Copyright 2008 by the Genetics Society of America

DOI: 10.1534/genetics.107.079624

Mitochondrial DNA Transfer to the Nucleus Generates Extensive Insertion

Site Variation in Maize
Ashley N. Lough,*,1 Leah M. Roark,*,1 Akio Kato, Thomas S. Ream, Jonathan C. Lamb,
James A. Birchler* and Kathleen J. Newton*,2
*Division of Biological Sciences, University of Missouri, Columbia, Missouri 65211, Faculty of Agriculture, Kyoto Prefectural University,
Sakyo-ku, Kyoto-shi, Kyoto-fu, 606-0823, Japan, Biology Department, Washington University, St. Louis, Missouri 63130
and Department of Biochemistry and Biophysics, Texas A&M University, College Station, Texas 77843
Manuscript received July 27, 2007
Accepted for publication October 23, 2007
ABSTRACT
Mitochondrial DNA (mtDNA) insertions into nuclear chromosomes have been documented in a number of eukaryotes. We used fluorescence in situ hybridization (FISH) to examine the variation of mtDNA
insertions in maize. Twenty overlapping cosmids, representing the 570-kb maize mitochondrial genome,
were individually labeled and hybridized to root tip metaphase chromosomes from the B73 inbred line. A
minimum of 15 mtDNA insertion sites on nine chromosomes were detectable using this method. One site
near the centromere on chromosome arm 9L was identified by a majority of the cosmids. To examine
variation in nuclear mitochondrial DNA sequences (NUMTs), a mixture of labeled cosmids was applied to
chromosome spreads of ten diverse inbred lines: A188, A632, B37, B73, BMS, KYS, Mo17, Oh43, W22, and
W23. The number of detectable NUMTs varied dramatically among the lines. None of the tested inbred
lines other than B73 showed the strong hybridization signal on 9L, suggesting that there is a recent
mtDNA insertion at this site in B73. Different sources of B73 and W23 were examined for NUMT variation
within inbred lines. Differences were detectable, suggesting either that mtDNA is being incorporated or
lost from the maize nuclear genome continuously. The results indicate that mtDNA insertions represent a
major source of nuclear chromosomal variation.

ITOCHONDRIAL DNA (mtDNA) is thought to

have originated from a bacterial ancestor with
the transfer of most genes to the nucleus in a process
that continues (Richly and Leister 2004b; Timmis
et al. 2004). Within the sequenced nuclear genomes of
Arabidopsis (Arabidopsis thaliana) and rice (Oryza sativa),
many mtDNA and chloroplast DNA (cpDNA) transfers
of varying sizes have been identified. The mtDNA fragments, called NUMT (nuclear mtDNA), were identified
by sequence similarity to the mitochondrial genomes,
and some of the insertions have been analyzed by fluorescence in situ hybridization (FISH) techniques. The
largest mitochondrial insertion described to date in flowering plants is 620 kb in length and located near the
centromere on the short arm of Arabidopsis chromosome 2 (Stupar et al. 2001). It was first identified by Lin
et al. (1999) in the Arabidopsis sequencing project, and
Stupar et al. (2001) using fiber-FISH to determine the
NUMTs organization. Several other smaller NUMTs also
exist within the Arabidopsis nuclear genome (Arabidopsis
Genome Initiative 2000). Sequence comparisons of the
rice nuclear genome have shown that between 0.180.19%
1

These authors contributed equally to this work.

Corresponding author: Division of Biological Sciences, 324 Tucker Hall,
University of Missouri, Columbia, MO 65211-7400.
E-mail: newtonk@missouri.edu
2

Genetics 178: 4755 ( January 2008)

of the genome is composed of mtDNA (International

Rice Genome Sequencing Project 2005). Rice chromosome 10 alone includes 57 NUMTs, varying from 80
to 2552 bp distributed across the chromosome (Rice
Chromosome 10 Sequencing Consortium 2003).
The majority of the previous work has been performed
within one line or ecotype. Only a few studies have examined variation of NUMTs or nuclear cpDNA transfers
(NUPTs) among lines or ecotypes of the same species
and these have focused on specific insertions. It was
reported that a 3.9-kb mtDNA insertion in a polyubiquitin gene of the Arabidopsis ecotype Columbia (Sun
and Callis 1993) did not exist in the Be-0, Ler, No-0,
RLD-0, or WS-0 ecotypes. Part of this insertion was found
in the Arabidopsis ecotypes Eifel, Enkheim, and Hilversum
nuclear genomes by Ullrich et al. (1997). In addition, a
131-kb NUPT on chromosome 10 of O. sativa subsp.
japonica was not identified in the O. sativa subsp. indica
or O. rufipogon nuclear genomes (Huang et al. 2005).
No systematic survey of NUMT variation within a species has been reported. The use of fluorescently labeled
mtDNA segments as hybridization probes to chromosomes provides a tool to assay NUMT variation. Maize is
a good system for such studies because it has a wellcharacterized karyotype and many inbred lines of known
pedigree. In this study, significant NUMT variation between and within inbred lines was found. Our results

