Mitochondrial DNA deletions in mice in men: Substantia nigra is much less affected in the mouse

Biochimica et Biophysica Acta 1797 (2010) 1159–1162

Contents lists available at ScienceDirect

Biochimica et Biophysica Acta j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / b b a b i o

Mitochondrial DNA deletions in mice in men: Substantia nigra is much less affected in the mouse Xinhong Guo a,b, Elena Kudryavtseva a, Natalya Bodyak a, Alexander Nicholas a, Igor Dombrovsky a,c, Deye Yang a,d, Yevgenya Kraytsberg a, David K. Simon a, Konstantin Khrapko a,⁎ a

Harvard Medical School and Beth Israel Deaconess Medical Center, Boston, MA, USA Institute of Life Science and Technology, Hunan University, Changsha, P.R. China Boston University, Boston, MA, USA d Division of Cardiology, The First Affiliated Hospital, Wenzhou Medical College, Wenzhou P.R. China b c

a r t i c l e

i n f o

Article history: Received 2 February 2010 Received in revised form 7 April 2010 Accepted 7 April 2010 Available online 11 April 2010 Keywords: Aging Mutations Brain Human Mice Mitochondrial DNA

a b s t r a c t Mitochondrial DNA (mtDNA) deletions have been reported to accumulate to high levels in substantia nigra of older humans, and these mutations are suspected of causing age-related degeneration in this area. We have compared levels of mtDNA deletions in humans and mice and report here that levels of deletions in the mouse are very significantly lower than in humans. While human mtDNA from substantia nigra contained more than 5% of deleted molecules, mouse substantia nigra contained less than 0.5%. These results imply that mtDNA deletions are unlikely to play any significant role in of murine substantia nigra aging and further call for caution in using mouse models in studies of the role of mtDNA deletions in aging and neurodegeneration. On a more general note, these results support the view that critical targets of the various aging processes may differ significantly between distant species. © 2010 Elsevier B.V. All rights reserved.

1. Introduction Mitochondrial DNA mutations have been long suspected to cause aging in mammals, though their role remains controversial [1]. Endogenous mitochondrial mutations sustain surprisingly high mutational rates [2], and are capable of efficient intracellular expansion, which eventually results in defect in oxidative phosphorylation [3] and potentially make age-related accumulation of these mutations a fair candidate for a significant component of the aging process. Of all mitochondrial mutations (i.e. point mutations and large rearrangements or deletions), the latter are recently receiving particular attention in studies from rodents to humans as potential culprits of the agerelated deterioration of various tissues including muscle [4,5], substantia nigra [6,7], and kidney [8]. Substantial similarities have been discovered in the distribution of mtDNA deletions in human and rodent muscle fibers [4,5]. This latter observation seems to support a general presumption that the mouse is a reasonable model system for human aging from the mtDNA perspective. The importance of this presumption is highlighted by recent work using models of mice with increased

⁎ Corresponding author. E-mail address: [email protected] (K. Khrapko). 0005-2728/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.bbabio.2010.04.005

rates of mtDNA deletions and a syndrome reminiscent of premature aging [9–12] as a model of mammalian aging [13]. It has been argued though that it is key for validation of these models to determine whether the distribution of deletions among various tissues faithfully represents the distribution that occurs in normal aging, either in the mouse or in human [14]. We have been interested in this issue since many years ago when we failed to see any deletions in a study of a large number of single cardiomyocytes from old murine hearts (Khrapko, unpublished data). This contrasted with human cardiomyocytes, of which a substantial proportion showed clonally expanded deleted mtDNA molecules [15]. In human heart, however, a low fraction of these deletions in each of the affected cardiomyocytes makes these deletions poor candidates for a culprit of the aging process. Much higher overall fractions and cell-bycell levels of mtDNA deletions have been discovered in the human substantia nigra [6,7], a compact midbrain area rich in dopaminergic pigmented neurons. High deletion levels make this area one of the most probable candidate tissues where age-related deterioration may be driven by mtDNA deletions. We therefore decided to explore the hypothesis that mice are similar to humans with respect to the accumulation of mtDNA deletions in substantia nigra. To do the comparison, we isolated substantia nigra tissue from aged murine and human brains and measured mutant fractions of deleted mtDNA. To our surprise, our experiments demonstrated that the levels of deletions in aged murine substantia nigra were much lower that those in aged humans.

