DISCUSSION
Embryo-endosperm relationship is one of the determining factors
contributing to seed development and subsequently the yield and quality
of cereal grains. However, signals or regulators that coordinate the
developmental processes of the two compartments remain largely unknown.
This study took advantages of the NB mutant to evolve a novel comparison
method for quantifying the influence of embryo on endosperm development,
revealing that the embryo has a dragging effect on the developmental
transition of the endosperm. To our knowledge, it is the first attempt
to directly disclose the evidence of embryo-endosperm interaction during
rice seed development. The findings reveal new aspects of the role of
embryo in the formation process of rice quality, and help to draw an
integrative picture of seed development from the perspective of agronomy
and crop physiology.
1. The dragging effect of embryo on endosperm development
Seed development is an orchestrated progression through a series of
stages, which can be described by two types of ‘time’: the chronological
time and the developmental time defined by the sequence of stages
(Ebisuya & Briscoe, 2018). There is evidence of bidirectional
interaction between embryo and endosperm throughout development (An et
al., 2020). Notably, two recent studies in Arabidopsis suggest an
independent relationship between the two tissues. By analyzing mutants
with defective endosperm cellularization,
O’Neill, Colon, and Jenik (2019)
found that this endosperm process is not required for the onset of
embryo maturation. Using single-fertilization mutants, Xiong, Wang, and
Sun (2021) demonstrated that in the absence of embryo, endosperm
develops the same way as in the wild type, suggesting that endosperm
development is an autonomously programmed process independent of
embryogenesis. These two studies in combination indicate that in terms
of developmental time, the mechanisms controlling endosperm and embryo
development act independently of each other (O’Neill et al., 2019). On
the contrary, in terms of chronological time, the current study show
that duration of endosperm filling in the bottom part of the NB grains
was prolonged by the embryo (Figure 2j), suggesting a substantial
interaction between embryo and endosperm. Similarly, Xiong et al. (2021)
showed that rapid embryo expansion can significantly accelerate
endosperm breakdown, thus shortening the lifespan of the transient
endosperm. Further, it should be noted that cereal seeds like rice have
a persistent endosperm, whereas dicots like Arabidopsis have an
ephemeral endosperm that degrades as the embryo grows. As a result,
whether the independence of endosperm development from embryo in
Arabidopsis is applicable to rice awaits further investigation.
Plants have evolved a variety of timing mechanisms that integrate
chronological time with developmental time to ensure proper development
(O’Neill et al., 2019). For the endosperm, these include internal timers
of molecular oscillators based on hormones or metabolites, and external
timers dependent on environmental signals or emanating from a different
tissue like the embryo. This study reveals a dragging effect of embryo
on endosperm development in chronological time, i.e. extending the
storage accumulating stage whereas delaying the maturation stage. This
finding provides direct evidence for the role of embryo as the external
timer controlling endosperm development. In addition, hormones like GA,
auxins, and ABA were unevenly distributed in rice seed, with the embryo
generally having higher contents at the early and middle stages, as also
reported by Zhang et al. (2020a). Interestingly, GA20ox(GA20ox2 and GA20ox3 ), GA3ox (OsGA3ox1 andOsGA3ox2 ), KS (OsKS1 and OsKS4 ) were
predominantly expressed in embryo at 5-20 DAF, indicating that embryo
may be the GA-synthetic site whereas endosperm is the GA-acting site
(Figure 4f). It is well established that embryo-derived GA modulates the
secretion of starch-degrading enzymes like α-amylase from the aleurone
and scutellum upon germination. But it is still uncertain whether the
degradation of starch during seed development is analogous to the
germinating process. Our results imply that one of the external timers
coordinating rice grain development might be the hormone GA released
from endosperm, which needs further investigation.
For internal timers modifying endosperm development, the T6P-SnRK1
signaling pathway may be the key component. At early stage of 5 to 10
DAF, sucrose in endosperm was lower, probably due to the deprivation by
the growing embryo. In response to the reduced sucrose content, T6P was
decreased simultaneously in endosperm, thus relieving the inhibition of
SnRK1 activity. The increased SnRK1 activities promoted the catabolism
or suppressed the anabolism of starch and proteins, as reflected by the
lower content of starch and prolamins as well as the enhanced gene
activity of amylase, lipase, and protease in endosperm (Figure 7).
Conversely, at middle stage between 20 and 25 DAF, the T6P-SnRK1
signaling showed an opposite trend relative to that between 5 and 10
DAF, indicating it may be involved in the transition of developmental
stages in endosperm (Figure 7). Taken together, the anticorrelation
between T6P and SnRK1 activity opens the possibility that the T6P-SnRK1
pathway may be a master regulator coordinating the communication between
embryo and endosperm during rice grain formation.