Heteroplasmy, Length and Sequence Variation in the mtDNA Control Regions of Three Percid Fish Species (Perca fluviatilis, Acerina cernua, Stizostedion lucioperca)

Corresponding author: Kjetill S. Jakobsen, Division of General Genetics, Department of Biology, University of Oslo, P.O. Box 1031 Blindern, N-0315 Oslo, Norway, k.s.jakobsen{at}bio.uio.no (E-mail).

Communicating editor: A. G. CLARK

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The nucleotide sequence of the control region and flanking tRNA genes of perch (Perca fluviatilis) mtDNA was determined. The organization of this region is similar to that of other vertebrates. A tandem array of 10-bp repeats, associated with length variation and heteroplasmy was observed in the 5' end. While the location of the array corresponds to that reported in other species, the length of the repeated unit is shorter than previously observed for tandem repeats in this region. The repeated sequence was highly similar to the Mt5 element which has been shown to specifically bind a putative D-loop DNA termination protein. Of 149 perch analyzed, 74% showed length variation heteroplasmy. Single-cell PCR on oocytes suggested that the high level of heteroplasmy is passively maintained by maternal transmission. The array was also observed in the two other percid species, ruffe (Acerina cernua) and zander (Stizostedion lucioperca). The array and the associated length variation heteroplasmy are therefore likely to be general features of percid mtDNAs. Among the perch repeats, the mutation pattern is consistent with unidirectional slippage, and statistical analyses supported the notion that the various haplotypes are associated with different levels of heteroplasmy. The variation in array length among and within species is ascribed to differences in predicted stability of secondary structures made between repeat units.

MITOCHONDRIAL DNA heteroplasmy is at least a transient stage in the propagation of any new mutation in the organelle DNA. While heteroplasmy for point mutations is rare (MORITZ et al. 1987 ), heteroplasmy for length variants frequently occurs in natural populations (e.g., BENTZEN et al. 1988 ; WILKINSON and CHAPMAN 1991 ; ARNASON and RAND 1992 ; BROWN et al. 1992 ; BROUGHTON and DOWLING 1994 ; STEWART and BAKER 1994 ; FUMAGALLI et al. 1996 ; MUNDY et al. 1996 ). These intraspecific and intraindividual length polymorphisms are most often observed in tandem repeated structures in the control region (reviewed in MORITZ et al. 1987 ; Rand 1993 ). The tandemly repeated arrays are typically located in the parts of the control region that coincide with the 5' end of the D-loop DNA where replication is initiated, and the 3' end where the D-loop DNA is terminated (LEE et al. 1995 ). Moreover, the repeated sequences are usually associated with secondary structures (e.g., WILKINSON and CHAPMAN 1991 ; ARNASON and RAND 1992 ; LEE et al. 1995 ). The precise mechanisms causing mtDNA length variation heteroplasmy are not known; however, several models have been suggested (e.g., BUROKER et al. 1990 ; BROWN et al. 1996 ).

MtDNA length variations, caused by tandem repeats, have previously been identified in a number of fish species: several species of sturgeon (e.g., Acipenser transmontanus, Acipenser oxyrhynchus desotoi) (BROWN et al. 1992 ; MIRACLE and CAMPTON 1995 ); cod (Gadus morhua) ( JOHANSEN et al. 1990 ; ARNASON and RAND 1992 ); European sea bass (Dicentrarchus labrax) (CECCONI et al. 1995 ); and American shad (Alosa sapidissima) (BENTZEN et al. 1988 ). With few exceptions, the repeats in these species tend to be long (e.g., 81 bp in sturgeon, 1.5 kbp in American shad), and located in the 5' end of the control region flanking the trnP gene.

The order Perciformes is among the most species-rich vertebrate groups, and thus provides an opportunity to study the evolutionary stability of mitochondrial arrays. In a previous study, REFSETH et al. 1998 have investigated the population structure and biogeographic history of perch (Perca fluviatilis) by sequencing the control region of perch mtDNA from several Norwegian and Swedish populations. The results revealed relatively low levels of mtDNA sequence variation and high levels of geographical structuring among Scandinavian perch populations. In this paper, using direct sequencing and TA-cloning of PCR products, we demonstrate the presence of length variation heteroplasmy in perch mtDNA due to a 10-bp tandemly repeated sequence. We have addressed the origin and maintenance of heteroplasmy among perch populations by analyzing a large sample (n = 149) for the occurrence of heteroplasmy, the phylogenetic relationship among different units and arrays, and the distribution of array lengths. Furthermore, to investigate the mechanisms maintaining heteroplasmy, we have analyzed mature oocytes from heteroplastic females. The evolutionary stability of the array was assessed by sequencing the 5' part of the control regions of two other species belonging to the Percidae family, zander (or pike perch; Stizostedion lucioperca) and ruffe (Acerina cernua), as well as other fish species both within and outside the order Perciformes.

Our data demonstrate that most perch show heteroplasmy for different array lengths. We also observed similar arrays in both zander and ruffe. Moreover, results from cloning of PCR products suggest that units are added and deleted by unidirectional slippage, and that maternal transmission plays a major role in the maintenance of heteroplasmy.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Sample collection and preparation:
A total of 143 perch, four ruffe, and two zander were used. The perch used in this study were collected from 18 east Norwegian and four Swedish populations. 132 of the perch samples were also used in REFSETH et al. 1998 . In addition, three oocytes were analyzed from each of two perch.

Tail fin or liver was removed in field and stored in 96% ethanol or at -80°. Mature oocytes were removed from two perch specimens and stored in 96% ethanol. Total genomic DNA was extracted using a standard proteinase K phenol-chloroform method with ethanol precipitation (SAMBROOK et al. 1989 ), or by the DNA-direct kit (Dynal AS, Oslo, Norway) (RUDI et al. 1997 ).

