Content uploaded by Michael Hofreiter
Author content
All content in this area was uploaded by Michael Hofreiter
Content may be subject to copyright.
Content uploaded by Michael Hofreiter
Author content
All content in this area was uploaded by Michael Hofreiter
Content may be subject to copyright.
© 2005 Nature Publishing Group
Multiplex amplification of the mammoth
mitochondrial genome and the evolution
of Elephantidae
Johannes Krause
1
, Paul H. Dear
2
, Joshua L. Pollack
3
, Montgomery Slatkin
3
, Helen Spriggs
2
, Ian Barnes
4
,
Adrian M. Lister
4
, Ingo Ebersberger
5
, Svante Pa
¨
a
¨
bo
1
& Michael Hofreiter
1
In studying the genomes of extinct species, two principal limi-
tations are typically the small quantities of endogenous ancient
DNA and its degraded condition
1
, even though products of up to
1,600 base pairs (bp) have been amplified in rare cases
2
. Using
small overlapping polymerase chain reaction products, longer
stretches of sequences or even whole mitochondrial genomes
3,4
can be reconstructed, but this approach is limited by the number
of amplifications that can be performed from rare samples. Thus,
even from well-studied Pleistocene species such as mammoths,
ground sloths and cave bears, no DNA sequences of more than
about 1,000 bp have been reconstructed
5–7
. Here we report the
complete mitochondria l genome sequence of the Pleistocene
woolly mammoth Mammuthus primigenius.Weusedabout
200 mg of bone and a new approach that allows the simultaneous
retrieval of multiple sequences from small amounts of degraded
DNA. Our phylogenetic analyses show that the mammoth was
more closely related to the Asian than to the African elephant.
However, the divergence of mammoth, African and Asian ele-
phants occurred over a short time, corresponding to only about
7% of the total length of th e phylogenetic tree for the three
evolutionary lineages.
We have developed a multiplex polymerase chain reaction (PCR)
approach that in principle allows an entire mitochondrial genome to
be amplified from ancient DNA using just two initial amplifications.
This is accomplished by using primer pairs that define overlapping
DNA sequence fragments representing the complete mitochondrial
genome. These primer pairs are combined into two sets, each
containing every second primer pair. Each of these two sets is used
in a multiplex PCR amplification that requires the same amount of
ancient template DNA as would be used for amplifying a single target
sequence. Subsequently, the two primary amplifications are diluted
and used as templates in secondary PCR reactions, in which each
product is amplified individually.
To test whether this approach works for Pleistocene DNA, we
designed 46 primer pairs, which together we expected to amplify the
entire mitochondrial (mt) genome of the woolly mammoth (Fig. 1;
see Supplementary Information). Forty-one of the secondary ampli-
fications yielded products of the expected size (Fig. 2 and Sup-
plementary Information). Of the five PCR amplifications that failed,
four were successful in later attempts, suggesting that the initial
failure was due to the absence of even a single template molecule
in th e PCR amplification. The fif th PCR failed repeatedly, and
inspection of the sequences of the overlapping amplimers revealed
differences at the priming sites between the mammoth and elephant
sequence s that can account for the fai lure. New primers were
designed for this amplimer on the basis of the flanking mammoth
sequence. The sequences of all 46 amplimers were found to be similar
to (but distinct from) corresponding sequences of extant elephants.
To ensure the authenticity of the ancient DNA sequences
1
,
we repeated the primary and secondary amplifications and th e
sequencing of the amplimers (see Methods). In 17 of the amplimers,
we found differences at 1–5 nucleotide positions between the two
independent experiments, presumably due to cytosine deamination
of the template
8
(Supplementary Information). In these cases, a third
round of primary PCR amplifications was performed, and the
consensus of the three sets of sequences was inferred to represent
the correct sequence. In total, fourteen primary PCR amplifications,
corresponding to , 200 mg of mammoth bone, were required for
determination and confirmation of the entire mtDNA sequence.
To further confirm the reproducibility of the results, samples of the
bone were extracted, amplified and sequenced in laboratories in
Cambridge and London, UK, that did not have access to the initial
results obtained in Leipzig. In total, 5,024 bp of the mitochondrial
sequence were independently reproduced. The reproduced sequences
LETTERS
Figure 1 | Map of the mammoth mitochondrial genome. Circular genome
(yellow), showing the positions of the control region and the genes encoding
22 transfer RNAs (grey boxes), 2 ribosomal RNAs and 13 proteins. The
positions and relative lengths of the 46 amplification products used are
depicted in blue (first set) and red (second set).
1
Max Planck Institute for Evolutionary Anthropology, Deutscher Platz 6, D-04103 Leipzig, Germany.
