Copyright 0 1991 by the Genetics Society of America

Evolutionary Dynamics of Mitochondrial DNA Duplications in Parthenogenetic Geckos, Heteronotia binoei

Craig Moritz Department of Zoology, University of Queensland, Brisbane, Queensland, Australia 4072

Manuscript received August 22, 1990 Accepted for publication May 1 1 , 199 1

ABSTRACT Mitochondrial DNA (mtDNA) from triploid parthenogenetic geckos of the Heteronotia binoei

complex varies in size from 17.2 to 27.6 kilobases (kb). Comparisons of long us. short genomes using restriction endonucleases revealed a series of tandem direct duplications ranging in size from 1.2 to 10.4 kb. This interpretation was supported by transfer-hybridization experiments which also demon- strated that coding sequences were involved. Some of the duplications have been modified by deletion and restriction site changes, but no other rearrangements were detected. Analysis of the phylogenetic and geographic distribution of length variation suggests that duplications have arisen repeatedly within the parthenogenetic form of H. binoei. The parthenogens, and thus the duplications, are of recent origin; modifications of the duplicated sequences, particularly by deletion, has therefore been rapid. The absence of duplications from the mtDNA of the diploid sexual populations of H. binoei reinforces the correlation between nuclear polyploidy and duplication of mtDNA sequences reported for other lizards. In comparison to the genomes of sexual H. binoei and of most other animals, the mtDNA of these parthenogenetic geckos is extraordinarily variable in length and organization.

C ONTRARY to previous assertions of extreme economy and conservative organization (WAL-

LACE 1982; SEDEROEF 1984; ATTARDI 1985), there have been several recent reports of animal mitochon- drial DNAs (mtDNAs) with large-scale size variation attributable to repetitive sequences (Table 1). To- gether with numerous reports of intra-individual size polymorphism (heteroplasmy, Table 1 and BIRMING- HAM, LAMB and AVISE 1986) and variation in gene order (WOLSTENHOLME et al. 1985; DUBIN, HSU-CHEN and TILLOTSEN 1986; HAUCKE and GELLISSEN 1988; GAREY and WOLSTENHOLME 1989; DESJARDINS and MORAIS 1990) these studies reveal a genome more plastic than previously supposed. It now seems that the duplication and transposition of sequences is an important mode of mtDNA evolution (MORITZ, DOWLING and BROWN 1987; JACOBS et al. 1989; CAN- TATORE et al. 1987).

Most of the length variation that has been charac- terized is due to variation in the number of copies of tandemly repeated noncoding sequences from the control region (Table 1). Repeat units of this type vary in size from 64 base pairs (bp) in mtDNA from Cnemidophorus lizards (DENSMORE, WRIGHT and BROWN 1985) to 2.0 kb in bark weevils (BOYCE, ZWICK and AQUADRO 1989). In two groups with exception- ally large mtDNA, scallops (SNYDER et al. 1987; LA ROCHE et al. 1990) and bark weevils, the additional DNA is largely due to these tandemly repeated se- quences. Several studies have suggested that copy number of the noncoding repeats changes rapidly,

(knetics 129: 221-230 (September, 1991)

acting against the strong genetic drift of oocyte mtDNA, to maintain heteroplasmy (DENSMORE, WRIGHT and BROWN 1985; RAND and HARRISON 1986; BUROKER et al. 1990). Analysis of segregation among the progeny of heteroplasmic females suggests that smaller molecules may be at an advantage (RAND and HARRISON 1986), although in Drosophila the bias varies with the age of the female (SOLIGNAC et al. 1987). Comparisons among related species of lizards (DENSMORE, WRIGHT and BROWN 1985; DENSMORE et al. 1989; MORITZ, WRIGHT and BROWN 1989), crick- ets (RAND and HARRISON 1989) and bark weevils (BOYCE, ZWICK and AQUADRO 1989) revealed that these noncoding tandem repeats are often maintained through speciation events. Thus, any selection against large molecules must be balanced by recurrent muta- tion.

A different form of length variation, involving the tandem duplication of coding sequences, has been reported in mtDNAs from newts, nematodes, and lizards (Table 1). These duplications vary in size from 1.1 to 8.0 kb and typically span or flank the control region. All known duplications of animal mtDNA coding sequences are tandem and direct, except in the nematode Romanomermis (HYMAN, BECK and WEISS 1988) where there appears to be a partial, inverted copy disjunct from three tandem direct repeats. In contrast to the noncoding repeats, duplications of coding sequences are rarely heteroplasmic and appear to be ephemeral in that not one is shared among

222 C. Moritz

TABLE 1

Characteristics of repeated sequences (>20 bp) in animal mtDNA

Shared between

Species Size species? Copy No. content Heteroplasmy Location or

Reference

1. Noncoding sequences Cnemidophorus spp. Acipens transmontanus Gryllus spp. Drosophila spp. Pissodes spp. Alosa sapidissma Placopecten magellias

2. Coding sequences Cnemidophurus spp. Triturus Romanomermis

64 bp 82 bp

206 bp 470 pb

1.5 kb 1.4 kb

1.5-8.0 kb

3.0 kb

0.8-2.0 kb

1.1-8.5 kb

Yes 3-9 ? 1-4

Yes 2-7 Yes 2-6 Yes

? 2-3 2-7

No 2 NO 2-3

? 3-5

Control region Control region Control region Control region Control region Control region

?

Variable Variable

?

15/92 521128

1471319 17/92

2191219 301244 181250

1/43

None 213

DENSMORE, WRIGHK and BROWN (1 985) BUROKER et al. (1 990) RAND and HARRISON (1989) SOLIGNAC et al . (1986); HALE and SINGH (1 986) BOYCE, ZWICK and AQUADRO (1 989) BENTZEN, LEGGETT and BROWN ( 1 988) LA ROCHE et a l . (1990)

MORITZ and BROWN (1 987) WALLIS (1 987) HYMAN, BECK and WEISS (1988)

closely related species (Table 1) (MORITZ, DOWLING and BROWN 1987).

