Origin and Evolution of Tandem Repeats in the Mitochondrial DNA Control Region of Shrikes (Lanius Spp.)

J Mol Evol (2004) 59:250–257 DOI: 10.1007/s00239-004-2619-6

Origin and Evolution of Tandem Repeats in the Mitochondrial DNA Control Region of Shrikes (Lanius spp.)

Nicholas I. Mundy,1 Andreas J. Helbig2

1 Department of Zoology, University of Cambridge, Downing Street, Cambridge, CB2 3EJ, UK 2 Vogelwarte Hiddensee, Universitat Greifswald, D-18565 Kloster/Hiddensee, Germany

Received: 25 September 2003 / Accepted: 12 February 2004 [Reviewing Editor: Martin Kreitman]

Abstract. The origin and evolution of a 128-bp number of taxa (reviewed in Hoelzel 1993; Rand 1993; tandem repeat in the mtDNA control region of Lunt et al. 1998). In birds, control region tandem re- shrikes (Lanius: Aves) were investigated. The tandem peats have been documented at the sequence level in repeat is present in only two species, L. excubitor and waders, gulls, auks, shrikes, penguins, and eagles L. ludovicianus. In contrast to the variation in repeat (Wenink et al. 1994; Berg et al. 1995; Mundy et al. number in L. ludovicianus, all individuals of three 1996; Ritchie and Lambert 2000; Va¨ li 2002). These subspecies of L. excubitor had three repeats. Com- repeats can be categorized according to size of the parative analysis suggests that a short direct repeat, basic repeat unit and the position of the array within and a secondary structure including the tandem repeat the control region (Hoelzel 1993). Interest in repeats and a downstream inverted repeat, may be important has focused on the light they shed on mechanisms of in the origin of the tandem repeat by slipped-strand mtDNA mutation and turnover and on the practical mispairing and its subsequent turnover. Homogeni- use of repeats as markers in population genetics. zation of repeat sequences is most simply explained by The origin of mtDNA tandem repeats is generally expansion and contraction of the repeat array. Sur- thought to involve slipped-strand mispairing during prisingly, mtDNA sequences from L. excubitor were mtDNA replication (Levinson and Gutman 1987; found to be paraphyletic with respect to L. ludovici- Buroker et al. 1990; Hayasaka et al. 1991; Madsen anus. These results show the utility of a comparative et al. 1993; Fumagalli et al. 1996; Broughton and analysis for insights into the evolutionary dynamics of Dowling 1994; Wilkinson et al. 1997). Other mecha- mtDNA tandem repeats. nisms for repeat generation, such as intermolecular recombination, are considered rare (reviewed in Ro- Key words: Mitochondrial DNA — Control region kas et al. 2003). However, the details of how slipped- — Tandem repeats strand mispairing works are poorly understood and there has been little work on why repeats form in some lineages and not others. As well as being of interest to control region repeats themselves, such mechanisms are also relevant to other examples of Introduction mtDNA turnover. For example, one favored hy- pothesis for the generation of mitochondrial gene Tandemly repeated DNA in the control region of the order rearrangements involves a large duplication mitochondrial DNA is reported from an ever-growing event followed by gene loss (Macey et al. 1997). In previous work we demonstrated that there is a 128-bp tandem repeat in the control region of the Correspondence to: Nicholas I. Mundy; email: [email protected] loggerhead shrike (Lanius ludovicianus) (Mundy et al. 251

