Association between chloroplast and mitochondrial lineages in oaks

1321

Mol. Biol. Evol. 15(10):1321–1331. 1998q 1998 by the Society for Molecular Biology and Evolution. ISSN: 0737-4038

Association Between Chloroplast and Mitochondrial Lineages in Oaks

S. Dumolin-Lapegue, M.-H. Pemonge, and R. J. PetitInstitut National de la Recherche Agronomique, Laboratoire de Genetique et Amelioration des Arbres Forestiers, Cestas,France

Patterns of chloroplast DNA (cpDNA) and mitochondrial DNA (mtDNA) variation were studied in 378 populationsof oak trees sampled throughout the southern half of France. Six cpDNA haplotypes detected in a previous Europeansurvey and three new cpDNA haplotypes were found in this region. Two mitochondrial polymorphisms detectedearlier by restriction analysis of PCR-amplified fragments alone, or in combination with single-strand conformationpolymorphism (SSCP), were compared with the cpDNA data. Sequencing revealed the nature of the two mito-chondrial mutations: a single-base substitution and a 4-bp inversion associated with a 22-bp hairpin secondarystructure. The single-base substitution was then analyzed by allele-specific amplification. Results for the two cy-toplasmic genomes were combined, which allowed the identification of 12 cpDNA-mtDNA haplotypes. The 4-bpmtDNA inversion has appeared independently in different cpDNA lineages. Given the peculiar nature of this mtDNAmutation, we suggest that intramolecular recombination leading to repeated inversions of the 4-bp sequence (ratherthan paternal leakage of one of the two genomes) is responsible for this pattern. Furthermore, the geographiclocations of the unusual cpDNA-mtDNA associations (due to the inversion) usually do not match the zones ofcontact between divergent haplotypes. In addition, in southern France, the groupings of populations based on themtDNA substitution were strictly congruent with those based on cpDNA. Because many populations that are poly-morphic for both cpDNA and mtDNA have remained in contact since postglacial recolonization in this area withoutproducing any new combination of cytoplasms involving the mitochondrial substitution, we conclude that paternalleakage is not a significant factor at this timescale. Such results confirm and expand our earlier conclusions basedon controlled crosses.

Introduction

In plants such as conifers, in which chloroplastsand mitochondria are transmitted by different parents,no strong association between the two genomes is ex-pected at the intraspecific level (Schnabel and Asmussen1989; Dong and Wagner 1994). In angiosperms, on thecontrary, chloroplasts and mitochondria are both mater-nally inherited in most species (Reboud and Zeyl 1994).As a consequence, the two organellar genomes shouldremain associated; that is, they should behave as if theyare completely linked (Schnabel and Asmussen 1989)and are expected to give similar information on dispersaland gene flow by seeds.

Occasional events of paternal leakage are difficultto observe by conventional inheritance studies (Milligan1992). In oaks, maternal inheritance for cpDNA andmtDNA was demonstrated in up to 143 progenies fromseveral controlled crosses (Dumolin, Demesure, and Pet-it 1995). However, statistical considerations indicate thatpaternal contributions as high as 2% cannot be excludedwith these sample sizes. In studies of organelle trans-mission in angiosperms, the bias toward maternal in-heritance is apparently stronger for mtDNA than forcpDNA, with more cases of biparental or paternal in-heritance reported for cpDNA and very few cases ofbiparental inheritance for mitochondria (Reboud and

Abbreviations: mtDNA, mitochondrial DNA; cpDNA, chloroplastDNA.

Key words: mitochondrial DNA, chloroplast DNA, inversion,Quercus robur complex, allele-specific amplification, postglacial re-colonization.

Address for correspondence and reprints: R. J. Petit, Institut Na-tional de la Recherche Agronomique, Laboratoire de Genetique etAmelioration des Arbres Forestiers, B.P. 45, F-33611 Gazinet Cedex,France. E-mail: [email protected].

Zeyl 1994). However, for most species, including thosefor which plastids seem to be completely absent fromthe male gametes (as in oaks; Hagemann and Schroder1989), some mitochondria from the male parent seem toreach the egg (Chapman 1986). It therefore seems dif-ficult to rule out a priori paternal leakage for either ofthe two organelles.

Theoretical studies indicate that rare events of pa-ternal transmission can have disproportionate impor-tance. For example, Petit, Kremer, and Wagner (1993a)have shown that paternal transmission rates as low as1% in cytoplasmic genomes can have a significant effecton their genetic structures, especially when there is anasymmetry between pollen and seed flow, a likely sit-uation in many plants, including oaks. According to Bir-ky (1978): ‘‘if there are very low levels of paternal genetransmission and recombination, these must be mea-sured for they may become important over evolutionarytime scales even though they are negligible when welook at the results of a single mating.’’ If paternal geneflow is so low that it cannot be measured in controlledcrosses, indirect assessments must be proposed.

If it could be demonstrated that both organelle lin-eages followed a strict parallel and clonal type of evo-lution over a long period, the hypothesis that they areboth strictly maternally inherited would be strengthened.Such a test should be possible by simultaneously ana-lyzing cpDNA and mtDNA polymorphisms for numer-ous individuals and populations in geographic regionswhere new combinations of the organelles would bereadily detected. However, to interpret such results, aprecise knowledge of the mutations involved, especiallyin the case of the mtDNA genome, is necessary. Indeed,due to the frequent and complicated rearrangements ofthe genome that are known to occur during plant mt-DNA evolution (Palmer 1992), recurrent mutationevents could easily compromise the interpretations.

