Evidence for maternal control of seed size in maize from phenotypic and transcriptional analysis
Journal of Experimental Botany
Evidence for maternal control of seed size in maize from phenotypic and transcriptional analysis
Xia Zhang 2
Candice N. Hirsch 1
Rajandeep S. Sekhon 0
Natalia de Leon 2 3
Shawn M. Kaeppler 2 3
Cristobal Uauy, John Innes Centre
0 Department of Genetics and Biochemistry, Clemson University , Clemson, SC 29634 , USA
1 Department of Agronomy and Plant Genetics, University of Minnesota , St. Paul, MN 55108 , USA
2 Department of Agronomy, University of Wisconsin-Madison , 1575 Linden Drive, Madison, WI 53706 , USA
3 DOE Great Lakes Bioenergy Research Center, University of Wisconsin-Madison , 1575 Linden Drive, Madison, WI 53706 , USA
Seed size is an important component of grain yield and a key determinant trait for crop domestication. The Krug Yellow Dent long-term selection experiment for large and small seed provides a valuable resource to dissect genetic and phenotypic changes affecting seed size within a common genetic background. In this study, inbred lines derived from Krug Large Seed (KLS) and Krug Small Seed (KSS) populations and reciprocal F1 crosses were used to investigate developmental and molecular mechanisms governing seed size. Seed morphological characteristics showed striking differences between KLS and KSS inbred lines, and the reciprocal cross experiment revealed a strong maternal influence on both seed weight and seed size. Quantification of endosperm area, starchy endosperm cell size, and kernel dry mass accumulation indicated a positive correlation between seed size, endosperm cell number, and grain filling rate, and patterns of grain filling in reciprocal crosses mirrored that of the maternal parent. Consistent with the maternal contribution to seed weight, transcriptome profiling of reciprocal F1 hybrids showed substantial similarities to the maternal parent. A set of differentially expressed genes between KLS and KSS inbreds were found, which fell into a broad number of functional categories including DNA methylation, nucleosome assembly, and heat stress response. In addition, gene co-expression network analysis of parental inbreds and reciprocal F1 hybrids identified co-expression modules enriched in ovule development and DNA methylation, implicating these two processes in seed size determination. These results expand our understanding of seed size regulation and help to uncover the developmental and molecular basis underlying maternal control of seed size in maize.
Endosperm; gene expression; maize; maternal effect; seed development; seed size
Seed size is a key determinant for evolutionary fitness and is
also a crucial agronomic trait selected during crop
(Doebley et al., 2006)
. Seeds produced by cereal crops
are a major source of staple food, livestock feed, and
. Seed size has been proposed to be a
key contributor to grain yield in crop plants
(Kesavan et al.,
2013; Zhang et al., 2014)
. In maize (Zea mays L.) breeding
programs, seed size is an important breeding target because
of the requirements of both end-use quality and consumer
(Gupta et al., 2006)
. Understanding determinants
of seed size is, therefore, essential to meet increasing demand
for food staples and renewable energy by the ever-growing
Seeds in angiosperms consist of three genetically distinct
constituents: the embryo, the endosperm, and the seed coat.
Seed development begins with a double fertilization event in
which one sperm nucleus fuses with the haploid egg and
produces the diploid embryo, and the other sperm nucleus
fertilizes with the diploid central cell to give rise to the triploid
endosperm, which is responsible for the transfer of nutrients
to the embryo. The embryo and endosperm develop within
the maternal tissues of the ovule, and the integuments of
the ovule ultimately give rise to the coat of the mature seed
(Chaudhury et al., 2001)
. Seed size is co-ordinately
determined by the growth of the triploid endosperm, the diploid
embryo, and resources and developmental cues provided by
the maternal plant
In monocots, the endosperm constitutes the majority of
the mature seed, and endosperm size has been found to play
a major role in determining seed size
(Berger et al., 2006)
Seed size frequently depends on the development and the
amount/size of the endosperm, and a relationship between
endosperm cell number and seed size has been observed
(Chojkecki et al., 1986)
. Final seed size and weight are
influenced by a number of cellular processes, as well as genetic and
environmental factors. Genetic factors that regulate seed size
zygotically or maternally have been identified in Arabidopsis
as well as in crop plants
(reviewed by Kesavan et al., 2013;
Li and Li, 2015)
. Epigenetic marks in the genome are also
important factors affecting seed size, and genomic
imprinting, primarily conveyed by DNA methylation, has been
proposed as another important phenomenon affecting seed size
(Gehring et al., 2004; Jiang and Köhler, 2012; Fatihi et al.,
The maternal parent contributes to the offspring seed
phenotype in multiple ways including providing
photosynthate and nutrients to support development, co-ordinating
developmental timing, as well as imprinting of maternal
gametes. The maternal plant affects seed size via (i) the seed
coat, which comprises the maternal genotype and imposes
mechanical constraints on seed development; (ii) maternal
provisioning during seed development; (iii) maternal
determination of progeny plasticity in response to developmental
signals and environmental cues; and (iv) the effect of the triploid
endosperm where gene imprinting occurs most often (Roach
and Wulff, 1987;
Platenkamp and Shaw, 1993
2009; Donohue, 2009
Fang et al., 2012
nutrient allocation is important for seed development. Highly
specialized maize cells in the basal endosperm transfer cell
layer facilitate the transport of maternal solutes and nutrients
at the interface between maternal tissues and the endosperm
Gómez et al., 2002
). Zea mays MYB-related protein-1
(ZmMRP-1) is a gene encoding a known endosperm transfer
cell-specific transcriptional activator, which is involved in the
expression of basal endosperm transfer layer (BETL)-specific
genes including BETL-1, Basal layer-type antifungal protein 2
(BAP2), and Maternally expressed gene 1 (Meg1), and plays
a central role in the regulatory pathways controlling transfer
cell differentiation and associated maternal nutrition
allocation (Gómez et al.,
2009; Costa et al., 2012
Xiong et al.,
Although a few known maize genes acting in endosperm
and maternal tissues to affect seed development have been
identified such as miniature 1 (mn1) and shrunken-2 (sh2)
(Miller and Chourey, 1992; Hannah et al., 2012)
little is known about the maternal genetic factors and molecular
mechanisms that regulate seed size in maize. Previous studies
have documented developmental timing contributions and
possible genetic regions under selection in the Krug Yellow
Dent long-term selection experiment for small and large seed
(Hirsch et al., 2014; Sekhon et al., 2014)
. In this study, we
used inbred lines derived from the Krug Large Seed (KLS)
and Krug Small Seed (KSS) populations and their reciprocal
F1 hybrids to explore maternal determinants underlying seed
size regulation, and presented evidence that the maternal
parent plays an important role in determining seed size via seed
morphological, cytological, and transcriptional analyses.
