Plant seeds are remarkable vessels of life, encapsulating the potential for growth and ensuring the survival and dispersal of the species. They originate from the maturation of ovules within flowering plant, containing an embryo surrounded by a protective outer layer called the seed coat. The genetic repository within the embryo controls the entire life cycle of the plant and the seed coat shields it from environmental stresses and pathogens securing the unfolding of the plant’s next generation in interaction with the environment (Kigel, 1995). Seeds stores the vital components of nutrient that nourish the development of a plant until it can photosynthesize on its own. As the seedling germinates, the embryo and cotyledons surrounded by the endosperm start to grow. Endosperm is a nutrient-rich tissue that mainly contains starch and provides essential energy to the germinating seedling. This resource allocation mechanism diverges between monocots and dicots. Monocots, characterized by a single cotyledon which is an embryonic leaf in the germinating seed, tend to retain their endosperm to nourish the growing seedling. Conversely, dicots often transfer nutrients from the endosperm to the cotyledons as the seed matures (Sabelli, 2012; Sreenivasulu and Wobus, 2013).

Approximately 60% of human energy supply is derived from three monocots cereal species — wheat, rice and maize, which all belong to the Poaceae family making Poaceae the most culturally and economically important plant family in the world (Linder et al., 2018). Therefore, seed development in the Poaceae family has been investigated through various approaches, including biochemistry, molecular biology and omics studies. There are studies describing the genotype/trait associations, genetics and transcriptional regulatory network of seeds development in wheat, rice and maize respectively (Dai et al., 2021; Shikha et al., 2021; Chen et al., 2022; Wang and Sun, 2023). However, no comparative analyses of omics data sets between wheat and maize in Poaceae family exist to the best of our knowledge. Therefore, understanding molecular and genetic mechanisms, modeling of biological networks by interrogation of entire pools of genomic, transcriptomic, metabolomic and proteomic data sets from the two crops with extensive comparative analysis remains a substantially important goal to further seed yield improvements.

Furthermore, it is well-established that the C4 pathway involves the incorporation of carbon into 4-carbon metabolites like malate and oxaloacetate, while the C3 photosynthetic pathway fixes CO ₂ into 3-carbon metabolites such as 3-phosphoglycerate (PGA) through the Calvin cycle. There has been an ongoing debate surrounding whether wheat, a representative C3 photosynthetic crop, utilizes the C4 photosynthetic pathway during grain development like maize, in recent decades (Hibberd and Furbank, 2016). Omics studies have the potential to provide insights on certain questions regarding to this topic.

In the past decade, high-throughput sequencing technologies have revolutionized entire branches of the life sciences including plant improvement, human disease, pharmaceutical engineering. The term ‘omics’ is derived from genomics and signifies a holistic approach to the study of biological systems. It involves the comprehensive investigation of entire set of biological molecules and processes on a large scale, providing an in-depth understanding of complex biological systems (Hasin et al., 2017). There are many types of omics, each focusing on a specific aspect of biological data. Several of them have widely applied in plant science as described below.

1. Genomics is a study of the complete genome of a particular species. It focuses on sequencing of an entire genome and identifying genetic variants associated with plant traits or responses. In the plant research field, genome-wide association studies (GWAS) and quantitative trait locus (QTL) analysis are the most popular approaches to discover variants of interest associated with various plant traits (Korte and Farlow, 2013).

2. Transcriptomics is a study of entire RNA molecules transcribed in a cell or organism. It is mainly used for analysis of gene expression levels, alternative splicing, and non-coding RNA. RNA-Seq, a widely used short-read sequencing technology, offers significantly more accurate transcriptome profiling, enabling the detection of specific isoforms with greater precision compared to other methodologies like microarrays (Wang et al., 2009).

3. Proteomics is a comprehensive analysis of the entire protein components in a specific tissue from a particular species. Utilizing mass spectrometry (MS)-based proteomics, it is possible to characterize and quantify thousands of proteins simultaneously, and uncover their post-translational modifications (PTMs) in a parallel manner (Bantscheff et al., 2012; Dupree et al., 2020). Presently, the insights gleaned from proteomic research has substantially enhanced our comprehension of biological complexity. This advancement has improved our understanding of the molecular mechanisms driving plant responses to environmental stimuli and various developmental stages.

4. Metabolomics focuses on the analysis of small molecules (metabolites) within a biological system (cell, tissue, organism, etc). It offers a comprehensive snapshot of a plant’s metabolic profile, encompassing the identification and quantification of diverse compounds like sugars, organic acids, and secondary metabolites. Notably, these plant secondary metabolites confer a multitude of advantages, including protective functions for the plants and health-promoting properties for consumers (Zhu et al., 2016; Kumar et al., 2017). By characterizing novel metabolites and examining their dynamic flows, we can significantly deepen our understanding of the pathways through which plants synthesize and regulate these vital compounds. These compounds are crucial not only for plant growth and development but also for their ability to respond effectively to various stressors.