PCR on total DNA:
Amplifications were carried out in a 50-µl final volume containing 1–5 ng template DNA, 1x PCR buffer, 2.5 mM MgCl₂, 1 µmol of each primer, and 0.5 units Taq-polymerase (Advanced Biotechnology, Surrey, UK). The following primers were used to amplify segments of the mtDNA control region: HV2: TTCCCCGGTCTTGTAAACC (modified from HOELZEL et al. 1991 ); CSB-2: AAACCCCCCTACCCCCC (SHEDLOCK et al. 1992 ); 12SR: CGGTGACTTGCATGTGTAATGTCA (REFSETH et al. 1998 ); CSB-3: TATTCCTGTTTCCGGGG; and CSB-D: GGAACCAAATGCCAGGAA (designed from Champsochromis spilorhychus, LEE et al. 1995 ). The locations of the primers in the control region are shown in Figure 1. The HV2 primer is located in the trnT gene, the CSB-2, -3, and -D span the conserved sequence block (CSB) 2, 3, and D, respectively, and the 12SR primer is located in the 12SRNA gene. For direct sequencing of PCR products, one of the primers was biotinylated. The reactions were submitted to an initial denaturation at 96° for 5 min, and then 30 cycles each consisting of denaturation at 96° for 1 min, annealing at 51° (HV2-CSB3), 54° (HV2-CSBD), 55° (CSB2-12SR), or 46° (HV2-12SR) for 1 min, and extension for 2 min at 72°. The reaction was performed on a Biometra Trio-Thermo block TB1 (Biomedizinische Analytik GmbH, Göttingen, Germany). Biotinylated PCR products were sequenced by a direct solid phase approach using Dynabeads M-280 Streptavidin (Dynal AS) (HULTMAN et al. 1989 ), applying the nonbiotinylated PCR-primer as sequencing primer.

View larger version (8K): In this window In a new window Download PPT slide   Figure 1. —Overview of the perch mtDNA control region and flanking tRNA genes. Conserved sequence elements and tRNAs are indicated by boxes. Primer positions are indicated by arrows. The HV2-CSB3 primer combination gave rise to an approximately 800-bp long fragment, and the HV2-CSBD combination gave rise to an approximately 550-bp fragment. The PCR product obtained using the CSB2 and 12SR primer was 300 bp. For ruffe and zander only the HV2-CSBD combination was used.

Cloning of PCR products and detection of heteroplasmy:
Nonbiotinylated PCR products from 11 adult perch, three mature perch oocytes, and two ruffe were cloned using the pGEM-T Vector System (Promega, Madison, WI). From each ligation between eight and 10 individual clones were PCR-amplified. Sequencing of the PCR products was performed either on an ABI 373 DNA Sequencer using ABI PRISM Dye Primer Cycle Sequencing Kit (Applied Biosystems, Foster City, CA), or on a Vistra DNA Sequencer 725 using Thermo Sequenase sequencing kit (Molecular Dynamics and Amersham Life Science, Buckinghamshire, UK).

Data analysis:
Sequence analysis was performed using the GCG package of computer programs (Version 7.0; Genetics Computer Group Inc., Madison, WI). Database searches were carried out using the FASTA program. Localization of tRNA genes and conserved sequences were done using BESTFIT with sequences from a cichlid (Champsochromis spilorhychus ; acc. no. U12553), European sea bass (Dicentrarchus labrax ; acc. no. X81472), tuna (Thunnus thynnus ; acc. no. X82653), cod (Gadus morhua ; acc. no. X99772), minnow (Cyprinella spiloptera ; acc. no. L07753), rainbow trout (Oncorhynchus mykiss ; acc. no. S68946), Arctic char (Salvelinus alpinus ; acc. no. X68659), white sturgeon (Acipenser transmontanus ; acc. no. X54348), a lungfish (Protopterus dolli; acc. no. L42813), Xenopus laevis (acc. no. M13046), chicken (Gallus domesticus ; acc. no. X52392), and humans (Homo sapiens; acc. no. V00662).

Potential secondary structures in the DNA were analyzed using the FOLDRNA program in GCG and visualized using loopDloop (GILBERT 1992A ). The multiple alignment of nucleotide sequences was carried out with the PILEUP program in GCG and SeqApp (GILBERT 1992B ). Statistical analyses were performed using the insight mode of the SAS program package (SAS 6.12 for windows; SAS Institute Inc., Cary, NC).

Phylogenetic analyses were performed in PAUP Version 3.1.1 (SWOFFORD 1991 ). The maximum parsimonious relationship among the three percid species was estimated using the "branch and bound"-search algorithm in PAUP. To estimate the phylogenetic relationship among perch control region haplotypes we used the algorithm described in TEMPLETON et al. 1992 . This cladogram estimation procedure is specifically designed to estimate intraspecific gene trees where most of the nodes (haplotypes) are present in the population under investigation and are few mutations apart. A set of cladograms that have a high probability (>95%) of being true is obtained. The resulting cladograms can be converted into a nested design as described in TEMPLETON et al. 1987 and TEMPLETON et al. (1993). The units defining the various nested branches of the cladogram are called n-step clades, where n indicates the number of mutational steps necessary to define the clade. 0-step clades refer to haplotypes. Given the n-step clades, the n + 1 clades are defined by the union of all n-step clades that can be joined together by moving back one mutational step from the terminal n -step clades. Hence, haplotypes are grouped into 1-step clades (i.e., 1-1, 1-2, 1-3, and 1-4), and these clades are further grouped into 2-step clades (i.e., 2-1 and 2-2 in Figure 3).

View larger version (41K): In this window In a new window Download PPT slide   Figure 2. —The perch mtDNA control region, including the flanking tRNAs, aligned with the sequenced fragments from ruffe and zander. Identical positions are indicated by dots (except for the tandem repeated array), and indels by dashes. The sequence corresponds to the L-strand of the human mtDNA (ANDERSON et al. 1981 ). The perch sequence is 1080 bp. Conserved sequence motifs are underlined. The borders of the tRNAs are indicated by arrows. In ruffe, the repeated array between trnP and TAS1 varied between six and 23 units, and in zander the array consisted of six units. Only four units (including the 3' degenerated unit) are shown to obtain the optimal alignment with the perch sequence. The sequences are submitted to GenBank's databases under acc. no. Y 14724-6.

View larger version (28K): In this window In a new window Download PPT slide   Figure 3. —The cladogram is modified from REFSETH et al. 1998 and represents a 99% plausible set of haplotypes networks, estimated using the maximum parsimony algorithm given in TEMPLETON et al. 1992 . The cladogram was converted into a nested design as described in TEMPLETON et al. 1987 (Templeton et al. (1993); haplotypes one mutation apart are grouped into 1-step clades (1-1, 1-2, 1-3 and 1-4), and these clades are further grouped into 2-step clades (2-1 and 2-2). The array types found in the haplotypes are indicated; e.g., first A - last GC means that the first repeat was type A and the last degenerated repeat was type GC (Table 2). The A haplotype is given, and the number in brackets corresponds to the position in the sequence (Figure 2). Only variable sites are included.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The perch mtDNA control region; conserved sequences and homology to other vertebrate mtDNA:
The complete sequence of the perch control region and its flanking tRNA genes has been determined. Figure 1 depicts overall structure of the perch control region including the conserved sequence blocks (CSBs), tRNA genes, and the repeated array. The nucleotide sequence of the perch control region is aligned with the 5' parts of the zander and ruffe control regions in Figure 2. The perch mtDNA control region revealed a length of 882 bp. The general organization of the perch control region was similar to that reported for other vertebrates (e.g., CHANG and CLAYTON 1984 ; FORAN et al. 1988 ; BUROKER et al. 1990 ; JOHANSEN and JOHANSEN 1993 ; BROUGHTON and DOWLING 1994 ).

