Evolution of Repeated Sequence Arrays in the D-Loop Region of Bat Mitochondrial DNA

Gerald S. Wilkinson,* Frieder Mayer; Gerald Ked*and Barbara Petri5

*Department of Zoology, University of Maryland, College Park, Maryland, tInstitut fur Zoologie II, Universitat Erlangen-Nurnberg, Erlangen, Germany, Theodor-Bovmi-Institut, UniversitatWurzburg, Wurzburg, Germany and Zoologisches Institut, Universitat Munich, Munich, Germany Manuscript received September 1, 1996 Accepted for publication March 17, 1997

ABSTRACT Analysis of mitochondrial DNA control region sequences from 41 species of bats representing 11 families revealed that repeated sequence arrays near thetRNA-Pro gene are present allin vespertilionine bats. Across 18 species tandem repeats varied in size from 78 to 85 bp and contained two to nine repeats. Heteroplasmy ranged from15% to 63%. Fewer repeats among heteroplasmic than homoplasmic individuals in a species with up to nine repeats indicates selection may act against long arrays. A lower limit of two repeats and morerepeats among heteroplasmic than homoplasmic individuals in two species with few repeats suggests length mutations are biased. Significant regressions of heteroplasmy, 0 and n, on repeat number furthersuggest that repeatduplication rate increases with repeat number.Comparison of vespertilionine bat consensus repeats to mammal control region sequences revealed that tandem repeats of similar size, sequence and numberalso occur inshrews, cats and bighorn sheep. Thepresence of two conserved protein-binding sequences in all repeat units indicates that convergent evolution has occurred by duplication of functional units.We speculate that D-loop region tandem repeats may provide signal redundancy anda primitive repair mechanism in the event of somatic mutations to these binding sites.

NIMAL mitochondria contain circulara l6kb DNA two sites in the central,conserved portion of the control A molecule, encoding 13 protein, 22 transfer RNA region (CHANGet al. 1985; KING and LOW 1987; CLAY- (tRNA) and two ribosomal RNA genes (ANDERSON et TON 1992). Each strand of the mtDNA molecule, re- al. 1981). The small size and compact organization of ferred to as heavy (H) and light (L) based on differ- themitochondrial DNA (mtDNA) genome has been ences in base composition, has different promoter re- suggested to be the result of selection for rapid organ- gions that bind nuclearcoded proteins (GHMZZANIet elle replication (HARRISON 1989; RAND 1993).However, al. 1993b; NASS 1995) and that differ in nucleotide se- recent discovery of length variation in the noncoding quence between species (KING and LOW1987). Replica- control region, which lies between the tRNA-Pro and tion of the H-strand is primed by RNA transcribed be- tRNA-Phe genes, in a varietyof vertebrate species tween the L-strand promoter (LSP) and the H-strand (DENSMOREet al. 1985; MORITZand BROWN1987; BURO- origin of replication (OH,CHANG and CLAWON 1985). KER et al. 1990; HAYASAKA et al. 1991; WILKINSONand H-strand replication is usually terminated shortly there- CHAPMAN1991; ARNMON and RAND 1992; BROWNet al. after at termination-associated sequences (TAS) re- 1992, 1996; HOELZELet al. 1993,1994; STEWART and sulting in short 7s DNA strands (DODAet al. 1981; CLAY- BAKER 1994; xu and hMON1994; YANG et al. 1994; TON 1991). 7s DNA strands remain associated with the CECCONIet al. 1995; PETRIet al. 1996; FUMAGALLIet al. L-strand and displace the original H-strand to create a 1996) is not consistent with this hypothesis and has not three-stranded structure known as the displacement or yet been adequately explained (WOLSTENHOLME1992; D-loop. In mice, only 5% of replication events continue RAND 1993). Because the proteins encoded by mtDNA beyond the control repon (BOGENHAGENand CLAWON genes play critical roles in oxidative metabolism and 1978). Sequence-specific DNA binding proteins inter- control region length might influencethe rate of act with TAS elements (MADSEN et al. 1993b) between mtDNA transcription or replication (ANNEX and WIL two conserved regions, mt 5 (OHNOet al. 1991), which LJMS 1990), the metabolic rate of the organism and is also referred to as region J (KING and LOW 1987), possibly its survival could be affected by length varia- and mt 6 (KUMAR et al. 1995). Initiation of replication tion. of the L-strand occurs only when H-strand replication Transcription of mitochondrial genes is initiated at is two-thirds complete and the conserved 012sequence, which in mammals lies between tRNA-Cy s and tRNA- Corresponding author: Gerald S. Wilkinson, Department of Zoology, University of Maryland, College Park, MD 20742. Asn, is exposed (CLAYTON1982). E-mail: [email protected] While the function of the D-loop is not well under-

Genetics 146: 1035-1048 (July, 1997) 1036 G. S. Wilkinson et al. stood (WOLSTENHOLME1992), its structure and size are plasmy among seven species of vespertilionine bats in likely to influence mtDNA replication. A high propor- order to identify evolutionary processes that influence tion of triplex to duplex forms correlates with mtDNA repeat array size. If the mutational process that gives copy number, mtRNA abundance and therate of oxida- rise to heteroplasmy is unbiased, we would expect ho- tive metabolism in different tissues (ANNEX and WIL- moplasmic and heteroplasmic individuals to have equal LIAMS 1990). The length of the 7s DNA strand, and numbers of R1 repeats (BROWNet al. 1996). Deviations therefore the size of the D-loop, varies depending on from this expectation indicate mutational bias or selec- which TAS site is used for termination (DODA et al. tion. Similarly, the proportionof heteroplasmic individ- 1981). Consequently, tandem repeats containing TAS uals is expected tobe determined by a balance between elements should alter D-loop size. TAS elements occur mutation and organelle segregation (CWUC1988; BIRKY within tandem repeats of evening bats (WILKINSONand et al. 1989) since paternal transmission is rare (HAR- CHAPMAN1991), shrews (STEWARTand BAKER1994; Fu- MSON 1989; SKIBINSKIet al. 1994). Thus, variation in MAGALLI et al. 1996),bighorn sheep (ZARDOYA et al. heteroplasmy should reflect variation in repeat muta- 1995), treefrogs FANGet al. 1994), minnows (BROUCH- tion rate if the number of organelles per cell does not TON and DOWLING1994), sturgeon(BROWN et al. 1996), vary. Finally, we compare consensus sequences from cod (ARNASON and RAND 1992; LEE et al. 1995), and vespertilionine bats with repeats to control region se- seabass (CECCONIet al. 1995). Despite the distant taxo- quences of other mammals with and without repeats to nomic affiliations among these species, in most cases determine if R1 repeated arrays have evolved multiple these R1 repeats (FUMAGALLIet al. 1996), Figure 1) are times in mammals and might influence organism func- -80 bp in length. In some fish and frogs the 80-bp tion. repeat contains two or more smaller units. In several species, R1 repeats have been predicted to form ther- MATERIALSAND METHODS modynamically stable secondary structures (BUROKERet al. 1990; WILKINSONand CHAPMAN1991; STEWARTand Sampling locations: Bats were captured by netting atroost- ing and foraging sites in Europe, Malaysia, United States, BAKER1994; FUMAGALLIet al. 1996). R1 repeat duplica- Central America, South America, and Africa. Nycticeius humer- tions and deletions are thoughtto occur by competitive alis were captured at six attic nursery colonies in Missouri strand displacement among the three strands of the D- and one in North Carolina (WILKINSONand CHAPMAN1991). loop (BUROKERet al. 1990) resulting in a unidirectional Eptesicus fuscus and Myotis lucifugus were captured in a single mutational process (WILKINSONand CHAPMAN1991). barn near the town of Princeton, Missouri. Leptonycteris cura- soae and L. nivalis were captured in day roosts in Mexico; Short, tandem repeats on the opposite side of the Glossophaga soricinawas netted in Guanacaste, Costa Rica (WIL- central conserved portion of the control region have KINSON and FLEMING1996). Four species were captured in also been reported in a variety of mammals including the Transvaal, South Africa: Epomophorus cvpturus and N. several carnivores (HOELZELet al. 1994), pinnipeds(AR- schle@nii were netted over streams near the town of Skukuzu in Kruger National Park while Nycteris thebaica and Rhinolophus NASON et al. 1993; HOELZELet al. 1993), pigs (GHMZ- clivosus were captured in a mine tunnel just southof Kruger ZANI et al. 1993a), horses (ISHIDAet al. 1994; XU and National Park. Four species were also captured from a cave ARNASON 1994), rabbits (MIGNOTTE et al. 1990; BIJU- on Tamana Hill in Trinidad, West Indies: Pteronotus parnelli, DWAL et al. 1991),shrews (FUMAGALLIet al. 1996), mar- Momoops megalophylla, Natalustumidirostris and Phyllostomus supials UANKE et al. 1994) and bats (PETRIet al. 1996; hastatus. Saccoptqx bilineata were captured at La Selva, Costa E. PETIT,personal communication). These R2 repeats Rica. Sixspecies were captured in peninsular Malaysia: Hippos- ideros diadema, R afJinis, R. sedulus, Murina suilla, Nyctophilus (FUMAGALLIet al. 1996) typically involve variable num- gouldii and Keriwoula papillosa. Seven species were collected in bers of short 6 to 30-bp units, which often contain the Germany: Nyctalus noctula in Brandenburg and Bavaria, and 4bp motif GTAC, and exhibit length variation similar E. nilssoni, M. myotis, M. bechsteini, Pipistrellus Pipistrellus, P. to that described for nuclear microsatellite loci nathusii and Vespertilio murinus in Bavaria. Seven species were (CHARLESWORTHet al. 1994). Because these short re- netted in Greece: E. serotinus, N.leisleri, N. lasiopterus, Miniopt- ems schreibersi, P. kuhli, Tadarida teniotis and R. fmmequinum. peats occur upstream from the origin of H-strand repli- Samples from R. fmmequinum were obtained from Switzer- cation, they probably do not influenceD-loop size. Con- land and Luxemburg. sequently, their formation is more likely to be caused DNA extraction,amplification and sequencing: A small by replication slippage (LEVINSONand GUTMANN1987; piece of patagia1 membrane, -10 mm‘,was excised from each individual with biopsy punches and stored either in a MADSEN et al. 1993a) than competitive strand displace- concentrated salt solution (SEUTINet al. 1991) or 95% ethanol ment. in the field. DNA was extracted from a tiny portion of each In this paper we present data on the presence and wing membrane sample using either Chelex (WALSH et al. number of tandem R1 repeats among 41 species of bats 1991), a modified salting out procedure (MILLERet al. 1988) representing most families in the order Chiroptera. By or a Qlagen DNA extraction kit following the manufacturer’s comparingsequence similaritybetween species with protocol. Control region mtDNA was amplified using two 22-bp prim- and without repeats we provide evidence for the evolu- ers, P and F (WILKINSONand CHAPMAN1991). The P primer tionary origin of R1 repeats in vespertilionine bats. We begins at position 15975 in the human proline tRNA gene then compare the number of R1 repeats and hetero- (ANDERSONet al. 1981), while the F primer ends at position Repeat Array Evolution in Bat mtDNA 1037

