J Mol Evol (1988) 28:87-97
Journal of
Plant Mitochondrial DNA Evolves Rapidly in Structure,
but Slowly in Sequence
Jeffrey D. Palmer and Laura A. Herbon
Department of Biology,Universityof Michigan,Ann Arbor, Michigan48109, USA
Summary. We examined the tempo and mode of
mitochondrial DNA (mtDNA) evolution in six
species of crucifers from two genera, B r a s s i c a and
R a p h a n u s . The six mtDNAs have undergone nu-
merous internal rearrangements and therefore differ
dramatically with respect to the sizes of their
subgenomic circular chromosomes. Between 3 and
14 inversions must be postulated to account for the
structural differences found between any two species.
In contrast, these mtDNAs are extremely similar in
primary sequence, differing at only 1-8 out of every
1000 bp. The point mutation rate in these plant
mtDNAs is roughly 4 times slower than in land plant
chloroplast DNA (cpDNA) and 100 times slower
than in animal mtDNA. Conversely, the rate o f rear-
rangements is extraordinarily faster in plant mt-
DNA than in cpDNA and animal mtDNA.
Key words: Genome evolution -- Rearrangement
-- Inversion -- B r a s s i c a - - Base substitution
Introduction
Mitochondrial and chloroplast genomes contain two
vital sets of genes. First, they encode many proteins
crucial to the fundamental bioenergetic processes of
the cell. Second, they encode many of the compo-
nents necessary for the proper expression of their
own genes. Given the vital importance of these two
sets o f organellar genes, one might expect t h e m to
change very slowly during the course of evolution.
Indeed, chloroplast genes have a low rate of point
mutations (reviewed in Palmer 1985a; Wolfe et al.
Offprint requests to." J . D . P a l m e r
MolecularEvolution
~) Springer-VerlagNew York Inc. 1988
1987; Zurawski and Clegg 1987). Animal mito-
chondrial DNA (mtDNA), however, has a very rap-
id rate o f point mutations. First clearly measured
for primates (Brown et al. 1979), this rapid rate now
appears to be a general feature of most groups of
vertebrates (reviewed in Brown 1983, 1985; Wilson
et al. 1985).
Sequence studies of several plant mitochondrial
genes suggest a much lower rate of point mutations
than in animal genomes (reviewed in Palmer 1985a;
Sederoff 1987; Wolfe et al. 1987). However, several
factors limit the conclusions that could be derived
from such studies: (1) Comparison of a few kilobases
of gene sequences provides little insight upon mu-
tation rates for the entire genome, which in plants
is quite large (Ward et al. 1981) and mostly non-
coding (Makaroffand Palmer 1987). (2) These se-
quence comparisons have mostly been between dis-
tantly related angiosperms; comparisons between
more closely related species are likely to yield more
precise and informative measurements of evolu-
tionary patterns and rates (Brown et al. 1982). (3)
None of these studies have allowed adequate com-
parisons with rates of change in chloroplast and nu-
clear genomes from the same groups of closely re-
lated plants.
Rates of sequence and structural evolution can
be remarkably different for a given organelle ge-
home. Animal mtDNAs, which change so rapidly
in primary sequence, are essentially invariant in gene
order among all vertebrates (Brown 1983, 1985).
Land plant chloroplast DNAs (cpDNAs) evolve
slowly in base sequence and rarely undergo internal
rearrangement (Palmer 1985a,b). Plant mtDNAs
may also change slowly in sequence, however, they
are extremely variable in size (200-2400 kb; Ward
et al. 1981; Palmer 1985a; Pring and Lonsdale 1985),
88
contain n u m e r o u s pieces o f c p D N A (Lonsdale 1988),
a n d usually exist as a collection o f different-sized
circles (Lonsdale 1988). This diversity implies a high
rate o f length m u t a t i o n s (deletions, insertions, du-
plications). H o w e v e r , there is little i n f o r m a t i o n
available concerning rates o f inversions a n d trans-
positions, particularly as m e a s u r e d o v e r the entire
plant m i t o c h o n d r i a l genome. Extensive internal
r e a r r a n g e m e n t has b e e n suggested for m t D N A s in
Z e a (Sederoff et al. 1981) a n d N i c o t i a n a (Bland et
al. 1985), but the evidence is limited to inferences
f r o m Southern blot hybridizations, in the absence
o f a n y supporting physical m a p s . Specific m t D N A
r e a r r a n g e m e n t s h a v e been characterized in great de-
tail in several plants (Schardl et al. 1985; Bailey-
Serres et al. 1986; D e w e y et al. 1986; Y o u n g a n d
H a n s o n 1987), but these events h a v e not been placed
in the larger context o f g e n o m e - w i d e rates a n d pro-
cesses o f rearrangement.
Here, we p r o v i d e the first m e a s u r e m e n t s o f the
t e m p o a n d m o d e o f e v o l u t i o n a r y change o v e r the
entire m i t o c h o n d r i a l g e n o m e in a group o f closely
related plants. W e h a v e chosen to study species in
the genus B r a s s i c a ~ for several reasons. First, the
relatively small size ( 2 0 8 - 2 4 2 kb) o f the m i t o c h o n -
drial g e n o m e in B r a s s i c a considerably facilitates
analysis o f the entire g e n o m e ( P a l m e r a n d Shields
1984). Second, c o m p a r e d to other groups o f plants
( W a r d et al. 1981), g e n o m e size is relatively invar-
iant in B r a s s i c a (Lebacq a n d Vedel 1981; P a l m e r et
al. 1983b), m a k i n g it easier to study processes o f
r e a r r a n g e m e n t a n d p o i n t m u t a t i o n in the absence
o f the c o n f o u n d i n g effects o f extensive length m u -
tation. Third, the t e m p o a n d m o d e o f c p D N A evo-
lution are already w o r k e d out for B r a s s i c a species
( P a l m e r et al. 1983a), enabling direct c o m p a r i s o n s
o f e v o l u t i o n a r y processes in the t w o plant cyto-
p l a s m i c genomes.
W e report a r e m a r k a b l e a n d n o v e l pattern o f
e v o l u t i o n a r y change in B r a s s i c a m i t o c h o n d r i a l ge-
n o m e s . C o m p a r e d to all p r e v i o u s l y studied orga-
helle genomes, these molecules change extraordi-
narily slowly in base sequence, yet v e r y rapidly in
structure a n d a r r a n g e m e n t .
Materials and Methods
Mitochondrial DNA was isolated from 6-wk-old green leaves of
Brassica campestris cv. Purple Top White Globe (turnip), B.
napus cv. American Purple Top (rutabaga), B. oleracea cv. Bruns-
wick (cabbage), B. hirta (white mustard; USDA PI 195, 922), B.
nigra (black mustard; 84/row 1307, gift of D. Cohen), and Ra-
We shall use "'Brassica'" to refer to all five species compared
herein that are classified in the genus Brassica and also for Raph-
anus sativus (radish), because cytoplasmically it too belongs in
Brassica (Palmer et al. 1983a)
phanus sativa cv. Scarlet Knight (fertile radish) by the DNase I
procedure of Kolodner and Tewari (1972). Restriction endonu-
clease digestions, agarose gel electrophoresis, bidirectional blot
transfers to Genescreen (New England Nuclear) and Zetabind
(AMF Cuono) hybridization membranes, nick-translations, filter
hybridizations, and cloning of mtDNA fragments into plasmid
vectors were performed as described (Palmer and Shields 1984;
Palmer 1986). Complete physical maps were constructed for
mtDNAs of B. oleracea and B. napus according to a previously
described mapping strategy (Palmer and Shields 1984). Probes
used in mapping hybridizations consisted of plasmids containing
almost the entire B. campestris mitochondrial genome (Palmer
and Shields 1984) and selected portions of the B. oleracea ge-
nome, and also gel-isolated fragments from both the B. oleracea
and B. napus genomes.
Results
Tricircular M i t o c h o n d r i a l D N A s
W e previously showed that the m i t o c h o n d r i a l ge-
h o m e s o f three o f the six species e x a m i n e d here, B.
c a m p e s t r i s , B. nigra, a n d R a p h a n u s sativa, exist in
a tripartite organization as the result o f i n t r a g e n o m i c
r e c o m b i n a t i o n across a pair o f direct repeats. T h i s
r e c o m b i n a t i o n interconverts a m a s t e r c h r o m o s o m e
with two s u b g e n o m i c circles ( P a l m e r a n d Shields
1984; P a l m e r a n d H e r b o n 1986). In contrast, B.
hirta m t D N A lacks a n y r e c o m b i n a t i o n repeats 2 a n d
so exists as a single circular c h r o m o s o m e ( P a l m e r
a n d H e r b o n 1987). F u r t h e r m o r e , d i f f e r e n t se-
quences are repeated a n d r e c o m b i n e in B. c a m p e s -
tris relative to B. nigra a n d R . sativa ( P a l m e r a n d
H e r b o n 1986). W e h a v e n o w investigated the re-
c o m b i n a t i o n a l activity o f m t D N A s f r o m B. oleracea
a n d B. n a p u s and constructed c o m p l e t e restriction
m a p s o f their g e n o m e s (Fig. 1) to allow e v o l u t i o n a r y
c o m p a r i s o n s a m o n g all six o f these B r a s s i c a m t -
DNAs.
T h e s e m a p p i n g studies reveal that the s a m e 2-kb
e l e m e n t that is duplicated as a direct repeat a n d en-
gaged in i n t r a m o l e c u l a r r e c o m b i n a t i o n in B. c a m -
p e s t r i s m t D N A ( P a l m e r a n d Shields 1984) has a
similar organization in b o t h B. oleracea a n d B. na-
p u s (Fig. 1). This implies a similar tricircular struc-
ture for all three g e n o m e s (Fig. 2). Note, however,
two m a j o r differences in the organization o f the re-
c o m b i n a t i o n repeats in the three m t D N A s . First,
although overall g e n o m e size is highly conserved,
at 218 kb for tl. c a m p e s t r i s , 219 kb for B. oleracea,
2 We use the term "recombination repeat" as defined previously
(Stem and Palmer 1984) to denote a sequence present in two
copies relative to unique sequences in the genome, but which
exists in four different genomic environments as the result of
recombination occurring at high frequency (sufficient to produce
an equilibrium mixture consisting of essentially equal levels of
the recombinational isomers) between the two repeat copies. This
operational definition does not include short repeats and other
sequences that only recombine rarely
 89
83 kb -~ 135 kb --