Conserved regulatory elements: Three TAS (Termination Associated Sequence) motifs—TACAT—were found in the 5' part of the control region, indicated in Figure 2 as TAS-1, -2, and -3. A motif similar to the conserved motif in the 5' region of lungfish and other vertebrates (ZARDOYA and MEYER 1996 ) was also identified (CM5' in Figure 2 and Table 1). This motif forms a potential stemloop (G = -4.4 kcl/mol) and the most conserved parts are the nucleotides predicted to form the stem, suggesting that this is a conserved secondary structure that may contribute to TAS function.

A motif with high similarity (52% total identity and 100% identity at the last five bases) to human CSB1 is present at position 772–794 (Figure 2 and Table 1). Somewhat surprisingly, the CSB1 element has been reported present in only a few fish species, e.g., minnow (BROUGHTON and DOWLING 1994 ) and sturgeon (BUROKER et al. 1990 ). However, alignment of the perch D-loop with other fish demonstrates that this element is also found in other species (Table 1), and thus seems to be universally distributed among fish species.

In vertebrates, the H- and L-strand promoters are located near the trnF gene. In perch, only one segment with homology to previously reported promoter sequences (reviewed in TRACY and STERN 1995 ) was found (Figure 2; Table 1), indicating a single putative bidirectional promoter. The only mitochondrial bidirectional promoter reported so far is in chicken (L'ABBÉ et al. 1991; TRACY and STERN 1995 ), and the perch promoter is similar to this sequence (Table 1). Moreover, the perch sequence is able to form a cruciform structure similar to that in chicken. The distance between CSB3 and the trnF gene is only 68 bases, which is considerably shorter than what is previously reported for other fish (e.g., LEE et al. 1995 ).

Finally, a tandemly repeated array located between the TAS 1 and trnP, composed of two to five repeated units in perch, six units in zander, and six to 23 units in ruffe, was observed. The repeated sequence is highly similar to a putative control element, Mt5 (Figure 2; Table 1), observed in several vertebrates including humans (OHNO et al. 1991 ). This element has been shown to specifically bind a protein possibly involved in regulation of D-loop DNA termination (KUMAR et al. 1995 ).

The anatomy of the tandem repeats of percid fish:
A short 10-bp tandem repeated sequence in the 5' end of the percid control region: All individuals examined among the three percid species studied possessed the tandemly repeated array consisting of 10-bp repeats, between the trnP gene and TAS 1 (Figure 2). Throughout this paper the repeat unit nearest the trnP gene is referred to as the first repeat. The remaining perfect repeats are referred to as second, third, fourth, or array repeats. Furthermore, in all perch sequenced, the array was flanked by a 3' degenerated repeat unit (hereafter referred to as the last repeat) that differed by several mutations from the "standard" perch array unit (Figure 2).

The first repeat unit extends four bases into the trnP gene. In general, the repeat units are highly similar both within and among the three species (identical in ruffe and zander; see Figure 2 and Table 2). The repeats contain perfect palindromic sequences (TTGCAA) and imperfect palindromes (AA(G)TATT).

In perch, two types of first and second repeat were observed (Table 2). Among the 149 individuals sequenced, this variable position defining the A- and T-type arrays (i.e., A or T at position 6 in the first repeat, Table 2), and an identical substitution in the second repeat unit, were the only sequence polymorphisms observed in the array (see also REFSETH et al. 1998 ). Similar levels of intraspecific conservation have also been observed in several other species (e.g., BUROKER et al. 1990 ; BROWN et al. 1996 ). The sequence of the last degenerated repeat also varies between individuals, and, using the same haplotype definitions as in REFSETH et al. 1998 , can be separated into two major classes of units: GC (last unit 1) and TT (last unit 2J; 2D, G, L, M; 3H; and 3I) (see Table 2). The two substitutions defining the TT and GC groups (G₁₃₀ C₁₃₁ T₁₃₀ T₁₃₁) apparently represent convergent mutations. The T₁₃₀ T₁₃₁ are present in the haplotypes D, G, L, and M (Figure 3). These haplotypes are phylogenetically separated by haplotypes possessing GC arrays (Figure 3), indicating recurrent mutations at these sites. Supporting this, both GC and TT arrays were found in about equal amounts within the same individual (Eikeren #3) (see below; Table 3). The last repeat unit in the ruffe and zander array did not differ as much from the consensus unit as the perch last unit (Table 2).

Arrays containing more than one repeat unit are predicted to form stable secondary structures: Since secondary structures have been implied in formation and maintenance of tandemly repeated structures (see BUROKER et al. 1990 ; WILKINSON and CHAPMAN 1991 ; ARNASON and RAND 1992 ; BROWN et al. 1992 ; BROUGHTON and DOWLING 1994 ; FUMAGALLI et al. 1996 ), the stability of potential secondary structures was predicted by the FOLDRNA algorithm in GCG program package. As can be inferred from the presence of palindromic sequences in the repeats, two or more single-stranded units were able to form stable secondary structures in all three species studied (Figure 4). The stability of the predicted structures varied, however. The perch internal units are predicted to produce more stable structures compared to the zander/ruffe units (Figure 4A). Moreover, in perch, the T-type first repeat forms stronger secondary structures than the A-type repeat (Figure 4B). The degenerated last units are also able to form stable secondary structures when folded with an internal unit, but the estimated G values suggest lower stability (Figure 4B).

View larger version (25K): In this window In a new window Download PPT slide   Figure 4. —Predicted secondary structures and their G values (kcl/mol). (A) Two perch array-units and two zander/ruffe units. (B) Different types of perch first and last units folded with one array repeat. For all structures, except the last imperfect units, the sequences correspond to the L-strand, and the G values for the H-strand are given in brackets. Note that the FOLD algorithm calculates G values for RNA sequences.

Phylogenetic topologies based on the repeat unit and the control region are congruent:
In order to investigate the evolutionary stability of the percid tandem repeated sequence, we estimated the phylogenetic relationship both among the flanking control region sequences and the different repeat unit sequences observed.