1160

X. Guo et al. / Biochimica et Biophysica Acta 1797 (2010) 1159–1162

2. Materials and methods 2.1. Tissues and DNA preparation Substantia nigra has been identified on human midbrain sections by position and dark color (neuromelanin). The area containing pigmented neurons was scraped from the section saturated with glycerol for better visibility as described elsewhere [16]. In the mouse, substantia nigra was identified by its position in the midbrain by inspecting a large series of frozen sections of the entire mouse brain. The identity of the area was further confirmed by tyrosine hydroxylase immunostaining of adjacent sections. The tissue was homogenized in a minimal volume (about 10 μl) of proteinase K/SDS buffer (0.1 mg/ml proteinase K, 0.5% SDS, 10 mM EDTA). We used two 80-year-old human brains, one normal and one with Parkinson disease with typical Lewy body pathology. We also analyzed the brains from a 34-month-old and a 27-month-old mouse. This very limited number of samples in this comparative study was justified based on the similarity of deletion levels in these diverse human samples (as far as the disease state is considered) compared to the huge difference in deletion levels between human and murine samples, in addition to practical considerations relating to available samples and the large amount of work involved in analyzing large mtDNA deletion levels in each sample. 2.2. Single molecule PCR The tissue homogenates were diluted appropriately for single molecule PCR (smPCR) as described elsewhere [17], i.e. about 3 orders of magnitude, to the point where about half of multiple parallel PCR reactions set up using aliquots of the diluted sample were producing PCR products and half were not. The appropriate dilution was determined empirically in pilot experiments. Counting the number of positive reactions allows an estimate of the concentration of molecules in the original homogenate and of the fraction of mtDNA deletions as described below. The primers used for PCR were for human 13-kb PCR: first stage nested PCR: 2999F and 16450R, 20 cycles, second stage: 3204F and 16201R, 30 cycles, and for 6-kb PCR. For mouse mtDNA PCR (13 kb): first stage 2417F and 15833R, 20 cycles, second stage 2496F and 1588R, 30 cycles. LaTaq thermostable DNA polymerase from TaKaRa was used according to manufacturer's recommendations. Each cycle was 20 sec 94C, 13 min 68C for 13-kb PCR and 6 minutes for 6-kb PCR. 3. Results 3.1. Estimation of fraction of deletions, humans vs. mice Estimating of the fraction of mtDNA deletions by single molecule PCR is done by counting the number of molecules corresponding to the wild type mtDNA (13 kb fragment, encompassing the entire major arch and about half of the minor arch) and the number of shorter PCR fragments (b11 kb), that correspond to mtDNA molecules harboring large deletions, as determined by agarose gel separation of the PCR products. The approach we used however was different for human and mouse samples. Our preliminary estimates demonstrated that fractions of deletions in human samples of substantia nigra were relatively high, so our approach was to count, isolate and sequence these deletions. In contrast, preliminary estimates in the mouse revealed that murine samples bear very low levels of deletions, so our strategy was to make an upper estimate of the deletion levels for the purpose of a statistically sound and conservative comparison to human, but we did not aim at isolating and studying these rare and likely functionally irrelevant deletions. The procedures described below are designed so that the estimates are conservative for human samples (i.e. our estimates tend to be underestimates), whereas the