Materials and methods
Plant materials, growth conditions, and sampling details
Thirty cycles of divergent mass selection for seed size were
performed in the open pollinated population Krug Yellow Dent to
generate KLS30 (selected for large seed size; PI 636488) and KSS30
(selected for small seed size; PI 636489)
(Odhiambo and Compton,
1987; Russell, 2006)
. Inbred lines were subsequently developed from
the populations by self-pollination for at least seven generations
without any selection for seed characteristics. This study included
two KLS30-derived inbred lines (KLS_S6_1-1 and KLS_S5_2-1-1,
abbreviated to L1L1 and L3L3, respectively) and two KSS30-derived
inbred lines (KSS_S4_4-1-1 and KSS_S4_3-2-1, abbreviated to S1S1
and S3S3, respectively). Experiments were planted at the University
of Wisconsin, West Madison Agricultural Research Station during
the 2013 and 2014 growing season under the same field growing
conditions previously described (Sekhon et al., 2014). Briefly, genotypes
were arranged in three-row plots with three replications and
handplanted in 2.9 m long rows with row and plant spacing of 0.76 m and
0.24 m, respectively. Eight F1 reciprocal hybrids (L1S1, L1S3, L3S1,
L3S3, S1L1, S3L1, S1L3, and S3L3) were generated from these four
parental inbred lines by manual pollination. These hybrids are coded
such that the character on the left signifies the maternal parent and
the character on the right signifies the paternal parent. Ample pollen
was used for each pollination to ensure well-filled ears for consistent
kernel phenotyping. Kernels from the center of three different
primary ears per plot were either dried for seed weight, fixed in ethanol
for imaging, or stored at –80 °C for transcriptional analysis.
Seed weight and seed size measurement
The bulk mature seeds from each plot were used for calculating
100seed weight measured in grams. To generate the grain filling rate, 10
kernels per ear were sampled at each time point [11, 14, 17, 20, 25,
and 28 days after pollination (DAP)] with six replications. Kernel
dry weight was determined after drying samples at 65 °C for 1 week.
To obtain the seed size distribution, images were captured using
an Epson Perfection V700 Photo desktop scanner and VueScan
scanning software without image enhancement and saved as TIFF
(tagged image file format) files. One hundred mature dry seeds
were spread uniformly with the embryo facing the glass platen and
scanned at a resolution of 1200 dpi. Using MatLab image analysis
software, the major (depth) and minor (width) axes and total area of
the kernels was quantified.
Microscopic examination of endosperm area and endosperm
Kernels freshly isolated from the middle of developing ears at 17
DAP were fixed in 70% ethanol (v/v) and stored in 4 °C. Kernels
were first rinsed with distilled water twice and trimmed on both sides
to form a 2–3 mm thick longitudinal median section containing the
embryo. The slices were imaged with a Zeiss AxioZoom fluorescence
stereo microscope to obtain the endosperm area. These slices were
also stained with 0.1% (w/v) berberine sulfate for 5–15 min
depending on the slice thickness to visualize the cell wall on a Zeiss LSM
510 META confocal microscope. A rectangular region of interest of
the approximate location covering most of the endosperm area at
17 DAP was selected and extracted from at least 10 kernels of each
of the KLS and KSS inbred lines. Endosperm size and endosperm
cell size were measured using ImageJ software (http://rsbweb.nih.
gov/ij/). Confocal imaging was performed at the Newcomb Imaging
Center, Department of Botany, University of Wisconsin-Madison.
Total RNA of whole developing kernels from the four
parental lines and eight reciprocal F1 hybrids at 14 DAP and 17 DAP
was extracted using TRIZOL reagent (Ambion, http://www.
lifetechnologies.com), checked for quality, and further purified
using an RNeasy MinElute Cleanup kit (Qiagen, http://www.
qiagen.com) following the manufacturer’s instructions. Isolation
of mRNA, cDNA synthesis, and construction and sequencing
of RNA sequencing (RNA-Seq) libraries were performed at the
University of Wisconsin Biotechnology Center (Madison, WI)
using the Illumina TruSeq RNA sample preparation kit v2
protocol (Illumina, http://www.illumina.com). Sixteen samples were
pooled per lane and sequenced using an Illumina HiSeq 2500 to
generate 151 nucleotide paired-end sequence reads. Raw sequence
reads are available through the National Center for Biotechnology
Information Sequence Read Archive (BioProject PRJNA287557).
Quality of the raw sequences was checked using the FastQC
and reads were trimmed to 100 nucleotides to remove low quality
bases with the fastx_trimmer program within the FASTX toolkit
(http://hannonlab.cshl.edu/fastx_toolkit/index.html). Reads were
mapped to the B73 version 2 reference sequence
(Schnable et al.,
using Bowtie version 0.12.7 (Langmead et al., 2009) and
TopHat version 1.2.0
(Trapnell et al., 2009)
, setting a minimum
intron length of five nucleotides and a maximum intron length
of 60 000 nucleotides. Fragments per kilobase of exon model
per million fragments mapped (FPKM) values were estimated by
Cufflinks version 0.9.3
(Trapnell et al., 2010)
using the version 5b
annotation and providing genome assembly, and requiring a
minimum intron size of five nucleotides.