2
MRC Laboratory of Molecular Biology, Hills Road, Cambridge CB2 2QH, UK.
3
Department of Integrative Biology, University of California, Berkeley, California 94720-3140, USA.
4
Department of Biology, University College London, London WC1E 6BT, UK.
5
WE Informatik, Heinrich Heine Universita
¨
t, Universita
¨
tsstr. 1, D-40225 Du¨sseldorf, Germany.
doi:10.1038/nature04432
1
© 2005 Nature Publishing Group
were identical to those initially determined, except for one position
determined from a single PCR reaction in London that shows a
thymine (T), when the sequence determined from several PCR
amplifications in Leipzig shows a cytosine (C). This discrepancy is
most likely due to cytosine deamination of the template DNA
8
.
One concern when determining mtDNA sequences is that nuclear
insertions of mtDNA fragments might be mistaken for the organelle
copies
9
. This problem can be particularly severe in some species,
including elephants
10
. However, the sequence we have determined is
circular and therefore cannot represent a nuclear insertion. Although
it is conceivable that some of the amplimers could have arisen from
nuclear insertions, no mismatches were observed in a total of
2,030 overlapping base pairs between adjacent amplimers. We thus
conclude that the results obtained indeed represent the complete
mtDNA sequence of the mammoth from Berelekh.
The phylogenetic relationship of the mammoth to its closest living
relatives, the African and Asian e lephants, has un til now been
unresolved. Numerous studies using small mitochondrial and
nuclear DNA
11
sequences have variously suggested that mammoths
were the sister group of African
12–14
or of Asian elephants
7,11,15
.It
has been argued that the failure to resolve this phylogenetic relation-
ship is due to the small amount of DNA sequence information
available
13
, because long mtDNA sequences are often necessary to
obtain a robust phylogenetic tree
16
. To address this question, we
aligned the entire 16,770-bp mammoth mtDNA with mtDNA from
the two elephant species and two potential outgroup species, dugong
and hyrax. Phylogenetic trees identified either the Asian or African
elephant as the sister taxon of the mammoth, depending on the
outgroup and tree reconstruction method used (Supple mentary
Table S2).
Two lines of evidence indicate that this lack of resolution results
from a too-distant relationship of elephants and mammoths to the
two outgroup species, both of which diverged from the proboscidean
lineage at least 65 million years ago
17
.First,abouthalfofthe
third-codon positions in the protein-coding sequences differ when
the mammoth or either elephant species is compared to dugong or
hyrax. Second, the ratio of transitions to transversions among
elephants and mammoth is 20:1, but is 2:1 between these species
and dugong or hyrax. Thus, the phylogenetic signal is blurred by
multiple substitutions when dugong or hyrax is included in the
analysis
18
.
As no closer outgroup species exists, we restricted our analysis to
the mammoth and the two elephant species using three different
methods that assume a molecular clock (that is, a uniform rate of
nucleotide substitutions in all lineages). We first tested whether the
mtDNAs of the three species evolved at equal rates using a clock test
that uses only three taxa and treats transitions and transversions
separately
19
, as well as by a likelihood ratio test. The assumption of a
molecular clock among the mammoth and two elephant mtDNAs
could not be rejected (P ¼ 0.61). We therefore estimated a maximum-
likelihood tree under the assumption of a molecular clock, applying
midpoint rooting. This analysis supports a sister-group relationship
between mammoth and Asian elephant with a 97% bootstrap value.
We then calculated the likelihood of the data assuming a star-like tree
topology. A likelihood ratio test reveals that the more complex
model, resulting in the resolved sequence tree, explains the data
significantly better, and thus rejects the simpler model of a star-like
tree (P , 0.01). Notably, the tree that places the Asian elephant and
mammoth as the closes t relatives to each other h as a post erior
probability of 99.8%. This indicates that the two alternative sce-
narios, each having posterior probabilities of ,0.1%, can be neg-
lected. Finally, we used Felsenstein’s
20
method for testing whether the
most parsimonious tree is significantly different from a tree with a
‘trifurcation’ (Fig. 3a). Again, we rejected the null hypothesis of a
trifurcation (P ¼ 0.032 and P ¼ 0.035 for the C and S statistics,
respectively; see Methods). Thus, the monophyly of the mammoth
and the Asian elephant, to the exclusion of the African elephant, is
significantly supported.
In this context, two points should be noted. First, the length of the
internal branch leading to the mammoth and Asian elephant mtDNA
is onl y 7.3% of that leading to the African elephant (Fig. 3b).
Paleontological data
21
suggest a divergence of Asian and African
elephants and mammoths about six million years ago in Africa. This
Figure 2 | Principle of the multiplex approach and typical results.
