MicroRNAs Regulate Grain Development in Rice

Abstract

Ensuring food security is a challenge for humans. Rice grain yield and quality must urgently be increased to overcome this challenge. MicroRNA (miRNA) is an important regulatory module in plant development and stress responses. Grain yield and quality are pleiotropic traits that employ cooperative genetic factors, including miRNA and its regulatory mechanisms. This review provides an overview of plant miRNAs and the composition and development process of rice grains. It also summarizes the research progress in miRNA regulation for agronomically important rice grain traits, providing a basis for further identifying miRNAs related to rice grain development and elucidating their regulatory mechanisms.

1. Introduction

1.1. Importance of Rice Grain Development

Rice grain, the harvested organ, determines both the final yield and quality, and serves as a vital energy source for humans. Grain is also the reproductive organ, making it essential for rice growth and development. While the waste is mainly used for low value-added production, it is being improved [1]. Today, an increasing global population, uneven regional rice yields, and a lack of rice with special functional properties have led to food security being the biggest global challenge for humans [2]. Increasing grain yield and quality can help solve this challenge. Therefore, grain yield and quality, which are complex agronomic traits that depend on multiple plant development features, are urgent considerations for improvement through breeding [3].

As the final sink organ, grain development directly determines the final grain yield and quality. The hull supplies a limited space for grain filling. The caryopsis, which is surrounded by the hull, is the edible part of the grain. The endosperm occupies most of the caryopsis. The endosperm accumulates storage substances and provides the nutrition for embryo germination and growth. Therefore, the hull limits grain weight, and the development and filling of the endosperm determine the grain weight and quality. The hull and endosperm are two prominent target tissues for rice improvement.

1.2. Discovery of miRNAs and Their Important Roles in Plants

miRNA is a class of endogenous, small, noncoding RNAs (sRNAs) that mainly suppress their target gene expression in the post-transcriptional level. miRNAs are approximately 20–24 nucleotides (nt) long, and 21 nt miRNAs are the most abundant in plants [4]. The first discovery of miRNAs was in Caenorhabditis elegans in 1993, which opened up a new avenue for research [5]. In plants, miRNA was first found in Arabidopsis in 2002 [6]. Subsequently, an increasing number of plant miRNAs were identified. High-throughput sequencing technology promoted the genome-wide discovery of miRNAs, but their biological functions are still being explored. Currently, 738 mature miRNAs produced from 604 precursors have been deposited in the miRBase v22 database (www.mirbase.org, accessed on 28 June 2025) for rice.

The discovery of miRNA revealed a new level of gene expression regulation. miRNAs are potent regulators of plant development, including meristem identity, organ morphogenesis, and floral transition, thereby affecting the development of roots, leaves, flowers, seeds, and crop agronomic traits [7,8,9,10]. In addition, abiotic and biotic stresses, such as disease, insect, drought, high temperature, cold, high salinity, and nutrient deficiency can induce the altered miRNA expression in plants. This indicates that miRNAs also play an indispensable role in plant stress resistance [11,12,13]. For instance, rice miR319 [14], miR166 [15,16], miR399 [17], miR827 [18], miR395 [19], miR398 [20], and miR12477 [21] are involved in various abiotic stress responses. While miR398b [20,22], miR7695 [23], miR164a [24], miR393 [25], and miR444 [26] respond to biotic stress. However, little study has focused on the specific regulation of how rice grain development responds to stress.

Due to the influence of miRNAs in modulating plant development, they are being extensively exploited to improve agronomic traits in crops, particularly grain yield and quality. Here, we summarized the function, biosynthesis, and action mechanism of miRNAs in plants, and the composition and development process of rice grain. The miRNAs identified for regulating agronomically important grain traits in rice are emphasized and listed in Table 1.

2. Biosynthesis and Action Mechanism of miRNA in Plants

2.1. Biosynthetic Pathways of miRNAs in Plants

The miRNA biogenesis pathway in plants belongs to the canonical type, which undergoes the following processes: transcription of miRNA genes, formation of miRNA duplexes, and miRNA maturation and assembly [12,27,28]. Specifically, plant miRNAs are produced from endogenous genes known as miRNA genes (MIRs). These MIR genes are transcribed by RNA polymerase II to form primary miRNA (pri-miRNA) in the nucleus [29,30]. Then, pri-miRNAs are processed by Dicer-like 1 (DCL1) with RNA-binding proteins HYPONASTIC LEAVES 1 (HYL1), SERRATE (SE), and nuclear cap-binding protein complex (CBC) to form precursor miRNAs (pre-miRNAs), and the pre-miRNAs are further cleaved by DCL1 to form miRNA duplexes (short double-stranded RNA, dsRNA), including a guide strand (mature miRNA) and a passenger strand (miRNA*) [12,31]. miRNA duplexes are 2′-O-methylated by Hua Enhancer1 (HEN1) for stabilization. Subsequently, the duplexes are transported from the nucleus to the cytoplasm, and one mature miRNA, as the guide strand, can bind to the effect protein Argonaute (AGO), forming miRNA-induced silencing complexes (RISCs) [32], which guide the cleavage, translational repression, or epigenetic modification of target mRNAs. However, the passenger strand miRNA* is gradually degraded [33,34,35].

Pleiotropic phenotypes are caused by protein mutations in miRNA biogenesis and function [27]. For example, DCLs play a critical role in miRNA processing. The loss of function of OsDCL1 showed overall shoot and root abnormalities [36]. OsDCL2 knockdown exhibited a dwarf phenotype and lower spikelet fertility [37]. OsDCL3a knockdown caused defective phenotypes, including dwarfism, larger flag leaf angle, and fewer secondary branches [38]. OsDCL3b knockdown reduced pollen fertility, the seed setting rate, and grain yield, but increased grain quality [39]. Mutation in OsDCL4 caused the abnormal formation of the shoot apical meristem (SAM) [40]. WAF1, an ortholog of Arabidopsis HEN1, is essential for rice development, especially SAM maintenance and leaf morphogenesis [41].

