochondria. Scale bars, 10 μm. n ≥ 22
individual cells from 3 independent experiments, *P < 0.01 using one-way
ANOVA with Holm–Sidak correction. t, time in minutes. g, Schematic of
mitoKATP channels.
612 | NAtUre | VOl 572 | 29 AUGUSt 2019