Recombination in Mitochondrial DNA of European

0 downloads 0 Views 415KB Size Report
Haplotypes. DNA was first checked by PCR with the universal AB15- ..... Mga. MedE mf1. 11a/15+mf2. 1a. MtrB. Mga mm mm. MedA mm. 1/1c. MedE. MedE. B. A.
J Mol Evol (2008) 67:377–388 DOI 10.1007/s00239-008-9157-6

Recombination in Mitochondrial DNA of European Mussels Mytilus Monika Filipowicz Æ Artur Burzyn´ski Æ Beata S´mietanka Æ Roman Wenne

Received: 22 February 2008 / Accepted: 5 August 2008 / Published online: 9 September 2008 Ó Springer Science+Business Media, LLC 2008

Abstract Mitochondrial DNA was long believed to be purely clonal and free from recombination. Major phylogenetic studies still depend on such assumptions. The peculiar genetic system of marine mussels Mytilus in which two divergent mitochondrial genomes exist provides a unique opportunity to study mtDNA recombination. Previous reports showed the existence of a few haplotypes having very strong recombination signal in the control region of mtDNA. Those recombinant variants have been found in a Baltic Sea population of Mytilus trossulus as well as in Mytilus galloprovincialis from the Black Sea. In both cases the mosaic genomes switched their transmission route and have been inherited paternally. In the present study rearranged mtDNA genomes found in all three European Mytilus species are described. The structure of their control region is a result of intra- and intermolecular recombination between mitochondrial genomes. Together with the phylogenetic reconstruction and geographic distribution, this suggests that two interlineage recombination events have occurred in the control region of mtDNA of European mussels Mytilus. Contrary to earlier observations, some of the mosaic genomes do not show any gender bias, which has important implications regarding the transmission and evolution of blue mussel mitochondrial genomes.

M. Filipowicz  A. Burzyn´ski (&)  B. S´mietanka  R. Wenne Department of Genetics and Marine Biotechnology, Polish Academy of Sciences, Institute of Oceanology, Powstan´co´w Warszawy 55, 81-712 Sopot, Poland e-mail: [email protected]

Keywords mtDNA recombination  D-loop  Control region  Doubly uniparental inheritance

Introduction There have been several reports on recombination in animal mtDNA recently (Ladoukakis and Zouros 2001; Hoarau et al. 2002; Burzyn´ski et al. 2003, 2006; Rawson 2005; Gantenbein et al. 2004; Piganeau et al. 2004; Kraytsberg et al. 2004; Breton et al. 2006; Shao et al. 2005; Zsurka et al. 2005; Tsaousis et al. 2005). The presence of more than one type of mtDNA (heteroplasmy) in an individual mitochondrion seems to be the prerequisite for recombination to occur. Sufficient divergence between recombining molecules is needed to enable the detection of recombinant molecules, but in the case of homologous recombination, an excessively big difference ([20%) suppresses this type of recombination (Rayssiguier et al. 1989). The existing methods of detecting recombination are not very powerful. Successful detection is achieved for sequences that have diverged by [5% and a number of recombination events greater than three. The higher the divergence of mitochondrial genomes in heteroplasmic cells, the greater the chance for detecting recombination (Wiuf et al. 2001). In the majority of animals mtDNA is transmitted maternally to the progeny. One way of generating heteroplasmy is paternal leakage of mtDNA, occasionally observed in some invertebrate and vertebrate species: Drosophila melanogaster (Kondo et al. 1990), mice (Gyllensten et al. 1991), honeybee (Meusel and Moritz 1993), great tit (Kvist et al. 2003), and anchovies (Magoulas and Zouros 1993). Schwartz and Vissing (2002) presented the first human case of paternal leakage in muscle tissue of a 28-year-old man. Two years later

123

378

Kraytsberg et al. (2004) showed that there had been recombination between maternal and paternal DNA in that individual. Another case of heteroplasmy is associated with doubly uniparental inheritance (DUI) of mtDNA, which is observed in at least seven families of bivalves: marine mussels Mytilidae (Skibinski et al. 1994; Zouros et al 1994), freshwater mussels Unionidae (Hoeh et al. 1996; Liu et al. 1996), Margaritiferidae (Hoeh et al. 2002), and Hyriidae (Curole and Kocher 2005), as well as clams Veneridae (Passamonti and Scali 2001), Danacidae, and Solenidae (Theologidis et al. 2008). In DUI two diverged mitochondrial molecules are transmitted to the progeny through the male (M-type) and female (F-type) lines separately. All the embryos contain both types of mtDNA, but within 24 h after fertilization the M-type is almost completely eliminated from female embryos (Sutherland et al. 1998; Sano et al. 2007). It is maintained in the germline of male embryos, leaving their somatic tissue dominated by the F-type (Cao et al. 2004b). Occasionally the paternal genomes are replaced by maternal genomes, resetting the divergence to zero (masculinization events [Hoeh et al. 1997]). Three cases of recombination in different parts of mussel mtDNA have been shown so far. Ladoukakis and Zouros (2001) provided evidence that homologous recombination occurred in the co3 region of male mussels M. galloprovincialis from the Black Sea. They examined heteroplasmic males possessing a typical F genome and Mf-masculinized genome, which played the role of the M genome. These molecules differed by about 4%. Several different co3 sequences have been detected in a single individual. No such recombinant haplotypes have been found in other individuals, and it has been speculated that recombination occurred just in the individual tested. Mytilus trossulus from the Pacific and Atlantic coasts of North America have exclusively recombinant genomes in the F lineage. Their control region (CR) contains a large insertion which has a high sequence similarity to the M. trossulus M lineage CR (Rawson 2005). Similar structures have been found in the only fully sequenced genome of American M. trossulus (Breton et al. 2006). In our earlier work (Burzyn´ski et al. 2003, 2006) several recombinant haplotypes belonging to three phylogenetically related groups were described. They occur in natural Baltic populations of M. trossulus at high frequencies, and are transmitted paternally. It has also been shown that rearranged structures of their CRs are the result of a series of multiplication and deletion events with at least one interlineage recombination. Here we ask how prevalent such rearranged haplotypes are in other European mussels, what their evolutionary relationship is, and whether they are also inherited paternally. We attempt to address these questions by screening for rearranged haplotypes by a set of sensitive PCR assays and sequencing CRs of representative genomes.