In addition, AGOs are a core component involved in the miRNA effect pathway. AGOs interact with various miRNAs to form RISC complexes, regulating rice development. The rice AGO family contains 19 members, and numerous gene duplications have been identified in this family. For example, OsAGO1 contains four homologs, including OsAGO1a/b/c/d, which cause functional diversification. The knockdown of four OsAGO1s results in various degrees of developmental defects and an increased number of miRNA targets [42]. OsAGO1a/b/c and OsAGO18 confer the dwarf virus in rice [43,44]. OsAGO1b/GSNL4 regulates grain and leaf development [45], and OsAGO1d determines temperature-sensitive male sterility [46]. OsAGO2 regulates anther development [47], rice immunity to blast disease and black-streaked dwarf virus infection [48,49], grain length, and salt tolerance [50]. OsMEL1, an AtAGO5 homologue, is essential for the progression of premeiotic mitosis and meiosis during sporogenesis in rice [51]. OsPNH1, an AtAGO10 homologue, regulates SAM and leaf development [52]. OsAGO7 regulates shoot meristem initiation [53,54]. DH2/AGO7 affects leaf and lemma development in rice [55]. OsAGO17 regulates grain size and stem development [56].

2.2. Action Modes of miRNAs in Plants

Once miRNAs are loaded onto RISC, they target complementary sequences and silence target genes through the specific cleavage of the target mRNA or translational repression of target genes, which mainly depends on the sequence similarity between the miRNA and target mRNA sequences. When miRNA perfectly pairs with the target mRNA, the AGO protein in the RISC complex cleaves the phosphodiester bonds of the target mRNA. The excised mRNA fragments are then degraded, blocking the target gene expression [11,12,57]. In addition, when miRNA shows imperfect sequence similarity with the target mRNA, the action mode can prevent ribosome binding and moving on the mRNA, thereby inhibiting protein translation [11]. In addition to these two main action modes, miRNAs may regulate gene expression through epigenetic modification [12]. However, most plant miRNAs match their target mRNAs closely. miRNA sequences are also conserved so that they typically have fewer targets, and the former action mode through cleavage is the main one identified in plants [58], enabling researchers to predict miRNA modules [10,12].

3. Composition and Development Processes in Rice Grains

Rice grains consist of the hull and caryopsis. The hull is an organ unique to the grass family and comprises the palea and lemma, protecting seeds and forming a filling container that sets the upper limit for the size of the caryopsis (Figure 1). The molecular pathways of grain size that refer to hull development have been well summarized [59].

The rice caryopsis is composed of three genetically different tissues: the diploid embryo formed by the fusion of the egg cell and sperm; the triploid endosperm formed by the fusion of two polar nuclei and sperm; and the diploid maternal tissues surrounding the embryo and endosperm, including the pericarp and testa. This is a typical phenomenon of double fertilization. Therefore, grain development is influenced by various factors through the synergistic control of maternal and zygotic tissues [60]. The early seed morphogenetic phase is marked by the orderly division of the zygote, embryonic patterning formation, and endosperm’s synchronized proliferation. This is followed by the maturation phase, which is characterized by the accumulation of storage substances, such as starch, protein, and lipid. It finally enters the drying phase, during which seeds enter and maintain dormancy until external conditions are suitable for seed germination. Understanding these critical, fine-tuned developmental changes and identifying the molecular controls for manipulating embryo and endosperm development is pivotal [60,61,62,63,64].

The proportion of the embryo and endosperm varies among species. Within a species, this ratio is relatively stable. The endosperm typically accounts for 90%, the embryo accounts for only 2–3%, and the pericarp, testa, aleurone layer, and other tissues account for 6–7% in the caryopsis of mature rice. This indicates that there is genetic regulation that maintains a balance between the endosperm and embryo size within the limited seed space of maternal tissue. The endosperm is the main edible part of mature grains, which determines the grain size to a certain extent and is the main place for starch and protein storage. The embryo is enriched in various nutrients, such as proteins, lipids, vitamins, minerals, and phytochemicals. However, brown rice is not popular due to its taste. White rice with the bran removed (i.e., endosperm) is the most consumed. Therefore, grain size and quality are strongly influenced by maternal and zygotic tissues, namely the hull and endosperm. As a reproductive organ, the endosperm can also provide nutrients for embryo germination and growth. Therefore, there are physical, nutritional, and signal transduction interactions between the embryo and endosperm.

Many efforts have been made to elucidate the processes of embryogenesis and endosperm development [65,66]. It usually takes about 30 days for rice from pollination to grain maturity. The first key developmental event is fertilization, which occurs within a few hours after pollination [66,67]. The development of the rice caryopsis begins with double fertilization; the egg cell and central cell separately fuse with the sperm cell to form the zygote and polar nucleus, which then develop into the diploid embryo and triploid endosperm. Caryopsis development involves synchronized processes, such as signaling exchange, interactions, and nutrient transfer between the embryo and endosperm.

For embryogenesis, the first division of the rice zygote is transverse and asymmetric, occurring about 4 h after fertilization. Embryogenesis occurs rapidly, and the embryo consists of 5–8 cells within 24 h. The multicellular globular embryo forms via cell division within 3 days after pollination (DAP). The embryo undergoes morphogenesis over the next 4–10 DAP, and therein the onset of the coleoptile and the differentiation of the SAM and radical are observed at 4 DAP. At 5 DAP, the first leaf primordium forms, and at 6 DAP, the vascular bundle and embryonic root tip differentiate. At 7–8 DAP, three foliage leaves are recognizable in the shoot apex. Subsequently, only the embryo volume increases, without other morphological changes. Finally, the embryo gradually matures and enters dormancy at 20 DAP [66,68].