48

A. N. Lough et al.
TABLE 1
Karyotyping markers included in the eight-probe mix

DNA sequences
Knob 180 bp
4-12-1
NOR-173
Cent C
2-3-3
Cent 4
1-26-2
2-3-3

Fluorescent molecules

Individual cosmid use final

concentration (ng/ml)

19-cosmid mix use final

concentration (ng/ml)

Coumarin or Cascade Blue

FITC or Alexa Fluor 488
FITC or Alexa Fluor 488
FITC or Alexa Fluor 488
FITC or Alexa Fluor 488
Cy5
Cy5
Cy5

12.5
12.5
0.0625
0.5
3.125
12.5
6.25
6.25

15.1
15.1
0.08
0.6
3.8
15.1
7.5
7.5

suggest that large portions of the maize mitochondrial

genome have been recently inserted into the nucleus.

MATERIALS AND METHODS

Cosmid DNA: A set of 20 cosmids provided by C. Fauron
(Fauron and Havlik 1988) that sequentially cover the entire
NB mitochondrial genome (Clifton et al. 2004) was used as
mtDNA probes, either individually or in a 19-cosmid mix. Chloramphenicol was added to bacterial cultures that had been
grown overnight, and the cosmids were allowed to amplify for
an additional 6 hr. Cosmid DNA was isolated from each culture
using the QIAGEN (Valencia, CA) Plasmid Maxi-prep protocol.
Karyotyping probes: A standardized cocktail of eight different repeated DNA sequences (Table 1) was used for karyotyping (Kato et al. 2004). The probes used were labeled by
nick translation with coumarin-5-dUTP or Cascade Blue-7dUTP, fluorescein isothiocyanate-12-dUTP (FITC) or Alexa
Fluor 488-5-dUTP and Cyanine 5-dUTP (Cy5).
Labeling of cosmid and karyotyping probes: The mtDNAcontaining cosmids were labeled with Texas red-5-dCTP. The
optimized nick translation protocol of Kato et al. (2006) was
used with the following specifications. For 5 mg DNA (25 ml,
200 ng/ml), the nick translation mix included 1 ml 100
milliunits/ml DNase (2 ml for 19-cosmid mix), 20 ml 10
units/ml DNA polymerase I (Texas red, Cy5), 5 ml 103 nick
translation buffer, 5 ml nonlabeled dAGC mix (2 mm,
karyotyping probes) or dATG mix (2 mm, mtDNA probes),
and 1.25 ml labeled 1 mm dUTP (karyotyping probes) or dCTP
(mtDNA probes). After labeling, each of the Texas red- and
Cy5-labeled probes were purified on Bio-gel P-60 columns
(Kato et al. 2006). After discarding two washes (50 and 350
ml), eluates from 23 additional 350 ml 13 TE washes (yielding
25 mg DNA) were collected. The labeled DNA was ethanol
precipitated, resuspended in 23 SSC/13 TE (TE, pH 8.0),
and stored at 20. Individual mtDNA-containing cosmid
probes were resuspended at a final concentration of 100 ng/
ml in 23 SSC/13 TE. The 19-cosmid mix probe was at a final
concentration of 400 ng/ml. For the nick translation reaction
with other fluorochromes (coumarin, Cascade Blue, FITC, and
Alexa Fluor 488) 6.25 ml DNA polymerase I (10 units/ml) was
used in each reaction. These probes were not column purified
but were ethanol precipitated immediately, dissolved in the
23 SSC/13 TE buffer, and stored at 20.
Preparation of root tip chromosome spreads: The protocol
for the preparation of root tip mitotic metaphase chromosome spreads was performed as described (Kato et al. 2006)
with the following modifications. Root tips were not incubated
on ice during enzymatic digestion. After dispersing the cells in

