Mitochondrial Diseases in
Man and Mouse
Douglas C. Wallace
Over the past 10 years, mitochondrial defects have been implicated in a
wide variety of degenerative diseases, aging, and cancer. Studies on
patients with these diseases have revealed much about the complexities of
mitochondrial genetics, which involves an interplay between mutations in
the mitochondrial and nuclear genomes. However, the pathophysiology of
mitochondrial diseases has remained perplexing. The essential role of
mitochondrial oxidative phosphorylation in cellular energy production, the
generation of reactive oxygen species, and the initiation of apoptosis has
suggested a number of novel mechanisms for mitochondrial pathology.
The importance and interrelationship of these functions are now being
studied in mouse models of mitochondrial disease.
Mitochondrial defects occur in a wide variety of
degenerative diseases, aging, and cancer. One
of the first indications that mitochondria may
play a role in pathogenesis was the report nearly
40 years ago of a patient with hypermetabolism
whose skeletal muscle contained large numbers
of abnormal mitochondria, a condition now
known as mitochondrial myopathy (Fig. 1) (1).
It is now clear that mitochondrial diseases en-
compass an extraordinary assemblage of clini-
cal problems, commonly involving tissues that
have high energy requirements such as heart,
muscle, and the renal and endocrine systems.
The genetic and molecular complexities of
these diseases, which typically display a bewil-
dering array of inheritance patterns, have been
studied intensively over the past decade (2–5).
The first mitochondrial diseases to be un-
derstood at the molecular level were the ma-
ternally inherited Leber’s hereditary optic
neuropathy (LHON), a sudden-onset blind-
ness resulting from a mitochondrial DNA
(mtDNA) missense mutation (6), and a spon-
taneously occurring group of neuromuscular
diseases, now classified as chronic progres-
sive external ophthalmopelia (CPEO) and
the Kearns-Sayre Syndrome (KSS), which
result from mtDNA deletions (7 ) (Fig. 2).
In the 10 years since these discoveries, over
50 pathogenic mtDNA base substitution
mutations and hundreds of mtDNA rear-
rangement mutations (deletions and inser-
tions) have been identified in a variety of
degenerative diseases (8). Other mitochon-
drial diseases have been linked to nuclear
genes whose inactivation either inhibits mi-
tochondrial bioenergetics or disrupts mito-
chondrial or mtDNA biogenesis.
Here I review our current understanding
of the genetics of mitochondrial disease, a
knowledge base derived primarily from pa-
tient studies, and then describe new mouse
models that are providing important insights
into pathophysiology.
Mitochondrial Genetics
Mitochondria generate cellular energy in the
form of ATP (adenosine triphosphate) by the
process of oxidative phosphorylation (OX-
PHOS). Most cells contain hundreds of mito-
chondria. These cytoplasmic organelles are
thought to have arisen about 1.5 billion years
ago from a symbiotic association between a
glyolytic proto-eukaryotic cell and an oxidative
bacterium. Modern mitochondria retain a num-
ber of features that reflect their endosymbiotic
origin. These include a double membrane struc-
ture and a circular mitochondrial genome with
mitochondria-specific transcription, translation,
and protein assembly systems. However, the
mitochondrion has also adapted to its intracel-
lular niche. To increase its replication rate, and
thus ensure transmission to the two daughter
cells at cytokinesis, the mammalian mitochon-
drial genome has been reduced in size to about
16,500 base pairs (9 –11). Presumably, this was
accomplished by the deletion of nonessential
genes and the transfer of many essential genes
to the nucleus where the proteins are now tran-
scribed into mRNAs, translated on cytoplasmic
ribosomes, and selectively imported back into
the mitochondrion (12).
The modern mammalian mtDNA retains
only 13 polypeptide genes, all of which encode
essential components of OXPHOS. It also en-
codes the 12S and 16S rRNA genes and the 22
tRNA genes required for mitochondrial protein
synthesis (Fig. 2). The remaining mitochondrial
OXPHOS proteins, the metabolic enzymes, the
DNA and RNA polymerases, the ribosomal
proteins, and the mtDNA regulatory factors,
such as mitochondrial transcription factor A
The author is at the Center for Molecular Medicine,
Emory University, 1462 Clifton Road, Suite 420, At-
lanta, GA 30322, USA. E-mail: dwallace@gmm.gen.
