A Conserved, Mg 2+ -Dependent Exonuclease Degrades Organelle DNA during Arabidopsis Pollen Development C W Ryo Matsushima, a,1 Lay Yin Tang, a,1 Lingang Zhang, a Hiroshi Yamada, a David Twell, b and Wataru Sakamoto a,2 a Institute of Plant Science and Resources, Okayama University, Kurashiki, Okayama 710-0046, Japan b Department of Biology, University of Leicester, Leicester LE1 7RH, United Kingdom In plant cells, mitochondria and plastids contain their own genomes derived from the ancestral bacteria endosymbiont. Despite their limited genetic capacity, these multicopy organelle genomes account for a substantial fraction of total cellular DNA, raising the question of whether organelle DNA quantity is controlled spatially or temporally. In this study, we genetically dissected the organelle DNA decrease in pollen, a phenomenon that appears to be common in most angiosperm species. By staining mature pollen grains with fluorescent DNA dye, we screened Arabidopsis thaliana for mutants in which extrachromosomal DNAs had accumulated. Such a recessive mutant, termed defective in pollen organelle DNA degrada- tion1 (dpd1), showing elevated levels of DNAs in both plastids and mitochondria, was isolated and characterized. DPD1 encodes a protein belonging to the exonuclease family, whose homologs appear to be found in angiosperms. Indeed, DPD1 has Mg 2+ -dependent exonuclease activity when expressed as a fusion protein and when assayed in vitro and is highly active in developing pollen. Consistent with the dpd phenotype, DPD1 is dual-targeted to plastids and mitochondria. Therefore, we provide evidence of active organelle DNA degradation in the angiosperm male gametophyte, primarily independent of maternal inheritance; the biological function of organellar DNA degradation in pollen is currently unclear. INTRODUCTION Mitochondria and plastids originate from the endosymbiosis of rickettsia-like a-proteobacteria and cyanobacteria-like photo- synthetic bacteria, respectively (Gray et al., 1999; Dyall et al., 2004; Keeling, 2010). Most genes in the primitive endosymbi- onts were transferred to the plant nuclear genome, yet both organelles retain remnant genomes and carry out DNA repli- cation, transcription, and translation (Gray, 1999; Kleine et al., 2009). For example, coordinated expression of nuclear and organelle genes is a central subject in the studies of chloroplast biogenesis. Despite its smaller genome size and limited genetic informa- tion, organelle DNA sometimes account for a substantial amount of total DNA because it is present as multiple copies (for review, see Sakamoto et al., 2008). For example, total DNAs from leaf tissues in higher plants often contain >20% of plastid DNAs (ptDNAs) (Bennet and Smith, 1976; Lamppa and Bendich, 1979; Arabidopsis Genome Initiative, 2000; Rauwolf et al., 2010). The copy number of ptDNA appears to correlate with nuclear ploidy and appears to vary among species or even among different tissues and during developmental stages (Herrmann and Kowallik, 1970; Kowallik and Herrmann, 1972; Lamppa and Bendich, 1979; Scott and Possingham, 1980; Kuroiwa et al., 1981; Boffey and Leech, 1982; Tymms et al., 1983). Multiplication of plastids by division during leaf development further complicates the ptDNA amount per organelle. Such a complex polyploid nature of the plastid genome (also of mitochondrial genome) has raised the question of whether organelle DNA levels are controlled spatially or temporally. Although several proteins are known to play roles in maintaining the configuration of plant organelle genomes (Abdelnoor et al., 2003; Edmondson et al., 2005; Zaegel et al., 2006; Shedge et al., 2007; Mare ´ chal et al., 2008, 2009; Rowan et al., 2010), very little is understood about DNA degradation at the molecular level. ptDNAs (and also mitochondrial DNAs [mtDNAs]) are cytolog- ically detectable as nucleoids by staining tissues with DNA fluorescent dye, such as 49,6-diamidino-2-phenylindole (DAPI) or SYBR green I (SYBR) (Kuroiwa, 1991, 2010). The ptDNAs exist as a complex with proteins that constitute plastid nucleoids (Sato et al., 1998, 2001, 2003; Murakami et al., 2000; Jeong et al., 2003). The cytological detection of organelle DNAs is frequently used to estimate DNA levels together with DNA gel blot hybrid- ization, quantitative PCR, and colorimetric detection of DNA hydrolysates. These methods enable us to investigate the num- ber, morphologies, and behavior of plastid nucleoids during leaf development (Rauwolf et al., 2010 and references therein). In Arabidopsis thaliana, cytological observation demonstrated that amounts of ptDNA increase more than 10-fold during leaf de- velopment (Fujie et al., 1994). By contrast, contradictory results were reported for ptDNA levels in mature leaves: constant or declining ptDNA amounts have been reported by different lab- oratories (Rowan et al., 2004, 2009; Li et al., 2006; Zoschke et al., 2007). Studies of numbers of mtDNAs per cell have revealed that 1 These authors contributed equally to this work. 2 Address correspondence to [email protected]. The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantcell.org) is: Wataru Sakamoto ([email protected]). C Some figures in this article are displayed in color online but in black and white in the print edition. W Online version contains Web-only data. www.plantcell.org/cgi/doi/10.1105/tpc.111.084012 The Plant Cell, Vol. 23: 1608–1624, April 2011, www.plantcell.org ã 2011 American Society of Plant Biologists
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
A Conserved, Mg2+-Dependent Exonuclease DegradesOrganelle DNA during Arabidopsis Pollen Development C W
Ryo Matsushima,a,1 Lay Yin Tang,a,1 Lingang Zhang,a Hiroshi Yamada,a David Twell,b and Wataru Sakamotoa,2
a Institute of Plant Science and Resources, Okayama University, Kurashiki, Okayama 710-0046, Japanb Department of Biology, University of Leicester, Leicester LE1 7RH, United Kingdom
In plant cells, mitochondria and plastids contain their own genomes derived from the ancestral bacteria endosymbiont.