70% ethanol, the cells were pelleted by centrifuging for 1020

sec instead of 2 min. After dropping the cell suspension on
slides, the slides were UV crosslinked (120-5 mJ/cm2) and
fixed with 10% formaldehyde for 10 min, washed with 100%
ethanol, and allowed to dry. Additional modifications to the
Kato et al. (2006) protocol were used for the comparisons
among the B73 (B), B73 (C), W23-B, and W23 A 3 B hybrid
lines. The root tips were washed in 100% ethanol three times
(without a 70% ethanol treatment) and then cells were dispersed by agitation in acetic acid. These samples were not subjected to formaldehyde fixation.
Denaturation protocol: The procedure for the denaturation
of probe and chromosomal DNA separately was the same as
described in Kato et al. (2004) with minor modifications.
Briefly, after formaldehyde treatment of the slides, 5 ml of 23
SSC/13 TE/salmon sperm DNA (1 mg/ml) was dropped onto
the center of the cells and a plastic coverslip applied. The
probe was prepared by mixing the various components in a
microcentrifuge tube to a final volume of 5 ml. When the 19cosmid mix was used, the probe for each slide consisted of
2.5 ml 19-cosmid mix, 1.7 ml karyotyping mix listed in Table 1, and
0.8 ml 23 SSC/13 TE. For individual cosmids, the probe for
each slide consisted of 1 ml of the individually labeled cosmid
and 4 ml of karyotyping mix (with 23 SSC/13 TE). Each slide
was incubated at 55 for 316 hr in a humid chamber. After
hybridization, the slide was washed with 23 SSC at room temperature for 5 min, followed by a wash with 23 SSC at 55 for
20 min in Coplin jars. The slide was mounted with Vectashield
(Vector Laboratories, Burlingame, CA) containing 49,6-diamidino2-phenylindole (DAPI). The DAPI was diluted to 1/20 strength
using Vectashield. Vectashield with no DAPI was used with
Cascade Blue probes because DAPI interferes with the probe
signal. Slides were stored at 20.
Image capture and data processing of FISH images: The
protocol for capturing and processing FISH images was as
previously described (Kato et al. 2004, 2006) with modifications.
Color assignments were as follows: blue (coumarin or Cascade
Blue), green (FITC or Alexa Fluor 488), red (Cy5), and white
(Texas red). Using Adobe Photoshop, background was reduced by subtracting an image from an empty area of the slide
taken at approximately the same exposures for each channel.
RESULTS

Analysis of NUMTs using individual mtDNA-containing

cosmids: The maize NB mitochondrial genome has been
sequenced from a normal male-fertile B37 line and consists of 569,630 bp (Clifton et al. 2004). A set of numbered cosmid clones (Fauron and Havlik 1988) covering

NUMT Variation in Maize

49

Figure 1.MtDNA segments hybridized to B73 chromosomes. (a) Linear map of the NB maize mitochondrial genome with
cosmid locations. A linearized version of the 569,630 bp NB maize mitochondrial genome is shown (Clifton et al. 2004; condensed from Allen et al. 2007) with the cosmid positions outlined. Mitochondrial genes, rRNAs, and tRNAs (half-length lines)
are represented by vertical lines. Repeats are indicated by colored arrows. The cpDNA insertions are indicated by green rectangles.
A scale in kilobases is shown along the bottom. (b) Individual mtDNA-containing cosmids hybridized to B73 chromosomes. Individual cosmid probes containing mtDNA were labeled with Texas red and hybridized to B73 chromosomes. The white arrowheads mark mtDNA insertion sites and the green arrowheads mark potential cpDNA insertion sites. The eight mix of karyotyping
probes (Kato et al. 2004) was used to identify each chromosome. Only the layer with the Texas red-labeled mitochondrial probes is
shown. The marked mtDNA insertion sites were identified on .88% of individual chromosomes examined.

the whole NB genome on 20 overlapping mtDNA fragments (averaging 37 kb each, Figure 1a) was used to
make FISH probes. It should be noted that there are
two large insertions of cpDNA in the NB mitochondrial
genome, as well as multiple, smaller insertions (Stern
and Lonsdale 1982; Lonsdale et al. 1983; Clifton et al.
2004). Cosmid 13 (39.5 kb) contains the two largest
chloroplast DNA insertions (Figure 1a) of 12.6 kb and
4.1 kb (comprising 34% of the cosmid). Thus, we expected the mtDNA from cosmid 13 to identify some of
the cpDNA as well as mtDNA insertions in nuclear
chromosomes.
The DNA from each of the 20 cosmids was individually labeled with Texas red and hybridized to B73 mitotic metaphase chromosomes to locate NUMTs (Figure
1b). The B73 inbred line was chosen for this analysis
because its nuclear genome is currently being sequenced
(Bennetzen et al. 2001) and it has the NB mitochondrial
genotype.
Individual mtDNA insertion sites were determined by
examining multiple chromosome preparations (at least
11; see supplemental Table 1 at http://www.genetics.
org/supplemental/). Due to hybridization of the mtDNA-

containing cosmids with organelles present on the

chromosome spreads and, less significantly, to chromosomal twisting, it is sometimes difficult to identify a specific NUMT in every case. The mtDNA insertions we have
indicated in Figure 1b were consistently present. The
minimum target size documented using FISH on maize
chromosomes is 3 kb (Yu et al. 2007). Thus, the insertions detected in this study are larger than this lower
limit.
The mtDNA contained in the individual cosmids
identified a total of 58 insertion sites within the B73
chromosomes (Figure 1b). When all of the NUMTs were
compared, it was determined that there was a minimum
of 21 unique mtDNA insertion sites. The NUMTs were
arbitrarily classified by their position on the chromosome: proximal (near centromeres), distal (near telomeres), and interstitial (between the centromere and a
telomere). Of the 58 total sites, 32 (55.2%) were proximal and 10 (17.2%) were distal. Most of the cosmids (14
of the 20) identified a specific site located proximally on
the long arm of chromosome 9. This insertion apparently includes a large region of mtDNA that is colinear
in the mitochondrial genome (contained in cosmids 20,