Previously reported duplications of mtDNA coding sequences have been taxonomically or geographically isolated, offering little scope for analysis of their evo- lutionary dynamics. In particular, there is no evidence on the form and rate of sequence evolution within repeated mtDNA genes. Is there selection for a return to small genome size? Does the presence of redundant sequences permit forms of sequence evolution (re- viewed in BROWN 1985) not usually seen in animal mtDNA?

This report concerns variation in the size and dis- tribution of large tandem duplications of coding se- quences in mtDNA from parthenogenetic (all-female) geckos of the Heteronotia binoei complex. These par- thenogens are triploid and arose via multiple inde- pendent hybridization events involving two chromo- some races (“CA6” and “SM6”) of sexual H. binoei (Figure 1) (MORITZ 1983). The parthenogenetic line- ages are now distributed throughout most of the central and western deserts of Australia and, for a parthenogenetic vertebrate, have extraordinarily high genetic (allozymic and chromosomal) diversity within and between lineages (MORITZ et al. 1989). There are two major chromosome lineages that differ in the dosage of the parental genes and which are thought to represent the result of different hybridizations (Fig- ure 1). In contrast to the variability of the nuclear genome, mtDNAs from the parthenogens have few restriction site differences other than those due to length changes. These data suggest that the parthen- ogens arose recently, i e . , within the last few thousand years, from a small geographic area, probably in the west (MORITZ 1991).

The studies on the parthenogens provide a histori- cal framework for interpreting changes in their mt- DNAs. Similar mtDNAs occur in two very different nuclear backgrounds; the triploid non-recombining

SEXUAL lcAsl S M6 DIPLOID (0153)

9 (0150)

PARTHENOGENETIC CABISM6

DIPLOID 7 (7)

J PARTHENOGENETIC ‘A’

TRIPLOIDS m l C A 6 1 S M B

(25154)

‘B + C’

m l S M 6 l S M 6 ( 3 2 1 3 3 )

FIGURE 1 .-Evolutionary history of parthenogenetic H . binoei and their mtDNA (boxed). Analyses of chromosome and allozyme variants suggest that the parthenogens arose through multiple hybridization events between the “CA6” and ”SM6” sexual races (MORITZ 1984; MORITZ et al. 1989). The predicted diploid-hybrid intermediate has never been found despite intensive collecting and may no longer exist. The mtDNA of the parthenogens analyzed in this report are most similar to those from western Australian populations of the CA6 type (MORITZ 199 1). The “A” and “B + C” are designations for the two major chromosome classes of triploids that are derived from different types of backcross. The frequency of duplications in each type of lizard is given in parentheses.

hybrid genome of the parthenogens, and the diploid recombining nonhybrid background of their maternal sexual parents. Comparisons of mtDNAs in sexual and parthenogenetic lizards, and in different types of parthenogenetic lineages may illuminate the processes involved. In particular, the presence in the partheno- gens of mtDNAs that differ only by length mutations provides a simple system for mapping and character- izing the mutations. This paper presents an analysis of the physical properties and the geographic and phylogenetic distributions of mtDNA duplications. This revealed a highly dynamic system of duplication and deletion, which is in stark contrast to the stability

Evolution of mtDNA Duplications 223

TABLE 2

Localities sampled and results obtained

Sample Length size variants" Locality

Aileron Stn., N T Alice Springs, N T 70km W Alice Springs, NT Bullabulling Stn., WA Bullardoo Stn., WA Coondambo Stn., SA Cunyu Stn.. WA De Rose Hill Stn., SA Faraheedy Stn., WA Glenayle Stn., WA Granite Downs Stn., SA Granite Peak Stn., WA Kathleen Valley, WA Kirkalocka Stn., WA Lake Violet Stn., WA Laverton Downs Stn., WA Leinster Downs Stn., WA Mt Willoughby Stn., SA Millbillillie Stn., WA Mt Ebenezer Stn., N T Munarra Stn., WA Neds Creek Stn., WA New Springs Stn., WA Oakden Hills Stn., SA Tea Tree, N T Wirraminna Stn., SA Yellowdine, WA Yundamindra, WA Yunndaga Stn., WA

1 9 1 4 4 3 1 2 2 1 5 1 2 1 6

13 2 5 1 1 1 4 3 2 1 3 1 2 5

The designation of length variants is described in Tables 3 and 4.

of mtDNA structure in sexual If. binoei and most other animals.

MATERIALS AND METHODS Lizards were captured from homestead rubbish tips in

the central and western deserts of Australia (Table 2). Each individual was karyotyped from short-term leucocyte cul- tures (MORITZ 1984) and the preserved specimens were lodged with the University of Michigan Museum of Zoology and the Queensland Museum.

mtDNA was purified by ultracentrifugation of CsCI-pro- pidium iodide gradients using a scaleddown method appro- priate to the smaller volumes of the Beckman TLS-55 swing bucket rotor in the TL-100 benchtop ultracentrifuge (DOWLING, MORITZ and PALMER 1990). mtDNAs were di- gested with restriction endonucleases following the sup- plier's instructions, end-labeled with ["PIdNTPs, and elec- trophoresed through 1.2% agarose and 3.5% polyacryl- amide gels (BROWN 1980; DOWLINC, MORITZ and PALMER

Southern transfer-hybridization experiments were done using acid depurination followed by alkali transfer (REED and MANN 1985) from agarose gels to charged nylon mem- branes (Gene Screen Plus, Du Pont). The filter-immobilized DNA was hybridized sequentially with nick-translated mtDNA that had no duplication and with a cloned 1.4-kb fragment of the 12s rRNA from gorilla mtDNA (HIXSON and BROWN 1986) at 2 X SSC, 65" for 18 hr with 10% dextran sulfate.