Bg~, l.*"'~llJ~lt~,l~"l"l'~" I,IPI " I ~ I "-I'~lt~oll ='''" Ho:~,1218.4

Pst, Jl,,I -,"l!.~lJ,,I-Illgl ~ II~,-PH-PIP;,,II~,H - i-I,,r ~7.~
Kp. I r r ~'l~,~,.~l"l~ " H'~ ~"=.1"1 ma.6
s.,, ~=~ " "'IlJ"I'~I'IW,,.r;I"I ~ :""I"PPIPH't~,I ~,6.,
Nru, I " 1'"l'4~~ " I"I'J"I~ 11~4
B, c a m p e s t r i s A',~o ~
A~k~ *

"~ 170 kb -~ 49 kb *"

~,, I,~'~|~o- I ~ I,#,'-It:~.l'::H'~,l'~,l P,~,I.I~ ~.~ 1#~1216., Fig. 1. Restriction maps o f the master
Pall I ' H '~= T-kd~l,.,~,H~ll~,~ll,.,I,,llgl~ II~f~.s[~l,l[ 218.1
chromosomes of three Brassica mtDNAs,
The circular maps are shown linearized at
a BglI site within the 2-kb repeat. Arrows
Z8 4k7 84
Nru . I ~ I.,I-,Itl~lo, l O ' , - I , ~ t l . , I , l - I ~ l - P l - l - J , . ~ l ,, I'l-~ 218.7 indicate the two copies of this repeat.
C - 5 0 0 bp D - 5 0 bp Numbers between arrows indicate sizes of
B. o l e r a c e a h ' kb
A-6 ~ ~ b*
subgenomic circles (cf. Fig. 2). Restriction
-.- 98 kb -,, 123 kb -.- fragment summations are given at the right
of the maps. Filled triangles indicate eight
length mutations, which are arbitrarily
Pstl 1'17-7]~:alJ~a[[~u ~ 1 ~ I"1~'1 " I~"1 I~-PII~i~"T"III~ 22oA shown only as insertions. Variable restric-
K,,., ,'*Pt~i ~:~'IT,. ~:'~,I"I"*H~, ~' ~,laeilJ"IH~'~ol4~i~l.,1221.7 tion sites resulting from length mutation
S,'' ~-=I~'HI"'T=H'I'~,i"i"I'IP-o~'~I ~'~'~ I"1 "="Ill 220.6 and point mutation are indicated with
" ~'
N r u I [ ~ I'"1"1~1,,1"1~'1~'1 " = 1~. HVlI-,I,~,*'I'~ I'l~,~gl 22o.s filled circles and stars, respectively. The B.
&tl" 9 campestris map is modified from Palmer
B. napus 9
E-S kb
r.~o~, e-~b, G-3 kb
bp
and Shields (1984).