1322 Dumolin-Lapegue et al.

Simultaneous information on the polymorphism ofboth organellar genomes has been obtained at a varietyof taxonomic levels for angiosperms (e.g., Berthou, Ma-thieu, and Vedel 1983; Holwerda, Jana, and Crosby1986; Ishikawa et al. 1992; Ishii et al. 1993; Laurent,Risterucci, and Lanaud 1993; Saumitou-Laprade et al.1993; Tsunewaki 1993). However, for plant specieswhere both organellar genomes are at least predomi-nantly maternally inherited, the association between thetwo lineages has not been explicitly studied, even whenapparently incongruent results were observed (Timothyet al. 1979).

In this study, we asked whether paternal transmis-sion of cpDNA or mtDNA genomes has occurred sincerecolonization of southern France by white oaks, about9,000 years ago. To answer this question, we screened378 populations of oaks with both cpDNA and mtDNAmarkers and studied the association between the two ge-nomes. Previous studies have identified several cpDNAlineages in these oaks that likely originated from differ-ent glacial refugia (Dumolin-Lapegue et al. 1997). Allthe main lineages meet in southern France. Furthermore,preliminary studies indicated that some of these cpDNAlineages also differed at two mtDNA variants. Hence,the prospect of detecting ‘‘recombinant’’ associations ofcpDNA and mtDNA in this region seemed to be partic-ularly good, if indeed paternal leakage of one of the twoorganelles does occur at low frequencies.

Materials and MethodsPlant Material

Four or five individuals from each of 378 oak pop-ulations (a total of 1,749 samples) were sampledthroughout southern France and analyzed with bothcpDNA and mtDNA markers. About 150 samples fromthe same area, which were included in a previouscpDNA study (Dumolin-Lapegue et al. 1997), were an-alyzed with the mtDNA fragments and the results re-ported here. The four oak species studied belong toQuercus subgenus Quercus, section Quercus (the whiteoaks), in the family Fagaceae (Nixon 1993). The mostabundant species were Quercus robur L. (pedunculateoak; 716 individuals) and Quercus pubescens Willd.(pubescent oak; 720 individuals), followed by Quercuspetraea (Matt.) Liebl. (sessile oak; 274 individuals) andQuercus pyrenaica Willd. (Pyrenean oak; 39 individu-als). The sampled trees were separated by at least 50 m;populations were located 20 to 30 km apart. As far aspossible, the material (acorns, buds, or leaves) was col-lected from old trees in ancient forests to reduce the riskof sampling planted populations. In this paper, no ad-ditional reference will be made to the species of the treesanalyzed, because a nearly systematic local sharing ofcytoplasms between species has been documented, aconsequence of extensive interspecific hybridization andintrogression (Petit, Kremer, and Wagner 1993b; Du-molin-Lapegue et al. 1997; Petit et al. 1997).

Restriction Analysis of PCR-Amplified FragmentsTotal DNA was extracted following the procedure

described by Dumolin, Demesure, and Petit (1995), ex-

cept that 1,4-dithiothreitol was used in place of b-mer-captoethanol. This DNA was used as a template in PCRreactions involving a set of conserved primers locatedin coding regions of cpDNA or mtDNA separated bypotentially more variable noncoding regions (mostly in-tergenic regions for cpDNA and introns for mtDNA).Three cpDNA primer pairs, trnD/trnT (DT), psaA/trnS(AS) (described by Demesure, Sodzi, and Petit 1995),and trnT/trnF (TF) (described by Taberlet et al. 1991),and two mtDNA primer pairs, nad1 exon2/nad1 exon3(nad1-2/3) and nad4 exon1/nad4 exon2 (nad4-1/2) (De-mesure, Sodzi, and Petit 1995) were used. The threepairs of cpDNA primers were chosen because they al-lowed the identification of all cpDNA haplotypes thathad been detected in France in a previous study (Du-molin-Lapegue et al. 1997). The two pairs of mtDNAprimers were selected because they allowed detection ofpolymorphisms in this part of the geographic range ofoaks, either by PCR-RFLP (restriction fragment lengthpolymorphism), for nad4-1/2, or by PCR-RFLP-SSCP,for nad1-2/3 (Dumolin, Demesure, and Petit 1995; Du-molin-Lapegue, Bodenes, and Petit 1996). The PCR pro-cedure was performed as described by Demesure, Sodzi,and Petit (1995). The PCR products (5 ml) were digestedwith 4-bp restriction endonucleases (5 U): TaqI and AluIfor fragment DT and HinfI for the other four fragments.The resulting restriction fragments were separated byelectrophoresis on 8% polyacrylamide gels as describedby Dumolin, Demesure, and Petit (1995).

DNA Sequencing

Sequencing of the mtDNA variants was performedin order to help distinguish between recurrent mutationsand true paternal contamination when new cpDNA-mtDNA associations were detected. To sequence themtDNA fragments, new pairs of primers were synthe-sized by adding a 17-bp forward M13/pUC sequence atthe 59 ends of the primers nad1 exon2 and nad4 exon1,and adding an 18-bp reverse M13/pUC sequence at the59 ends of the primers nad1 exon3 and nad4 exon2. PCRparameters were left unchanged, and mtDNA PCR prod-ucts were purified using a QIAquick PCR purificationkit (Qiagen). DNA sequencing was carried out using aLI-COR model 4000L automated DNA sequencer. Se-quencing reactions were performed with the SequithermEXCEL Long-Read DNA Sequencing Kit-LC (Epicen-tre Technologies), and fragments were separated onLongRanger (FMC) or Sequagel XR (National Diag-nostics) gels at a concentrations of 6.0% (for 41-cm-long gels) and 4.3% (for 66-cm- long gels). Sequencingwas performed in both directions with the M13/pUCforward and reverse primers. Additional internal se-quencing primers, designed using the OLIGO software,version 4.1 (Rychlik and Rhoades 1989), were used (onefor nad1 and two for nad4) because of the large sizesof the fragments.