emory.edu
A B
C D
Fig. 1. Mitochondrial
myopathy in man and
mouse. (A) and (C) are
skeletal muscle sam-
ples from a patient
with myoclonic epilep-
sy and ragged-red fiber
disease (MERRF), which
is caused by a mutation
in the mitochondrially
encoded tRNA
Lys
gene
(38, 39). (B) and (D) are
skeletal muscle sam-
ples from a mouse with
mitochondrial myop-
athy and hypertrophic
cardiomyopathy result-
ing from the targeted
inactivation of the
gene encoding the
heart-muscle isoform
of the adenine nucleo-
tide translocator (Ant1)
(84). Frozen sections
showing a single fiber
(A) or several fibers
(B) were stained with
Gomori modified tri-
chrome to show the
ragged-red muscle fi-
bers (RRFs). Electron micrographs show (C) an abnormal mitochondrion with paracrystalline arrays
in a human RRF, and (D) the abnormal proliferation of mitochondria and degeneration of the
contractile elements in a mouse RRF.
5 MARCH 1999 VOL 283 SCIENCE www.sciencemag.org1482
M ITOCHONDRIA
REVIEW
(Tfam) are all encoded by nuclear genes (Fig.
2) (4, 5). The transfer of mtDNA sequences to
the nucleus continues to this day (13, 14). How-
ever, as the mitochondrial genome became
smaller, its genetic code began to drift. Today’s
mtDNA genes are no longer “intelligible” to the
nucleocytosolic system and the mammalian mi-
tochondrial genome is functionally stable (9).
The hundreds of mitochondria (and thou-
sands of mtDNAs) within each cell’s cytoplasm
are transmitted through the oocyte’s cytoplasm
at fertilization and thus are strictly maternally
inherited (15). The semiautonomous nature of
the mitochondria and mtDNA has been demon-
strated in experiments with cultured cells. A
mtDNA mutation imparting resistance to the
mitochondrial ribosome inhibitor chloramphen-
icol (CAP) has been transferred from CAP-
resistant (CAP
R
) cells to CAP-sensitive (CAP
S
)
cells by fusion of enucleated cytoplasmic frag-
ments (cytoplasts) from the donor cell to the
recipient cells (16 –18). This cytoplasmic hy-
brid or “cybrid” transfer technique is now a
standard assay for determining whether cellular
defects associated with mitochondrial disease
are due to mtDNA mutations (19, 20).
Mitochondrial DNA has a very high muta-
tion rate. When a mutation arises, cells initially
contain a mixture of wild-type and mutant
mtDNAs, a state known as heteroplasmy. Dur-
ing division of a heteroplasmic cell, the mutant
and wild-type mtDNAs are randomly distribut-
ed into the daughter cells, such that over many
generations the mtDNA genotype of a cellular
lineage can drift toward predominantly mutant
or wild-type mtDNAs (homoplasmy), a process
known as replicative segregation. As the per-
centage of mutant mtDNAs increases, the cel-
lular energy capacity declines until it falls be-
low the bioenergetic threshold, the minimum
energy output necessary for a cell or tissue to
function normally. Beyond this point disease
symptoms appear and become progressively
worse (4, 5).
The stochastic and quantitative nature of
mtDNA genetics means that the inheritance and
expression of heteroplasmic mutations are high-
ly variable. Indeed, it has been repeatedly dem-
onstrated that the same mtDNA mutation can
produce markedly different symptoms among
members of the same family, in association
with the chance fluctuation in the percentage of
mutant mtDNAs that each individual inherits.
Mutations in nuclear genes can also affect OX-
PHOS, often resulting in Mendelian diseases
with phenotypes similar to those caused by
mtDNA mutations. Hence, the genetics of mi-
tochondrial diseases and the associated issues
this raises in the context of genetic counseling
can be extremely perplexing (4, 5).
Mitochondrial Biology
The vagaries of mitochondrial genetics are
made even more complex by the multiple cel-
lular functions performed by the mitochondri-
on. Three of the more important aspects of
mitochondrial OXPHOS for disease pathogen-
esis are: (i) energy production, (ii) generation of
reactive oxygen species (ROS), and (iii) regu-
lation of programmed cell death, or apoptosis.
The proteins involved in OXPHOS are
located within the mitochondrial inner mem-
brane and include the electron transport chain
(ETC) components, ATP synthase, and the
adenine nucleotide translocator (ANT) (Fig.
