Maternal Epigenetic Pathways
Control Parental Contributions
to Arabidopsis Early Embryogenesis
Daphne ´ Autran,1,6Ce ´lia Baroux,2,6Michael T. Raissig,2Thomas Lenormand,3Michael Wittig,4Stefan Grob,2
Andrea Steimer,2Matthias Barann,4Ulrich C. Klostermeier,4Olivier Leblanc,1Jean-Philippe Vielle-Calzada,5
Phillip Rosenstiel,4Daniel Grimanelli,1,* and Ueli Grossniklaus2,*
1Diversite ´, Adaptation et De ´veloppement des Plantes, Institut de Recherche pour le De ´veloppement, Universite ´ de Montpellier, UMR 232,
Montpellier 34394, France
2Institute of Plant Biology and Zu ¨rich-Basel Plant Science Center, University of Zu ¨rich, Zu ¨rich CH-8008, Switzerland
3Centre d’Ecologie Fonctionnelle et Evolutive, CNRS, Universite ´ de Montpellier, UMR 5175, Montpellier 34293, France
4Institute of Clinical Molecular Biology, Christian Albrechts University Kiel, Kiel 24105, Germany
5National Laboratory of Genomics for Biodiversity, CINVESTAV Guanajuato, Irapuato CP36500, Guanajuato, Mexico
6These authors contributed equally to this work
*Correspondence: email@example.com (D.G.), firstname.lastname@example.org (U.G.)
critical to understand the fundamental processes
tive isolation. To determinetheparental contributions
and their regulation during Arabidopsis embryo-
genesis, we combined deep-sequencing-based RNA
profiling and genetic analyses. At the 2–4 cell stage
there is a strong, genome-wide dominance of ma-
ternal transcripts, although transcripts are contrib-
uted by both parental genomes. At the globular stage
to a gradual activation of the paternal genome. We
identified two antagonistic maternal pathways that
control these parental contributions. Paternal alleles
are initially downregulated by the chromatin siRNA
pathway, linked to DNA and histone methylation,
whereas transcriptional activation requires maternal
activity of the histone chaperone complex CAF1. Our
results define maternal epigenetic pathways control-
In most animal species, the zygote is transcriptionally quiescent,
and early embryogenesis is governed by maternal products
stored in the oocyte prior to fertilization (Ande ´ol, 1994). Depend-
ing on the species, zygotic genome activation (ZGA) takes place
after one to several cell divisions. ZGA is a gradual process that
relies on large-scale chromatin reprogramming leading to an
increasing number of zygotically expressed genes (Tadros and
Lipshitz, 2009). Maternal transcripts and proteins inherited
gotic transcripts gradually take over the control of development.
As a result, parental contributions to the embryonic transcrip-
tome dynamically change during early development, with an
initial maternal control that is of variable duration (1 to 15 cell
cycles) (Baroux et al., 2008; Tadros and Lipshitz, 2009). Strik-
ingly, such a maternal influence occurs in animals as evolution-
arily divergent as insects, amphibians, and mammals. Under-
standing the parental contributions and the regulation of
zygotic genome expression during early embryogenesis is
a key question in developmental and evolutionary biology.
In flowering plants, the knowledge about the regulation and
dynamics of parental contributions during early embryogenesis
remains fragmented, despite its importance in understanding
hybrid vigor, hybrid viability, parent-of-origin-dependent inter-
ploidy, and nonself pollination (xenia) effects determined by
interactions of parental genomes after fertilization (Bushell
et al., 2003; Jahnke et al., 2010; Meyer and Scholten, 2007; Pah-
tion produces the zygote, which develops into the embryo
(Movie S1 available online), and the endosperm, an embryo-
nurturing tissue. Both fertilization products develop within
maternal integuments, forming the seed. Genetic studies have
shown that seed development is under maternal influence
(Chaudhury and Berger, 2001), but the composite nature of the
seed makes determining the origin of maternal effects complex.
Recently, downregulation of RNA Polymerase II (PolII) in the
mature Arabidopsis egg cell revealed thatthe embryo developed
to the preglobular stage in absence of significant de novo tran-
scription (Pillot et al., 2010b). Thus, de novo transcription of
parental genomes is not an absolute requirement for early
embryogenesis, but the timing, dynamics, and mechanisms of
zygotic genome activation have yet to be elucidated. Reporter
have identified several transcripts with a dominant maternal
representation at early stages, whereas paternal transcripts
were detected only later (Baroux et al., 2001; Grimanelli et al.,
Cell 145, 707–719, May 27, 2011 ª2011 Elsevier Inc. 707
2005; Vielle-Calzada et al., 2000). In contrast, a biparental
expression in the zygote and early embryo was shown for certain
other genes (Aw et al., 2010; Meyer and Scholten, 2007; Ron-
ceret et al., 2005, 2008; Scholten et al., 2002; Weijers et al.,
2001). Furthermore, near-saturation mutagenesis screens iden-
tified a plethora of mutations affecting embryo development at
or before the globular stage. The majority of these segregate
as zygotic recessive traits, indicating biparental contributions
to early embryogenesis (Tzafrir et al., 2004). Thus, to date there
is no clear understanding of the relative parental contributions
to plant embryogenesis nor are the mechanisms regulating the
respective contributions known.
We performed allele-specific profiling of the embryonic tran-
scriptome and quantified relative transcript contributions of the
paternal and maternal genomes in Arabidopsis embryos. We
represented. This finding reconciles the observation made by
several laboratories of both maternal and zygotic effects during
early embryogenesis. Using reporter and profiling analyses, we
found an increasing contribution of paternal products as embryo
basis. We identified two maternal epigenetic pathways, involving
the chromatin siRNA pathway and the histone chaperone
contribution. Importantly, we show that these pathways are
distinct from those regulating genomic imprinting.
Analysis of Parental Contributions in Isolated
To unambiguously define the parental contributions in the early
embryo, we profiled the transcriptome of early-stage embryos
in an allele-specific manner. We dissected 2–4 cell and globular
stage embryos (Figure 1A) derived from a cross between the
polymorphic Landsberg erecta (Ler) and Columbia (Col) acces-
sions. The embryonic transcriptome was sequenced using
a SOLiD v3 platform (Figure 1; see Table S1). Reads covering
single-nucleotide polymorphisms (SNPs) (Borevitz et al., 2007)
were extracted and are referred to hereafter as informative reads
with respect to parental origin. Informative reads were used to
quantify parental contributions both globally (Figure 1A), and
for each gene individually. Genes were considered biparentally
expressed at a given stage whenever both parental alleles
were identified in the corresponding transcriptome. The classifi-
cation of genes for which only a single allelic variant was de-
tected was more ambiguous. Transcripts might be detected
from only one parent because of uniparental expression or
because of low sampling rates, which is of particular concern
for genes expressed at low levels. To circumvent this difficulty,
we made a probabilistic model describing the distribution of
genes according to the proportion of maternal transcripts, q,
Figure 1. Parental Contributions to the Arabidopsis Early Embryo
(A) Distribution of maternal versus paternal reads covering the embryonic
transcriptome of isolated embryos. Embryos were derived from a cross
between polymorphic parents. Informative SOLiD reads (Table S1) mapping to
known SNPs were assigned to the maternal (Ler) or paternal (Col) parent.