50

A. N. Lough et al.

and 110) and which could be present on a single

mitochondrial subgenome. However, our experiments
cannot distinguish whether the mtDNA organization is
preserved in the NUMT.
After comparing the B73 NUMTs to the mitochondrial genome map (Figure 1a) (Clifton et al. 2004), no
preference was seen for insertions of gene-rich vs. genepoor regions of mtDNA. No visible NUMTs were identified using cosmid 11, which contains both gene-rich
and gene-poor regions (Figure 1b). The mtDNA contained in cosmid 11 is 40 kb, of which 9.7 kb does not
overlap cosmids 10 and 12 (Figure 1a). The region of
overlap with cosmid 10 is relatively gene rich, containing
nad5 exons 35, rps12, nad3, nad1 exon 5, mat-r, and 283
bp of rps1. Detectable NUMTs did include regions of
gene-poor mtDNA. For example, cosmid 8 includes 38
kb of mtDNA, with 10.7 kb not overlapping cosmids
7 and 9 (Figure 1a). It contains only two known genes:
ccmC (723 bp) and nad5 exons 12 (2316 bp) (Clifton
et al. 2004). This cosmid detects two mtDNA insertions,
located proximally on 2L and 9L (Figure 1b). Cosmid 15
contains 38 kb mtDNA, with 32.3 kb not overlapping
cosmids 14 and 16. It includes only one gene: rrn26
(3552 bp) (Clifton et al. 2004). The mtDNA within
cosmid 15 detects three NUMTsin 4L proximal, 5L
interstitial, and 9S distal. Collectively, these data suggest
that gene content does not affect which regions of mtDNA
are inserted into the nuclear genome.
Cosmid 13 identified more apparent NUMTs in the
B73 nuclear genome than any other cosmid (Figure 1b).
However, because the mtDNA contained within this
cosmid includes large segments of integrated cpDNA,
we attempted to infer which insertion sites could be due
to cpDNA. We compared the sites identified with cosmid
13 to those detected using the other cosmids. The two
NUMTs on chromosome 4 and one on chromosome 3
were identified by other cosmids, indicating that they
were likely mtDNA insertion sites. The insertion on chromosome 2L matches one identified only by the overlapping cosmid 14, which contains 4.1 kb of cpDNA,
indicating that this site is potentially a NUPT. A second
site identified by cosmid 14 on 4L does appear to be a
NUMT because the same site was detected by other
cosmids. The remaining seven sites identified by cosmid
13 are potentially cpDNA insertions because no other
mtDNA cosmids hybridize to those locations. However,
our data set does not allow us to determine conclusively
whether the insertions originated from mitochondria or
plastid DNA.
If we exclude the sites identified by hybridization with
cosmid 13 that potentially represent cpDNA insertions,
a total of 51 different mtDNA insertion sites were detected in our FISH analyses. Of the 51 mtDNA insertion
sites, 29 (60.4%) were classified as proximal and 9 (18.8%)
as distal. When the relative positions of all sites identified by all of the cosmids were compared, a minimum
of 15 unique NUMT locations were found to exist in

B73. Excluding cosmid 13 from the analyses, NUMTs

were not detected on chromosomes 1, 8, or 10 in B73.
Analysis of NUMTs in 10 maize inbred lines: To
assess variability for NUMTs in maize, mtDNA insertions
were identified in a set of 10 maize inbred lines using the
19-cosmid mix probe labeled with the most sensitive
fluorochrome in our system, Texas red (Figure 2). Cosmid
13 was excluded from the cosmid mix because, as noted
above, it carries a region of mtDNA that includes large
insertions of cpDNA. The karyotypes of the 10 inbred
lines had been previously characterized using FISH with
a cocktail of repetitive DNA probes (Kato et al. 2004). A
karyotyping cocktail (see materials and methods) was
included in the hybridization with the 19-cosmid mix
in order to identify the chromosomes of the different
inbred lines.
The numbers and locations of the mtDNA hybridization sites on chromosomes varied greatly among the
maize inbred lines (Figure 2; see supplemental Table 2
at http://www.genetics.org/supplemental/ for the numbers of chromosomes examined). The number of detected NUMTs ranged from 9 in A188 and A632 to 19 in
Oh43. Several of these NUMTs were observed at proximal and distal locations on chromosomes (Table 2).
When all the insertions are compared (Figure 2), only
one on 2S proximal appeared to be shared among all 10
lines examined. However, due to the limitations inherent
in analyzing highly condensed metaphase chromosomes,
it is possible that the 2S proximal signals in different
lines are at different positions.
The B73 mtDNA insertion sites identified using the
19-cosmid mix (Figure 2) were compared to those detected by the individual mtDNA-containing cosmid probes
(Figure 1b). Using the 19-cosmid mix, 13 distinct mtDNA
insertion sites were detected in B73. There were two
fewer NUMTs identified using the 19-cosmid mix than
with the individual mtDNA-containing cosmid probes;
they were located on 7S (using cosmid 1) and on 5L (using cosmid 6). This difference in detection ability might
be due to the fact that the individual cosmid mtDNA
probes contained a larger amount of mtDNA per cosmid
than the 19-cosmid mix. The same concentration of
mtDNA per cosmid could not be used in the 19-cosmid
mix because at high concentrations (.400 ng/ml), the
labeled probe DNA becomes very difficult to dissolve.
The final concentration of each individual cosmid in
the 19-cosmid mix probe cocktail was approximately
half the concentration used when individual cosmids
were applied.
One additional B73 mtDNA insertion site was identified on 1S using the 19-cosmid mix. It was not observed
using individual mtDNA-containing cosmid probes. We
hypothesize that this location contains portions of mtDNA
from across the mitochondrial genome (contained in
different cosmids) that were connected through random
end-joining prior to insertion into the nuclear genome
(Noutsos et al. 2005). These portions would be too small