Despite their limited genetic capacity, these multicopy organelle genomes account for a substantial fraction of total cellular
DNA, raising the question of whether organelle DNA quantity is controlled spatially or temporally. In this study, we
genetically dissected the organelle DNA decrease in pollen, a phenomenon that appears to be common in most angiosperm
species. By staining mature pollen grains with fluorescent DNA dye, we screened Arabidopsis thaliana for mutants in which
extrachromosomal DNAs had accumulated. Such a recessive mutant, termed defective in pollen organelle DNA degrada-
tion1 (dpd1), showing elevated levels of DNAs in both plastids and mitochondria, was isolated and characterized. DPD1
encodes a protein belonging to the exonuclease family, whose homologs appear to be found in angiosperms. Indeed, DPD1
has Mg2+-dependent exonuclease activity when expressed as a fusion protein and when assayed in vitro and is highly active
in developing pollen. Consistent with the dpd phenotype, DPD1 is dual-targeted to plastids and mitochondria. Therefore, we
provide evidence of active organelle DNA degradation in the angiosperm male gametophyte, primarily independent of
maternal inheritance; the biological function of organellar DNA degradation in pollen is currently unclear.
INTRODUCTION
Mitochondria and plastids originate from the endosymbiosis of
rickettsia-like a-proteobacteria and cyanobacteria-like photo-
synthetic bacteria, respectively (Gray et al., 1999; Dyall et al.,
2004; Keeling, 2010). Most genes in the primitive endosymbi-
onts were transferred to the plant nuclear genome, yet both
organelles retain remnant genomes and carry out DNA repli-
cation, transcription, and translation (Gray, 1999; Kleine et al.,
2009). For example, coordinated expression of nuclear and
organelle genes is a central subject in the studies of chloroplast
biogenesis.
Despite its smaller genome size and limited genetic informa-
tion, organelle DNA sometimes account for a substantial amount
of total DNA because it is present as multiple copies (for review,
see Sakamoto et al., 2008). For example, total DNAs from leaf
tissues in higher plants often contain >20% of plastid DNAs
(ptDNAs) (Bennet and Smith, 1976; Lamppa and Bendich, 1979;
Arabidopsis Genome Initiative, 2000; Rauwolf et al., 2010). The
copy number of ptDNA appears to correlate with nuclear ploidy
and appears to vary among species or even among different
tissues and during developmental stages (Herrmann andKowallik,
Scott and Possingham, 1980; Kuroiwa et al., 1981; Boffey and
Leech, 1982; Tymms et al., 1983). Multiplication of plastids by
division during leaf development further complicates the ptDNA
amount per organelle. Such a complex polyploid nature of the
plastid genome (also of mitochondrial genome) has raised the
question of whether organelle DNA levels are controlled spatially
or temporally. Although several proteins are known to play roles
in maintaining the configuration of plant organelle genomes
(Abdelnoor et al., 2003; Edmondson et al., 2005; Zaegel et al.,
2006; Shedge et al., 2007; Marechal et al., 2008, 2009; Rowan
et al., 2010), very little is understood about DNA degradation at
the molecular level.