16425 in a conserved sequence region found in the middle of R1 repeats the control region (Figure 1, (SOUTHERNet al. 1988). Double- lad middle first stranded amplifications using PCR were performed as described in WILKINSONand CHAPMAN(1991) using AmpliTaq Conserved regions HSP (Perkin Elmer) and 40 cycles of 95” for 1 min,55” for 1.5 min, and 72” for 2 min in a Peltier thermal cycler. Amplification I “ products were purified and concentratedusing either ethanol tRNA tRNA precipitation or a silica gel-based method (Geneclean kit, Pro Control Region Phe QIAEX or Qlagen PCR-prep kit) following the manufacturers’ instructions. FIGURE1.-Schematic organization of the bat mitochon- Double-stranded PCR products were sequenced by the di- drial control region. R1 repeatedsequences (78-85 bp in deoxy chain termination method using either y””SATP and length) in vespertilionine bats are located in the left domain Sequenase 2.0 (Amersham) or by cycle sequencing with Ther- of the control regon near the tRNA-Pro. Filled squares indi- mosequenase (Amersham) using flourescent labeled primers cate termination-associated sequences (TAS) similar to those and automated sequencers (LI-COR automated sequencerin identified in humans, mice, and cows (DODAet al. 1981; KING Erlangen or an AB1 automated sequencer at the Molecular and LOW 1987; MADSEN et al. 1993b). Conserved sequence Genetics Instrumentation Facility at the University of Geor- areas in the central region (letters) and conserved sequence gia). Cycle sequencing was performed according to the manu- blocks (CSBs) are after SOUTHERNet al. (1988) and WALBERC facturer’s protocol.A nested primer (P* 5’-CCCCACCAT- and CLAY~ON(1981), respectively. The origin of H-strand rep CAACACCCAAAGCTGA-3’) was used to sequence PCR prod- lication (OH),the displacement loop (D-loop), and the L ucts generated with primersC and F (WILKINSONand and H-strand promotors (LSP, HSP) indicateapproximate CHAPMAN1991) in a single direction for S. bilineata, H. dia- locations determined for human, cow, rat and mouse (CHANG dm,R affinis, R sedulus, R fmmequinum, T. teniotis, M. et al. 1985; SACCONEet al. 1991). suilla, K. papillosa, M. schreibersi, E. nilssoni, E. serotinus, M. myotis, M. bechsteini, P. pipistrellus, P. kuhli, P. nathwii, V. muri- To assess the possibility that mutation rate or population nus, N. leisleri, N. lasiopterus, and N. gouldii. The threemegader- size may influence the number ofR1 repeats, we used the matid sequences were provided by J. WORTHINGTON-WILMER. last repeat in the array for seven vespertilionine species to Control region sequence was obtained in both directions us- estimate 8, the proportion of segregating nucleotide sites ing both the P and F primers to initiate the sequencing reac- (WATTERSON1975), and A, the heterozygosity per nucleotide tion for the remaining 18 species. site (NEI 1987). K = C X,X~A~~,where x, and xi are the frequen- repeat estimation and comparison: The number of R1 R1 cies of the ith and jth type of sequences, respectively, and A~ repeats in the arrays of homoplasmic and heteroplasmic indi- is the proportion of different nucleotides between the ith viduals was inferred by comparing PCR product sizes to a and jthtype of sequence. Both of these statistics estimate the 100-bp ladder after agarose gel electrophoresis and ethidium neutralparameter, L (NEI 1987), which for mitochondrial bromide staining under UV. Expected repeat length was esti- DNA is equal to 2N,,p (RAND et al. 1994) at equilibrium where mated from sequence information for each species. To test N4is the effective population size of females and p is the rate for differences in the frequencies of heteroplasmic and home of mutation. We compared sequences from the last repeat in plasmic genotypes between species, we used contingency chi- the array because this repeat appears to be more conserved square tests. We also used analysis of variance (ANOVA) to than any other repeat (see below) and can be more easily determine if the number of R1 repeats differed between het- aligned between species, unlike the single copy sequence in eroplasmic and homoplasmic individuals among the eight the D-loop region, which differs markedly in length between vespertilionid species for which the DNA of eight or more species (Table 1).Because the sequence of the last repeat in individuals was amplified. When determining repeat number the array could change as a consequence of a deletion event we assumed that heteroplasmic individuals contained equal that removed the last repeat, I9 and A are influenced both by amounts of each repeat array detected on the gel. the rate of nucleotide substitutions and by the rate of repeat Sequencecomparison andanalysis: All sequences were duplication and deletion. In contrast, the proportion of het- aligned with the help of the Higgins algorithm using the eroplasmic individuals is influenced only by the rate of repeat program MACDNASISand were improved by subsequent man- duplication and deletion. ual alignment. When more than one individual of a species To compareR1 repeat sequences between species with and was sequenced, a consensus sequence was generated and then without multiple R1 repeats we generated a consensus vesper- used for among species comparisons. In species having multi- tilionine sequencefor first, middle and last repeats using ple R1 repeats, the flanking single copy region, as well as the those eight species (P.pipistrellw, M. bechsteini, M. lucifuffu, first and last repeats (Figure l),were aligned in a similar way. M. adversus, N. humeralis, E. fuscus, N. gouldi, N. noctula) for Throughout this paper we refer to the repeat nearest the which the entire repeatregion was sequenced (Figure 2). We central portion of the control region as the first repeat be- then used the Lipman-Pearson algorithm and calculated the cause it undergoes replication first. The last repeat refers to maximum percentage similarity between the three corre- the repeat nearest the tRNA-Pro gene. Because prior studies sponding repeats in all other bats with multiple R1 repeats. To of bats (WILKINSON1992; WILKINSONand CHAPMAN1991) determine if repeat sequence similarity differs across repeats with R1 tandem repeats have demonstrated that some process, among bats with multiple R1 repeats we used the nonparamet- such as competitive stranddisplacement (BUROKERet al. ric Friedman test. For bat species without R1 repeats, we iden- 1990), homogenizes repeat sequences in the middle of the tified similar sequences to consensus first, middle and last array, we compared sequences between species with variable vespertilionine repeat sequences by calculating maximum numbers of repeats by aligning all repeats between the first percentage similarities for all three repeats. To determine and last repeat to generate a single middle repeat consensus sequence similarity between bats and othermammals, we used sequence. This method resulted in a consensus sequence for the three vespertilionine consensus repeat sequences to each of eight species of vespertilionine bats with repeats con- search GenBank using the Blastn algorithm. We report maxi- sisting of a first, middle and last repeat and flanking single mum similarity values and, when available, the probability of copy sequences on each side of the repeats (Figure 2). obtaining such similarity by chance (ALTSCHUI. et al. 1990). 1038 G. S. Wilkinson et al.