Identification of differentially expressed genes, gene annotation,
and functional enrichment
Hierarchical clustering was conducted on log2-transformed FPKM
values of expressed genes with FPKM values >1 in all samples using
the hclust command in R. Differentially expressed genes (DEGs)
were identified by pairwise comparisons using edgeR
et al., 2010)
and read counts calculated with the coverageBed
program within BEDTools version 2.17.0 (Quinlan and Hall, 2010).
Only genes with read counts >1 were used for differential expression
analysis, and the significance of differences in expressed genes was
judged on two criteria: FDR (P-value after adjusting for false
discovery rate) ≤0.05 and |log2 fold change| ≥1. A heatmap with
dendrograms was produced with the pheatmap R package
Annotation of transcriptional factor family members was based on
information from GrassTFDB of GRASSIUS (Gray et al.,
Yilmaz et al., 2009
). Gene Ontology (GO) enrichment analyses
of the DEGs and weighted gene co-expression network analysis
(WGCNA)-generated co-expression modules were performed with
the goseq package in R using the Wallenius approximation method
(Young et al., 2010)
. GO term annotations for maize genes were
obtained from Gramene (ftp://ftp.gramene.org/pub/gramene/
CURRENT_RELEASE/data/ontology/go/). All calculations and
plotting were performed in R.
Identification of gene co-expression modules
Gene co-expression module assignments were determined using
the WGCNA protocol
(Zhang and Horvath, 2005; Langfelder and
based on FPKM data. Genes with FPKM <1 for
all samples were filtered out, and a coefficient of variation cut-off
of 0.25 was used to filter genes with low variation among samples.
The Dynamic Tree Cut algorithm with a minimum module size of
50 genes was used to cut the hierarchal clustering. The soft
threshold power beta was set to nine. Significant module–trait associations
were identified by correlating module eigengenes with seed weight,
and the modules with P-values <0.001 were selected for GO
enrichment analysis with the goseq package in R using the Wallenius
(Young et al., 2010)
Maternal parent has a significant effect on seed weight
and seed size
Significant variation for seed size among the KLS30 and
KSS30 populations has previously been shown
and Compton, 1987; Russell, 2006; Sekhon et al., 2014)
understand the maternal contribution to this variation, two
KLS inbred lines (named L1L1 and L3L3) and two KSS
inbred lines (named S1S1 and S3S3), derived from KLS30
and KSS30, respectively, and their reciprocal F1 crosses were
developed (Fig. 1A). The dry weight of the mature seeds of
parental inbred lines showed that KLS inbreds had 267–377%
of the seed weight of KSS inbreds. A strong maternal
influence on seed weight was remarkable as hybrid seeds produced
with KLS inbreds as the mother plants, irrespective of the
genotype of the pollen donor, were consistently heavier than
those produced by maternal KSS plants (Fig. 1B). The
significant maternal effects on seed weight were further revealed
by plotting samples based on the maternal or paternal
parents they had in common (L1, L3, S1, and S3) (Fig. 1C).
When evaluated among maternal parent groups (Fig. 1C,
upper panel), seed weights of maternal group L (L1 and L3)
and maternal group S (S1 and S3) were significantly different
(Tukey test, P<0.05) and there was no overlap in the spread
between group L and group S. However, for the paternal
groups, the spread overlapped across all four group medians
(Fig. 1C, lower panel), and no significant differences were
observed for paternal effects (Tukey test, P>0.05).
We further quantified seed size by image analysis.
Consistent with seed weight, kernel width and depth of KSS
inbreds were significantly smaller than those of KLS inbreds
(Student’s t-test, P<0.05). Comparisons between group
samples that shared either the same KSS maternal parent or the
same KLS maternal parent only showed a significant
difference in kernel width (Fig. 1D). Frequency histograms of
kernel area corroborated the maternal effect, showing the clear
skew trend that discriminated the KLS and KSS maternal
groups (Fig. 1E). Together, the seed morphology analysis
revealed a striking difference in seed weight and seed size
between KLS and KSS parental inbreds as well as the
significant contribution of the maternal parent to seed weight and
seed size in reciprocal hybrids.
KLS inbred lines have larger endosperms and smaller cells than KSS inbred lines, and hybrids mirror the developmental rate of the maternal parent
The endosperm in cereals is the main nutrient sink where
storage materials are deposited during grain filling. To explore
if variation in seed weight and seed size was associated with
changes in endosperm characteristics, we compared endosperm
area and starchy endosperm cell size between KSS and KLS
inbreds. The median longitudinal sections that contained the
embryo of kernels collected at 17 DAP were selected for
measuring endosperm area, and a large rectangular region covering
the majority of the starchy endosperm was used for
measuring cell size (Fig. 2A). Overall, KLS inbreds had a 30–38%
larger endosperm area compared with KSS inbreds (Fig. 2B).
Surprisingly, cell area was smaller in KLS inbreds (Tukey test,
P<0.05). L1L1 endosperm had smaller cell area than S1S1 and
S3S3 by 32% and 9%, respectively, and this reduction for L3L3
was 26% and 4% in comparison with S1S1 and S3S3,
respectively (Fig. 2C, D). The larger endosperms and smaller cell size
of KLS inbreds compared with KSS inbreds indicates that
KLS inbreds have a higher number of endosperm cells, and
that the difference in seed size between KLS and KSS inbreds
would be mainly explained by cell number rather than cell size.