3). The ETC oxidizes hydrogen derived from
the oxidation of organic acids such as pyru-
vate and fatty acids with atomic oxygen to
generate water. The electrons, borne on
NAD
1
(nicotinamide adenine dinucleotide),
are transferred to respiratory complex I
(NADH dehydrogenase) and then to coen-
zyme Q
10
(CoQ), and the electrons from suc-
cinate in the tricarboxylic acid (TCA) cycle
are transferred to complex II (succinate de-
hydrogenase, SDH) and to CoQ. From CoQ,
the electrons are passed to complex III, then
to cytochrome c (cyt c), then to complex IV
(cytochrome c oxidase, COX), and finally to
1/2 O
2
to give H
2
O. The energy released is
used to pump protons (H
1
) out of the mito-
chondrial inner membrane to create an elec-
trochemical gradient (DC) that is positive
and acidic on the outside and negative and
alkaline on the mitochondrial matrix side.
This creates a capacitor that can be depolar-
ized by the transport of protons back into the
matrix through a proton channel in the F
0
mem-
brane component of the ATP synthase. The
proton flux drives the condensation of ADP
(adenosine diphosphate) and P
i
(inorganic
phosphate) to make ATP, which is then export-
ed to the cytosol in exchange for the spent ADP
by the ANT. In this way, oxygen consumption
by the ETC is coupled to ADP phosphorylation
by the ATP synthase through the electrochem-
ical gradient, DC (3–5).
OXPHOS is the major endogenous source
of the ROS (O
2
z2
,H
2
O
2
, and OH
z
), which are
toxic by-products of respiration (Fig. 3). This
is because when the ETC is inhibited, the
electrons accumulate in the early stages of the
ETC (complex I and CoQ), where they can be
donated directly to molecular oxygen to give
superoxide anion (O
2
z2
). Superoxide anion is
detoxified by the mitochondrial Mn superox-
ide dismutase (MnSOD) to give hydrogen
peroxide (H
2
O
2
), and H
2
O
2
is converted to
H
2
O by glutathione peroxidase (GPx). H
2
O
2
,
in the presence of reduced transition metals,
can also be converted to the highly reactive
hydroxyl radical (OH
z
) by the Fenton reac-
tion. Chronic ROS exposure can result in
oxidative damage to mitochondrial and cel-
lular proteins, lipids, and nucleic acids, and
acute ROS exposure can inactivate the iron-
sulfur (Fe-S) centers of ETC complexes I, II,
and III, and TCA cycle aconitase, resulting in
shutdown of mitochondrial energy produc-
tion (3, 4).
Mitochondria also provide a major switch
Fig. 2. The human mtDNA
map, showing the location
of selected pathogenic mu-
tations within the 16,569-
base pair genome. Human
mtDNA codes for seven of
the 43 subunits of complex
I (ND1, 2, 3, 4, 4L, 5, and 6),
shown in red; one of the 11
subunits of complex III (cy-
tochrome b, cyt b), shown
in orange; three of the 13
subunits of complex IV
(COI, II, and III), shown in
purple; and two of 16 sub-
units of complex V (ATPase
6 and 8), shown in yellow.
It also codes for the small
and large rRNAs, shown in
green; and 22 tRNAs,
shown in beige, with the
adjacent letters indicating
the cognate amino acids.
The heavy (H)-strand ori-
gin of replication (O
H
) and
the H-strand and light (L)-
strand promoters, P
H
and
P
L
, are indicated in the con-
trol region. The mitochon-
drial transcription factor
Tfam binds between the promoters. Tfam is important in initiating transcription and in generating the
primer from the L-strand transcripts to initiate H-strand DNA replication at O
H
. The L-strand origin of
replication (O
L
) is located two-thirds of the way around the genome. The positions of representative
pathogenic point mutations are shown on the inside of the circle, with the nucleotide position and
disease acronym. All acronyms are defined in the text except “DEAF,” which signifies sensory neural
deafness and “ADPD,” which signifies late-onset Alzheimer’s disease (4, 5, 8).
www.sciencemag.org SCIENCE VOL 283 5 MARCH 1999 1483
M ITOCHONDRIA
for the initiation of apoptosis. This switch is
thought to involve the opening of a nonspe-
cific mitochondrial inner membrane channel,
the mitochondrial permeability transition
pore (mtPTP) (Fig. 3) (21–23). The mito-
chondrial inner membrane space contains a
number of cell death-promoting factors, in-
cluding cyt c, apoptosis-inducing factor (AIF,
a flavoprotein), and latent forms of special-
ized proteases called caspases. Opening of
the mtPTP causes collapse of DC, swelling
of the mitochondrial inner membrane, and
release of these death-promoting factors. The
cyt c activates the cytosolic caspase protein
degradation pathway, leading to destruction
of the cytoplasm. AIF translocates to the
nucleus, inducing chromatin destruction (24,
25). Opening of the mtPTP and the accom-
panying death of the cell can be initiated by
the mitochondrion’s excessive uptake of
Ca
21
, increased exposure to ROS, or decline
in energetic capacity (22–26). Thus, a marked
reduction in mitochondrial energy production
and a chronic increase in oxidative stress
could theoretically activate the mtPTP and
initiate apoptosis.