Topologies based on control region sequences: In this analysis we also sequenced the control region of American yellow perch (Perca flavescens; acc. no. Y 14728), as well as spotted wolffish (Anarhichas minor ; family Anarhicadidae, acc. no. Y 14775), and flounder (Plantichthys flesus; order Pleuronectiformes, acc. no. Y 14730), to obtain outgroup information. The tree obtained by applying the "branch and bound"-search algorithm in PAUP, with results from 1000 branch and bound bootstrap replicates is presented in Figure 5A. High bootstrap values were obtained for both perch-yellow perch and ruffe-zander monophyletic groups. However, when omitting the outgroup species, only the perch-yellow perch cluster was supported (bootstrap value at 100%, data not shown). The same topology was also suggested when using only the flounder sequence as outgroup (bootstrap value at 85%, data not shown). In these analyses the tandem repeat was included. Leaving out the repeats, however, did not change the topology of the trees.

View larger version (17K): In this window In a new window Download PPT slide   Figure 5. —Phylogenetic relationships inferred from the control region fragment sequenced in perch, zander, and ruffe (A), and among the observed repeat units (B). The bootstrap option of PAUP was used to demonstrate confidence in positions of tree nodes, and 1000 branch and bound bootstrap replicates were performed. The values obtained are shown in bold italic. (A) The most parsimonious tree based on the control region sequences. The yellow perch sequence (acc. no. Y 14728) was also included. The consistency index (ci) was 0.933, and the retention index (ri) was 0.659. The flounder (acc. no. Y 14730) and the spotted wolffish (acc. no. Y 14775) sequences were used as outgroups. (B) The majority-rule consensus tree estimated using the repeat sequences. Ten equally parsimonious trees were obtained, and the percentages of the trees containing the branch are indicated in addition to the bootstrap values (roman and bold italic values, respectively). The ci for each most parsimonious tree was 0.8, and the ri was 0.909. The "midpoint rooting" option was used.

Phylogenetic relationships among different repeats: We also estimated the phylogenetic relationships among the different repeat units observed, combining the perch units and the ruffe/zander units (see Table 2). A total of 10 equally parsimonious trees were found, and the 50% rule consensus tree is shown in Figure 5B with the results from 1000 bootstrap replicates. The low bootstrap value, due to short sequences and low divergence, indicates a high degree of uncertainty associated with the suggested topology. However, the "majority-rule" values are high (between 67 and 100%). All perch last units are clustered together, and were connected to the remainder of the tree through the perch array unit. Moreover, the perch type A first unit is the one closest to the ruffe and zander sequences. This is in agreement with our previous findings based on intraspecific phylogenetic relationship and frequencies of perch haplotypes (see Figure 3 and REFSETH et al. 1998 . Hence, the PAUP analysis suggests that the ancestral unit resembles the ruffe/zander unit or the type A first perch repeat.

Evolutionary stability of the tandem repeats: The sequence obtained from American yellow perch shows that this species has an array similar to the perch A-GC type. The standard array length also here seems to be three units (not shown). However, sequences highly similar to the percid repeats were not found among previously reported mtDNA control regions, including other members of the order Perciformes (e.g., sea bass, cichlids; as determined by FASTA searches in the GenBank data bases), possibly suggesting that this repeat is confined to the family Percidae. However, by sequencing a number of fish both within and closely related to the order Perciformes, a distantly related array was observed in flounder (Figure 6). In this species the repeated unit is 19 bases, seemingly composed of two percid units. Remnants of this repeat were observed in two other flatfishes, American plaice (Hippoglossides platessoides; acc. no. Y 14727) and brill (Scophtalamus rhombus ; acc. no. Y 14729), in which we have sequenced this region (data not shown). Furthermore, an array, showing similarity to the flounder repeats, but composed of 74 -bp units, is present in left eye flounder (Paralichys olivaceus; acc. no. AB000668).

View larger version (23K): In this window In a new window Download PPT slide   Figure 6. —Alignment of the percid and the flounder repeats. One flounder unit is aligned against two percid units.

Characterization of length variation and heteroplasmy associated with the tandem repeat:
Different array types show different levels of heteroplasmy: Cloning of PCR products and sequencing of the individual clones revealed variation among molecules in copy number of the tandem repeats causing length variation and heteroplasmy in the mtDNA of perch.

Occurrence of heteroplasmy could also be detected by direct sequencing of PCR products. Many of the perch sequences displayed a double sequence on the autoradiogram; i.e., at least two sequence ladders were superimposed on each other after the repeat array (Figure 7). Sequencing of individual clones obtained from double sequence PCR products always revealed at least two length variants. Furthermore, of the 20 clones obtained from two PCR products not showing double sequences, only one possessed a four-repeat array (the rest were three-repeat arrays, Suluvatn #9 and Store Lauarvann #6 in Table 3).

View larger version (38K): In this window In a new window Download PPT slide   Figure 7. —Autoradiogram showing double sequences; i.e., at least two sequence ladders superimposed on each other after the repeat array. The end of the array is indicated. The band of the "second" sequence is indicated by a H and reads: ... TTGCTAGCACTTGCTA ..., i.e., at least one extra repeat unit.

Analyzing the 149 directly sequenced perch PCR products revealed that 74% showed length variation heteroplasmy. Among the A-GC arrays (type A first unit and GC last degenerated unit) 80% showed heteroplasmy. 75% of the A-TT arrays, 68% of the T-GC arrays, and 50% of the T-TT arrays showed heteroplasmy. A log linear model (based on Poisson distribution and log link) suggested that these differences were significant (deviance/degrees of freedom = 1.02, P = 0.0001; see Table 3). The parameter estimates (Table 3) supported that there is a higher degree of heteroplasmy in individuals possessing GC last units in contrast to TT last units (P = 0.023). There was no significant effect of type of first repeat (P = 0.09) and no dependence between the first and last unit combined and the frequency of heteroplasmy (Chi square "first unit x last imperfect unit x homo/heteroplasmy" = 1.3, P = 0.24). The relationship between the first and last unit (P = 0.072) simply reflects that most of the sampled individuals possessed arrays with the A-GC combination.

We sequenced 8–10 individual clones from each of 14 different cloned PCR products and found that all individuals contain molecules with three-repeat units (Table 3). This was also generally the major array length variant. For haplotype L (see Figure 3), however, the major length variant was four- or five-unit arrays. Among the 14 PCR products, only one contained molecules with less than three repeats (one sequence from Ravalsjøen #5; Table 3). No directly sequenced PCR products showed "double sequences" before the third repeat unit, and all individuals classified as homoplastic possessed three-unit arrays. This is consistent with the results from cloning; molecules with less than three repeats are rare.

The ruffe and zander arrays contained identical repeat unit sequences (Table 2). The length of the array varied, however. The two cloned ruffe PCR products possessed arrays with six, seven, and 10, and 16, 17, 18, 22, and 23 units, respectively (Table 4). Furthermore, two directly sequenced ruffe PCR products showed double sequences beyond the seventh unit. The two zanders sequenced both possessed six-unit arrays, and neither showed double sequences.