Grain filling is an important agronomic trait in cereals, resemblance to the maternal parent in the rate of grain filling
where a number of cell layers in the endosperm play a critical (Fig. 2E).
role. We systematically evaluated the performance of
parental inbreds and derived hybrids for the grain filling rate by Global gene expression pattern of Krug reciprocal F1
measuring dry mass accumulation throughout kernel devel- hybrids exhibits substantial similarities with maternal
opment beginning at 11 DAP. KLS inbreds and hybrids with parents
a KLS inbred as the maternal parent accumulated dry matter
faster than KSS inbreds and hybrids with a KSS inbred as the In the context of understanding the molecular
mechamaternal parent. Importantly, hybrids exhibited remarkable nisms and genetic regulation underlying seed size control,
we profiled the transcriptomes of developing kernels of
KSS and KLS inbred lines and their F1 reciprocal hybrids
collected at 14 DAP and 17 DAP. We generated 11 × 106–
18 × 106 raw reads for each sample, of which 81.5–84.7%
aligned to the B73 v2 reference genome assembly
et al., 2009)
; the unique aligned reads accounted for ~75%
of the sequenced raw reads (Supplementary Table S1 at
JXB online). The relative abundance of transcripts
calculated as the FPKM was provided (Supplementary Table
S2). We found that 19 909–22 443 of 39 456 maize gene
models were expressed with FPKM >1 among the samples
(Supplementary Table S1). Intriguingly, hierarchical
clustering of the transcriptome profiles grouped the samples
into four major clades corresponding to the four maternal
parents, and reciprocal crosses always formed the primary
cluster with their maternal parent (Fig. 3). This similarity
in expression pattern between reciprocal crosses and their
maternal parents provides valuable insights into the
regulatory role of the maternal transcriptome in phenotypic
variation for seed size and seed weight.
Identification of differentially expressed genes between
KLS and KSS inbreds
Treating inbred lines of each genotype as replicates, we
identified 402 and 691 DEGs between KLS and KSS inbreds at 14
DAP and 17 DAP, respectively. Of these, 187 and 229 showed
higher transcript abundance in KLS inbreds at 14 DAP and
17 DAP, and 215 and 462 DEGs were more abundant in
KSS inbreds at 14 DAP and 17 DAP, respectively (Fig. 4A;
Supplementary Table S3). GO enrichment analysis was
performed on these four groups of DEGs to discover
overrepresented functional categories (Fig. 4B; Supplementary
Table S3). No significant GO terms were found enriched in
DEGs up-regulated in KSS inbreds at 14 DAP, while
upregulated DEGs in KSS inbreds at 17 DAP were enriched in
major metabolic processes such as ‘carbohydrate metabolic
process’ and ‘ֲfatty acid metabolic process’; nutrient reservoir
activity was one subcategory of molecular function
showing the most marked over-representation in KSS inbreds at
both 14 DAP and 17 DAP (see Supplementary Table S3).
Significantly over-represented GO terms among DEGs
up-regulated in KLS inbreds at 14 DAP were all related to
stimulus responses (‘response to stress’, ‘response to high
light intensity’, ‘response to heat’, ‘protein folding’, ‘response
to hydrogen peroxide’, ‘heat acclimation’, and
‘hyperosmotic response’) (Fig. 4B), suggesting improved responses
or adaptation to environmental cues. Significant functional
over-representations enriched in up-regulated genes in KLS
inbreds at 17 DAP included ‘glycolytic process’, ‘DNA
methylation’, ‘glucose metabolic process’, and ‘nucleosome
assembly’ (Fig. 4B). Genes enriched for nucleosome assembly, a
biological process also involved in DNA replication and cell
division, mainly included histone superfamily genes, which
are crucial for packaging of DNA and cell cycle regulation
(Marzluff and Duronio, 2002)
. Up-regulated genes in KLS
at 17 DAP related to DNA methylation were linked to three
genes required for maintenance of CG methylation in plants,
namely VARIANT IN METHYLATION 103 (VIM103)
and two DNA methyltransferase genes (MET8 and MET1)
(Feng et al., 2010; Law and Jacobsen, 2010; Candaele et al.,
. DNA methylation is a key epigenetic determinant that
regulates gene imprinting in plants, and imprinting has been
proposed to be involved in maternal control of nutrient
distribution in plant seeds
(Feil and Berger, 2007; Costa et al.,
We also examined differential accumulation of
transcription factors between KLS and KSS inbred lines. Thirty-three
transcription factors representing 16 families were identified
as differentially expressed in KLS and KSS inbreds (Fig. 4C;
Supplementary Table S4). Of the transcription factors
upregulated in KSS inbreds, Prolamin-box binding factor1
(Pbf1) and WRINKLED1 transcription factor 2 (ZmWri1b)
from the DOF family and AP2/EREBP gene family,
respectively, have been documented as regulators of storage
protein and seed oil
(Pouvreau et al., 2011; Lang et al., 2014;
Zhang et al., 2015)
. Their up-regulation in KSS inbreds
largely corroborated GO enrichment analysis of DEGs. In
addition, transcription factors that function in response
to environmental cues and nutrient uptake and transport,
including three heat shock factors (ZmHSF17, ZmHSF20,
(Yilmaz et al., 2009)
and one MYB-related
protein ZmMRP-1 (Gómez et al., 2002), were up-regulated in
KLS inbreds. ZmMRP-1 is known as a primary endosperm
transfer cell-specific transcriptional activator that plays a
central role in the regulatory pathways controlling transfer cell
differentiation and associated maternal nutrition allocation
Gómez et al., 2009
). Together, functional characterization of
DEGs between KLS and KSS inbreds indicated that different
adaptive responses to environmental and developmental cues
could influence their ability to provision seeds and therefore
affect offspring phenotypes when serving as maternal plants.
Gene co-expression network identifies biological processes and candidate genes important for maternal effects on seed size
To identify networks of co-expressed genes, especially those
that are correlated with seed size, we performed a WGCNA.