(B) Informative reads identified 3973 and 3078 genes in the 2–4 cell and
globular transcriptomes, respectively. The graph shows a likelihood-based
gene distribution according to the proportion of maternal transcripts, q. q = 1
and q = 0 represent genes contributed only maternally or paternally, respec-
tively. Genes with 0 < q < 1 are contributed biparentally. y axis: proportion of
genes (log scale) ; x axis:q values along 1% quantiles. The full-colored circles
indicate extreme quantiles (0 % q < 0.01 and 0.99 < q % 1).
(C) Composite diagram representation of the gene distribution as drawn in
(B) according to q intervals as labeled. Related to Figure S1, Table S1, and
708 Cell 145, 707–719, May 27, 2011 ª2011 Elsevier Inc.
which was adjusted to best-fit the observations (Experimental
Procedures). From this likelihood-based distribution, we esti-
mated the frequency of genes fitting the biparental class (0 <
q < 1, both parental alleles were detected), the paternal class
(q = 0, only the paternal alleles were detected), or the maternal
class (q = 1, only the maternal alleles were detected). In addition,
each gene could be assigned a probability of falling within each
class (Table S2). This integrated approach allowed a quantitative
and qualitative analysis of parental contributions to the embry-
The Transcriptome of 2–4 Cell Embryos Is Maternally
Dominant despite Significant Contributions from Both
Strikingly, at the 2–4 cell embryo stage 88.4% of the informative
reads (n = 135,142) were of maternal origin (Figure 1A). This was
confirmed in an independent biological replicate (Figure S1). The
informative reads represented 3973 loci located throughout the
genome. The gene distribution drawn for q quantiles showed
a strong bias toward maternal overrepresentation (Figure 1B)
with 85% of the genes described by q > 0.75 (Figure 1C; Table
S1). Transcripts of the biparental class (0 < q < 1; 68.2% of the
identified genes) contributed strongly to this maternal domi-
nancewith 54.7% genesdescribed by0.75<q <1,i.e.,maternal
overrepresentation (Figure 1C and Table S1). Furthermore, our
analysis revealed 30.2% transcripts of the maternal class
(q = 1) against only 1.6% transcripts of the paternal class
(q = 0) at the 2–4 cell stage. Thus, in Arabidopsis both parental
genomes contribute to the early embryonic transcriptome but
overall it is clearly dominated by maternal transcripts.
The Paternal Contribution Is Higher at the Globular
To determine how parental transcript contributions change
during embryogenesis, we extended our allele-specific profiling
to embryos at the globular stage. Although maternal dominance
was maintained, the paternal contribution increased, as shown
by 35.9% paternal reads versus 11.6% at the 2–4 cell stage (Fig-
ure 1A and Table S1). These informative reads identified 3078
loci, for which the q distribution remained skewed toward
maternal overrepresentation, although to a lesser extent than
at the 2–4 cell stage (Figure 1B). The maternal class (q = 1) rep-
resented 13.8% of the genes (versus 30.2% at the 2–4 cell
stage), and the paternal class (q = 0) increased marginally to
2.3% (versus 1.6% at the 2–4 cell stage) (Table S1). Concomi-
tantly, the biparental class increased, now comprising 83.9%
of genes, with 34.5% showing maternal overrepresentation
(0.75 < q < 1) (versus 54.7% at the 2–4 cell stage) (Figure 1C).
Importantly, the globular stage transcriptome shared 2417
genes (78.5%) with that of the 2–4 cell stage, representing
95% of the reads (Figure 2A). We analyzed the changes of
parental contributions among these shared genes by quantifying
class transitions (Figure 2B). For instance a transition from the
maternal class to the biparental or paternal class indicates de
novo activation of the paternal allele and represents 21.5%
(515) of the genes (Figure 2C). De novo activation of the maternal
allele was less prominent with only 0.8% genes, mostly because
few paternally expressed genes were identified at the 2–4 cell
stage. This analysis identified loci with a decay of one parental
transcript, with 2.5% and 9.7% showing loss of their maternal
or paternal transcripts, respectively (Figure 2C). However, most
quantitative changes occurred in the biparentally expressed
class (54.4%, 1315 genes) where the majority showed a marked
increase in the relative paternal contribution (778 genes, Fig-
ure 2D). This could result from decay of maternal RNAs, de
novo transcription of paternal alleles, or a combination of both.
To refine the timing of activation of paternal alleles, we moni-
tored paternal activity of six marker lines expressing a reporter
gene under diverse promoters active during early embryogen-
esis. The lines reflect genes with diverse cellular functions (Table
S3) and showed either early, intermediate, or late paternal
activity, respectively (e.g., RPS5A, CYCB1;1, and ET1041), but
did not appear in our allele-specific transcriptome due to the
absence of a referenced SNP in their sequence. The markerlines
clearly showed distinct expression depending on whether they
were maternally or paternally inherited (Figure 3A and Fig-
ure S2A). We scored the number of F1 embryos showing
paternal marker expression at the same developmental stage
and in a wild-type maternal background (three to seven biolog-
ical replicates each, Table S4). For all markers the proportion
of stained embryos increased with developmental progression
(Figure 3B and Figure S2B). Consistent with our RNA profiling
results, paternal expression of the markers displayed gene-
specific activation timing (developmental stage at which the first
expression was detected) and kinetics (incremental increase in
the fraction of progeny showing expression). For instance, in
a Ler maternal background the paternal ET1041 marker showed
only 4% stained embryos at the 2–4 cell stage, whereas the
(Figure 3B). In contrast, maternally transmitted markers showed
consistent expression in essentially all embryos, even at earliest
stages (Figures S2A and S2C).
Taken together with the transcriptome study, these findings
strongly suggest that paternally inherited alleles—even those
that are detectable at a very early stage—are activated gradually
after fertilization. This is consistent with the previous observa-
tions made for individual genes in Arabidopsis, and reconciles
earlier, apparently conflicting reports. Whether maternal loci
follow similar or different activation kinetics cannot be easily
resolved because of the potential importance of maternal carry-
over. Nevertheless, our observations are reminiscent of the
gradual, de novo, expression of zygotic genes reported in
animals (Tadros and Lipshitz, 2009).
Maternal KRYPTONITE Activity Controls Paternal
Contribution in the Early Embryo
In our marker analyses we observed that expression was influ-
enced by the maternal genotype (i.e., the accession) (Fig-
ure S2B). This indicated a maternal control of paternal expres-
sion as suggested previously (Ngo et al., 2007). The gradual
increase in paternal allele expression might reflect the progres-
sive release of a silencing mechanism. In Arabidopsis, silent
chromatin is enriched in histone H3 dimethylated at lysine 9
(H3K9me2), a modification principally deposited by the SUVH4
histone methyltransferase KRYPTONITE (KYP) (Jackson et al.,
Cell 145, 707–719, May 27, 2011 ª2011 Elsevier Inc. 709
2002). To investigate the possibility that maternal KYP regulates
the activity of paternal alleles, we quantified reporter gene
expression in seeds resulting from a cross between a maternal
kyp mutant and wild-type pollen carrying the marker. For all
reporters tested, lack of maternal KYP activity significantly
increased the proportion of embryos showing early paternal
reporter activity (before the 16-cell embryo stage) (Figures 4A
and 4B, Figure S2A, and Table S4). We confirmed the maternal
effect of the kyp mutation on the endogenous AGP18 locus by
allele-specific RT-PCR (Figure 4C). These data strongly suggest
a role for maternal KYP activity in repressing the transcription of
paternal alleles during early embryogenesis.