Amplification results for the mammoth ancient DNA extract, with 15 of 16
amplifications yielding positive results (a). Negative controls for the same 16
primer pairs are shown in b. Gel lanes marked with S contain molecular size
markers. The size of the amplification products ranges from 306 to 554 bp.
Figure 3 | Comparison of three-species trees. Comparisons based on the
complete mtDNA sequences for human, chimpanzee, gorilla, mammoth and
African and Asian elephants. a, Absolute number of phylogenetically
informative positions on which the three-species test is based. b, Relative
length of the internal branch in the two groups of species for clock-like
phylogenetic trees using the substitution data from a and rooted by
midpoint rooting. Note the differences in the absolute numbers of
substitutions and relative branch length of the intern al branch between the
two species groups (see also Supplementary Information).
LETTERS NATURE
2
© 2005 Nature Publishing Group
date implies that the divergence between the mammoth and Asian
elephant took place only 440,000 years after the divergence of the
African elephant. Second, this time is short enough that polymorph-
isms in the ancestral species may have persisted between the two
speciation events
22
, as has been observed for humans, chimpanzees
and gorillas
23
. As the probability of such events increas es with
increasing effective population size
24
, mtDNA is more likely to reflect
species phylogeny than autosomal genes that have an approximately
four times larger effective population size. The sister-group relation-
ship of the mtDNAs of the mammoth and the Asian elephant
therefore suggests that these two species shared a common ancestor
after their separation from the African elephant (Supplementary
Information). Thus, any morphological similarit ies between the
African elephant and mammoth
13
are likely to be either ancestral
or convergent. However, owing to the short time between the two
divergence events, this is also likely to be true for many morpho-
logical similarities between the Asian elephant and mammoth.
In summary, the multiplex approach described here allows the
retrieval of complete mtDNA sequences from Late Pleistocene fossils,
enabling robust phylogenetic inferences to be made. The small
amounts of sample material necessary allow complete mtDNA
sequences to be determined even from very valuable specimens.
This approach makes it possible to sample the mitochondrial
genomes of Pleistocene animals as widely as those of extant species,
and provides the opportunity to answer detailed questions about the
structure and history of extinct populations.
METHODS
Detailed descriptions of the methods used to date the mammoth bone and
extract, amplify and sequence its DNA, together with the phylogenetic methods
used, are provided in the Supplementary Information.
DNA extraction and amplification. For DNA extraction we used 747 mg bone
powder from a mammoth bone found in Berelekh, Yakutia (Supplementary
Information), yielding 70
m
l of extract. We used 1.5
m
l of the DNA extract for
each multiplex PCR reaction, in a total volume of 20
m
l. After 27 cycles of PCR,
the appropriate multiplex product was diluted and 0.625% of this material was
used for each of 46 individual amplifications. Amplification products of the
correct size were cloned using the TOPO TA cloning kit (Invitrogen), and a
minimum of three clones were sequenced on an ABI3730 capillary sequencer
(Applied Biosystems).
We designed new primers for those fragments that gave only weak (or no)
amplification products in the first attempts and for which the sequences of
adjacent fragments showed differences from the elephant sequence in the primer
sites. These new primers, together with the previously successful primers, were
used to amplify the remaining segments of the mammoth mtDNA and to
replicate all positions at least twice from independent primary amplifications
(see also Supplementary Information). Extraction protocols and PCR con-
ditions for the laboratories in Cambridge and London are available in the
Supplementary Information.
Phylogenetic analyses. We initially aligned the mtDNA sequence of the
mammoth to the corresponding sequences from Asian (Elephas maximus) and
African (Loxodonta africana) elephant and dugong (Dugong dugon), excluding
the control region in all analyses. Applying Modeltest
25
on this alignment, we
obtained a general time-reversible substitution model with gamma-distributed
substitution rates across sites. We estimated neighbour-joining and maximum-
likelihood trees correcting for multiple substitutions using this substitution
model and maximum parsimony trees using the program package PAUP*
26
, and
bayesian trees using MrBayes
27
. Depending on the tree-building method and
outgroup, we recovered both the Asian and African elephant as sister taxa of the
mammoth. We therefore added the mtDNA sequence from hyrax (Procavia
capensis) as additional outgroup in an attempt to gain resolution. Again, we
could not resolve the phylogeny (Supplementary Table S2).
Our analyses indicate that both dugong and hyrax are too distantly related to
the mammoth and African and Asian elephants to resolve the phylogeny.