After CV filtering, 6349 expressed genes (FPKM ≥1) in our
transcriptome profiling fell into 52 modules, with each
containing at least 50 genes (Supplementary Fig. S1; Supplementary
Table S5). The identified modules were then selected on the
basis of the module–trait relationship, which was calculated
by correlating the module’s eigengene value to the mature seed
weight. Thirteen of these modules were found strongly
correlated with seed weight (correlations ranging from –0.87 to
0.89, P<0.001), containing between 57 (M13) and 168 (M5)
genes (Fig. 5A), and including two (M4) to eight (M5 and
M6) transcription factor genes (Fig. 5A, B). Among the eight
modules negatively correlated with seed weight, four of them
(M1, M5, M10, and M13) showed differential expression
patterns between the KSS group (KSS inbreds and hybrids with a
KSS inbred as the maternal parent) and the KLS group (KLS
inbred lines and hybrids with a KLS inbred as the maternal
parent) (Supplementary Fig. S2). The molecular functions
and biological processes that were most significantly enriched
in these modules negatively correlated with seed weight were
protein autophosphorylation (M1), nutrient reservoir
activity (M5), hexose catabolic process (M10), and trehalose
biosynthetic process (M13) (Table 1). M4 and M12 were two of
the five modules positively correlated with seed weight, both
of which showed consistently lower expression in the KSS
group compared with the KLS group at both 14 DAP and
17 DAP (Fig. 5C). Consistent with GO enrichment analysis
of DEGs, M4 was significantly enriched in DNA methylation
(Table 1), which included MET1 and MET8. M12 contained
114 genes including seven transcription factor genes from the
families ARF, bZIP, G2-like, MADS-box, and Orphans.
Overrepresented biological processes of M12 were related to floral
organ development including maintenance of floral organ
identity, carpel development, and ovule development (Table 1).
This module mainly involved three MADS-box transcription
factor genes: Zea mays AGAMOUS homolog 2 (ZAG2), Zea
mays MADS1 (ZMM1), and ZMM2. According to the maize
gene expression atlas
(Sekhon et al., 2011; Stelpflug et al.,
, ZAG2 and its paralogous gene ZMM1 are largely
restricted to reproductive organs, whole seed, endosperm, and
pericarp (Supplementary Fig. S3). Interestingly, ZAG2 has
been identified as a maternally expressed imprinted gene in the
maize endosperm (Liu et al., 2015). Together, the WGCNA
analysis largely corroborated findings from standard
differential gene expression analyses, and also identified possible
meta-networks composed of multiple GO categories
underlying the observed maternal effect on seed size.
Variation in seed size is common within and among plant
species. Underlying this variation, and thus regulation of seed size, is
a complex array of interactions involving genetic factors,
developmental signals, and environmental cues. Although maternal
effects in plants have long been recognized
(Roach and Wulff,
, the mechanisms whereby maternal effects affect seed size
remain largely unknown, which is especially true for maize. By
using a unique genetic resource derived from the Krug Yellow
Dent long-term selection experiment for seed size in maize, we
identified remarkable reciprocal differences due to large maternal
effects on seed weight and seed size. Integrative analysis of seed
morphogenesis (endosperm size and grain filling) and
transcriptome profiling further provides insights into developmental and
molecular events underlying the maternal control of seed size.
Our observation that reciprocal F1 crosses closely
mirrored the phenotype of the self-pollinated maternal parent
in terms of seed weight, seed size, and seed development
provides strong support for a maternal influence on seed size in
maize. The endosperm in cereals serves as the primary
nutrient source for embryo and seed development. The endosperm
development depends on both sink capacity and assimilates
supplied by sporophytic maternal tissues, thus implicating the
maternal genotype in the process. The endosperm’s strength
as a nutrient sink is proposed to be the function of the
number of endosperm cells and/or the number of starch granules
formed during grain filling
(Capitanio et al., 1983; Reddy and
. Maternal effects on kernel mass are thought
to be due to changes in the number of endosperm cells formed
(Jones et al., 1996). In our study, KLS inbred lines have larger
endosperms but smaller cells compared with KSS inbreds
(Fig. 2), indicating more endosperm cells in the large seed
genotype. Thus, maternal sink constraint determined by the
number of endosperm cells appears to be one developmental
determinant contributing to seed weight variation between
KLS and KSS inbreds and the associated maternal effect on
seed weight/size and grain filling in the reciprocal hybrids.
Consistent with the maternal contribution to seed weight/
size, transcriptome profiles of reciprocal F1 hybrids showed
substantial similarities to the maternal parents (Fig. 3).
Comparative transcriptional profiling analysis of KSS and
KLS inbreds identified a number of DEGs involved in
important biological processes. ZmMRP-1, one up-regulated gene
in KLS inbreds, is so far the only known endosperm transfer
cell-specific transcription activator that regulates transfer cell
differentiation and associated maternal nutrition allocation
Gómez et al., 2009
Lopato et al., 2014
). Thus, the differential
expression of ZmMRP-1 indicated that the divergence of seed
size in KLS and KSS might be related to maternally controlled
nutrient uptake and allocation during seed development. This
also corroborated our hypothesis that maternal sink
constraints would set the basis for maternal effect on seed size.
Interestingly, GO analysis of DEGs identified that the
significantly enriched biological processes in up-regulated DEGs of
KLS at 14 DAP were all related to stimulus responses
including heat stress. Heat stress imposes limitation on endosperm
enlargement, and thus seed size and yield
(Folsom et al., 2014)
Kernel sink capacity determined by endosperm cell number
and/or the number of starch granules is often disrupted by
(Wilhelm et al., 1999; Commuri and Jones, 2001)
Thus, the enhanced expression of heat response genes in KLS
inbreds may endow the kernels with improved intrinsic
ability for thermotolerance, which could contribute to endosperm
enlargement and thus more efficient grain filling.
Differential expression and WGCNA analysis both
identified DNA methylation as a key process distinguishing large
and small seed. We also found a robust association between the
DNA methylation GO term and seed size when we compared
the current meta-analysis with the previous transcriptional
data from Sekhon et al. (2014) in which they profiled the
transcriptome of the developing endosperm of three large Krug
inbreds and three small Krug inbreds. Despite the differences
in the exact genetic stocks used and in the tissues sampled
between these two studies, we identified largely common GO
terms including DNA methylation that were enriched in DEGs
between the endosperm of KLS and KSS inbreds at both 15
DAP and 18 DAP (data not shown). DNA methylation is a
major epigenetic mark underlying gene imprinting which has
been hypothesized to regulate seed size by affecting nutrient
uptake and allocation during endosperm development
et al., 2012; Xin et al., 2013; Bai and Settles, 2014)
examining the overlap between DEGs with imprinted genes that were
previously identified in developing endosperm
(Waters et al.,
2013; Xin et al., 2013)
, we found that a subset of genes
differentially expressed at 17 DAP significantly overlapped with a
subset of the previously described maternally expressed genes
(Supplementary Table S6). Therefore, while our study did not
focus on identification of imprinted genes, transcriptional
differences in genes controlling DNA methylation provide indirect
support for the role of gene imprinting as a molecular
mechanism underlying the observed maternal effect on seed size.