The interaction of these three OXPHOS
processes may explain some of the perplex-
ing pathophysiological features of mitochon-
drial disease. For example, mutations that
disrupt mitochondrial OXPHOS would re-
duce energy output and presumably impair
multiple cellular processes. Inhibition of the
ETC is likely to increase ROS production and
oxidative stress. Increased oxidative stress
and decreased energy levels might activate
the mtPTP, leading to apoptosis. Consistent
with this scenario, in patients with Complex I
defects, the severity of the enzyme deficiency
has been correlated with increased O
2
z2
pro-
duction and induction of MnSOD mRNA
(27). Moreover, in skeletal muscle samples
from patients with CPEO and MELAS (mi-
tochondrial encephalomyopathy, lactic acido-
sis, and stroke-like episodes), the RRFs
showed localized increases in MnSOD pro-
tein (28).
Mitochondrial Diseases
As noted earlier, mitochondrial diseases can
have a wide variety of inheritance patterns—
maternal, Mendelian, and a combination of
the two. Adding to this complexity is the fact
that the phenotypes of mitochondrial diseases
can be both diverse and overlapping; that is,
the same mtDNA mutation can produce quite
different phenotypes, and different mutations
can produce similar phenotypes. Hence, for
mitochondrial diseases it has been more pro-
ductive to classify patients by genetic defect
rather than by clinical manifestation.
Pathogenic mtDNA mutations include
both base substitutions and rearrangement
mutations (8). The base substitutions can be
subdivided into missense mutations that af-
fect the 13 protein-encoding genes and those
that affect the rRNA or tRNA genes, which
have global effects on mitochondrial protein
synthesis (4, 5).
The mtDNA mutations that cause LHON,
dystonia, and Leigh’s disease provide good ex-
amples of the clinical variability that can ac-
company a heteroplasmic missense mutation.
LHON and dystonia can be caused by the same
MTND6*LDYT14459A mutation,aGtoA
transition in the mitochondrial ND6 gene,
which encodes a subunit of NADH dehydroge-
nase. The mutation converts a highly conserved
alanine at codon 72 to a valine. Patients with
LHON present in midlife with sudden-onset
blindness caused by death of the optic nerve.
Patients with dystonia present early in life with
a generalized movement disorder, impaired
speech, mental retardation, and short stature,
frequently accompanied by degeneration of the
brain’s basal ganglia. LHON is thought to be
associated with a lower percentage of mutant
mtDNA and dystonia with a higher percentage
(29–31). The MTND6*LDYT14459A muta-
tion causes a substantial reduction in cellular
complex I activity and this defect can be trans-
ferred between cells in cybrid fusions (32).
However, it is not clear why the same biochem-
Fig. 3. Diagram of a mito-
chondrion, illustrating the
relationships between mito-
chondrial oxidative phospho-
rylation and (i) the produc-
tion of energy (ATP), (ii) the
generation of reactive oxy-
gen species (ROS), and (iii)
the initiation of apoptosis
through activation of the
mitochondrial permeability
transition pore (mtPTP). The
respiratory enzyme com-
plexes involved in OXPHOS
are complex I [NADH:
ubiquinone oxidoreductase],
which includes a flavin
mononucleotide and six Fe-S
centers (designated with a
cube); complex II [succinate:
ubiquinone oxidoreductase],
which includes a flavin-ade-
nine dinucleotide, three Fe-S
centers, and a cytochrome b;
complex III [ubiquinol: cyto-
chrome c oxidoreductase],
which includes cytochrome
b, cytochrome c1, and the
Rieske Fe-S center; complex
IV [cytochrome c oxidase], which includes cytochromes a1a
3
, CuA,
and CuB; and complex V [H
1
-translocating ATP synthase]. Pyruvate
from glucose enters the mitochondria via pyruvate dehydrogenase
(PDH), generating acetyCoA, which enters the tricarboxylic acid (TCA)
cycle by combining with oxaloacetate (OAA). cis-Aconitase converts
citrate to isocitrate and contains an 4Fe-4S center. Lactate dehydro-
genase (LDH) converts excess pyruvate plus NADH to lactate (3–5).