To investigate the maternal effect of the kyp mutation at the
genome-wide level, we performed allele-specific profiling of
2–4 cell embryos dissected from crosses between maternal
kyp (Ler) and paternal wild-type (Col) parents (Table S1).
Embryos inheriting maternal kyp (kypm/KYPp) showed a strong
increase in the proportion of paternal reads (35.9% versus
11.6% in 2–4 cell stage wild-type embryos) (Figure 4D and repli-
cate Figure S1) resulting in a similar paternal contribution as in
Figure 2. Dynamic Changes of Parental
Contributions during Embryo Development
(A) Venn diagrams showing the number of genes
between 2–4 cell and globular embryo tran-
scriptomes. Global coverage for specific genes is
by the majority of reads (90%–95%).
(B) The transition tables describe the changes for
common genes in the parental class (P, B, M, see
legend) that occurred during development (2–4
(C) Subsets of class transitions illustrate dynamic
changes in allele representation as indicated
(activation/de novo expression and decreased
expression/decay of one parental allele) during
developmental progression. Note that ‘‘decay’’
may correspond to a decrease in SNP coverage
falling below detection threshold rather than an
absolute loss of transcript. The % are the sum of
the % genes in (B) falling into the gray transitions.
(D) A vast majority of common genes (54.4%) is
biparentally represented at both stages. The
proportion of maternal transcripts (q) was calcu-
lated for 1125 common genes sequenced on both
alleles and plotted as indicated (left). The inter-
pretation of relative changes toward higher
paternal or maternal representation is shown
(right). The inset shows the number of genes with
changes in q values > 10% (dashed lines). A total
of245genes showednoor<10% change. Related
to Table S1 and Table S2.
wild-type embryos at the globular stage
(36.1%; Figure 1A). Informative reads in
genes (Table S1) for which the q distribu-
tion was shifted toward a higher paternal
representation compared to wild-type
embryos at the 2–4 cell stage (Figure 4E).
Notably, the proportion of the paternal class increased markedly
(10.3% at q = 0, versus 1.6% in wild-type 2–4 cell embryos, Fig-
ure 4F) whereas the proportion of genes in the maternal class
diminished (Figure 4F). In addition, genes in the biparental class
(0 < q < 1) showed a higher paternal contribution compared to
wild-type embryos at the 2–4 cell stage (shifted distribution Fig-
ure 4E; less maternally dominant genes Figure 4F). Thus, the
maternal kyp mutation affected a large number of genes
throughout the genome, resulting in a higher paternal contribu-
tion to the early embryonic transcriptome.
Importantly, 2461 of these 3125 genes were common to the
wild-type transcriptome of embryos at the same stage and
were covered by 94.6% of the informative reads (Figure S3B).
This suggests that the maternal kyp mutation does not drasti-
cally alter the 2–4 cell stage embryonic transcriptome, but
instead modifies the relative parental contributions of genes
normally expressed at this stage. Interestingly, the maternal
kyp mutation induced class transitions similar to those induced
by developmental progression (Figures S3C–S3E compared to
Figures 2B–2D). The changes in the relative contribution of
710 Cell 145, 707–719, May 27, 2011 ª2011 Elsevier Inc.
for a subgroup (158 genes) showing a higher increase of the
paternal contribution in kypm/KYPp2–4 cell embryos as com-
pared to wild-type globular embryos (Figure 4G). The analysis
kyp mutation induces an increased paternal and maternal contri-
bution for 1307 and 117 genes, respectively (Figure S3F).
Taken together these results strongly suggest that maternal
KYP activity downregulates early transcription of many paternal,
and some maternal, alleles of loci throughout the genome. This
maternal control may be progressively released during embryo-
The Chromatin siRNA Pathway Maternally Controls
the Paternal Contribution to the Early Embryo
Because KYP-dependent H3K9me2 in Arabidopsis is linked
to non-CG DNA methylation, we verified the effects of muta-
tions in DOMAIN REARRANGED METHYLTRANSFERASE2
(DRM2) and CHROMOMETHYLASE3 (CMT3), two genes that
control non-CG methylation (Feng et al., 2010). Similar to kyp,
both mutations inherited maternally resulted in precocious
activation of the paternally inherited markers (Figure 5A and
Figures S4A and S4B). By contrast, maternal mutations in
either METHYLTRANSFERASE1 (MET1) or DECREASED DNA
METHYLATION1 (DDM1), involved in the maintenance of CG
DNA methylation (Feng et al., 2010), had no consistent effect
(Figure S4C). Thus, non-CG but not CG DNA methylation partic-
ipates in the transcriptional repression of paternal marker genes.
In Arabidopsis, DNA and histone methylation can be mediated
by small-interfering RNAs (siRNAs) via the chromatin siRNA
pathway, also known as the RNA-directed DNA methylation
(RdDM) pathway (Brodersen and Voinnet, 2006). To determine
whether the RdDM pathway plays a role in the maternal repres-
sion of paternal alleles, we tested mutations in the following
RdDM components: NRPD1a (PolIV), NRPD1b (PolV/NRPE1),
DCL3, RDR2, and AGO4 (Table S3). When inherited maternally,
all these mutations allowed an earlier and stronger detection
of paternally transmitted markers (Figure 5A). By contrast,
the dcl2-1 mutant, which affects a distinct siRNA-dependent
Figure 3. Gradual Activation of Paternal Markers during Early Embryo Development
(A) Representative panel showing differential expression of the marker tested (Table S3 and Table S4), here MET333 reporting AGP18 expression, when
transmitted maternally (left) or paternally (right), as monitored by histochemical detection of GUS (blue substrate).
(B) Expression of the paternal markers was scored as the proportion of embryos showing GUS staining at a given developmental stage in a wild-type maternal
background. Two-tailedFisher’s exacttests wereusedtoassessdifferences between twoconsecutivedevelopmentalclasses:*p<0.05;**p<0.01;***p<0.001;
****p < 0.0001. Error bars represent standard error between independent biological replicates.
Related to Figure S2, Table S3, and Table S4.
Cell 145, 707–719, May 27, 2011 ª2011 Elsevier Inc. 711
Figure 4. Maternal KYP Activity Controls Parental Contributions to the Early Embryo
(A) Histochemical staining forGUS activityfrom apaternallyinherited embryo marker (MET333),showing the typical stained and unstainedseeds as scored inthe
graphs in (B).