1996) (Fig. 1A). All individuals of this species either have two repeats, have three repeats, or are heteroplasmic for two and three repeats. In contrast, the one other species of shrike examined (L. collaris) did not have the repeat. The 128-bp tandem repeat in the loggerhead shrike falls into the relatively small category of large repeats that are present in a 3¢ position in the light strand of the control region. Other examples in this category include repeat arrays in rabbits (Mignotte et al. 1990), macaques (Hayasaka et al. 1991), minnows (Broughton and Dowling 1994), redfish (Bentzen et al. 1998), and crocodiles (Ray and Densmore 2003). An important feature of these tandem repeat arrays is that since they presumably occur Fig. 1. A Structure of mtDNA in species with (L. ludovicianus; top) and without (L. collaris; bottom) the 128-bp tandem repeat. 3¢ to the origin of heavy strand replication (OH) they CSB-1 is conserved sequence block 1. DLL2 and FTPH2 are the do not form part of the D-loop, and thus mechanisms primers used in this study. B Model of tandem repeat formation of slipped-strand mispairing that involve the D-loop (Mundy et al. 1996). The light strand is the lower strand. Toward the end of heavy-strand replication, the leading strand spontane- (e.g., Buroker et al. 1990) are not relevant to their ously denatures (top). Stem-loop formation by the inverted repeat origin and expansion/contraction. occurs and the strand reanneals onto the direct repeat sequence Other sequence motifs in and around the tandem (middle). Continued replication generates the tandem repeat (bot- repeat led to the proposal of a model of how the tom). tandem repeat arose by slipped-strand mispairing in shrikes, toward the end of heavy strand replication collected during banding procedures. Abbreviations follow genus, (Mundy et al. 1996; Fig. 1B), which was a variation species, and subspecies for species with more than one subspecies (e.g., Lei for Lanius excubitor invictus) and genus, species for spe- of a model originally proposed by Broughton and cies with no subspecies (e.g., Lco for Lanius collaris). The single Dowling (1994). In the model, short direct repeats exception involves L. ludovicianus mearnsi and L. l. gambeli, whose were proposed to be important for reannealing fol- haplotypes are labeled Llu1, Llu2, and Llu3, to be consistent with lowing strand slippage, and an inverted repeat 3¢ to nomenclature used in the previous study (Mundy et al. 1996). the tandemly repeated region was proposed to form a stable stem loop structure following slippage that Laboratory Methods stabilized the slipped strand to promote reannealing to a different position. However, data from two All laboratory procedures closely followed those used previously species were insufficient to critically appraise this (Mundy et al. 1996). Total genomic DNA was extracted from single model. feather shafts using Chelex and from blood using a salting-out procedure. Primers used to amplify the region containing tandem Key questions to test the model are: In the line- repeats were DLL2 (5¢ ATGCACTTTTACCCCATTCATGGTGG age(s) in which the tandem repeat originated, were 3¢) and FTPH2 (5¢ CCATCTTGACATCTTCAGTGCCATGC 3¢) the direct repeats present? Were the inverted repeats (see Fig. 1A for positions). Polymerase chain reaction (PCR) am- present? and Were predicted free energies of second- plifications were performed in a Hybaid thermal cycler in a 25-ll ary structures comparatively high? Absence of any of total volume containing 0.5–1.0 units Taq polymerase (Perkin–El- mer), 1· PCR buffer, a 50 lM concentration each dNTP, 1.5 mM these would lead to rejection of the model. In the MgCl2, and 25 lg bovine serum albumin (Fraction V, Sigma), and present study we have obtained an expanded dataset 10–100 ng template DNA. Cycling parameters were as follows: 1 · of control region sequences from 8 species and 14 94C, 3 min; 30–40 · 94C, 30 s; 50–60C, 60 s, 72C, 90 s; and 1 · subspecies of shrikes, including the great grey shrike 72C, 10 min. Different samples required different annealing tem- (L. excubitor), which is generally considered to be the peratures for optimization. Direct double-stranded sequencing of PCR products was performed with Sequenase (USB) and 35S- sister species of L. ludovicianus. The results show that dATP. Products of sequencing reactions were separated on 8% the tandem repeat is only present in L. excubitor and polyacrylamide gels (Long Ranger; J. T. Baker). Dried gels were L. ludovicianus, support the model of repeat origi- exposed to X-ray film (X-Omat; Kodak Eastman) for 1–7 days. nation, and shed light on mechanisms of repeat Sequences were read by eye. turnover. DNA Sequence Analysis