PCR Amplification of Specific Alleles

The availability of the sequence of the nad1-2/3intron made it possible to use another technique toscreen for the nad1-2/3 mtDNA polymorphism in the

Association Between cpDNA and mtDNA Lineages 1323

FIG. 1.—Allele-specific amplification performed to study a substitution in the nad1-2/3 fragment. The three primers used are A1, A2, andA3. After amplification, the control and specific fragments are separated in agarose gels. The specific fragment is present only when the individualhas the nad1-2/3 b type.

oak populations. No commonly available restriction en-zyme would differentiate between the two alleles; there-fore, we adapted a technique of allele-specific amplifi-cation (Okayama et al. 1989; Wu et al. 1989). The so-lution that we selected is described in figure 1. We in-cluded a control fragment that spans the polymorphicsite in a priming competition with the specific fragment.As in the original protocol, two primers were designedto amplify one specific allele based on the sequence ob-tained for the nad1-2/3 fragment. Therefore, the first oli-gonucleotide (A1; fig. 1) was chosen such that the poly-morphic base was at its 39 end. A second primer (A2)was similarly designed for the other strand. Furthermore,we included a third primer (A3), which was designedfor the same strand as A1. PCR was performed (1 cycleof 4 min at 948C; 30 cycles of 45 s at 948C, 45 s at588C, 45 s at 728C) using these three primers after op-timization of the annealing temperature with a Robo-cycler Gradient 96 (Stratagene). Two fragments cantherefore be amplified: A1A2, if the 39 end of A1 match-es the DNA template, and A1A3, which should alwaysbe amplified and which acts as a PCR control (fig. 1).Further optimizations were needed because a single-basemismatch is not always sufficient to hinder elongation.First, MgCl2 concentration was decreased to 1 mM. Sec-ond, because a competition was observed between prim-ers A1 and A3 (yielding an excess of the fragmentA1A2), the concentration of A1 was decreased. The op-timized concentrations were 0.20 mM for A2, 0.02 mMfor A1, and 0.18 mM for A3. PCR products were sep-

arated in 1% agarose gels. High throughput wasachieved by using large agarose gels that could be load-ed up to three times with 96 samples each time (fig. 2).However, once the conditions had been optimized andmost samples had been analyzed, a modification of thepolymerase by the manufacturer resulted in nonspecificamplifications. Nevertheless, most trees were screenedfor this mutation, as detailed below.

Screening of the Populations

Fragments DT, TF, and nad4-1/2 were studied forall individuals (1,749) by PCR-RFLP with one restric-tion enzyme each, TaqI for DT and HinfI for the twoother fragments. Fragment AS was used only to distin-guish between two cpDNA haplotypes (1 and 2, see Du-molin-Lapegue et al. 1997). The unusual haplotypes de-tected in the survey were studied with the combinationDT-AluI to check whether they had a specific point mu-tation that groups cpDNA haplotypes of a particular lin-eage (Dumolin-Lapegue et al. 1997). The substitutionlocated in fragment nad1-2/3 was studied by allele-spe-cific amplification for a subset of 1,613 individuals(92% of all individuals).

ResultscpDNA Polymorphisms and cpDNA Haplotypes

Six of the 23 cpDNA haplotypes identified in anearlier study of cpDNA diversity in oak species through-out Europe (Dumolin-Lapegue et al. 1997) were found


FIG. 2.—Allele-specific amplification: separation of specific and control PCR fragments in 1% agarose gel. The picture shows one third ofthe agarose gel that was loaded three times successively. The molecular marker (M) was loaded only once during the first run. The specificfragment (A1A2) is present only when the individual has the nad1-2/3 b type, while the control fragment (A1A3) is always present.

in the region studied. In the previous study, cpDNA hap-lotypes were grouped into lineages (A, B, and C) on thebasis of numerous (47) cpDNA polymorphisms. Twolineages (A and C) are supported by high bootstrap val-ues based on parsimony analysis (99%; see fig. 3). Here,two new polymorphisms were detected in the DT frag-ment, distinguishing new cpDNA haplotypes, 25 and 26,attributed to lineages B and A, respectively. AnothercpDNA haplotype (24) belonging to lineage B was alsocharacterized from a new combination of fragments (ta-ble 1 and fig. 3). However, the three new cpDNA hap-lotypes (24, 25, and 26), in addition to cpDNA haplo-type 21, are very rare compared with cpDNA haplotypes1, 7, 10, 11, and 12. cpDNA haplotype 21 was found in15 individuals from 5 populations, cpDNA haplotype 24in 8 individuals from 3 populations, cpDNA haplotype25 in 1 individual, and cpDNA haplotype 26 in 3 indi-viduals from a single population.

Association Between the mtDNA Polymorphisms andthe cpDNA Haplotypes

Preliminary studies revealed a length variation innad4-1/2 digested by HinfI (Dumolin, Demesure, andPetit 1995) and a conformation polymorphism (SSCP)for nad1-2/3 digested by HinfI (Dumolin-Lapegue, Bod-enes, and Petit 1996). For most of the 23 cpDNA hap-lotypes from the previous European study, one individ-ual per haplotype had been screened for these twomtDNA mutations (Dumolin-Lapegue, Bodenes, andPetit 1996). To further study the association between thecpDNA and mtDNA haplotypes, the individuals sam-pled in southern France and already analyzed for cp-DNA variation were screened for these two mtDNApolymorphisms. The number of individuals character-ized by each cpDNA-mtDNA combination is given intable 2. Based on results for the nad1-2/3 fragment, an-alyzed by allele-specific amplification, all studied indi-

viduals (663) that belong to cpDNA lineage A (i.e.,cpDNA haplotypes 7 and 26) have the same mtDNAhaplotype (a), while all individuals that do not belongto lineage A possess the alternative mtDNA haplotype(b) (table 2). Similarly, all 702 individuals characterizedby cpDNA haplotypes 7 and 26 have the c mtDNA hap-lotype for the nad4-1/2 fragment, while all individualscharacterized by cpDNA haplotypes 12, 21, 24, and 25possess the d mtDNA haplotype. For cpDNA haplotypes1, 10, and 11, however, individuals have either type cor type d (table 2). In total, 12 cpDNA-mtDNA haplo-types were detected in this study (table 1). To distin-guish these haplotypes, the more abundant chloroplast-mitochondrial combinations are named according to thecpDNA haplotypes, whereas the three others are desig-nated 1 bis, 10 bis, and 11 bis. The distribution of thetwo variants for each of the two mtDNA markers isrepresented on the phylogenetic tree of the cpDNA hap-lotypes, redrawn from Dumolin-Lapegue et al. (1997),with the addition of the three new cpDNA haplotypes(fig. 3a and b).