However, by assuming a molecular clock, rooted phylogenies can be obtained
without outgroups. We first tested the clock assumption by applying a simple test
that uses only three taxa
19
. We also performed a clock test based on a likelihood
ratio test with the program TREE-PUZZLE
28
, using the Hasegawa–Kishino–
Yano (HKY) model of DNA sequence evolution
29
and a gamma-distribution
using eight gamma rate categories to model substitution rate heterogeneity
among sites. Neither of the two tests could reject the clock assumption. We
subsequently used a parsimony method developed to estimate the phylogenetic
relationship of three sequences that evolve under a molecular clock
20
and a
likelihood ratio test (Supplementary Information) in order to resolve the
relationship between mammoth, African and Asian elephant.
The parsimony method uses two statistics, C and S in Felsenstein’s notation
20
.
Both are based on the number of positions at which one of the three sequences
differs from the other two (which are identical to each other). C is the highest
number of such positions obtained from one of the three possible comparisons
n
1
, n
2
and n
3
(n
1
, species A differs from species B and C; n
2
, B differs from A and
C; n
3
, C differs from A and B). S is described by n
1
2 n
2
, given that n
1
and n
2
are
the largest and second largest numbers obtained for n, respectively. Excluding the
control region, 403 (n
1
) positions support a sister-group relationship between
the mammoth and Asian elephant, 357 (n
2
) positions support a sister-group
relationship between the mammoth and the African elephant, and 339 (n
3
)
positions support a sister-group relationship between the two elephant species.
This yields a C statistic of 403 (C ¼ n
1
) and an S statistic of 46 (S ¼ n
1
2 n
2
)
20
.
On the basis of a Monte Carlo randomization test (10
6
simulations), the
monophyly of the mammoth and the Asian elephant, to the exclusion of the
African elephant, is significantly supported by both the C and the S statistic
(P ¼ 0.032 and P ¼ 0.035, respectively).
To test whether the phylogenetic signals in the data are strong enough to
warrant resolution of the sequence tree under a likelihood framework, we
proceeded as follows. T he likelihood of the data was assessed under two
alternative models: (1) a simple model with only a single free parameter,
corresponding to a star-like tree topology, and (2) a more complex model
with two free parameters, resembling the resolved tree topology. A subsequent
likelihood ratio test revealed that the more complex model, yielding the resolved
sequence tree, explains the data significantly better, and thus rejects the simpler
model of a star-like tree (P , 0.01).
Received 1 July; accepted 14 November 2005.
Published online 18 December 2005.
1. Pa
¨
a
¨
bo, S. et al. Genetic analyses from ancient DNA. Annu. Rev. Genet. 38,
645–-679 (2004).
2. Lambert, D. M. et al. Rates of evolution in ancient DNA from Adelie penguins.
Science 295, 2270–-2273 (2002).
3. Cooper, A. et al. Complete mitochondrial genome sequences of two extinct
moas clarify ratite evolution. Nature 409, 704–-707 (2001).
4. Haddrath, O. & Baker, A. J. Complete mitochondrial DNA genome sequences
of extinct birds: ratite phylogenetics and the vicariance biogeography
hypothesis. Proc. R. Soc. Lond. B 268, 939–-945 (2001).
5. Ho
¨
ss, M., Dilling, A., Currant, A. & Pa
¨
a
¨
bo, S. Molecular phylogeny of the
extinct ground sloth Mylodon darwinii. Proc. Natl Acad. Sci. USA 93, 181–-185
(1996).
6. Loreille, O. et al. Ancient DNA analysis reveals divergence of the cave bear,
Ursus spelaeus, and brown bear, Ursus arctos, lineages. Curr. Biol. 11, 200–-203
(2001).
7. Yang, H., Golenberg, E. M. & Shoshani, J. Phylogenetic resolution within the
Elephantidae using fossil DNA sequences from the American mastodon
(Mammut americanum) as an outgroup. Proc. Natl Acad. Sci. USA 93, 1190–-1194
(1996).
8. Hofreiter, M., Jaenicke, V., Serre, D., von Haeseler, A. & Pa
¨
a
¨
bo, S. DNA
sequences from multiple amplifications reveal artifacts induced by cytosine
deamination in ancient DNA. Nucleic Acids Res. 29, 4793–-4799 (2001).
9. Bensasson, D., Zhang, D. X., Hartl, D. L. & Hewitt, G. M. Mitochondrial
pseudogenes: evolution’s misplaced witnesses. Trends Ecol. Evol. 16, 314–-321
(2001).
10. Greenwood, A. D. & Pa
¨
a
¨
bo, S. Nuclear insertion sequences of mitochondrial
DNA predominate in hair but not in blood of elephants. Mol. Ecol. 8, 133–-137
(1999).
