DEGENERATED LEMMA (DEL) Is a New Allele of OsRDR6 That Regulates Lemma Development and Affects Rice Grain Yield


 The lemma and palea are floral organ structures unique to grasses, and their development affects grain size. However, information on the molecular mechanism of lemma development is limited. In this study, we investigated a rice spikelet mutant, degenerated lemma (del), which developed florets with a slightly degenerated or rod-like lemma. The results indicate that the mutation of the DEL gene interfered with lemma development. In addition, del also showed a significant reduction in grain length and width, seed setting rate, and 1000-grain weight, which led to a reduction in yield. The results indicate that the mutation of the DEL gene further affects rice grain yield. Map-based cloning shows a single-nucleotide substitution from T to A within Os01g0527600/DEL, causing an amino acid mutation of Leu-34 to His-34 in the del mutant. DEL is an allele of OsRDR6, encoding the RNA-dependent RNA polymerase 6, and is highly expressed in the spikelet. RT-qPCR results show that the expression of some floral organ identity genes was changed, which indicates that the DEL gene regulates lemma development by modulating the expression of these genes. The present results suggest that DEL plays an important role in lemma development and rice grain yield.


Introduction
Rice (Oryza sativa L.) is a model monocotyledonous plant and is one of the most important food crops in the world. The oral organs of rice are closely associated with yield and quality. An individual oret is composed of the pistil, stamen, lodicules, palea, and lemma from the innermost to the outermost whorl.
The lemma and palea, also known as the glumes, are external oral organs unique to grasses and provide photosynthetic products for the oral organs in the early stages of development to protect the grain from attack by diseases and insects after maturity. In addition, the glume encloses the mature grain, and the size of the lemma and palea determines the length and width of the grain, thus the normal development of the glume is important for ensuring the normal development of the grain, and is closely related to the yield and quality (Ren et al., 2019;Shomura et al., 2008;Xing and Zhang, 2010).
The characteristic genes that regulate oral meristem identity maintain the normal development of the in orescence, and activate the expression of the related oral organ identity genes to produce specialized oral organ primordia and ultimately form the complete oret. The so-called ABCDE model for the regulation of oral organ development in higher plants has been proposed (Theißen, 2001). Class A genes are involved in the regulation of the lemma, three of which have been isolated in rice, namely OsMADS14, OsMADS15, and OsMADS18, which all belong to the SQUA gene family and the FUL subfamily. Expression of OsMADS14 can be detected in the glume and sterile lemma at the early stage of oral organ development and is expressed in the stamen and pistil at the later stage. Ectopic expression of OsMADS14 leads to an early-owering phenotype . In situ hybridization indicates that OsMADS15 is expressed in the lemma and palea; the mutant shows degeneration and albinism of the lemma, and overexpression of OsMADS15 results in early owering (Kyozuka and Shimamoto, 2002;Wang et al., 2010). OsMADS18 is widely expressed in all tissues and its absence does not cause phenotypic abnormalities, but its overexpression leads to early owering and accelerated tillering of the stem meristem (Moon et al., 1999). In addition, Class E genes are involved in the regulation of the lemma.
Class E genes in rice comprise OsMADS1 (LHS1), OsMADS5, OsMADS7, OsMADS8, and OsMADS34, which predominantly regulate the development of organs in the entire oret. OsMADS1 may maintain the development of the palea and lemma and the establishment of spikelet determinacy through the production of specialized apical meristems and the differentiation and proliferation of glume cells (Malcomber and Kellogg, 2004).
The development of the lemma and palea is also regulated by the trans-acting small interfering RNA (ta-siRNA) synthesis pathway. In contrast to the rst ve types of genes, the ta-siRNA synthesis pathway predominantly regulates lemma and palea development by affecting the establishment of polarity. The production of ta-siRNA is caused by the speci c miRNA-ARGONAUTE (AGO) complex, cleaving the miRNA targets on TRANS-ACTING SIRNA transcripts (Allen et al., 2005). Subsequently, the broken single strand is synthesized into dsRNA by RDR6 under the protection of the SGS3 protein (Adenot et al., 2006;Garcia et al., 2006). Then, under the action of the DCL4 and DRB4 proteins, it is cut into siRNA to inhibit the expression of downstream target genes (Yoshikawa et al., 2005). Therefore, ta-siRNA synthesis is controlled by three types of major proteins that all play important roles, namely the AGO protein that speci cally binds to miRNA; the RDR6 and SGS3 proteins involved in dsRNA formation; and the DCL4 and DRB4 proteins performing cleavage functions (Allen et al., 2005). Previous research has shown that mutations in the ta-siRNA synthesis pathway can cause defective polarity development of the glume. SHL2 encodes a protein in rice similar to Arabidopsis RDR6, which mediates the production of dsRNA. In the shl2 mutant, the stamens and lemma are defective in adaxial-abaxial polarity development, causing the lemma to become lamentous or stick-shaped, or to fail to develop entirely (Toriba et al., 2010). Rice SHO1 and SHO2 encode proteins closely related to Arabidopsis DCL4 and AGO7, respectively. Leaf growth of the sho1 mutant is accelerated and the leaves are deformed, showing short and narrow or lamentous shapes, and lemma development is hindered by the absence of adaxial surfaces (Nagasaki et al., 2007). The sho2 mutant exhibits leaf phenotypes similar to those of sho1, but lemma development has not previously been described (Song et al., 2012). Thus, genes associated with the ta-siRNA pathway are likely to be involved in the establishment of adaxial-abaxial polarity in rice.
In this study, we identi ed a rice DEGENERATED LEMMA (DEL) gene, a novel allele of OsRDR6, which encodes an RNA-dependent RNA polymerase, and is highly expressed in the spikelet, especially in the lemma. The del developed orets with slightly degenerated or rod-like lemma. Analysis of agronomic traits shows that grain length and width, seed setting rate, and 1000-grain weight of the del mutant were reduced signi cantly compared with the wild type. This suggests that the mutation of DEL further affected rice grain yield. RT-qPCR results show that the expression of some oral organ identity genes was changed, which indicates that the DEL gene regulates lemma development by modulating the expression of these genes. Taken together, these results suggest that DEL regulates lemma development in rice, revealing the important role of DEL in rice spikelet development and breeding.