NUMT Variation in Maize

51

Figure 2.MtDNA insertion sites in the chromosomes of 10 maize inbred lines. The karyotypes of 10 different maize inbred
lines are shown. Sites of mtDNA hybridization on the chromosomes, probed with the 19-cosmid mix of mtDNA, are indicated by
arrowheads. The eight mix of karyotyping probes was used to identify the chromosomes. The marked mtDNA insertion sites were
identified on .88% of the chromosomes examined.

to detect using only one cosmid probe, but the 19cosmid mix would hybridize simultaneously to all of the
small sequences, potentially allowing detection of the
insertion site.
A detailed comparison was performed between B73
and three other inbred lines. The first was Oh43 because this line contained the largest number of mtDNA
insertion sites. Five NUMTs detected in B73 were not
TABLE 2
Numbers and relative positions of NUMTs in
four maize inbred lines

No. of mtDNA
insertion sites
No. (%) proximal
No. (%) distal

B73

Oh43

A188

B37

13

19

14

5 (38.5)
4 (30.8)

5 (26.3)
3 (15.8)

3 (33.3)
3 (33.3)

4 (28.6)
3 (21.4)

observed in Oh43 (Figure 2). Oh43 contained at least

eight NUMTs that were not observed in B73.
The B73 NUMTs were also compared to those from
the A188 inbred line. B73 and A188 differ in their mitochondrial genotypes; A188 carries the rarer NA mitochondrial genotype, of which 5.11% is not represented
in the NB mitochondrial genome (Fauron and Casper
1994; Clifton et al. 2004; Allen et al. 2007). A minimum
of nine mtDNA insertion sites were detected in B73 that
were not identified in A188 (Figure 2). A188 contained
five NUMTs that were not detected in B73.
B73 and B37 are the most closely related inbred lines
in our study. These two lines were both developed from
the Iowa Stiff Stalk Synthetic (BSSS) lines. B37 was selected 14 years before B73 (Troyer 1999). Five NUMTs
were detected in B73 that were not observed in B37.
Fourteen mtDNA insertion sites were identified in B37,
with at least one on every chromosome, while no NUMTs
were observed on B73 chromosomes 8 and 10 (Figure 2).

52

A. N. Lough et al.

Figure 3.Different B73 sources

probed with the 19-cosmid mix. The
karyotypes of B73 from three different
sources are shown. Arrowheads indicate
mtDNA insertion sites. White arrowheads
indicate insertion sites seen in .88% of
the chromosomes observed. Open arrowheads indicate insertion sites seen in
6087% of the chromosomes observed.
Chromosomes were identified as described in Figure 2.

B37 contained six NUMTs that were not observed in

B73.
The strong hybridization signal detected in the B73
9L proximal region (Figure 1b, Figure 2) was seen to be
much weaker in B37 (Figure 2). Indeed, B73 was the only
line that had such a strong signal from the 19-cosmid
mix in this chromosomal region; a lesser signal was observed at this position in seven other lines. KYS had no
apparent NUMT at this location. In A188, a NUMT was
observed that appeared to be coincident with the centromere of 9. These comparisons suggest that the 9L
proximal insertion, containing most of the mitochondrial genome, is not common in maize inbreds and thus
is likely a recent event in the B73 lineage.
Analysis of NUMTs within the B73 inbred line: After
observing the extensive diversity of NUMTs among inbred lines we were interested in assaying variation within
an inbred line (Figure 3). B73 is a commonly used inbred that is maintained by self and sib pollinations in
many laboratories. We obtained the B73 inbred from
three different laboratories and have arbitrarily designated the different sources A, B, and C. B73 (A) is the
source used for our previous experiments, B73 (B) is
from G. Davis (University of Missouri), and B73 (C) is from
S. Gabay-Laughnan (University of Illinois). All mtDNA
insertion sites indicated previously had been observed
in at least 88% of the chromosomes examined. For the
purpose of comparing closely related materials, we also
included sites that were observed less frequently (6379%
of the examined chromosomes; see supplemental Table
3 at http://www.genetics.org/supplemental/).
Using the 19-cosmid mix, B73 (A) had 13 detectable
NUMTs in at least 88% of the chromosomes examined,
and one other site was observed less frequently (on 77%
of the chromosomes examined; Table 3 and supplemental Table 3). The insertions were the same as those
seen in the earlier analyses (Figure 2). B73 (B) had 13
NUMTs detected in at least 88% of chromosomes examined and 3 less frequently detected sites (Table 3; supplemental Table 3). B73 (C) was more distinctive; 16