123

J Mol Evol (2008) 67:377–388

Materials and Methods Specimen Collection, DNA Extraction, PCR, and Sequencing Mussels Mytilus were collected from the Azov Sea, Black Sea, Mediterranean Sea, Bay of Biscay, North Sea, and coast of Iceland in years 2003–2005 unless otherwise indicated (Table 1 and Fig. 1). Mussels were sexed by microscopic examination of mantle tissue. The procedure succeeded for 197 individuals; 93 males and 104 females were identified. Unfortunately, in the remaining 573 individuals, immaturity of gonads prevented unambiguous identification. DNA was isolated from a small biopsy of tissue from the inner mantle surface using the CTAB method according to Hoarau et al. (2002). The isolated DNA was used in an array of PCRs, targeting the CR and rnl, rns, cob, and co2 genes. PCR primers were designed using appropriate GenBank records and sequences obtained during this study. The position, sequence, and annealing temperature (optimized in gradient PCR; T-gradient cycler from Biometra) for all primers are summarized in Table 2. All PCRs were carried out as follows, unless otherwise indicated. Approximately 20 ng of total DNA was used in a Table 1 Location of sampling sites, taxonomic identity (G, M. galloprovincialis; E, M. edulis; T, M. trossulus), total number of individuals (N), and number of individuals having a rearranged mitochondrial genome (n) Acronym

Sampling site

Taxon

GDA

Baltic; Gulf of Gdan´sk

E/Tb

49

10

E

50

0

a

N

n

ICE

Iceland

TJA

Danish Straits; Tjarno¨

E

48

0

BAL

North Sea; Balgzand

E

42

2

WES

North Sea; Westerschelde

E

47

0

SEI

English Channel; Seine

E

24

0

LOI

Bay of Biscay; Loire Estuary

E

24

0

CAU GAL

Ireland; Giant’s Causewaya Ireland; Galwaya

G/Ec G/Ed

18 19

0 0

RE

Bay of Biscay; Ile de Re

E

27

14

LAR

Bay of Biscay; La Rochelle

E

50

6

BID

Bay of Biscay; Bidasoa

G

24

0

BAN

Mediterranean Sea; Banyuls

G

26

0

ORI

Mediterranean Sea; Oristano

G

67

5

LAS

Ligurian Sea; LaSpezia

G

28

0

ODE

Black Sea; Odessa

G

48

41

SEV

Black Sea; Crimea, Sevastopol

G

73

65

FEO

Black Sea; Crimea, Feodosiia

G

85

47

AZO

Sea of Azova

G

21

14

770

204

Total a b

From S´mietanka et al. (2004). Frequency of ‘‘edulis’’ allele at 0.71, c 0.32, and d 0.22, respectively

J Mol Evol (2008) 67:377–388

379

ICE

TJA other

CAU

GDA mm

mf1 mf2

GAL BAL WES ODE

SEI LOI LAR+RE

AZO

SEV

FEO

LAS BID BAN ORI

Fig. 1 Location of sampling sites and frequencies of rearranged haplotypes. The position of each pie chart corresponds exactly with the sampling site; the area is proportional to the number of individuals sampled. The reference at the upper right represents 100 individuals. The scale bar is 500 km. Visible meridians are 20W, prime, 20E, and 40E; three parallels—40th, 50th, and 60th—of north latitude are also visible

final reaction volume of 20 ll with primers at 0.4 lM each, dNTPs at 200 lM each, 1.5 mM MgCl2, and 1 U of DyNAzymeEXT2 DNA polymerase (Finnzymes Oy) in a buffer supplied by the manufacturer. The initial denaturation at 94°C lasted 5 min, followed by 30 cycles of denaturation at 93.5°C for 1 min, annealing for 30 s

(Table 2), and extension at 72°C. Final extension at 72°C for 5 min concluded the reaction. For long-range PCR the Expand Long Template PCR System was used according to the protocol supplied by Roche, the manufacturer. PCR products were separated by agarose gel electrophoresis in 1 9 TAE (Sambrock et al. 1989) buffer and visualized with ethidium bromide in UV light. Gel images were captured using a videocamera and a framegrabber. Lengths of the PCR products were estimated using ONE-Dscan software (ver. 1.33; Scanalytics, Rockville, MD). Selected PCR products were sequenced after alkaline phosphatase/ exonuclease I treatment. Direct sequencing was performed using the BigDye terminator cycle sequencing method. An ABI 3730 automatic sequencer was used to resolve products. The Gap4 program from Staden Package (Bonfield et al. 1995) and Phred (Ewing et al. 1998) facilitated sequence assembly. All sequences have been deposited in GenBank under accession numbers EF434631–EF434653. Identification and Classification of Rearranged Haplotypes DNA was first checked by PCR with the universal AB15AB16 primer pair. Expected patterns of PCR products were obtained from all samples. Rearranged genomes were then detected by PCR with primers AB32 and AB16 (PCR-1). Obtaining a PCR product with this pair of primers indicates that at least part of the CR is multiplicated (Burzyn´ski et al. 2006). Several other PCR reactions were conducted in order to characterize these mtDNA variants further. PCR

Table 2 Summary of information on PCR primers Name

Region

Strand

Lineage

Anneal (°C)

Sequence

GenBank reference

AB15b AB16b

rnl CD

? -

F, M F, M

66 66

TTGCGACCTCGATGTTGG CAGGCTATAGAGCATAATCTAAAACG

NC_006161 1330 AY115479 759 AY115479 818

AB18

VD1

-

F

53

CAGCCGCATAGGACC

AB20b

VD1

-

M

65

GCCTTTTCCTCAGCCATCT

AY115482 1034

AB23b

rnl

?

M, F

66

AAGATTGCGACCTCGATGTTGG

NC_ 006161 1364

AB25b

VD1

?

M

61

CGCTTAACTTCCCTGCCA

AY115482 1034

AB32b

CD

?

F, M

66

TGTCAGAGTCATGTGAGACTTAACC

NC_006161 2320

AB35b

bp2

-

1/1cb

65

GCCTTTTCCTCCGCCATC

AY115481 236

AB36

bp2

-

11a/15b

65

CGTTTTTTCCTCAGCCATCT

DQ198248 1286

AB37

bp1

-

mm

65

ACACTTTTTCCTCAGCCATCTT

EF434634 1329

AB40

rnl

?

F, M

64

GACGACAAGACCCTATGAAGC

NC_006161 1156

AB48

VD1

-

M

59

TGGCAGGGAAGTTAAGCGTGTAGA

AY115482 379

AB49Rv2

VD1

?

F

59

TTGTTTGGTGATAGGTTGTTAAGTGTGG

AY115479 214

CBM2a

cob

-

F, M

61

ACCTTCACCAGGCGTTTAAG

NC_006161 2946

CBM5

VD1

?