11. Greenwood, A., Capelli, C., Possnert, G. & Pa
¨
a
¨
bo, S. Nuclear DNA sequences
from late Pleistocene megafauna. Mol. Biol. Evol. 16, 1466–-1473 (1999).
12. Noro, M., Masuda, R., Dubrovo, I. A., Yoshida, M. C. & Kato, M. Molecular
phylogenetic inference of the woolly mammoth Mammuthus primigenius, based
on complete sequences of mitochondrial cytochrome b and 12S ribosomal RNA
genes. J. Mol. Evol. 46, 314–-326 (1998).
13. Thomas, M. G., Hagelberg, E., Jone, H. B., Yang, Z. & Lister, A. M. Molecular
and morphological evidence on the phylogeny of the Elephantidae. Proc. R. Soc.
Lond. B 267, 2493–-2500 (2000).
14. Debruyne, R., Barriel, V. & Tassy, P. Mitochondrial cytochrome b of the
Lyakhov mammoth (Proboscidea, Mammalia): new data and phylogenetic
analyses of Elephantidae. Mol. Phylogenet. Evol. 26, 421–-434 (2003).
15. Ozawa, T., Hayashi, S. & Mikhelson, V. M. Phylogenetic position of mammoth
and Steller’s sea cow within Tethytheria demonstrated by mitochondrial DNA
sequences. J. Mol. Evol. 44, 406–-413 (1997).
16. Cummings, M. P., Otto, S. P. & Wakeley, J. Sampling properties of DNA sequence
data in phylogenetic analysis. Mol. Biol. Evol. 12, 814–-822 (1995).
NATURE LETTERS
3
© 2005 Nature Publishing Group
17. Shoshani, J. Understanding proboscidean evolution: a formidable task. Trends.
Ecol. Evol. 13, 480–-487 (1998).
18. Felsenstein, J. Inferring Phylogenies 196–-221 (Sinauer Associates, Sunderland,
2004).
19. Tajima, F. Simple methods for testing the molecular evolutionary clock
hypothesis. Genetics 135, 599–-607 (1993).
20. Felsenstein, J. Confidence-limits on phylogenies with a molecular clock. Syst.
Zool. 34, 152–-161 (1985).
21. Tassy, P. in Lothagam: The Dawn of Humanity in Eastern Africa (eds Leakey, M. G. &
Harris, J. M.) 331–-358 (Columbia Univ. Press, New York, 2003).
22. Nei, M. in Evolutionary Perspectives and the New Genetics (eds Gershowitz, H.,
Rucknagel, D. L. & Tashian, R. E.) 133–-147 (Alan R. Liss, New York, 1986).
23. Chen, F. C. & Li, W. H. Genomic divergences between humans and other
hominoids and the effective population size of the common ancestor of
humans and chimpanzees. Am. J. Hum. Genet. 68, 444–-456 (2001).
24. Tajima, F. Evolutionary relationship of DNA sequences in finite populations.
Genetics 105, 437–-460 (1983).
25. Posada, D. & Crandall, K. A. MODELTEST: testing the model of DNA
substitution. Bioinformatics 14, 817–-818 (1998).
26. Swofford, D. L. PAUP*. Phylogenetic Analysis Using Parsimony (*and Other
Methods) (Sinauer Associates, Sunderland, 2003).
27. Ronquist, F. & Huelsenbeck, J. P. MRBAYES 3: Bayesian phylogenetic inference
under mixed models. Bioinformatics 19, 1572–-1574 (2003).
28. Schmidt, H. A., Strimmer, K., Vingron, M. & Haeseler, A. TREE-PUZZLE:
Maximum likelihood phylogenetic analysis using quartets and parallel
computing. Bioinformatics 18, 502–-504 (2002).
29. Hasegawa, M., Kishino, H. & Yano, T. Dating of the human-ape splitting by a
molecular clock of mitochondrial DNA. J. Mol. Evol. 22, 160–-174 (1985).
Supplementary Information is linked to the online version of the paper at
www.nature.com/nature.
Acknowledgements We thank the members of our laboratories for discussions
and support, G. Khlopatchev for providing the mammoth bone and
K. Finstermeier for help with figure design. This work was supported by the Max
Planck Society. J.L.P. and M.S. were supported by an NIH grant (to M.S.).
Author Information The complete mammoth mitochondrial DNA sequence has
been deposited in GenBank under accession number DQ188829. Reprints and
permissions information is available at npg.nature.com/reprintsandpermissions.
The authors declare no competing financial interests. Correspondence and
requests for materials should be addressed to M.H. (hofreiter@eva.mpg.de).
LETTERS NATURE
4