80 loo Consensus Al'GTATAATTGTAC--L-AT TAAATPATAT T-CCACATGA ATATTAAGCAWTACATACA TATATTAATA TTACATAATA CATTA-TATG TATAATIVFA P.p. ....G..... -...----.. A .....T.. A .-...... T...... T .....C...... -...... C.T.. N.n...... "" ...... T..A .-...... C..A...... T ...... A. ....T-A...... E. f...... A-. ....---- .. A...C...C. A...... A...... C.... N.h. A..T-"-...... C...... -T .....A...... A...... A...... AT-A... G...... N.g...... ----.. A .....TA T. .-...... A. ..CC...... B. ...C.-...... T..A... M. 1 ...... C. AC..ATTA...... C. .A..C...... T ...... AT-...... C... M.a...... C A.G.----.. A ...... C. C-...... A...... TTA.C...... C...... -...... C... M.b...... "" ...... -...... T A.....G... C ...... C...... -...... M.su...... C ..G.----.. .-..C.G.TA AT ..C...... A.G.T.C.A. .TG...-... G ...... -A... CT.T..C... K.P...... C A...---- .. .-..A.G.TC .AGA...... C..... A...... T . -...G....G C..---.CGC TCA..C.... .G....C... M.sc...... C.-.C ..G.----.. .-..C.T... .T..C...... GT...... Tp. .GATC.TD...... G.GT ACA.TA...T ATAT..C...

1 Middle repeat \

-5110 1Ib-~:i74?~70 160 10 90 200 CATTAAACTA rm:: Consensus TATT-CCACA TGAATATTM GCATOTACAT ACTTATATTA ATATTACATAATACATTA-T ATGTATAATT OrACATTAAA TTATATT-CC P.P. ...A...T.T ..A.-...... C...... C...... -...... C. T .....A. T. ..T..A.-.. N.n...... T.T ..A.-...... C.. A...... A ...... T ...... A .....T-A ...... T..A.-.. E.f. ...A...... C.A-...... AA ...... A...... A... C...C.A-,. N.h...... -...... A ...... AA...... G....AT-A ...... C-.. N.g. ...A.....C .C.A-...T...... AA.. CC ...... C.-...... T..A ...... A... CCC.T.A-.. M.l...... T.. .C..A..C...... AT-...... C ...... C..A.. M.a. ...A...... C..-...... A...... T.A.C...... C ...... -...... C ...... A... C...C.C-.. M.b...... T...... -...... A.....G ...... C...... -...... C..C-.. M.SU. """"- """"" """"" """"" """"" """"" """"" """"" """"" """"" K.P. """"- """"" """"" """""""""" """"" """"" """"" -""""- """"" M.w. """"- """"" """"" """"" """"" """""""""" "-"""- """"" """"" +First repeat '

rl? TAs 240 250 260 270 280 290 300 COnSenSUs ACATGAATAT 7TAAGCATOTA CATArrPATA TTAATATTAC ATAATACATA CAATOCOTAAmACATAC CCCATACAAT ------TU P.p...... C3-7 ...... T.. C ...... T .G ...... ATA...T. ------A.T..T- N.n...... C..A...... A,...... T...... C ...... T .A ...... G. ------...T.T. E.f...... T...AG.--- .....GC..G .-..G....T ...... G...... A.T. ------T.T.T. N.h...... A...... AC...... G ...... G ...... G. ------e.. .TT. N.g...... TA..G. ....e...... T A...... G...... A.....G... ------G ...Tc. M.l. C ...... T ...... G T.. .. ------T. .TT.C. M.a...... A..A.C. ....TCG...... G. .T...... TA.G. ------CTl'.T.. M.b...... - - ...... T.A... ..G ...... TI!... T...... AT.C..G. ------TT.T.G M.su. """"" """"" """"" """"" """"" """"__ """ ...... TAT.. ------.. .C. .T K.p. """"" """"" """"" """""""""" """"" """ .... AA ...... GGAGTTCTAA CTA... C..T M.sc. """"" """"" """"" """""""""" """"" """...A A....CE.. ------GTG.T..

7region F 310 340 320 330 350 370 180 Consensus AWCCAGACA ACATGACTAT CCACAAACCAAAGTT-AG'TC TCAE-ATCT ACCTACClCC G"ACCAA CAACCCGCCC P.P. ..TAWTI!.. ...C.G.... T..TCTC ... ..A.--.T.G AAT.--..TC .G...... T ...... N.n. ....WTl'.. .T.C...... T.-.... C..G.-.A...... C..A. GA...... G ...... E.f...... TCT.. ...C.GA...... -.... T.T..-G..T .AT.A-...... N.h. .C..ACA...... -.... .TAC.-G... .A..A-...... G ...... A... N.g...... -. .TAA.C.AA...... -...... T...... T...... M.l...... T...... T..A.TC. CA...-.... .G...... A... M.a. ....ATA... ..TC...... T.T.... .CTG.-G..A ....A-..T. .G ...... M.b. .ATIT...... T.C..+...... G-.... .T.GA-.A.. C....- ...... M.8U. ..T.T..T.. .T.C...... T...C.GTA. T.C.--G..T .....-..T. ...AT...... K.P. .CT.-..C...... GC.T... .C..-GT...... -...... G ...... A.A M.sc. ..T....T...... T.TGC..G.. CT ...--.GT .T..TA...... AT...... CG .C ...... FIGURE2.-Best alignment of consensus D-loop sequences for 11 vespertilionid species between the highly variable region nearest the tRNA-Pro gene and the end of conserved sequence block F. Only the consensus first, middle and last repeat are given for the eight species having multiple repeats (Nh, N. humeralis; Nn, N. noctula; Ef, E. fuscus; Ng, N. gouldi; Mb, M. bechsteini; Ma, M. adversus; M1, M. luc@gus; Pp, P. pipistrellus). The most similar region in three species having no repeats (Msc, M. schreibersi; Msu, M. suilla; Kp, K. papillosa) is aligned to the last, most conserved repeat in the other eight species. mt 5 and mt 6 indicate conserved sequences identified in humans, cows, pig, rat and mouse. TAS, termination-associated sequences. Array Evolution in Bat mtDNA Bat in RepeatEvolution Array 1039