Materials and Methods Sequences were aligned using Clustal X (Higgins and Sharp 1989). A control region sequence from a rook (Corvus frugilegus) was used Samples as the closest available outgroup (GenBank Accession No. Y18522 [Ha¨ rlid and Arnason 1999]). Phylogenetic analyses were performed Species and subspecies sampled in the study are shown in Table 1. in PAUP* (Swoﬀord 1999). Neighbor-joining trees were con- Samples from L. ludovicianus and L. excubitor invictus were structed using HKY distances. Branch support in neighbor-joining plucked feathers; those from other species were blood samples and parsimony was assessed using 1000 bootstrap replicates. 252

Table 1. Shrike taxa sampled in this study

Taxon Origin Lab ID

Loggerhead shrike Lanius ludovicianus mearnsi San Clemente I., USA Haplotypes Llu1, Llu2 (Mundy et al. 1996)a L. l. gambeli California, USA Haplotypes Llu2, Llu3 (Mundy et al. 1996)a L. l. excubitorides Alberta, Canada Lle6 Great grey shrike L. e. excubitor Germany Lee1, Lee2, Lee3 Northern shrike L. e. invictus Oregon, USA Lei1, Lei2, Southern grey shrike L. e. meridionalis Spain Lem1, Lem2, Lem3 Fiscal shrike L. collaris Guinea Lco1 (Mundy et al. 1996) Woodchat shrike L. s. senator Italy Lss1, Lss3 L. s. badius Italy Lsb4 L. s. niloticus Israel Lsn5 Red-backed shrike L. c. collurio Italy Lcc3, Lcc6 Isabelline shrike L. isabellinus Mongolia Lis1, Lis2 Masked shrike L. n. nubicus Israel Lnn1 Long-tailed shrike L. s. schach Hong Kong Lsc1, Lsc2 a Haplotypes Llu1, Llu2, and Llu3 were obtained in the previous study (Mundy et al. 1996), from numerous individuals of L. l. mearnsi and L. l. gambeli, and these designations are retained here for ease of comparison.

Maximum likelihood analyses were performed using an HKY plasmy was detected in this species. The 3¢ copy of the model of sequence evolution. Ancestral sequences of the L. ludo- tandem repeat was 128 bp in length in all cases in L. vicianus/L. excubitor lineage were inferred by parsimony. Free excubitor, in contrast to L. ludovicianus, in which the binding energies of single-stranded DNA secondary structures were estimated in RNA structure 3.7 (Mathews et al. 1999). Sequences 3¢ copy of the repeat array always has a 2-bp deletion. reported here have been deposited in GenBank (accession Nos. The single individual of the new subspecies of L. lu- AY599851–AY599870). dovicianus (L. l. excubitorides) to be sequenced had Following the standard convention, the terms upstream, down- two copies of the repeat, and its tandem repeats were stream, 5¢, and 3¢ are used throughout with reference to the identical to those of sequence Llu3 previously ob- mtDNA light strand. Tandem repeats are numbered in the 5¢–3¢ direction. tained from both L. l. mearnsi and L. l. gambeli. In all species of shrike which did not contain the tandem repeat, the sequence that forms the tandem repeat is Results present. Direct repeats, occurring at the beginning of the Taxonomic Distribution of Tandem Repeats and duplicated region and immediately after it, are pre- Other Sequence Motifs sent to some extent in all species (Fig. 2A). However, the strongest similarity between these sequences oc- The segment of control region containing the tandem curs in L. ludovicianus and L. excubitor. In particular, repeat, from the conserved sequence block 1 (CSB-1) Phe these two species were the only ones to have a perfect to the boundary with the tRNA gene, was se- copy of a CATTTT hexamer. quenced from 6 additional species and 11 additional Inverted repeats downstream of the tandem repeat subspecies of shrike, leading to a total dataset from 8 region are present in all species, but there is again species and 14 subspecies (Table 1). Tandem repeats variation among species (Fig. 2B). The length of were present in two species of shrike—Lanius ludo- perfect inverted repeats varies from 22 bp in L. lu- vicianus and L. excubitor—and absent in all other dovicianus to 16 bp in L. collurio. shrike species examined (Table 2). Tandem repeats were also absent from 17 other genera of Passerines (Ruokenen and Kvist 2002). The tandem repeats in Phylogenetic Reconstructions L. ludovicianus and L. excubitor are of identical length (128 bp) and occur at the same position. All In order to exclude the possibility that atypical evo- eight individuals from three subspecies of L. excubi- lution of the tandem repeat region would bias results, tor had three copies of the repeat, and no hetero- phylogenetic analysis was ﬁrst performed on a 203-bp 253