ptDNAs (and also mitochondrial DNAs [mtDNAs]) are cytolog-
ically detectable as nucleoids by staining tissues with DNA
fluorescent dye, such as 49,6-diamidino-2-phenylindole (DAPI) or
SYBR green I (SYBR) (Kuroiwa, 1991, 2010). The ptDNAs exist as
a complex with proteins that constitute plastid nucleoids (Sato
et al., 1998, 2001, 2003; Murakami et al., 2000; Jeong et al.,
2003). The cytological detection of organelle DNAs is frequently
used to estimate DNA levels together with DNA gel blot hybrid-
ization, quantitative PCR, and colorimetric detection of DNA
hydrolysates. These methods enable us to investigate the num-
ber, morphologies, and behavior of plastid nucleoids during leaf
development (Rauwolf et al., 2010 and references therein). In
Arabidopsis thaliana, cytological observation demonstrated that
amounts of ptDNA increase more than 10-fold during leaf de-
velopment (Fujie et al., 1994). By contrast, contradictory results
were reported for ptDNA levels in mature leaves: constant or
declining ptDNA amounts have been reported by different lab-
oratories (Rowan et al., 2004, 2009; Li et al., 2006; Zoschke et al.,
2007). Studies of numbers of mtDNAs per cell have revealed that
1 These authors contributed equally to this work.2 Address correspondence to [email protected] author responsible for distribution of materials integral to thefindings presented in this article in accordance with the policy describedin the Instructions for Authors (www.plantcell.org) is: Wataru Sakamoto([email protected]).CSome figures in this article are displayed in color online but in blackand white in the print edition.WOnline version contains Web-only data.www.plantcell.org/cgi/doi/10.1105/tpc.111.084012
The Plant Cell, Vol. 23: 1608–1624, April 2011, www.plantcell.org ã 2011 American Society of Plant Biologists
mtDNA levels also fluctuate during leaf development. More
importantly, some mitochondria might lack a complete genome
(Preuten et al., 2010). A decrease in mtDNAs has also been
reported during pollen development (Wang et al., 2010). These
circumstantial observations prompted us to study organelle DNA
degradation by a forward genetic approach.
Here, we specifically examine organelle DNA levels in mature
pollen of Arabidopsis. Much attention has been given to organelle
DNA levels in male reproductive organs because the decrease of
organelle DNAs in male tissues is suggested to correlate with
their maternal inheritance: ptDNAs that are detectable by DAPI
stains in male germ cells are often associated with biparental
inheritance of ptDNAs (Hagemann and Schrodoer, 1989; Nagata
et al., 1999; Birky, 2001; Hagemann, 2004; Kuroiwa, 2010).
Irrespective of the inheritance mode, however, we noticed that
organelle DNAs are cytologically absent in pollen vegetative cells
(Matsushima et al., 2008a; Sakamoto et al., 2008). A survey of
numerous mature pollen grains using DAPI staining has revealed
that almost all angiosperm species lack cytologically detect-
able organelle DNAs in pollen vegetative cells (Corriveau and
Coleman, 1988;Mogensen, 1996; Zhang et al., 2003;Wang et al.,
2010). It is noteworthy that pollen vegetative cells do not con-
tribute to fertilization, but they do contain numerous plastids and
mitochondria, which might be necessary for a pollen tube to
germinate, elongate, and deliver spermcells into the embryo sac.
Given that the lack of DAPI signals in pollen vegetative cells is so
clear and consistent in many species (irrespective of inheritance
mode), we reasoned that organelle DNA levels are strictly down-
regulated by a dominant mechanism. We also exploited the
relative ease in examining organelle DNAs in mature pollen,
rather than in leaf tissues, which require time-consuming sec-
tioning or protoplast isolation.
For this study, we performed extensive forward genetic anal-
ysis and isolated Arabidopsis mutants in which organelle DNA
was retained in mature pollen grains. Characterization of the
gene responsible for the mutants led us to identify a nuclease
that is expressed preferentially during pollen development. It is
particularly interesting that this DNase is localized in both the
plastid andmitochondria, providing evidence for the existence of
an organelle nuclease in eukaryotes. Our data reveal an active
mechanism of organelle DNA degradation in a tissue-specific
manner, which is primarily independent of maternal inheritance.
Our cytological analysis of pollen grains in Arabidopsis revealed
that when mature pollen was stained with DAPI, almost no
signals corresponding to organelle DNAs were detected (Figures
1 and 2; see also Sakamoto et al., 2008). Numerous plastids and
mitochondria exist in pollen vegetative and sperm cells. There-
fore, we inferred that organelle DNAs decrease during pollen
development and that it is feasible to screen mature pollen for
mutants that exhibit altered levels of organelle DNAs. This
strategy presents the additional advantage that the pollen phe-
notype will segregate in M1 pollen and thus can be found by
screening mature pollen grains from M1 flowers (Chen and
McCormick, 1996). In our screening method, pollen grains col-
lected from M1 or M2 flowers were fixed briefly with glutaralde-
hyde with subsequent gentle squashing over a cover slip and
DAPI staining, which allowed careful observation of DAPI sig-
nals (Figure 1A). Mature pollen grains of ;2000 individual ethyl
methanesulfonate–mutagenized M1 and 2000 M2 Arabidopsis
plants were screened using this method. As a consequence, we
isolated five mutant lines that exhibited unusual DAPI signals
within the cytoplasmof vegetative cells. TheseDAPI signalswere
distinct from those corresponding to vegetative and spermnuclei
and rather resembled organelle DNAs (Figure 1B). Mutants
isolated in M1 population were further characterized to obtain
M2 individuals where all pollen showed the phenotype. These
mutants were designated as defective in pollen organelle DNA
degradation (dpd). This dpd phenotype was characterized ge-
netically in the subsequent generations. Overall, our screening
and genetic analysis identified two recessive mutations, dpd1
and dpd2; this work specifically examined dpd1. All mutants
except for dpd2 were the alleles of dpd1 (designated as dpd1-1
to dpd1-4; see below). The recessive nature of dpd mutations
Figure 1. Experimental Strategy to Isolate Mutants Defective in Organ-
elle DNA Degradation in Arabidopsis Pollen.