(B)Expressionof paternal markers, inwild-type Ler or kyp-2 maternal backgrounds scored asdescribed inFigure 3.See also FigureS3A,Table S3,and Table S4.
(C) Allele-specific RT-PCR of endogenous gene AGP18 paternal transcripts in siliques harvested 1–5 days after pollination (dap) inheriting a maternal kyp
mutation, as compared tothe wild-type. Selective amplificationof the paternal allele(p,top), amplification of maternal and paternal alleles (m+ p, middle), control
amplification of paternally expressed PHE1 mRNA (bottom).
(D–G) Allele-specific transcriptome profiling in isolated 2–4 cells embryos inheriting amaternal kyp-2 mutation compared to 2–4 cell wild-type embryos. Data and
legends are as in Figure 1.
712 Cell 145, 707–719, May 27, 2011 ª2011 Elsevier Inc.
silencing pathway (Brodersen and Voinnet, 2006), did not alter
the activation kinetics of the paternal reporters (Figure 5A).
Importantly, paternal inheritance of mutant KYP, CMT3, or
NRPD1b components showed no effect on paternal marker
expression (Figure S4D), confirming a specific maternal role for
the RdDM pathway in paternal marker regulation.
Several lines of evidence indicate that derepression of the
paternal reporters was notlinked to their transgenic nature. First,
we confirmed precocious detection of paternal transcripts for
the endogenous AGP18 locus in embryos inheriting a maternal
kyp mutation using allele-specific RT-PCR (Figure 4C). Second,
several transgenic reporters under the control of transposon
enhancers active in pollen (Slotkin et al., 2009) remained pater-
nally undetectable in embryos inheriting a maternal cmt3 muta-
tion (Figure S4E), whereas ectopic maternal activation of the
same reporters was reported in cmt3 embryo sacs (Pillot et al.,
2010a). Consistently, the profiling confirmed that the kyp muta-
tion did not massively derepress transposons and repeats in
the embryo (Table S1). Together with the genome-wide analysis
of over 3000 endogenous loci in kypm/KYPpembryos, these
results indicate that the maternally inherited components of the
RdDM pathway are involved in controlling the genome-wide
transcriptional dynamics of paternally inherited alleles.
Consistent with the proposed global role for the RdDM
pathway in transcriptional control in early embryos, we observed
that nrpd1a1b mutant zygotes showed an abnormally high level
of active PolII in their nucleus (Figure 5B). This was associated
with abnormal deposition of the repressive H3K9me2 marks
(Figure S5). The epigenetic and transcriptional states of zygotic
nuclei in this RdDM mutant is thus in stark contrast to wild-
type zygotes, which have a relatively quiescent transcriptional
state (Pillot et al., 2010b).
Surprisingly, despite their effect on gene expression, RdDM
mutants have not been reported to cause embryo lethality.
However, this does not exclude subtle defects and, indeed,
transient patterning defects were observed in embryos lacking
maternal CMT3 activity (Pillot et al., 2010b). Similarly, early kyp
embryos showed abnormal division planes in the embryo and
suspensor cells, suggesting a role in early embryonic patterning
Ovules Are Enriched in 24 nt siRNAs Targeting
Our results suggest a novel role for the RdDM pathway in the
regulation of genic regions, as this pathway had previously
been associated mostly with transposon and repeat silencing.
To verify the presence of maternal small RNAs targeting
protein-coding regions, we profiled a library of small RNAs
generated from manually dissected mature ovules before fertil-
the siRNAs had to be produced before fertilization. Our analysis
revealed a large fraction of siRNAs targeting genic regions
(comprising protein-coding sequences [CDS] and 500 bp of
putative 50regulatory regions of genes) in ovules as compared
to whole inflorescence (Lu et al., 2005) (Figures 6A and 6B).
This increase was not due to 21 nucleotide (nt) sRNAs (repre-
sented mainly by DCL1-dependent miRNAs and siRNAs) but
was associated with the 24 nt fraction, whose biogenesis is
dependent on PolIV, RDR2, and DCL3 (Brodersen and Voinnet,
2006). Overrepresentation of 24 nt siRNAs derived from CDS in
ovules was correlated with the transcriptional control of indi-
vidual loci mediated by maternal KYP activity in the embryo:
genes showing a transition from maternal expression in the
wild-type to biallelic or paternal expression in kypm/KYPp
embryos (kyp-responsive genes; Figure 6C) showed significantly
more matching siRNAs than genes that remained maternally ex-
pressed in kypm/KYPpembryos (kyp-unresponsive; Figure 6C).
This finding is consistent with a role of the 24 nt siRNAs in regu-
lating paternally inherited alleles.
Maternal CAF-1 and Histone H3 Turnover Regulate
Transcriptional Activation of Paternal Alleles
Mutant analyses showed that lack of maternal RdDM compo-
nents induced the precocious transcriptional activation of
paternal alleles. It is unknown whether paternal alleles are in
a state permissive to transcription or whether additional epige-
netic reprogramming events are necessary for their activation.
In the course of our genetic screen for mutants affecting paternal
reporter expression, we identified MULTIPLE SUPPRESSOR
OF IRA1 (MSI1) (Hennig et al., 2003) to be necessary for their
expression. In embryos inheriting a maternal msi1 mutation,
the activation of paternal reporter genes was markedly delayed
(Figures 7A and 7B and Figure S7A) compared to wild-type
embryos. Reduction of paternal transcript levels in msi1 mutants
was confirmed by RT-PCR for the endogenous GRP23 locus
MSI1 participates in several protein complexes including the
mitotic stability (Onoetal.,2006). CAF1isformedbyFASCIATA1
(FAS1), FAS2, and MSI1, and functions as an H3/H4-specific
chaperone facilitating nucleosome assembly during replication
(Hennig et al., 2005; Kaya et al., 2001). Consistently, maternal
loss of another subunit of the CAF1 complex, FAS2, showed
a similar effect as msi1 (Figure 7A and Figure S7B). MSI1 is
also a subunit of the MEA-FIE Polycomb group (PcG) complex
active in seeds (Ko ¨hler et al., 2003), but a mutation affecting
(D) Distribution of maternal versus paternal reads as in Figure 1A.
(E) q distribution as in Figure 1B. A total of 3125 genes were identified by informative reads in mutant embryos.
(F) Composite diagram representation of the gene distribution in (E) as in Figure 1C.
(G) Scatter plotdistribution of811biparental-class genescommonly detected inwild-type 2–4 cell embryos, globular embryos, and 2–4 cell kypm/KYPpembryos.
The difference in maternal contribution between stages (qglobular? q2–4 cell) or genotype (q2–4 cell WT? q2–4 cell kyp/KYP) is plotted on the x and y axis, respectively. q
values were calculated for each transcript as follows: maternal reads/maternal + paternal reads. Differences >0 mean a higher paternal contribution compared to
wild-type 2-4 cell embryos. Blue frame: genes with a correlated increased paternal contribution in both globular and kypm/KYPp2–4 cell embryos. Two groups of
genes are delineated according to their relative maternal contribution (q) in kypm/KYPpembryos as indicated (red, blue). Linear regressions: (i), y = 0.47x + 0.54 ;
R2= 0.17; (158 genes); and (ii), y = 0.41x + 0.02 ; R2= 0.20 (653 genes).