NUMTs were observed in at least 88% of chromosomes

examined and 2 sites were observed less frequently
(Table 3). In particular, a NUMT was observed in the
distal region of chromosome 5 that was not observed
in B73 (A) and (B). This site could represent a new insertion of mitochondrial sequences that has occurred
over the course of a few generations or it could result
from the segregation of a preexisting polymorphism.
Most of the mtDNA insertion sites were shared by
the various B73 sources (Figure 3). A number of these
NUMTs were identified at proximal and distal chromosomal locations (Table 3). In addition, all the B73
sources had equally strong signals on proximal 9L. This
relative consistency of detectable NUMTs among different
sources of an inbred indicates that the NUMT composition is characteristic for the karyotype of each inbred
line. However, the differences within an inbred line illustrate the dynamic nature of NUMT insertions.
A sampling of individual cosmid probes was used to
confirm that reproducible differences exist among the
mtDNA insertion sites on the chromosomes from different sources of B73 (Figure 4; number of chromosomes examined are given in supplemental Table 4 at
http://www.genetics.org/supplemental/).
Analysis of NUMTs within the W23 inbred line: The
differences observed among the sources of B73 suggest
that mtDNA insertions into the nuclear genome of maize
can occur frequently. To investigate this phenomenon
TABLE 3
Numbers of mtDNA insertion sites in three sources of B73

Sites frequently observed

(.88%)
Sites less frequently
observed (6087%)
Total no. (%) proximal
Total no. (%) distal

B73 A

B73 B

B73 C

13

13

16

5 (35.7)
4 (28.6)

5 (31.3)
4 (25.0)

5 (27.8)
5 (27.8)

NUMT Variation in Maize

53
Figure 4.Different B73 sources (A, B, and C) probed with
mtDNA segments. Six individual
mtDNA-containing cosmids labeled
with Texas red were hybridized to
B73 chromosomes of different
sources. The white arrowheads
mark mtDNA insertion sites and
the green arrowheads mark potential cpDNA insertion sites. The
eight mix of karyotyping probes
was used to identify each chromosome. Only the layer with the Texas
red-labeled mitochondrial probes
is shown. The marked mtDNA insertion sites were identified on
.88% of individual chromosomes
examined. Chromosome 8 contained no mtDNA insertion sites
and was not included.

further, the nuclear chromosomes of two W23 sources

were examined for variation in mtDNA insertion sites
(Figure 5). Both W23 lineages were originally from the
Wisconsin Breeding Program. W23-A (see also Figure 2)
was obtained from E. Coe (University of Missouri) and
W23-B was obtained from J. Kermicle (University of
Wisconsin). Seedlings from each source were characterized for NUMTs and an F1 hybrid of the A and B
sources was generated and analyzed. These karyotypes
included mtDNA sites that were observed frequently (at
least 88% of chromosomes examined) and those that
were more difficult to detect (6383% of chromosomes
examined; supplemental Table 5 at http://www.genetics.
org/supplemental/).
For each source and the hybrid between the two, the
numbers and locations of mtDNA insertion sites were
examined and compared (Figure 5). W23-A plants had
15 mtDNA insertion sites observed in at least 88% of
chromosomes examined and 2 less frequently observed
NUMTs (Table 4). W23-B plants had 18 mtDNA in-

sertion sites detected in at least 88% of chromosomes

observed and 1 less frequently detected NUMT (Table
4). A majority of the NUMTs within these sources were
commonly located proximally and distally.
W23-B had NUMTs located on 1S, 3L, and 6L that
were not detected in W23-A. W23-A had only one NUMT
(4S) that was not detected in W23-B. The changes occurred in both directions; each source contained at
least one mtDNA insertion site that the other source was
missing. To confirm the validity of these differences,
NUMTs were characterized in the hybrid between the
two lines. When a signal was present on a chromosome
in one parent and absent in the other, both homologs
could be clearly distinguished in the F1 hybrid (Figure 5).
Detecting the differences in the hybrid under identical
conditions of tissue fixation and chromosome image
capture confirms that these distinctions occur reproducibly. The difference for the mtDNA insertion site on
the short arm of chromosome 1 was confirmed with a
sampling of individual cosmid probes (Figure 6; numbers

Figure 5.Different W23 sources probed with the 19-cosmid mix. The karyotypes of both W23 sources and their F1 hybrid are
included. The top two karyotypes of the separate sources show both the mtDNA insertion sites and karyotyping probes (color).
The bottom karyotype of the hybrid shows only the mtDNA insertion sites. Arrowheads indicate mtDNA insertion sites and are
labeled as described in Figure 3. The eight mix of karyotyping probes was used to identify each chromosome.