(A) Schematic representation of the experiment. Arabidopsis wild-type
Col were mutagenized using ethyl methanesulfonate. Mature pollen
grains from M1 or M2 plants were screened using the pollen squash
method (middle) for easier observation of DAPI signals derived from
organelle DNAs. Examples of DAPI-stained pollen grains before and after
the squash are shown on the right. Bars = 20 mm.
(B) Isolation of dpd mutants. DAPI-stained squashed pollen from Col
(wild type), dpd1, and dpd2 are shown.
Organelle DNA Degradation in Pollen 1609
Figure 2. Observation of Organelle DNAs in Developing Pollen and Leaf Mesophyll Cells of Col and dpd1-1.
(A) Schematic representation of pollen development in Arabidopsis.
(B) DAPI-stained Technovit sections of bicellular and tricellular pollens. Strong signals are indicated by arrows in tricellular pollen and sperm nuclei;
other signals in dpd1 correspond to extrachromosomal DNAs. Bars = 5 mm.
(C) Electron micrographs of mature pollen grains from wild-type ecotype Col and dpd1-1. Bars = 2 mm.
(D) DAPI-stained mesophyll protoplasts from Col and dpd1-1. Protoplasts were gently squashed to detect DAPI signals within chloroplasts. Bars =
20 mm.
1610 The Plant Cell
implied a dominant mechanism to decrease organelle DNAs in
mature pollen, as we expected.
Organelle DNAs Do Not Decrease during dpd1
Pollen Development
To observe organelle DNA levels more carefully, we prepared
Technovit-embedded thin sections (0.5mm) of developing pollen
grains fromwild-type Columbia (Col) and dpd1 and stained them
with DAPI. In Arabidopsis, mature pollen is formed after two
characteristic mitoses (Figure 2A) (Borg et al., 2009). Asymmetric
division of uninucleate microspores at pollen mitosis I produces
bicellular pollen comprising a vegetative cell and a germ cell.
Subsequent division of the germ cell at pollen mitosis II (PMII)
produces a pair of sperm cells forming tricellular pollen. Our thin
section analysis showed that, in Col, cytoplasmic DAPI signals
start to decrease in the late bicellular stage. They disappear
completely at the tricellular stage in Col, indicating that the
decrease (or degradation) of organelle DNA normally occurs
during PMII (Figure 2B). By contrast, dpd1 showed strong
cytoplasmic DAPI signals in pollen, which were retained even
at the tricellular stage. Segregation of the dpd phenotype in the
F2 population from a cross between dpd1 and Col revealed that
dpd1 behaved as a single recessive trait and that the dpd
phenotype showed complete penetrance in the selfed progeny
(see Supplemental Tables 1 and 2 online). No ultrastructural
abnormality was observed in dpd1 organelles (Figure 2C),
suggesting that the membrane integrity of both mitochondria
and plastids is maintained. Examination of in vitro–germinated
pollen by DAPI also revealed that organelle DNAs were retained
even after germination in dpd1 (see Supplemental Figure 1 on-
line). These results demonstrated that dpd1 appears to compro-
mise DNA reduction during pollen development.
Vegetative and Reproductive Growth of dpd1
We next examined whether dpd1 displays any visible pheno-
types not only in reproductive growth but also in vegetative
growth. None of the dpd1 alleles showed differences in their
vegetative growth under normal conditions (Figure 3A), suggest-
ing that DPD1 primarily affects organelle DNA levels in pollen
grains but not those in other tissues. To test this possibility, we
examined organelle DNA levels by staining protoplasts derived
from 6-week-old mature leaves with DAPI. Results showed that,
unlike mature pollen grains, no apparent difference in DAPI-
detectable organelle DNAs was detected in chloroplasts of leaf
cells (Figure 2D). These results imply that, in dpd1, organelle DNA
levels are increased in pollen grains but not in other somatic
tissues. Subsequently, visible phenotypes in dpd1 male repro-
ductive organs were also examined. Visual inspection showed
Figure 3. Plant Architecture and Reproductive Organs of dpd1 Mutants.
(A) Five-week-old dpd1-1. Bar = 1 cm.
(B) Immature seeds from wild-type Col and dpd1-1. Bars = 1 mm.
(C) Floral organs from Col and dpd1-1. Bars = 2 mm.
(D) Single flower from Col and dpd1-1. Bars = 2 mm.
(E) Single anther from Col and dpd1-1 stained with Alexander solution. Bars = 100 mm.