Sequence Analysis of the mtDNA Polymorphisms

The nad4-1/2 fragment was sequenced for 12 in-dividuals, one from each of the cpDNA-mtDNA hap-lotypes. All sequences were 2,064 bp in length. The onlydifference detected between the sequences was at posi-tions 1096–1099, where mtDNA haplotype c has thesequence GAAA and mtDNA haplotype d has TTTC.The presence of a hairpin secondary structure was in-ferred in the polymorphic region (see fig. 4) using OLI-GO, version 4.1. The hairpin stem was 22 bp in length,which corresponds to a free- energy change (DG) of226.5 kcal/mol, that is, a high probability of sponta-neous formation. Such hairpins are likely to increase theprobability of intramolecular recombination (Kelchnerand Wendel 1996). In the case of recombination inside


FIG. 3.—a, Distribution of the mitochondrial polymorphism detected in fragment nad1-2/3 (substitution) on the neighbor-joining chloroplasttree. Numbers correspond to the 26 cpDNA haplotypes. Those underlined were found in southern France. Lineage A includes five cpDNA types(4, 5, 6, 7, and 26), lineage B includes five (10, 11, 12, 24, and 25), and lineage C includes three (1, 2, and 3). The two highest bootstrapvalues based on parsimony analysis are indicated. Allele a of fragment nad1-2/3 is represented by black squares, and allele b is rpresented bywhite squares (table 1). Although the mtDNA haplotype a seems to have appeared two or three times, it must be noted that the positions ofhaplotypes 20, 21, and 22 are poorly resolved, and they may therefore form a single clade with lineage A. In contrast to the other mtDNAmutation (see b), no cpDNA haplotype is polymorphic for the two alleles. b, Distribution of the mitochondrial polymorphism detected infragment nad4-1/2 (inversion) on the neighbor-joining chloroplast tree. Notations are as in a, except that allele c of fragment nad4-1/2 isrepresented by white circles, allele d is represented by black circles, and half-filled circles indicate that both alleles were found to be associatedwith the corresponding cpDNA haplotype. It it clear that the cpDNA and the mtDNA trees are not congruent.

Table 1Description of the cpDNA, mtDNA, and Combined cpDNA-mtDNA Haplotypes Found in this Study

CPDNALINEAGEa

CPDNAb

DTTaq1

DTTaq2

DTTaq3

DTTaq3/ac

DTAlu1d

ASHinf2

TFHinf8

cpDNAHaplotype

MTDNA

nad1-2/3PASAe

nad4-1/2Hinf1

CP-MTDNAHAPLOTYPEf

C. . . . . . . . . . . . . 9 1 2 — 1 1 2 1 b c 19 1 2 — 1 1 2 1 b d 1 bis

A. . . . . . . . . . . . . 1 1 9 1 1 1 2 7 a c 71 1 9 2 1 1 2 26 a c 26

B. . . . . . . . . . . . . 1 2 3 — 9 1 2 10 b d 101 2 3 — 9 1 2 10 b c 10 bis1 1 3 — 9 1 2 24 b d 241 2 2 — 9 1 2 25 b d 251 2 3 — 9 1 1 11 b c 111 2 3 — 9 1 1 11 b d 11 bis1 2 4 — 9 1 2 12 b d 12

— . . . . . . . . . . . . 1 1 3 — 1 1 2 21 b d 21

a cpDNA lineages defined by phylogenetic analyses according to Dumolin-Lapegue et al. (1997).b The abbreviated name of the PCR fragment is followed by the name of the restriction enzyme used and the number of the polymorphic fragment, ranked from

the longest (1) to the shortest. The same principle is used to label the length variants, except that 9 indicates a point mutation. New cpDNA haplotypes andcorresponding mutations are shown in boldface.

c The largest of the two fragments generated by the point mutation in fragment DT Taq3 has two length variants.d The point mutation detected in the DT fragment at an AluI site was used to distinguish haplotype 24 from haplotype 21, a polymorphism that distinguishes

the haplotypes of the B lineage from those of the other lineages.e This mtDNA fragment was studied with PCR-amplification of a specific allele.f As long as only one chloroplast-mitochondrial combination is detected, the cpDNA-mtDNA haplotype takes the name of the cpDNA haplotype. New combi-

nations are called ‘‘bis’’ and are underlined.


Table 2Number of Individuals Having Each mtDNA HaplotypeAccording to Their cpDNA Haplotype

LineagecpDNA

Haplotypenad1-2/3 a1

nad1-2/3 b1

nad4-1/2 c

nad4-1/2 d

C . . . . . . . 1 0 142 92 51A . . . . . . . 7 660 0 699 0

26 3 0 3 0B . . . . . . . 10 0 470 4 518

24 0 8 0 825 0 1 0 111 0 225 250 712 0 93 0 101

—a. . . . . . 21 0 11 0 15

NOTE.—Cases of polymorphic mtDNA within a cpDNA haplotype areshown in boldface.

a A haplotype not grouped in a lineage.