Endosperm development is categorized into cellular, helobial, and nuclear types. Rice endosperm is of the nuclear type, which is the most common form. According to cell morphology and physiological characteristics, the rice endosperm development process can be divided into four main stages: free nuclear (1–2 DAP), cellularization (3–5 DAP), storage substance accumulation (6–20 DAP, of which 6–9 DAP is aleurone and starchy endosperm differentiation), and maturity stage (21–30 DAP) [62,69].

The primary endosperm begins cell division 3.5 h after fertilization, undergoing a series of repeated, rapid nuclear divisions without cytoplasmic division to form a single endosperm cell with multiple free nuclei. The generated nuclei and surrounding cytoplasm are localized at the periphery of the embryo sac, and a large central vacuole occupies the central region of the embryo sac. This process lasts nearly 2 days. From 3 DAP, the cytoplasmic division begins from the periphery of the embryo sac to the center, and cell walls are produced; this process is known as cellularization. This process lasts about 3 days and is basically complete in 5 days, resulting in endosperm cells that fully fill the embryo sac and are beginning to synthesize storage substances, such as starch and lipids [65,69]. From approximately 6 DAP, after endosperm cellularization, rice endosperm cells begin to differentiate. The outermost endosperm cells differentiate into the aleurone layer, whereas starch granules begin to accumulate rapidly from the middle to the sub-aleurone layer of the endosperm, developing into the inner starchy endosperm [69,70]. During the storage substance accumulation phase, the inner starchy endosperm accumulates starch and a small amount of protein. Starch accumulation in rice usually starts from the central area of the starchy endosperm and moves towards the peripheral sub-aleurone layer. In contrast, the outer aleurone layer mainly accumulates lipids, proteins, and minerals, with almost no starch. Storage substance accumulation continues until 20 DAP, after which the endosperm enters the maturity phase, during which dry matter growth stagnates and water content decreases until maturity [65].

Synchronously, caryopsis morphology and size also change during development. Due to water and nutrient absorption, the internal cell expansion drives the longitudinal growth of the caryopsis within 6 DAP, while lateral transverse expansion mainly occurs from 5 to 12 DAP [67,69]. The rapid accumulation of endosperm storage substances increases the grain width and thickness, which peak at 9 and 12 DAP, respectively, with no further significant changes [67,69]. Grain filling is mostly completed by 21 DAP, and it strongly affects the final rice yield and quality [69]. In addition, rice endosperm growth is driven by cell proliferation within 9–10 DAP, and endosperm growth then mainly depends on cell volume expansion [71,72].

Although the aleurone layer and starchy endosperm share the same developmental origin, their cellular fates are distinct. In the later stage of caryopsis development, with starch accumulation and increased grain fullness, the organelles and nuclei in the cells of starch endosperm begin to undergo programmed death. The starch endosperm cells lose activity and eventually disappear, but this process can help integrate carbohydrates, amino acids, and storage proteins more effectively, making the grains fuller and increasing the grain weight. In contrast to the inner starchy endosperm, aleurone cells do not undergo programmed death or degradation during grain maturation; thus, the aleurone layer is a biologically active endosperm tissue that plays an important physiological role in seed dormancy and germination. Finally, in the mature caryopsis, the embryo and aleurone layer remain living tissues, while the starchy endosperm, and maternal tissues such as the pericarp and testa, are dead [64,73].

4. miRNA-Mediated Regulation of Agronomically Important Grain Traits

In rice, almost all agronomical traits are regulated by miRNAs. Grain development is a complex biological process, and miRNA regulatory modules play critical roles in grain development. The modules that determine multiple functional traits are potential tools in future breeding programs. The hull and endosperm are two prominent targets for improving grain yield and quality. Through an exhaustive literature survey, we collected and screened all functional miRNAs based on their regulation of agronomically important rice grain traits (Table 1; Figure 1).

4.1. miRNAs Regulate Hull Development to Affect Grain Size

The development of rice floral organs, especially the hull, plays a decisive role in the formation of grain morphology. The spikelet hull sets the volume of the cavity, provides limited space for the caryopsis to develop, and determines the final grain size. An intact rice floret is generally divided into four whorls of floral organs from the outside to the inside: palea and lemma, serosa, stamen, and pistil. In rice, the characteristic development of floral organs is regulated by the ABCDE model. Class A genes mainly regulate lemma and palea development and jointly regulate serosa development with class B genes. Class B genes, along with class C genes, regulate stamen development. Pistil development is mainly controlled by class C genes. Ovule formation is determined by class D genes. Class E genes are essential for the development of the floral organs of all whorls, rather than the characteristics of a certain whorl. Class A genes include OsMADS14, OsMADS15, and OsMADS18 [74,75], which regulate normal lemma and palea structures.

On the basis of normal flower development, there are some phenotypic variations in the hull size and color. Hull color depends on pigment accumulation in the hull. Rice pigments are controlled by internal genetic factors via diverse systems in different tissues. Pigment-related genes refer to structural and regulatory genes. Structural genes are responsible for catalyzing pigment biosynthesis, while regulatory genes regulate the expression of structural genes in different tissues. The pigment content in hulls alters hull color from purple, red, or brown to yellow, which might be useful for mechanized commercial hybrid rice seed production [76,77].