Table 2. Distribution of tandem repeats and direct repeats and predicted binding energies of single-stranded DNA secondary structuresa

Minimum free energy (kJ/mol) of

Presence Tandem Presence of tandem of CATTTT Inverted repart Tandem repeat repeat to inverted Taxon repeat (No. of repeats) direct repeat (66–70 bp) (128 bp) repeat (218–231 bp)

L. ludovicianus (Llu1, Llu2, +(2, 3, 2 + 3) + )14.9 )4.0 )24.7 Llu3, Llu6) L. excubitor Lem1, Lem2, Lem3, Lei2 +(3) + )15.1 )9.2 )25.0 Lei1 +(3) + )7.3 )9.2 )21.8 Lee1, Lee3 +(3) + )8.5 )9.4 )19.5 Lee2 +(3) + )11.2 )9.4 )17.6 L. ludovicianus/ ±? + )8.5 )4.2 )19.6 L. excubitor ancestor L. collaris (Lco1) )))7.0 )5.0 )13.9 L. senator (Lsb4, )))3.8 )3.1 )11.8 Lsn5, Lss1, Lss3) L. collurio (Lcc3, Lcc6) )))3.6 )3.1 )7.4 L. nubicus (Lnn1) )))7.3 )4.5 )12.3 L. schach (Lsc1, Lsc2) )))8.5 )3.3 )15.1 L. isabellinus Lis1 )))8.6 )3.1 )12.7 Lis2 )))3.6 )3.1 )7.4 a Taxa with identical sequences are grouped together.

Fig. 2. Control region sequence alignments. A Direct repeats. B Inverted repeats. alignment excluding the tandem repeat region. L. lu- L. ludovicianus/L. excubitor clade, L. niloticus, dovicianus and L. excubitor sequences consistently L. collaris, and L. senator. formed a monophyletic clade in neighbor-joining, In order to examine the evolutionary history of the parsimony, and maximum likelihood analyses that tandem repeat region itself, a separate analysis was had strong bootstrap support (Fig. 3A). Interesting- performed on a 128-bp alignment of the tandem re- ly, there was strong evidence for a paraphyletic repeat region (Fig. 3B), with each repeat in the tandem lationship between L. excubitor and L. ludovicianus, array, where present, treated separately. Results with L. e. excubitor sequences forming a well-sup- among species without the tandem array are con- ported sister clade to a clade containing L. ludovici- cordant with those obtained with the 203-bp align- anus, L. e. invictus,andL. e. meridionalis. The ment. For tandem repeats in L. ludovicianus and grouping of L. e. invictus and L. e. meridionalis is L. excubitor, many relationships are poorly resolved, strongly supported. Relationships among L. schach, but some structure is nevertheless present. All repeats L. isabellinus, and L. collurio are well resolved but in L. e. excubitor form a well-supported clade, with there is poor resolution of basal relationships be- repeats 2 and 3 in this taxon forming a clade that has tween the clade containing these three species, the a sister relationship to a clade containing repeat 254