54

A. N. Lough et al.
TABLE 4
Numbers of mtDNA insertion sites in two W23 sources
W23 A

W23 B

Sites frequently observed (.88%)

15
18
Sites less frequently observed (6087%)
2
1
Total no. (%) proximal
7 (41.2) 7 (36.8)
Total no. (%) distal
3 (17.6) 5 (26.3)

of chromosomes examined are given in supplemental

Table 6 at http://www.genetics.org/supplemental/). These
findings indicate that the NUMTconstitution can change
within an inbred line.
DISCUSSION

In the studies reported here, we used FISH to visualize

variation for mtDNA integration sites into nuclear chromosomes among and within maize inbred lines. MtDNA
insertion sites were first examined using a series of
20 individual mtDNA-containing cosmids as hybridization probes onto B73 chromosomes. Discrete, localized
mtDNA signals were detected on all but three of the B73
chromosomes.
A background level of hybridization was also observed
on all the chromosomes. This background suggests that
the mtDNA probes may hybridize to many small NUMTs
throughout the genome that cannot be distinguished as
discrete signals. Previous studies have found that there
are many small NUMTs in the Arabidopsis and rice nuclear genomes; the average NUMT size is 346 bp in Arabidopsis and 206 bp in rice (Richly and Leister 2004a).
Such small or highly degenerate fragments could con-

tribute to the background hybridization, but they would

not be detectable as discrete FISH signals. Using our
method, 3 kb of complete homology is the approximate
lower limit of detection (Yu et al. 2007).
Overall, 19 of the 20 mtDNA-containing cosmids hybridized to the B73 nuclear chromosomes, which suggests
that any section of the mitochondrial genome potentially could be integrated. No insertion preference of
gene-rich or gene-poor regions of the mitochondrial genome was observed. Our conclusion concurs with those
from previous studies that also found no bias for which
organellar sequences were inserted into nuclear genomes
(Lin et al. 1999; Arabidopsis Genome Initiative 2000;
Stupar et al. 2001; Yuan et al. 2002; Timmis et al. 2004;
International Rice Genome Sequencing Project
2005; Matsuo et al. 2005).
Of particular interest is the major mtDNA insertion
site located near the centromere of chromosome 9 in
the maize B73 inbred. This NUMT was shown to include
mtDNA from 14 of the 20 mtDNA-containing cosmid
probes. It could have arisen from an insertion of a postulated circular subgenome, identified by cosmid 20 and
cosmids 110. If continuous, this insertion would include 400 kb of the mapped NB mitochondrial genome.
Additional mtDNA fragments (identified by cosmids
1618) could have inserted in the same region at another time. Long uninterrupted sequences of cpDNA
have been identified in the rice nuclear genome, including 131-kb and 33-kb fragments on chromosome 10
(Noutsos et al. 2005).
In this study we have used FISH to document the
variability of major mtDNA insertions into the nucleus.
When NUMTs were compared for several inbred lines
of maize using FISH analysis, significant variation was

Figure 6.Two W23 sources (A and B) probed

with mtDNA segments. Six individual mtDNAcontaining cosmids labeled with Texas red were
hybridized to W23 chromosomes of two sources.
The white arrowheads mark mtDNA insertion
sites. The eight mix of karyotyping probes was
used to identify each chromosome. Only the layer
with the Texas red-labeled mitochondrial probes
is shown. The marked mtDNA insertion sites were
identified on .88% of individual chromosomes
examined.

NUMT Variation in Maize

observed. A small amount of variation in NUMT locations

could be detected even within maize inbreds, specifically
in the B73 and W23 lines obtained from different laboratories (sources). Such variation could result from an
insertion of a new fragment of mtDNA, from amplification of a preexisting mtDNA insertion, or from losses
and degradation of sequences at a particular site. The
amount of variation we have observed suggests that insertions of mtDNA into the maize nuclear genome are
ongoing and frequent.
The potentially frequent organellar DNA insertions
might also contribute to the spontaneous mutation frequency of maize, especially considering that only insertions of large size can be detected in our assays. Many
additional mtDNA fragments below the detection limit
are likely and would also contribute to the overall insertion frequency. The extensive variation found indicates
that NUMTs constitute a significant contributor to maize
chromosomal diversity.
We thank G. Davis, S. Gabay-Laughnan, E. Coe, and J. Kermicle for
maize lines, and C. Fauron for the mtDNA-containing cosmids. We
thank J. Allen, T. Langewisch, and L. Meyer for helpful comments
throughout the project. This work was funded by the National Science
Foundation grant DBI-0423898. A.L. was supported by a University of
Missouri Life Sciences fellowship.