However, the final grain size and weight are determined by hull development and grain filling. A number of quantitative trait loci (QTLs) that affect grain size and weight have been identified [59,78,79]. Many miRNAs have been reported for grain size and weight, often showing pleiotropic effects on other yield-related traits, such as plant and panicle architecture [3]. miRNAs that affect grain size and weight in rice have been summarized here (Table 1).

4.1.1. SPL-Related miRNAs

miR156 is an important grain yield regulator in rice. Most squamosa promoter binding protein-like (SPL) genes are putative targets of miR156, and the miR156-SPL module plays an important role in plant growth and development [80,81,82]. miR156 contains 11 members that can be divided into two subfamilies: one includes miR156a/b/c/k/l, a CRISPR/Cas9-mediated mutation that enhances seed dormancy with little effects on yield traits, including grain size; whereas another includes miR156d/e/f/g/h/i, the same mutation that modifies the plant architecture and increases the grain size but has little effect on seed dormancy [83].

miR156 overexpression resulted in abnormal plant morphology, including dwarfism, an increased number of tillers, and a reduced number of grains per panicle [80]. For grain size, miR156 targets the 3′-untranslated region of OsSPL13/GLW7 and downregulates GLW7 expression in panicles. GLW7 positively increases the grain length and yield by regulating the cell size in the hull [84]. OsSPL16/GW8 positively the regulates grain width but negatively regulates the grain length. An allelic variant at the miR156 target site in OsSPL16/GW8 abolishes the suppression of miR156 [85].

OsSPL18 increased the grain width and thickness by affecting cell proliferation in spikelet hulls. OsSPL18 binds to the DEP1 promoter, positively regulating DEP1 expression. OsSPL18 can be cleaved by miR156k [86]. OsSPL4 regulates the grain size by promoting cell division in the hull and is cleaved by miR156 [87,88].

miR529 shares 14-nt homology with miR156, and, therefore, has similar functions to miR156 in rice, but it also has differences in the number of members, the targeted SPL genes, and the expression pattern [88]. The overexpression of miR529a, but not miR529b, produces similar phenotypes in rice yield traits to that of miR156 [88,89]. miR156 and miR529a are dominantly expressed in different stages and thereby coordinately control the number of rice tillers, plant height, panicle architecture, and grain size by targeting OsSPL14 [88]. miR529a may promote the grain length by altering the expression of OsSPL2/7/14/16/17/18 [90].

miR535 shares high similarity with miR156 and miR529 and thus targets the same SPL transcription factor gene family. miR535 is highly expressed in young panicles. miR535 overexpression confers rice blast resistance, modifies plant architecture, and increases grain length, which might be through repressing the expression of OsSPL7/12/16 [91] and OsSPL14 [92].

4.1.2. GRF-Related miRNAs

In addition, most growth-regulating factors (GRFs) are targets of miR396. miR396-GRF modules play important roles in grain shape determination and grain yield. Rice contains five miR396 isoforms, encoded by eight genes. miR396e and f have the highest expressions in the inflorescence and spikelet, and miR396c has some expression in these tissues [93].

miR396e and miR396f are important regulators of grain size and plant architecture, and mutations in them result in an increased grain size, weight, and yield [94]. miR396c overexpression decreases the grain size and weight [95]. OsGRF4/GL2/GS2/GLW2, interacting with OsGIFs, positively regulates grain size by promoting both cell division and cell expansion in spikelet hulls. When the 2-bp substitution perturbs the cleavage of GL2/GS2/GLW2 by miR396, this results in an elevated expression of GL2/GS2/GLW2 and thus the increase in grain size [95,96,97,98].

Downregulating miR396 increases grain length, while the grain size of OsGRF8 overexpression was most similar to that of miR396 suppression, suggesting that OsGRF8 is a target of miR396 in grain size regulation [93,99]. miR408 is an embryo-specific miRNA that positively regulates grain size. OsGRF8 directly binds the miR408 promoter, and the mutant phenotype of miR396 is partially complemented by the mutation in miR408, suggesting that the miR396-GRFs-miR408 module affects the grain shape [99]. miR408 targets the uclacyanin gene OsUCL8, downregulating its expression and therefore affecting photosynthesis [100]. miR408 overexpression in rice enhances photosynthesis. Therefore, miR408 overaccumulation in plants results in higher rates of vegetative growth and has an effect on grain enlargement [101].

miR396d overexpression results in defects in floral organ development, including open husks, long sterile lemmas, and altered floral organ morphology. miR396d regulates the expression of OsGRF6/10 genes, which function with OsGIF1 by activating OsJMJ706 and OsCR4, affecting spikelet development [102].

4.1.3. Other TFs Related to miRNAs

GAMYB, which encodes an MYB transcription factor (TF), is a component of gibberellin (GA) signaling in the aleurone and anther, where miR159 has not been observed. But in vegetative tissues, miR159 is co-expressed, and GAMYB is silenced. GAMYB is the only conserved target of miR159 [103]. miR159-GAMYB has different functions in vegetative tissues. miR159 positively regulates organ size, including grain size, by promoting cell division, which might involve the cell cycle and hormone homeostasis [104]. miR159 and GAMYB regulate both brassinosteroid (BR) and GA responses, including grain size alteration [105]. miR159 targets GAMYB and GAMYBL2. GAMYBL2 directly binds to the GS3 promoter and represses its expression. Suppressing miR159 or overexpressing GAMYBL2 and GS3 produces similar dwarf and small grains [106].