estimates for the mouse samples tend to be overestimates. These biases would result in an overall underestimation of the differences in deletion levels between humans and mice. Thus, our conclusion that murine levels are much lower than those in humans is highly reliable as it is based on a “double conservative” estimate. Dilutions of DNA isolated from human substantia nigra homogenates were treated with the restriction enzyme SnaBI. This treatment rendered wild type mtDNA molecules non-amplifiable (because SnaBI cuts in the middle of the PCR amplicon). Thus the only amplifiable molecules remaining are mtDNA deletions, i.e. those of them in which the deleted part includes the SnaBI site. Our previous studies showed that essentially all deletions in human substantia nigra remove the SnaBI restriction site [7]. As expected, almost all single molecule PCR products represent mtDNA deletions (except for a few wild type molecules that managed to escape restriction). Deletions were counted and the resulting numbers were used in statistical analysis. 30 molecules from one of the samples were mapped and sequenced (data not shown). Each DNA sample was further subjected to single molecule PCR using a different set of primers which result in a 6-kb-long PCR fragment that does not encompass the SnaBI site, so that wild type mtDNA molecules are readily amplified, and therefore the number of positive reactions in the corresponding smPCR represented the number of wild type molecules in the DNA sample. Comparison of the smPCR counts of the deletions in the 13-kb PCR to the counts of all molecules in the 6-kb PCR yielded an underestimate of the fraction of deletions in human samples. This is because the efficiency of amplification of shorter PCR fragments is higher that that of longer ones (see Note on controls for artifacts below for explanation). The vast majority of mtDNA deletions yield PCR fragments between 6 and 11 kb, and hence comparison to the more efficiently amplified reference 6 kb fragment would tend to result in an underestimate of the fraction of deletions. The 13-kb and 6-kb experiments described above were repeated at least four times for each sample, with over 100 deletions and 6-kb single molecule PCR fragments harvested per sample, and concentrations of deleted (13-kb PCR) and wild type (6-kb PCR) mtDNA molecules were determined. Using these estimates of the concentrations of the wild type and the deleted molecules, we calculated that the lower estimates of the percentage of mtDNA molecules harboring large deletions in human substantia nigra homogenates to be 4 × 10−2 (95% confidence limits: 2.8 × 10−2 to 5.2 × 10−2) and 5 × 10−2 (3.5 × 10−2 to 6.5 × 10−2) in the two samples we studied. Mouse homogenates were used without prior restriction enzyme digestion and were subjected to PCR at single molecule and higher concentrations. Initial experiments at single molecule concentrations yielded no deleted PCR products, indicating very low levels of deletions. The number of wild type molecules in these experiments was used to estimate mtDNA concentrations in the samples and then PCR was performed at higher concentrations, corresponding to about 10 wild type molecules per well (see Table 1). We have run PCR reactions equivalent to about 1500 amplified wild type genomes, and detected one deleted molecule, or 7 × 10−4 (95% conf limits: 0.025–5.6 molecules, or 1.7 × 10−5 to 3.7 × 10−3) in the 27 month old mouse and 3 deleted molecules or 2 × 10−3 (95% conf limits: 0.62–8.8 molecules or 4 × 10−4 to

Table 1 Fractions of mtDNA substantia nigra deletions in mice are much lower than in humans. Note that lower confidence limits of the fraction of deletions in human tissue are at least five fold higher than the upper confidence limits for the mouse (the numbers to compare are in bold). Samples

Human s. nigra Mouse s. nigra

Fraction of deletions

80 years old, norm. 80 years old, PD 27 months old 34 months old

0.04 0.05 0.0007 0.002

95% Conf. limits Lower

Upper

0.028 0.035 0.000017 0.0004

0.052 0.065 0.0037 0.006

X. Guo et al. / Biochimica et Biophysica Acta 1797 (2010) 1159–1162

6 × 10−3) in the 34-month-old mouse. Confidence limits were estimated assuming Poisson distribution of the counts of molecules [18]. Thus our estimates of mutant fractions in the mouse are more than an order of magnitude lower than those for humans and even the upper 95% confidence limits for both samples are five times lower than the lower confidence limits of human estimates.