Organelle DNA Degradation in Pollen 1611
that flowers and seed set appeared normal in dpd1 (Figures 3B
to 3D). The dpd phenotype was completely penetrant in dpd1
selfed progeny and segregated normally in the F2 (see Supple-
mental Tables 1 and 2 online). Mature pollen grains in dpd1
exhibitedmorphology and viability (Alexander stain, Figure 3E; see
Supplemental Table 3 online) that were indistinguishable from
those of the wild type. No difference in pollen size was found
between the wild type and dpd1 (see Supplemental Table 4
online). Collectively, we inferred that the dpd phenotype does
not significantly influence overall plant growth and pollen
morphology.
Both ptDNA and mtDNA Are Retained in dpd1 Pollen
The dpd phenotype in dpd1 raised the question of which DNA-
containing organelles––plastids, mitochondria, or both––emitted
the DAPI signals in pollen vegetative cells. To address this
question, we visualized plastids and mitochondria in mature
pollen by expressing organelle-targeted fluorescent proteins
(green fluorescent protein in plastids [ptGFP] and red fluorescent
protein in mitochondria [mtRFP]; Figure 4). We reported previ-
ously that these organelle-targeted GFP/RFPs, when expressed
under the control of a vegetative cell-specific promoter LAT52
from tomato (Solanum lycopersicum; Twell et al., 1990), can show
plastids or mitochondria in pollen vegetative cells (Matsushima
et al., 2008b; Tang et al., 2009). We introduced the correspond-
ing transgenes (Lat52pro:PTS:GFP and Lat52pro:MTS:RFP) into
dpd1. Mature pollen grains from these transgenic lines were
examined using DAPI or SYBR stain (depending on the fluores-
cent protein used) and simultaneous detection of fluorescent
proteins. The results demonstrated that DAPI signals in dpd1
(Lat52pro:PTS:GFP) colocalized with ptGFP. Similarly, SYBR
signals in dpd1 (Lat52pro:MTS:RFP) colocalized with mtRFP
(Figure 4). These results indicate that the extrachromosomal
DNA signals in dpd1 mutants were derived from both plastids
and mitochondria.
Both ptDNA and mtDNA Increased in Pollen but Not in
Somatic Cells
To confirm that both ptDNA and mtDNA levels were altered in
pollen but not in somatic tissues of dpd1, we performed a PCR-
based assay. Total DNAs were prepared from pollen and young
seedlings and assayed by quantitative real-time PCR. Levels of
ptDNAs (psbA) or mtDNAs (cox1) in these tissues of Col and
dpd1 were normalized based on the level of nuclear DNAs (18S
rDNA; see Methods and Rowan et al., 2009). As expected,
ptDNAs andmtDNAs had significantly increased in dpd1-1 com-
pared with Col (n = 3, Welch’s t test, P = 0.0041 and 0.0226,
respectively) (Figures 5A and 5B). A large difference in mtDNA
levels was detected, probably because of an extremely low level
of mtDNA in Arabidopsis wild-type pollen grains (Wang et al.,
2010). By contrast, young seedlings showed no significant
differences between Col and dpd1-1 in the level of ptDNA and
mtDNA (P = 0.2887 and 0.2692, respectively) (Figures 5C and
5D). Together, these results verified our cytological analysis of
dpd1 pollen stained with DAPI. We concluded that both ptDNA
and mtDNA levels are increased in dpd1 pollen.
Map-Based Cloning of the DPD1 Locus
We identified the DPD1 gene based on conventional map-based
cloning. Wemapped the dpd1-1mutation within a 623-kb region
on chromosome 5 (summarized in Supplemental Figure 2 online).
A survey of genes that potentially encode proteins targeted to
plastids and/ormitochondria allowed us to select possible genes
responsible for the dpd1 phenotype. Subsequent sequencing of
the candidate genes identified a base change in the At5g26940
gene, which results in an amino acid substitution (Figure 6A).
Figure 4. Colocalization of DNA Signals with Plastids and Mitochondria in dpd1 Pollen.
A transgene that expresses plastid-targeted GFP or mitochondria-targeted RFP in pollen vegetative cells (presented on the left) was introduced into
dpd1-1. Colocalization of plastidial GFP signals with DAPI signals (top panels) or of mitochondrial RFP signals with SYBR (bottom panels) was
examined. Arrowheads indicate colocalization of fluorescent organelles and DAPI-stained or SYBR-stained signals. The asterisk denotes nuclear-
derived DAPI signals. Bars = 5 mm.
1612 The Plant Cell
Sequencing of this gene in three other dpd1 alleles (dpd1-2,
dpd1-3, anddpd1-4; Figure 6B) revealed that all these alleles had
base changes in At5g26940 (Figure 6). The base changes in
dpd1-1, dpd1-2, and dpd1-3 caused amino acid substitutions
(see Supplemental Figure 3 online), whereas dpd1-4 had a base
change close to the border of intron1/exon2. Furthermore, we
obtained two T-DNA insertion mutants (dpd1-5 and dpd1-6) in
At5g26940 and examined whether they showed the dpd1 phe-
notype. As expected, they all exhibited the dpd phenotype that
was indistinguishable from other dpd1 alleles.