Mutation mechanisms in the repeated array inferred from the distribution of point mutations and potential secondary structures: Along with the length variation polymorphism, site heteroplasmy was observed in five of the cloned PCR products (see Table 3). Eikeren #3 possessed arrays with both TT and GC last units (Table 3). Since all other individuals from this location (and from the other populations within the same geographical region) possessed GC last units (REFSETH et al. 1998 ), it seems most likely that the TT array arose from deleting the last GC unit and half the fourth unit in a four-unit array. Folding of a GC last repeat and the array repeat confirm this (Figure 4B); i.e., formation of such a structure in the L-strand during H-strand replication, would result in an array with a TT last unit.

In Røysjø #8, Svartvann #4, Mjær #11 and Mjær #15 another mutation occurred (Table 3); in Røysjø #8 all five-unit arrays possessed an A-type second unit while the three unit arrays possessed a T-type second unit. This mutation was also present in the two four-unit arrays observed in Svartvann #4 and Mjær #11, and in one of the two four-unit arrays in Mjær #15. Finally, this was the major type of array in one individual [three-unit array; haplotype G in REFSETH et al. 1998 ; Figure 3]. All arrays with an A-type second repeat also had an A-type first repeat. Thus, this mutation presumably occurred by duplication of the first unit. Consequently, both incidents of site heteroplasmy can be explained by duplications and deletions of repeats.

The phylogenetic topology inferred from the different repeats (Figure 5B), the minimum spanning network in Figure 3, and REFSETH et al. 1998 suggests that the type A perch first unit (i.e., A at position 6 in the repeat, Table 2) represents the ancestral form. However, the perch standard array unit has a T in this position. Thus, the A T transversion has spread to all units (i.e., by concerted evolution), except to the first unit in 109 of the 149 perch individuals sequenced. This observation strongly suggests that the repeats are mainly duplicated and deleted by a 5' to 3' (relative to the sequence in Figure 2) unidirectional slippage mechanism.

Imperfect duplications or deletions of units were not observed. Moreover, the last degenerated unit was never duplicated. Taken together, all the above observations imply that the duplication and deletions of units mainly occur after two or three array units are replicated, and that the secondary structures formed primarily involve one array unit and the last degenerated unit. Furthermore, as noted by BROWN et al. 1996 the high level of sequence conservation within the arrays implies high addition and deletion rates.

Transmission of heteroplasmy; distribution of array lengths in mature oocytes: To examine the possible mechanisms maintaining heteroplasmy, we analyzed the distribution of repeats in perch oocytes from a heteroplasmic individual by cloning of PCR products obtained from PCR on a single egg. The results are presented in Table 3. Both major length variants were transmitted to the three oocytes examined. In addition, PCR products obtained from three eggs from a different female were directly sequenced, and both the adult and the eggs showed double sequences (not shown). Thus, these results suggest that the high level of heteroplasmy observed may be caused by low levels of drift during oogenesis and maternal transmission of the heteroplastic condition.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Most repeats observed in the 5' end of the D-loop tend to be composed of long units (BENTZEN et al. 1988 ; JOHANSEN et al. 1990 ; BROWN et al. 1992 ; MIRACLE and CAMPTON 1995 ). In contrast, the percid repeat units are only ten bases long. Due to the short length, movement of mutations in the perch array could easily be traced. This information is particularly valuable in the investigation of possible length mutation mechanisms. In addition, the short size of the units made it possible to compare the predicted stability of secondary structures associated with different arrays. Three patterns of length variation polymorphism were observed among the three species studied. In perch, no more than five units were found among the individuals studied, and three-unit variants were dominating. The two zander individuals both possessed six-unit arrays, and the ruffe possessed arrays with six or more units. Both the perch and ruffe showed extensive heteroplasmy.

Phylogeny and origin of the tandem repeat unit:
The overall phylogenetic arrangement of the three species based on the repeated sequences alone is similar to that obtained using the entire control region fragment (Figure 5). The same topology was also suggested from phylogenetic analyses of the mtDNA cytochrome b genes from perch, yellow perch, zander, and ruffe (acc. nos. Y 14776, AJ001521, AJ001512, and AJ001511, respectively) using the mackerel sequence as outgroup (Scomber scombrus, family Scombridae; acc. no. X81564; data not shown). This indicates that the sequence of the repeat unit possesses some phylogenetic information.

The localization of the array between two conserved secondary structures, CM5' and the trnP gene, suggests that the array originated from a duplication event induced by secondary structures formed in one, or both, of these sequences. Furthermore, the relatively high level of sequence identity between the flatfish and the percids (Figure 6) indicates that the array originated either once early in the evolution of Perciformes and Pleuronectiformes lineages, or alternatively, it may have been formed several times independently by a common mechanism. We find it most likely that it has originated from a single event and has subsequently been lost in some lineages (e.g., tuna, cichlids, sea bass).

Addition and deletion of repeats in the array:
Several models have attempted to explain the cause and persistence of heteroplasmy (Rand AND HARRISON 1989 ; BUROKER et al. 1990 ; HAYASAKA et al. 1991 ; MADSEN et al. 1993A ; BROUGHTON and DOWLING 1994 ; BROWN et al. 1996 ; MUNDY et al. 1996 ; WILKINSON et al. 1997 ). The distribution of point mutations in the perch arrays and the predicted secondary structures suggest that the repeats are deleted and duplicated by unidirectional slippage. The percid repeat units differ in some respects compared to repeats observed in the 5' end of the control region among other fish species; they are shorter and do not contain the TAS element. Since the TAS is thought to be involved in termination of the D-loop DNA [i.e., a nascent H-strand terminated after a couple of hundred bases (e.g., MADSEN et al. 1993B )], the occurrence of a TAS sequence within the repeated sequence has been a major point in some of the suggested mutation mechanisms (e.g., BUROKER et al. 1990 ; BROWN et al. 1996 ). However, the percid repeats show high similarity with the Mt5 element (see Table 1), which has been associated with protein binding and regulation of the termination of D-loop DNA (OHNO et al. 1991 ; KUMAR et al. 1995 ).

BROWN et al. 1996 proposed a biochemical model emphasizing selection against tandem repeats containing binding sites for proteins involved in D-loop DNA termination. This model predicts that selection should lead to array lengths skewed around one repeat, as observed in sturgeon species (BROWN et al. 1996 ). However, as also observed among different bat species (WILKINSON et al. 1997 ), the persistence of arrays over two repeats long in the four percid species investigated suggests that this model cannot account for the array length distributions observed in this study.