Related to Figure S1, Figures S3B–S3F, Table S1, and Table S2.
Cell 145, 707–719, May 27, 2011 ª2011 Elsevier Inc. 713
the MEDEA (MEA) subunit had no effect on paternal expression
at the globular stage (Figure S7C). These results strongly
suggest thatthe CAF1complex ismaternally required toactivate
transcription of the paternal genome, likely via histone turnover.
CAF1 may regulate the incorporation of specific histone variants
controlling transcriptional activity in plants, as it is the case
in animals. For example, in the animal germ line H3.3 variants
are incorporated at actively transcribed loci (Ooi et al., 2006).
Figure 5. A Functional RdDM Pathway Is
Required for Maternal Control of Paternal
Embryo Markers and Maintenance of Tran-
scriptional Quiescence in the Zygote
(A) Expression of paternal markers in embryos,
scored as in Figure 3 and Figure 4, in maternal
mutant backgrounds lacking activity of RdDM
components as indicated. See also Figure S4,
Figure S6, Table S3, Table S4, and Table S5.
(B) Immunolocalization of the active form of RNA
polymerase II (H5) in PolIV/PolV mutant zygotes
(nrpd1a1b) compared to the wild-type (WT). DNA
wascounterstained withDAPI. See alsoFigure S5.
A similar role may be proposed in plants,
because we observed that maternal
mutations affecting the H3.3 variants
HTR4 and HTR5 (Okada et al., 2005)
significantly delayed the activation of
paternal markers in the embryo (Fig-
ure S7D). These results indicate a role
for CAF1 and H3/H3.3 turnover in the
transcriptional activation of paternal—
We provided a genome-wide view of the
parental contributions to early embryo-
genesis in Arabidopsis. At early stages,
the maternal transcriptome clearly pre-
dominates: although 68% of the genes
arebiparentally expressed, their maternal
transcripts are overrepresented. Further,
>30% of the genes show an exclusively
maternal contribution. During the transi-
tionfromthe 2–4cellto the globularstage
the paternal contribution increases. We
showed that these parental contributions
are maternally controlled by two antago-
nistic regulatory pathways regulating the
onset of paternal and, at least partially,
maternal zygotic transcription. Maternal
dominance at early stages results from
downregulation of paternal alleles at
loci throughout the genome via the chro-
matin siRNA pathway, linked to RNA-
directed DNA and histone methylation.
In addition, transcriptional activation of
paternal alleles involves histone exchange, possibly via the
replacement of H3.3 variants, for which rapid turnover is
observed after fertilization (Ingouff et al., 2010). Release of
silencing might also involve passive DNA-demethylation during
mitoses or the activity of DNA- or histone-demethylases.
Although additional investigations are required to refine the
mechanistic role of these events in the control of the zygotic
genome, our results suggest that flowering plants evolved
714 Cell 145, 707–719, May 27, 2011 ª2011 Elsevier Inc.
Figure 6. Ovules Are Enriched in Small RNAs Targeting Genic Regions
(A) Deep sequencing of small RNA (sRNAs) from mature ovules reveals increased targeting to genic regions (comprising protein coding sequences (CDS) and
putative 50-regulatory regions 500 bp upstream of ATG) as compared to sRNAs from inflorescence (pie charts). Size distribution of CDS-targeted sRNAs showed
an increase in the 24 nt fraction in ovules as compared to inflorescence sRNA libraries (histogram).
(B) Comparative mapping of sRNA distribution in inflorescence (upper line) and ovules (lower line) libraries, exemplified by chromosome 1 (top), with a 400 kb
zoom(middle),and 40kbzoom(bottom). Boxesrepresent protein codingunits(genes). CDS-specific 24ntsRNAweredistributed between +and?strandsinthe
proportion of 63(+):37(?) in the ovule library.
(C) Average number of 24 nt siRNA per locus for maternal- and biparental-class genes showing no significant changes (kyp-unresponsive genes) or reduced q
values(increased paternalcontribution)inkypm/KYPpembryoscompared towild-type(kyp-responsivegenes). ttestpvalues(tableand graph,*p<0.01)refersto
differences in siRNA mean number between groups.
Cell 145, 707–719, May 27, 2011 ª2011 Elsevier Inc. 715
strategies to regulate early embryogenesis that are similar to
those described in animals (Baroux et al., 2008; Tadros and
Maternal Effect and Zygotic Functions during Early
The relative contribution and dynamic changes of parental tran-
scripts we observed in early Arabidopsis embryos may result
from a combination of de novo transcription postfertilization and
transcripts carried over from the egg or sperm, although their
respective abundance is unknown. 1331 and 621 genes with
respective maternal and paternal contributions (of 3399 and
1482 represented on the ATH1 microarray) were consistently de-
tected as present in egg- or sperm-specific microarray experi-
ments (Borges et al., 2008; Wuest et al., 2010). Although these
overlaps between pre- and postfertilization transcriptomes indi-
cate a carryover from egg and sperm cells, potentially influencing
early embryo development (Bayer etal., 2009;Pillotet al., 2010b),
the expression status of these genes in the early embryo remains
likely coexist during early embryogenesis (Tadros and Lipshitz,
2009), as it was shown by chromosomal deletions in Drosophila
pressed, the vast majority of transcripts are derived from the
maternalgenome, providing extensive maternalcontrol overearly
development, explaining the existence of numerous maternal
clear that there is a paternal contribution to early embryogenesis
from many loci, sometimes exclusively, consistent with paternal
effect and zygotic genes acting early after fertilization (Bayer
Figure 7. The Maternal CAF1 Nucleosome Assembly Complex Controls Paternal Gene Activation
(A) Paternal marker expression in msi1 or fas2 maternally mutant embryos compared to wild-type.
(B) Paternal marker expression scored as in Figure 3, Figure 4, and Figure 5, in msi1 or fas2 maternal background. See also Figure S7, Table S3, Table S4, and
(C) Allele-specific RT-PCR shows reduced endogenous GRP23 paternal transcript levels in siliques collected 2–5 dap inheriting a maternal msi1-2 mutation, as
compared to the wild-type. Selective amplification of the paternal allele (p), or of both maternal and paternal alleles (m + p), control amplification of paternally
716 Cell 145, 707–719, May 27, 2011 ª2011 Elsevier Inc.
Scholten et al., 2002; Tzafrir et al., 2004; Weijers et al., 2001).