Small molecules diffuse through the outer membrane via the voltage-
dependent anion channel (VDAC) or porin. VDAC together with the
adenine nucleotide translocator (ANT), Bax, and cyclophilin D (CD)
are thought to come together at the mitochondrial inner and outer
membrane contact points to create the mtPTP (22, 23, 26). Bax is
pro-apoptotic and is thought to interact with the anti-apoptotic
protein Bcl2 and the benzodiazepine receptor (BD). The opening of the
mtPTP is associated with the release of AIF and cyt c, and activation
by cyt c of Apaf-1 and pro-caspase-9. The activated caspase-9 then
initiates the proteolytic degradation of cellular proteins, leading to
cell death (21–25).
5 MARCH 1999 VOL 283 SCIENCE www.sciencemag.org1484
M ITOCHONDRIA
ical defect can result in such different clinical
phenotypes.
Leigh’s syndrome can be caused by the
MTATP6*NARP8993G mutation,aTtoG
transition in the ATP6 gene (encoding sub-
unit 6 of mitochondrial ATP synthase) that
changes the conserved leucine at codon 156
to an arginine (33). This mutation is invari-
ably heteroplasmic and, when present in a
small percentage (,75%) of mtDNAs, it can
cause neurogenic muscle weakness, ataxia,
and retinitus pigmentosa (NARP). However,
when present in a higher percentage (.95%)
of mtDNAs, it can cause Leigh’s syndrome
(34 –36), an early-onset disease that is fre-
quently lethal. Leigh’s syndrome is associat-
ed with ataxia, hypotonia, spasticity, devel-
opmental delay, optic atrophy, ophthalmople-
gia (paralysis of the extra-ocular eye mus-
cles), developmental delay, and subsequent
regression. The end-stage disease is generally
associated with degeneration of the basal
ganglia accompanied by vascular prolifera-
tion (5). The MTATP6*NARP8993G muta-
tion causes a block in the ATP synthase F
O
proton channel, a defect that can be trans-
ferred in cybrid experiments (37). Thus, vari-
ation in the percentage of mutant mtDNAs
between patients must change the ATP output
and cause the variation in clinical symptoms.
Mutations in mitochondrial protein synthe-
sis genes can produce a complex array of symp-
toms. In severe cases the mutations are hetero-
plasmic and frequently cause central nervous
system (CNS) abnormalities as well as mito-
chondrial myopathy with ragged-red fibers
(RRFs) (Fig. 1) (2–5), an association referred to
as mitochondrial encephalomyopathy. CNS
manifestations can include sensory neural
hearing loss, epilepsy, stroke-like episodes,
and progressive dementia. Cardiomyopathy,
lactic acidosis, and endocrine disorders, in-
cluding diabetes mellitus, are also common (4).
Although the mitochondrial encephalomyopa-
thies frequently share certain clinical features
such as RRFs, specific mutations are often as-
sociated with specific clinical manifestations.
The MTTK*MERRF8344G mutation in the
tRNA
Lys
gene can cause myoclonic epilepsy
and RRFs (hence the acronym MERRF) (38,
39). The MTTL1*MELAS3243G mutation in
the tRNA
Leu
gene is often associated with
stroke-like activity and mitochondrial myop-
athy (MELAS) when it is present in a high
percentage (.85%) of mtDNAs (40), but it is
associated with maternally inherited diabetes
mellitus and deafness when present in a low
percentage (5 to 30%) of mtDNAs (41, 42).
Cultured cells from patients harboring either of
these mutations have reduced levels of mito-
chondrial protein synthesis and complex I and
IV activities, and these defects can be trans-
ferred along with the mutant mtDNA in cybrid
transfer experiments (43– 45). Milder mito-
chondrial protein synthesis mutations can be
homoplasmic and only affect the CNS. One
example is the MTTQ*ADPD4336C mutation
in the tRNA
Gln
gene, which is consistently
homoplasmic and has been associated with late-
onset Alzheimer’s disease (46, 47).