Additionally, a range of omics disciplines, such as epigenomics, phenomics, and lipidomics, play a pivotal role in advancing research on plant seed development. This review is dedicated to offering a thorough overview of the current understanding of the molecular facets of wheat and maize seed development. Specifically, it focuses on delineating both the commonalities and differences between these two vital crops within the context of the aforementioned four omics domains.

The development of seeds is a complex process encompassing several maternal and filial tissues with a series of intricate events. These events begin with the fertilization of the ovule where one sperm cell fuses with the egg cell to form a zygote marking the start of the embryonic development. The cellularization events followed after fertilization and lead to the formation of triploid endosperm (two polar nuclei and one sperm cell) with continuous cell division (Kowles and Phillips, 1988; Berger, 1999). Unlike dicot species, that are dominated by the developing embryo, the endosperm in cereals like wheat and maize will continuously accumulates starch and storage proteins leading to a high-calories content. While the endosperms of cereal grains, like wheat, generally achieve full cellularization through ongoing cell division and alveolation ( Figure 1 ), the endosperm of maize displays a somewhat distinct pattern of cellularization. Contrasting with wheat, where maize individual cells form distinct cell walls, the primary endosperm cell undergoes several rounds of nuclear divisions without the formation of cell walls. As a result, the endosperm remains multinucleate, with many nuclei sharing a common cytoplasm (Olsen et al., 1995). Hence, maize demonstrates an irregular final partitioning of the central vacuole at the base of its endosperm, which significantly influences the kernel size and contributes to its distinctive morphology (Leroux et al., 2014) ( Figure 1 ). At the end of mitotic division, the development of the outer and inner layers, known as the pericarp and aleurone layer, respectively, signals the commencement of the grain filling stage. Owing to the reduced mitotic activity and limited cell size, the growth of the endosperm is predominantly driven by cell enlargement, characterized by the accumulation of starch, lipids, and proteins (Randolph, 1936; Kowles and Phillips, 1985; Li et al., 2006; Kong et al., 2015). Studies on the distribution and retention of radioactivity of ¹⁴ CO ₂ on cereal grains suggested that photosynthesis plays a crucial role in sustaining the majority of the grain weight (Merah and Monneveux, 2015; Yang et al., 2016; Estévez-Geffriaud et al., 2020). However, the sources of reserves remobilized for seed grain development vary between wheat and maize. In maize, although stored carbohydrates are available for seed growth, they are primarily allocated to maintenance processes, especially under significant stress. As a result, maize plants exhibit limited efficiency in utilizing these reserves for grain filling prior to flowering (Kiniry et al., 1992). Instead, these reserves are more effectively used in post-flowering stages, such as during leaf senescence. In contrast, wheat demonstrates a three- to four-fold higher efficiency in the remobilization of assimilates stored before flowering for seed growth, in comparison to maize (Borrás et al., 2004). Initial stages of seed development, including cellularization and grain filling, are characterized by intense transcriptional regulation across various pathways. These pathways encompass primary metabolism, cell division, stress response mechanisms, and protein synthesis and degradation (Sprunck et al., 2005; She et al., 2011; Chen et al., 2017a). However, as seeds advance toward maturity, they develop a suite of crucial physiological characteristics that ensure successful seedling establishment once they are independent of the parent plant. These traits hinge on the ability to undergo complete desiccation without losing viability, a phenomenon referred to as desiccation tolerance (Righetti et al., 2015). Therefore, the focus of transcriptional regulation shifts from primary metabolism pathways to processes like chromatin condensation and nuclear size reduction (Baroux et al., 2007). At this stage, a mature grain is composed of starchy endosperm (accounting for 83–84% of dry weight), embryo (3%), aleurone layer (6.5%), and pericarp (7–8%), with only minor differences observed between wheat and maize (Barron et al., 2007; Sethi et al., 2021).

QTL mapping is a statistical analysis to identify a region of DNA associated with a particular trait that exhibits variability within a population based on a genetic linkage map (Lander and Botstein, 1989). The mapping process requires the population generated between two genetically distinct parents with respect to the trait of interest (Myles and Wayne, 2008). In most of cases, a significant portion of the phenotypic variation within this region can be attributed to a small number of loci with substantial effects (Remington and Purugganan, 2003). The precisely mapping of QTL to a fine genetic position generally requires sufficient map resolution (Dinka et al., 2007).