Plant materials
The del mutant was derived from the progeny of a rice indica restorer line, Xinong 1B, treated with EMS and was stably inherited through seven successive generations of self-crossing. The

Molecular mapping of del and linkage map construction
To localize the target gene, the bulk segregant analysis method was used (Michelmore et al., 1991). All DNA used to localize the target gene was extracted using the etyltrime thylammonium bromide method (Murray and Thompson, 1980

Quantitative real-time PCR analysis
RNAprep Pure Plant RNA Puri cation Kit (Tiangen, Beijing, China) was used to isolate the total RNA of the lemma, rod-like lemma, palea, pistil, and stamen of the wild type and the del mutant. The extraction process followed the kit instructions and kept RNase free environment throughout the whole process. The purity and concentration of RNA were determined by NanoDrop™ One/One C (Thermo) and RNA integrity was determined by agarose gel electrophoresis. 2 ug of puri ed RNA was reverse-transcribed into cDNA using the PrimeScript® Reagent Kit with gDNA Eraser (TaKaRa). Half a microliter of the reversetranscribed RNA was used as a PCR template with gene-speci c primers (Table S4). Quantitative real-time PCR analysis was performed in three replicates using the SYBR® Premix Ex Taq™ II Kit (TaKaRa, Dalian, China) with the CFX Connect™ Real-Time System (Bio-Rad). ACTIN was used as an internal reference gene (Table S4). The complete reaction conditions followed the kit instructions. The average expression level was calculated for each gene. All samples were obtained from fresh plants in the eld, stored in liquid nitrogen, and transported to the laboratory. RNA extraction, reverse transcription and RT-qRCR experiments were completed within two days. Residual RNA samples were stored at -80°C and cDNA at -20°C.

In situ hybridization
Young panicles from the wild type were xed in 70%FAA (RNase free), then dehydrated through a series of alcohols/xylenes, before being embedded in para n (Sigma-Aldrich). For the DEL probe, gene-speci c cDNA was ampli ed and labeled using the DIG RNA Labeling Kit (Roche, Basel, Switzerland). The in situ hybridization was performed as described previously (Zhang et al., 2017). The primers are shown in Supplemental Table S5.