1990).

kb s L1 L2 L3 L5 L7 L7 0 2 Ls LS L11

3' i

1

FIGURE 2.-Autoradiogram ofBg/lI-digested mtDNAs from par- thenogenetic H. binoei showing the effects of different size dupli- cations on fragment patterns. The smallest duplication, Lt , results in a type 1 change; in others the duplicated sequence is present as an additional fragment (type 11 modification). The sample labeled Ls has a 6.1-kb duplication but, because of a site loss, the original 1.9- and 6.1-kb fragments have combined to produce an 8.0-kb fragment. The L7 samples had a 7.2-kb duplication, but have lost 0.4 kb from the 6.0-kb fragment, resulting in a net gain of 6.8 kb (Ds, Table 4). Similarly, the LS sample has an additional 8.8-kb fragment (comigrating with the largest fragment) which in D2 has been reduced by a 0.38-kb deletion. The designation of length mutations follows Tables 3 and 4. Arrows indicate novel or altered fragments relative to S mtDNA.

RESULTS

Size and nature of insertions: Analysis of the frag- ment patterns produced by digestion with a variety of restriction endonucleases revealed that mtDNA of Heteronotia typically varies between 17.0 and 17.4 kb and that heteroplasmy for minor length variation is common (MORITZ 1991; unpublished data). How- ever, some parthenogenetic individuals had larger genomes, with the size of insertions ranging from 1.2 to 10.4 kb (Figure 2). The total genome size therefore varies from 17 to 27 kb.

The nature of these insertions was investigated by comparing fragment patterns of normal length ( S ) and long ( L ) genomes using restriction endonucleases for which the cleavage sites in the S genome had been mapped. Three types of modification were observed (Table 3) as in previous studies (MORITZ and BROWN 1986, 1987). For some enzymes, one fragment was larger in the L genome (type I). For others, there was one additional fragment in the L genome (type 11). Finally, digestion with some enzymes produced mul- tiple additional fragments in the L genome, all but one of which comigrated with fragments also seen in the S genome (type 111). The total size increment for each enzyme was the same within experimental error; i.e., the length increase for type I changes was the same size as the single novel fragment for type I1 changes and as the sum of the additional fragments

224 C. Moritz

TABLE 3

Characteristics of large mtDNA genomes in parthenogenetic H. binoei

Duplication LI L2 Ls L. Ls h L7 LS L L,o LI I

Size (kb) 1.2 2.8 5.8 6.0 6.1 6.6 6.8" 7.6 8.8 9.4 10.4 Frequency

"A" 1 0 3 0 4 0 4 0 11 0 2 " B + C " 0 6 3 1 0 2 0 3 10 7 0

mtDNA hap D D I, A M . L J K A N N 0 lotype

Am1 (a) I 5.9 I 1 7.2 7.7 8.8 9.0 11.0 Bcll (c) 1.2 2.8 4.3, 1.5 4.3, 1.5 4.3, 1.5 3.9.2.15 4.3,2.4 4.3.3.7 4.3.4.3 4.3,3.9

0.78 0.78 0.78 0.78 0.78 0.78 0.78 BdII (9) I 2.8 5.8 5.9 6.1 6.6 7.2 7.6 8.8 9.0 10.5 EcoO 109 (e) 1.2 I 4.7, 1.0 5.1, 1.0 4.3, 1.3' 3.9, 1.95 4.3, 2.3 4.3, 2.0 4.3, 2.3 4.3, 2.0

0.78, 0.22 1 .O 1 .o 1.0, 0.82 2.0, 1.0 1.0.0.82 0.78 0.48

EcoRV (v) 5.8 6.1 7.2 5.1.3.8 5.1.4.3 5.6, 5.1 Hind111 (h) 6.1 I I I I

MluI (m) I 1 6.1 6.8 8.8 Ncol (0) 1 1 6.1 6.8 8.8 9.4 NheI (n) 1 I I 1 p v U I I (P) I I 6.1 I 7.5 8.8 8.4, 1.15 9.5 SPeI(d) 1.2 1.5, 1.25 2.6, 1.5 3.0, 1.5 3.4, 1.5 4.4.2.5 4.4, 1.5 5.6, 1.5 6.1, 1.5 5.9, 1.5

SStII (s) 2.8 4.6, 1.7 6.8 7.1, 1.7 1.3,0.3 1.3, 0.3 1.3, 0.3 1.3,0.3 1.3, 0.3 1.3, 0.3 3.1, 1.25, 0.3

XbaI (x) 1.2 I 5.8 6.1 6.1 6.6 6.8 7.6 8.8 9.0 8.6, 1.2

The number observed for each of the major chromosome types ("A" or "B + C") is given. mtDNA haplotypes are defined by the gain and loss of cleavage sites (MORITZ 1991). The modifications in fragment pattern relative to digests of S genomes are shown against each enzyme. "I" indicates a type I modification. Numbers are the size in kh of additional fragments-those in bold are novel fragments, others comigrate with fragments also present in the S digests.

These samples have superimposed a 400-bp deletion on a 7.2-kb duplication (see D2. Table 4). 'Site gain 1.0 + 0.78 + 0.22.

for type 111 modifications. For example, the total amount of additional DNA in the LS variant ranged from 5.7 to 5.9 kb (Table 3). However, for the two samples with the L I I variant, the sum of additional fragments in digests with BcZI and EcoO109 (type I11 changes) were lower than expected (Table 3). Further analysis of these genomes is necessary to characterize and map the length mutations. Some other exceptions are discussed below.

These three types of modification are consistent with the presence of direct tandem duplications and are inconsistent with inversions or inverted repeats (see DOWLING, MORITZ and PALMER 1990). Under the duplication model, type I modifications are pre- dicted for enzymes that have no site in the duplicated sequence, type I for enzymes with one site duplicated, and type I11 for enzymes with multiple sites within the duplication.

The possibility that the additional DNA is exoge- nous was tested by probing digests of S and L genomes with S mtDNA (Figure 3). The additional fragments present in type I1 and I11 digests of L genomes hy- bridized with the S probe, confirming their origin from mtDNA. It is also extremely unlikely that an

FIGURE 3.-Autoradiograms of filter-hybridhation experiments where short (S, lane 1) and long (D4, lane 2; Ls, lane 3) mtDNAs were hybridized with S mtDNA (A) and with 12s rDNA sequences from primate mtDNA (B) after digestion with four restriction endonucleases. The arrowheads indicate the position of duplication bearing fragments.

exogenous mtDNA sequence would mimic the regular fragment changes expected for a duplication.