To prove that At5g26940 encodes DPD1, a genomic fragment
encompassing At5g26540 was transformed into the dpd1-1
mutant, we generated five transgenic lines. One of the lines
was homozygous for the transformed sequence. It fully comple-
mented the dpd phenotype (Figure 7A) when mature pollen
grains were examined using our squash method. Moreover,
we conducted a PCR-based assay to verify the altered or-
ganelle DNA levels in dpd1 and the complemented line. PCR
products corresponding to plastid and mitochondria DNAs
were more abundant in dpd1-1 pollen than in Col and the
complemented line. Collectively, we concluded that At5g26940
encodes DPD1.
DPD1 Encodes a Protein Belonging to the Exonuclease
Family and Dual Targeted to Plastids and Mitochondria
DPD1 is present in a single copy inArabidopsis, where it encodes
a protein (316 amino acids) belonging to the exonuclease family
(Pfam: PF00929, ExonulX-T), which is included in the large
ribonuclease H-like superfamily (Clan CL0219). Three domains
included within the protein family members, ExoI, ExoII, and
ExoIII«, appeared to be conserved in DPD1 and other members
(see Supplemental Figure 3 online). A BLAST search and sys-
tematic analysis of DPD1 homologs using the entire DPD1 amino
acids and the SALAD Database (http://salad.dna.affrc.go.jp/
salad/en/) revealed that DPD1 homologs are present in small
green algae (Ostreococcus and Micromonas), moss (Physcomi-
trella), and higher plants (Figure 6C). By contrast, no suchhomolog
was detectable in a green alga (Chlamydomonas), a red alga
(Cyanidioschyzon), or a fungus (Saccharomyces). Although pro-
teins detectedas related toDPD1 in the green algae andmosshad
conserved exonuclease domains, they were much larger (>1260
amino acids) and apparently had additional DNA helicase do-
mains, suggesting a role that is distinct from organelle DNA
degradation. These results imply that DPD1 had evolved with
the appearance of anisogamous male reproductive organs in the
plant lineage.
It was tempting to speculate that DPD1 is targeted to both
plastids andmitochondria. To examine the cellular localization of
DPD1, we transiently expressed DPD1-GFP in Arabidopsis pro-
toplasts prepared from mesophyll cells, as described previously
(Miura et al., 2007). TheDPD1-GFP fusion protein colocalized not
only with chlorophyll autofluorescence, but also with the mito-
chondria-specific dyeMitotracker Red (see Supplemental Figure
4 online). These results indicate that DPD1 is localized in plastids
and mitochondria, each of which shows defective organelle
DNA degradation in dpd1 mutants.
DPD1 Is a Mg2+-Dependent Exonuclease
Sequence information and thedpd1phenotype together strongly
suggest that DPD1 has exonuclease activity (Figure 6A). To study
this possibility in vitro, a recombinant DPD1-His protein (6 3histidine tagged at C terminus) was generated in Escherichia coli
and affinity purified as described in Methods. In addition to the
wild-type DPD1-His gene, we expressed twoDPD1-His genes in
which a point mutant was introduced (Figure 8A). Because of a
nonsense mutation at the 186th amino acid, DPD1-His(Y186*)
has DPD1 that is C-terminally truncated and lacks His. Similarly,
DPD1-His(A236V) has an amino acid change equivalent to
dpd1-1 (Figure 6A; see Supplemental Figure 3 online). A majority
of these DPD1 proteins expressed in E. coliwere detected as ag-
gregated, although we purified DPD1-His and DPD1-His(A236V)
from soluble fractions (Figures 8B to 8D). Purified DPD-His
proteins were detected as two bands: The lower band was likely
a degradation product. Actually, DPD1-His(Y186*) was not pu-
rified because of the absence of the His tag, but we used this
fraction as a negative control.
Figure 5. Quantitative Analysis of Organelle DNA Levels in Pollen Grains
and Young Seedlings.
Levels of organelle DNAs per nuclear DNAs in Col and dpd1-1 were
determined using real-time PCR. Mean values of normalized organelle
DNA levels of Col were determined as 1.0. The relative values of dpd1-1
were calculated (n = 3, SD as error bars).
(A) ptDNA in pollen grains.
(B) mtDNA in pollen grains.
(C) ptDNA in 17-d-old seedlings.
(D) mtDNA in 17-d-old seedlings.