a b c d
and B. napus (123 kb and 98 kb) are more similar
091 I
PSI I (Fig. 2). Second, different repeat configurations are
Sa! I
present on different circles in one species relative to
: ': another. (There are two copies of the repeat per
B. campestris ~_ genome equivalent, but recombination places these
in a total o f four configurations with respect to unique
flanking sequences.) The " a " and "'b'" configurations
b a d c
are present on the master chromosome and the "c "
Pst I and " d " configurations on the subgenomic circles
Sal I
in both B. campestris and B. oleracea, whereas in
B. napus the reverse is true (Fig. 2). Moreover, in
B. campestris the " c " configuration is present on the
B.oleracea ~--
larger subgenomic circle and the " d " configuration
on the smaller one, whereas the opposite holds for
d c b a
B. oleracea (Fig. 2). These differences in circle size
Pst I and repeat location have the same cause, i.e., the
Sail
',: ',, : inversions described below.
Our map for B. oleracea m t D N A agrees with that
B. napus i ~ ~- + reported by Chetrit et al. (1984) in the placement
o f 90-95% o f the sites for the three enzymes mapped
in common. However, Chetrit et al. describe only
Fig. 2. Tricircular organization o f three Brassica mtDNAs. Ar- a single master chromosome, on which they place
rows denote the position and orientation o f a 2-kb sequence that our "a'" and "c'" configuration repeats, whereas ac-
is duplicated and engaged in high frequency intragenomic recom- cording to our mapping studies these two repeat
bination in each mtDNA. Letters distinguish each repeat in terms
of its configuration of flanking unique sequences. Cleavage maps
configurations can only exist on different genome
for three restriction enzymes are shown in regions encompassing isomers and never on the same chromosome. Be-
each of the repeats. cause the B. oleracea m t D N A s show identical pat-
terns with the three enzymes used in c o m m o n in
and 221 kb for B. napus, distances between repeats the two studies, strain differences cannot account
are different. This results in subgenomic circles of for these discrepancies. Therefore, we feel their map
unequal size in B. oleracea (170 kb and 49 kb), is in error, and in fact describes a highly improbable
whereas those in B. campestris (135 kb and 83 kb) genome arrangement.
90
5 4 3 2 1
:;~ I ,Itl ,I t
170 kb 1,.o ]" 4 9 kb
I 9 11 I I i 3.1 *
B. oleracea , I lull I 17.21 I le~lli ; ;I I/ /11 I 11.8111 I 1 lo.1,

B. campestris
83 kb ~ 17.o 135 kb
~! LI-M -" ', -~1 I

171 kb inversion 1 I 2 3 4 5tl

9 l, I. I~'I, l
6 6 kb inversion 1 5 r 4 i3 2 1

,l ~ I -t.el J l
8 4 kb inversion 1 5 4 2 t 1
tl' I t I 'e'l ' 'I
1 5 4 3 2 1

1 5 3 4 2
I ,I ,I ,I, I ,I
"- 9 8 kb " 123 kb
B. napus 1~1t, 28
I'o"lql 1. I 10.1 33
11"1 ,o II:i-I - I"1 z

Fig. 3. Relative arrangement of

similar sequences in three Brassica
mtDNAs. Numbered horizontal
campestris
,~I .1 I,~ arrows above and below the com-
8 3 kb ~-- 135 kb ~,- plete SalI maps indicate blocks o f
I ~1 t-I Jl '1 "1 sequences whose arrangement has
5 7 kb inversion 1 I 2 3 I 4 5 been conserved between genomes.
The crossing lines between Sail
I 'I" I' I .I 'I
fragments connect regions of
123 kb inversion 1 3 t 2 4 5 t
cross-hybridization. Double-head-
I .I' I. I' I 'I ed vertical arrows indicate inver-
70 kb inversion 1 t 3 5 I 4 2 sion endpoints. Horizontal arrow-
I ,I tl LI, I ,I heads indicate recombination
1 5 3 4 2 repeats.

few detectable length mutations in these genomes,

Inversions in Brassica mtDNAs
which will be discussed in a later section.
The relative arrangement of sequences in the six Second, the genomes differ significantly in linear
Brassica mtDNAs was assessed by hybridizing sequence order. The B. carnpestris and B. oleracea
cloned fragments covering 96% o f the B. campestris genomes can be divided into five linkage groups,
genome (Palmer and Shields 1984) to filter-bound where sequences within each group have the same
mtDNAs from the other five species, each digested arrangement in the two genomes, but where the rel-
with the five enzymes used in mapping. In addition, ative order and orientation o f linkage groups differ
selected clones from these five genomes were hy- between the two species (Fig. 3). A m i n i m u m o f
bridized to filters containing digests ofB. campestris three separate inversions must be postulated to ac-
mtDNA. We first compared the three most closely count for these linkage disruptions, i.e., to convert
related mtDNAs, from B. campestris, B. oleracea, one genome arrangement into the other. The B.
and B. napus. Alignment of cross-hybridizing SalI campestris-B, napus comparison also breaks the ge-
fragments from these three mtDNAs leads to two nomes into five linkage groups and postulates a min-
major conclusions (Fig. 3). First, the three genomes i m um o f three inversions (Fig. 3).
have essentially the same sequence content. Every In each of the two comparisons, two of the pos-
fragment from the B. carnpestris reference genome tulated inversions share a c o m m o n endpoint. For
hybridizes to fragments from the B. oleracea and B. example, in the B. oleracea-B, campestris compar-
napus genomes, and vice versa. There are only a ison, the right endpoints o f the 171-kb and 84-kb