Fig. 3. mtDNA phylogeny. Trees shown were obtained using sequence. B Results from 128-bp alignment of tandem repeat se- maximum likelihood. Numbers above and below branches repre- quence and homologous sequences in species lacking the tandem sent percentage bootstrap support from neighbor-joining and repeat. Repeats are labeled as follows: L. excubitor (Lei, Lem, parsimony analyses, respectively. A Results from 203-bp alignment Lee)—r1–r3, repeat 1–repeat 3. L. ludovicianus (Llu, Lle)—p1 and not containing the tandem repeat, rooted using Corvus frugilegus p2, perfect repeats 1 and 2; im, imperfect (3¢) repeat. number 1. Tandem repeats from L. e. invictus and verted repeat predicted in the ancestor of L. excubitor L. e. meridionalis cluster together in two separate and L. ludovicianus ()8.5 kJ/mol) is toward the top clades. Tandem repeats from L. ludovicianus occupy end of this latter range. basal positions but there is poor support for their For the tandemly repeated region itself, the pre- relationships to each other and to other clades. dicted free binding energies for L. excubitor ()9.2 to )9.4 kJ/mol) are the highest. The values for L. lu- ) Binding Energies of Potential Secondary Structures dovicianus ( 4.0 kJ/mol) and the L. ludovicianus/L. ) During Tandem Repeat Formation and/or Turnover excubitor ancestor ( 4.2 kJ/mol) fall within the range for taxa lacking the tandem repeat ()3.1 to )5.0 kJ/ An important part of the model previously proposed mol). was the potential of inverted repeats and tandem re- For the sequence from the tandem repeat region peats to form stable secondary structures. The through to the inverted repeat region there is a clear strength of binding of secondary structures formed by separation between species with and without the inverted and/or tandem repeats was estimated for tandem repeat. The range of free binding energies for ) ) each taxon and for the reconstructed sequence of L. ludovicianus and L. excubitor ( 17.6 to 24.7 kJ/ ) the ancestor of L. ludovicianus and L. excubitor mol) is greater than that for the other taxa ( 7.4 to ) (Table 2). This ancestral sequence was unambigu- 15.1 kJ/mol), and the predicted L. ludovicianus/L. ) ously reconstructed using parsimony. excubitor ancestor ( 19.6 kJ/mol) is clearly within the There is large variation in predicted free binding range of L. ludovicianus and L. excubitor. energy of the inverted repeat region among taxa ranging from )3.6 kJ/mol in one L. isabellinus indi- Divergence in Tandem Repeat Arrays vidual (Lis2) to )15.1 kJ/mol for some L. excubitor sequences. The range of binding energy in taxa pos- Sequence comparisons among the diﬀerent repeats sessing the tandem repeat is higher than, but overlaps within the tandem repeat array can provide insights with, the range for taxa without the tandem repeat into tandem repeat evolution and turnover and are ()3.6 to )8.6 kJ/mol). The binding energy of the in- shown in Fig. 4. In all cases, the ﬁrst and last repeats 255