3.2. Note on controls for artifacts As in any study of low frequency somatic mutations, this study is potentially at risk of artifacts. Previously we have argued that single molecule PCR approach is relatively resistant to PCR-derived artifacts [17]. Nevertheless, appropriate controls are necessary. Possible artifacts are artificially generated deleted PCR fragments that may originate in the jumping PCR [19,20]. This artifact is particularly difficult to control for, because artificial deletions may potentially be induced by DNA damage, so the use of control DNA is not a sufficient measure. Indeed, the absence of deletions in a control DNA sample does not guarantee that the procedure is artifact-free, unless one can demonstrate that the levels of DNA damage are similar between the sample and the control, which is not a trivial task. In our system, an efficient internal control is provided by the distribution of breakpoints. Sequencing of deletion breakpoints demonstrates that breakpoints as determined by our approach are tightly constrained by biological boundaries, i.e. the frequency of 5' deletion breakpoints drops dramatically outside socalled major arch of the mitochondrial genome. Any deletions with 5' breakpoints outside this area would have removed the light strand origin of replication and therefore the resulting molecules presumably would not be able to efficiently propagate themselves in vivo. There is no rational explanation why artificial DNA damage-induced breakpoints would not form outside the major arch. In fact there is a significant pressure on artificial breakpoints to form beyond the light strand origin as corresponding deletions will result in shorter PCR products and thus would have a selective advantage in initiating PCR amplification because of DNA damage (discussed below). The absence of breakpoints beyond light strand origin of replication in our set of deletions is thus a strong argument against a significant generation of artificial deletions in our experiments. Another issue related to artifacts is PCR bias. It is well established that amplification efficiency decreases with the fragment length, which may significantly distort quantitative PCR results. This problem has been circumvented in our experiments by the use of smPCR. Amplification of a single molecule obviously ensures that there is no competition between molecules of different length as they are segregated into different PCR reaction tubes [17]. Note that this problem does not pertain to quasi-smPCR conditions used in this study: we did run PCR reactions with multiple wild type molecules for the mouse mtDNA. Since in this case our goal was to count shorter deleted molecules among an excess of wild type molecules, their better ability to amplify merely made our task easier and in no way could have distorted our results. A more difficult to address source of bias is the variable DNA damage. To illustrate this phenomenon let us assume that human DNA is more damaged that that of the mouse. If a lesion completely prevents a DNA molecule from PCR amplification [21], then a higher incidence of such lesions in human DNA could result in a higher relative number of amplifiable deleted mtDNA molecules in humans than in the mouse, because shorter deleted molecules have proportionally a lower probability of sustaining a PCR impassable lesion. In that case, deleted mtDNA molecules would have an advantage in single molecule PCR. Note that this advantage is realized not during all amplification cycles, but rather at the very first duplication only. This would have risen the possibility that the higher proportion of deletions in humans is simply an artifactual consequence of the higher DNA damage in human samples. Our approach of comparing human deletions with a shorter, 6-kb