The most crucial feature of the illegitimate elongation model suggested by BUROKER et al. 1990 is the inclusion of the repeated array in the D-loop DNA. This model emphasizes the competition between the D-loop DNA and the H-strand. Due to secondary structures in the D-loop DNA, or the H- and L-strand, units will be added or deleted, respectively (see BUROKER et al. 1990 ). Apart from containing the Mt5 element, the percid arrays are flanked by two TAS sequences and the CM5' between TAS1 and TAS2 (see Figure 2). Since the D-loop DNA in most vertebrates terminates 60–80 bases 5' (relative to the sequence in Figure 2) to the TAS sequence (e.g., CLAYTON 1982 ; FORAN et al. 1988 ; MADSEN et al. 1993B ), the percid repeat array is most likely included in the D-loop DNA, which would be consistent with the model of BUROKER et al.

The illegitimate elongation model predicts that with a minimum of two repeats needed for hairpin formation, the minimum number of units in the array is three (ARNASON and RAND 1992 ). Two-unit molecules were nevertheless observed at low frequency in perch (Ravalsjøen #5, Table 3). However, this does not invalidate the model; formation of a secondary structure involving the last imperfect unit and one array repeat results in deletion of one unit (see Figure 4B). A similar structure in the D-loop DNA may account for the addition of one unit to a three unit molecule, hence explaining that most arrays were composed of three or four units. Furthermore, the illegitimate elongation model, as opposed to replication slippage, can explain the low occurrence of duplications of the first unit. The distribution of mutations in the array indicated that most duplications and deletions involved secondary structures between the last degenerated repeat and one array unit. This, taken together with the observation that most arrays were three or four units long, would imply that most perch D-loop DNAs are terminated after the replication of the last imperfect unit and two array units. Some of the D-loop DNAs must, however, also extend to the first unit, since the A-type first repeat was duplicated in some individuals.

However, other mechanisms causing length mutations must exist as the tandem repeat of cod fails to meet several criteria of the model of BUROKER et al. (ARNASON and RAND 1992 ). Deviation from the model was also observed in the perch array. According to the illegitimate elongation model, mutations can only move by duplications in the 5' to 3' direction (relative to the sequence in Figure 2; BUROKER et al. 1990 ; WILKINSON and CHAPMAN 1991 ). The transversion observed in the 38 perch individuals possessing a T-type first unit is therefore either a parallel substitution, or, perhaps more likely, this mutation has "jumped" by another mechanism. One possibility is that deletion of the first A-type unit has occurred during L-strand replication. All haplotypes (C, L, H, I) with this particular mutation are connected in the maximum parsimony network in Figure 3, suggesting that this is a rare mutation, and confirming that unidirectional slippage and illegitimate elongation is the major mechanism.

Mechanisms causing differences in level of heteroplasmy and mean array lengths among percid fish:
CLARK 1988 showed by deterministic models that subtle differences in mutation rate can be responsible for large differences in level of heteroplasmy. BROWN et al. 1996 suggest that the relative magnitude of the mutation rates associated with deletion and addition of units plays a major role in determining distribution of copy numbers in sturgeon. Considering the special mode of replication in mitochondria (see CLAYTON 1991A ; CLAYTON b for reviews), different probabilities for addition and deletion of units is to be expected. Assuming that secondary structures preferentially will form in the D-loop strand and that illegitimate elongation is the major mutation mechanism in the arrays, the critical step will be deletion of units. The ruffe/zander units form less stable secondary structures compared to perch units (see Figure 4A). Thus, we predict that the ability to delete units should be reduced, and the equilibrium array length should be longer compared to perch—and this is exactly what we observe.

The log linear model suggested that the last imperfect unit had the most significant effect on the differences in level of heteroplasmy. Individuals possessing GC last units were more heteroplastic than individuals possessing TT last units. As demonstrated in Figure 4B, the same G value is obtained for both the GC- and TT-units when folding the H-strand sequences. Folding of the L-strand sequences however, gives a lower G value for the GC last imperfect unit. This implies that the elevated frequencies of heteroplasmy are caused by a higher deletion rate, which might seem contradictory to our previous findings. However, due to the weaker secondary structures (see Figure 4), the deletion rate would be much lower, and will not be able to match the duplication of array units. Hence, we would expect arrays containing a type A first unit, since these units form weaker structures compared to T-type units, and a GC last unit to be the most variable, which was what we indeed observed. This also explains the observation that four or longer arrays as major type (observed in haplotype L in Figure 3) tended to be associated with T-TT arrays.

Similar mechanisms might also account for differences in length distribution and amount of heteroplasmy among tandem repeated arrays in other species. FUMAGALLI et al. 1996 reported different levels of heteroplasmy in two species of shrews; Crocidura russula showed higher levels of heteroplasmy than Sorex araneus, where the units form stronger secondary structures (G of -16 compared to -7.5 in C. russula). However, the shrew units also contain TAS sequences, which are suggested to affect length distribution as well (BROWN et al. 1996 ). The array length and level of heteroplasmy in these species might therefore be influenced by both factors.

The results from cloning PCR products obtained from oocytes suggests that maternal transmission of molecules with different length variants is at least partly responsible for the high levels of heteroplasmy among perch populations. High degree of heteroplasmy in gonad tissue compared to other tissue types has previously been observed in rabbits, and has been attributed to stabilizing selection (CASANE et al. 1994 , CASANE et al. 1997 ). Low levels of drift during oogenesis, combined with high mutation rates, might thus be a general feature in the evolution of heteroplasmy.

Concluding remarks:
Since the tandemly repeated array was observed in species representing both subfamilies in the family Percidae (Percinae: ruffe, perch and yellow perch; Luciopercinae: zander), the array and the associated length variations are general features of percid mtDNAs. Investigating more percid species might therefore further elucidate the mutation mechanisms in the repeat arrays. Similar arrays, but composed of longer repeat units, have also been observed in flounder (19 bp) and Japanese flounder (74 bp), which belong to another order, Pleuronectiformes. The flounder repeat showed high similarity to the percid repeats and the 5' end of the Japanese flounder repeat. Hence, surveying more species among both percids and flatfishes could unravel the mechanisms organizing short repeats into higher order structures.

	FOOTNOTES

¹ Present address: Department of Toxicology, National Institute of Occupational Health, P.O. Box 8149 Dep., N-0033 Oslo, Norway.

	ACKNOWLEDGMENTS

We thank KJARTAN ØSTBYE, TOR ERIK NESBØ, UNN HILDE REFSETH, ANDREW LAMBERTSSON, BILL DAVIES, KNUT RUDI, JOHN E. STACY and two anonymous reviewers for comments on earlier drafts of this work. We thank UNN HILDE REFSETH for sequencing some of the clones from Mjær #11. We also want to thank EMELITA KARLSEN for excellent lab assistance and TONJE FOSSHEIM for sequencing the yellow perch. This work was supported by grant no. 107622/420 from the Research Council of Norway (NFR) to K.S.J.