inversions map to approximately the same position
within a B. campestris probe fragment o f 4.0 kb.
Only by sequencing these endpoints can we be cer-
tain o f whether they fall precisely at the same place.
Certain regions o f the genome, perhaps containing
short dispersed repeats, may promote inversional
recombination.
Knowledge o f the positions of inversions enables
us to derive simple explanations for the differences
noted in the preceding section regarding circle size
and repeat location. The net effect o f two successive
inversions, each one flipping one repeat relative to
the other (as diagrammed in each o f the compari-
sons in Fig. 3), is to retain a direct orientation of
the repeats on the master chromosome, but to move
them either farther apart or closer together. Hence,
a tricircular organization will be retained, but the
sizes o f the subgenomic circles will differ from one
species to the other.
Such a two-step inversion o f repeat-containing
linkage groups postulates an evolutionary inter-
mediate genome in which the repeats are oriented
inversely relative to one another (Fig. 4). These ge-
nomes would be direct physical analogs o f the "flip-
flop" inversion pairs found in cpDNA (reviewed in
Palmer 1985a, b) and the 2-micron plasmid o f yeast
(Broach 1982). The two inversion isomers in the
hypothesized intermediate genome are both "mas-
ter chromosomes," i.e., they contain the entire se-
quence complexity of the mitochondrial genome.
Therefore, the second evolutionary inversion in these
two-step progressions is as likely to occur in the
inversion isomer o f the circle that sustained the first
evolutionary inversion as in the unripped circle.
The net effect o f this first outcome would be to gen-
erate a tricircular genome with direct repeats, in
which the two repeat configurations o f the original
master chromosome are now on the subgenomes
and vice versa. This is exactly the situation de-
scribed in the preceding section for B. napus com-
pared to the other two genomes (Fig. 2). Figure 4
shows an evolutionary model postulating exactly
this series o f events--a "flip-flop" inversion occur-
ring in between (in a phylogenetic sense) two evo-
lutionary inversions--the net effect o f which is to
convert the B. campestris genome into a B. napus
genome.
The other three mtDNAs, from B. hirta, B. nigra,
and R. sativa, are much more highly rearranged rel-
ative to each other and to the reference genome,
from B. campestris, than are the three genomes com-
pared thus far. Relative to B. campestris, these
mtDNAs can be divided into no fewer than 11 (B.
hirta; Palmer and Herbon 1987), 12 (B. nigra; Fig.
5), and 15 (R. sativa; Fig. 5) blocks o f linkage con-
servation. Rearrangement is so extensive in these
genomes that we cannot derive a single unambig-
uous and parsimonious series o f individual muta-
91
a5) t2b ~
123 kb
d2
Inversion
4v
[I II
1 3 1 3
a , ~ , b --... d , ~ , c
kL/' kb 2 4 ~ 5
4
l
57 kb TInversion
l
70 kb ~ Inversion
~~3b 1 5
dt ~ ~ c
4 4
B. campestris B, napus
Fig. 4. Evolutionary model for the interconversion of the B.
campestris (A) and B. napus (D) mitochondrial genomes. Mol-
ecules B and C represent hypothesized evolutionary intermedi-
ates which themselves are interconvertible by a nonevolutionary
inversion--a flip-flop inversion occurring within the paired re-
combination repeats as shown in the equivalent molecules B' and
C'. Numbers and arrows denote linkage groups shown in Fig. 3.
Small letters denote the four configurations of the recombination
repeat (cf. Fig. 2).
tions that would convert one genome form into
another, as was feasible for the more closely related
genomes (Figs. 3 and 4). If, by analogy to these
simpler comparisons, we assume that all rearrange-
ments (excluding length mutations) are inversions,
then a m i ni m um o f 10, 11, and 14 inversions must
be postulated to convert the B. campestris genome
into a B. hirta, B. nigra, and R. sativa genome, re-
spectively. If each o f these putative inversions oc-
curred independently (in location) of the others, then
20, 22, and 28 inversion endpoints would be pre-
dicted for each o f the three comparisons. Because
only 11, 12, and 15 rearrangement breakpoints were,
in fact, detected (Fig. 5 and Palmer and Herbon
1987), many o f the inversions must share com m on
endpoints. This reinforces the conclusion derived
above from the more straightforward and inter-
pretable comparisons among the closely related
Brassica species (Fig. 3) and suggests that certain
regions of the mitochondrial genome are relatively
active in promoting recombination.
Nucleotide Sequence Divergence
We estimated levels of nucleotide sequence diver-
gence among the six mtDNAs based on the pro-
portion of shared restriction sites. Comparisons over
the entire genome were possible only for the three
most closely related mtDNAs, from B. campestris,
B. oleracea, and B. napus. Excluding 10 sites located
within regions that are not present in all three ge-
nomes, a total of 133 mapped sites could be com-
92
R. s a t i v a --~ 139kb 103kb
Bgll 4o t'{'11"2"3[[l"' 2, 1-~+~-'1f2-~
l"~i~']l'7. 1 ~..4 ~'~? ,~:~ ,, ,.,+ 1~,.,
] 14.3 /I 11317
23
[
2~ 2.,tl f 2.s
, 6.711140
240.4
7.51ll 9.7 i 110.3 I

Ps t I H i2tl
7. 20.5 l'" 2.,t.iF.o
2.,~I e.'l
2..'1 12.,t ,,~o,
" li9.6 ~,~
110.: ,I.Illl~l.l'.'llfo~ i ''~.., b4l~ 241.3
12.2 3. 0.6 1.4 4.2 17.8 3.7
Kpn I 1.~i5"7i1~.~1'" N. It'.' 1.1]t22
t:.2l ,.1 ] 12~1'2'8tl 1"1 ,2 I,.+.212.;-1l'1 242.s
,~,rl-,,il,,:.
i,9 17,51,1 14.2 Ii 2,
,6"155
t l I,.,I .,12.,o.s'la.o1,., 12.a-I. .so. 4.o 2.1 4.2
11,1,4 I.I .P 2 4 1 . 3
~o /1~,1":" ,1 4.5 4.8
Sal I 3..+1 ~ 34 ~.oil~.~,,.,
3.0 ,,I I,';.I, t>,41,.,I1"I
21 i t,t,~112,~;.~i..i F,,5,o2f,.,ll~,.oi
.s " " 1'7.21[117.Sl
15., 11,.51 2s 11~t1:~.,?" 243~
4.44.9

1 6 12 2,4 14 9 513 7 11 3 10 15 8

1 2 3 45 6 7 8 9 1 0 11 12 13 14 15
B. c a m p e s t r i s
-~- 3.7 "~ "~
Bgll
Pst I ]t,.~l1. ,~" t,~;tl,.;I;~lb.4
<,, :~ :" 3.~t ".21fl "'-~ 3., ,-io~l ~~ "
-I'1t,,~1 ~'PLL]]218.4
~7-3
Kpnl ,~oll~ 2. I=1'<"~-'1,-11 ,-s - 4~11.11-~1.1...~' :~,,o=~,,,.~
Sal I
I i,,.o z,,t .o ,., 8,1,.14,24 ..o I i,.2 33 "1o,.I,~101
~-~6-"
Nru I 15.s ,a ,4 a,lll 2s '.'l':ll@<'l~,s 14'1 :~t 1~,,o., I1,, cz.8.4

5 9 8 4 6 12 11 10 2 7 3
B. n i g r a 96 kb --~ 135 k b --"
2.1 2 2,6 42 3.9 5 1.6 1.2 117
Bgll ,, l;,o.sl..~lf 3.=3~ 0,,tl 1..o I,.:141].~1~ ,H 2,d,o21'M3.= 2 3 0 . 5
Pst ' "'§162 ,2 1 ,,.ob~ll,.%bol,s.~ 11..slJ,:,ll,:oQH.[-~ ~.41,II 229.7
,,.' ,,., t1,~1'.11'-'1 ,'.'1,., ~I~ [' I2"' [~ ' 4"8 I8"01 4l0428 ~1 23 ],4U~.,U.IP,,, 1.,1 23o.,

Nrul 66
' [ 15.4 I 24 3 '0 12231"1
45 87 l { ~t7~ 93 [ 154
" 111:'111G~fl:
0 ' 28 ] J7.~]"75 I lg [ 165
. i .93 J 111. f. 3 231.1
4.44.7

Fig. 5. Relativearrangement of similar sequences in three highly rearranged Brassica mtDNAs. Restriction site maps for B1 nigra
and R. sativa (fertile radish) are from Palmer and Herbon (1986); the map for B. campestris is from Palmer and Shields (1984). Arrows
directly above the three restriction site maps indicate recombination repeats. Numbered horizontal arrows above and below the B.
campestris map indicate blocks of sequences whose arrangement has been conserved relative to the genomes of R. sativa and B. nigra,
respectively. The crossing lines connect these cross-hybridizing, arrangement-conserved regions. Squiggles in the arrows indicate
sequences found in only one genome in each comparison. Restriction fragment summations are given at the right of the maps.