Co-expression network analysis also revealed potential
biological processes and candidate genes involved in seed
development and gene imprinting which could underlie the observed
maternal effect on seed size. Co-expression module M12,
which was positively correlated with seed weight, contained
genes significantly enriched in ovule development. Key genes
in M12 included AGAMOUS-LIKE type I MADS-box
transcription factor (AGL) genes including ZAG2, its paralogous
gene ZMM1, and the C-type MADS gene ZMM2 (Fig. 5).
AGL genes are mostly expressed in female gametophytes or
developing seeds, and have been shown to affect endosperm
development and regulate seed size
(Lu et al., 2012)
in Arabidopsis demonstrated that down-regulation of AGL
genes in the endosperm due to increased levels of
homologous siRNAs caused decreased seed size
(Lu et al., 2012)
Another AGL gene, AGL62, acting as a dosage-sensitive seed
size regulator, correlated positively with seed size
et al., 2013)
. ZAG2 of M12 is highly similar to AGL5, which
was shown to be the direct target of the complex formed
by AGAMOUS (AG) and SEPALLATA (SEP) in the
control of carpel and ovule development in Arabidopsis. ZAG2
was also identified as a maternally expressed imprinted gene
(Waters et al., 2011; Zhang et al., 2011; Liu et al.,
and its expression was largely restricted to
developing seeds and endosperm
(Supplementary Fig. S3; Sekhon
et al., 2011; Stelpflug et al., 2015)
. Interestingly, module M12
genes were found to be enriched in maternal expressed genes
(Supplementary Table S6) by examining the overlap between
WGCNA-generated co-expression modules and the imprinted
genes identified in developing endosperm (Xin et al., 2013).
Considering the similar expression of ZAG2/ZMM1 between
reciprocal hybrids and their maternal plants in addition to
previously described imprinted genes contained in module
M12, we predict that ZAG2 is probably a promising candidate
gene functioning in regulating seed size through its imprinting
role in endosperm. Identifying and separating imprinted loci
from reciprocal endosperms is of great interest for future
studies and will be greatly beneficial for deciphering the genetic
mechanism underlying maternal control of seed size in maize.
Our comprehensive analyses of seed morphology, endosperm
cytology, and seed transcriptome revealed a notable role for the
maternal parent in determining seed size. The identification of
DEGs and co-expression module genes involved in maternal
source constraints extends our understanding of the complex
molecular and cellular events in this process and provides a
foundation for future studies on seed size in crops.
Supplementary data are available at JXB online.
Figure S1. Gene clustering tree (dendrogram) for
identifying consensus modules obtained by hierarchical clustering of
adjacency-based dissimilarity based on FPKM values of all
Figure S2. Heatmaps and barplots of eigengenes for
WGCNA-generated co-expression modules (M1, M5, M10,
and M13) that were negatively correlated with seed weight.
Figure S3. Spatial and temporal expression of ZAG2 and
ZMM1 identified by WGCNA-generated co-expression
module M12 based on the Maize B73 Gene Atlas.
Table S1. Number of reads, mapping percentage, and
number of expressed genes in 24 RNA-seq samples which include
four parental inbreds and eight reciprocal F1 hybrids
collected at 14 DAP and 17 DAP.
Table S2. Gene expression values of B73 RefGen_v2
Filtered Gene Set (FGS) in each sample.
Table S3. Differentially expressed genes and Gene Ontology
in KLS and KSS inbred lines.
Table S4. Transcription factors identified in differentially
expressed genes between KLS and KSS inbreds.
Table S5. Gene models in co-expression modules
significantly associated with seed weight.
Table S6. Relationships of the differentially expressed
genes and co-expression modules to the maternally expressed
gene sets previously described.
We thank Tiezheng Yuan for assistance with WGCNA in R, Nathan D. Miller
and Nicholas Haase for assistance with the individual kernel size
measurements, Marisa Otegui for suggestions on the endosperm cell size experiment,
and Sarah Swanson at the Newcomb Imaging Center at the University of
Wisconsin for technical support.
2009. Local maternal control of seed size by kluh/cyp78a5-dependent growth signaling . Proceedings of the National Academy of Sciences, USA 106 , 20115 - 20120 .
Bai F , Settles AM . 2014 . Imprinting in plants as a mechanism to generate seed phenotypic diversity . Frontiers in Plant Science 5 , 780 .
Berger F , Grini PE , Schnittger A. 2006 . Endosperm: an integrator of seed growth and development . Current Opinion in Plant Biology 9 , 664 - 670 .
Candaele J , Demuynck K , Mosoti D , Beemster GTS , Inzé D , Nelissen H. 2014 . Differential methylation during maize leaf growth targets developmentally regulated genes . Plant Physiology 164 , 1350 - 1364 .
Capitanio R , Gentinetta E , Motto M. 1983 . Grain weight and its components in maize inbred lines . Maydica 28 , 365 - 379 .
Chaudhury AM , Koltunow A , Payne T , Luo M , Tucker MR , Dennis ES , Peacock WJ . 2001 . Control of early seed development . Annual Review of Cell and Developmental Biology 17 , 677 - 699 .
Chojkecki AJS , Gale MD , Bayliss MW . 1986 . The number and sizes of starch granules in the wheat endosperm, and their association with grain weight . Annals of Botany 58 , 819 - 831 .