Location and gene content: The location of the duplications was determined by comparing the frag- ment changes for each enzyme to those predicted from the location of cleavage sites in the S genome map (Figure 4A). For example, L2 (Table 3) is a 2.8- kb duplication that spans the 1.5-kb SpeI fragment (type I11 modification) and includes single sites for BgZII, BcZl and SstII (type I1 changes). This duplication

Evolution of mtDNA Duplications

A. DUPLICATIONS

225

L1 1.2kb L2 2.8kb

La 5.8kb

Le &lkb Le &6kb L, 7.2kb

La 7.6kb L, 6.0kb

Lm OAkb Lo 8.8kb

"* ...-... ......." ....... ... ... ....... ....... .... ....

GENE CONTENT NO2 ND1 res 12s CR Cyt bND6 NDS ND4 1 - 2 ..... - 4

..... ..... Da OAkb

..E.. D~ o . 3 w

D~ 0 . 2 ~ ...... .....a

excludes sites for Ec00109 and NcoI (type I changes). These data localize the L2 duplication to the region shown in Figure 4A. The survey of all mtDNAs re- vealed 1 1 different duplications that varied in size and location (Table 3 , Figure 4A). Only one type of du- plication was observed in any one individual and there was no evidence for multiple overlapping duplications.

Comparison of the ten fully characterized duplica- tions (Figure 4A) reveals three notable features. First, the duplications are clustered in a single region of the genome. Second, most of the left-hand boundaries of the duplications are located in a small region (0.3 kb) between a PvuII and a BcZl site, whereas the right hand boundaries appear to be more widely dispersed. Third, the region included in each of the smaller duplications is encompassed by the location of larger duplications. Only the two smallest duplications, L 1 and Lp, do not overlap in content.

These duplications span between 7 and 61 % of the S mtDNA. Given the tight packing of genes in animal mtDNA, it seems probable that each must include sequences for one or more genes. The presence of the highly conserved SstII sites within several of the du- plications (Table 3) suggests that these include rRNA gene sequences. This was confirmed for the L9 dupli- cation by the hybridization of a mitochondrial 12s rDNA probe to novel fragments in the L genome (Figure 3B) . On the basis of their location, it appears that all of the duplications include at least some rDNA sequences (Figure 4A). The identity of the other genes

..-.. 0, 2;5kb

FIGURE 4.-(A) Size and location of mtDNA duplications in relation to the cleavage map for S mtDNA from the parthenogens and to part of the genetic map for vertebrate mtDNA. The cross-hatched bar indicates the region definitely included within the duplication and the adjacent dots in- dicate the maximum extension in either direction. The sawtooth in the map indicates the position of minor high frequency length variation. The predicted gene content of the region covered by the duplications is shown below the cleavage map, ND = NADH dehydrogenase, CR = control region, Cytb = cytochrome b, 16s and 12s are the large and small ribosomal RNAs; respectively. Enzyme abbrevi- ations and duplication designations follow Table 3. (B) Location of dele- tions in relation to the restriction map of the region duplicated in the L.9

variant.

included in the various duplications can be inferred on the assumption that the size and order of mtDNA genes is the same in the geckos as in frogs, mammals (reviewed by BROWN 1985), whiptail lizards (D. STAN- TON, C. MORITZ and W. M. BROWN, unpublished data) and fish (JOHANSEN, GUDDAL and JOHANSEN 1990). All but the smallest of the duplications ( L I ) border on or span the control region. Downstream from the control region, most duplications appear to include both ribosomal genes and all or most of N D I . The gene content upstream is less certain because of the unknown size of the control region, but the larger duplications may extend as far as the 3' end of ND5 (Figure 4A).

Modifications of duplicate sequences: The frag- ment comparisons provided evidence for both dele- tions and changes in restriction sites within the dupli- cated sequences. The presence of a deletion within one copy of the duplication was suggested by (1) a consistent reduction in size of novel fragments or of fragments also present in the S genome, and (2) by the elimination of some cleavage sites with a concom- itant reduction in the size of the new combined frag- ment. Four deletions ranging in size from 0.38 to 2.5 kb were characterized in detail (Table 4). Three of these, Dl, 0 2 and Dq, occurred in genomes with the LS duplication. The Ds deletion appears to have oc- curred in one copy of a 7.2-kb duplication, resulting in a net gain of 6.8 kb (6 L7 in Table 3).

The location of these deletions was determined

226 C. Moritz

TABLE 4

Alterations in fragment patterns used to define deletions within duplicated sequences

Length variant

Duplication Ls Deletions

Enzyme +8.8 D, -0.19 De -0.38 D , -0.40 0 4 -2.5

AvaI 8.8 8.8 + 8.7 8.8 + 8.5 7.4 + 7.2 9.5 + 7.2 BglII 8.8 8.8 + 8.6 8.8 --f 8.4 6.1 + 5.6 9.2 + 6.8 Bcll 3.7 4.3 0.78 3.7 + 2.4 4.3 ”-* 3.9 7.5 + 5.3 EcoO 109 0.78 4.3 2.0 2.0 + 1.81 2.0 + 1.62 4.3 + 3.9

EcoRV 3.8 5.1 5.1 + 4.7 8.4 + 8.0 (3.4 + 5.1) + 6.1 NcoI 8.8 9.6 + 7.3 PvulI 8.8 8.8 + 8.2 + 14.8 SstII 7.1 1.7 (1.7 + 5.5) + 6.8 11.0 + 8.5 Spe I 5.6 1.6 1.3 5.6 + 5.1 (1.6 + 1 . 3 ) - + 2 . 5

Xba I 8.8 13.0 + 10.5

1.0 0.82

0.3

The additional fragments for the LS duplication are listed for comparison. Numbers in bold are novel fragments altered by deletion.

from patterns of overlap in the affected fragments or restriction sites (Figure 4B). The 0.4-kb deletion in- cludes the SstII site located in the 16s rRNA gene. The location and gene content of the 0.19-, 0.38- and 2.5-kb deletions is less well defined, but they appear to be involve sequences in the ND5-ND6 region. The location of the 2.5-kb D4 deletion in the right hand copy of the L g duplication was confirmed by the hy- bridization experiments (Figure 3). Had the deleted sites for EcoRV and PuuII been in the left hand copy, then the larger of the AuaI, BgZII and NcoI fragments that hybridize with the rDNA probe would have been reduced in size. These were unaffected (Figure 3B).