F

53

TGGCAAAGAAAGGTTTAG

AY115479 669

MF12S MFCO2

rns co2

? -

F, M F, M

59 59

TGGTTGTCAAAGAGAAATAAATAGGCG CACCAAAATATCGACTCCCATAAAAAG

NC_006161 16067 NC_006161 4137

Note: From

a

Burzyn´ski et al. (2003) and

b

Burzyn´ski et al. (2006)

123

380

with the AB49Rv2 (F-specific) and AB48 (M-specific) primer pair (PCR-2) indicated the presence of both VD1F and VD1M domains in one genome. PCR with the universal AB23 primer together with the AB48 primer (PCR3) revealed haplotypes (rearranged and as well as typical) containing M-specific sequences only. Based on the presence and length of these PCR products, all rearranged haplotypes have been assigned to one of three groups. Variants giving 750-bp-long PCR-1, 1,230-bp-long PCR-2, and 1,500-bp-long PCR-3 products were assigned to the mf1 group. Variants yielding [800-bp-long PCR-1, 1,430bp-long PCR-2, and 1,700-bp-long PCR-3 product were assigned to the mf2 group. All other variants, with a variable-length PCR-1 product, 560-bp-long PCR-3 product, and negative for PCR-2, were assigned to the mm group. The association between gender and presence of rearranged variants was tested with the Monte Carlo method of Roff and Bentzen (1989). The same procedure has been used by Quesada et al. (2003) to draw conclusions regarding the mode of transmission of tested genotypes. Resolution of the CR Structure Selected haplotypes from each group were subject to further analysis. First, the AB32-AB16 PCR products were sequenced from both ends. The sequence provided the clue as to which primers to use to obtain PCR products overlapping the repeat on both ends (50 and 30 ). Appropriate PCR products for 50 (AB40 with either AB20 or AB36 or AB37) and 30 (always AB25-CBM2) were obtained. Direct sequencing of those PCR products was carried out. The sequences of those three PCR products were combined, yielding the sequence of the CR for a rearranged genome. This procedure has been used successfully for Baltic Mytilus trossulus (Burzyn´ski et al. 2006). The sequence obtained in such a way usually contained only one copy of a repeated fragment corresponding to the AB16/AB32 PCR product. In the typical M and F genome, 40 bp separates primers AB16 and AB32. Assuming that this distance is conserved in genomes with tandemly arranged copies of the fragment containing both primer binding sites, the estimate of the length of the repeat unit can be obtained by adding 40 bp to the length of the AB16/AB32 PCR product. To get a rough estimate of the number of repeats involved, long-range PCR with primers MF12S and MFCO2 spanning the region from rns to co2 was performed. By dividing the apparent increase in length of the obtained amplicon over that of the typical genome by the length of the repeat estimated previously, it was possible to estimate the number of repeats. To confirm the structure of recombinant haplotypes Southern blotting analysis was conducted as described previously (Burzyn´ski et al. 2006), with minor modifications. One additional

123

J Mol Evol (2008) 67:377–388

restriction enzyme was used (MboI) and the F-specific probe located within the VD1F domain substituted for the universal probe. The probe was obtained and labeled by PCR with the CBM5-AB18 primer pair. To annotate sequences properly, and to pinpoint the exact location of recombination breakpoints, all complete CR sequences were subject to the following procedure. First, breakpointcontaining regions were identified by dotplot comparison of rearranged sequences with reference M and F sequences, as described previously (Burzyn´ski et al. 2006). Then the relevant fragments were aligned and analyzed by a series of recombination detection programs: Geneconv (Padidam et al. 1999), MaxChi (Maynard Smith 1992), Chimaera (Posada and Crandal 2001), SiScan (Gibbs et al. 2000), Bootscan (Martin et al. 2005a), and 3SEQ (Boni et al. 2007) as implemented in RDP version 3 beta 24 (Martin et al. 2005b). Phylogenetic Reconstruction Recombination processes affect the evolutionary history of the CR so that it cannot be examined as a whole. A fragment of the variable domain VD1 was used in phylogenetic analysis because it was not interrupted by recombination. Conveniently the selected fragment also exhibits enough variation. Unfortunately this is also the most diverged part between the M and the F genomes (Burzyn´ski et al. 2003; Cao et al. 2004a). These differences are so large (up to 80% sequence divergence, 50% difference in sequence length) that the alignment of VD1 fragments from both the M and the F genomes cannot be reliable. Therefore, it was decided not to do a global M-F alignment but to analyze M-specific and F-specific VD1 fragments separately. In the case of M-specific sequences the 255-bp fragment of VD1 was analyzed. For the F genome a 650-bp-long fragment was used. Relevant fragments were aligned with ClustalW (version 1.83; Higgins and Sharp 1989; Thompson et al. 1994), with a gap opening penalty at 6.66 and gap extension penalty at 15. Differences in sequence length were small and attributed to different lengths of homopolymeric fragments, hence the alignments obtained were reliable. Phylogenetic relationships were reconstructed using several methods. For neighbor joining (NJ) and minimum evolution (ME) analysis, MEGA version 3.1 (Kumar et al. 2004) was used with default parameters, and trees obtained were bootstrapped with 1,000 iterations. For maximum likelihood (ML), PAUP*, version 4.0 beta 10 (Swofford 2003), was used. The best model of sequence evolution was chosen according to Modeltest (Posada and Crandall 1998). The HKY ? G model was used with gamma shape parameter at 1.875 for M-specific alignment, and the TrN ? G model with gamma shape at 0.231 was used for F-specific alignment. For Bayesian analysis, MrBayes

J Mol Evol (2008) 67:377–388

version 3.1 (Ronquist and Huelsenbeck 2003) was used. The Markov chain was run with eight plausible models (sets of parameters) for both alignments (all combinations of nst = 2 or 6, rates equal or gamma-distributed, with or without covarion). It was run for the number of generations sufficient to be in the stationary phase for at least 106 generations (up to 108 generations total). Data for the nonstationary phase were discarded. Bayes factor comparison of models was performed, favoring the model defined by nst = 2, gamma-distributed rates of substitutions with covarion. The trees obtained using this model are presented. Genetic distances were calculated in MEGA with the Kimura two-parameter model.

381

3000

2000

1000

540

M

mf1

mm

F

mf2

3000

Results 2000

Recombinant Structure of the CR In this study 17 sequences of CRs with parts of flanking genes—rnl, trnY, and cob—were obtained. Comparisons of sequences obtained with the F and M reference sequences led to the conclusion that some of the examined sequences are recombinant and contain parts of the M and F genomes. The presence of an M-specific fragment within rearranged haplotypes was confirmed by Southern blotting analysis (Fig. 2), consistent with PCR and sequencing results. Strong recombination signals were detected in all sequences. Examples of p-values for breakpoints estimated by RDP and Bootscan for all sequences analyzed are reported in Table 3. The precise location of breakpoints within low-divergence regions (such as CD) may still be unreliable. Consequently, some recombination detection programs indicated it ambiguously. In such cases the range of possible breakpoint locations has been defined. The structures of representatives for each of three groups of haplotypes are shown in Fig. 3. The mm haplotypes contained a variable number of tandem repeats consisting of M-specific sequences only. Each repeat contained at least a part of VD1, the whole CD, VD2, and usually a fragment of trnY. For each mm haplotype only one sharp breakpoint has been found. It marked the current extent of tandem repeat units present in those haplotypes and was situated between the fragment of trnY and the truncated VD1 domain (repeat length, 790–1000 bp; 800 bp in most cases). The sequence of mf haplotypes contained both F- and M-specific parts. All of the mf haplotypes have an F-specific VD1 domain fragment of similar length and a fragment of M-specific VD1 domain of variable length. This fragment was shorter in mf1 than in most mf2 haplotypes, although the mf2 group was heterogeneous with regard to the length of its VD1M domain. All mf haplotypes encompassed the fragment of trnY before the M-M