With the emergence of high-throughput genotyping technologies, millions of markers can be identified at more affordable costs. GWAS has now become one of the most potent tools at the disposal of geneticists (Flint-Garcia et al., 2003). GWAS aims to identify associations between genotypes and traits of interest by leveraging phenotypic variations across diverse populations. These populations consist of unrelated individuals whose ancestry can be traced back to a common progenitor (Uffelmann et al., 2021). Consequently, the genetic distance between these individuals, influenced by numerous recombination events, leads to a faster decay in linkage disequilibrium, requiring extensive marker coverage to sufficiently capture the genome-wide genetic variations (Alqudah et al., 2020).

In recent decades, QTL mapping and GWAS have emerged as successful approaches, significantly advancing our understanding of gene functions in a variety of crops, including wheat, maize, and rice. It has been widely used to investigate a range of biological and physiological traits, as well as to uncover the genetic underpinnings of natural selection and population divergences within a species. There are numerous comprehensive reviews that delve into population design and statistical models used in breeding research (Bordes et al., 2014; Xiao et al., 2017; Shikha et al., 2021; Dwiningsih and Al-Kahtani, 2022; Gupta et al., 2022; Saini et al., 2022). Plant geneticists focus on a wide array of traits, ranging from disease resistance and grain yield to flowering time, biomass, and seed size, among others. Our review of literature from 2011 to the present reveals that disease resistance and grain yield are particularly prominent, accounting for over half of the studies published (773/1004 for wheat, 501/884 for maize). This emphasis highlights the critical importance of these two traits in agricultural research (Kaya and Akcura, 2014; Millet et al., 2019). (Bhatta et al., 2018; Qaseem et al., 2019). (Qaseem et al., 2019). (Muhammad et al., 2020).

Grain yield is a multifaceted trait, intricately linked to both the number of grains per unit area and the thousand-kernel weight (TKW). Furthermore, aspects like grain shape, spike architecture, plant height, and flag leaf size play a crucial role in influencing grain yield by affecting photosynthesis. In wheat research, a multitude of candidate loci have been identified, each associated with grain yield, size, and TKW. In the study conducted by Yu and colleagues, 768 bread wheat accessions were analyzed to pinpoint the genetic loci linked to grain size and weight. Notably, a principal locus related to grain length, designated as qGL-2D, exhibited the most robust association across multiple years and encompasses 125 annotated genes. Through transcriptome analysis, the researchers validated the gene TraesCS2D02G414800, which is responsible for encoding an oleosin—a protein crucial in seed maturation and germination—as a candidate gene within the qGL-2D locus (Yu et al., 2022). In a separate study conducted by Tekeu and team, three quantitative trait loci (QTLs) related to grain size were identified on chromosomes 1D, 2D, and 4A. This discovery was made through a GWAS that included 157 international wheat accessions and 71 Canadian wheat accessions. Notably, a candidate gene located on chromosome 2D, TraesCS2D01G331100, was identified as an ortholog of the D11 gene in rice. This gene, which encodes a cytochrome P450, displayed high expression levels in the developing embryo, particularly during embryogenesis and grain development in wheat, affecting both endosperm and pericarp tissues (Tekeu et al., 2021).

Although a vast array of candidate genes linked to grain yield have been discovered through GWAS and QTL mapping, the molecular cloning of these genes in wheat is still relatively unexplored (see Tables 1 , 2 ). Unlike wheat, maize is a diploid species, the research on seeds development has greatly benefited from the investigation of mutator-induced mutant collections such as defective kernel (dek) mutants with affected development in both embryo and endosperm (Neuffer and Sheridan, 1980; McCarty et al., 2005), empty pericarp (emp) mutants (Scanlon et al., 1994), defective endosperm (de) or endosperm-specific mutant (Consonni et al., 2022) and embryo specific (emb) mutants (Clark and Sheridan, 1991). Those mutant collections provide abundant material for the molecular genetic analysis on the mechanisms of embryo and endosperm development allowing maize to become the leading system for molecular cloning of functional genes in plants. For instance, numerous dek mutants, such as Dek1 (encoding a membrane protein of the calpain gene superfamily) (Lid et al., 2002), Dek2, Dek35, Dek36, Dek37, Dek39 (encode a pentatricopeptide repeat protein) (Qi et al., 2017; Wang et al., 2017; Chen et al., 2017b; Dai et al., 2018; Li et al., 2018), and Dek* (encoding AAA-ATPase controlling 60S ribosome export) were characterized. The associated pathways involve cell fate specification in the endosperm (Dek1), ribosome use efficiency (Dek*) and RNA editing (Dek2, Dek35–39). However, compared to the mutant collections, the number of genes identified still represents only a small fraction of the entirety of the collections.