that are broadly supportive of the previously proposed model of repeat origination (Fig. 1). In particular, the three tests of the model mentioned in the introduction are confirmed. First, the direct repeats flanking the duplicated region are longest in L. ludovicianus and L. excubitor, and, in particular, these are the only taxa to have a perfect copy of a CAT- TTT motif, which is also predicted to be present in their ancestor. The cytosine may be especially relevant, as it is in a position to form a GC clamp when the denatured newly replicated heavy strand reanneals to the light strand template. Second, the inverted repeat is present in the two species containing the repeat, and their common ancestor. Third, predicted binding energies of secondary structures show differences between L. ludovicianus, L. excubitor, and their ancestor and other species. Fig. 4. Tandem repeat array structure in L. excubitor (repeats 1 to Free binding energies for the inverted repeat and 3) and L. ludovicianus (p1 and p2, perfect repeats; im, imperfect repeat). Only variable sites are shown. tandem repeat are generally higher in the L. ludovicianus/L. excubitor lineage, but the most striking re- sult is that the entire sequence from the tandem in an array are the most divergent, with the middle repeat to the inverted repeat has a higher free energy repeat being identical to either the first or the last or in all members of this lineage than in taxa without the intermediate between the two. The degree of diver- tandem repeat. Overall, therefore, the best-supported gence between the first and the last repeats varies scenario is similar to the model in Fig. 1 except that among taxa. In L. e. invictus and L. e. meridionalis, both the inverted repeat and the tandem repeat re- the first and third repeats differ by 6.3% (8 bp), in gion participate in secondary structure formation to L. e. excubitor they differ by 3.2% (4 bp), while in reposition the denatured newly replicated heavy L. ludovicianus, the first and last repeats differ by only strand over the direct repeat. This model can be a 2-bp indel and a single nucleotide substitution at contrasted with others involving tandem repeat in the the 3¢ end. Comparison with the reconstructed an- 3¢ part of the control region that emphasize second- cestral sequence of L. ludovicianus and L. excubitor ary structure of the tandemly repeated region (Ha- shows that in all cases the last repeat has diverged yasaka et al. 1991; Broughton and Dowling 1994). A more than the first repeat. model proposed for crocodilians (Ray and Densmore 2003) also involves secondary structure formation outside the tandem repeat region, but here the stem Discussion loop region is upstream of the tandem repeat, and slipped-strand mispairing is predicted to occur during The 128-bp tandem repeat has a restricted distribu- light strand replication. tion in two species of shrike (L. ludovicianus/L. ex- Following the origin of the tandem repeat, gain or cubitor). Our phylogenetic reconstructions using the loss of copies could occur by a similar mechanism of control region strongly suggest that these two species slipped-strand mispairing. Intrahelical recombination form a monophyletic group, and this is supported by is an alternative mechanism by which copies could be phylogenies using whole cytochrome b sequences (A. lost. Good evidence for recombination in animal Helbig, unpublished data). As the shrike taxa we mtDNA has recently been obtained in a few species sampled are representative of the diversity of (e.g., nematodes [Lunt and Hyman 1997]), including shrikes, the tandem repeat is probably absent from one example in a tandem repeat in a flatfish (Hoarau other unsampled species of Laniidae. The tandem et al. 2002), but the overall significance of mtDNA repeat is also absent in 17 other genera, in five recombination remains unknown and controversial families of passerines (Ruokenen and Kvist 2002). (Rokas et al. 2003). This can be contrasted with the arrays in similar How many times did the tandem repeat arise in the positions in the control region of crocodilians (Ray L. ludovicianus/L. excubitor lineage? Parsimony sug- and Densmore 2003) and lagomorphs (Casane et al. gests a single origin, but if our model is correct, the 1997), which have a broad distribution across appropriate conditions for tandem repeat origination families. existed throughout the evolution of this lineage. In- Comparison of sequences in taxa with and without dependent origins of mtDNA tandem repeats have the tandem repeat reveals some interesting differences been documented in macaques (Hayasaka et al. 1991) 256 and across mammalian lineages (Wilkinson et al. As in all PCR-based studies of mtDNA, the possi- 1997). Comparisons of the relationships of repeat bility that nuclear pseudogenes of mtDNA (numts) sequences within the tandem array suggest that the have inadvertently been amplified should be considered repeats in L. e. invictus and L. e. meridionalis origi- (Bensasson et al. 2001). It is very unlikely that numts nated in or before their most recent common ances- have affected