We tested for possible sequence convergence by comparing dividuals indicate significant differences in three of four sequence similarities between genera with and without re- species with extreme repeat numbers, but not in four peats using the nonparametric Mann-Whitney U test. species with intermediate repeat numbers (Figure 4b). The two species with fewest R1 repeats exhibit more RESULTS repeats among heteroplasmic than homoplasmic indi- Variation in array length among bats:PCR amplifica- viduals (N. schleiffenii: Fl,6= 7.7, P = 0.0055; M. bech- tion and sequencing of D-loop mtDNAfrom 41 species steini: F1,243 = 10.4, P = 0.0013), while one of the two of bats in 11 different families reveals that tandem re- species with the highest number of R1 repeats exhibits peats occur only in 18 vespertilionid bats (Table 1),all fewer repeats among heteroplasmic than homoplasmic of which are typically placed in the subfamily Vespertili- individuals (M. Zucifugus: F1,,, = 21.6, P < 0.0001). oninae (HILLand HARRISON 1987; KOOPMAN1993). The proportion of heteroplasmic individuals should Species from each of the other threevespertilionid sub- correlate with the rate of length mutation in the ab- families exhibit significant sequence similarity with the sence of paternal transmission and with similar patterns consensus repeat sequence (Table l),but do not con- of selection (CLARK1988). Thus, a significant regres- tain multiple R1 repeats. Among the vespertilionine sion between heteroplasmy and average repeat number bats sampled, the repeated sequencevaries between 78 would suggest that length mutation changes with num- and 85 bp with most species exhibiting 81-bp repeats. ber of R1 repeats. Even if the rate of length mutation The size ofthe repeatchanges independently of phylog- is unbiased, a minimum number of repeats will exert a eny as length differences occur within three different directional bias to the mutation process that should genera, Myotis, Eptesicus and Nyctalus, that are placed result in higher repeat numbers as mutation rates in- in different tribes (HILL andHARRISON 1987; VOLLETH crease. A weighted least squares regression reveals that and HELLER 1994). heteroplasmy increases additively with repeat number Comparison of repeat array lengths among eightspe- (Figure 5a, p = 0.73, fi,5 = 13.4, P = 0.015). Signifi- cies of vespertilionine bats reveals that the modal num- cance of this regression was estimated by weighting each ber of repeats vanes from five in M. bechsteini and N. species by the square root of the proportionof individu- schleiffeniito eight in M. luciji~gus(Figure 3). A nested als represented by that species in the data set (WILKIN- analysis of variance shows that the mean number of SON 1992). R1 repeats per individual differs significantly among Additional evidence for an effect of mutation rate species within tribes (F4,790= 210.9, P < O.OOOl), but on repeat number comes from comparison of repeat not among tribes (F3,4= 0.12, P = 0.94). Post hocTukey number to 0 and x,with both statistics calculated from tests indicate that the mean number of repeats differs last repeat sequences. A weighted least squares regres- between the three species of Myotis and between two sion of 0 on repeat numberexhibited a significant posi- Pipistrellini species (Figure 4a). Thus, mean number tive relationship (Figure 5b, = 0.70, Fl,5= 11.7, P = of R1 repeats also varies independently of phylogeny. 0.0187) as did the weighted regression of x on repeat The proportion of individuals that were heteroplas- number (Figure 5c, l? = 0.90, = 43.7, P= 0.0012). mic differs between the eight species of vespertilionine R1 repeat sequence similarities among bats: To de- bats (Table 2, contingency x* = 57.0, d.f. = 7, P < termine if repeat position influences the rate ofse- O.OOOl), but shows no consistent phylogenetic pattern. quence divergence, as would be expected if mutation Partitioningthe contingency table among species rates differ at opposite ends of the array, we compared within tribes reveals that much of this effect is caused consensus first, middle, and last vespertilionine repeats by heteroplasmy differences among species of Myotis (Figure 2) to other species. Significant differences in (x2= 41.5, P< 0.001) and between the two Nycticeiini sequence similaritywere detected when each of the species (x2= 4.34, P= 0.037). Only the two Pipistrellini consensus repeat sequences was compared to the corre- species failed to show any difference in heteroplasmy sponding repeat from other vespertilionine species (x2 frequency (x2= 1.3, P = 0.26, average heteroplasmy = = 7.8, d.f. = 2, P = 0.020, Friedman Test). The last 43.3%). repeat showed the highest median similarity (87%) fol- Selection and mutation of R1 repeats: Either selec- lowed by the middle (85%)and first (84%) repeats tion or biased mutation could cause the number of R1 (Table 1).Thus, among those bats containing multiple repeats to differ between heteroplasmic and homoplas- R1 repeats, the repeat furthest from the origin of H- mic individuals. A two-way ANOVA on the number of strand replication (Figure 1) appears to be more highly R1 repeats in heteroplasmic and homoplasmic individu- conserved. als among species reveals a significant interaction be- In contrast, when each of the three consensus vesper- tween species and heteroplasmy (F7.782 = 6.2, P < tilionine repeats are aligned to maximize similarity to 0.0001) as well as the significant effect of species noted non-vespertilionine bat sequences, a single repeat is de- above (F7,782 = 83.9, P < O.OOOl), but no main effect tected, but percentagesimilarity does not differ among of heteroplasmy (Fl,782= 1.1, P = 0.29). Single species repeats (x2= 4.3, d.f. = 2, P = 0.115, Friedman Test). contrasts between heteroplasmic and homoplasmic in- Maximum similarities for an 81-bp region ranged from 1040 G. S. Wilkinson et al.

TABLE 1 Distribution of R1 tandem repeats in the Chiroptera

Family, subfamily, and Base pairs from No. of Repeat Maximumeat re species n" tRNA-Pro torepeats repeats size similarityR Pteropodidae Epomophorus cTpturus 2 55 1 64 Emballonuridae Saccoptoyx bilineata 1 114 1 75 Nycteridae Nycteris thebaica 3 69 1 73 Megadermatidae Megaderma gagas 6 26 1 61 Megaderma spasma 2 24 1 57 Megaderma lyra 2 22 1 62 Rhinolophidae Rhinolophus clivosus 1 106 1 57 Rhinolophus fmmequinum 11 106 1 70 Rhinolophus sedulus 4 106 1 74 Rhinolophus afJinis 1 106 1 61 Hipposideridae Hipposiderus diadema 1 106 1 6:! Mormoopidae Pteronotus parnelli 3 90 1 59 MormoOps megalophylla 17 I02 1 63 Phyllostomidae Phyllostomus hastatus 5 29 1 65 Glossophaga soricina 1 24 1 68 Ltptonycteris nivalis 4 24 1 68 LeptonyctPris curasaae 49 24 1 68 Mollossidae Molossus molossus 2 29 1 77 Tadarida teniotis 1 41 1 72 Natalidae Natalus tumidirostris 20 31 1 78 Vespertilionidae Kerivoulinae Kvrivoula papillosa 1 226 1 74 Murinae Murina suilla 2 100 1 77 Miniopterinae Miniopteras schreibersi 1 52 1 75 Vespertilioninae Barbastella barbastellus 1 98 2 82 Myotis myotis 191 155 3-7 89 Myotis bechsteini" 245 203 3-7 93 Myotis adversus* 4 162 4 89 Myotis lucifugus" 19 31 5-9 96 Nyctireius humeralis* 195 71 5-8 99 Nycticeinops schlei@nii 8 63 4-7 74 Eptesicus fuscus* 20 67 5-6 94 Eptesicus nilssoni 2 80 3-4 86 Eptesicus serotinus 1 79 4 88 Nyctophilus gouldi' 1 36 3 85 Vespertilio murinus 1 66 8 82 Pipistrellus pipistrellus" 8 44 5-9 91 Pipistrellus nnthusii 1 44 8 91 Pipistrellus kuhli I 92 6 93 Nyctalus nactula" 112 121 4-9 88 Nyctalus lasioptems 1 139 6 86 Nyctalus leisleri 1 141 7 89 GenBank accession numbers for sequences reported in this table are U95318-U95355.

I' Number of individuals scored for R1 repeat number. Maximum percentage similarity calculated using the Lipman-Pearson algorithm between each species D- loop sequence and threeconsensus sequence arrays estimated from eightvespertilionid species (*) as illustrated in Figure 2. All similarity percentages are adjusted to match an 81-bp sequence. Repeat Array Evolution in Bat mtDNA 1041

Nyctalus noctuia 0.8

Pipistrellini I

3456789 3456789

Eptesiws fuscus 0.8 - )r 0 Eptesicini cal 0.6- "3 0.4- 2 FIGURE S.-Frequency of 0.2- R1 repeats in eight vespertilionine species. Hetero- 0 , , I , , , , 3456789 plasmic individuals were counted as contributing 1, , 11 I equally to each repeat class. Nycticeinops schleiffenii Nycticeius humeralis Phylogeneticrelationships I o.8i among four tribes (VOL LETH and HELLER1994) are indicated by a clado iycticeiini gram.

Jlyotini 34567893456789 1 Myotis bechsteini Myotis lucifugus 3.8

3.6

3.4 3.2 , l::j,, 0.20 0 "iL3456789 3456789 3456iii Number of R1 Repeats

57% to 77% (Table 1). When sequences of vespertilie all bat species withR1 repeats also exhibit a highly nids with and without multiple R1 repeats are aligned conserved 14bp partial repeat after the last repeat (Fig- (Figure 2), three conserved regions, previously identi- ure 2). Subsequent sequence between this partial re- fied in the D-loop of other mammals as TAS (DODAet peat and the tWA-Pro geneis difficult to align between al. 1981), mt 5 (OHNOet al. 1991) and mt 6 (KUMAR et species and exhibits considerable length variation, e.g., al. 1995), can be recognized within every repeat of each from 36 bp in N. gouldi to 203 bp in M. bechsteini (Table species. As has been reported for othermammals, some 1). Similar length variation in this end of the control sequence differences occur among species in the TAS region also occurs in bats without multiple repeats (Ta- (Figure 2). Furthermore, in almost all species both the ble l).Thus, the amountof single copy DNA in the left 5' H-strand end and the middle of each repeated se- domain of the control region does not correlate with quence contain the 4bp palindrome GTAC (Figure 2). the numberof R1 repeats. A partial repeat is not evident With the exceptionof a 4bp insertion in M. luciji~gus, at the opposite end of the array. While the amount of 1042 G. S. Wilkinson et al.