FIG. 4.—Association of the 4-bp inversion detected in the nad4-1/2 fragment with a 22-bp stem-loop hairpin. After the formation of thehairpin, an intramolecular recombination at this level leads to the inversion of the 4 bp at the edge, which generates the two alternativemitochondrial types c and d.

the hairpin, this would lead to an inversion of the 4-bploop, yielding the two observed mtDNA haplotypes. In-deed, the 4-bp loop exactly matches the four polymor-phic bases, and GAAA is the reverse complement of thesequence TTTC.

The nad1-2/3 fragment was completely sequencedfor four individuals characterized by the followingcpDNA-mtDNA haplotypes: 7 (two individuals), 1, and1 bis (one individual each). Again, the fragment had thesame length for the four individuals (i.e., 1,661 bp). Theonly difference between them was a 1-bp substitution at

position 941, with an A in mtDNA haplotype a and aT in mtDNA haplotype b.

Geographic Distribution of the Haplotypes

The geographic distribution of the 12 cpDNA-mtDNA haplotypes is clearly not random (fig. 5): hap-lotypes from lineage B are restricted to the western partof the studied region, while those from lineage A are inthe eastern part, and those from lineage C are in thesoutheastern part. Furthermore, although both haploty-pes 1 and 1 bis from lineage C are mostly restricted tothe southeastern part of France, each occupies a separatearea. The three most common haplotypes of lineage B(i.e., 10, 11, and 12) have similar, but not identical, dis-tributions: haplotype 11 is abundant in a region from themouth of the Garonne to the Massif Central and up tothe Rhone valley, whereas haplotype 12 is frequentalong the Atlantic coast, and haplotype 10 has a muchwider distribution in the western part of the studied re-gion. Finally, haplotype 10 bis is found in two distantpopulations, one in the center of the studied region andone north of the mouth of the Garonne, and haplotype11 bis is found in three populations located close to oneanother to the north of the mouth of the Garonne. Al-though haplotype 7 of lineage A is very abundant ineastern France, it was also found as far west as the At-lantic coast, in the region where lineage B is widespread.Such an overlap between lineages A and B is also ap-


FIG. 5.—Distribution of the 12 cpDNA-mtDNA haplotypes detected in this study of oaks in southern France. Each color corresponds to ahaplotype, and the size of the circle is proportional to the number of individuals analyzed. The ‘‘bis’’ haplotypes are all distinguished by stripes.

parent in the northern part of the studied area, wherehaplotype 10 appears to have penetrated a region inwhich haplotype 7 is abundant. Only a few exceptionsto this overall pattern were apparent, with some haplo-types of one lineage located far from their main distri-bution, often mixed with the presumably local haploty-pe. Moreover, populations fixed for haplotype 1 arefound in the northeastern part of the studied area andaround the Mediterranean coast.

Discussion

Until recently, assessment of inter- and intraspecificcpDNA diversity has largely relied on RFLP analysisusing Southern hybridization experiments. In plantmtDNA, intraspecific variation is detectable by RFLPanalysis, but the observed mutations likely result fromstructural rearrangements. Although this variation maybe very useful for other applications, it is not appropri-ate for phylogenetic or other types of evolutionary stud-ies, even at low taxonomic levels (Palmer 1992). Poly-morphisms revealed by Southern hybridization are dif-ficult to characterize further by sequencing, especially ifthey fall outside of the sequence that is homologous to

the probe. Here, we used only PCR-derived techniquescombined, when necessary, with sequencing. However,even PCR polymorphisms are difficult to find at lowtaxonomic levels. Chloroplast DNA evolves four timesas slowly as the plant nuclear genome (Wolfe, Li, andSharp 1987), and this trend is even more pronounced inlong-lived species such as those of Fagaceae (Frascariaet al. 1993). Nevertheless, intraspecific cpDNA variationin angiosperms has been demonstrated for many species,including forest trees, using more sensitive moleculartechniques (Taberlet et al. 1991; Demesure, Sodzi, andPetit 1995, Dumolin-Lapegue et al. 1997; Soltis et al.1997). It must be pointed out that such techniques detectnot only point mutations, but also insertion/deletions incpDNA, and some sites of noncoding cpDNA sequencesactually appear to be very labile to small insertion/de-letion mutations (Clegg et al. 1994), which can lead toparallel or recurrent mutation events at those sites. Nev-ertheless, length mutations have provided assessmentsof intraspecific phylogenetic relationships that are oftenconcordant with those obtained via restriction site anal-ysis (e.g., Doebley, Ma, and Renfroe 1987; Gielly andTaberlet 1994).


FIG. 6.—Schematic distribution of the main lineages and probable directions of expansion from the inferred refugia (Spanish, Italian, andBalkanic). The locations of the ‘‘bis’’ haplotypes corresponding to the unusual cpDNA-mtDNA combinations are represented.

Methods such as PCR-RFLP, which were success-fully used for cpDNA, proved to be less useful formtDNA (Dumolin-Lapegue, Bodenes, and Petit 1996).Indeed, in contrast to animal mtDNA, plant mtDNA isvery conserved in primary sequence, evolving threetimes as slowly as cpDNA (Wolfe, Li, and Sharp 1987).However, the SSCP technique can be combined withPCR-RFLP to detect rare polymorphisms in plantmtDNA (Dumolin-Lapegue, Bodenes, and Petit 1996).Once detected, SSCP variants can be sequenced to allowthe subsequent design of a rapid method of allele detec-tion.