the conclusions of this study. In the log- tor, whereas the sequences in L. e. excubitor and L. gerhead shrike, the presence of heteroplasmic individ- ludovicianus are taxon-specific (Fig. 3B). Therefore, a uals with varying ratios of control-region amplicons reasonable alternative scenario would be three inde- with two and three repeats is a clear indication that pendent origins of the tandem repeat in these three nuclear pseudogenes are not involved (Mundy et al. lineages. An intermediate situation of two origins in 1996). The same PCR primers, which were designed L. e. excubitor and the common ancestor of L. e. from loggerhead shrike mtDNA, were used in other invictus/L. e. meridionalis/L. ludovicianus is also pos- species. PCR products of varying sizes (which would be sible. consistent with a numt plus a mtDNA containing one Homogenization of repeat sequences in a tandem or more tandem repeats) were never found, and direct array is evident in all cases in shrikes and is a per- sequencing never yielded mixed sequences. Finally, the vasive feature of mtDNA tandem repeats. The sim- phylogenetic reconstruction (Fig. 3A) is consistent plest mechanism to explain such homogenization is with that obtained independently using cytochrome b expansion and contraction of the array, and there are (A. Helbig, unpublished data). no examples in the dataset (Fig. 4) in which the In conclusion, we have shown that the 128-bp pattern of variation among repeats contradicts this tandem repeat in shrikes has a restricted phylogenetic possibility. If this mechanism is the explanation in distribution. Sequence analysis provides support for a shrikes then the lack of variation in repeat number in model of repeat formation in which the presence of L. excubitor is striking and suggests stabilizing se- direct repeats and secondary structure involving in- lection. Selection on array number has been com- verted repeats are important. It would be interesting monly inferred in other control region tandem to apply a similar approach to other cases of mtDNA repeats (e.g., rabbits [Casane et al. 1997], sturgeon tandem repeats. [Buroker et al. 1990], bats [Wilkinson et al. 1997]). It has been frequently suggested in these cases that selection is attributable to the presence of functional Acknowledgments. We thank D.S. Woodruff for laboratory fa- cilities for part of this work. sequences in the tandem repeat. In shrikes, no such sequences have been identified. In our previous study (Mundy et al. 1996), a poorly conserved sequence References block 3 (CSB-3) was tentatively assigned to the tandem repeat region, but a recent comparative analysis Bensasson D, Zhang DX, Hartl DL, Hewitt GM (2001) Mito- of the avian control region concluded that CSB-3 chondrial pseudogenes: Evolution’s misplaced witnesses. Trends (and CSB-2) are not present in birds (Ruokenen and Ecol Evol 16:314–321 Kvist 2002). Bentzen P, Wright JM, Bryden LT, Sargent M, Zwanenburg KCT In all cases the first repeat in the array is more (1998) Tandem repeat polymorphism and heteroplasmy in the mitochondrial control region of redfishes (Sebastes: Scor- similar to the ancestral sequence than the last repeat paenidae). J Hered 89:1–7 in the array. Such a pattern has been documented in Berg T, Moum T, Johansen S (1995) Variable numbers of simple minnows (Broughton and Dowling 1997) and pro- tandem repeats make birds of the order Ciconiiformes hetero- vides additional support for a heavy strand model of plasmic in their mitochondrial genomes. Curr Genet 27:257–262 repeat expansion, as new repeats will be added at the Broughton RE, Dowling TE (1994) Length variation in mitochondrial DNA of the minnow Cyprinella spiloptera. Genetics 5¢ end of the array. 138:179–190 Although only a short sequence segment was an- Broughton RE, Dowling TE (1997) Evolutionary dynamics of alyzed, there was strong evidence for paraphyly of tandem repeats in the mitochondrial DNA control region of the L. excubitor in comparison with L. ludovicianus. The minnow Cyprinella spiloptera. Mol Biol Evol 14:1187–1196 topology is surprising on geographic grounds since Buroker NE, Brown JR, Gilbert TA, O’Hara PJ, Beckenbach AT, Thomas WK, Smith MJ (1990) Length heteroplasmy of stur- the Mediterranean taxon L. e. meridionalis is placed geon mitochondrial DNA: An illegitimate elongation model. as sister taxon to the North American L. e. invictus, Genetics 124:157–163 while L. e. excubitor has a basal position in the Casane D, Dennebouy N, De Rochambeau H, Mounolou JC, phylogeny, but its Palearctic range lies between those Monnerot M (1997) Nonneutral evolution of tandem repeats in of the other subspecies. Similar results have been the mitochondrial DNA control region of lagomorphs. Mol Biol Evol 14:779–789 obtained using whole cytochrome b sequences from Fumagalli L, Taberlet P, Favre L, Hausser J (1996) Origin and these taxa (A. Helbig, unpublished data) and nuclear evolution of homologous repeated sequences in the mito- sequences will be required to confirm this phylo- chondrial DNA control region of shrews. Mol Biol Evol 13:31– genetic arrangement. 46 257