1000

540

Fig. 2 An example of Southern hybridization analysis. DNA was digested with MboI and hybridized with the M-specific probe (top). A second hybridization with the F-specific probe (bottom) was performed after stripping off the first probe. Marker fragment sizes (bp) are given at the left. Both probes are located at the 30 end of VD1 and do not cross-hybridize. MboI does not cut F-specific VD1 but cuts M-specific VD1. The 3,000-bp-long fragment hybridizing with the Fspecific probe is characteristic for the typical F genome, and a 540bp-long fragment hybridizing with the M-specific probe is characteristic for the M genome. Additional bands (1,800—2,000 bp long) hybridizing with both probes are characteristic for mf haplotypes. The bands at ca. 800—1,000 bp represent repeated M-specific fragments (see Fig. 3)

recombination breakpoint and contained a multiplicated region including the part of trnY, VD1M, CD, and part of VD2 (mf1 repeat unit length, 790 bp; mf2, 750–1000 bp). Haplotypes mm and mf2 have variable numbers of repeats (based on the length of MF12S/MFCO2 and AB16/AB32 PCR products), ranging from 4 to 12, for the total length of the CR exceeding 10000 bp in extreme cases. Haplotypes from mf1 group have a uniform repeat length of 800 bp. The number of repeats was difficult to ascertain due to poor PCR efficiency. In one case the length of the PCR product was consistent with the presence of 13 repeats, for a total CR length of [ 11,000 bp. For haplotype mf1 and mf2 three recombination breakpoints have been identified. The first one marked the change from F to M and was located in either the 30 or the 50 part of CD in mf1 or mf2, respectively. The second breakpoint was found between

123

382

J Mol Evol (2008) 67:377–388

Table 3 Statistical support for recombination breakpoints

Haplotype

Breakpoint

RDP

mm

bp1

3.1 9 10-3–6.03 9 10-17

mf1

-2

bp1

2.24 9 10-7–9.73 9 10-9

-9

1.48 9 10-2–8.29 9 10-9

5.39 9 10 –3.89 9 10

bp2

-8

-15

-2

-4

1.2 9 10-2–3.66 9 10-5

6.1 9 10 –3.5 9 10

bp3 mf2

Bootscan

2.4 9 10 –9.62 9 10

6.65 9 10-2–8.44 9 10-3

-5

5.9 9 10-4

bp1

6.5 9 10

bp2

4.08 9 10-8

bp3

-5

1 9 10-4 1.08 9 10-3

2.56 9 10

trnY AB32

AB16

cob

200bp

F

VD1

rnl

CD

VD2

trnY AB32

AB16

VD2

M

VD1

rnl

trnY AB16

mm

rnl

AB32

bp1

VD2

AB32

AB16

VD2

trnY cob

CD

VD1

CD

VD1

cob

CD

RPT

AB16

bp1

AB32

trnY

bp2

bp3 AB16

AB32

mf1

CD

VD1

rnl

VD1

trnY cob

VD2 CD

VD2

RPT

AB16

mf2

rnl

trnY VD2

AB32

CD

VD1 bp1

bp2

AB16

bp3

AB32

cob CD

VD1

trnY

VD2

RPT

Fig. 3 Diagrams of haplotypes identified in this study. Typical M and F genomes are shown for reference. Dark-gray and white boxes represent M-related and F-related regions, respectively, identified without ambiguity by at least two recombination detection programs. Back-to-back arrows indicate the position of the AB16 and AB32 PCR primers used to detect repeated units. The convention for naming CR domains follows that of Mizi et al. (2005) and Burzyn´ski et al.

(2006). All variants are aligned at the VD1/CD boundary. Doubleheaded arrows indicate spans of tandem repeat units (RPT); only one unit is shown for simplicity. Arrowheads on boxes denote the ends of rnl and trnY genes. Potential breakpoints identified by the RDP suite of programs are indicated as bp1, bp2, and bp3 for the relevant haplotypes

the part of trnY and the VD1M domain, similarly to mm haplotypes. The third breakpoint was positioned within the VD2 domain. All positions of recombination breakpoints are shown in Fig. 3 and are annotated in GenBank records.

(Burzyn´ski et al. 2006) were used. They belonged to three separate groups: the 1a group, with the shortest and simplest CR structure; the 11a/15 group, with a CR structure identical to that of mf2 genomes; and the 1/1c group, with a complex CR structure. Genomes from the 1a and 1/1c groups did not contain VD1F fragments and consequently were included in the VD1M tree only. The trees for M- and F-specific VD1 domains were constructed separately because of the high sequence divergence. Both phylogenetic trees were built with several tree-building methods: NJ, ME with MEGA, ML with PAUP*, and MrBayes. Each method gave a similar topology. The Bayesian trees are shown in Fig. 4. For the VD1M two well-supported recombinant clades could be identified. The first one contains mf1 and Baltic type 1/1c haplotypes.

Evolutionary History of Rearranged Genomes To ascertain the origin of rearranged haplotypes a comparative phylogenetic analysis was performed. The longest fragments (and also unbroken by recombination) of VD1M and VD1F were selected for analysis. Several reference sequences were used, including sequences from typical M and F as well as recombinant genomes. In particular, relevant fragments of all recombinant genomes from Baltic Sea M. trossulus described previously

123

J Mol Evol (2008) 67:377–388

383

A

26 27

91

24 25 10 9 11 67 7

65 100

72

0.1

8 2 3 29

100 94

1 89

100

18

5 19

100

11a/15+mf2

4 30 12

86

37a 37 36

75

55

92 100

33 20 32a 31 32

24

39 26 38

mf1

25 100 100

14

13

16 15 17 6

1/1c

MedA

89 100

34a 34

53

75

B

28

72

67

mf1

100

mm

100

43 100

MedE

44 42

28 29 40 2 41

11a/15+mf2

3 4 30 1

Mga MedE

mm

MtrB 1a

MedE 90

99

100

23 22 21

Mga

35

mm

Fig. 4 Phylogenetic trees of haplotypes based on VD1 sequences constructed using Bayesian analysis. (A) VD1M tree based on a 255bp-long alignment. (B) VD1F tree based on a 650-bp-long alignment. The trees are unrooted. Posterior probabilities of bipartitions [ 0.5 multiplied by 100 are shown as support for clades. Appropriate fragments of several database sequences are used for comparison. The identity of each leave/clade is as follows: mf1, mf2, and mm— haplotypes identified in this study; 1/1c, 1a, and 11a/15—masculinized recombinant haplotypes described previously in Baltic M. trossulus (Burzyn´ski et al. 2003, 2006). Reference M and F sequences come from the following: Mga, M. galloprovincialis; MedA, M. edulis from North America; MedE, M. edulis from Europe; MtrB, M. trossulus from the Baltic Sea. An arbitrary label is given to each haplotype. The same label is given to both VD1M and VD1F haplotypes derived from the same record, whenever possible. In some instances two fragments representing repeats from the same genome are presented; the second repeat is denoted by an ‘‘a’’ suffix. The