pared. Mutations were observed at only 2 o f these 19 PstI fragments from B. c a m p e s t r i s m t D N A , which
133 sites (Fig. 1). c o v e r 130 kb o f the g e n o m e that is colinear a m o n g
T o obtain a m o r e accurate measure o f divergence, the three m o s t closely related m t D N A s , were serially
we next c o m p a r e d all six m t D N A s with the 5 en- hybridized to filter blots containing all six m t D N A s
z y m e s used for m a p p i n g and with 15 additional d o u b l y digested with PstI a n d each o f the a b o v e
enzymes, A v a I , B a m H I , BanI, BanII, BclI, BglII, enzymes. D o u b l e digests were analyzed in order to
ClaI, EcoRI, EcoRV, HaeII, H i n d I I I , NciI, NsiI, reduce the size o f the fragments being c o m p a r e d and
XbaI, and X m n I . In order to study restriction site therefore allow better resolution o f point m u t a t i o n s
divergence without the c o n f o u n d i n g effects o f rear- f r o m small length mutations. Furthermore, because
rangement, we focused our attention o n unrear- the PstI sites defining the 19 probe fragments arc,
ranged regions o f the genome. Plasmids containing with two exceptions, absolutely conserved in all six

Table 1. Sequence differences among Brassica mtDNAs and cpDNAs
campestris napus oleracea
campestris -- 3/4002 5/4002
napus 0.07/0.33 -- 4/4002
oleracea 0.12/0.30 0.10/0.33 --
sativa 0.26/1.00 0.26/1.00 0.21 / 1.00
nigra 0.51/2.49 0.51/2.49 0.45/2.49
hirta 0.37/2.08 0.37/2.08 0.32/2.08
The number o f restriction site mutations between any two mtDNAs is given in the upper right-hand section of the matrix along with
the number of base pairs compared (these numbers take into account the sites cut by both KpnI and BanI). Values in the lower leR-
hand section correspond to the percentage sequence difference between pairs of mtDNAs (left of slash) and cpDNAs (right of slash)
and are calculated as 100p, where p is the estimated number of substitutions per base pair from Eq. 3 of Brown et al. (1979). The
m t D N A data are from this study and the cpDNA data are from Palmer et al. (1983a)
genomes, we were better able to compare regions
lying very close to breakpoints without any of the
confounding effects of the rearrangements.
The six Brassica mtDNAs are very similar in
nucleotide sequence as estimated by their propor-
tion of shared restriction sites (Table l). The most
divergent pair of species are over 99% identical in
primary sequence, whereas the three most closely
related genomes, those ofB. campestris, B. oleracea,
and B. napus, are roughly 99.9% identical. More-
over, even these very low sequence divergence val-
ues (0.07-0.80%; Table 1) may be overestimated,
because some of the site changes could be due to
very small length mutations rather than base sub-
stitutions. Therefore, our estimates of sequence di-
vergence should be considered maximal ones.
The entire, roughly 220-kb genomes of three
species (B. campestris, B. oleracea, and B. napus)
were compared with the five enzymes used in map-
ping, and 130 kb of each genome was compared
with the 15 additional enzymes, producing a total
sample size of 675 restriction sites and 4002 nu-
cleotide pairs, or almost 2% of the genome. At the
other extreme, only about one-fifth (45.6 kb) of the
B. nigra and R. sativa mitochondrial genomes could
be compared, producing a sample size of 253 sites
and 1496 bp. Because roughly equal proportions of
site mutations were observed in the genomewide
comparisons of the three closely related Brassica
species as in the more extensive sampling of their
unrearranged regions, it appears that rearranged and
unrearranged regions are equally susceptible to point
mutations. Hence, the limited (in physical terms)
sampling of the more rearranged genomes should not
produce any bias in detection of point mutations.
Size Variation and Length Mutations
An accurate assessment of length mutations (dele-
tions/insertions) over the entire mitochondrial ge-
93
sativa nigra hirta
5/1900 9/1759 8/2163
5/1900 9/1759 8/2163
4/1900 8/1759 7/2163
-- l 0/1496 l 0/1900
0.67/2.20 -- 14/1759
0.53/1.90 0.80/1.00 --
nome was possible only for the three most similar
mtDNAs, which are also very similar in overall size
(218-221 kb; Fig. 1). Length mutations among these
three mtDNAs were detected by comparing frag-
ment sizes in two ways-- first, by hybridizing clones
covering the entire mitochondrial genome to blots
containing adjacent lanes of the three mtDNAs di-
gested singly with the five mapping enzymes, and
second, by the double digest hybridizations de-
scribed in the preceding section. This latter assay
was more sensitive than the first because the average
fragment being compared was smaller. Thus, our
study was biased to detect small length mutations
in this unrearranged 130-kb portion of the genome.
A m i n i m u m of eight length mutations, ranging
in size from 50 bp to 6 kb, distinguishes among the
mtDNAs ofB. campestris, B. oleracea, andB. napus.
Six of the eight length mutations distinguish the B.
napus genome from the other two genomes (Fig. 1),
and three of these mutations are fairly large (3, 5,
and 6 kb), so that the absolute value of these size
differences far exceeds the overall genome size dif-
ference. For example, the 2-kb size difference be-
tween the 221-kb B. napus and 219-kb B. oleracea
genomes reflects the combined insertion and dele-
tion of over 14 kb of DNA in a minimum of seven
separate events. In contrast, the B. campestris and
B. oleracea genomes differ by only 550 bp in size
and by only two detectable length mutations.
The other mitochondrial genomes cover a much
wider range of sizes--208 kb for B. hirta (Palmer
and Herbon 1987), and 231 kb for B. nigra and 242
kb for R. sativa (Palmer and Herbon 1986). Rear-
rangement and size variation in these genomes were
too extensive to allow analysis of most length mu-
tations. However, Fig. 5 does show the largest length
mutations in the B. nigra and R. sativa genomes
relative to that of B. campestris. These are of two
types, sequence duplications (creating recombina-
tion repeats; Palmer and Herbon 1986) and dele-
tions/insertions of unique sequences (Fig. 5).
94

B. oleracea 1 I 2 31 4 I ~ I 1

B. napus : 1 I 2 In 4 I 5 :

B. hirta =1 1 I= 2 I 3 41 s 16171 a 191 lo Im., - k

B. nigra =1 1 -I 2 I 3 4 I 5 6 I 7 =1ol , I ,o Ill I'~'

R. sat iva | 1 121 3 ~4~s 161 7 8 ~1~ 11 [ 12 Bill3~ 14 [ 15

Composite =1 Bill II II I II Imll I I -'11 I~= I I f I=

2 3 456 7 8 9 1011 12 13 14 15 16 17 18 19 20 21 22
Map 1

B.campestris coxl! atp6 coxl cob'c~xll/ /nbhl ~tp9 atpA 26S 18S cox.
eoxlll rps13

t, 1,1, t, I ~ 1 , 1 , 1 , I,1,11i i I J l l l , I,I, I, 1,1,1, I,I

o lo 20 30 40 so 6o 70 s o 90 loo 11o 12o 13o 14o 1so 1so 1to ,ao19o 200 21o 22o

Fig. 6. Summary o f m t D N A rearrangement breakpoints relative to the reference genome, from B. campestris. Each individual species
map shows its rearrangement breakpoints (indicated with horizontal slashes), deletions larger than 1 kb in size (indicated with solid
boxes), and blocks of unrearranged sequences (numbered regions between breakpoints) relative to the map of B. campestris (cf. Figs.
3 and 5). The composite map summarizes rearrangements and deletions for all five comparisons. The B. campestris gene map is from
Makaroff and Palmer (1987). Size scale is in kilobases.