EpigeneticPathways Controlling ParentalContributions
Are Distinct from Those Regulating Genomic Imprinting
In plants and mammals, certain genes are regulated by genomic
imprinting and are expressed monoallelically only from one
that for maternally expressed, imprinted loci, the RdDM pathway
repressing paternal alleles stays in place throughout develop-
ment, thus leading to monoallelic maternal expression. However,
the maternal regulatory pathways we uncovered complement—
and act beyond—the regulation of genes by genomic imprinting:
neither KYP nor CMT3 regulates the imprinted FIS2 locus (Jullien
et al., 2006b). Conversely, we determined that dcl3, a mutation
affecting the RdDM pathway, does not alter paternal silencing of
the imprinted MEA gene, nor do mutations affecting the CAF1
subunit FAS2 (Table S5). Instead, the MEA-FIE PcG complex
maintains silencing of the paternal MEA allele via H3K27 methyl-
ation (Baroux et al., 2006; Jullien et al., 2006a). This complex is
a mutation affecting the MSI1 PcG subunit did not show preco-
cious activation of paternal markers, but instead had a delaying
effect. Thus, the maternal mechanisms controlling the timing of
regulating genomic imprinting. Furthermore, although parental
contributions were described in the late endosperm (Hsieh
et al., 2011), their regulation at early stages still needs to be
elucidated at a genome-wide level. Possibly, different transcrip-
tional controls are in place because the endosperm is actively
transcribed soon after fertilization (Aw et al., 2010; Pillot et al.,
2010b), owing to its differential targeting by a global DNA
demethylase pathway that may counteracts the RdDM pathway
(Gehring et al., 2009; Hsieh et al., 2009).
The RdDM Pathway Plays a Role in Early Seed
We have shown that maternal mutations affecting the PolIV
subunit NRPD1a de-repressed the activity of paternal markers
and modified the transcriptional status of the zygotic genome.
Thus, we propose that PolIV-dependent 24 nt maternal siRNAs
epigenetically control the transcriptional status of paternal, and
possibly also maternal, loci throughout the genome. Whether
an epigenetic dimorphism establishes differential susceptibility
of the parental alleles to siRNA-based regulation is a challenging
question that remains to be addressed. Maternal siRNAs were
recently detected at later stages of seed development in the
endosperm (Mosher et al., 2009), and thus our findings extend
their role to early stages of embryogenesis, particularly in regu-
lating protein-coding sequences. The endosperm and its pro-
genitor, the central cell, or alternatively maternal sporophytic
tissue (Olmedo-Monfil et al., 2010) have recently been proposed
asa potential source of mobile maternal siRNAs drivingsilencing
in the egg cell and the embryo (reviewed, e.g., in Bourc’his and
Voinnet, 2010; Feng et al., 2010), an attractive hypothesis await-
Perturbation of the maternal RdDM pathway leads to transient
patterning defects as reported for embryos lacking maternal
CMT3 (Pillot et al., 2010b) or, as described here, KYP activity.
Thus, the pathway seems to fine-tune the expression of early
patterning genes. Alternatively, phenotypic aberrations might
be revealed in embryos inheriting a divergent paternal genome
distinct from the maternal background, which provides the
epigenetic control on zygotic genome expression. Further
studies are awaited to determine if RdDM mutants might repre-
sent sensitized backgrounds to hybridization and how these
maternal pathways affect out-breeding species. Maternal
siRNAs, particularly those targeting transposable elements,
havebeen proposed to actinheterosis and inter-specific hybrid-
ization (Chen, 2010; Martienssen, 2010). Twenty-four nucleotide
siRNAs targeting coding regions are also downregulated in
hybrid offspring (Groszmann et al., 2011). Our data suggest
that siRNA-based mechanisms also target protein-coding
sequences of the early embryonic genome, to control chro-
matin-based parental interactions during the epigenetic reprog-
ramming that occurs after fertilization.
Arabidopsis thaliana accessions Columbia-0 (Col), Landsberg erecta (Ler),
C24, WS or Nossen (No) were used as wild-type controls according to the
mutant’s background. The marker lines and mutants are listed in Table S3
and genotyping assays are described in Extended Experimental Procedures.
Embryonic cDNA Libraries, Sequencing, and Allele-Specific
The full method is described in Extended Experimental Procedures. In brief,
embryos were released from the seeds by gentle pressure and isolated under
an inverted microscope using a microcapillary. Total RNA was extracted using
the PicoPure RNA Isolation Kit (Arcturus) and 300–700 pg was amplified in
a linear fashion using the WT-Ovation Pico RNA Amplification System (NuGEN
Technologies). Aftersecond strandcDNA synthesisandlibrarypreparation, 50
bases sequence reads were generated by SOLiD v3 (Applied Biosystems) and
aligned to the Arabidopsis Col genome (TAIR8.0). Unique reads mapping full
length in the exome were sorted for the presence of Ler annotated polymor-
phisms (Borevitz et al., 2007).
model fitting best the observed distribution of maternal and paternal reads per
transcripts. A detailed explanation is provided in Extended Experimental
Marker Gene Analysis
The activity of paternal markers following crosses to wild-type or mutant
females was assayed by histochemical staining of the uidA reporter gene
product (the GUS enzyme) as described in Extended Experimental
Procedures. The number of GUS-positive progeny was scored for each devel-
opmental stage anddifferenceswereassessedusingtwo-tailed Fisher’s exact
test. For allele-specific RT-PCR, LNA-modified primers targeting a SNP from
one parental transcript were used. Details on the reactions and primer
sequences are provided in Extended Experimental Procedures.
Profiling of Small RNAs in Ovules
In brief, total RNA was extracted from dissected mature ovules (Peiffer et al.,
2008) and a small RNA library was prepared and sequenced using Illumina
Genome Analyzer. After filtering, reads were mapped against Arabidopsis
Col reference sequence (TAIR 8) and compared to inflorescence small RNA
reads (Lu et al., 2005), analyzed using the same procedure. Mapping, occur-
renceinformation, normalization,and graphical displays werecomputed using
R. Genic regions and repeat target analysis was done using TAIR 8 genome
release and ftp://ftpmips.helmholtz-muenchen.de/plants/cress/.
Cell 145, 707–719, May 27, 2011 ª2011 Elsevier Inc. 717
Immunodetection was performed essentially as described (Pillot et al., 2010b),
using antibodies from Abcam: anti-H3K9me2 (#ab1220) and anti-phosphoS2
RNA PolII [H5] (#ab24758) specific to the active form of PolII (Palancade and
Bensaude, 2003). Images were captured on a laser scanning confocal micro-
scope (Leica SP2) and maximum-intensity projections of selected optical
Data from embryo SOLiD profiling are accessible under GEO accession
number GSE24198. Data from ovule small RNA profiling are accessible under
GEO accession number GSE28627. Sequence data from this article can be
the following accession numbers: At5g13960 (KRYPTONITE), At5g14620
(DRM2), At5g15380(DRM1), At1g69770
At5g66750 (DDM1), At4g11130 (RDR2), At3g43920 (DCL3), At3g03300
(DCL2), At2g27040 (AGO4), At1g63020 (NRPD1A/POLIVA), At2g40030
(NRPD1b/NRPE1/POLV), At5g58230 (MSI1), At5g64630 (FAS1), At5g64630
(FAS2), At1g02580(MEDEA), At4g40030
At3g11940 (RPS5a), At2g241500 (LACHESIS), At1g10270 (GRP23).
figures, five tables, and one movie and can be found with this article online at
We thank the Arabidopsis Stock Centers (NASC/ABRC), J. Carrington, R.