Ha¨ rlid A, Arnason U (1999) Analyses of mitochondrial DNA nest Moritz C, Brown WM (1986) Tandem duplications of D-loop and ratite birds within the Neognathae-supporting a neotenous or- ribosomal RNA sequences in lizard mitochondrial DNA. Sci- igin of ratite morphological characters. Proc R Soc Lond B ence 233:1425–1427 266:305–309 Mundy NI, Winchell CS, Woodruff DS (1996) Tandem repeats and Hayasaka K, Ishida T, Horai S (1991) Heteroplasmy and poly- heteroplasmy in the mitochondrial DNA control region of the morphism in the major noncoding region of mitochondrial loggerhead shrike (Lanius ludovicianus). J Hered 87:21–26 DNA in Japanese monkeys: Association with tandemly re- Quinn TW, Wilson AC (1993) Sequence evolution in and around peated sequences. Mol Biol Evol 8:399–415 the mitochondrial control region in birds. J Mol Evol 37:417– Higgins DG, Sharp PM (1989) CLUSTAL. CABIOS 5:151–153 425 Hoarau G, Holla S, Lescasse R, Stam WT, Olsen JL (2002) Het- Rand DM (1993) Endotherms, ectotherms and mitochondrial ge- eroplasmy and evidence for recombination in the mitochondrial nome size variation. J Mol Evol 37:281–295 control region of the flatfish Platichthys flesus. Mol Biol Evol Ray DA, Densmore LD (2003) Repetitive sequences in the cro- 19:2261–2264 codilian mitochondrial control region: Poly-A sequences and Hoelzel AR (1993) Evolution by DNA turnover in the control heteroplasmic tandem repeats. Mol Biol Evol 20:1006–1013 region of vertebrate mitochondrial DNA. Curr Opin Genet Dev Ritchie PA, Lambert DM (2000) A repeat complex in the mit- 3:891–895 ochondrial control region of Adelie penguins from Antarctica. Levinson G, Gutman GA (1987) Slipped-strand mispairing: A Genome 43:613–618 major mechanism for DNA sequence evolution. Mol Biol Evol Rokas A, Ladoukakis E, Zouros E (2003) Animal mitochondrial 4:203–221 DNA recombination revisited. Trends Ecol Evol 8:411–417 Lunt DH, Hyman BC (1997) Animal mitochondrial DNA recom- Ruokenen M, Kvist L (2002) Structure and evolution of the avian bination. Nature 387:247 mitochondrial control region. Mol Phylogenet Evol 23:422–432 Lunt DH, Whipple LE, Hyman BC (1998) Mitochondrial DNA Savolainen P, Arvestad L, Lundeberg J (2000) mtDNA tandem variable number tandem repeats (VNTRs): Utility and prob- repeats in domestic dogs and wolves: Mutation mechanism lems in molecular ecology. Mol Ecol 7:1441–1455 studied by analysis of the sequence of imperfect repeats. Mol Macey JR, Larson A, Ananjeva NB, Fang Z, Papenfuss TJ (1997) Biol Evol 17:474–488 Two novel gene orders and the role of light-strand replication in Stanton DJ, Dahler LL, Moritz CC, Brown WM (1994) Sequences rearrangement of the vertebrate mitochondrial genome. Mol with the potential to form stem-and-loop structures are asso- Biol Evol 14:91–104 ciated with coding-region duplications in animal mitochondrial Madsen CS, Ghivizzani SC, Hauswirth WW (1993) In vivo and in DNA. Genetics 137:233–241 vitro evidence for slipped mispairing in mammalian mito- Va¨ li U¨ (2002) Mitochondrial pseudo-control region in old world chondrial DNA. Proc Natl Acad Sci USA 90:7671–7675 eagles (genus Aquila). Mol Ecol 11:2189–2194 Mathews DH, Sabina J, Zuker M, Turner DH (1999) Expanded Wenink PW, Baker AJ, Tilanus MGJ (1994) Mitochondrial con- sequence dependence of thermodynamic parameters improves trol-region sequences in two shorebird species, the turnstone prediction of RNA secondary structure. J Mol Biol 288:911–940 and the dunlin, and their utility in population genetic studies. Mignotte G, Gueride M, Champagne A-M, Mounolou J-C (1990) Mol Biol Evol 11:22–31 Direct repeats in the non-coding region of rabbit mitochondrial Wilkinson GS, Meyer F, Kerth G, Petri B (1997) Evolution of DNA: Involvement in the generation of intra- and inter-indi- repeated sequence arrays in the D-loop region of bat mito- vidual heterogeneity. Eur J Biochem 194:561–571 chondrial DNA. Genetics 146:1035–1048