1161

fragment, and mouse deletions with a longer, 13-kb fragment (see Results section), addresses this problem . 4. Discussion Our observation that the incidence of mtDNA deletions in murine substantia nigra is over an order of magnitude lower than that in humans came as a surprise, as the general expectation was that mouse aging is similar to human aging. Our findings clearly imply that this is not always the case and calls for caution in interpretation of data obtained in model animals onto human case. It is important to study the levels and distributions of mtDNA mutations before using a particular model. Interestingly, in some cases, the murine and the human distribution and dynamics of mtDNA mutations may be very similar, as documented in most impressive work of the Aiken group for muscle [4,5]. The message of this study is that this similarity, however, cannot be, in general, taken for granted. This result is particularly interesting because the substantia nigra in humans is considered one of the tissue types most affected by mtDNA deletions. We still do not know whether these high levels of deletions are of any relevance to aging and/or neurodegeneration of substantia nigra neurons in humans [1], but much lower levels of deletions in mice imply that these deletions most likely are irrelevant for mouse aging. Interestingly, these lower levels of mitochondrial deletions in the murine substantia nigra appear to correlate with another nigral aging phenomenon in substantia nigra: in humans, dopaminergic neurons of the substantia nigra start accumulating neuromelanin, which is thought to be related to high oxidative stress in these cells. This process starts around 20 years of age in humans when the substantia nigra noticeably turns dark in color. In contrast to humans, the mouse substantia nigra does not turn dark even at oldest ages, and the amount of melanin in murine dopaminergic neurons appears much lower and requires special staining techniques to be detected [22]. It is tempting to speculate that the relative slowness of both processes, i.e. accumulation of neuromelanin and deletions in mice reflect some fundamental difference in the aging of substantia nigra in humans and mice. We have previously reported that a majority of pigmented neurons of the substantia nigra in humans at old age contain over 50% of mtDNA deletions. This result is fully compatible with our lower current estimates of deletion levels in human substantia nigra homogenates. The difference probably results from two factors. First, homogenates contain many other cells besides pigmented neurons, and we have demonstrated earlier [7] that most other cell types in the brain contain much lower levels of deletions. Therefore, the presence of other cells results in effective “dilution” of deletions. Second, as we argued above, the estimates in this study are only lower estimates of the true level of deletions in the human substantia nigra. It should be noted that in our calculations we assumed that relative dopamine neuronal density is similar in the mouse and humans. If, for example, the relative number of dopaminergic neurons (compared to other cell types presumably not carrying mtDNA deletions) in the mouse is lower than in humans, then higher effective “dilution” of their mitochondrial DNA might be responsible in part for the apparent lower fraction of deletions in the mouse. While visually cell densities appear to be similar in mouse and humans, it would be important to perform a detailed cell-by-cell investigation in the mouse and compare the results to the same in humans [7] to fully resolve this issue. Acknowledgments This work has been supported in part by NIA grant AG019787 to KK and by NINDS grant NS058988 to DKS and visiting scholar fellowships from the Chinese Government to XG and DY. Human brain tissue was provided by the Queen Square Brain Bank for Neurological Disorders, Institute of Neurology, University College, London. We thank Tanya Leibman for helping in the experiments.

1162

X. Guo et al. / Biochimica et Biophysica Acta 1797 (2010) 1159–1162

References [1] K. Khrapko, J. Vijg, Mitochondrial DNA mutations and aging: devils in the details? Trends Genet. 25 (2009) 91–98. [2] L.A. Marcelino, P.C. Andre, K. Khrapko, H.A. Coller, J. Griffith, W.G. Thilly, Chemically induced mutations in mitochondrial DNA of human cells: mutational spectrum of N-methyl-N'-nitro-N-nitrosoguanidine, Cancer Res. 58 (1998) 2857–2862. [3] H.A. Coller, N.D. Bodyak, K. Khrapko, Frequent intracellular clonal expansions of somatic mtDNA mutations: significance and mechanisms, Ann. N. Y. Acad. Sci. 959 (2002) 434–447. [4] E. Bua, J. Johnson, A. Herbst, B. Delong, D. McKenzie, S. Salamat, J.M. Aiken, Mitochondrial DNA-deletion mutations accumulate intracellularly to detrimental levels in aged human skeletal muscle fibers, Am. J. Hum. Genet. 79 (2006) 469–480. [5] A. Herbst, J.W. Pak, D. McKenzie, E. Bua, M. Bassiouni, J.M. Aiken, Accumulation of mitochondrial DNA deletion mutations in aged muscle fibers: evidence for a causal role in muscle fiber loss, J. Gerontol. A Biol. Sci. Med. Sci. 62 (2007) 235–245. [6] A. Bender, K.J. Krishnan, C.M. Morris, G.A. Taylor, A.K. Reeve, R.H. Perry, E. Jaros, J.S. Hersheson, J. Betts, T. Klopstock, R.W. Taylor, D.M. Turnbull, High levels of mitochondrial DNA deletions in substantia nigra neurons in aging and Parkinson disease, Nat. Genet. 38 (2006) 515–517. [7] Y. Kraytsberg, E. Kudryavtseva, A.C. McKee, C. Geula, N.W. Kowall, K. Khrapko, Mitochondrial DNA deletions are abundant and cause functional impairment in aged human substantia nigra neurons, Nat. Genet. 38 (2006) 518–520. [8] S.H. McKiernan, V.C. Tuen, K. Baldwin, J. Wanagat, A. Djamali, J.M. Aiken, Adultonset calorie restriction delays the accumulation of mitochondrial enzyme abnormalities in aging rat kidney tubular epithelial cells, Am. J. Physiol. Renal Physiol. 292 (2007) F1751–F1760. [9] D. Zhang, J.L. Mott, S.W. Chang, G. Denniger, Z. Feng, H.P. Zassenhaus, Construction of transgenic mice with tissue-specific acceleration of mitochondrial DNA mutagenesis, Genomics 69 (2000) 151–161. [10] A. Trifunovic, A. Wredenberg, M. Falkenberg, J.N. Spelbrink, A.T. Rovio, C.E. Bruder, Y.M. Bohlooly, S. Gidlof, A. Oldfors, R. Wibom, J. Tornell, H.T. Jacobs, N.G. Larsson,