Gross-Hardt, R. Martienssen, F. Meins, T. Lagrange, J. Paszowski, and E. Ri-
chards for seeds; C. Michaud, N. Duran-Figueroa, and V. Olmedo-Monfil for
technical support; T. Nawy for advice on embryo microdissection; S.
Schreiber, L. Bossen, and M. Schilhabel for sequencing support at Kiel; and
M. Schmid for help in GEO data submission. We are grateful to M. Arteaga-
Vazquez,O.Hamant,and T.Lagrange forhelpfulcommentsonearlier versions
of the manuscript and to S. Kessler and S. Gillmor for critical reading. C.B.,
M.R., S.G., and A.S. were supported by the University of Zu ¨rich and grants
of Swiss National Science Foundation and the European Union (FP5 ApoTool
Project) to U.G. D.A., O.L., and D.G. were supported by the Institut de
Recherche pour le De ´veloppement and the Agence Nationale de la
Recherche, and T.L. by CNRS and a starting grant of the European Research
Council. P.R. is supported by the Clusters of Excellence: ‘‘The Future Ocean’’
and ‘‘Inflammation at Interfaces.’’ J.-P.V.-C. is supported by CONACyT and is
an International Scholar of the Howard Hughes Medical Institute.
Received: January 21, 2011
Revised: March 28, 2011
Accepted: April 15, 2011
Published: May 26, 2011
Ande ´ol,Y. (1994). Early transcriptionindifferentanimal species:implication for
transition from maternal to zygotic control in development. Roux. Arch. Dev.
Biol. 204, 3–10.
Aw,S.J., Hamamura, Y.,Chen,Z., Schnittger, A.,and Berger, F. (2010). Sperm
entry is sufficient to trigger division of the central cell but the paternal genome
is required for endosperm development in Arabidopsis. Development 137,
Baroux, C., Autran, D., Gillmor, C.S., Grimanelli, D., and Grossniklaus, U.
(2008). The maternal to zygotic transition in animals and plants. Cold Spring
Harb. Symp. Quant. Biol. 73, 89–100.
Baroux, C., Blanvillain, R., and Gallois, P. (2001). Paternally inherited trans-
genes are down-regulated but retain low activity during early embryogenesis
in Arabidopsis. FEBS Lett. 509, 11–16.
Baroux, C., Gagliardini, V., Page, D.R., and Grossniklaus, U. (2006). Dynamic
regulatory interactions of Polycomb group genes: MEDEA autoregulation is
required for imprinted gene expression in Arabidopsis. Genes Dev. 20,
Bayer, M., Nawy, T., Giglione, C., Galli, M., Meinnel, T., and Lukowitz, W.
(2009). Paternal control of embryonic patterning in Arabidopsis thaliana.
Science 323, 1485–1488.
Borevitz, J.O., Hazen, S.P., Michael, T.P., Morris, G.P., Baxter, I.R., Hu, T.T.,
Chen, H., Werner, J.D., Nordborg, M., Salt, D.E., et al. (2007). Genome-wide
patterns of single-feature polymorphisms in Arabidopsis thaliana. Proc. Natl.
Acad. Sci. USA 104, 12057–12062.
Borges, F., Gomes, G., Gardner, R., Moreno, N., McCormick, S., Feijo, J.A.,
and Becker, J.D. (2008). Comparative transcriptomics of Arabidopsis sperm
cells. Plant Physiol. 148, 1168–1181.
esis, fertilization, and early zygotic development. Science 330, 617–622.
Brodersen, P.,and Voinnet,O.(2006). The diversity of RNA silencingpathways
in plants. Trends Genet. 22, 268–280.
Bushell, C., Spielman, M., and Scott, R.J. (2003). The basis of natural and arti-
ficial postzygotic hybridization barriers in Arabidopsis species. Plant Cell 15,
Chaudhury, A.M., and Berger, F. (2001). Maternal control of seed develop-
ment. Semin. Cell Dev. Biol. 12, 381–386.
Chen, Z.J. (2010). Molecular mechanisms of polyploidy and hybrid vigor.
Trends Plant Sci. 15, 57–71.
De Renzis, S., Elemento, O., Tavazoie, S., and Wieschaus, E.F. (2007).
Unmasking activation of the zygotic genome using chromosomal deletions
in the Drosophila embryo. PLoS Biol. 5, e117.
Feng, S., Jacobsen, S.E., and Reik, W. (2010). Epigenetic reprogramming in
plant and animal development. Science 330, 622–627.
Gehring, M., Bubb, K.L., and Henikoff, S. (2009). Extensive demethylation of
repetitive elements during seed development underlies gene imprinting.
Science 324, 1447–1451.
Grimanelli, D., Perotti, E., Ramirez, J., and Leblanc, O. (2005). Timing of the
maternal-to-zygotic transition during early seed development in maize. Plant
Cell 17, 1061–1072.
Groszmann, M., Greaves, I.K., Albertyn, Z.I., Scofield, G.N., Peacock, W.J.,
and Dennis, E.S. (2011). Changes in 24-nt siRNA levels in Arabidopsis hybrids
suggest an epigenetic contribution to hybrid vigor. Proc. Natl. Acad. Sci. USA
Hennig, L., Bouveret, R., and Gruissem, W. (2005). MSI1-like proteins: an
escort service for chromatin assembly and remodeling complexes. Trends
Cell Biol. 15, 295–302.
Hennig, L., Taranto, P., Walser, M., Schonrock, N., and Gruissem, W. (2003).
Arabidopsis MSI1 is required for epigenetic maintenance of reproductive
development. Development 130, 2555–2565.
Hsieh, T.F., Ibarra, C.A., Silva, P., Zemach, A., Eshed-Williams, L., Fischer,
R.L., and Zilberman, D. (2009). Genome-wide demethylation of Arabidopsis
endosperm. Science 324, 1451–1454.
Hsieh, T.F., Shin, J., Uzawa, R., Silva, P., Cohen, S., Bauer, M.J., Hashimoto,
M., Kirkbride, R.C., Harada, J.J., Zilberman, D., et al. (2011). Regulation of im-
Ingouff, M., Rademacher, S., Holec, S., Soljic, L., Xin, N., Readshaw, A., Foo,
S.H., Lahouze, B., Sprunck, S., and Berger, F. (2010). Zygotic resetting of the
HISTONE 3 variant repertoire participates in epigenetic reprogramming in
Arabidopsis. Curr. Biol. 20, 2137–2143.
718 Cell 145, 707–719, May 27, 2011 ª2011 Elsevier Inc.
Jackson, J.P., Lindroth, A.M., Cao, X., and Jacobsen, S.E. (2002). Control of
CpNpG DNA methylation by the KRYPTONITE histone H3 methyltransferase.
Nature 416, 556–560.
Jahnke, S.,Sarholz, B.,Thiemann,A.,Kuhr,V.,Gutierrez-Marcos,J.F.,Geiger,
H.H., Piepho, H.P., and Scholten, S. (2010). Heterosis in early seed develop-
ment: a comparative study of F1 embryo and endosperm tissues 6 days after
fertilization. Theor. Appl. Genet. 120, 389–400.