Manuscript received October 12, 1997; Accepted for publication January 7, 1998.

	LITERATURE CITED

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

ANDERSON, S., A. T. BANKIER, B. G. BARELL, M. H. L. DE BRUIJN, and A. R. COULSON et al., 1981  Sequence and organization of the human mitochondrial genome. Nature 290:457-465[Medline].

ÁRNASON, E. and D. M. RAND, 1992  Heteroplasmy of short tandem repeats in mitochondrial DNA of Atlantic cod, Gadus morhua.. Genetics 132:211-220[Abstract].

BENTZEN, P., W. C. LEGGETT, and G. G. BROWN, 1988  Length and restriction site heteroplasmy in the mitochondrial DNA of American Shad (Alosa sapidissima). Genetics 118:509-518[Abstract/Free Full Text].

BOGENHAGEN, D. F. and B. K. YOZA, 1986  Accurate in vitro transcription of the Xenopus laevis mitochondrial DNA from two bi-directional promoters. Mol. Cell. Biol. 6:2543-2550[Abstract/Free Full Text].

BROUGHTON, R. E. and T. E. DOWLING, 1994  Length variation in mitochondrial DNA of the Minnow Cyprinella spiloptera.. Genetics 138:179-190[Abstract].

BROWN, J. R., A. T. BECKENBACH, and M. J. SMITH, 1992  Mitochondrial DNA length variation and heteroplasmy in populations of white sturgeon (Acipenser transmontanus). Genetics 132:221-228[Abstract].

BROWN, J. R., K. BECKENBACH, A. T. BECKENBACH, and M. J. SMITH, 1996  Length variation, heteroplasmy and sequence divergence in the mitochondrial DNA of four species of sturgeon (Acipenser). Genetics 142:525-535[Abstract].

BUROKER, N. E., J. R. BROWN, T. A. GILBERT, P. J. O'HARA, and A. T. BECKENBACH et al., 1990  Length heteroplasmy of Sturgeon mitochondrial DNA: an illegitimate elongation model. Genetics 124:157-163[Abstract].

CASANE, D., N. DENNEBOUY, H. DE ROCHAMBEAU, J. C. MOUNOLOU, and M. MONNEROT, 1994  Genetic analysis of systematic mitochondrial heteroplasmy in rabbits. Genetics 138:471-480[Abstract].

CASANE, D., N. DENNEBOUY, H. DE ROCHAMBEAU, J. C. MOUNOLOU, and M. MONNEROT, 1997  Nonneutral evolution of tandem repeats in the mitochondrial DNA control regions of Lagomorphs. Mol. Biol. Evol. 14:779-789[Abstract].

CECCONI, F., M. GIORGI, and P. MARIOTTINI, 1995  Unique features in the mitochondrial D-loop region of the European sea bass Dicentrarchus labrax.. Gene 160:149-155[Medline].

CHANG, D. D. and D. A. CLAYTON, 1984  Precise identification of individual promoters for transcription of each strand of human mitochondrial DNA. Cell 36:635-643[Medline].

CLARK, A. G., 1988  Deterministic theory of heteroplasmy. Evolution 42:621-626.

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

CLAYTON, D. A., 1991a  Nuclear gadgets in mitochondrial DNA replication and transcription. Trends Biochem. Sci. 16:107-111[Medline].

CLAYTON, D. A., 1991b  Replication and transcription of vertebrate mitochondrial DNA. Annu. Rev. Cell Biol. 7:453-478.

FABER, J. E., and C. A. STEPIEN, 1997 The utility of mitochondrial DNA control region sequence for analysing phylogenetic relationships among populations, species, and genera of the Percidae, pp. 129–143 in Molecular systematics of fishes, edited by T. D. KOCHER and C. A. STEPIEN. Academic Press, San Diego.

FORAN, D. R., J. E. HIXSON, and W. M. BROWN, 1988  Comparison of ape and human sequences that regulate mitochondrial DNA transcription and D-loop DNA synthesis. Nucleic Acids Res. 16:5841-5861[Abstract/Free Full Text].

FUMAGALLI, L., P. TABERLET, L. FAVRE, and J. HAUSSER, 1996  Origin and evolution of homologous repeated sequences in the mitochondrial DNA control region of shrews. Mol. Biol. Evol. 13:31-46[Abstract].

GILBERT, D. G., 1992a loopDloop, a Macintosh program for visualizing RNA secondary structure. Published electronically on the Internet, available via anonymous ftp to ftp.bio.indiana.edu.

GILBERT, D. G., 1992b SeqApp; A biosequence editor and analysis application. Preliminary release 1.9a169. Apple Computer Inc.

HAYASAKA, K., T. ISHIDA, and S. HORAI, 1991  Heteroplasmy and polymorphism in the major noncoding region of mitochondrial DNA of Japanese monkeys: association with tandemly repeated sequences. Mol. Biol. Evol. 8:399-415[Abstract].

HOELZEL, A. R., J. M. HANCOCK, and G. A. DOVER, 1991  Evolution of the cetacean mitochondrial D-loop region. Mol. Biol. Evol. 8:475-493[Abstract].

HULTMAN, T., S. STÅHL, E. HORNES, and M. UHLÉN, 1989  Direct solid phase sequencing of genomic and plasmid DNA using magnetic beads as solid support. Nucleic Acids Res. 17:4937-4946[Abstract/Free Full Text].

JOHANSEN, S. and T. JOHANSEN, 1993  The putative origin of heavy strand replication (oriH) in mitochondrial DNA is highly conserved among the teleost fishes. DNA Sequencing 3:397-399.

JOHANSEN, S., P. H. GUDDAL, and T. JOHANSEN, 1990  Organization of the mitochondrial genome of Atlantic cod, Gadus morhua.. Nucleic Acids Res. 18:411-419[Abstract/Free Full Text].

KUMAR, S., H. SUZUKI, S. ONOUE, S. SUZUKI, and N. HATTORI et al., 1995  Rat mitochondrial mtDNA binding proteins to inter-specifically conserved sequences in the displacement loop region of vertebrate mtDNAs. Biochem. Mol. Biol. Int. 36:973-981[Medline].

L'ABBÉ, D., J.-F. DUCHAINU, B. F. LANG, and R. MORAIS, 1991  The transcription of DNA in chicken mitochondria initiates from one major bi-directional promoter. J. Biol. Chem. 266:10844-10850[Abstract/Free Full Text].

LEE, W.-J., J. CONROY, W. H. HOWELL, and T. D. KOCHER, 1995  Structure and evolution of teleost mitochondrial control region. J. Mol. Evol. 41:54-66[Medline].