In contrast to the deletions, gains and losses of restriction sites were enzyme specific (Table 3). One of the four samples with the D2 variant had gained an EcoO109 site in one copy of the duplication (4.3 kb 4 4.1 + 0.16). Similarly, both L6 samples had an extra EcoO109 site in one copy (1.0 kb + 0.78 + 0.22). The L I 1 samples had modifications for SpeI and XbaI suggestive of site gains (Table 3), but this interpreta- tion is tentative until other ambiguities for these sam- ples are resolved.

A notable feature of both types of modification is that the changes were restricted to one or other copy of the duplicated sequence. In any one individual deletions were restricted to one copy of the duplica- tion, but across all the individuals assayed, deletions were observed in both the upstream and downstream copies. Restriction site changes were also only present in one or other copy. Changes in both copies would have been obvious in comparison to the S genome but were not seen.

Phylogenetic and geographic distribution: Of the 87 parthenogens surveyed, 57 had duplications. Du- plications occurred in mtDNA from both major chro-

mosome types of parthenogens (Figure l), but were significantly more common in the “B + C” group than in the “A” group (x2 = 29.7, P < 0.001). In contrast, mtDNA duplications were absent among 53 surveyed representatives of the sexual race that provided the mtDNA in the parthenogens, i.e., the CA6 race (Fig- ure 1) (see MORITZ 199 1). This difference in incidence is highly significant (x2 = 58.6, d.f. = 1, P < 0.001).

A more detailed study of the phylogenetic distri- bution of length variants was done using the results of a survey of the same mtDNAs with a suite of 4-bp- recognizing restriction endonucleases (MORITZ 199 1). This complementary study, which specifically ex- cluded fragment changes due to the length mutations, revealed 15 different haplotypes among the parthen- ogens. These haplotypes differed by between one and nine restriction sites and three, a, d and n, were common and present in both major chromosome types of triploid. The relationships among the various hap- lotypes was estimated by arranging them into a mini- mum length network (Figure 5). Placing the length variants onto this network reveals that samples with and without duplications are scattered throughout the network. Among the individuals with two of the com- mon mtDNA haplotypes, a and d , some lacked dupli- cations, whereas others had a range of duplication sizes. The other common haplotype, n, was character- ized by the Lg variant and deletions derived from it (Dl, D2 and D4). Each specific length variant is local- ized within the mtDNA phylogeny, suggesting that each has a separate origin, either by deletion from a larger duplication, or by de novo duplication from an S genome.

Heteroplasmy for the presence of duplications (ie., S and L genomes in the same individual) was rare. Only four of the 87 samples appeared to be hetero-

02

Evolution of mtDNA Duplications

S I

f-6 S i 0 S

227

0 # Lll

-0.38

S L7

FIGURE 5.-Minimum length network of mtDNA from parthenogenetic H. binoei based on the gain and loss of cleavage sites for 4-bp recognizing endonucleases independent of length changes (details in MORITZ 1991). Each node is identified by haplotype (lower case letter) and length mutation(s) (as in Tables 3 and 4). The L, variant was not analyzed in sufflcient detail to be included here. Boxed changes on the branches indicate the inferred location of some duplication or deletion events. Narrow bars indicate changes in cleavage sites.

plasmic for S and L genomes. Also, aside from the ubiquitous minor length variation, only one size of duplication was seen in any one individual.

Parthenogens from both central and western Aus- tralia had mtDNA duplications (Figure 6). However, there is a striking geographic pattern in that individ- uals from western Australia tended to have small duplications or to lack them altogether, whereas the central Australian parthenogens all had large dupli- cations. Each variant had a well defined geographic distribution. At one extreme, the LS variant was wide- spread in central Australia (Table 2, Figure 6). At the other, several variants (e.g., L1 and L4) were only found in one place. Three of the deletion variants were found in central Australia and the fourth was from the southwestern limit of the distribution (Figure 6).

DISCUSSION

The presence in Heteronotia of mtDNAs with a range of duplication sizes and of identical genomes without duplications offers an unusual opportunity to investigate the evolutionary origins and consequences of excess, presumably redundant, coding sequences in an otherwise economical genome. A novel feature of this study is that previous investigations on the evo- lution of the parthenogenetic form provide a historical framework for interpreting the changes in mtDNA.

Gene content and location of the duplications: Except for the smallest variant ( L 1 ) , all of the dupli- cations border on or include the control region. The general location and content of the duplications is therefore similar to that reported for coding sequence duplications in other animal mtDNAs (Table l), es- pecially those in Cnemidophorus lizards (MORITZ and BROWN 1987). In all, these duplications span an ap- proximately 1 1-kb region bounded on one side by the origin of light strand replication and on the other by the ND4 sequence. The only duplication so far re- ported to include the origin of light strand replication is heteroplasmic and is associated with mitochondrial dysfunction in humans (POULTON, DEADMAN and

GARDINER 1989). In contrast, the mtDNA duplica- tions in Cnemidophorus (MORITZ and BROWN 1987) and Heteronotia have no obvious phenotypic effect.