haplotypes (VD1F and VD1M respectively) derived from the following GenBank records are labeled as follows: DQ198244, 1; DQ198245, 2; DQ198247, 3; DQ198248, 4; DQ198236, 5; DQ198238, 6; DQ198239, 7; DQ198240, 8; DQ198242, 9; DQ198243, 10; DQ198241, 11; AY350791; AY823623, 12; AY823624, 13; AY350793, 14; DQ198222; DQ198223, 15; DQ198225, 16; DQ198224, 17; EF434632, 18; EF434633, 19; AY350794, 20; AY363687, 21; AY629164, 22; AY629163, 23; EF434643, 24; EF434648, 25; EF434642 and EF434644, 26 and 24; EF434639, EF434641, and EF434646, 26; EF434647, 38 and 26; EF434645, 39 and 26; EF434640, 27 and 26; EF434649, 28; EF434653, 29; DQ198249, 40 and 29; DQ198246, 41 and 29; EF434638, 30; EF434635 and EF434635, 31; AY115480, DQ198237, and EF434634, 32; EF434636, 33; EF434650, 34; EF434652, 35; EF434631 and EF434637, 36; EF434651, 37; AY484747, 42; AY497292, 43; AY629165, 44. Records in boldface were generated during this study

This apparently close relationship between mf1 and previously described 1/1c haplotypes is also supported by dotplot analysis (Fig. 5). The second clade contains mf2 as well as Baltic 11a/15 haplotypes. Noticeably, the support for common origin of those two clades was high (posterior probability of bipartition, 1.0). Bootstrapping of ML and NJ trees also supported all highly probable bipartitions with support value ranging from 80% to 100% (data not shown). Haplotypes from the mm group did not resolve well on the tree and formed a loose group in the center, together with typical European M as well as Baltic 1a haplotypes. One exceptional mm haplotype grouped with M. galloprovincialis M sequences. For the VD1F tree each of the mf groups was again recovered as monophyletic. The close relationship of mf2 with 11a/15 was also fully confirmed by VD1F data. It was, however, impossible to recover both recombinant groups as a single clade: some

reference F haplotypes of M. galloprovincialis are more closely related to mf2 than are mf1 haplotypes. Geographic and Sex Distribution of Rearranged Haplotypes Most Atlantic M. edulis populations did not contain any rearranged haplotypes (Table 1). The geographic distribution of all three groups is presented in Fig. 1. Haplotypes from the mf1 group were abundant in Black and Azov Seas. Up to 89% of individuals from those populations carry the mf1 haplotype. Haplotypes from the mm group were rather rare; they never reached frequencies comparable to mf1’s, although at their main locations in the Bay of Biscay, up to 25% of animals contain such haplotypes. Haplotypes scored as mf2 in Baltic (20%) likely represent previously described 11a/15 CR length variants (Burzyn´ski et al.

123

384

J Mol Evol (2008) 67:377–388

Discussion

1400 1200

Relationship with Baltic Recombinants and Evolutionary Scenario

1000 800 600 400 200 0 1500

1000

500

0

2000

mf1

Fig. 5 Dotplot comparison of the mf1 CR structure with that of the previously described Baltic haplotype 1. The dotpath program from the EMBOSS package (Rice et al. 2000) was used with word size 4 to obtain the graph. The comparison starts within the rnl gene, ends within the cob gene, and spans the area depicted in Fig. 3. Only one copy of a repeat unit in mf1 is included in comparison. The haplotypes compared differ by two features: one large deletion encompassing most of VD1F, the first CD, and part of the first VD2 and a lack of repeats in the shorter haplotype. The sequences were derived from the following GenBank records: haplotype 1, from DQ198242; haplotype mf1, from EF434647 Table 4 Haplotype frequency in sexed animals; combined data from four selected sampling sites (FEO, GDA, RE, ORI)a Haplotype

Male

Female

Undetermined

Total

n

n

n

n

%

%

%

%

mf1

21

22.6

27

26.0

0

0.0

48

mf2

10

10.8

8

7.7

2

6.5

20

8.8

mm Other

6 56

6.5 60.2

0 69

0.0 66.3

2 27

6.5 87.1

8 152

3.5 66.7

Total

93

100.0

104

100.0

31

100.0

228

100.0

a

21.1

See Table 1 for location acronyms

2006). Haplotypes belonging to this group were rare elsewhere, again with the exception of one location in the Bay of Biscay (25%). To determine if mf and mm haplotypes exhibit any sex bias the haplotype 9 sex contingency table was analyzed by the Monte Carlo method of Roff and Bentzen (1989) (Table 4). Four samples were sexed successfully enough to be included in this analysis. The distribution of rearranged haplotypes among sexes was significantly nonrandom. The association of each haplogroup with gender was tested separately by evaluating a 2 9 2 (haplotype 9 sex) contingency table. The group in question was tested against all other pooled haplotypes. Haplotypes from the mm group were never present in females and their association with males was significant (P \ 0.01). However, neither the mf1 nor the mf2 group exhibited a significant sex bias (P [ 0.1). The mf1 haplotypes were actually scored more frequently in females (26.0%) than in males (22.6%) but the association was nonsignificant.

123

In a previous study (Burzyn´ski et al. 2006) several groups of masculinized haplotypes from a Baltic population of M. trossulus having mosaic F/M/F CRs were described. Haplotypes from two related groups (11a and 15) are apparently undistinguishable from haplotypes from the mf2 group presented here. They have similar structures and are situated together on the same clade of the phylogenetic tree (posterior probability, 0.89). The scenario proposed previously to explain the origin of Baltic recombinant genomes involved initial duplication of the CR fragment in an M genome, subsequent recombination of such rearranged genome with an F genome, and, finally, a series of deletions and secondary multiplications. The scenario called for the existence of two intermediates: one similar to the typical M genome, but with a multiplicated fragment of the CR, and the second recombinant with a shorter VD1M fragment than in mf2 (Burzyn´ski et al. 2006, Fig. 5, p. 1090). The structures proposed there have hereby been found: the CR of mm genomes is almost identical to that previously postulated for intermediate 1, and the CR of mf1 corresponds well with the postulated intermediate 2. Moreover, the comparison of Baltic recombinant type ‘‘1’’ with typical mf1 haplotype (Fig. 5) unequivocally confirms that they differ solely by one large deletion within the VD1F domain. The genetic distance for the remaining part of the studied mtDNA segment is also very small, at 0.02. It is surprising to note that this relatively complex pathway, involving at least two tandem duplication events, recombination, and a series of deletions, led to the haplotype’s (type ‘‘1’’) VD1 domain being of exactly the same length as a typical M genome. The close relationship between Baltic type ‘‘1’’ recombinants and mf1 genomes gets further support from phylogenetic reconstruction (Fig. 4), in which those two are on the same clade with a probability of 1.0. Altogether these facts confirm that mf1 haplotypes indeed represent previously postulated intermediate 2. The phylogenetic analysis does not, however, support a common origin of recombinants from both mf groups and mm haplotypes, despite their structural similarity. Moreover, the monophyletic origin of the mm group itself is not supported at all. While this may be due primarily to the inevitably poor resolution achievable by analysis based on too few polymorphisms (short alignment), it most likely means that this group is truly heterogeneous and new haplotypes with tandemly duplicated CR fragments are quite frequently generated within paternal mtDNA lineages of Mytilus mussels. Therefore, it is likely that one such haplotype, now extinct, previously postulated as intermediate 1, gave rise to both the mf1 and