TABLE 2 Frequencies of R1 repeats for eight vespertilionine species

f(3) f(4) f(5) f(6) f(7) f(8) f(9)Speciesf(8) f(7) f(6) f(5) f(4) f(3) f(345) f(45) f(456) Pipistrellus pipistrellus 1 2 Nyctalus noctula 30 31 4 1 2 Nycticeius humeralis 40 97 3 2 1 1 1 Nycticeinops schleiffenii 2 Eptesicus fuscus 15 2 Myotis myotis 12 2 64 25 1 2 2 Myotis lucifugus2 7 4 Myotis bechsteini 21 53 122 5 27 1

f(56)f(57)f(567) f(67) f(678) f(78)f(789)f(89) Prop.het. Pipistrellus1 pipistrellus 1 1 2 0.63 Nyctalus noctula Nyctalus 27 10 2 1 4 0.42 Nycticeius humeralis 38 3 12 2 0.28 3 1 Nycticeinops schleqfmii 3 0.63 Eptesicus fuscus 3 0.15 Myotis myotis 24 2 16 2 37 2 0.47 3 1 1 Myotis1 lucifugus 3 1 0.32 Myotis bechsteini 15 1 0.18 f(45), number of individuals that are heteroplasmic for four and five repeats; f(5), number of individuals that are homoplasmic for five repeats. Prop. het, proportion of individuals heteroplasmic. sequence between CSB-F and the first repeat is similar quence regions-mt 5, mt 6 and TAS-occur in the among vespertilionid species, with the exception of a same order and relative position among repeats, even 14bp insertion in K. papillosa, little sequence conserva- though their location within a repeat varies between tion is apparent in this 75- to 85-bp region (Figure 2). orders (Figure 6). R1 repeatsequence similarities between bats and other mammals: A search of GenBank using the con- DISCUSSION sensus vespertilionine first, middle and last repeats un- covered mtDNA sequences from nine additional orders Processes influencing the numberof R1 repeats The of mammals with significantly similar sequences in the number of R1 repeats is not strongly influenced by his- control region (Table 3). Examination of these se- torical factors because the modal numberof R1 repeats quences revealed the presence of R1 repeats between vanes extensively among closely related species, e.g., 74 and 80 bp in length in three additional orders: In- Myotis. The evidence presented here is more consistent sectivora, two generaand several species of shrews with repeat array length within a species being deter- (STEWARTand BAKER1994; FUMAGALLIet al. 1996); Car- mined by a balance between selection and mutation. nivora, domestic cat (LOPEZet al. 1996) and mountain Selection is implicated both by the limited distribution lion (M. CULVER,personal communication); andArtio- of repeats within a species and by comparison of repeat dactyla, bighorn sheep (ZARDOYA et al. 1995). The maxi- number among heteroplasmic and homoplasmic indi- mum sequence similarity between any of the three con- viduals. A relatively ancient origin of R1 repeats in ves- sensus vespertilionine repeats and each mammal genus, pertilionine bats (see below) and a high rate of length excluding all vespertilionid genera, differs between or- mutation (WILKINSONand CHAPMAN1991) should re- ders (H= 18.4, d.f. = 9, P = 0.031, Kruskal-Wallis Test) sult in extensive variation in repeat number amongspe- and between genera with and without R1 repeats (Z = cies in the absence of stabilizing selection (unpublished 2.71, P = 0.0066, Mann-Whitney U Test). The median simulation results). In contrast,R1 repeats in vespertili- maximum sequence similarity between vespertilionine onine bats contain between two and nine repeats with repeats and other mammal genera with repeats is 79% every species exhibiting a unimodal distribution of re- (n = 5) and without repeats is 68.5% (n = 38). Al- peats (Figure 3). Furthermore, fewer R1 repeats among though our sample of genera is not without phyloge- heteroplasmic than homoplasmic M. lucifigu, one of netic bias, with the exception of bighorn sheep, which the two species with high median repeat number, indi- are in the order exhibiting the highest sequence simi- cates that mitochondria with more than nine tandem larity to the vespertilionine repeats, evolution of R1 repeats are at some selective disadvantage. Whether P. repeats in mammals appears to involve sequence con- pipistrellus, the otherspecies with high repeat numbers, vergence. Sequence comparisons of species with and actually differs from this pattern cannot be determined without repeats reveals thatthe three conserved se- with confidence due to small sample size. R epeat Array Evolution in Bat in Evolution Array Repeat mtDNA 1043

I M. rnyotis- 1a 0.44 Nn (112)0 Mrn(191) 0 hurnemlis- N. MI (19)

N. nocfula - " Mb(245) El (20) E. fuscus- I

"I N. schleiffenii- I I v) M. bechst8ini- I 0.04 .-a, 4 5 6 7 0 x 0.03

I I a 0.02

0.01

01 I I I I

0.04

0.03 e 0.02 u 4 5 6 7 8 Number of R1 Repeats 0.01 Nh (55) FIGURE4.-(a) Mean (+SE) number of repeats for each 0 4 I I I of the eight vespertilionine species illustrated in Figure 3. 5 6 7 Means that do not differat the 5% level according to post hoc Number of R1 Repeats Tukey comparisons are connected by horizontal lines. (b) ! Average (+SE) number of repeats for heteroplasmic andho- FIGURE5.- (a) Least squares regression of proportion of moplasmic individuals from each species. heteroplasmicindividuals on mean number of R1 repeats estimated fromPCR product lengths for seven vespertilionine species (Nh, N. humeralis; Nn, N. noctula; Ef, E. fuscus; Mb, M. Two repeats appear to represent the lower limit to bechsta'ni; Mm, M. myotis; M1, M. lucifugus; Pp, P. pipistrellus). R1 repeat number in bats with multiple repeats (Table Sample size is indicated in parentheses for each species. (b) 1). Such a limit will cause biased length mutation to- Least squares regression of proportion of segregating sites,0, ward increasingrepeat number because duplication estimated from thelast repeat nearest the tRNA-Pro onmean events will bemore common than deletion events number ofR1 repeats. (c) Least squares regression of 7r on mean number of R1 repeats. The number of individuals se- among individuals with two repeats. A lower limit to quenced is indicated in parentheses in b and c for each the repeat number does not,by itself, predict the significant seven species. positive regressions observed between heteroplasmy, 0 or x,and repeat number. These results are, however, terminationproteins. Assuming that 5% of D-loop consistent with a fixed probability that any repeat in an strands lead to complete H-strand replication (BOGEN- array will fold and either beduplicated or deleted dur- HAGEN and CLAYTON1978), theprobability of H-strand ing replication. Such a process would cause length mu- replication should equal p" where p is the proportion tation rates to increase additively with repeat number. of D-loopstrands initiatingreplication and n is the num- Additional data are needed to determine if the regres- ber of repeats with TAS elements (BROWNet al. 1996). sion of 0 and x on number of repeats are also influ- Thus, D-loop strands from genomes with high numbers enced by variation in population size. of repeats should rarely lack bound protein andconse- BROWNet al. (1996) recently proposed a biochemical quently shouldbe outreplicated by genomes containing mechanism for how selection operates against mtDNA few repeats in heteroplasmic individuals. This process genomescontaining multiple R1 repeats. If protein should lead to a distribution of repeat numbers that is binding to conserved TAS sequences halts initiation of strongly skewed toward a single repeat (BROWNet al. H-strand synthesis (MADSEN et al. 1993b), thenmultiple 1996). Unfortunately, this mechanism, as described, TAS sequences would be morelikely to bindreplication does not account for variation in the number of R1 1044 G. S. Wilkinson et al.