The localization of duplications to this part of the genome may reflect a bias in the mutation process. MORITZ and BROWN (1987) noted that this region remains single stranded for a lengthy period during the highly asymmetric replication process and that most of the rearrangement boundaries coincided with the predicted location of tRNA genes. They suggested that the secondary structure of the tRNAs may act as a signal for the duplication process, perhaps through illegitimate initiation of replication (see also JACOBS et al. 1989). In Heteronotia, the left boundary of many duplications occurs in the same small region, but the gene content of the boundaries is not certain. Further information on the duplication process in Heteronotia mtDNA will be gained through sequencing studies now in progress.

The origin of duplications: The mtDNA duplica- tions are present in some, but not all, parthenogens and are absent from their sexual relatives (Figure 1). This provides strong evidence that the duplications arose in the parthenogenetic lineages, rather than in their sexual ancestors. Further, the dispersion of the L and S genomes throughout the restriction-site based mtDNA network and the presence of both S and L genomes within particular haplotypes suggests that duplications have arisen repeatedly within the par- thenogens.

It is not clear whether these events occurred in the triploid lineages or in their hypothetical diploid- hybrid parthenogenetic precursors (Figure 1). The presence of some variants (e.g., L,) in both major chromosome types of triploid suggests that they arose in the common diploid ancestor. However, the differ- ence in the incidence of duplications in the “A” vs. “B + C” chromosome types and the strong geographic patterns suggest that at least some of the duplications arose in the triploid lineages.

The association between parthenogenesis and du-

228 C. Moritz

FIGURE 6.-Map of southwestern Australia showing the geographic dis- tribution of mtDNA length variants. mtDNAs without duplications are in- dicated as open boxes, duplications by filled boxes and the numbers 1 to 11 (L1 to LII) and deletions by circled numbers 1 to 4 ( D l to D4).

plication in H. binoei may be related to the fundamen- tal changes in nuclear genes that accompany the change from sexual to parthenogenetic reproduction. In particular, the parthenogens are polyploid and have non-recombining hybrid genomes. In Cnemido- phorus, mtDNA duplications were found in both sex- ual and parthenogenetic lizards, but were significantly more common in polyploids than in diploids (MORITZ and BROWN 1987). In that study, there was no signif- icant difference in the incidence of duplications in hybrid us. nonhybrid nuclear backgrounds. The ap- parent predisposition of parthenogenetic lizards to mtDNA duplication is reinforced by the presence of a duplication in one population of another triploid parthenogenetic species, Hemidactylus garnotii (C. MORITZ, unpublished data).

The mechanism of duplication of mtDNA coding sequences is not known. The association between fun- damental changes in the nuclear genome of lizards and duplications suggests that the nuclear control of mtDNA replication has been perturbed. The replica- tion and transcription of mtDNA depends on the import of nuclearly encoded enzymes, transcription factors and RNA moeties (CLAYTON 1982, 1984; CHANG and CLAYTON 1987). That the nuclear geno- type can have a direct effect on mtDNA structure is evident from the observation that in humans, mito- chondrial myopathy, which involves large deletions of mtDNA, is inherited as an autosomal dominant trait (ZEVIANI et al. 1989). It is conceivable that the mtDNA duplications in these lizards result from changes in cell cycle time or gene expression that may be associ- ated with the transition from diploidy to triploidy or from some effect of hybridity. This hypothesis can, in principle, be tested by manipulating cell lines.

Evolutionary dynamics of duplicated sequences: Parthenogenetic Heteronotia have very little restric- tion site variation other than that due to the length mutations, suggesting that they arose recently, prob- ably within the past few thousand years (MORITZ 1991). This means that the duplications, which are only found in the parthenogens, must also be recent. It follows that the modification of the duplicated sequences by deletion and restriction site change has been rapid. Similarly, HYMAN and SLATER (1990) reported a deletion within one copy of a duplicated sequence that had arisen within 170 generations of a nematode lineage. These observations contrast strongly with the usual structural stability of animal mtDNA.

The small size of animal mtDNA is often attributed to strong selection against the accumulation of DNA (WALLACE 1982; SEDEROFF 1984; ATTARDI 1985; CLARK-WALKER 1985), perhaps because smaller mol- ecules can replicate more efficiently (WALLACE 1982; RAND and HARRISON 1986; WALLIS 1987). The evi- dence for Heteronotia is consistent with a strong bias toward smaller genome size. Although genome size can be increased by 6 1 % without obvious phenotypic effects (but see POULTON, DEADMAN and GARDINER 1989), the phylogenetic analysis suggests that the du- plicated sequences are prone to rapid deletion. This was most obvious for the L9 variant where three different deletions were found in one or other copy of the 8.8-kb duplication. It is also likely that some of the smaller duplications are derived from larger ones by deletion of sequences from the internal junction; such changes are indistinguishable from de novo du- plication using restriction fragment analysis. The ob- servation that the additional sequences are rapidly

Evolution of mtDNA Duplications 229

reduced by deletion is consistent with their restricted phylogenetic distribution (MORITZ and BROWN 1987; MORITZ, DOWLING and BROWN 1987) and provides further evidence that large coding sequence duplica- tions are ephemeral.

Each of the modifications of the duplicated se- quences encountered in this survey was restricted to one copy of the duplication. The evolutionary inde- pendence of each copy of a duplicated sequence was also observed for Cnemidophorus mtDNAs (MORITZ and BROWN 1987). This differs from the situation for the phylogenetically conserved inverted repeat of land-plant chloroplast DNA where copy-correction is virtually instantaneous (reviewed by PALMER 1985). It also differs from the situation for non-coding re- peats in animal mtDNA which appear to show con- certed evolution (SOLIGNAC, MONNEROT and MOUN- OLOU 1986; BUROKER et al. 1990). The apparent homogenization of these smaller noncoding repeats may be due to their high frequency of deletion and reduplication. This process would not operate in the large coding sequence duplications because they are generated less often and because only one additional copy is present.