LITERATURE CITED
Allen, J. O., C. M. Fauron, P. Minx, L. Roark, S. Oddiraju et al.,
2007 Comparisons among two fertile and three male-sterile
mitochondrial genomes of maize. Genetics 177: 11731192.
Arabidopsis Genome Initiative, 2000 Analysis of the genome sequence of the flowering plant Arabidopsis thaliana. Nature 408:
796815.
Bennetzen, J. L., V. L. Chandler and P. Schnable, 2001 National
Science Foundation-sponsored workshop report. Maize genome
sequencing project. Plant Physiol. 127: 15721578.
Clifton, S. W., P. Minx, C. M. Fauron, M. Gibson, J. O. Allen et al.,
2004 Sequence and comparative analysis of the maize NB mitochondrial genome. Plant Physiol. 136: 34863503.
Fauron, C. M., and M. Casper, 1994 A second type of normal maize mitochondrial genome: an evolutionary link. Genetics 137: 875882.
Fauron, C. M., and M. Havlik, 1988 The BamHI, XhoI, SmaI restriction enzyme maps of the normal maize mitochondrial genome
genotype B37. Nucleic Acids Res. 16: 1039510396.
Huang, C. Y., N. Grunheit, N. Ahmadinejad, J. N. Timmis and W.
Martin, 2005 Mutational decay and age of chloroplast and mitochondrial genomes transferred recently to angiosperm nuclear
chromosomes. Plant Physiol. 138: 17231733.
International Rice Genome Sequencing Project, 2005 The
map-based sequence of the rice genome. Nature 436: 793800.

55

Kato, A., J. C. Lamb and J. A. Birchler, 2004 Chromosome painting using repetitive DNA sequences as probes for somatic chromosome identification in maize. Proc. Natl. Acad. Sci. USA 101:
1355413559.
Kato, A., P. S. Albert, J. M. Vega and J. A. Birchler, 2006 Sensitive
fluorescence in situ hybridization signal detection in maize using
directly labeled probes produced by high concentration DNA
polymerase nick translation. Biotech. Histochem. 81: 7178.
Lin, X., S. Kaul, S. Rounsley, T. P. Shea, M. I. Benito et al.,
1999 Sequence and analysis of chromosome 2 of the plant Arabidopsis thaliana. Nature 402: 761768.
Lonsdale, D. M., T. P. Hodge, C. J. Howe and D. B. Stern, 1983 Maize
mitochondrial DNA contains a sequence homologous to the ribulose-1,5-bisphosphate carboxylase large subunit gene of chloroplast
DNA. Cell 34: 10071014.
Matsuo, M., Y. Ito, R. Yamauchi and J. Obokata, 2005 The rice
nuclear genome continuously integrates, shuffles, and eliminates
the chloroplast genome to cause chloroplast-nuclear DNA flux.
Plant Cell 17: 665675.
Noutsos, C., E. Richly and D. Leister, 2005 Generation and evolutionary fate of insertions of organelle DNA in the nuclear genomes of flowering plants. Genome Res. 15: 616628.
Rice Chromosome 10 Sequencing Consortium et al., 2003 Indepth view of structure, activity, and evolution of rice chromosome 10. Science 300: 15661569.
Richly, E., and D. Leister, 2004a NUPTs in sequenced eukaryotes
and their genomic organization in relation to NUMTs. Mol. Biol.
Evol. 21: 19721980.
Richly, E., and D. Leister, 2004b NUMTs in sequenced eukaryotic
genomes. Mol. Biol. Evol. 21: 10811084.
Stern, D. B., and D. M. Lonsdale, 1982 Mitochondrial and chloroplast genomes of maize have a 12-kilobase DNA sequence in common. Nature 299: 698702.
Stupar, R. M., J. W. Lilly, C. D. Town, Z. Cheng, S. Kaul et al.,
2001 Complex mtDNA constitutes an approximate 620-kb insertion on Arabidopsis thaliana chromosome 2: implication of potential sequencing errors caused by large-unit repeats. Proc. Natl.
Acad. Sci. USA 98: 50995103.
Sun, C. W., and J. Callis, 1993 Recent stable insertion of mitochondrial-DNA into an Arabidopsis polyubiquitin gene by nonhomologous recombination. Plant Cell 5: 97107.
Timmis, J. N., M. A. Ayliffe, C. Y. Huang and W. Martin, 2004 Endosymbiotic gene transfer: organelle genomes forge eukaryotic chromosomes. Nat. Rev. Genet. 5: 123135.
Troyer, A. F., 1999 Background of U.S. hybrid corn. Crop Sci. 39:
601626.
Ullrich, H., K. Lattig, A. Brennicke and V. Knoop, 1997 Mitochondrial DNA variations and nuclear RFLPs reflect different
genetic similarities among 23 Arabidopsis thaliana ecotypes. Plant
Mol. Biol. 33: 3745.
Yu, W., J. C. Lamb, F. Han and J. A. Birchler, 2007 Cytological visualization of DNA transposons and their transposition pattern in
somatic cells of maize. Genetics 175: 3139.
Yuan, Q., J. Hill, J. Hsiao, K. Moffat, S. Ouyang et al., 2002 Genome
sequencing of a 239-kb region of rice chromosome 10L reveals a
high frequency of gene duplication and a large chloroplast DNA
insertion. Mol. Genet. Genomics 267: 713720.
Communicating editor: T. P. Brutnell