LETTER
doi:10.1038/nature13620
Non-equivalent contributions of maternal and
paternal genomes to early plant embryogenesis
Gerardo Del Toro-De Leo ´n
1
, Marcelina Garcı ´a-Aguilar
1
& C. Stewart Gillmor
1
Zygotic genome activation in metazoans typically occurs several
hours to a day after fertilization, and thus maternal RNAs and pro-
teins drive early animal embryo development
1
. In plants, despite sev-
eral molecular studies of post-fertilization transcriptional activation,
the timing of zygotic genome activation remains a matter of debate.
For example, two recent reports that used different hybrid ecotype
combinations for RNA sequence profiling of early Arabidopsis em-
bryo transcriptomes came to divergent conclusions. One identified
paternal contributions that varied by gene, but with overall maternal
dominance
2
, while the other found that the maternal and paternal
genomes are transcriptionally equivalent
3
. Here we assess paternal
gene activation functionally in an isogenic background, by perform-
ing a large-scale genetic analysis of 49 EMBRYO DEFECTIVE genes
and testing the ability of wild-type paternal alleles to complement
phenotypes conditioned by mutant maternal alleles. Our results dem-
onstrate that wild-type paternal alleles for nine of these genes are com-
pletely functional 2 days after pollination, with the remaining 40 genes
showing partial activity beginning at 2, 3 or 5 days after pollination.
Using our functional assay, we also demonstrate that different hybrid
combinations exhibit significant variation in paternal allele activation,
reconciling the apparently contradictory results of previous transcrip-
tional studies
2,3
. The variation in timing of gene function that we observe
confirms that paternal genome activation does not occur in one early
discrete step, provides large-scale functional evidence that maternal
and paternal genomes make non-equivalent contributions to early
plant embryogenesis, and uncovers an unexpectedly profound effect
of hybrid genetic backgrounds on paternal gene activity.
In animals, maternal RNAs and proteins drive early embryo develop-
ment. The transition to zygotic control requires both clearing of maternal
transcripts and zygotic genome activation, a phenomenon collectively
known as the maternal-to-zygotic transition
1
. Research in plants has
focused almost exclusively on zygotic genome activation, and has led to
differing conclusions on its timing. Studies in maize and in tobacco sug-
gested that large-scale zygotic transcription does not begin until 3–4
days after fertilization
4,5
. In agreement with this, an Arabidopsis RNA
sequence profiling and reporter line study
2
and three smaller-scale mole-
cular studies
6–8
also came to the conclusion that many paternal alleles are
either quiescent or transcribed at lower levels than corresponding mater-
nal alleles in the first few days after fertilization.
Other work has suggested a contrary view, namely that zygotic tran-
scription begins immediately or soon after fertilization. Significant dif-
ferences have been found between the transcript populations of egg cells
and two-celled embryos in wheat
9
, and equivalent maternal and pater-
nal expression for 25 different genes was observed in maize zygotes at
1 day after pollination
10
. In Arabidopsis, three genes have been shown by
genetic means to be expressed from the paternal allele in early embryo-
genesis
11–13
, and a recent genome-wide transcriptional study found
biparental transcription for nearly all of the ,8,000 genes that could be
assayed at the one- to two-cell zygote stage
3
. Even taking into account that
timing of zygotic genome activation may differ among plant species, it
has been difficult to reconcile these conflicting results. However, when
evaluating the two RNA profiling experiments arguing for
3
or against
2
immediate zygotic genome activation in Arabidopsis, it is important to
note that each study used a unique combination of polymorphic strains
to distinguish maternal and paternal transcripts, such that the results
are based on gene expression in embryos hybrid for different combina-
tions of ecotypes.
To study paternal allele activation in an isogenic background, we used
a genetic assay with strict functional criteria: a test of the ability of wild-
type (WT) paternal alleles to complement early embryo mutant pheno-
types conditioned by mutant maternal alleles. While this assay is necessarily
limited in the number of genes that can be tested (and in the corresponding
conclusions that can be drawn), it presents two significant advantages
over RNA sequencing approaches: by design, we only evaluate genes with
a proven role in embryo development, leading to biologically relevant
results. Second, we avoid the complex effects on quantitative traits—such
as gene expression values—that necessarily result from crossing poly-
morphic, distantly related ecotypes (the basis for allele specific profil-
ing by RNA sequencing)
14–17
.
We screened a collection of mutants in several hundred EMBRYO
DEFECTIVE (EMB) genes required for early embryo development in
Arabidopsis
18,19
, and selected genes that were associated with clearly
recognizable anatomical defects segregating in emb/1 plants at pre-
globular stages (Supplementary Table 1 and Supplementary Data Set 1).
For 49 different EMB genes, we then crossed heterozygous mutant plants
with WT pollen, and scored the resulting embryos for mutant pheno-
types at 2, 3, 5 and ,14 days after pollination (Fig. 1; complete data shown
in Supplementary Table 2). We identified nine genes where the WT pater-
nal allele complemented the mutant maternal allele as soon as there was a
requirement for the gene product, resulting in no observed mutant phe-
notypes at 2 days after pollination or later (AtSWI3A shown as an example
in Fig. 1d–f, j). For the remaining 40 genes, the WT paternal allele gradual-
ly complemented the mutant phenotype conferred by the mutant mater-
nal allele (EMB2804 shown as an example in Fig. 1g–i, j). To determine
more precisely the time point of the onset of paternal allele activation,
we studied a subset of 13 genes in more detail (Supplementary Table 3).
Comparison of mutant ratios resulting from hand-pollinated emb/13
emb/1 and emb/13 Col crosses identified genes whose activation began
at either 2, 3 or 5 days after pollination. By contrast, Col 3 emb/1 control
crosses resulted in no mutant phenotypes, showing that the delayed com-
plementation we observed does not reflect haploinsufficiency expressed
only in the first few days of embryogenesis, and that the maternal allele
for these genes is absolutely required in early embryogenesis, while the
paternal allele is not (Supplementary Table 3).
We next sought to correlate function and expression, selecting the
GCT and CCT genes as examples
20
. Figure 2a–g, i–o shows that mater-
nally derived gGCT-GUS and pCCT::GUS reporter constructs drove
embryo expression at 2 days after pollination, whereas paternally derived
constructs drove embryo expression beginning at 3 days after pollina-
tion, and reached full expression at 4 days after pollination. Phenotypic
complementation of gct and cct mutant maternal alleles by the WT
paternal allele began at 5 days after pollination (Fig. 2h, p), 1 day after
1
Laboratorio Nacional de Geno ´ mica para la Biodiversidad (Langebio), Unidad de Geno ´ mica Avanzada, Centro de Investigacio ´ n y de Estudios Avanzados del Instituto Polite ´ cnico Nacional (CINVESTAV-IPN),
Irapuato, Guanajuato 36821, Me ´ xico.
624 | NATURE | VOL 514 | 30 OCTOBER 2014
Macmillan Publishers Limited. All rights reserved ©2014