Discussion with ambiguous polarities in the phylogeny depicted

General Features of mtDNA Rearrangement
in Brassica
This study is the first to measure mutation rates
over an entire plant mitochondrial genome and doc-
uments a unique tempo and mode o f evolution for
plant mitochondrial DNA. Nucleotide substitutions
occur at a very slow rate in these Brassica mtDNAs,
whereas internal rearrangements, predominantly in-
versions, occur frequently. Thus, the Brassica mi-
tochondrial genome may be viewed as a collection
o f unchanging sequences whose relative arrange-
ment is extremely fluid.
The overall extent and nature of m t D N A rear-
rangements in Brassica are summarized in a spatial
framework in Fig. 6 and in a temporal, phylogenetic
framework in Fig. 7. Figure 6 shows that many o f
the rearrangement breakpoints found in the five ge-
homes that were compared to the B. campestris ge-
nome map to the same approximate locations rel-
ative to B. campestris. This probably reflects two
factors. First, as discussed in the Results section on
inversions, many rearrangements that occurred in-
dependently in time are likely to have been fostered
by recombination at a limited number o f shared
endpoint sequences. Second, some rearrangements
undoubtedly occurred in the c o m m o n ancestors of
two or more o f these species. Unfortunately, the
complexity o f rearrangement in these genomes, to-
gether with the apparent high incidence of shared
rearrangement breakpoints, makes impossible an
unambiguous determination of phylogenetic direc-
tion o f individual rearrangement events. It is for
this reason that most rearrangements are portrayed
in Fig. 7.
The composite rearrangement map shown in Fig.
6 reveals that rearrangement breakpoints occur on
average every l0 kb in the 208-257-kb Brassica
mitochondrial genome and that the longest block o f
unrearranged sequences (no. 9) is only about 10%
o f the length o f the genome. Many of the identified
plant mitoehondrial genes map singly within indi-
vidual rearrangement units and several o f these genes
(atp6, coxI, atpA) are closely flanked on both sides
by rearrangements in one genome or another. Fur-
thermore, although cob and coxII are linked in all
Brassica genomes, they are located far apart in
m t D N A s o f maize (Lonsdale 1988) and sunflower
(Siculella and Palmer 1988). In addition, o f the four
genes (coxIII, rps 13, ndh 1, atp9) that remain linked
on sequence block 9 in the composite Brassica map
(Fig. 6), only two, rps 13 and ndh l, are also linked
in such plants as maize, tobacco, and sunflower
(Bland et al. 1986; Lonsdale 1988; Siculella and
Palmer 1988). In general, then, gene order is incon-
sequential to the proper function o f plant mito-
chondrial genes.
There are two indications that rearrangement has
been most extensive in the lineage leading to radish,
R. sativa. First, the fertile radish line examined in
this study is more rearranged relative to B. cam-
pestris than any other Brassica species, including
two species (B. nigra and B. hirta) that are phylo-
genetically more distantly related (Fig. 7). Second,
the mtDNAs o f this fertile radish line and a cyto-
plasmic male sterile (CMS) radish line have been
shown to be extensively rearranged (approx. 10 in-
versions), in spite of a high degree o f primary s e -