The overexpression of miR172 forms did not show differences in plant height or tillering but a severe phenotype was observed in the panicle [89], including an elongated lemma and palea and unclosed hull, resembling the osmads1 mutant phenotype. Overexpressing each AP2 gene causes a shortened lemma and palea, and overexpressing AP2 partially rescues the phenotype of OsMADS1-RNAi plants. OsMADS1 acts upstream of miR172s to suppress their expression. OsMADS1, miR172s, and AP2s form a regulatory network for rice floral organ development, especially lemma and palea elongation [107].

miR5506 plays an essential role in the regulation of the number of floral organs, spikelet determinacy, and female gametophyte development in rice. miR5506 overexpression caused pleiotropic abnormalities, including an over- or underdeveloped palea. miR5506 targeted LOC_Os03g11370, a member of the reproductive meristem transcription factor gene family, regulating reproductive development [108].

miR166 knockdown resulted in multiple phenotypic alterations, including rolled leaf, plant dwarfism, and grain size changes. miR166 targets OsHB4, which is a member of the HD-Zip III transcription factor family. A rolled leaf phenotype has been observed in seedlings overexpressing OsHB4 [15].

4.1.4. Auxin-Related miRNAs

Auxins play a crucial role in grain size determination. miR167 has been implicated in auxin signaling through the regulation of the expression of ARF, which participates in grain development [109]. The miR167a-OsARF6-OsAUX3 module regulates grain length. OsARF6 binds directly to the auxin response elements of the OsAUX3 promoter, thereby negatively controlling grain length by altering longitudinal expansion and the auxin distribution in glume cells. Furthermore, miR167a positively regulates grain length by directing OsARF6 mRNA silencing. miR167a overexpression resulted in longer grains, similarly to osarf6 mutants. In contrast, the targeted mutation of miR167a resulted in a lower grain weight [110]. miR167 targets OsARF12 for rice grain filling and grain size regulation. OsARF12 activates the expression of OsCDKF;2, which mediates auxin and BR signals and is involved in cell cycle processes, resulting in increased grain filling and grain size. In addition, the miR167-OsARF12 module acts downstream of miR159 through GAMYBL2 binding to the miR167 promoter [111].

miR160 and miR393 also contribute to grain development. miR160 targets OsARF18, whereas a long noncoding RNA temporally and competitively binds to miR160, thereby attenuating the repression of miR160 on OsARF18, and influencing rice seed setting and seed size. ARF18 overexpression has pleiotropic effects including small grains [112,113]. miR393 overexpression causes a severe phenotype, including hulls that fail to close, resulting in a smaller dehusked caryopsis. miR393 targets OsTIR1 and OsAFB2, which interact with OsIAA1. The suppressed expression of OsTIR1 and OsAFB2 causes phenotypes that resemble miR393 overexpression [114].

4.1.5. Functional Protein-Related miRNAs

miR397 is highly expressed in young panicles and grains. miR397 overexpression increased rice yield by promoting grain number, and larger and heavier grains, similarly to the knockdown of laccase OsLAC. miR397 mostly negatively regulates the expression of OsLAC, which functions in lignin synthesis and various developmental processes by regulating BR signaling [115,116]. The miR397-OsLAC module promotes grain filling through BR signaling [117]. OsLAC directly binds to OsTTL and prevents phosphorylation-mediated degradation. OsTTL is a negative regulator of BR signaling. Overexpressing OsTTL decreased the BR sensitivity in rice, and the loss of function of OsTTL led to enhanced BR signaling and increased grain yield [118]. The expression pattern of OsAGO17 is consistent with miR397b. OsAGO17 can increase the grain size and weight. OsAGO17 and miR397b form a RISC, affecting rice development by suppressing LAC gene expression [56].

miR1848 targets OsCYP51G3, which encodes obtusifoliol 14α-demethylase, influencing phytosterol and BR biosynthesis. The overexpression of miR1848 or repression of OsCYP51G3 causes typical phenotypic changes related to phytosterol and BR deficiency, including shorter grains [119].

miR5504 has a pleiotropic effect on the plant height, grain yield, and quality, and regulates grain size by affecting cell expansion. An unknown expressed protein LOC_Os08g16914 was a target of miR5504. LOC_Os08g16914 overexpression showed a significant decrease in plant height and grain weight. Thus, miR5504 affect grain yield-related traits by targeting LOC_Os08g16914 [120].

miR530 negatively regulates grain size, as miR530 overexpression significantly decreases grain size by controlling cell division and expansion in spikelet hulls. OsPIL15 activates miR530 expression by directly binding to its promoter. miR530 targets OsPL3, which encodes a PLUS3 domain-containing protein. OsPL3 knockout decreases the grain yield [121].

miR398 overexpression increases the panicle length, grain number, and grain size in rice, while the OsCSD2 transgenic line shows some phenotypes similar to miR398 suppression on panicle traits but not on grain size. Therefore, miR398 has compounded effects through different target genes [122].

In conclusion, many miRNAs have been identified for grain size and weight, revealing that miRNA likely interacts with hormone signals and transcription factors, therefore regulating grain yield.

4.2. miRNAs Regulate Storage Substance Accumulation in the Endosperm

Rice endosperm development refers to the process of differentiation and the development of endosperm cells. In contrast, grain filling focuses on the process whereby photoassimilates are transported to the developing caryopsis and converted into storage substances, principally in the form of starch and protein that accumulate in the endosperm, directly determining rice yield and quality. Endosperm cellularization can provide a well sink for grain filling. Therefore, large grains may be an optimized way to improve yield in breeding practices (see Section 4.1). In addition, the accumulation and composition of storage substances during grain filling directly affect rice quality.

Many genes and hormones involved in endosperm development have been identified, which is also influenced by environmental factors [62,72]. In rice, technologies for identifying miRNAs include genetic screening, direct cloning, bioinformatics prediction, microarray, and high throughput sequencing, by which several sets of miRNAs specifically expressed in grains have been identified related to the different processes of grain development and filling [123,124,125,126,127,128,129,130,131]. Furthermore, degradome sequencing has been used for global miRNA target identification in plants [132]. However, only a small number of miRNAs involved in this process have been functionally validated. The complexity of these miRNAs determines the requirements for further study.