J Mol Evol (2008) 67:377–388

the mf2 groups. This is entirely consistent with the phylogenetic reconstruction based on the M-specific fragment. There is, however, discord between the VD1M and the VD1F trees. According to the VD1M tree one recombination event may have led to both mf1 and mf2 haplotypes. However, according to the VD1F tree there are also apparently nonrecombinant genomes within the clade encompassing both mf1 and mf2. This is not consistent with the previously postulated scenario and can be explained in several ways. One possibility is that the extinct haplotype, ancestral for the M-specific parts of the mf genomes (most likely mm—like intermediate 1), existed long enough before recombination to diverge substantially from other M genomes and then, in two separate recombination events, with two different F genomes, gave rise to two independent recombinant lines. This would explain the different lengths of VD1M in the mf1 and mf2 genomes by different extents of those two recombination events. A second explanation would call for a single recombination event leading to one ancestral recombinant mf haplotype, with a subsequent series of deletions eventually leading to current mf1 and mf2. In this scenario the complete loss of inserted M-specific sequences in some evolutionary lineages effectively re-created the structure of the typical F genome. This hypothesis also gets support from the observation that the CD part of the CR in some M. galloprovincialis F genomes clusters with CD domains of M rather than other F genomes in the NJ tree (Cao et al. 2004a). The sequence from one such genome intervenes between mf1 and mf2 in our VD1F tree. The third possibility would also require a series of two recombination events: primary recombination between ancestral mm-like M and typical F genomes, leading to an mf2-like ancestor, and later a secondary recombination transferring the shorter VD1M fragment from this genome to another F genome, leading to mf1. Consequences for DUI Haplotypes from the mm group represent a certain instability of mtDNA CR, very similar to that commonly found in other species (Kumazawa et al. 1996; Gach and Brown, 1997; Eberhard et al. 2001; Lavrov et al. 2002; Mueller and Boore 2005; Shao et al. 2004; Abbott et al. 2005). The fact that they have been found mostly near an apparent hybridization zone between M. edulis and M. galloprovincialis (Bierne et al. 2003), supports the view that hybridization may favor such instabilities. The first case describing CR duplication in mtDNA also involved a mixed-species context (Stanton et al. 1994). Tandem duplications of even greater length have been postulated to explain the structure of certain paternal haplotype in M. galloprovincialis (Burzyn´ski 2007), suggesting that the presence of structurally similar but not directly related mm

385

haplotypes at low frequencies may be a feature of European Mytilus populations. These haplotypes stay within the paternal lineage and do not violate DUI. Haplotypes from the mf2 group are quite rare and can be found mainly on the Atlantic coast of France and in the Baltic Sea. They are most likely identical to Y haplotypes reported previously from southwestern England (Fisher and Skibinski 1990). EcoRI digests of Y haplotypes are in good agreement with our data and consistent with Y haplotypes having very large CRs. Even though the mf2 haplotypes found on the Atlantic coast and Baltic masculinized genomes described previously (11a/15 group [Burzyn´ski et al. 2006]) are undistinguishable, their sex distribution is different. These haplotypes are present exclusively in males in the Baltic Sea, whereas they can be easily found in females elsewhere. It is tempting to speculate that mf2 haplotypes originated within the apparent hybridization zone and invaded the paternal line of inheritance in Baltic Mytilus very recently. Haplotypes from the mf1 group are limited to the Black Sea and they dominate there. Contrary to the sexbiased distribution of related Baltic haplotypes (1/1c [Burzyn´ski et al. 2003, 2006]), mf1 do not seem to be associated with any gender. They are frequently found in heteroplasmic males together with the typical M genome as well as with the typical F genome and are also common in homoplasmic females. They can also be found in sperm. It is likely that they represent genomes previously reported as masculinized from the Black Sea basin (Ladoukakis et al. 2002; S´mietanka et al. 2004). Moreover, the recently described genome found in highly purified sperm (Venetis et al. 2007) also belongs to the mf1 group: it has the same CR structure and sequence (i.e., it has only three substitutions compared to one of our partial mf1 sequences— EF434642—over the whole alignable 2,500 bp). The mf1 haplotypes are present at high concentrations, readily detectable by Southern hybridization and not just by targeted PCR—a striking difference from the trace amounts of paternally inherited genomes detectable in various tissues and sexes only by much more sensitive techniques (Garrido-Ramos et al. 1998; Sano et al. 2007). Since mf1 genomes found in males and females are very similar sequence-wise (overall nucleotide divergence, 0.004), and they do not form separate clades in the tree, their current status as fully masculinized genomes is debatable at best. Despite being present in sperm, they do not have all the features of the masculinized genome. It has been shown previously that paternally inherited genomes are associated with males in population studies (e.g., Skibinski et al. 1994; Quesada et Al. 2003; Burzyn´ski et al. 2006). The distribution of genomes from the mf1 group among sexes is different. This can be explained only by invoking some kind of disruption in the DUI system. The mf1 genomes could originate from an ongoing or very recent

123

386

masculinization with near-zero divergence between paternally and maternally inherited genomes, both belonging to the recognizable mf1 group. Should that be the case, it would be comparable with the case of American M. trossulus, in which haplotypes with recombinant CR structures have been reported as maternally inherited genomes (Rawson 2005) and also as genomes postulated to be recently masculinized (Breton et al. 2006). The situation in American M. trossulus is complicated by the lack of nonrecombinant F genomes. There is apparently only one phylogenetic group of recombinant genomes, and therefore it is impossible to trace back the history of role reversals in this population unless non-recombinant F haplotypes are found there. In European Mytilus spp. the situation is different, as the three species share the same mtDNA haplotypes of M. edulis and several nonrecombinant F lineages exist. The evolutionary relationship with and distance of mf1 haplotypes from other F haplotypes suggest that they represent a relatively old group. Thus they must have evolved within one lineage for some time already. It is unclear whether this was the maternal or paternal lineage. There is, however, no doubt that Baltic 1/1c paternal haplotypes originated from the ancestor closely related to mf1. The parsimonious explanation would therefore be that the ancestral mf haplotype was inherited paternally and the present lack of association with males, particularly in the case of mf1 haplotypes, represents a secondary feminization event. Alternatively mf1 haplotypes may not obey strict DUI rules and be inherited truly biparentally, perhaps due to the replicational advantage gained by virtue of multiplicated regulatory sequences. This would still be consistent with the hypothesis that the M-type CR sequences may favor paternal inheritance of genomes bearing them (Burzyn´ski et al. 2003), although in a nondeterministic fashion. Future Studies Heteroplasmy for two moderately divergent genomes and fusion of mitochondria seem to favor mtDNA recombination in mussels. Nevertheless, the process has far fewer consequences than could have been expected—as few as two such events suffice to explain all CR recombinant mtDNA variants present in European Mytilus mussels. To find out exactly how the extant recombinant genomes are inherited, more detailed analyses of mitochondrial genomes from the Black Sea population of M. galloprovincialis are required, including rigorous comparison of genomes from somatic tissues and gametes of both sexes. Further comparative analysis and precise identification of the closest nonrecombinant ancestors of mf genomes will also help to elucidate the exact pathways, mechanisms, and time scales involved.