Our sequence data confirm this low rate of evolu-tion of plant mtDNA. The sequencing of 3,725 bp fromintervening noncoding sequences from several oak treesbelonging to different cpDNA lineages did not revealpolymorphisms in addition to the two that had been de-tected previously. The two mtDNA polymorphisms ap-pear radically different in terms of their nature and theirrelationships with the cpDNA polymorphisms. In south-ern France, the polymorphism observed in the nad4-1/2 fragment shows at least three cases of uncoupling withthe three common cpDNA haplotypes (table 2 and figs.5 and 6). When plotted onto the cpDNA phylogenetictree, this polymorphism also appears to be quite inde-pendent of the cpDNA lineages. Although all individu-als from cpDNA lineage A have the same mtDNA vari-ant, lineages B and C differ for this mtDNA polymor-phism, and four cpDNA haplotypes are polymorphic forthe mtDNA marker (fig. 3b). Such disassociation be-tween cpDNA and mtDNA variation may result fromoccasional paternal leakage in the transmission of oneor both organelles. However, the nature of the mtDNAmutation involved can be a key to understanding thisdisassociation. The mtDNA polymorphism is very likelythe result of an inversion associated with a particularlylong hairpin secondary structure. Given the length of the

hairpin, intramolecular recombination is likely to occurrepeatedly, probably even at the intraspecific level. Sim-ilar, although shorter, inversions have been identified inthe noncoding regions of cpDNA (Golenberg et al.1993; Kelchner and Wendel 1996), but none has beenreported so far in plant mtDNA. Hence, we consider thatthis observed cpDNA-mtDNA uncoupling may resultfrom repeated mtDNA recombination and is likely notdue to occasional events of paternal leakage.

The geographical distribution of the haplotypesand, especially, the localization of the new cpDNA-mtDNA combinations (involving the 4-bp mtDNA in-version) may provide additional insights. Indeed, ifthese new combinations were systematically found inareas of contact between complementary haplotypes, thehypothesis of paternal leakage could nevertheless re-ceive support.

The overall genetic structure that exists today ap-pears to have been built up largely during the last post-glacial recolonization. The most likely postglacial re-colonization routes for oaks throughout Europe were in-ferred in a previous study (Dumolin-Lapegue et al.1997). Three main cpDNA lineages were identified, andthese were probably initially located in different refugiaduring the last glaciation before spreading over the restof Europe. In the present study focusing on southernFrance, the probable routes of recolonization are furtherrefined (fig. 6). Haplotypes from lineage B have beentreated together, because they are likely to have origi-nated from the same Iberian refugium. However, eachof the three main haplotypes (10, 11, and 12) that belongto this lineage can dominate locally, yielding a patchydistribution as previously described (Petit et al. 1997).The distribution of haplotype 7 from lineage A confirmsits eastern origin (possibly in the Balkans), with a prob-able northern skirting of the Alps. Two extensions ofhaplotype 7 farther west are also apparent, in regions


where haplotypes of lineage B are abundant. Recipro-cally, haplotypes of lineage B have also extended fartherto the east (fig. 6). Haplotypes 1 and 1 bis are principallylimited to the southeastern part of France, in agreementwith the hypothesis that they originate from an Italianrefugium and that they migrated in France along theMediterranean coast. However, two isolated populationsfixed for haplotype 1 are found farther north, near theSwiss border. In this case, a Swiss origin is possible,because this haplotype has crossed the Alps to extendnorthward in Europe (Dumolin-Lapegue et al. 1997).Populations present farther west, on the other hand,could indicate another expansion along the Mediterra-nean coast up to northeastern Spain, where haplotype 1has been detected (Dumolin-Lapegue et al. 1997). How-ever, these hypotheses should be confirmed by addition-al analyses in these regions and in the neighboring coun-tries.

The localization of the new cpDNA-mtDNA com-binations can be examined within this general frame-work (fig. 6). In our earlier work (Dumolin-Lapegue etal. 1997), we concluded that most cpDNA mutations hadoccurred before recolonization. Here, given the limiteddistribution of all but one cpDNA-mtDNA haplotype(haplotype 1 bis, also present in Italy, data not shown),a more recent origin of the new combinations is morelikely. If these new combinations have appeared locally,then an examination of the surrounding haplotypes infigures 5 and 6 may give some hints as to their origins.Oak trees characterized by haplotype 11 bis near themouth of the Garonne are surrounded by individuals ofhaplotypes 10 and 12 that could have contributed theirmitochondria to haplotype 11 (also present in the area),yielding this new combination. Conversely, haplotype10 bis occurs in two disconnected areas in which nohaplotype could have contributed the new mitochondrialhaplotypes. First, in the northwestern portion of thestudied area, haplotype 10 bis is mixed with haplotype10. The second region, in which haplotype 10 bis wasfound (in the south-central portion of the studied area;see figs. 5 and 6), is dominated by haplotype 7. Al-though in this case, a local origin for haplotype 10 bisis dubious, it remains that a combination involving hap-lotype 10 with haplotype 7 is unlikely to have yieldedhaplotype 10 bis. Indeed, this would mean that haplo-type 7 had contributed only part of its mtDNA (cf. table1). In all, there is only one out of three situations wherea mixing of preexisting haplotypes may have yielded thenew combination. Hence, based on this mtDNA inver-sion, it is difficult to conclude whether or not paternaltransmission or recurrent mutation is responsible for theuncoupling of cpDNA and mtDNA haplotypes. How-ever, the likelihood of paternal transmission seems scantat best, given the molecular structure of the region sur-rounding the inversion and the apparently random geo-graphic localization of the new combinations.