Organelle DNA Degradation in Pollen 1613
Nuclease activity of these recombinant proteins was assayed
in a standard buffer withmagnesiumbivalent cation as a cofactor
and DNA fragments (PCR-amplified ptDNA). The result showed
that DNAs were degraded by DPD-His during the first 20 min of
incubation (Figure 8E). By contrast, neither heat-denatured
DPD1-His, DPD1-His(A236V), nor the negative control fraction
degraded PCR fragments, even after 60 min. Also, EDTA inhib-
ited nuclease activity of DPD1-His, indicating that magnesium is
a necessary cofactor. By contrast, none of three other bivalent
cations (copper, manganese, or zinc) was observed to act as a
cofactor for DPD1 DNA degradation (Figure 8F). These data
indicate that DPD1-His has magnesium-dependent nuclease
activity. Complete degradation of PCR fragments indicates that
DPD1-His has exonucleolytic activity, but its endonucleolytic
activity was not ruled out completely. To test this possibility, we
performed the same nuclease assays with circular plasmids.
Intact plasmids were stable even after 60-min incubation with
DPD1-His. However, once nicked or linearized by restriction
enzymes (EcoRI, EcoRV, and KpnI), the plasmids became com-
pletely degradable by DPD1-His (Figure 8G). These data reflect
that DPD1 is an exonuclease that requires accessible ends to
degrade double-stranded DNAs.
Pollen-Enhanced Expression of DPD1
We next performed RT-PCR analysis to examine tissue-specific
expression of DPD1 (Figure 9A). RNAs were prepared from
Figure 6. Identification of the DPD1 Gene.
(A) Schematic representation of At5g26940 gene (DPD1) and the predicted protein encoded. Gray boxes show coding regions. Adenine of the
translation start codon (ATG) is designated as +1. Arrows indicate positions of mutations in dpd1 alleles. Here, dpd1-1, dpd1-2, and dpd1-3 are located
within the coding region; dpd1-4 is located at the splicing acceptor site of the first intron. In addition, dpd1-5 and dpd1-6 are T-DNA–inserted alleles.
Red boxes (ExoI, ExoII, and ExoIIIe) are functional domains conserved among 39-59 exonucleases. a.a., amino acids.
(B) DAPI-stained squashed pollen grains in dpd1 mutant alleles other than dpd1-1. A part of the squashed pollen cytoplasm is shown. Bars = 10 mm.
(C) A phylogenetic tree of DPD1 and its homologs. Multiple alignment was performed using ClustalW as described inMethods (see Supplemental Figure
7 and Supplemental Data Set 1 online).
1614 The Plant Cell
seedlings, the shoot apex, flowers, and various stages of pollen
development; RT-PCR was conducted as described previously
(Honys and Twell, 2004). As expected, DPD1 was highly ex-
pressed in flowers compared with whole seedlings and the shoot
apex. Further dissection of pollen development at the stage of
unicellular microspore, bicellular pollen, tricellular pollen, and
mature pollen grains revealed that DPD1 starts to express at the
bicellular pollen stage. It reaches a peak at the tricellular pollen
stage. This temporal expression of DPD1 in pollen coincides
perfectly with the disappearance of DAPI-stained organelle
DNAs (Figure 2B). It also coincides with the fact that organelle
DNA levels in vegetative tissues are not altered in dpd1. To
further characterize DPD1 expression in vivo, transgenic Arabi-
dopsis plants were generated that expressed the previously
described DPD1-GFP fusion under the DPD1 promoter. Char-
acterization of these transgenic plants showed that the DPD1
promoter is highly active in mature pollen grains (Figures 9B and
9C). Furthermore, we crossed this transgenic plant with another
line harboring Lat52pro:MTS:RFP, thereby expressing mtRFP in
vegetative cells. Careful observation of this line revealed smaller
GFP signals colocalized withmtRFP and additional larger signals
corresponding to plastids (Figure 9D). This observation con-
firmed the results of our transient assay in mesophyll cells (see
Supplemental Figure 4 online) that DPD1 is dual-targeted to both
organelles. When characterized by confocal microscopy, we
occasionally found smaller GFP mitochondrial signals that did
not merge with mtRFP in a hollow area that likely corresponds to
sperm cells (Figure 9E). The size and distribution of these small
GFP signals resembled those observed in sperm mitochondria
(Matsushima et al., 2008b), indicating that DPD1 is expressed
not only in vegetative but also in sperm cells. Collectively, these
results suggest that DPD1 is predominantly expressed in the
male gametophyte.
Pollen Viability and Transmission Efficiency in dpd1
Our results thus far indicated that DPD1 is a pollen-specific
exonuclease responsible for organelle DNA degradation. Sup-
porting this assumption, dpd1 showed no detectable phenotype
in plant vegetative growth (Figure 3). To characterize the effect of
organelle DNA degradation in pollen vegetative cells, we further
examined dpd1 pollen viability. We first examined the transmis-
sion efficiency (TE) of the dpd1mutation. A heterozygous DPD1/
dpd1-1 plant was subjected to a reciprocal cross with Col. The
genotype of F1 seeds was determined using PCR-based geno-
typing, as described in Methods, to estimate TE of dpd1 in the
next generation. This result indicated that TE of dpd1 through the
male (TEmale) was not reduced to a statistically significant degree
(Table 1). We also performed conditional TE tests to examine
whether TEmale is affected by the position of seeds in fertilized
siliques. After a reciprocal cross betweenCol andDPD1/dpd1-1,
fertilized siliques were cut into upper and lower halves. Then, the
genotype for DPD1 was determined separately (Table 2). Again,
these results showed no significant difference in TEmale. We
subsequently examined pollen germination ability in vitro. De-
spite the unaffected TE in dpd1, we found that the germination
rate is slightly lower in dpd1-1 than in Col and the complemented
line (Table 3). This reduction, which is likely to be too slight to
Figure 7. Complementation of the dpd1 Phenotype.