Diseases resulting from mtDNA rear-
rangements include CPEO and KSS, mater-
nally inherited diabetes mellitus and deafness
(48, 49), and the spontaneously occurring
Pearson’s marrow-pancreas syndrome, a fatal
disease associated with pediatric pancytope-
nia (loss of all blood cells) (50). CPEO and
the KSS patients present with mitochondrial
myopathy including ophthalmoplegia and
ptosis (droopy eyelids), and a subset have
cardiac defects, renal problems, diabetes mel-
litus, and other symptoms. Histological anal-
ysis of CPEO and KSS muscle has revealed
bands of COX-deficient (COX
2
) and SDH-
hyperreactive (SDH
1
) activity along the
muscle fibers. These COX
2
and SDH
1
fibers
generally correspond to regions where there
is mitochondrial proliferation with RRFs,
high levels of mutant mtDNA, and induced
expression of OXPHOS genes (51–54).
Mitochondrial DNA diseases commonly
have a delayed onset and a progressive
course. This implies that the phenotypic ex-
pression of these diseases may involve two
factors, the predisposing mutation and an
age-related factor that causes a decline in
mitochondrial function, which exacerbates
the inherited defect (4).
Mitochondrial diseases resulting from
mutations in nuclear OXPHOS genes exhibit
Mendelian inheritance patterns, yet share
many of the clinical features of mtDNA mu-
tations. For example, a mutation in the gene
encoding the 18-kD structural protein of
complex I has been reported in a child man-
ifesting hypotonia, mental retardation, con-
vulsions, and basal ganglia degeneration (55).
A mutation in a mitochondrial protease-like
adenosine triphosphatase (ATPase) has been
associated with autosomal dominant spastic
paraplegia (56). Mutations in a number of
nuclear genes have been associated with
Leigh’s syndrome. These include the genes
encoding the E1a subunit of pyruvate dehy-
drogenase (57 ), the NDUF8 N-2 Fe-S center
protein of complex I (58), the flavoprotein
subunit of complex II (59), and the SURF-1
protein associated with complex IV defects
(60, 61).
Mutations in nuclear genes can also exert
their phenotypic effects by indirectly inacti-
vating OXPHOS or destabilizing the
mtDNA. Friedreich’s ataxia is an autosomal
recessive disease that results in cerebellar
ataxia, peripheral neuropathy, and hypertro-
phic cardiomyopathy. The mutant protein,
frataxin, is targeted to the mitochondrial
inner membrane and functions to transport
iron out of the mitochondrion. With the loss
of this protein, iron accumulates in the
mitochondrial matrix, stimulating the con-
version of H
2
O
2
to OH
z
by the Fenton
reaction. This inactivates the mitochondrial
Fe-S center enzymes (complexes I, II, III,
and aconitase), which in turn reduces mi-
tochondrial energy production (62).
Other diseases that arise from nucleocy-
toplasmic interactions include the autosomal
dominant-progressive external ophthalmople-
gia (AD-PEO), the mtDNA depletion syn-
drome, and the MNGIE syndrome. AD-PEO
families inherit a dominant nuclear mutation
that predisposes them to accumulate multiple
mtDNA deletions in their skeletal muscle,
and to develop mitochondrial myopathy and
PEO. The disease has been linked to two
different chromosomal loci, but the responsi-
ble genes have not been identified (63, 64).
The mtDNA depletion syndrome is associat-
ed with severe reductions in the mtDNA lev-
els of muscle, liver, or kidney, resulting in
organ failure and death. This syndrome is
likely caused by a nuclear mutation that dis-
rupts the regulation of mtDNA copy number
during development, resulting in random loss
of the mtDNA (65, 66).
The mitochondrial neurogastrointestinal
encephalomyopathy (MNGIE) syndrome is
associated with mitochondrial myopathy with
RRFs and abnormal mitochondria, decreased
respiratory chain activity, and multiple
mtDNA abnormalities. This autosomal reces-
sive disease has been linked to mutations in
TP, a nuclear gene encoding thymidine phos-
phorylase, although the clinical symptoms
probably result from the destruction of the
mtDNA. It has been hypothesized that inac-
tivation of TP alters cellular thymidine pools
that are important in mtDNA maintenance
(67).
Somatic mtDNA Mutations in
Aging and Cancer
The delayed onset and progressive course of
mitochondrial diseases suggests that mito-
chondria function may decline with age. This
hypothesis is supported by multiple reports of
age-related declines in primate mitochondrial
OXPHOS enzyme activities in skeletal mus-
cle, liver, and brain (3, 68), and the associated
accumulation of somatic mtDNA rearrange-
ments in these same postmitotic tissues. For
example, polymerase chain reaction (PCR)
experiments have shown that skeletal muscle
from human subjects under the age of 40
contains primarily intact mtDNAs, whereas
skeletal muscle from subjects over the age of
50 shows an accumulation of a wide array of
mtDNA rearrangements (69). In addition, the
skeletal muscle of elderly subjects has been
found to have RRFs, with each COX
2
and
SDH
1
fiber containing a different mtDNA
mutation (3, 70). This confirms that each of
the mutations arose de novo and was selec-
tively amplified within the cell to create the
www.sciencemag.org SCIENCE VOL 283 5 MARCH 1999
1485
M ITOCHONDRIA
regional respiratory defects.