TABLE 3 Vespertilionine bat consensus repeat similarity to other mammals

Maximum Repeat Repeat Order, species nameCommon similarity" Probability no. size Artiodactyla Alces alces Moose 69 1 Bison bison Bison 83 1.1 e-06 1 Bos taurus cow 79 2.0 e-07 1 Cervus elaphus Red deer 77 3.1 e-05 1 Ceruus nippon Sika deer 73 1 Odocoileus hemionus Mule deer 74 1 Odocoileus virginiana White-tailed deer 72 1 Owis canadensis Bighorn sheep 75 4 74 Sus scrofa Pig 80 1.8 e47 1 Carnivora Canis familiaris Dog 67 1 Felis catus Cat 82 6.1 e-12 4 80 Puma concolor Mountain lion 79 4-9 80 Cetacea Balaenoptera physalus Fin whale 77 2.1 e-06 1 Cephalorhynchus hectori Southern dolphin 74 4.0 e-05 1 Delphinus delphis Common dolphin 73 4.1 e-05 1 Globicephala melas Pilot whale 73 9.5 e-04 1 Megaptera novaeangliae Humpback whale 74 2.2 e-06 1 Phoecena phoecena Common porpoise 72 0.02 1 Tursiops truncatus Bottle-nosed dolphin 73 9.5 e-04 1 Insectivora Erinaceus europeus Hedgehog 63 1 Crocidura russula White-toothed shrew 74 5.6 e-04 2-9 78 Sorex araneus European shrew 80 4.9 e-10 5-6 78 Sorex cinereus Shrew 81 1.4 e-13 5-7 79 Sorex haydeni Shrew 80 2.0 e-05 5 79 Smex hoyi Pygmy shrew 80 3.4 e-1 1 5 79 Marsupialia Didelphis virginiana Common oppossum 65 1 Perissodactyla Diceros bicornis Black rhinoceros 61 0.008 1 Equus caballus Horse 79 1.1 e-08 1 Pinnipedia Arctocephalus forsten' New Zealand fur seal 62 1 Halichoms gypus Grey seal 62 1 Mirounga angurostris No. elephant seal 65 1 Primate Homo sapiens Human 54 1 Rodentia Clethrionomys rufocanus Bank vole 63 1 Mus musculus House mouse 65 1 '' Maximum sequence similarity percentages are relative to an 81-bp sequence. repeats in vespertilionine bats (Figure 3). Even ignoring tion increasing with repeat number noted above. Com- that the minimal repeat numberin vespertilionine bats parison of replication rates in mtDNA genomes dif- is two rather than one, neither individuals with eight fering in R1 repeat number,such as occur in vespertilio- repeats nor repeat distributions skewed toward larger nine bats, is clearly needed to test these ideas. repeat number, as occur in M. lucifugus, would be pre- Origin and evolution of R1 repeats: We found multi- dicted (Figure 3). However, if p were to increase with ple R1 repeats in all species of vespertilionine bats, but repeat number, perhaps because a slightly larger D- detected only a single R1 sequence in the three other loop somehow facilitates replication initiation, then p" vespertilionid subfamilies: Murinae, Miniopterinae and need not be maximal at n = 1. Furthermore, a positive Kerivoulinae. Phylogenetic reconstruction of genera in relationship between p and repeat number would be these subfamiliesbased on chromosomal characters consistent with the rate of repeat duplication and dele- suggests that vespertilionine species with R1 repeats are Repeat Array Evolution in Bat mtDNA 1045

human FIGURE6.-Location of conserved elements, Pro SBmouse Pro F mt 5 and mt 6, within the left domain of the mammalian mitochondrial control region. For spe- PigCSB Pro F cies having multiple R1 repeats a single repeat is shown and underlined (-, for whole repeat; Pro SBcow Pro F ---,partial repeat).The figure is based on sequences from human Pro CSB sheep Pro F (ANDERSON et al. 1981), mouse (PRAGER et al. 1993), pig (IMACKAY et al. mt 5 1986), cow (ANDERSON et Pro CSB shrew Pro F al. 1982), bighorn sheep I (ZARDOYA et al. 1995), shrew, Crocidura msula (FUMAGALLIet al. 1996), Pro CSB cat Pro F cat (LOPEZet al. 1996) and bat, N. noctula (this study). mt 6 Pro CSB bat Pro F I monophyletic (VOLLETHand HELLER1994). Unfortu- examined further suggests that R1 repeats arose from nately, the characters used by VOLLETHand HELLER sequence duplication of functional unitswithin the mi- (1994) do notcontain sufficient information to resolve tochondrial genome rather than from recent genetic the placement of the genusMyotis either within or out- exchange between the mitochondrial and nuclear ge- side a vespertilionine clade. Nevertheless, recent phylo- nomes as has recently been notedfor other taxa (LOPEZ genetic analysis using 803 bp ofND1 mitochondrial et al. 1996; SORENSONand FLEISCHER1996). Although DNA sequence for 15 vespertilionid species (F. WYER, the function of these sequence elements in regulating unpublished data) confirms that Myotis is the sister ge- mtDNA replication is unclear, at least two different nu- nus to monophyletica clade containing all other vesper- clearcoded proteins have been identified that bind to tilionine genera used by VOLLETHand HELLER(1994). these elements (MADSEN et al. 1993b; KUMARet al. 1995). Thethree remaining vespertilionid subfamilies join Furthermore, while the sequence of the repeated unit basal to Myotis in this analysis. Thus, current evidence differs in vespertilionine bats, shrews, cats, and bighorn strongly supports monophyly of R1 repeat arrays in bats. sheep, the order and spacing of the mt 5, mt 6 and In contrast, the presenceof multiple R1 repeats with TAS sequence elements in two repeats is identical in similar sequences in vespertilionine bats, shrews, cats these and other mammalian species excepthumans and bighorn sheep suggests recurrent evolution of re- (Figure 6). The orderof these conserved elements may peat arrays in mammals. The alternative hypothesis of be critical for forming stable secondary structures. Al- R1 repeat array loss in most daughter taxa ofa common though R1 repeats from fish, shrews and vespertilionine ancestor to vespertilionine bats, shrews, cats and sheep bats differ in sequence, each have been predicted to is unlikely for two reasons. R1 repeat sequences among form stable secondary structures with remarkably simi- vespertilionine bats, shrews and cats have converged, lar size and shape (BUROKERet al. 1990;WILKINSON and not diverged, with phylogenetic distance. Furthermore, CHAPMAN1991; STEWARTand BAKER1994; PETRIet dl. we found no evidence that multiple R1 repeats have 1996). These observations suggest that successful pro- ever been been lost in any species of vespertilionine tein binding in this part of the control region probably bat, cat (M. CULVER,personal communication) or involves similar secondary structures in all vertebrates. shrew, where sequences forseveral related species have Greater sequence similarities between the last repeats been examined. in the array within a species (WILKINSONand CHAPMAN The presence of three conserved sequence ele- 1991), as well as among different vespertilionine spe- ments-TAS (DODAet al. 1981), mt 5 (OHNOet al. cies, further suggest that the last repeat may be the 1991) and mt6 (KUMARet al. 1995)-in all cases of bat most important functional unit in the array. R1 repeats and in all mammalian control regions we Initial duplication of an R1 repeating unit may have 1046 G. S. Wilkinson et al. occurredthrough a modification of the competitive and D. RAND for useful discussion, and two reviewers for helpful strand displacement model (BUROKERet al. 1990) in comments. This research was supported by grants from the American Philosophical Society, Arizona Game and Fish Department, and the which a partial repeat near the tRNA-Pro gene partici- National Science Foundation to G.S.W. and a grant from the Federal pated in duplication. If the proto-repeat folded into a Agency for Nature Conservation to F.M. stem-loop structure during replication, then thepartial repeat sequence could anchor the new H-strand to the beginning of the repeat on theL-strand, thereby yield- LITERATURE CITED ing a duplication.Although the beginning and ending ALTSCHUL, S. F., W. GISH,W. MILLER,E. W. MYERSand D. J. LIPMAN, point of the repeat unit was defined differently in ves- 1990 Basic local alignment search tool. J. Mol. Biol. 215: 403- pertilionine bats and shrews (WILKINSONand CHAPMAN 410. ANDERSON,S., A. T. BANNER,B. G. BmxI., M. H. L. DE BRUIJN,A. R. 1991; STEWARTand BAKER1994; FUMAGALLIet al. 1996), COULSONet al., 1981 Sequence and organization of the human if sequences arealigned in the direction thatreplication mitochondrial genome. Nature 290: 457-465. occurs, a partial repeat of similar length can be identi- ANDERSON,S., M. H. DE BRUIN, A. R. COULSON,I. C. EPERON,F. SANCERet al., 1982 Complete sequence of bovine mitochon- fied in bats, shrews, cats and bighorn sheep after the drial DNA. Conserved features of the mammalian mitochondrial last repeat (Figure 6). A partial repeat is also found in genome. J. Mol. Biol. 156: 683-717. the closest relatives (Miniopterinae, Kerivoulinae and ANNEX, B. H., and R. S. WILLIAMS,1990 Mitochondrial DNA structure and expression in specialized subtypes of mammalian stri- Murininae) of those bats having multiple R1 repeats ated muscle. Mol. Cell. Biol. 10: 5671-5678. (Figure 3). ARNASON, E., and D. M. RAND, 1992 Heteroplasmy of short tandem Possible selection on repeats: R1 repeat se- repeats in mitochondrial DNA of atlantic cod, Gadus morhua. R1 Genetics 132: 211-220. quence, size and number convergence between vesper- ARNASON, U., A. GUI.I.BERG,E. JOHNSSON and C. LEQIE, 1993 The tilionine bats, shrews, cats and bighorn sheep, as well nucleotide sequence of the mitochondrial DNA molecule ofthe as the absence of array loss, suggest that multiple R1 grey seal, Halichoerus gypus, and a comparison with mitochondrial sequence of other true seals. J. Mol. Evol. 37: 323-330. repeats may provide some selective advantage, rather BAUMER,A,, C. ZHANC, A. W. LINNANEand P. NAGLEY,1994 Age- than just represent an example of selfish replicating related human mtDNA deletions: a heterogeneous set of dele- elements. Exactly how selection operates, however, is tions arising at a single pair of directly repeated sequences. Am. J. Hum. Genet. 54: 618-630. unclear because R1 repeat sequences may undergo se- BIJ~I-DLVAL,C., H. ENNAFAA,N. DENNEROLTY,M. MONNEROT, F. MI(; lection at multiple levels due to competition among NOTI'E et ul., 1991 Mitochondrial DNA evolution in lago- mitochondria within individuals, as well ascompetition morphs: origin of systematic heteroplasmy and organization of diversity in European rabbits. J. Mol. Evol. 33 92-102. among individuals with potentially different metabolic BIRM, C. W. J., P. FUERSTand T. MARUYAMA,1989 Organelle gene abilities. To the extent that successful organelle trans- diversity under migration, mutation, and drift: equilibrium ex- mission depends on replication rate, larger organelle pectations, approach toequilibrium, effects of heteroplasmic cells, and comparison to nuclear genes. Genetics 121: 613-627. genomes containing many R1 repeats should be at a BOGENHAGEN,D., andD.A. CIAITON, 1978 Mechanism of mito- selective disadvantage compared to smaller genomes chondrial DNA replication in mouse L-cells: kinetics of synthesis within an individual. and turnover of the initiation sequence. J. Mol. Biol. 119: 49- 68. In contrast, selection among individuals may favor an BROUGHTON,R. E., and T. E. DOWLING,1994 Length variation in increase in R1 repeat numbers for at least two reasons. mitochondrial DNA ofthe minnow Cy@'nella spilopteru. Genetics One possibility is that multiple R1 repeats could com- 138: 179-190. BROWN,J. R., A. T. BECKENBACHand M. J. SMITH,1992 Mitochon- pensate for deleterious mutations during the lifetime drial DNA length variation and heteroplasmy in populations of of an individual. Mitochondrial DNA is well known for white sturgeon (Acipmser transmontanus).Genetics 132: 221-228. its high mutation rate (BROWN1985) and lack of repair BROWN,J. R., K. BECKENRACH,A. T. BECKENBACHand M. J. SMITH, 1996 Length variation, heteroplasmy and sequence divergence mechanisms (WOLSTENHOLME1992). Multiple R1 re- in the mitochondrial DNA of four species of sturgeon (Acipmser). peats may provide a redundant signal if a mutation in Genetics 142: 525-535. one repeat alters the binding ability of a regulatory BROWN,W. M., 1985 The mitochondrial genome of animals, pp. 95-130 in Molecular EuolutionaT Genetics, edited by R. J. MACIN- protein. Alternatively, concerted evolution caused by m.Plenum Press, New York. repeat duplication and deletion could eliminate dam- BUROKER,N. E., J. R. BROWN,T. A. GILBERT,P. J. ~'HARA,A. T. BECK- aged repeat sequences. If either process occurred dur- ENBACH et al., 1990 Length heteroplasmy of sturgeon mitochondrial DNA an illegitimate elongation model. Genetics 124: 157- ing the lifetime of the animal, then multiple repeats 163. might increase longevity. Some effect on longevity CECCONI,F., M. GIORGIand P. MAKIOTTINI, 1995 Unique features seems likely because the rate of deletions and point in the mitochondrial D-loop region of the European seabass Dicentrarchus labrux. Gene 160: 149-155. substitutions in the mtDNA genome increases with age CHANG,D. D., and D. A. CIAY~ON,1985 Priming of human mito- in humans (BAUMERet al. 1994; LEE et al. 1994; KADEN- chondrial DNA replication occurs at the light-strand promoter. Proc. Natl. Acad. Sci. USA 82: 351-355. BACH et al. 1995) and mice (TANHAUSERand LAIPIS CHANG,D. D., T. W. WON(;,J. E. HIXSONand D. A. CIAITON, 1985 1995). Regulatory sequences for mammalian mitochondrial transcrip tion and replication, pp. 135- 144 in Achievemats and Perspe'ctirl<.s We thank E. BARRATT,F. BONTADINA,T. FLEMINC,K-G. HELLER,0. of Mitochondrial Research, edited by E. QUAGLIAREIILO.E. C. VON HELVERSON,J. PIR, M. CULVER,C. VOIGT,and J. WORTHINGTON- Sls\TER, F.PAI.MIERI, C. SACCONF.and A. M. KROON. Elsevier, New WILMERfor contributing sequences, DNA, or tissues, A. DONOGHUE, York. E. PETITand M. DECKERfor assistance in the laboratory, W. STEPHAN CHARLESWORTH,B., P. SNIEGOWSKIand W. STEPHAN,1994 The evo- Repeat Array Evolution in Bat mtDNA 1047