Commuri PD , Jones RJ . 2001 . High temperatures during endosperm cell division in maize . Crop Science 41 , 1122 - 1130 .
Costa Liliana M , Yuan J , Rouster J , Paul W , Dickinson H , GutierrezMarcos Jose F. 2012 . Maternal control of nutrient allocation in plant seeds by genomic imprinting . Current Biology 22 , 160 - 165 .
Doebley JF , Gaut BS , Smith BD . 2006 . The molecular genetics of crop domestication . Cell 127 , 1309 - 1321 .
Donohue K. 2009 . Completing the cycle: maternal effects as the missing link in plant life histories . Philosophical Transactions of the Royal Society B: Biological Sciences 364 , 1059 - 1074 .
Fang W , Wang Z , Cui R , Li J , Li Y. 2012 . Maternal control of seed size by eod3/cyp78a6 in Arabidopsis thaliana . The Plant Journal 70 , 929 - 939 .
Fatihi A , Zbierzak AM , Dörmann P. 2013 . Alterations in seed development gene expression affect size and oil content of Arabidopsis seeds . Plant Physiology 163 , 973 - 985 .
Feil R , Berger F. 2007 . Convergent evolution of genomic imprinting in plants and mammals . Trends in Genetics 23 , 192 - 199 .
Feng S , Cokus SJ , Zhang X , et al. 2010 . Conservation and divergence of methylation patterning in plants and animals . Proceedings of the National Academy of Sciences, USA 107 , 8689 - 8694 .
Folsom JJ , Begcy K , Hao X , Wang D , Walia H. 2014 . Rice fie1 regulates seed size under heat stress by controlling early endosperm development . Plant Physiology 165 , 238 - 248 .
Gehring M , Choi Y , Fischer RL . 2004 . Imprinting and seed development . The Plant Cell 16 , S203 - S213 .
Gómez E , Royo J , Guo Y , Thompson R , Hueros G. 2002 .
Establishment of cereal endosperm expression domains: identification and properties of a maize transfer cell-specific transcription factor, ZmMRP-1 .
The Plant Cell 14 , 599 - 610 .
Gómez E , Royo J , Muñiz LM , Sellam O , Paul W , Gerentes D , Barrero C , López M , Perez P , Hueros G. 2009 . The maize transcription factor myb-related protein-1 is a key regulator of the differentiation of transfer cells . The Plant Cell 21 , 2022 - 2035 .
Gray J , Bevan M , Brutnell T , et al. 2009 . A recommendation for naming transcription factor proteins in the grasses . Plant Physiology 149 , 4 - 6 .
Gupta PK , Rustgi S , Kumar N. 2006 . Genetic and molecular basis of grain size and grain number and its relevance to grain productivity in higher plants . Genome 49 , 565 - 571 .
Hannah LC , Futch B , Bing J , Shaw JR , Boehlein S , Stewart JD , Beiriger R , Georgelis N , Greene T. 2012 . A shrunken-2 transgene increases maize yield by acting in maternal tissues to increase the frequency of seed development . The Plant Cell 24 , 2352 - 2363 .
Hirsch CN , Flint-Garcia SA , Beissinger TM , et al. 2014 . Insights into the effects of long-term artificial selection on seed size in maize . Genetics 198 , 409 - 421 .
Jiang H , Köhler C. 2012 . Evolution, function, and regulation of genomic imprinting in plant seed development . Journal of Experimental Botany 63 , 4713 - 4722 .
Jones RJ , Schreiber BMN , Roessler JA. 1996 . Kernel sink capacity in maize: genotypic and maternal regulation . Crop Science 36 , 301 - 306 .
Kesavan M , Song JT , Seo HS . 2013 . Seed size: a priority trait in cereal crops . Physiologia Plantarum 147 , 113 - 120 .
Kolde R. 2013 . Pheatmap: pretty heatmaps . R package version 0.7.7 .
Kradolfer D , Hennig L , Köhler C. 2013 . Increased maternal genome dosage bypasses the requirement of the fis polycomb repressive complex 2 in Arabidopsis seed development . PLoS Genetics 9 , e1003163 .
Lang Z , Wills DM , Lemmon ZH , Shannon LM , Bukowski R , Wu Y , Messing J , Doebley JF . 2014 . Defining the role of prolamin-box binding factor1 gene during maize domestication . Journal of Heredity 105 , 576 - 582 .
Langfelder P , Horvath S. 2008 . Wgcna: an r package for weighted correlation network analysis . BMC Bioinformatics 9 , 559 .
Langmead B , Trapnell C , Pop M , Salzberg S. 2009 . Ultrafast and memory-efficient alignment of short DNA sequences to the human genome . Genome Biology 10 , R25 .
Law JA , Jacobsen SE . 2010 . Establishing, maintaining and modifying DNA methylation patterns in plants and animals . Nature Reviews Genetics 11 , 204 - 220 .
Li N , Li Y. 2015 . Maternal control of seed size in plants . Journal of Experimental Botany 66 , 1087 - 1097 .
2015. Characterization of the imprinting and expression patterns of ZAG2 in maize endosperm and embryo . Crop Journal 3 , 74 - 79 .
Lopato S , Borisjuk N , Langridge P , Hrmova M. 2014 . Endosperm transfer cell-specific genes and proteins: structure, function and applications in biotechnology . Frontiers in Plant Science 5 , 64 .
Lu J , Zhang C , Baulcombe DC , Chen ZJ . 2012 . Maternal siRNAs as regulators of parental genome imbalance and gene expression in endosperm of Arabidopsis seeds . Proceedings of the National Academy of Sciences, USA 109 , 5529 - 5534 .
Makkar HPS (ed.). 2012 . Biofuel co-products as livestock feed: opportunities and challenges . Rome: FAO.
Marzluff WF , Duronio RJ . 2002 . Histone mRNA expression : multiple levels of cell cycle regulation and important developmental consequences .
Current Opinion in Cell Biology 14 , 692 - 699 .
Miller ME , Chourey PS. 1992 . The maize invertase-deficient miniature-1 seed mutation is associated with aberrant pedicel and endosperm development . The Plant Cell 4 , 297 - 305 .