Somatic mtDNA mutations also occur in
the brain. Quantitation of the common 5-kb
mtDNA deletion has shown that mtDNA de-
letions accumulate markedly in the basal gan-
glia and various cortical regions in humans
after age 75 (71, 72). An analogous age-
related accumulation of somatic mtDNA re-
arrangements also occurs in mouse tissues
(73), the extent of which is proportional to
life-span rather than absolute time.
The cause of the somatic mtDNA muta-
tions is likely to be oxidative damage, which
increases with age in the mtDNA of both man
and mouse (3, 74, 75). Patients with chronic
ischemic heart disease, which is associated
with cyclic bursts of mitochondrial ROS dur-
ing ischemia and reperfusion (76), have been
found to harbor 8 to 2000 times more mtDNA
deletions in the heart than age-matched con-
trols (77). Similarly, cortical mtDNA dele-
tion levels are elevated in patients with Alz-
heimer’s and Huntington’s disease (78, 79),
and mtDNA from the former group shows
increased oxidative damage (80).
These observations have led to the hypoth-
esis that somatic mtDNA mutations accumulate
in postmitotic tissues with age as a result of
mitochondrial ROS damage. The resulting age-
related decline in OXPHOS would ultimately
degrade the tissue’s bioenergetic capacity until
it falls below a certain threshold, resulting in
symptoms and senescence. This same age-
related decline in OXPHOS could interact with
inherited mitochondrial defects, which would
account for the delayed onset and progression
of mitochondrial diseases.
Somatic mtDNA mutations have also
been identified in various tumors and tumor
cell lines. These mutations include intragenic
deletions (81), missense and chain-termina-
tion point mutations (82), and alterations of
homopolymeric sequences that result in
frameshift mutations (83). In principle, these
mutations could contribute to neoplastic
transformation by changing cellular energy
capacities, increasing mitochondrial oxida-
tive stress, and/or modulating apoptosis.
Mouse Models of Mitochondrial
Disease
Patient studies have revealed much about the
genetics of mitochondrial disease, but the
pathophysiological mechanisms that underlie
the complex array of symptoms remain myste-
rious. Recently, new insights into this question
have been obtained by the creation and analysis
of mouse models for mitochondrial disease.
Four such models are discussed below.
Ant1-deficient mice. Targeted inactivation
of the nuclear-encoded Ant1 gene has provid-
ed a model for chronic ATP deficiency. Sur-
prisingly, ANT1-deficient (Ant1
2/2
) mice
are viable, although they develop classical
mitochondrial myopathy and hypertrophic
cardiomyopathy (84).
In mice, ANT is encoded by two genes.
Ant1 is expressed at high levels in skeletal
muscle and heart and at lower levels in brain,
whereas Ant2 is expressed in all tissues but
skeletal muscle (84). Thus, Ant1
2/2
mice are
completely deficient in ANT in skeletal mus-
cle, partially deficient in heart, and have nor-
mal ANT levels in liver. The skeletal muscle
of Ant1
2/2
mice exhibit classic RRFs (Fig.
1) and increased SDH and COX staining in
the Type I oxidative muscle fibers. These
elevated OXPHOS enzyme activities corre-
late with a massive proliferation of giant
mitochondria in the skeletal muscle fibers,
degeneration of the contractile fibers (Fig. 1),
and a marked exercise fatigability. The hy-
pertrophic cardiomyopathy is also associated
with mitochondrial proliferation.
Both nuclear and mitochondrial OX-
PHOS genes are upregulated in Ant1
2/2
mice. Among the transcripts overexpressed
in skeletal muscle are mRNAs for several
components of complex I and complex IV,
various mitochondrial transcripts, and the
mRNA for Mcl-1, a muscle homolog of
Bcl-2 (Fig. 3) (85). This implies that ANT1
deficiency may affect not only mitochon-
drial bioenergetics, but also oxidative stress
and possibly apoptosis.
