Mitochondrial DNA deletions are
abundant and cause functional
impairment in aged human
substantia nigra neurons
Yevgenya Kraytsberg1, Elena Kudryavtseva1, Ann C McKee2,3,
Changiz Geula1, Neil W Kowall2,3& Konstantin Khrapko1
Using a novel single-molecule PCR approach to quantify
the total burden of mitochondrial DNA (mtDNA) molecules
with deletions, we show that a high proportion of individual
pigmented neurons in the aged human substantia nigra contain
very high levels of mtDNA deletions. Molecules with deletions
are largely clonal within each neuron; that is, they originate
from a single deleted mtDNA molecule that has expanded
clonally. The fraction of mtDNA deletions is significantly
higher in cytochrome c oxidase (COX)-deficient neurons than
in COX-positive neurons, suggesting that mtDNA deletions may
be directly responsible for impaired cellular respiration.
Somatic mutations in mtDNA have been hypothesized to be respon-
sible for some aspects of the aging process1,2. Indeed, mtDNA
mutations are capable of producing aging-like phenotypes in mito-
chondrial diseases ranging from cardiomyopathy to neurodegenera-
tion: for example, in mitochondrial diseases. Recently, multiple aging
phenotypes have been demonstrated in transgenic mice with an
increased rate of somatic mtDNA mutations (reviewed in ref. 3).
However, a major challenge to the mtDNA mutational theory of aging
has been the presumably low abundance of mtDNA mutations in
normally aging tissues3,4, such that their physiological significance
may be questionable. It was shown over a decade ago that a particular
type of mtDNA mutations, mtDNA deletions, are distributed in a
highly nonuniform manner among different tissues and within the
same tissue, particularly among different areas of the brain5,6. Some
brain areas have been reported to sustain a few orders of magnitude
more deletions than others. It is tempting to hypothesize that mtDNA
mutations, although rare on average, may reach sufficiently high
concentrations in specific cell types to significantly impair cellular
processes in an age-dependent manner. Measurements of mtDNA
deletions are technically difficult, however, and reliable quantitative
studies have focused on a single ‘common’ type of deletion5,6. Thus,
the absolute burden of all possible mtDNA deletions, the most
important parameter with respect to physiological relevance of these
mutations, remained unknown.
The substantia nigra, the primary site of neurodegeneration in
Parkinson disease, sustains particularly high levels of ‘common’
mtDNA deletions compared with other brain areas6. Notably, a large
proportion of pigmented neurons in substantia nigra have been shown
by immunohistochemistry to lose COX, an mtDNA-encoded enzyme,
with increasing age7. No direct relationship between these defects and
mtDNA mutations has yet been reported, although our previous work8
suggested that individual pigmented neurons may contain very high
quantities of all mtDNAdeletions combined. To more fullyexplore this
relationship, we developed a new approach to quantify the total
cellular burden of mtDNA deletions using single-molecule PCR
(smPCR), as described in the Supplementary Note online. We then
used this quantitative smPCR method to directly measure mtDNA
deletions in individual neurons of substantia nigra of various ages and
compare the mutational loads of COX-positive and COX-deficient
neurons. We and others have previously used this smPCR approach to
perform mutational analysis in other systems (reviewed in ref. 9).
We collected 80 individual cells from human substantia nigra
specimens, aged 33–102 years, and determined the percentage
of deleted mtDNA molecules in each cell in order to assess the
Fraction of deleted mtDNA per neuron
Age of individual
Figure 1 Fraction of deleted mtDNA in individual
pigmented neurons of substantia nigra from
subjects of different ages. Each bar represents
the mutant fraction in a single cell as determined
by single-molecule PCR or extended PCR. Bars
representing neurons from the same individual
of a certain age are grouped, and the age is
indicated under the group. Error bars represent
standard error with respect to repeated
measurements of the same cell (n Z 3). Zero-
height bars represent cells that were determined
to be deletion-free by extended PCR. See
Supplementary Note for methodological details.
Received 13 December 2005; accepted 6 March 2006; published online 9 April 2006; doi:10.1038/ng1778
1Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts 02215, USA.2Boston University School of Medicine, Boston,
Massachusetts 02118, USA.3Geriatric Research Education and Clinical Center, Bedford Veterans Affairs Medical Center, Bedford, Massachusetts 01730, USA.
Correspondence should be addressed to K.K. (email@example.com).