123

J Mol Evol (2008) 67:377–388 Acknowledgment This work was partially supported by Grant N30300531/0226 from the Polish Ministry of Science and Higher Education to R.W.

References Abbott CL, Double MC, Trueman JWH, Robinson A, Cockburn A (2005) An unusual source of apparent mitochondrial heteroplasmy: duplicate mitochondrial control regions in Thalassarche albatrosses. Mol Ecol 14:3605–3613 Bierne N, Borsa P, Daguin C, Jollivet D, Viard F, Bonhomme F, David P (2003) Introgression patterns in the mosaic hybrid zone between Mytilus edulis and M. galloprovincialis. Mol Ecol 2:447–461 Bonfield JK, Smith KF, Staden R (1995) A new DNA sequence assembly program. Nucleic Acids Res 24:4992–4999 Boni MF, Posada D, Feldman MW (2007) An exact nonparametric method for inferring mosaic structure in sequence triplets. Genetics 176:1035–1047 Breton S, Burger G, Stewart DT, Blier PU (2006) Comparative analysis of Gender-associated complete mitochondrial genomes in marine mussel (Mytilus spp). Genetics 172:1107–1119 Burzyn´ski A (2007) Two events are responsible for an insertion in a paternally inherited mitochondrial genome of the mussel Mytilus galloprovincialis. Genetics 175:959–962 Burzyn´ski A, Zbawicka M, Skibinski DOF, Wenne R (2003) Evidence for recombination of mtDNA in the marine mussel Mytilus trossulus from the Baltic. Mol Biol Evol 20:388–392 Burzyn´ski A, Zbawicka M, Skibinski DOF, Wenne R (2006) Doubly uniparental inheritance is associated with high polymorphism for rearranged and recombinant control region haplotypes in Baltic Mytilus trossulus. Genetics 174:1081–1094 Cao L, Kenchington E, Zouros E, Rodakis GC (2004a) Evidence that the large noncoding sequence is the main control region of maternally and paternally transmitted mitochondrial genomes of the marine mussel (Mytilus spp.). Genetics 167:835–850 Cao L, Kenchington E, Zouros E (2004b) Differerential segregation patterns of sperm mitochondria in embryos of the blue mussel (Mytilus edulis). Genetics 166:883–894 Curole JP, Kocher TD (2005) Evolution of a unique mitotype-specific protein-coding extension of the cytochrome c oxidase II gene in freshwater mussels (Bivalvia: Unionoida). J Mol Evol 61:381–389 Eberhard JR, Wright TF, Bermingham E (2001) Duplication and concerted evolution of the mitochondrial control region in the parrot genus Amazona. Mol Biol Evol 18:1330–1342 Ewing B, Hillier L, Wendl MC, Green P (1998) Base-calling of automated sequencer traces using phred. I. Accuracy assessment. Genome Res 8:175–85 Fisher C, Skibinski DOF (1990) Sex-biased mitochondrial DNA heteroplasmy in the marine mussel Mytilus. Proc R Soc Lond Ser B 242:149–156 Gach MH, Brown WM (1997) Characteristics and distribution of large tandem duplications in brook stickeback (Culaea inconstans) mitochondrial DNA. Genetics 145:383–394 Gantenbein B, Fret V, Gantenbein-Ritter IA, Balloux F (2004) Evidence for recombination in scorpion mitochondrial DNA (Scorpiones: Buthidae). Proc R Soc Lond Ser B 272:697–704 Garrido-Ramos MA, Stewart DT, Sutherland BW, Zouros E (1998) The distribution of male-transmitted mitochondrial DNA types in somatic tissues of blue mussels: implications for the operation of doubly uniparental inheritance of mitochondrial DNA. Genome 41:818–824 Gibbs MJ, Armstrong JS, Gibbs AJ (2000) Sister-scanning: a Monte Carlo procedure for assessing signals in recombinant sequences. Bioinformatics 16:573–582

J Mol Evol (2008) 67:377–388 Gyllensten U, Wharton D, Josefsson A, Wilson AC (1991) Paternal inheritance of mitochondrial DNA in mice. Nature 352:255–257 Higgins DG, Sharp PM (1989) Fast and sensitive multiple sequence alignments on a microcomputer. Comput Appl Biosci 5:151–153 Hoarau G, Holla S, Lescasse R, Stam WT, Olsen JL (2002) Heteroplasmy and evidence for recombination in the mitochondrial control region of the flatfish Platichthys flesus. Mol Biol Evol 12:2261–2264 Hoeh WR, Stewart DT, Sutherland BW, Zouros E (1996) Multiple origins of gender-associated mitochondrial DNA lineages in bivalves (Mollusca: Bivalvia). Evolution 50:2276–2286 Hoeh WR, Stewart DT, Saavedra C, Sutherland BW, Zouros E (1997) Phylogenetic evidence for role-reversals of gender-associated mitochondrial DNA in Mytilus (Bivalvia: Mytilidae). Mol Biol Evol 14:959–967 Hoeh WR, Stewart DT, Guttman SI (2002) High fidelity of mitochondrial genome transmission under the doubly uniparental mode of inheritance in freshwater mussels (Bivalvia: Unionoida). Evolution 56:2252–2261 Kondo R, Satta Y, Matsuura ET, Ishiwa H, Takahatat N, Chigusa SI (1990) Incomplete maternal transmission of mitochondrial DNA in Drosophila. Genetics 126:657–663 Kraytsberg Y, Schwartz M, Brown TA, Ebralidse K, Kunz WS, Clayton DA, Vissing J, Khrapko K (2004) Recombination of human mitochondrial DNA. Science 304:981 Kumar S, Tamura K, Nei M (2004) MEGA3: integrated software for molecular evolutionary genetics analysis and sequence alignment. Brief Bioinform 5:150–163 Kumazawa Y, Ota H, Nishida M, Ozawa T (1996) Gene rearrangements in snake mitochondrial genomes: highly concerted evolution of control-region-like sequences duplicated and inserted into a tRNA cluster. Mol Biol Evol 13:1242–1254 Kvist L, Martens J, Nazarenko AA, Orell M (2003) Paternal leakage of mitochondrial DNA in the great tit (Parus major). Mol Biol Evol 20:243–247 Ladoukakis ED, Zouros E (2001) Direct evidence for homologous recombination in mussel (Mytilus galloprovincialis) mitochondrial DNA. Mol Biol Evol 18:1168–1175 Ladoukakis ED, Saavaedra C, Magoulas A, Zouros E (2002) Mitochondrial DNA variation in a species with two mitochondrial genomes: the case of Mytilus galloprovinicialis from the Atlantic, the Mediterranean and the Black Sea. Mol Ecol 11:759–769 Lavrov DV, Boore JL, Brown WM (2002) Complete mtDNA sequences of two millipedes suggest a new model for mitochondrial gene rearrangements: duplication and nonrandom loss. Mol Biol Evol 19:163–169 Liu H, Mitton JB, Wu S (1996) Paternal mitochondrial DNA differentiation far exceeds maternal mitochondrial DNA and allozyme differentiation in the freshwater mussel, Anodonta grandis grandis. Evolution 50:952–957 Magoulas A, Zouros E (1993) Restriction-site heteroplasmy in anchovy (Engraulis encrasicolus) indicates incidental biparental inheritance of mitochondrial DNA. Mol Biol Evol 10:319–325 Martin DP, Posada D, Crandall KA, Williamson C (2005a) A modified bootscan algorithm for automated identification of recombinant sequences and recombination breakpoints. AIDS Res Hum Retroviruses 21:98–102 Martin DP, Williamson C, Posada D (2005b) RDP2: recombination detection and analysis from sequence alignments. Bioinformatics 21:260–262 Maynard Smith J (1992) Analyzing the mosaic structure of genes. J Mol Evol 34:126–129 Meusel MS, Moritz RF (1993) Transfer of paternal mitochondrial DNA during fertilization of honeybee (Apis mellifera L.) eggs. Curr Genet 24:539–543