lutionary dynamics of repetitive DNA in eukaryotes. Nature 371: a major mechanismfor DNA sequence evolution. Mol. Biol. Evol. 215-220. 4: 203-221. CLARK,A. G., 1988 Deterministic theory of heteroplasmy. Evolution LOPEZ,J. V., S. CEVERIOand S. J. O’BRIEN,1996 Complete nucleo- 42 621-626. tide sequences of the domestic cat (Felis catw) mitochondrial CLA~ON,D. A,, 1991 Replication and transcription of vertebrate genome and the transposed mtDNA tandem repeat (Numt) in mitochondrial DNA. Annu. Rev. Cell. Biol. 7: 453-478. the nuclear genome. Genomics 33: 229-246. CLA~ON,D. A,, 1982 Replication of animalmitochondrial DNA. MACKAY,S. L., P. D. OLWO,P. J. LAIPIS and W. W. HAUSWIRTH,1986 Cell 28 693-705. Templatedirected arrestof mammalian mitochondrial DNA syn- CLAYTON,D. A., 1992 Transcription and replication of animal mito- thesis. Mol. Cell Biol. 6: 1261-1267. chondrial DNAs. Int. Rev. Cytol. 141: 217-232. MADSEN, C. S.,S. C. GHMZZANIand W. M. HAUWIRTH,1993a In DENSMORE,L. D., J. W. WRIGHTand W.M. BROWN,1985 Length vivo and in vitro evidence for slipped mispairing in mammalian variation and heteroplasmy in mitochondrial DNA from parthe- mitochondria. Proc. Natl. Acad. Sci. USA 90: 7671-7675. nogenetic and bisexual lizards (genus Cnemidophorus). Genet- MADSEN, C. S., S. C. GHMZZANI andW. M. HAUSWIRTH,1993b Pro- ics 110: 689-707. tein binding to a single termination-associated sequence in the DODA,J. N., C. T. WRIGHTand D. A. CLAYTON,1981 Elongation of mitochondrial DNA d-loopregion. Mol. Cell. Biol. 13: 2162- displacement-loop strands in human and mouse mitochondrial 2171. DNAis arrested near specific template sequences. Proc. Natl. MICNOTTE,F., M. GUERIDE,A”. CHAMPAGNEand J.-C. MOUNOLOU, Acad. Sci. USA 78: 6116-6170. 1990 Direct repeats in the non-coding region of rabbit mito- FUMAGALLI,L., P. TABERLET, L.FAVRE and J. HAUSSER,1996 Origin chondrial DNA. Eur. J. Biochem. 194 561-571. and evolution of homologous repeated sequences in the mito- MILLER,S. A,, D. D. DYKESand H. F. POLESKY,1988 A simple salting chondrial DNA control region of shrews. Mol. Biol. Evol. 13: 31- out procedure forextracting DNA from human nucleatedcells. 46. Nucleic Acids Res. 16: 1215. GHMZZANI,S. C., S. L. D. MACKAY, C. S. MADSEN, P. J. PIS and MORITZ,C., and W. M. BROWN,1987 Tandem duplications in ani- W. W. HAUSWIRTH,1993a Transcribed heteroplasmic repeated mal mitochondrial DNAs: variation in incidence and gene con- sequences in the porcine mitochondrial DNA D-loop region. J. tent among lizards. Proc. Natl. Acad. Sci. USA 84: 7183-7187. Mol. Evol. 37: 36-47. NASS, M. M., 1995 Precise sequence assignment of replication origin GHMZZANI,S. C., C. S. MADSEN and W.M. HAUSWIRTH, 1993b In in the control region of chick mitochondrial DNA relative to 5’ organello footprinting: analysis of protein binding at regulatory and 3’ D-loop ends, secondary structure, DNA synthesis, and regions in bovine mitochondrial DNA. J. Biol. Chem. 268: 8675- protein binding. Curr. Genet. 28: 401-409. 8682. NEI, M., 1987 Molecular Evolutionaly Genetics. Columbia University HARRISON, R. G., 1989 Animal mitochondrial DNA as a genetic Press, New York. marker in population and evolutionary biology. TREE 4 6-11. OHNO,K, M. TANAKA,H. SUZUKI,T. OHBAYASHI,S. IKEBEet al., 1991 HAYASAKA,R, T. ISHIDAand S. HORAL, 1991 Heteroplasmy and Identification of a possible control element, Mt5, in the major polymorphism in the major noncoding region of mitochondrial noncoding region of mitochondrial DNA by intraspecific nucleo- DNA in Japanese monkey: association with tandemly repeated tide conservation. Biochem. Int. 24 263-271. sequences. Mol. Biol. Evol. 8: 399-415. PETRI,B., A. VON HAESELERand S. Pmo, 1996 Extreme sequence HILL, J.E., and D. L. HARRISON, 1987 The baculum in the Vespertili- heteroplasmy in bat mitochondrialDNA. Biol. Chem. 377: 661 - oninae (Chiroptera: Vespertilionidae) with a systematic review, 667. a synopsis of Pipistrellus and Eptesicus, and the descriptions of PRAGER,E. M., R. D. SAGE, U. GYLLENSTEN,W. K. THOMAS,R. a new genus and subgenus. Bull. Br. Mus. (Nat. Hist.) Zool. 52: HUEBNERet al., 1993Mitochondrial DNA sequence diversity 225-305. and the colonization of Scandinavia by house mice from East HOELLEL,A. R., J. M. HANCOCK and G. A. DOVER,1993 Generation Holstein. Biol. J. Linn. SOC.Lond. 50: 85-122. of VNTRs and heteroplasmy by sequence turnover in the mito- RAND,D. M., 1993 Endotherms, ectotherms, and mitochondrial ge- chondrial control region of two elephant seal species. J. Mol. nome-size variation. J. Mol. Evol. 37: 281-295. Evol. 37: 190-197. RAND, D. M., M. DORFSMANand L. M. KANN, 1994 Neutral and non- HOELZEL,A. R., J. V. LOPEZ,G. A. DOVERand S. J. O’BRIEN,1994 neutral evolution of Drosophila mitochondrial DNA. Genetics Rapid evolution of a heteroplasmic repetitive sequence in the 138: 741-756. mitochondrial DNA control region of carnivores. J. Mol. Evol. SACCONE,G., G. PESOLEand E.SBISA, 1991 The main regulatory 39: 191-199. region of mammalianmitochondrial DNA structure-function ISHIDA,N., T. HASEGAWA,K. TAKEDA, M. SAKAGAMI, A.ONISHI et al., model and evolutionary pattern. J. Mol. Evol. 33: 83-91. 1994 Polymorphic sequencein the D-loop region of equine SEUTIN,G., B. N. WHITEand P. T. BOAG,1991 Preservation of avian mitochondrial DNA. Anim. Genet. 25 215-221. blood and tissue samples for DNA analysis. Can. J. Zool. 69: 82- JANKE, A., G. FELDMAIER-FUCHS, W.KELLEY THOMAS, A. VON HAESE- 90. L.ER and S. PM~O,1994 The marsupial mitochondrial genome SKIBINSKI,D.0. F.,C.GALLAGHERandC. M.BEYNON,1994 Mitochon- and theevolution of placental mammals. Genetics 137: 243-256. drial DNA inheritance. Nature 368 817-818. KADENBACH,B., C. MUNSCHER,V. FRANK,J. MULLER-HOCKERand J. SORENSON,M. D., and R. C. FLEISCHER,1996 Multiple independent NAPIWOTSKI,1995 Human aging is associated with stochastic transpositions of mitochondrial DNA control region sequences somatic mutations of mitochondrial DNA. Mutat. Res. 338: 161- to the nucleus. Proc. Natl. Acad. Sci. USA 93: 15239-15243. 172. SOUTHERN,S. O., P. J. SOUTHERNand A. E. DIZON,1988 Molecular KING, T. C., and R.L. Low, 1987 Mapping of control elements in characterization of a cloned dolphin mitochondrial genome. J. the displacement loop region of bovine mitochondrial DNA. J. Mol. Evol. 28 32-42. Biol. Chem. 262: 6204-6213. STEWART, D.T., and A. J. BAKER,1994 Patterns of sequence variation KOOPMAN,R F., 1993 Chiroptera, pp. 137-241 in MammalianSpecies in the mitochondrial D-loop region of shrews. Mol. Biol. Evol. of the World, Ed. 2, edited by D. E. WILSON and D. M. REEDER. 11: 9-21. Smithsonian Institution Press, Washington, DC. TANHAUSER,S. M., and P. J. LAIPIS,1995 Multiple deletions are de- KUMAR,S., H. SUZU~,S. ONOUE, S. SUZUKI,N. HATTORI et al., 1995 tectable in mitochondrial DNA of aging mice. J. Biol. Chem. Rat mitochondrial mtDNA-binding proteins to inter-specifically 270: 24769-24775. conserved sequences in the displacement loop region of verte- VOLLETH,M., and K-G. HELLER,1994 Phylogenetic relationships brate mtDNAs. Biochem. Mol. Biol. Int. 36: 973-981. of vespertilionid genera (Mammalia: Chiroptera) as revealed by LEE,H. C., C. Y. PANG, H. S. HSUandY. H. WEI, 1994 Ageing-associ- karyological analysis. Z. Zool. Syst. Evolutionsforsch. 32: 11-34. ated tandem duplications in the D-loop of mitochondrial DNA WALBERG,M. W., and D. A. CLAYTON,1981 Sequence and properties of human muscle. FEBS Lett. 354 79-83. of the human KB cell and mouse L cell D-loop regions of mito- LEE,W-J., J. CONROY,W. H. HOWELI.and T. D. KOCHER,1995 Struc- chondrial DNA. Nucleic Acids Res. 9: 5411-5421. ture and function of teleost mitochondrial control regions. J. WALSH,P. S., D. A. METZCERand R. HIGUCHI,1991 Chelex 100 as Mol. Evol. 41: 54-66. a medium for simple extraction of DNA for PCR-based typing LEVINSON,G., and G.A. GUTMANN,1987 Slipped-strand mispairing: from forensic material. Blotechniques 10: 506-513. 1048 G. S. Wilkinson et al.