Starch is the main component of the rice endosperm, accounting for about 80%. miR5519 overexpression reduces the grain size, weight, and seed setting rate due to inadequate grain filling. miR5519 regulates grain filling by targeting and downregulating sucrose synthase gene RSUS2 in rice [133].

Several miRNAs target starch synthesis- or plant hormone homeostasis-related genes that participate in grain filling, but their functions have not been verified [126]. Plant hormones are a key factor affecting starch synthase activity, thereby regulating grain filling in rice. miR159, miR319, miR812, and miR819 target the genes involved in the regulation of plant hormone homeostasis [126].

miR1861 is highly accumulated during grain filling [134], and miR1861 targets genes that encode β-amylase OsBAM2/LOC_Os10g32810 [135] and starch binding domain-containing protein SBDCP1/LOC_Os01g63810 [126]. miRn45-5p targets a sucrose-phosphate synthase SPS, which influences sucrose synthesis [126]. miR1867 is highly correlated with rice grain filling [134]. miR143 and miR1867 target granule-bound starch synthase GBSSII/LOC_Os07g22930, which participates in starch synthesis, regulating the grain filling rate [135]. miR1428e-3p acts more indirectly by targeting two sucrose non-fermenting-1 related kinases SnRK1, with one kinase expressed in the endosperm and aleurone and playing a role in regulating starch accumulation [136]. miR1432 might be the key miRNA in grain filling by targeting alpha-amylase RAmy3D/LOC_Os08g36910 [137]. miR172 targets rice starch regulator 1, RSR1/LOC_Os05g03040, a starch biosynthesis inhibitor [123].

Protein is the second most important component of the endosperm, accounting for 5–8% of the rice endosperm. miR5144 targets OsPDIL1;1 in rice seeds and seedlings, mediating the formation of protein disulfide bonds. OsPDIL1;1 is a key regulator in protein folding. miR5144 overexpression or OsPDIL1;1 downregulation results in a lower protein–disulfide bond content, affecting the accumulation of storage protein and starch and producing a floury endosperm [138].

In addition to starch and protein, the components of the rice endosperm include lipids, vitamins, minerals, and other metabolites. Lipid accounts for about 0.3–0.6% of the rice endosperm. These nutrients are mainly retained in the aleurone layer rather than the endosperm. Fatty acids are the fundamental building blocks of lipids, supplying energy for humans, and they can be esterified to glycerol molecules to form lipids. Suppressing the expression of miR1432 significantly improves the grain weight by enhancing the grain filling rate. OsACOT is a major target of miR1432, as OsACOT overexpression resembles the suppression of miR1432. OsACOT encodes acyl-CoA thioesterase, which is a regulator of fatty acid biosynthesis and desaturation in developing seeds. The miR1432-OsACOT module regulates the grain weight by determining the grain filling rate in rice through the modulation of auxin and abscisic acid levels [139].

In summary, the regulation mechanisms of miRNAs are multi-step and multi-level complex processes. Some miRNAs have shown crosstalk in rice yield and quality; for example, AP2 and RSR1 are targeted by miR172. AP2 and RSR1 have been functionally characterized by being involved in panicle and hull development and starch synthesis, respectively [107,123]. Thus, this crosstalk may provide a clue for better understanding miRNA regulation.

Table 1. miRNAs that have been functionally verified to regulate rice grain development.