Ant1
2/2
mice also have elevated serum
levels of lactate, alanine, and succinate, con-
sistent with inhibition of the respiratory chain
and TCA cycle. Their skeletal muscle mito-
chondria are completely resistant and their
heart mitochondria are partially resistant to
the stimulation of oxygen consumption by
exogenous ADP. This is consistent with the
reductions in ANT levels and the associated
inhibition of ADP/ATP exchange. The result-
ing limitation of available matrix ADP for the
ATP synthase would be expected to reduce
proton flux through its F
o
proton channel,
hyperpolarize the mitochondrial inner mem-
brane inhibiting the ETC, and redirect elec-
trons from the ETC into the ROS-generating
pathway. Indeed, H
2
O
2
production was in-
creased six- to eightfold in skeletal muscle
and heart mitochondria from Ant1
2/2
mice,
to levels comparable to those observed when
control mitochondria are treated with the
complex III inhibitor Antimycin A. This in-
crease in ROS production was paralleled by
an eightfold induction of the O
2
z2
detoxify-
ing enzyme MnSOD in skeletal muscle mi-
tochondria, but not in heart mitochondria
(86). In the heart, the increased O
2
z2
produc-
tion, without a compensating induction in
MnSOD, was associated with a dramatic in-
crease in heart mtDNA damage. In fact, the
heart mtDNAs of 16- to 20-month-old
Ant1
2/2
mice had mtDNA rearrangement
levels comparable to those seen in 32-month-
old Ant1
1/1
animals (86). Hence, inhibition
of OXPHOS not only reduces energy produc-
tion, but also elevates ROS production with
an associated increase in damage to the mito-
chondria and mtDNA.
MnSOD-deficient mice. The origins and
consequences of increased mitochondrial
ROS production have been confirmed by
studying mice carrying inactivating muta-
tions in the nuclear gene encoding mito-
chondrial MnSOD. Two mouse lines lack-
ing MnSOD (Sod2) have been constructed,
Sod2
tm1Cje
(87) and Sod2
tm1Leb
(88). The
Sod2
tm1Cje
mutation on the CD1 back-
ground results in neonatal death from dilat-
ed cardiomyopathy (87). The Sod2
tm1Leb
mutation on the C57BL/6 background re-
sults in death at about day 18, in associa-
tion with neuronal degeneration in the basal
ganglia and brain stem (88).
The Sod2
tm1Cje
mutation on the CD1
background has been extensively character-
ized. These MnSOD-deficient mice die from
dilated cardiomyopathy at about 8 days of
age (87, 89), and they exhibit striking lipid
deposits in the liver. The mice also have
significant reductions of complex II and com-
plex III activity in skeletal muscle and heart,
of complex I activity in heart, and of mito-
chondrial aconitase activity in heart and
brain. In addition, the animals’ urine contains
organic acids that are characteristically found
in the urine of patients with HMG (3-hy-
droxy-3-methylglutaryl)-CoA lyase deficien-
cy, and the activity of this enzyme is signif-
icantly reduced in the liver of the mutant
mice. Finally, DNA from the heart and brain
of the mice shows accumulation of oxidative
DNA damage (90).
These observations indicate that the high
levels of mitochondrial O
2
z2
anion resulting
from the MnSOD-deficiency can inhibit the
ETC (complexes I and II) and the TCA cycle
(aconitase), probably by inactivating the Fe-S
centers in these enzymes. Consistent with this
notion, the cardiac defect in the Sod2
2/2
mice
can be rescued by administration of the antiox-
idant MnTBAP [manganese 5,10,15,20-tetrakis
(4-benzoic acid) porphyrin], which converts
O
2
z2
to H
2
O
2
(89). Peritoneal injection of this
compound eliminated the dilated cardiomyop-
athy, reduced the liver lipid deposition, and
extended the mean life span of the mice from 8
to 16 days. However, MnTBAP does not cross
the blood-brain barrier, and by 12 days of age
the MnTBAP-treated mice began to develop
debilitating movement disorders in association
with neuronal cell loss in the brain’s motor
centers.
Although the acute pathology observed in
the Sod2
2/2
mice confirms the importance of
mitochondrial ROS toxicity in mitochondrial
dysfunction, many clinical syndromes such
as Friedreich’s ataxia are associated with
chronic ROS exposure. The analysis of het-
erozygous Sod2
1/2
mice (91), which exhibit
a 50% reduction in MnSOD levels, may pro-
5 MARCH 1999 VOL 283 SCIENCE www.sciencemag.org
1486
M ITOCHONDRIA