With the recent advancements of GWAS and QTL mapping, more natural genetic variations related to maize kernel traits had been identified ( Tables 2 , 3 ). By comparing the genes list between wheat and maize, there are some genes orthologs exist in both wheat and maize that effected the grain weight and size, such as Grain Size 5 (GS5) which encodes a serine-type carboxypeptidase. The overexpression of GS5 in wheat and maize both exhibit a positive regulation of grain weight suggesting a similar pathway controlling the yield (Liu et al., 2015; Ma et al., 2016). Remarkably, a recent GWAS, which examined a Chinese mini-core collection comprising 262 accessions—encompassing over 70% of the genetic diversity in Chinese wheat—uncovered a notable candidate gene, trehalose-6-phosphate phosphatase (TaTPP-7A). This gene is significantly correlated with the TKW in wheat (Liu et al., 2023). Overexpression of TaTPP-7A in wheat induced a phenotype featuring higher TKW, longer grains and early maturity, while RNAi or gene-edited lines with silenced TaTPP-7A expression exhibited a reversed phenotype, including lower TKW, shorter grains, and late maturity. Further transcriptome analysis revealed that TaTPP-7A primarily boosts starch synthesis via the T6P-SnRK1 pathway and the sugar-ABA interaction. Intriguingly, the RAMOSA pathway, first proposed in maize decades ago, consists of three RAMOSA genes, including two transcription factors. The RA1 gene encodes a Cys2-His2 zinc finger transcription factor, while RA2 is responsible for a Lateral Organ Boundary (LOB) domain transcription factor. Meanwhile, RA3 is identified as a gene encoding trehalose-6-phosphate phosphatase (TPP), which is expressed in the axillary inflorescence meristems (Wang et al., 2021). RA3 acts upstream of transcription factor RA1 to regulate inflorescence branching through modulation of treholose or T6P levels. In maize, two TPP genes, namely RA3 and TPP4, have been characterized. Mutants of both ra3 and tpp4 demonstrated a decrease in crop yield, characterized by abnormal branching and irregular seed row patterns. Notably, the double mutant combining ra3 and tpp4 exhibited an even more pronounced phenotype (Satoh-Nagasawa et al., 2006; Claeys et al., 2019). Moreover, the overexpression of a rice TPP gene in the female reproductive tissues of maize significantly improved yield under drought conditions during the flowering stage (Nuccio et al., 2015). This, along with molecular evidence from mutants and transgenic lines in both wheat and maize, suggests a conserved pathway in cereal crops related to trehalose-6-phosphate (T6P), impacting grain size and yield.

In addition, increasing evidence has indicated the ubiquitin-proteasome pathway plays a conserved role in controlling the grain size through restricting the cell expansion in plants (Li and Li, 2016; Shi et al., 2019). GW2 encodes a E3 ubiquitin ligase has been confirmed to be a negative regulators of grain size in rice (Song et al., 2007), maize (Li et al., 2010b) and wheat (Liu et al., 2020). Interestingly, a novel allele of TaGW2 has been shown to increases grain weight but decrease grain number in wheat, suggesting additional roles of E3 ubiquitin ligase in controlling grain weight and size in wheat (Zhai et al., 2018).

Ortholog-based gene discovery is a valuable strategy, researchers often resort to identify orthologous genes from well-studied model plants. In the realm of cereal research, rice was widely used as a model species. In most cases, leveraging orthologs presents an effective avenue for gene discovery offering insights on conserved pathways and gene functions. However, it’s important to acknowledge that relying solely on ortholog-based gene discovery has its limitations. As the evolutionary distance between species can influence the accuracy of orthologs predictions, potentially resulting mis-annotations or missed genes and overlooking species-specific adaptations and unique genetic features (Woodhouse and Hufford, 2019).

Despite the identification of numerous candidate genes through GWAS that exhibit promising influences on grain traits such as ear length and height in maize, the precise roles of these genes remain elusive. For example, the gene YIGE1, implicated in sugar and auxin signaling pathways, was discovered via GWAS. CRISPR/Cas9 knockout mutants of YIGE1 displayed reduced grain yield, accompanied by a decrease in female inflorescence meristem size and the number of kernels per row (Luo et al., 2022). Nevertheless, the specific function of the YIGE1 protein is yet to be fully characterized.

This review presents an in-depth exploration of the complex mechanisms involved in seed development in wheat and maize, with a particular focus on the molecular dimensions as revealed through genomics, transcriptomics, proteomics, and metabolomics. Despite considerable research on these individual species, there is a notable deficiency in comprehensive comparative studies between C3 and C4 plants within the Poaceae family. The conclusion emphasizes the essential need for a multidisciplinary approach to decode the complexities of photosynthetic contributions to grain filling in cereal crops. It draws attention to the ongoing debates within this field and underlines the critical importance of sustained research. Future research should aim to clarify the regulatory principles, particularly through an integrated lens of genomics, transcriptomics, proteomics and metabolomics.

YJ: Conceptualization, Writing – original draft, Writing – review & editing. TH: Writing – original draft, Writing – review & editing. AS: Supervision, Writing – review & editing. LJ: Conceptualization, Supervision, Writing – review & editing.