Jullien, P.E., Katz, A., Oliva, M., Ohad, N., and Berger, F. (2006a). Polycomb
in Arabidopsis. Curr. Biol. 16, 486–492.
Jullien, P.E., Kinoshita, T., Ohad, N., and Berger, F. (2006b). Maintenance of
DNA methylation during the Arabidopsis life cycle is essential for parental
imprinting. Plant Cell 18, 1360–1372.
Kaya, H., Shibahara, K.I., Taoka, K.I., Iwabuchi, M., Stillman, B., and Araki, T.
(2001). FASCIATA genes for Chromatin Assembly Factor-1 in Arabidopsis
maintain the cellular organization of apical meristems. Cell 104, 131–142.
Ko ¨hler, C., Hennig, L., Bouveret, R., Gheyselinck, J., Grossniklaus, U., and
Gruissem, W. (2003). Arabidopsis MSI1 is a component of the MEA/FIE Poly-
comb group complex and required for seed development. EMBO J. 22, 4804–
Lu, C., Tej, S.S., Luo, S., Haudenschild, C.D., Meyers, B.C., and Green, P.J.
(2005). Elucidation of the small RNA component of the transcriptome. Science
Martienssen, R.A. (2010). Heterochromatin, small RNA and post-fertilization
dysgenesis in allopolyploid and interploid hybrids of Arabidopsis. New Phytol.
Meyer, S., and Scholten, S. (2007). Equivalent parental contribution to early
plant zygotic development. Curr. Biol. 17, 1686–1691.
Mosher, R.A., Melnyk, C.W., Kelly, K.A., Dunn, R.M., Studholme, D.J., and
Baulcombe, D.C. (2009). Uniparental expression of PolIV-dependent siRNAs
in developing endosperm of Arabidopsis. Nature 460, 283–286.
Ngo, Q.A., Moore, J.M., Baskar, R., Grossniklaus, U., and Sundaresan, V.
(2007). Arabidopsis GLAUCE promotes fertilization-independent endosperm
development and expression of paternally inherited alleles. Development
Okada, T., Endo, M., Singh, M.B., and Bhalla, P.L. (2005). Analysis of the
histone H3 gene family in Arabidopsis and identification of the male-gamete-
specific variant AtMGH3. Plant J. 44, 557–568.
Olmedo-Monfil, V., Duran-Figueroa, N., Arteaga-Vazquez, M., Demesa-Are-
valo, E., Autran, D., Grimanelli, D., Slotkin, R.K., Martienssen, R.A., and
Vielle-Calzada, J.P. (2010). Control of female gamete formation by a small
RNA pathway in Arabidopsis. Nature 464, 628–632.
Ono, T., Kaya, H., Takeda, S., Abe, M., Ogawa, Y., Kato, M., Kakutani, T.,
Mittelsten Scheid, O., Araki, T., and Shibahara, K. (2006). Chromatin Assembly
Factor 1 ensures the stable maintenance of silent chromatin states in Arabi-
dopsis. Genes Cells 11, 153–162.
Ooi, S.L., Priess, J.R., and Henikoff, S. (2006). Histone H3.3 variant dynamics
in the germline of Caenorhabditis elegans. PLoS Genet. 2, e97.
Pahlavani, M.H., and Abolhasani, K. (2006). Xenia effect on seed and embryo
size in cotton (Gossypium hirsutum L.). J. Appl. Genet. 47, 331–335.
Palancade, B., and Bensaude, O. (2003). Investigating RNA polymerase II
carboxyl-terminal domain (CTD) phosphorylation. Eur. J. Biochem. 270,
Peiffer, J.A., Kaushik, S., Sakai, H., Arteaga-Vazquez, M., Sanchez-Leon, N.,
Ghazal, H., Vielle-Calzada, J.P., and Meyers, B.C. (2008). A spatial dissection
of the Arabidopsis floral transcriptome by MPSS. BMC Plant Biol. 8, 43.
Pillot, M., Autran, D., Leblanc, O., and Grimanelli, D. (2010a). A role for
CHROMOMETHYLASE3 in mediating transposon and euchromatin silencing
during egg cell reprogramming in Arabidopsis. Plant Signal. Behav. 5, 1167–
Pillot, M., Baroux, C., Vazquez, M.A., Autran, D., Leblanc, O., Vielle-Calzada,
J.P., Grossniklaus, U., and Grimanelli, D. (2010b). Embryo and endosperm
inherit distinct chromatin and transcriptional states from the female gametes
in Arabidopsis. Plant Cell 22, 307–320.
Raissig, M.T., Baroux, C., and Grossniklaus, U. (2011). Regulation and flexi-
bility of genomic imprinting during seed development. Plant Cell 23, 16–26.
Ronceret, A., Guilleminot, J., Lincker, F., Gadea-Vacas, J., Delorme, V.,
Bechtold, N., Pelletier, G., Delseny, M., Chaboute, M.E., and Devic, M.
(2005). Genetic analysis of two Arabidopsis DNA polymerase epsilon subunits
during early embryogenesis. Plant J. 44, 223–236.
Ronceret, A., Gadea-Vacas, J., Guilleminot, J., Lincker, F., Delorme, V.,
Lahmy, S.,Pelletier, G.,Chaboute, M.E., and Devic, M. (2008). The first zygotic
division in Arabidopsis requires de novo transcription of thymidylate kinase.
Plant J. 53, 776–789.
Scholten, S., Lo ¨rz, H., and Kranz, E. (2002). Paternal mRNA and protein
synthesis coincides with male chromatin decondensation in maize zygotes.
Plant J. 32, 221–231.
Slotkin, R.K., Vaughn, M., Borges, F., Tanurdzic, M., Becker, J.D., Feijo, J.A.,
and Martienssen, R.A. (2009). Epigenetic reprogramming and small RNA
silencing of transposable elements in pollen. Cell 136, 461–472.
Tadros, W., and Lipshitz, H.D. (2009). The maternal-to-zygotic transition:
a play in two acts. Development 136, 3033–3042.
Tzafrir, I., Pena-Muralla, R.,Dickerman, A., Berg, M., Rogers,R., Hutchens, S.,
Sweeney, T.C., McElver, J., Aux, G., Patton, D., et al. (2004). Identification of
genes required for embryo development in Arabidopsis. Plant Physiol. 135,
Vielle-Calzada, J.P., Baskar, R., and Grossniklaus, U. (2000). Delayed activa-
tion of the paternal genome during seed development. Nature 404, 91–94.
Weijers, D., Geldner, N., Offringa, R., and Ju ¨rgens, G. (2001). Seed develop-
ment: early paternal gene activity in Arabidopsis. Nature 414, 709–710.
Wuest, S.E., Vijverberg, K., Schmidt, A., Weiss, M., Gheyselinck, J., Lohr, M.,
Wellmer, F., Rahnenfuhrer, J., von Mering, C., and Grossniklaus, U. (2010).
Arabidopsis female gametophyte gene expression map reveals similarities
between plant and animal gametes. Curr. Biol. 20, 506–512.
Cell 145, 707–719, May 27, 2011 ª2011 Elsevier Inc. 719