miRNA Family	Target Gene	Coregulatory Genes	Grain traits Affected by miRNA	Effect of miRNA on Grain Size, Weight or Quality	Regulatory Module	References
SPL-related miRNAs
miR156	OsSPL13/GLW7		Grain length and yield		miR156-OsSPL13	[84]
miR156	OsSPL16/GW8	GW7	Grain length and width		miR156-OsSPL16-GW7	[85]
miR156k	OsSPL18	DEP1	Grain width and thickness		miR156k-OsSPL18-DEP1	[86]
miR156	OsSPL4		Grain size	Positive	miR156-OsSPL4	[87,88]
miR529a	OsSPL14		Grain size	Positive	miR529a-OsSPL14	[88,89]
miR529a	OsSPLs		Grain length and width	Positive	miR529a-OsSPLs	[90]
miR535	OsSPL7/12/16, SPL14		Grain length	Positive	miR535-OsSPL7/12/16/14	[91,92]
GRF-related miRNAs
miR396	OsGRF4	GIF1, brassinosteroid-related genes	Grain size and weight	Negative	miR396-OsGRF4-GIF1	[95,96,97,98]
miR396	OsGRF8	GIFs	Grain size and panicle branch	Negative	miR396-OsGRF4/6/8-GIF1/2/3	[93,99]
miR396	OsGRF8	miR408	Grain size	Negative	miR396-OsGRF8-miR408	[99]
miR408	Copper proteins, OsUCL8		Photosynthesis and seed size	Positive	miR408-OsUCL8	[100,101]
miR396d	OsGRF6/10	OsGIF1	Floral organ development		miR396d-OsGRF6/10-OsGIF1	[102]
Other TFs-related miRNAs
miR159	GAMYB, GAMYBL2	GS3, BR and GA response genes	Organ size including grain size	Positive	miR159-GAMYB-GS3	[103,104,105,106]
miR172	AP2	OsMADS1	Panicle phenotype including hull size	Positive	OsMADS1-miR172s-AP2s	[89,107]
miR5506	LOC_Os03g11370		Hull morphology		miR5506-LOC_Os03g11370	[108]
miR166	OsHB4		Multiple phenotypes including grain size and weight	Positive	miR166-OsHB4	[15]
Auxin-related miRNAs
miR167a	ARF6	OsAUX3	Grain size and quality	Positive	miR167-ARF6-AUX3	[109,110]
miR167	ARF12	OsGAMYBL2, OsCDKF;2	Grain size and grain filling	Negative	OsGAMYBL2-miR167-ARF12-OsCDKF;2	[111]
miR160	ARF18	LncRNA	Seed size and seed setting		LncRNA-miR160-ARF18	[112,113]
miR393	OsTIR1, OsAFB2	OsIAA1	Hull failed to close		miR393-OsTIR1/OsAFB2-OsIAA1	[114]
Functional protein-related miRNAs
miR397	OsLAC	brassinosteroid, OsTTL	Grain size and panicle branch	Positive	miR397-OsLAC-brassinosteroid	[115,116,117,118]
miR397b		OsAGO17	Grain size and weight	Positive	miR397b-OsAGO17	[56]
miR1848	OsCYP51G3	brassinosteroid	BR deficiency phenotype including grain size	Negative	miR1848-OsCYP51G3-brassinosteroid	[119]
miR5504	LOC_Os08g16914		Grain size	Positive	miR5504-LOC_Os08g16914	[120]
miR530	OsPL3	OsPIL15	Grain size and weight	Negative	OsPIL15-miRNA530-OsPL3	[121]
miR398	OsCSD2		Grain length, width, and weight	Positive	miR398-OsCSD2	[122]
miRNAs regulate storage substances accumulation in endosperm
mi5519	RSUS2		Grain filling	Negative	mi5519-RSUS2	[133]
miR159, miR319, miR812, miR819	Plant hormone homeostasis		Grain filling			[126]
miR1861	OsBAM2, SBDCP1		Grain filling		miR1861-OsBAM2, miR1861-SBDCP1	[126,134,135]
miR45-5p	SPS		Sucrose synthesis		miR45-5p-SPS	[126]
miR1867	GBSSII		Grain filling		miR1867-GBSSII	[134,135]
miR1428e-3p	SnRK1		Starch accumulation		miR1428e-3p-SnRK1	[136]
miR1432	RAmy3D		Grain filling		miR1432-RAmy3D	[137]
miR172	RSR1		Starch synthesis		miR172-RSR1	[123]
miR5144	OsPDIL1;1		Starch and protein content	Negative	miR5144-OsPDIL1;1	[138]
miR1432	OsACOT		Grain filling, fatty acid elongation	Negative	miR1432-OsACOT	[139]

Table 1. miRNAs that have been functionally verified to regulate rice grain development.

miRNA Family	Target Gene	Coregulatory Genes	Grain traits Affected by miRNA	Effect of miRNA on Grain Size, Weight or Quality	Regulatory Module	References
SPL-related miRNAs
miR156	OsSPL13/GLW7		Grain length and yield		miR156-OsSPL13	[84]
miR156	OsSPL16/GW8	GW7	Grain length and width		miR156-OsSPL16-GW7	[85]
miR156k	OsSPL18	DEP1	Grain width and thickness		miR156k-OsSPL18-DEP1	[86]
miR156	OsSPL4		Grain size	Positive	miR156-OsSPL4	[87,88]
miR529a	OsSPL14		Grain size	Positive	miR529a-OsSPL14	[88,89]
miR529a	OsSPLs		Grain length and width	Positive	miR529a-OsSPLs	[90]
miR535	OsSPL7/12/16, SPL14		Grain length	Positive	miR535-OsSPL7/12/16/14	[91,92]
GRF-related miRNAs
miR396	OsGRF4	GIF1, brassinosteroid-related genes	Grain size and weight	Negative	miR396-OsGRF4-GIF1	[95,96,97,98]
miR396	OsGRF8	GIFs	Grain size and panicle branch	Negative	miR396-OsGRF4/6/8-GIF1/2/3	[93,99]
miR396	OsGRF8	miR408	Grain size	Negative	miR396-OsGRF8-miR408	[99]
miR408	Copper proteins, OsUCL8		Photosynthesis and seed size	Positive	miR408-OsUCL8	[100,101]
miR396d	OsGRF6/10	OsGIF1	Floral organ development		miR396d-OsGRF6/10-OsGIF1	[102]
Other TFs-related miRNAs
miR159	GAMYB, GAMYBL2	GS3, BR and GA response genes	Organ size including grain size	Positive	miR159-GAMYB-GS3	[103,104,105,106]
miR172	AP2	OsMADS1	Panicle phenotype including hull size	Positive	OsMADS1-miR172s-AP2s	[89,107]
miR5506	LOC_Os03g11370		Hull morphology		miR5506-LOC_Os03g11370	[108]
miR166	OsHB4		Multiple phenotypes including grain size and weight	Positive	miR166-OsHB4	[15]
Auxin-related miRNAs
miR167a	ARF6	OsAUX3	Grain size and quality	Positive	miR167-ARF6-AUX3	[109,110]
miR167	ARF12	OsGAMYBL2, OsCDKF;2	Grain size and grain filling	Negative	OsGAMYBL2-miR167-ARF12-OsCDKF;2	[111]
miR160	ARF18	LncRNA	Seed size and seed setting		LncRNA-miR160-ARF18	[112,113]
miR393	OsTIR1, OsAFB2	OsIAA1	Hull failed to close		miR393-OsTIR1/OsAFB2-OsIAA1	[114]
Functional protein-related miRNAs
miR397	OsLAC	brassinosteroid, OsTTL	Grain size and panicle branch	Positive	miR397-OsLAC-brassinosteroid	[115,116,117,118]
miR397b		OsAGO17	Grain size and weight	Positive	miR397b-OsAGO17	[56]
miR1848	OsCYP51G3	brassinosteroid	BR deficiency phenotype including grain size	Negative	miR1848-OsCYP51G3-brassinosteroid	[119]
miR5504	LOC_Os08g16914		Grain size	Positive	miR5504-LOC_Os08g16914	[120]
miR530	OsPL3	OsPIL15	Grain size and weight	Negative	OsPIL15-miRNA530-OsPL3	[121]
miR398	OsCSD2		Grain length, width, and weight	Positive	miR398-OsCSD2	[122]
miRNAs regulate storage substances accumulation in endosperm
mi5519	RSUS2		Grain filling	Negative	mi5519-RSUS2	[133]
miR159, miR319, miR812, miR819	Plant hormone homeostasis		Grain filling			[126]
miR1861	OsBAM2, SBDCP1		Grain filling		miR1861-OsBAM2, miR1861-SBDCP1	[126,134,135]
miR45-5p	SPS		Sucrose synthesis		miR45-5p-SPS	[126]
miR1867	GBSSII		Grain filling		miR1867-GBSSII	[134,135]
miR1428e-3p	SnRK1		Starch accumulation		miR1428e-3p-SnRK1	[136]
miR1432	RAmy3D		Grain filling		miR1432-RAmy3D	[137]
miR172	RSR1		Starch synthesis		miR172-RSR1	[123]
miR5144	OsPDIL1;1		Starch and protein content	Negative	miR5144-OsPDIL1;1	[138]
miR1432	OsACOT		Grain filling, fatty acid elongation	Negative	miR1432-OsACOT	[139]