I <
i~ o t 10 kb
~lnmtion za~eat
I
o t 2 kb (oo~ zz)
rux~b~unatio. ~ - ~ t
I
<. 3
I
> laHm of
~ m m
< 3--~
~m.~slcm
D. c a n m s t : r t s D. o l _e~___ ]}. ha:us
cau:~je
218 219 221
135 + 83 170 + 49 123 + 98,
2 y-
quence similarity in both mtDNA and cpDNA of
the two radish lines (Makaroff and Palmer 1988).
We are presently sequencing the endpoints of a
number of these radish-derived rearrangements in
order to examine mutations potentially involved in
CMS; these studies may also provide clues con-
cerning the greater instability of m t D N A in radish.
Although it is impossible to polarize Brassica
mtDNA inversions, the evolutionary direction of
several large insertions and deletions can be inferred
with reasonable confidence. In particular, the du-
plication o f the coxlI gene as a recombination repeat
is a relatively recent event in Brassica evolution,
occurring in the common ancestor of the three most
closely related species examined, B. campestris, B.
oleracea, and B. napus (Fig. 7). The duplication rep-
resented by the 10-kb recombination repeat is pres-
ent in only some taxa in each of the two major clades
of the Brassica phylogeny (Fig. 7). We favor the idea,
expressed in Fig. 7, that this duplication was present
in the common ancestor of the genus and was sub-
sequently lost twice independently--once in B. hirta
and once in the 2-kb recombination repeat-contain-
ing lineage. We prefer this explanation to the equally
parsimonious one of two parallel gains, once in B.
nigra and once in R. sativa. It seems easier to imag-
ine parallel loss o f one copy of a duplicate (and hence
nonessential) sequence than its parallel duplication
in a genome of the size and complexity of the Bras-
95
10 k b r z~aJ~c
sir~jle ~ germ
< 10-11
14
z ~ ,'r.
zeca~b. Fig. 7. Phylogenetic history of m t D N A
rearrangements in Brassica. Top Cytoplas-
mic phylogeny for Brassica based on cp-
I DNA restriction site mutations (Palmer et
B. sat~va B. s a t t v a D. ld.~ al. 1983a). This phylogeny is cladisticaily
rebec.tie 04S black derived and is not intended to convey di-
mustazd vergence times. Numbers of rearrangments
are given relative to the reference genome,
B. campestris, except for the 10 rearrange-
242 257 231 208
ments that have been shown (Makaroff
and Palmer 1988) to distinguish the CMS
139 + 103 130 + 127 135 + 96 and fertile mtDNAs of R. sativa. Bottom
Sizes of master chromosomes and (where
present) subgenomic circles resulting from
11o
t'~twb. high frequency recombination at the indi-
z'qpmt cated recombination repeats.
sica mitochondrial genome. Although these large
duplications in plant mtDNA are relatively ephem-
eral compared to those in cpDNA, which can persist
for several hundred million years (Palmer 1985a,b),
they are long-lived relative to tandem duplications
in animal mtDNA, which are usually restricted to
particular individuals or populations (Moritz and
Brown 1987).
Novel Mode o f Evolution of Plant mtDNA
Comparisons of plant mtDNA with cpDNA and
animal mtDNA reveal striking contrasts in genome
evolution. The much smaller mtDNAs of animals
are extremely invariant in genome arrangement (all
vertebrate mtDNAs have the same gene order) yet
undergo extraordinarily rapid changes in primary
sequence (Brown 1983, 1985; Wilson et al. 1985).
The mean intraspecific divergence in animal mtDNA
ranges from 0.18% to 4.10% within 12 different
mammalian species (Wilson et al. 1985). This com-
pares to a mean interspecific divergence of 0.37%
(Table 1) and an intraspecific divergence of 0%
(Palmer 1988) in Brassica mtDNAs. We estimate
plant mtDNA sequence change to be on the order
of 100 times slower than for animal mtDNA. This
is based on calculations that cpDNA mutates 10--
100 times slower than animal m t D N A (Zurawski et
al. 1984) and on our own findings that m t D N A in
96
Brassica mutates 4 times slower than cpDNA (see
below). Overall then, plant and animal m t D N A evo-
lution are completely opposite with respect to both
primary sequence and genome arrangement.
Gene order is almost as highly conserved in land
plant cpDNA as in animal m t D N A (Palmer 1985a,b;
Gray 1986) and is therefore orders of magnitude
more constrained than in plant mtDNA. Indeed, we
earlier demonstrated (Palmer et al. 1983a) complete
colinearity among the cpDNAs of the six Brassica
species whose mtDNAs are shown to be so highly
rearranged (Figs. 3 and 5). This disparity in rear-
rangement rates for the two plant cytoplasmic ge-
nomes far exceeds their difference in point mutation
rates. Comparisons among the same Brassica species
reveal that substitution frequencies across the ge-
nome are about four times slower for plant m t D N A
than for cpDNA (Table 1). This number, which is
based on indirect sampling of substitutions by re-
striction site analysis, is in excellent agreement with
the recent estimate of a threefold slower rate of syn-
onymous substitutions for plant mitochondrial genes
compared to chloroplast genes (Wolfe et al. 1987).
Our total genome divergence estimates and those
based on single genes (Wolfe et al. 1987) are in dis-
agreement with those of McClean and Hanson
(1986). They estimated nucleotide substitutions
based on the shared fragment method, which as-
sumes that all fragment changes are due to substi-
tutions. However, they pointed out that their esti-
mates would not reflect actual accumulation of
substitutions if the mitochondrial genomes con-
tained many rather than a few rearrangements. The
assumption of few rearrangements is clearly not met
for Brassica mtDNAs, nor apparently for those in
Zea (Sederoffet al. 1981; Schardl et al. 1985; Dewey
et al. 1986) and in Nicotiana (Bland et al. 1985) and
Petunia (Young and Hanson 1987), two genera in
the same family as Lycopersicon. Because rearrange-
ment appears to be the major mode of plant mtDNA
change, we believe that the Lycopersicon divergence
values calculated by McClean and Hanson (1986)
are significantly overestimated.
Mechanisms of Plant mtDNA Evolution
Why does plant m t D N A rearrange so rapidly, yet
remain so invariant in primary sequence? The large
size of plant mitochondrial genomes and their abun-
dance of noneoding sequences suggest that a high
proportion of possible rearrangements is likely to
be tolerated, i.e., will not disrupt gene function. Our
transcriptional studies indicate that at most only
about one-third of the 218-kb mitochondrial ge-
nome ofB. campestris is likely to have coding func-
tions and that almost all genes are transcribed singly
(Makaroff and Palmer 1987). It seems reasonable
therefore to expect that only 10% or less of the DNA
is genie in the larger plant genomes (Ward et al.
1981). In contrast, land plant cpDNAs are densely
packed with genes, many of which are cotranscribed
(Gray 1986), whereas animal mtDNA lacks spacer
sequences almost entirely and in mammals at least
is transcribed into a single genome-sized RNA
(Clayton 1984; Brown 1985). Thus, gene packing
constraints on rearrangement should be extremely
relaxed in plant m t D N A compared to both cpDNA
and animal mtDNA. Furthermore, plant mitochon-
drial genomes seem able to tolerate at least certain
rearrangements that actually do affect gene struc-
ture; these generate novel, recombinant genes that
appear to be involved in cytoplasmic male sterility
(Bailey-Serres et al. 1986; Dewey et al. 1986; Young
and Hanson 1987).
A second major factor promoting rearrangement
in plant mtDNAs is the prevalence of short dis-
persed repeats that could serve as the points o f cross-
over for homologous recombination events. Such
repeats are virtually absent from animal m t D N A
and from most cpDNAs (Palmer 1985a,b; Brown
1985). In contrast, several families of short dis-
persed repeats exist in Brassica mtDNAs (Palmer
and Shields 1984; Shirzadegan and Palmer, unpub-
lished data). Furthermore, hybridization studies re-
veal that these repeats are located near or at the
precise breakpoints of the three inversions (Fig. 3)
that distinguish mtDNAs of B. campestris and B.
oleracea (M. Shirzadegan and J. Palmer, unpub-
lished data). Sequencing studies are now in progress
to clarify the relationship between repeats and in-
versions in Brassica mtDNA.
The slow rate of point mutations in plant m t D N A
is intriguing. The fact that the genomes are mostly
noncoding might cause one to expect high levels of
sequence variation, particularly when measured by
restriction site studies of the entire genome. Yet,
plant m t D N A accumulates substitutions 3--4 and
100 times more slowly than the densely gene-packed
DNAs of chloroplasts and animal mitochondria, re-
spectively. We therefore favor the idea that the mu-
tation rate in plant m t D N A is driven by factors
affecting mutation pressures, rather than by strin-
gent selection pressures. For example, we predict
that plant mitochondria are likely to contain highly
efficient systems for repairing damage to their DNA,
and perhaps, a relatively error-free DNA replication
system. In contrast, the high mutation rate in animal
mtDNA has been attributed, at least in part, to faulty
replication enzymes and a general absence of DNA
repair processes (Brown 1983).
The explanations we have offered for the high
rate of rearrangements and low rate of substitutions
in plant m t D N A are but prima facie biochemical
ones and do not address underlying selective factors.
What at present is completely mystifying is why
plant mtDNAs can tolerate lots of extra sequences,