MADSEN, C. S., S. C. GHIVIZZANI, and W. W. HAUSWIRTH, 1993a  In vivo and in vitro evidence for slipped mispairing in mammalian mitochondria. Proc. Natl. Acad. Sci. USA 90:7671-7675[Abstract/Free Full Text].

MADSEN, C. S., S. C. GHIVIZZANI, and W. W. HAUSWIRTH, 1993b  Protein binding to a single termination-associated sequence in the mitochondrial DNA D-loop region. Mol. Cell. Biol. 13:2162-2171[Abstract/Free Full Text].

MIRACLE, A. L. and D. E. CAMPTON, 1995  Tandem repeat sequence variation and length heteroplasmy in the mitochondrial DNA D-loop of the threatened Gulf of Mexico sturgeon, Acipenser oxyrhynchus desotoi.. J. Hered. 86:22-27[Abstract/Free Full Text].

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

MUNDY, N. I., C. S. WINCHELL, and D. S. WOODRUFF, 1996  Tandem repeats and heteroplasmy in the mitochondrial DNA control region of the loggerhead shrike (Lanius ludovicianus). J. Hered. 87:21-26[Abstract/Free Full Text].

OHNO, K., M. TANAKA, H. SUZUKI, T. OHBAYASHI, and S. IKEBE et al., 1991  Identification of a possible control element, Mt5, in the major noncoding region of mitochondrial DNA by intraspecific nucleotide conservation. Biochem. Int. 24:263-271[Medline].

RAND, D. M., 1993  Endotherms, ectotherms, and mitochondrial genome-size variation. J. Mol. Evol. 37:281-295[Medline].

RAND, D. M. and R. G. HARRISON, 1989  Molecular population genetics of mtDNA size variation in crickets. Genetics 121:551-559[Abstract/Free Full Text].

REFSETH, U. H., C. L. NESBØ, J. E. STACY, L. A. VØLLESTAD, E. FIELD, and K. S. JAKOBSEN, 1998  Genetic evidence for different migration routes of freshwater fish into Norway revealed by analysis of current perch (Perca fluviatilis) populations in Scandinavia. Mol. Ecol. in press.

RUDI, K., M. KROKEN, O. DAHLBERG, A. DEGGERDAL, and K. S. JAKOBSEN et al., 1997  A rapid, universal method to isolate PCR-ready DNA using magnetic beads. BioTechniques 22:506-511[Medline].

SAMBROOK, J., E. F. FRITSCH and T. MANIATIS, 1989 Molecular Cloning: A Laboratory Manual, Ed. 2, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY.

SHEDLOCK, A. M., J. D. PARKER, D. A. CRISPIN, T. W. PIETSCH, and G. C. BRUMER, 1992  Evolution of the salmonid mitochondrial control region. Mol. Phylogenet. Evol. 1:179-192[Medline].

STEWART, D. T. and A. J. BAKER, 1994  Patterns of sequence variation in the mitochondrial D-loop region of shrews. Mol. Biol. Evol. 11:9-21[Abstract].

SWOFFORD, D. L., 1991 PAUP: Phylogenetic analysis using parsimony, Version 3.1.1. Computer program distributed by Illinois Natural History Survey, Champaign, Illinois.

TEMPLETON, A. R. and C. F. SING, 1993  A cladistic analysis of phenotypic associations with haplotypes inferred from restriction endonuclease mapping. IV. Nested analysis with cladogram uncertainty and recombination. Genetics 134:659-669[Abstract].

TEMPLETON, A. R., E. BOERWINKLE, and C. F. SING, 1987  A cladistic analysis of the phenotypic associations with haplotypes inferred from restriction endonuclease mapping. I. Basic theory and an analysis of alcohol dehydrogenase activity in Drosophila. Genetics 117:343-351[Abstract/Free Full Text].

TEMPLETON, A. R., K. A. CRANDALL, and C. F. SING, 1992  A cladistic analysis of the phenotypic associations with haplotypes inferred from restriction endonuclease mapping and DNA sequence data. III. Cladogram estimation. Genetics 132:619-633[Abstract].

TRACY, R. L. and D. B. STERN, 1995  Mitochondrial transcription initiation: promoter structures and RNA polymerases. Curr. Genet. 28:205-216[Medline].

TURNER, T. F., 1997  Mitochondrial control region sequences and phylogenetic systematics of darters (Teleostei: Percidae). Copeia 2:319-338.

WILKINSON, G. S. and A. M. CHAPMAN, 1991  Length and sequence variation in evening bat D-loop mtDNA. Genetics 128:607-617[Abstract].

WILKINSON, G. S., F. MAYER, G. KERTH, and B. PETRI, 1997  Evolution of repeated sequence arrays in the D-loop region of bat mitochondrial DNA. Genetics 146:1035-1046[Abstract].

ZARDOYA, R. and A. MEYER, 1996  The complete nucleotide sequence of the mitochondrial genome of the lungfish (Protopterus dolli) supports its phylogenetic position as a close relative of land vertebrates. Genetics 142:1249-1263[Abstract].

	NOTE ADDED IN PROOF

The complete control region of several percid species, among them also yellow perch and zander, has been published recently (FABER and STEPIEN 1997 ; TURNER 1997 ) The repeat was found in all ancestral genera. In species with repeat unit similar to the ruffe and zander (A-type repeat), the motif is repeated at least seven times. The yellow perch in FABER and STEPIEN 1997 , which had similar repeat sequence to perch, possessed three repeat units and one GC-imperfect last unit.

This article has been cited by other articles:

				L. Kvist, J. Martens, A. A. Nazarenko, and M. Orell Paternal Leakage of Mitochondrial DNA in the Great Tit (Parus major) Mol. Biol. Evol., February 1, 2003; 20(2): 243 - 247. [Abstract] [Full Text] [PDF]

				G. Hoarau, S. Holla, R. Lescasse, W. T. Stam, and J. L. Olsen Heteroplasmy and Evidence for Recombination in the Mitochondrial Control Region of the Flatfish Platichthys flesus Mol. Biol. Evol., December 1, 2002; 19(12): 2261 - 2264. [Abstract] [Full Text] [PDF]

				A. Ludwig, B. May, L. Debus, and I. Jenneckens Heteroplasmy in the mtDNA Control Region of Sturgeon (Acipenser, Huso and Scaphirhynchus) Genetics, December 1, 2000; 156(4): 1933 - 1947. [Abstract] [Full Text]

				J. S. Taylor and F. Breden Slipped-Strand Mispairing at Noncontiguous Repeats in Poecilia reticulata: A Model for Minisatellite Birth Genetics, July 1, 2000; 155(3): 1313 - 1320. [Abstract] [Full Text]