[11]

[12]

[13]

[14]

[15]

[16]

[17] [18] [19] [20] [21]

[22]

Premature ageing in mice expressing defective mitochondrial DNA polymerase, Nature 429 (2004) 417–423. G.C. Kujoth, A. Hiona, T.D. Pugh, S. Someya, K. Panzer, S.E. Wohlgemuth, T. Hofer, A.Y. Seo, R. Sullivan, W.A. Jobling, J.D. Morrow, H. Van Remmen, J.M. Sedivy, T. Yamasoba, M. Tanokura, R. Weindruch, C. Leeuwenburgh, T.A. Prolla, Mitochondrial DNA mutations, oxidative stress, and apoptosis in mammalian aging, Science 309 (2005) 481–484. K.G. Bensch, W. Degraaf, P.A. Hansen, H.P. Zassenhaus, J.A. Corbett, A transgenic model to study the pathogenesis of somatic mtDNA mutation accumulation in beta-cells, Diabetes Obes. Metab. 9 (Suppl 2) (2007) 74–80. M. Vermulst, J. Wanagat, G.C. Kujoth, J.H. Bielas, P.S. Rabinovitch, T.A. Prolla, L.A. Loeb, DNA deletions and clonal mutations drive premature aging in mitochondrial mutator mice, Nat. Genet. 40 (2008) 392–394. K. Khrapko, Y. Kraytsberg, A.D.N.J. de Grey, J. Vijg, E.A. Schon, Does premature aging of the mtDNA mutator mouse prove that mtDNA mutations are involved in natural aging? Aging Cell 5 (2006) 279–282. N.D. Bodyak, E. Nekhaeva, J.Y. Wei, K. Khrapko, Quantification and sequencing of somatic deleted mtDNA in single cells: evidence for partially duplicated mtDNA in aged human tissues, Hum. Mol. Genet. 10 (2001) 17–24. Y. Kraytsberg, N. Bodyak, S. Myerow, A. Nicholas, K. Ebralidze, K. Khrapko, Collection of isolated cells for studying mitochondrial DNA mutations within individual cells, Methods Mol. Biol. 554 (2009) 315–327. Y. Kraytsberg, K. Khrapko, Single-molecule PCR: an artifact-free PCR approach for the analysis of somatic mutations, Expert Rev. Mol. Diagn. 5 (2005) 809–815. A. Goldstein, Biostatistics, Macmillian, New York, 1968. S. Pääbo, D.M. Irwin, A.C. Wilson, DNA damage promotes jumping between templates during enzymatic amplification, J. Biol. Chem. 265 (1990) 4718–4721. V.K. Viswanathan, K. Krcmarik, N.P. Cianciotto, Template secondary structure promotes polymerase jumping during PCR amplification, Biotechniques 27 (1999) 508–511. F.M. Yakes, B. Van Houten, Mitochondrial DNA damage is more extensive and persists longer than nuclear DNA damage in human cells following oxidative stress, Proc. Natl. Acad. Sci. U. S. A. 94 (1997) 514–519. S.T. Kim, J.H. Choi, D. Kim, O. Hwang, Increases in TH immunoreactivity, neuromelanin and degeneration in the substantia nigra of middle aged mice, Neurosci. Lett. 396 (2006) 263–268.