518VOLUME 38 [ NUMBER 5 [ MAY 2006 NATURE GENETICS
© 2006 Nature Publishing Group http://www.nature.com/naturegenetics
distribution of fractions of deleted mtDNA that accumulate in
individual pigmented neurons with age (Fig. 1). The number of
mtDNA deletions was significantly different between old and young
tissues (P o 0.0001 by two-sample, two-tailed, homeoscedastic t-test,
regardless of cutoff age, 40 to 69 years). Moreover, there was a very high
absolute prevalence of mtDNA deletions in neurons from aged sub-
stantia nigra (Fig. 1). In addition, in many neurons, the fraction of
deletions exceeded 60%, which is believed to be the phenotypic thresh-
old (the fraction above which mtDNA deletions impair respiratory
function; reviewed in ref. 10). Younger pigmented neurons occasionally
accumulate high fraction of deletions, as in one of the cells from a 38-
year-old (Fig. 1), but such events are rare. Several other cell types,
including pyramidal neurons of the cerebral cortex, cerebellar Purkinje
cells and large neurons of the dentate nucleus of aged individuals usually
contained undetectable deletion levels, consistent with the notion that
the accumulation of mtDNA deletions is highly cell type–specific5,6.
Similar results have been reported independently by another group using
a different approach (real-time PCR) to quantify mtDNA deletions11.
We then explored whether the highly abundant mtDNA deletions
that we observed have an effect on COX activity. A large representative
study7on four areas of substantia nigra from the brains of 36 subjects
aged 30–99 years old has shown that up to 30% of pigmented neurons
in the older brains are COX deficient, albeit with a large case-to-case
variation. Capitalizing on these data, we attempted to determine
whether COX deficiency develops in cells with the highest fractions
of mtDNA deletions. Because it is particularly important to address
this question in the most affected individuals, we selected an 80-year-
old brain with high frequency of COX-deficient neurons for a detailed
cell-by-cell study. A tissue section from this individual stained for COX
activity (Fig. 2a) shows that a local population of pigmented neurons
contains about 30% COX-deficient cells; that is, cells with decreased,
but not necessarily absent, COX activity (Supplementary Note online).
We collected individual COX-positive and COX-deficient neurons and
measured the fraction of deletions in each cell (Fig. 2b). In accord with
our hypothesis, COX-positive cells contained a significantly lower
fraction of deletions than COX-deficient cells (P o 0.0001; two-
sample, two-tailed, homeoscedastic t-test). Notably, the fractions of
deletions in all COX-deficient cells were not only higher than in COX-
positive cells but they actually exceeded the 60% threshold and thus
might be sufficient to cause the observed COX deficiency, although
more data will be necessary to confirm this observation.
The difference in distribution of mtDNA deletions between COX-
positive and COX-deficient neurons strongly implies that the mtDNA
deletions may be one of the primary causes of the COX defect.
However, an alternative possibility is that some other local primary
defect (such as increased free radical production or a defect in DNA
maintenance) might have caused secondary accumulation of deletions
and the development of respiratory deficiency. To evaluate this
possibility, we assessed, in individual neurons, the distribution
of mtDNA deletions by type in individual neurons. The DNA
of individual neurons was subjected to ‘extended’ PCR capable of
amplifying the entire mitochondrial genome12. Although extended
PCR is not quantitative, it helps to determine the specific types
of mtDNA deletions present in a cell. A typical example of such an
analysis is presented in the Supplementary Note online. In most cases,
a neuron either contained no deletions or contained a single species of
deletion that showed up repeatedly in several independent PCRs. The
presence of multiple identical molecules of deleted mtDNA in a cell
implies that mtDNA deletions are clonal; that is, they originate from a
single initial mutant molecule that has multiplied. Thus, expansion of
preexisting mutant molecules, rather than ongoing de novo mutational
events (as one would expect in the case of a local DNA maintenance
defect or increased free radical production), is primarily responsible
for accumulation of mtDNA deletions in individual neurons.
In conclusion, here we have provided evidence for direct involve-
ment of mtDNA deletions in the development of COX defects in the
aged human substantia nigra. The relationship between these COX
defects and aging awaits further investigation. It is worth noting,
however, that the incidence of mild parkinsonian signs (MPS), a
movement disorder common in the older population (prevalence of
25% over 65 years of age and 50% over 85 years of age) associated
with significant morbidity and excessive mortality13, is similar to the
percentage of aged individuals with an elevated fraction of COX-
deficient neurons in substantia nigra (B40% over 80 years old)7. COX
defects may affect performance of substantia nigra either by rendering
the affected neurons dysfunctional via disrupted energy supply or
by prompting neuronal death. Either of these end points is expected
to predispose to MPS, as MPS have been shown to correlate with
decreased neuron counts in substantia nigra14.