The tight packing of genes in most animals mtDNAs may usually preclude structural rearrangements, ef- fectively freezing the genome structure (WALLACE 1982; SEDEROFF 1984; ATTARDI 1985; CLARK- WALKER 1985; BROWN 1985). The presence of redun- dant sequences in Heteronotia mtDNA should relax selection against rearrangements and, with the appro- priate deletions, might even change gene order (MOR- ITZ, DOWLINC and BROWN 1987; CLARK-WALKER 1989; DFSJARDINS and MORAIS 1990). In the absence of redundant sequences, deletions of coding sequences are only tolerated in the heteroplasmic state and even then can be deleterious (reviewed by WALLACE 1989). Several Heteronotia with duplicate sequences were homoplasmic for deletions in one or other copy (but never both), confirming that there is redundancy of coding sequences. However, the survey of 87 mtDNAs did not reveal any inversions or transpositions. This suggests that even in the absence of constraints im- posed by tightly packed genes, large inversions and transpositions occur less often than deletions and the gain or loss of cleavage sites.

Thanks to S. LAVERY, A. HEIDEMAN and E. ZEVERING for assist- ance in the laboratory, to s. BALDAUF, D. CLARK-WALKER, R. SLADE and C. E. ZEVERINC for comments on the manuscript, and to the Australian Research Council, the National Science Foundation, the National Geographic Society (United States), 2nd the University of Queensland for support.

L ITERATURE CITED

ATTARDI, G., 1985 Animal mitochondrial DNA: an extreme of genetic economy. Int. Rev. Cytol. 93: 93-145.

BENTZEN, P., W. C. LEGCETT and G. G. BROWN, 1988 Length

and restriction site heteroplasmy in the mitochondrial DNA of American shad (Alosa sapidissima). Genetics 118: 509-5 18.

BIRMINGHAM, E., T . LAMB and J. C. AVISE, 1986 Size polymor- phism and heteroplasmy in the mitochondrial DNA of lower vertebrates. Heredity 77: 249-252.

BOYCE, T . M., M. E. ZWICK and C. F. AQUADRO, 1989 Mitochondrial DNA in the bark weevils: size, structure and heteroplasmy. Genetics 123: 825-836.

BROWN, W. M., 1980 Polymorphism in mitochondrial DNA of humans as revealed by restriction endonuclease analysis. Proc. Natl. Acad. Sci. USA 77: 3605-3609.

BROWN, W. M., 1985 The mitochondrial genome of animals, pp. 95-130 in Molecular Evotutionary Genetics, edited by R. J. MACINTYRE. Plenum Press, New York.

BUROKER, N. E., J. R. BROWN, T. A. GILBERT, P. J. O’HARA, A. T. BECKENBACH, W. K. THOMAS and M. J. SMITH, 1990 Length heteroplasmy of sturgeon mitochondrial DNA: an illegitimate elongation model. Genetics 124: 157-163.

CANTATORE, P., M. N. GADALETA, M. ROBERTI, C. SACCONE and A. C. WILSON, 1987 Duplication and remoulding of tRNA genes during the evolutionary rearrangement of mitochondrial genomes. Nature 3 2 9 853-855.

CHANG, D. D., and D. A. CLAYTON, 1987 A mitochondrial RNA processing activity contains nuclearly encoded RNA. Science

CLARK-WALKER, G. D., 1985 Basis of diversity in mitochondrial DNAs, pp. 277-297 in The Evolution of Genome Size, edited by T. CAVALIER SMITH. John Wiley & Sons, New York.

CLARK-WALKER, G. D., 1989 In vivo rearrangement of mitochon- drial DNA in Saccharomyces cerevisiae. Proc. Natl. Acad. Sci. USA 8 6 8847-8851.

CLAYTON, D. A,, 1982 Replication of animal mitochondrial DNA. Cell 28: 693-705.

CLAYTON, D. A,, 1984 Transcription of the mammalian mito- chondrial genome. Annu. Rev. Biochem. 53: 573-594.

DENSMORE, L. D., J. WRIGHT and W. M. BROWN, 1985 Length variation and heteroplasmy are frequent in mitochondrial DNA from parthenogenetic and bisexual lizards (genus Cnemidopho- rus). Genetics 110 689-707.

DENSMORE, L. D., C. MORITZ, J. WRIGHT and W. M. BROWN, 1989 Mitochondrial DNA analyses and the origin and relative age of parthenogenetic lizards (genus Cnemidophorus). IV. Nine sexlineatus group unisexuals. Evolution 43: 969-983.

DESJARDINS, P., and R. MORAIS, 1990 Sequence and gene orga- nization of the chicken mitochondrial genome. J. Mol. Biol.

DOWLING, T. E., C. MORITZ and J. PALMER, 1990 Nucleic acids 11: restriction site analysis, pp. 251-317, in Molecular System- atics, edited by D. M. HILLIS and C. MORITZ. Sinauer, Sunder- land, Mass.

DUBIN, D. T., C. C. HSU-CHEN and L. E. TILLOTSEN, 1986 Mosquito mitochondrial transfer tRNAs for valine, gly- cine and glutamate: RNA and gene sequences and vicinal genome organization. Curr. Genet. 1 0 701-707.

CAREY, J. R., and D. R. WOLSTENHOLME, 1989 Platyhelminth mitochondrial DNA: evidence for early evolutionary origin of a tRNA”‘ AGN that contains dihydrouridine arm replacement loop, and of serine-specifying AGA and AGG codons. J. Mol. Evol. 28: 374-387.

HALE, L. R., and R. S. SINGH, 1986 Extensive variation and heteroplasmy in size of mitochondrial DNA among geographic populations of Drosophila melanoguster. Proc. Natl. Acad. Sci. USA 83: 8813-8817.

HAUCKE, H.-R., and G. GELLISSEN, 1988 Diffferent mitochondrial gene orders among insects: exchanged tRNA gene positions in the COlI/COIIl region between an orthopteran and a dipteran species. Curr. Genet. 1 4 471-476.

HIXSON, J. E., and W. M. BROWN, 1986 A comparison of the

235: 1178-1 184.

212: 599-634.

230 C. Moritz

small ribosomal RNA genes from the mitochondrial DNA of great apes and humans: sequence, structure, and phylogenetic implications. Mol. Biol. Evol. 3: 1-18.