387 Mizi A, Zouros E, Moschonas N, Rodakis GC (2005) The complete maternal and paternal mitochondrial genomes of the Mediterranean mussel Mytilus galloprovincialis: implications for the doubly uniparental inheritance mode of mtDNA. Mol Biol Evol 22:952–967 Mueller RL, Boore JL (2005) Molecular mechanisms of extensive mitochondrial gene rearrangement in plethodontid salamanders. Mol Biol Evol 22:2104–2112 Padidam M, Sawyer S, Fauquet CM (1999) Possible emergence of new geminiviruses by frequent recombination. Virology 265:218–225 Passamonti M, Scali V (2001) Gender-associated mtDNA heteroplasmy in the venerid clam Tapes philippinarum (Mollusca Bivalvia). Curr Genet 39:117–124 Piganeau G, Gardner M, Eyre-Walker A (2004) A broad survey of recombination in animal mitochondria. Mol Biol Evol 21: 2319–2325 Posada D, Crandall KA (1998) Modeltest: testing the model of DNA substitution. Bioinformatics 14:817–818 Posada D, Crandall KA (2001) Evaluation of methods for detecting recombination from DNA sequences: computer simulations. Proc Natl Acad Sci USA 98:13757–13762 Quesada H, Stuckas H, Skibinski DOF (2003) Heteroplasmy suggests paternal co-transmission of multiple genomes and pervasive reversion of maternally into paternally transmitted genomes of mussel (Mytilus) mitochondrial DNA. J Mol Evol 57:1–10 Rawson PD (2005) Nonhomologous recombination between the large unassigned region of the male and female mitochondrial genomes in the mussel, Mytilus trossulus. J Mol Evol 61: 717–732 Rayssiguier C, Thayler DS, Radman M (1989) The barrier to recombination between Eschericha coli and Salmonella typhymurium disrupted in mismatch-repair mutants. Nature 342:396–401 Rice P, Longden I, Bleasby A (2000) EMBOSS: the European molecular biology open software suite. Trends Genet 16: 276–277 Roff DA, Bentzen P (1989) The statistical analysis of mitochondrial DNA polymorphisms: chi-square and the problem of small samples. Mol Biol Evol 6:539–545 Ronquist F, Huelsenbeck JP (2003) MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19:1572–1574 Sambrock J, Fritsch EF, Maniatis T (1989) Molecular cloning: a laboratory manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY Sano N, Obata M, Komaru A (2007) Quantitation of the male and female types of mitochondrial DNA in a blue mussel, Mytilus galloprovincialis, using real-time polymerase chain reaction assay. Dev Growth Differ 49:67–72 Schwartz M, Vissing J (2002) Paternal inheritance of mitochondrial DNA. N Engl J Med 347:576–580 Shao R, Barker SC, Mitani H, Aoki Y, Fukunaga M (2004) Evolution of duplicate control regions in the mitochondrial genomes of metazoa: a case study with Australasian Ixodes ticks. Mol Biol Evol 22:620–629 Shao R, Mitani H, Barker SC, Takahashi M, Fukunaga M (2005) Novel mitochondrial gene content and gene arrangement indicate illegitimate inter-mtDNA recombination in the chigger mite, Leptotrombidium pallidum. J Mol Evol 60:764–773 Skibinski DOF, Gallagher C, Beynon CM (1994) Mitochondrial DNA inheritance. Nature 368:817–818 S´mietanka B, Zbawicka M, Wołowicz M, Wenne R (2004) Mitochondrial DNA lineages in the European populations of mussels Mytilus. Mar Biol 146:79–92 Stanton DJ, Daehler LL, Moritz CC, Brown WM (1994) Sequences with the potential to form stem-and-loop structures are

123

388 associated with coding-region duplications in animal mitochondrial DNA. Genetics 137:233–241 Sutherland B, Stewart D, Kenchington ER, Zouros E (1998) The fate of paternal mitochondrial DNA in developing female mussels, Mytilus edulis: implications for the mechanism of doubly uniparental inheritance of mitochondrial DNA. Genetics 148:341–348 Swofford DL (2003) PAUP*: Phylogenetic analysis using parsimony and other methods. Version 4. Sinauer Associates, Sunderland, MA Theologidis I, Fodelianakis S, Gaspar MB, Zouros E (2008) Doubly uniparental inheritance (DUI) of mitochondrial DNA in Donax trunculus (Bivalvia: Donacidae) and the problem of its sporadic detection in Bivalvia. Evolution 62:959–970 Thompson JD, Higgins DG, Gibson TJ (1994) CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res 22:4673– 4680

123

J Mol Evol (2008) 67:377–388 Tsaousis AD, Martin DP, Ladoukakis ED, Posada D, Zouros E (2005) Widespread recombination in published animal mtDNA sequences. Mol Biol Evol 22:925–933 Venetis C, Theologidis I, Zouros E, Rodakis GC (2007) A mitochondrial genome with a reversed transmission route in the Mediterranean mussel Mytilus galloprovincialis. Gene 406: 79–90 Wiuf C, Christensen T, Hein J (2001) a simulation study of the reliability of recombination detection methods. Mol Biol Evol 18:1929–1939 Zouros E, Ball AO, Saavedra C, Freeman KR (1994) Mitochondrial DNA inheritance. Nature 368:818 Zsurka G, Kraytsberg Y, Kudina T, Kornblum C, Elger CE, Khrapko K, Kunz W (2005) Recombination of mitochondrial DNA in skeletal muscle of individuals with multiple mitochondrial DNA heteroplasmy. Nat Genet 37:873–877