Odhiambo MO , Compton WA . 1987 . Twenty cycles of divergent mass selection for seed size in corn . Crop Science 27 , 1113 - 1116 .
Platenkamp GAJ , Shaw RG . 1993 . Environmental and genetic maternal effects on seed characters in Nemophila menziesii . Evolution 47 , 540 - 555 .
Pouvreau B , Baud S , Vernoud V , Morin V , Py C , Gendrot G , Pichon JP , Rouster J , Paul W , Rogowsky PM . 2011 . Duplicate maize Wrinkled1 transcription factors activate target genes involved in seed oil biosynthesis .
Plant Physiology 156 , 674 - 686 .
Quinlan AR , Hall IM . 2010 . Bedtools: a flexible suite of utilities for comparing genomic features . Bioinformatics 26 , 841 - 842 .
Reddy V , Daynard T. 1983 . Endopsperm characteristics associated with rate of grain filling and kernel size in corn . Maydica 28 , 339 - 355 .
Roach DA , Wulff RD . 1987 . Maternal effects in plants . Annual Review of Ecology and Systematics 18 , 209 - 235 .
Robinson MD , McCarthy DJ , Smyth GK . 2010 . edgeR: a bioconductor package for differential expression analysis of digital gene expression data .
Bioinformatics 26 , 139 - 140 .
Russell WK . 2006 . Registration of KLS_30 and KSS_30 populations of maize . Crop Science 46 , 1405 - 1406 .
Schnable PS , Ware D , Fulton RS , et al. 2009 . The B73 maize genome: complexity, diversity, and dynamics . Science 326 , 1112 - 1115 .
Sekhon RS , Hirsch CN , Childs KL , Breitzman MW , Kell P , Duvick S , Spalding EP , Buell CR , de Leon N , Kaeppler SM . 2014 . Phenotypic and transcriptional analysis of divergently selected maize populations reveals the role of developmental timing in seed size determination . Plant Physiology 165 , 658 - 669 .
Sekhon RS , Lin H , Childs KL , Hansey CN , Buell CR , de Leon N , Kaeppler SM . 2011 . Genome-wide atlas of transcription during maize development . The Plant Journal 66 , 553 - 563 .
Stelpflug SC , Sekhon RS , Vaillancourt B , Hirsch CN , Buell CR , Leon Nd , Kaeppler SM . 2015 . An expanded maize gene expression atlas based on RNA-sequencing and its use to explore root development . Plant Genome doi: 10.3835/plantgenome2015.3804 .0025.
Sundaresan V. 2005 . Control of seed size in plants . Proceedings of the National Academy of Sciences, USA 102 , 17887 - 17888 .
Trapnell C , Pachter L , Salzberg SL . 2009 . Tophat: discovering splice junctions with RNA-seq . Bioinformatics 25 , 1105 - 1111 .
Trapnell C , Williams BA , Pertea G , Mortazavi A , Kwan G , van Baren MJ , Salzberg SL , Wold BJ , Pachter L. 2010 . Transcript assembly and abundance estimation from RNA-seq reveals thousands of new transcripts and switching among isoforms . Nature Biotechnology 28 , 511 - 515 .
Waters AJ , Makarevitch I , Eichten SR , Swanson-Wagner RA , Yeh CT , Xu W , Schnable PS , Vaughn MW , Gehring M , Springer NM. 2011 .
Parent-of-origin effects on gene expression and DNA methylation in the maize endosperm . The Plant Cell 23 , 4221 - 4233 .
Waters AJ , Bilinski P , Eichten SR , Vaughn MW , Ross-Ibarra J , Gehring M , Springer NM. 2013 . Comprehensive analysis of imprinted genes in maize reveals allelic variation for imprinting and limited conservation with other species . Proceedings of the National Academy of Sciences, USA 110 , 19639 - 19644 .
Wilhelm EP , Mullen RE , Keeling PL , Singletary GW . 1999 . Heat stress during grain filling in maize: effects on kernel growth and metabolism . Crop Science 39 , 1733 - 1741 .
Xin M , Yang R , Li G , et al. 2013 . Dynamic expression of imprinted genes associates with maternally controlled nutrient allocation during maize endosperm development . The Plant Cell 25 , 3212 - 3227 .
Xiong Y , Mei W , Kim E-D , Mukherjee K , Hassanein H , Barbazuk WB , Sung S , Kolaczkowski B , Kang B-H. 2014 . Adaptive expansion of the maize maternally expressed gene (Meg) family involves changes in expression patterns and protein secondary structures of its members .
BMC Plant Biology 14 , 204 .
Yilmaz A , Nishiyama MY , Fuentes BG , Souza GM , Janies D , Gray J , Grotewold E. 2009 . Grassius: a platform for comparative regulatory genomics across the grasses . Plant Physiology 149 , 171 - 180 .
Young M , Wakefield M , Smyth G , Oshlack A. 2010 . Gene ontology analysis for RMA-seq: accounting for selection bias . Genome Biology 11 , R14 .
Zhang B , Horvath S. 2005 . A general framework for weighted gene co-expression network analysis . Statistical Applications in Genetics and Molecular Biology 4 , Article 17.
Zhang M , Zhao H , Xie S , et al. 2011 . Extensive, clustered parental imprinting of protein-coding and noncoding RNAs in developing maize endosperm . Proceedings of the National Academy of Sciences, USA 108 , 20042 - 20047 .
Zhang Z , Liu Z , Hu Y , Li W , Fu Z , Ding D , Li H , Qiao M , Tang J. 2014 .
QTL analysis of kernel-related traits in maize using an immortalized F2 population . PLoS One 9 , e89645 .
Zhang Z , Yang J , Wu Y. 2015 . Transcriptional regulation of zein gene expression in maize through the additive and synergistic action of opaque2, prolamine-box binding factor, and O2 heterodimerizing proteins .
The Plant Cell 27 , 1162 - 1172 .