The other mtDNA fragment (the second intron ofthe nad1 gene) separates cpDNA haplotypes of one lin-eage (A) from all other haplotypes except 18, 20, and22 (fig. 3a). Note that the cpDNA tree is a neighbor-joining tree based on a majority of insertion/deletions

that are abundant and much easier to identify, althoughprobably more prone to recurrence than point mutations.Hence, the groupings that are not supported by highbootstrap values must be considered with caution (morebootstrap values are represented in fig. 3 of Dumolin-Lapegue et al. 1997). In southern France, no newcpDNA-mtDNA associations involving the nad1 substi-tution were detected among all individuals analyzed.This is remarkable, considering the importance of thezone of contact between haplotypes of lineage A (shownin blue in fig. 5) and all other haplotypes. Indeed, notethat when trees characterized by these contrasting hap-lotypes mate (and there is every reason to believe thatthis has occurred; see Petit et al. 1997), any case ofpaternal leakage involving either cpDNA or mtDNAwill necessarily result in a new haplotype. In our sample,there were as many as 33 cases of populations mixedwith such contrasting haplotypes. In addition, this con-tact must have existed since recolonization took place,approximately 9,000 years ago (Huntley and Birks1983). Despite the cumulative effects of at least 50–100generations of sexual reproduction, no new combina-tions involving the mtDNA substitution could be de-tected.

Conclusions

A population genetics survey of organellar ge-nomes of plants requires techniques to detect sufficientpolymorphism to subsequently be adapted for screeningsurveys. Here, polymorphisms in both organellar ge-nomes of oaks were studied in southern France basedon an extensive population sample, and the associationof the two maternal lineages was systematically inves-tigated. One mtDNA mutation (a 4-bp inversion locatedin the first intron of the gene nad4) was partly indepen-dent of the cpDNA polymorphisms. This partial inde-pendence is likely due to recurrent inversions that resultfrom the secondary structure of this region of the mi-tochondrial genome. On the other hand, greater congru-ence was observed between the chloroplast and mito-chondrial genomes for the second mtDNA mutationstudied (a substitution within the second intron of thegene nad1). Overall, it appears that, at least during aperiod that encompasses 50–100 generations of oaks,despite a zone of contact between complementary hap-lotypes that spans several hundred kilometers, no caseof well-established paternal leakage of either organellecould be demonstrated. This strengthens preliminary re-sults from controlled crosses that indicate that both cy-toplasmic genomes were, without exception, maternallyinherited. However, simultaneous studies of both mt-DNA and cpDNA from other angiosperm species maygive contrasting results, because inheritance patterns canvary among species. As expected, the distinctive fea-tures of plants characterized by two extranuclear ge-nomes can provide specific evolutionary insights absentfrom studies of other eukaryotes characterized by a sin-gle extranuclear genome.


Acknowledgments

We particularly thank the following people, whoparticipated in the collection of oak material: E. Bertoc-chi, S. Dumas, A. Ducousso, V. Le Corre, J.-M. Louvet,R. Roubeyrie, G. Roussel, and A. Zanetto. Commentson an earlier version of the manuscript from A. Gillies,J. Cottrell, L. Johnson, and one anonymous reviewerwere greatly appreciated. The research was supported bythe EC research program FAIR1-CT950297 and the Ac-tion Concertee Coordonnee (ACC), Sciences du Vivant,number 7, from the Ministere de l’Education Nationale,de l’Enseignement Superieur, de la Recherche et del’Insertion Professionnelle.

LITERATURE CITED

BERTHOU, F., C. MATHIEU, and F. VEDEL. 1983. Chloroplastand mitochondrial DNA variation as indicator of phylo-genetic relationships in the Genus Coffea L. Theor. Appl.Genet. 65:77–84.

BIRKY, C. W. JR. 1978. Transmission genetics of mitochon-dria and chloroplasts. Annu. Rev. Genet. 12:471–512.

CHAPMAN, G. P. 1986. Mitochondrial delivery via the malegametophyte and the prospects for recombination. Pp.61–68 in S. H. MANTELL, G. P. CHAPMAN, and P. F. S.STREET, eds. The chondriome. Longman, London.

CLEGG, M. T., B. S. GAUT, G. H. LEARN JR., and B. R.MORTON. 1994. Rates and patterns of chloroplast DNAevolution. Proc. Natl. Acad. Sci. USA 91:6795–6801.

DEMESURE, B., N. SODZI, and R. J. PETIT. 1995. A set ofuniversal primers for amplification of polymorphic non-coding regions of mitochondrial and chloroplast DNA inplants. Mol. Ecol. 4:129–131.

DOEBLEY, J. F., D. P. MA, and W. T. RENFROE. 1987. Inser-tion/deletion mutations in the Zea chloroplast genome.Curr. Genet. 11:617–624.

DONG, J., and D. B. WAGNER. 1994. Paternally inheritedchloroplast polymorphism in Pinus: estimation of diver-sity and population subdivision, and tests of disequilib-rium with a maternally inherited mitochondrial polymor-phism. Genetics 136:1187–1194.

DUMOLIN, S., B. DEMESURE, and R. J. PETIT. 1995. Inheri-tance of chloroplast and mitochondrial genomes in pe-dunculate oak investigated with an efficient PCR method.Theor. Appl. Genet. 91:1253–1256.

DUMOLIN-LAPEGUE, S., C. BODENES, and R. J. PETIT. 1996.Detection of rare polymorphims in mitochondrial DNAof oaks with PCR-RFLP combined to SSCP analysis.Forest Genetics 3:227–230.

DUMOLIN-LAPEGUE, S., B. DEMESURE, S. FINESCHI, V. LE

CORRE, and R. J. PETIT. 1997. Phylogeographic structureof white oaks throughout the European continent. Ge-netics 146:1475–1487.

FRASCARIA, N., L. MAGGIA, M. MICHAUD, and J. BOUSQUET.1993. The rbcL gene sequence from chestnut indicates aslow rate of evolution in the Fagaceae. Genome 36:668–671.

GIELLY, L., and P. TABERLET. 1994. The use of chloroplastDNA to resolve plant phylogenies: noncoding versusrbcL sequences. Mol. Biol. Evol. 11:769–777.