5. Conclusions and Prospects

5.1. miRNAs in Rice Agronomic Trait Improvement

Rice is a vital food crop that requires continuous improvement to meet the food demands of the growing global population. Grain development is one of the most important processes in determining the final grain yield and quality, and many miRNAs have been reported to participate in this process in rice. Therefore, miRNA can act as a breeding tool for developing good varieties by increasing stress tolerance, yield, and quality. If miRNA positively regulates a favorable trait, then it is preferable to overexpress miRNA. If an miRNA produces an undesirable trait, then the approach may be to generate miRNA cleavage-resistant target genes or to silence miRNA expression. The miRNAs summarized here (Table 1) positively or negatively regulate grain size, weight, or quality [3,123], with the impact of different miRNAs. Knocking out key genes in the development process often produces severe phenotypes, whereas miRNA can moderately regulate gene expression within a safe range, enabling plant development to be fine-tuned. Therefore, miRNA modules have great potential in improving the agronomic and economically important traits of rice [9]. Current approaches for miRNA fine-tuning are the use of genetic engineering tools, especially emerging clustered regularly interspaced short palindromic repeat (CRISPR)/Cas9 technology, which enables the identification of alleles for trait enhancement, thereby generating desirable and heritable trait changes and eliminating concerns about the safety of genetically modified crops. For instance, the CRISPR/Cas9-based miR396 mutation enhanced the rice grain yield [93,94].

However, candidate miRNAs for crop improvement may have pleiotropic effects. These effects may be beneficial for improving target traits but may be penalized for other traits. Although increased grain size means a large sink, it does not always bring gains due to the tradeoffs associated with source–sink relationships. A successful example is that the tissue-specific downregulation of miR167 and miR1432 produces larger grains without adversely affecting plant development [140]. The differences between the mir167a mutant and endosperm-specific knockdown of miR167 might be due to the change in the grain filling path [110,140].

In addition, plant stress has become a major challenge due to its increased frequency and the negative effect on plant development, yield, and quality. Rice, especially in the grain filling stage, is sensitive to stress responses. miRNAs have critical regulatory roles in attenuating stress effects [11,13]. Notably, some miRNAs do not work independently but appear to coordinately regulate plant development and stress responses [9,141]. Therefore, whether miRNAs function as node molecules for optimal development is worth further exploration.

5.2. Research on miRNA Mining and Regulatory Networks

To date, numerous miRNAs have been identified in crops, suggesting the divergence and conserved roles of miRNAs in modulating plant development [9,57,142]. These conserved miRNAs associated a target trait with further function validation across plants. In addition, other species-specific miRNAs tend to evolve neutrally. In rice, however, only a small number of miRNAs have been functionally validated, and the exact functions of most miRNA modules are not yet clear. Therefore, further studies should focus on uncovering novel miRNA target modules through integrative multi-omics and gene editing technologies, while deeply exploring their roles in regulating rice yield and quality, thereby contributing to food security.

These verified targets of miRNAs include diverse regulatory genes, with a large proportion being transcription factors that further regulate downstream functional genes [7,9,11]. The regulatory functions of miRNAs comprise a complex network that is composed of three layers: one miRNA targeting distinct genes; distinct miRNAs on one trait; and the harmonization of different traits [9]. For example, miR156 controls rice grain size and panicle branching by targeting multiple SPL genes, which further regulate the downstream functional genes, and SPL genes and functional genes may interact either synergistically or antagonistically. In addition, different miRNA families may converge to regulate the same trait; for example, miR156, miR396, and miR397 are all involved in regulating rice grain size.

Thus, miRNAs and targets often exist in a many-to-many relationship, forming a complex network of miRNA target modules that coordinate different developmental processes and stress responses. An in-depth analysis of these known modules aids in dissecting the regulatory mechanisms of traits of interest and offering breeders suitable targets to modify agriculturally important traits. The miRNA functional networks need to be investigated at finer tissue and cellular levels to reveal their specific regulatory mechanisms across different developmental stages.