WATTERSON,G. A., 1975 On the number of segregating sites in ge- XU, X., and U. ARNASON, 1994 The complete mitochondrial DNA netical modelswithout recombination. Theor. Popul. Biol. 7: sequence of the horse, Equus caballw: extensive heteroplasmy of 256-276. the control region. Gene 148: 357-362. WILKINSON,G. S., 1992 Communal nursing in the evening bat. Be- YANG, Y. G., Y-S. LIN,J-L. WU and C.-F. HUI,1994 Variation in mitc- hav. Ecol. Sociobiol. 31: 225-235. chondrial DNA and population structure of the Taipei treefog WILKINSON,G. S., and A.M. CHAPMAN,1991 Length and sequence Rhacophorus lazpeianus in Taiwan. Mol. Ecol. 3: 219-228. variation in evening bat d-loop mtDNA. Genetics 128 607-617. ZARDOYA,R., M. VILLALTA,M. J. LOPEZ-PEREZ,A. GARRIDGPERTIERRA, WILKINSON,G. S., and T. H. FLEMING,1996 Migration and evolution J. MONTOYAet aL, 1995 Nucleotide sequence of the sheep mito- of lesser long-nosed bats inferred from mtDNA. Mol. Ecol. 11: chondrial DNA D-loop and its flanking tRNA genes. Curr. Genet. WILKINSON,L., 1992 SKYTAT. 5.2 ed. SYSTAT, Inc., Evanston, IL. 28: 94-96. WOLSTENHOLME,D. R., 1992 Animal mitochondrial DNA structure and evolution. Int. Rev. Cytol. 141: 173-216. Communicating editor: A. G. CIARK