(A) Squashed mature pollen stained with SYBR from dpd1, Col, and
dpd1 complemented with At5g26940 genomic sequence (Comp.). A
trace of the picture from dpd1 is shown at the top right. VN, vegetative
of the DNA polymerase III holoenzyme (Scheuermann and Echols,
1984) and TREX1 that degrades retroelements and exogenous
DNAs inmammalian cells (Lehtinenet al., 2008). It canbeassumed
Table 1. Genetic Transmission Analysis of dpd1-1 Mutations
Parental Cross No. of Wild Type (+/+) No. of dpd1-1 (+/�) Total TE (%)a x2 (P Value) for 1:1
dpd1-1+/� 3 Col 99 77 176 77.8 2.8 (0.097)
Col 3 dpd1-1+/� 71 65 136 91.5 0.26 (0.61)
The genotype of each progeny was determined using PCR.aTE (%) = number of mutant/number of wild-type progenies 3 100.
Table 2. Conditional Genetic Transmission Analysis of dpd1-1 Mutations
Parental Crossa No. of Wild Type (+/+) No. of dpd1-1 (+/�) Total TE (%)b x2 (P Value) for 1:1
Col 3 dpd1-1+/�-U 49 39 88 80 1.1 (0.29)
Col 3 dpd1-1+/�-L 34 32 66 94 0.061 (0.81)
dpd1-1+/� 3 Col-U 39 48 87 123 0.93 (0.33)
dpd1-1+/� 3 Col-L 32 44 76 138 1.9 (0.17)
The genotype of each progeny was determined using PCR.aSiliques were cut in the middle and separated into upper (U) and lower (L) parts. Seeds collected from each part were subjected to genotyping
analyses.bTE (%) = number of mutant/number of wild-type progenies 3 100.
1618 The Plant Cell
that one of these nuclease members evolved into DPD1 to
degrade organelle DNAs, along with the emergence of pollen
vegetative cells in angiosperms. Conserved presence of DPD1
homologs in species with anisogamy is consistent with our result
that DPD1 expression is limited predominantly to pollen. Collec-
tively, these data demonstrate that organelle DNA degradation is
common in angiosperm pollen and that it is controlled by the
exonuclease DPD1.
Regulation of DPD1 Expression, Degradation, and
Relevance to General Organelle DNA Degradation
In addition to the circumstances described previously, it is
notable that most of the dpd mutants we isolated through our
extensive screening were dpd1 alleles. We therefore consider
that DPD1 plays a central role in DNA degradation in both
plastids and mitochondria. Apparently, several questions arose
from our findings. We first asked about whether the role of DPD1
in organelle DNA degradation can be generalized in other so-
matic tissues, such as those of leaves. This is apparently not the
case because of (1) a lack of detectable phenotype in dpd1
vegetative growth, (2) pollen-specific expression of DPD1, and
(3) a lack of detectable change of organelle DNA levels in dpd1
leaf tissues. Moreover, DPD1 does not appear in plastid pro-
teome data prepared from nonpollen tissues (e.g., The Plant
Protein Database, http://ppdb.tc.cornell.edu/). Actually, DPD1
has never been associated with plastid nucleoids (or transcrip-
tionally active chromosomes) derived from nonpollen tissues
(Pfalz et al., 2006). Based on these observations, we infer that
DPD1 functions specifically in pollen and does not participate in
general organelle DNA metabolism, including DNA replication
and/or degradation. However, it is possible that misregulation of
DPD1 expression affects organelle DNA levels in somatic cells.
Additional investigations are necessary to elucidate the function
of DPD1 in organelle degradation in tissues other than pollen.
We next examined the question of whether DPD1 acts on DNA
degradation by itself or requires additional factor(s). Given its
exonucleolytic activity, it is presumed that linear DNAs are the
substrate of DPD1, but circular DNAs are not. Although predom-
inant forms of both ptDNAs and mtDNAs are believed to consist
of circular DNAs, nicked or linear molecules have been found in
both plastid and mitochondria (Bendich, 2004; Oldenburg and
Bendich, 2004); those molecules can be DPD1 substrates. It is
therefore possible that DPD1 alone can degrade some, but not
all, of the organelle DNA population. A significant difference in
organelle DNA levels between Col and dpd1 pollen (Figure 5)
implies that most of the organelle DNA population might in fact
be nicked or linear molecules in pollen vegetative cells (Bendich,