The influence of mtDNA deletions may not be limited to substantia
nigra. The putamen and the caudate nucleus5,6have been reported to
contain high fractions of the ‘common’ mtDNA deletion and thus are
likely to contain cell types significantly affected by mtDNA deletions.
Our preliminary results indicate that respiratory defects in specific
neurons from other brain areas are also caused by clonal expansion
of mtDNA deletions. Therefore, we believe that a cell type–specific
quantitative study of mtDNA mutations in the brain may uncover
additional critical areas where aging is driven by mtDNA mutations.
These studies are of practical importance, as procedures aimed at
alleviating the effects of mtDNA mutations are currently in develop-
ment (reviewed in ref. 15).
Note: Supplementary information is available on the Nature Genetics website.
The authors are grateful to A. Griner, A. Kraytsberg, and A. Vaysburd for help
in experiments; E. Richfield (Rutgers University) for tissue samples and critical
Fraction of deleted DNA per neuron
COX+ cellsCOX– cells
2 3 4 5 6 7 8 9 10 11 12
Figure 2 Clonal expansions of mtDNA deletions are associated with
COX defects in individual neurons. (a) COX-specific immunostaining
of the substantia nigra pars compacta of an 80-year-old (Cresyl violet
counterstain). COX-positive cells appear brown, whereas COX-deficient
neurons appear violet owing to counterstaining with Cresyl violet. Dense
black granules are the neuromelanin aggregates characteristic of pigmented
neurons. Cells identified as COX-positive or COX-deficient were individually
collected by laser capture microdissection and analyzed for mtDNA
deletions. (b) Mutational analysis of individual neurons by single-molecule
PCR. Each bar represents the fraction of deletions in a single neuron;
brown and blue bars represent COX-positive and COX-deficient neurons,
respectively. Numbers under the bars correspond to the numbers assigned
to individual collected cells, some of which are visible in a, though not all
the collected cells fit within the field of view. Error bars represent standard
error (n Z 3).
NATURE GENETICS VOLUME 38 [ NUMBER 5 [ MAY 2006519
© 2006 Nature Publishing Group http://www.nature.com/naturegenetics
review of the manuscript; W. Kunz for communicating his histochemistry
protocols and O. Kocher (Beth Israel Deaconess Medical Center) for granting
access to a laser capture microscope. This work was supported in part by US
National Institutes of Health grants ES11343 and AG19787 to K.K. and AG13846
(Boston University Alzheimer Disease Center) to N.W.K. and the Department of
COMPETING INTERESTS STATEMENT
The authors declare that they have no competing financial interests.
Published online at http://www.nature.com/naturegenetics
Reprints and permissions information is available online at http://npg.nature.com/
1. Harman, D. J. Am. Geriatr. Soc. 20, 145–147 (1972).
2. Linnane, A.W., Marzuki, S., Ozawa, T. & Tanaka, M. Lancet 1, 642–645 (1989).
3. Khrapko, K., Kraytsberg, Y., de Grey, A., Vijg, J. & Schon, E.A. Aging Cell (in the press).
4. Jacobs, H.T. Aging Cell 2, 11–17 (2003).
5. Corral-Debrinski, M. et al. Nat. Genet. 2, 324–329 (1992).
6. Soong, N.W., Hinton, D.R., Cortopassi, G. & Arnheim, N. Nat. Genet. 2, 318–323
7. Itoh, K., Weis, S., Mehraein, P. & Muller-Hocker, J. Neurobiol. Aging 17, 843–848
8. Nekhaeva, E., Kraytsberg, Y. & Khrapko, K. Mech. Ageing Dev. 123, 891–898 (2002).
9. Kraytsberg, Y. & Khrapko, K. Expert Rev. Mol. Diagn. 5, 809–815 (2005).
10. Rossignol, R. et al. Biochem. J. 370, 751–762 (2003).
11. Bender, A. et al. Nat. Genet., advance online publication 9 April 2006 (doi:10.1038/
12. Khrapko, K. et al. Nucleic Acids Res. 27, 2434–2441 (1999).
13. Bennett, D.A. et al. N. Engl. J. Med. 334, 71–76 (1996).
14. Ross, G.W. et al. Ann. Neurol. 56, 532–539 (2004).
15. Khrapko, K. Rejuvenation Res. 8, 6–8 (2005).
520VOLUME 38 [ NUMBER 5 [ MAY 2006 NATURE GENETICS
© 2006 Nature Publishing Group http://www.nature.com/naturegenetics