GOLENBERG, E. M., M. T. CLEGG, M. L. DURBIN, J. DOE-BLEY, and D. P. MA. 1993. Evolution of a non-codingregion of the chloroplast genome. Mol. Phylogenet. Evol.2:52–64.

HAGEMANN, R., and M.-B. SCHRODER. 1989. The cytologi-cal basis of the plastid inheritance in angiosperms. Pro-toplasma 152:57–64.

HOLWERDA, B. C., S. JANA, and W. L. CROSBY. 1986. Chlo-roplast and mitochondrial DNA variation in Hordeumvulgare and Hordeum spontaneum. Genetics 114:1271–1291.

HUNTLEY, B., and H. J. B. BIRKS. 1983. An atlas of pastand present pollen maps for Europe, 0–13,000 years ago.Cambridge University Press, Cambridge, England.

ISHII, T., T. TERACHI, N. MORI, and K. TSUNEWAKI. 1993.Comparative study on the chloroplast, mitochondrial andnuclear genome differentiation in two cultivated rice spe-cies, Oryza sativa and O. glaberrima, by RFLP analyses.Theor. Appl. Genet. 86:88–96.

ISHIKAWA, S., S. KATO, T. IMAKAWA, and Y. SHIMAMOTO.1992. Organelle DNA polymorphism in apple cultivarsand rootstocks. Theor. Appl. Genet. 83:963–967.

KELCHNER, S. A., and J. F. WENDEL. 1996. Hairpins createminute inversions in non-coding regions of chloroplastDNA. Curr. Genet. 30:259–262.

LAURENT, V., A. RISTERUCCI, and C. LANAUD. 1993. Chlo-roplast and mitochondrial DNA diversity in Theobromacacao. Theor. Appl. Genet. 87:81–88.

MILLIGAN, B. G. 1992. Is organelle DNA strictly maternallyinherited? Power analysis of a binomial distribution. Am.J. Bot. 79:1325–1328.

NIXON, K. C. 1993. Infrageneric classification of Quercus(Fagaceae) and typification of sectional names. Ann. Sci.For. 50(Suppl. 1):25s–34s.

OKAYAMA, H., D. T. CURIEL, M. L. BRANTLY, M. D.HOLMES, and R. G. CRYSTAL. 1989. Rapid nonradioac-tive detection of mutations in the human genome by al-lele-specific amplification. J. Lab. Clin. Med. 114:105–113.

PALMER, J. D. 1992. Mitochondrial DNA in plant system-atics. Pp. 36–49 in P. S. SOLTIS, D. E. SOLTIS, and J. J.DOYLE, eds. Molecular systematics of plants, Chapmanand Hall, New York.

PETIT, R. J., A. KREMER, and D. B. WAGNER. 1993a. Finiteisland model for organelle and nuclear genes in plants.Heredity 71:630–641.

. 1993b. Geographic structure of chloroplast DNApolymorphisms in European oaks. Theor. Appl. Genet.87:122–128.

PETIT, R. J., E. PINEAU, B. DEMESURE, R. BACILIERI, A.DUCOUSSO, and A. KREMER. 1997. Chloroplast DNAfootprints of postglacial recolonization by oaks. Proc.Natl. Acad. Sci. USA 94:9996–10001.

REBOUD, X., and C. ZEYL. 1994. Organelle inheritance inplants. Heredity 72:132–140.

RYCHLIK, W., and R. E. RHOADES. 1989. A computer pro-gram for choosing optimal oligonucleotides for filter hy-bridization, sequencing and in vitro amplification ofDNA. Nucleic Acids Res. 17:8543–8551.

SAUMITOU-LAPRADE, P., G. J. A. ROUWENDAL, J. CUGUEN,F. A. KRENS, and G. MICHAELIS. 1993. Different CMSsources found in Beta vulgaris ssp maritima: mitochon-drial variability in wild populations revealed by a rapidscreening procedure. Theor. Appl. Genet. 85:529–535.

SCHNABEL, A., and M. A. ASMUSSEN. 1989. Definition andproperties of disequilibria within nuclear-mitochondrial-chloroplast and other nuclear-dicytoplasmic systems. Ge-netics 123:199–215.

SOLTIS, D. E., M. A. GITZENDANNER, D. D. STRENGE, andP. S. SOLTIS. 1997. Chloroplast DNA intraspecific phy-


logeography of plants from the Pacific Northwest ofNorth America. Plant Syst. Evol. 206:353–373.

TABERLET, P., L. GIELLY, G. PAUTOU, and J. BOUVET. 1991.Universal primers for amplification of three non-codingregions of chloroplast DNA. Plant Mol. Biol. 17:1105–1109.

TIMOTHY, D. H., C. S. LEVINGS III, D. R. PRING, M. F. CON-DE, and J. L. KERMICLE. 1979. Organelle DNA variationand systematic relationship in the genus Zea: teosinte.Proc. Natl. Acad. Sci. USA 76:4220–4224.

TSUNEWAKI, K. 1993. Genome-plasmon interactions inwheat. Jpn. J. Genet. 68:1–34.

WOLFE, K. H., W.-H. LI, and P. M. SHARP. 1987. Rates ofnucleotide substitution vary greatly among plant mito-chondrial, chloroplast, and nuclear DNAs. Proc. Natl.Acad. Sci. USA 84:9054–9058.

WU, D. Y., L. UGOZZOLI, B. K. PAL, and R. B. WALLACE.1989. Allele-specific enzymatic amplification of b-globingenomic DNA for diagnosis of sickle cell anemia. Proc.Natl. Acad. Sci. USA 86:2757–2760.

PAMELA SOLTIS, reviewing editor

Accepted June 29, 1998

Association between chloroplast and mitochondrial lineages in oaks

Documents

Transcript of Association between chloroplast and mitochondrial lineages in oaks