HYMAN, B. C., and T. M. SLATER, 1990 Recent appearance and molecular characterization of mitochondrial DNA deletions within a defined nematode pedigree. Genetics 124: 845-853.

HYMAN, B. C., J. L. BECK and K. C. WEISS, 1988 Sequence amplification and gene rearrangement in parasitic nematode mitochondrial DNA. Genetics 1 2 0 707-712.

.JACOBS, H. T., S. ASAKAWA, T. ARAKI, K. MIURA, M. J. SMITH and K. WATANABE, 1989 Conserved tRNA cluster in starfish mi- tochondrial DNA. Curr. Genet. 15: 193-206.

JOHANSEN, S., P. H. GUDDAL and T . JOHANSEN,

1990 Organization of the mitochondrial genome of Atlantic cod, Gadus morhua. Nucleic Acids Res. 18: 41 1-419.

L A ROCHE, J., M. SNYDER, D. I. COOK, K. FULLER and E. ZOUROS, 1990 Molecular characterization of a repeat element causing large-scale size variation in the mitochondrial DNA of the sea scallop Placopecten mugellanicus. Mol. Biol. Evol. 7: 45-64.

MORITZ, C., 1983 Parthenogenesis in the endemic Australian lizard Heteronotia binoei (Gekkonidae). Science 220: 735-737.

MORITZ, C., 1984 The origin and evolution of parthenogenesis in Heteronotia binoei (Gekkonidae). Chromosoma 8 9 15 1-162.

MORITZ, C., 1991 The origin and evolution of parthenogenesis in Heteronotia binoei (Gekkonidae): evidence for recent and restricted origins of widespread clones. Genetics 1 2 9 2 1 1-2 19.

MORITZ, C., and W. M. BROWN, 1986 Tandem duplication of D- loop and ribosomal RNA sequences in lizard mitochondrial DNA. Science 233: 1425-1427.

MORITZ, C., and W. M. BROWN, 1987 Tandem duplications in animal mitochondrial DNAs: variation in incidence and gene content among lizards. Proc. Natl. Acad. Sci. USA 84: 7183- 7187.

MORITZ, C., T. E. DOWLING and W. M. BROWN, 1987 Evolution of animal mitochondrial DNA: relevance for population biol- ogy and systematics. Annu. Rev. Ecol. Syst. 18: 269-292.

MORITZ, C., J. W. WRIGHT and W. M. BROWN, 1989 Mitochondrial DNA and the origin and relative age of parthenogenetic lizards (genus Cnemidophorus). 111. C. uelox and C. exsunpis. Evolution 43: 958-968.

MORITZ, C., S. DONNELLAN, M. ADAMS and P. R. BAVERSTOCK, 1989 The origin and evolution of parthenogenesis in Heter- onotia binoei (Gekkonidae): extensive genotypic diversity among parthenogens. Evolution 43: 994-1 003.

PALMER, J. D., 1985 Evolution of chloroplast and mitochondrial

DNA in plants and algae, pp. 131-240 in Molecular Evolution- ary Genetics, edited by R. MACINTYRE. Plenum Press, New York.

POULTON, J., M. E. DEADMANand R. M. GARDINER, 1989 Tandem direct duplication of mitochondrial DNA in mitochondrial myopathy: analysis of nucleotide sequence and tissue distribu- tion. Nucleic Acids Res. 17: 10223-10229.

RAND, D. M., and R. G. HARRISON, 1986 Mitochondrial DNA transmission genetics in crickets. Genetics 114 955-970.

RAND, D. M., and R. G. HARRISON, 1989 Molecular population genetics of mtDNA size variation in crickets. Genetics 121:

REED, K. C., and D. A. MANN, 1985 Rapid transfer of DNA from agarose to nylon membranes. Nucleic Acids Res. 13: 7207- 7221.

SEDEROFF, R. R., 1984 Structural variation in mitochondrial DNA. Adv. Genet. 22: 1-108.

SNYDER, M., A. R. FRASER, J. LA ROCHE, K. E. GARTNER-KEPKAY and E. ZOUR~S, 1987 Atypical mitochondrial DNA from deep-sea scallopPlacopecten mugellanicus. Proc. Natl. Acad. Sci.

SOLIGNAC, M., M. MONNEROT and J.-C. MOUNOLOU, 1986 Concerted evolution of sequence repeats in Drosophila mitochondrial DNA. J. Mol. Evol. 24: 53-60.

SOLIGNAC, M., J. GENERMONT, M. MONNEROT and J.-C. MOUNOLOU, 1987 Drosophila mitochondrial genetics: evolution of heter- oplasmy through germ line cell divisions. Genetics 117: 687- 696.

WALLACE, D. C., 1982 Structure and evolution of organelle ge- nomes. Microbiol. Rev. 4 6 208-240.

WALLACE, D. C., 1989 Mitochondrial DNA mutations and neu- romuscular disease. Trends Genet. 5: 9-13.

WALLIS, G. P., 1987 Mitochondrial DNA insertion polymorphism and germ line heteroplasmy in the Triturus cristatus complex. Heredity 5 8 229-238.

WOLSTENHOLME, D. R., D. 0. CLARY, J. L. MACFARLANE, J. A. WAHLEITHNER and L. WILCOX, 1985 Organization and evo- lution of invertebrate mitochondrial genomes, pp 61-69 in Achievements and Perspective in Mitochondriat Research, Vol. 11: Biogenesis, edited by E. QUAGLIARELLIO, E. C. SLATER, F. PALMIERI, C. SACCONE, et al. Elsevier, New York.

ZEVIANI, M., S. SERVIDEI, C. GELLERA, E. BERTINI, S. DIMAURO and S. DIDONATO, 1989 An autosomal dominant disorder with multiple deletions of mitochondrial DNA starting at the D-loop region. Nature 339: 309-31 1 .

551-569.

USA 84: 7595-7599.

Communicating editor: M. J. SIMMONS