including r e c o m b i n o g e n i c repeats, whereas animal
mtDNAs cannot. Similarly, why can animal mt-
D N A s tolerate m u c h higher substitution rates than
p l a n t m t D N A s , w h e n b o t h e n c o d e s i m i l a r sets o f
genes?
Acknowledgments. We thank J. Nugent for assistance with the
mtDNA isolations, C. Shields for assistance with the initial filter
hybridizations, and W. Brown, T. Bruns, P. Calie, C. Makaroff,
C. Moritz, and M. Zolan for critical reading of the manuscript.
This research was funded by NIH grant GM-35087 to J.D. Palm-
er.
References
Bailey-Serres J, Hanson DK, Fox TD, Leaver CJ (1986) Mi-
tochondrial genome rearrangement leads to extension and
relocation of the cytochrome c oxidase subunit I gene in Sor-
ghum. Cell 47:567-576
Bland MM, Matzinger DF, Levings CS III (1985) Comparison
of the mitochondrial genome of Nicotiana tabacum with its
progenitor species. Theor Appl Genet 69:535-541
Bland MM, Levings CS III, Matzinger DF (1986) The tobacco
mitochondrial ATPase subunit 9 gene is closely linked to an
open reading frame for a ribosomal protein. Mol Gen Genet
204:8-16
Broach JR (1982) The yeast plasmid 2 micron circle. Cell 28:
203-204
Brown WM (1983) Evolution of animal mitochondrial DNA.
In: Nei M, Koehn RK (eds) Evolution of genes and proteins.
Sinauer, Sunderland MA, pp 62-88
Brown WM (1985) The mitochondrial genome of animals. In:
MacIntyre ILl (ed) Monographs in evolutionary biology: mo-
lecular evolutionary genetics. Plenum, New York, pp 95-130
Brown WM, George M Jr, Wilson AC (1979) Rapid evolution
of mitochondrial DNA. Proc Nail Acad Sci USA 76:I967-
1971
Brown WM, Prager EM, Wang A, Wilson AC (1982) Mito-
chondrial DNA sequences of primates: tempo and mode of
evolution. J Mol Evol 18:225-239
Chetrit P, Mathieu C, Muller JP, Vedel F (1984) Physical and
gene mapping of cauliflower (Brassica oleracea) mitochon-
drial DNA. Curr Genet 8:413-421
ClaytonDA (1984) Transcriptionoftheanimalmitochondrial
genome. Annu Rev Biochem 53:573-594
Dewey RE, Levings CS III, Timothy DH (1986) Novel recom-
binations in the maize mitochondrial genome produce a unique
transcriptional unit in the Texas male-sterile cytoplasm. Cell
44:439-449
Gray J (1986) Wonders of chloroplast DNA. Nature 322:501-
502
Kolodner R, Tewari KK (1972) Physicochemical characteriza-
tion of mitochondrial DNA from pea leaves. Proc Natl Acad
Sei USA 69:1830-1834
Lebacq P, Vedel F (1981) Sat I restriction enzyme analysis of
chloroplast and mitochondrial DNAs in the genus Brassica.
Plant Sci Lett 23:1-9
Lonsdale DM (1988) The plant mitochondrial genome. In: Da-
vies DD (ed) The biochemistry of plants, vol. II, biochemistry
of metabolism. Academic Press, Orlando FL (in press)
MakaroffCA, Palmer JD (1987) Extensive mitochondrial-spe-
cific tranrcription of the Brassica campestris mitochondrial
genome. Nucleic Acids Res 15:5141-5156
Makaroff CA, Palmer JD (1988) Mitochondrial DNA rear-
rangements and transcriptional alterations in the male-sterile
cytoplasm of Ogura radish. Mol Cell Biol 8:1474-1480
97
McClean PE, Hanson MR (1986) Mitochondrial DNA se-
quence divergence among Lycopersicon and related Solanum
species. Genetics 112:649-667
Moritz C, Brown WM (1987) Tandem duplications in animal
mitochondrial DNAs: variation in incidence and gene content
among lizards. Proc Nati Aead Sci USA 84:7183-7187
PalmerJD (1985a) Evolutionofchloroplastandmitochondrial
DNA in plants and algae. In: MacIntyre ILl (ed) Monographs
in evolutionary biology: molecular evolutionary genetics. Ple-
num, New York, pp 131-240
Palmer JD (1985b) Comparative organization of chloroplast
genomes. Annu Rev Genet 19:325-354
Palmer JD (1986) Isolation and structural analysis of chloro-
plast DNA. Methods Enzymol 118:167-186
Palmer JD (1988) Intraspecific variation and multicircularity
in Brassica mitochondrial DNAs. Genetics I 18:341-351.
Palmer JD, Herbon LA (1986) Tripartite mitochondrial ge-
homes of Brassica and Raphanus: reversal of repeat config-
urations by inversion. Nucleic Acids Res 14:9755-9765
Palmer JD, Herbon LA (1987) Unicircular structure of the
Brassica hirta mitochondrial genome. Curr Genet 11:565-
570
Palmer JD, Shields CR (1984) Tripartite structure of the Bras-
sica campestris mitochondrial genome. Nature 307:437-440
Palmer JD, Shields CR, Cohen DB, Orlon TJ (1983a) Chlo-
roplast DNA evolution and the origin of amphidiploid Bras-
sica species. Theor Appl Genet 65:18 l-189
Palmer JD, Shields CR, Cohen DB, Orton TJ (1983b) An un-
usual mitochondrial DNA plasmid in the genus Brassica. Na-
ture 301:725-728
Pring DR, Lonsdale DM (1985) Molecular biology of higher
plant mitochondrial DNA. Int Rev Cytol 97:1-46
Schardl CL, Pring DR, Lonsdale DM (1985) Mitochondrial
DNA rearrangements associated with fertile revertants of
S-type male-sterile maize. Cell 43:361-368
SederoffRR (1987) Molecular mechanisms of mitochondrial-
genome evolution in higher plants. Am Nat 130:$30-$45
SederoffRR, Levings CS III, Timothy DH, Hu WWL (1981)
Evolution of DNA sequence organization in mitochondrial
genomes of Zea. Proc Natl Acad Sci USA 78:5953-5957
Siculella L, Palmer JC (1988) Physical and gene organization
of mitochondrial DNA in fertile and male sterile sunflower:
CMS-associated alterations in structure and transcription of
the atpA gene. Nucleic Acids Res 16:3787-3799
SternDB, PalmerJD (1984) Recombination sequences in plant
mitochondrial genomes: diversity and homologies to known
mitochondrial genes. Nucleic Acids Res 12:6141-6 157
Ward BL, Anderson RS, Bendich AJ (I 981 ) The mitoehondrial
genome is large and variable in a family of plants (Cucurbi-
taceae). Cell 25:793-803
Wilson AC, Cann RL, Carr SM, George M, Gyllensten UB, Helm-
Byehowski KM, Higuchi RG, Palumbi SR, Prager EM, Sage
RD, Stoneking M (1985) Mitochondrial DNA and two per-
spectives on evolutionary genetics. Biol J Linn Soe 26:375-
400
Wolfe KH, Li W-H, Sharp PM (1987) Rates ofnucleotide sub-
stitution vary greatly among plant mitochondrial, chloroplast,
and nuclear DNAs. Proc Natl Acad Sci USA 84:9054-9058
Young EG, Hanson MR (1987) A fused mitoehondrial gene
associated with cytoplasmic male sterility is developmentally
regulated. Cell 50:41-49
Zurawski G, Clegg MT (1987) Evolution of higher-plant chlo-
roplast DNA-encoded genes: implications for structure-func-
tion and phylogenetic studies. Annu Rev Plant Physiol 38:
391-418
Zurawski G, Clegg MT, Brown AHD (1984) The nature of
nucleotide sequence divergence between barley and maize
chloroplast DNA. Genetics 106:735-749
Received March 2, 1988/Revised and accepted April 7, 1988