Results
Morphological and histological analysis of the del mutant at the owering stage The spikelet of the wild type contains one terminal fertile oret, one pair of sterile lemmas, and one pair of rudimentary glumes in rice ( Fig. 1-A). The terminal fertile oret is bisexual and consists of one pistil and six stamens, as well as two lodicules and one pair of bract-like organs (lemma and palea) ( Fig. 1-B and C). There are ve and three vascular bundles in the lemma and palea, respectively. In addition, the lemma is composed of four cell layers: from the abaxial to the adaxial surface these are a silici ed upper epidermis, sclerenchyma cells, parenchyma cells, and a vacuolated inner epidermis, and the palea is composed of two fused parts: the body of the palea (bop) and two marginal regions of the palea (mrp) ( Fig. 1-D and E). The cellular structure of the bop is similar to that of the lemma, but the mrp displays a distinctive smooth epidermis. The lemma and palea are hooked together to protect the inner oral organs and grain ( Fig. 1-F and G).
In comparison with the widle type, in the del mutant 36% of spikelets developed a slightly degenerated lemma ( Fig.1-H and N), which was narrow and unable to hook with the palea, causing a cracked oret ( Fig.1-H and J). The palea and inner oral organs developed normally ( Fig.1-I, K and L). Inspection of para n sections revealed that the four-layer cell structure of the lemma persisted, but the number and volume of the cells were signi cantly lower in the mutant ( Fig.1-M and N). In 53% of spikelets, a rod-like lemma formed ( Fig.1-O and T), and in these spikelets, the lemma was extremely degenerated and transformed into an awn-like organ ( Fig.1-O, Q and R). The morphology of the lemma was highly similar to that of the awn and contained only one midvein. The silici ed cells that should have been present on the outside surface of the wild type were ectopic and found in the inner epidermis of the rod-like lemma. The internal cell structure was almost completely disordered and unrecognizable ( Fig.1-T and U). The palea and inner oral organs in the mutant were normal ( Fig.1-P, S and T), thus the del mutant showed degeneration of the lemma to differing degrees, which suggests that DEL plays an important role in lemma development.

The del mutant affects grain yield
The spike type of the del mutant was more erect and smaller than that of the wild type. Compared with the wild type, the plant height of the del mutant was increased by 6.55 cm, while the panicle length was reduced by 24.5% ( Fig. 2-A, B, E and F). While the number of primary branches of del showed no signi cant difference, the number of secondary branches was reduced by 10.7% (Fig. 2-G and H). The rate of seed setting in del was reduced by 70% ( Fig. 2-I). As well as impaired ower development, the mature grains of del were also affected.The glumes of the mature grains in del were still dehiscent, and the lemma was rod-like in the severe mutant ( Fig. 2-C).The mature brown rice grains were shaped like water droplets, being narrow at the top and wide at the bottom, and the rod-like spikelets produced few grains (Fig. 2-D). The grain length and the brown-rice grain length of the del were reduced by 7.7% and 27.0%, respectively (Fig. 2-J and M). The grain width of the del was increased by 16.3%, but the brownrice grain width of the del was reduced by 26.8% (Fig. 2-K and N). The 1000-grain weight and the 1000-grain brown-rice weight were also reduced by 71.3% and 83.0%, respectively (Fig. 2-L and O). The results reveal that the DEL gene plays an important role in spikelet development and affects rice grain yield.
Early morphological analysis of the del mutant Using scanning electron microscopy, we examined spikelets of the wild type and del mutant during the early developmental stages. In the wild-type oret, the primordia of the lemma and palea began to develop during the spikelet 4 (Sp4) stage (Fig.3-A). At Sp4, no signi cant difference between spikelets of the wild type and the del mutant was observed (Fig.3-A, E, and I). In the wild-type oret, the lemma and palea primordia were formed and the palea was smaller than the lemma, the ve spherical stamen primordia were formed synchronously during the Sp5 and Sp6 stages except for the primordium nearest the lemma, and the lodicule primordia were formed at a subsequent stage (Fig.3-B). In the del mutant, which was different from the wild type, the shape of the slightly degenerated lemma was normal (Fig.3-F), and the margin of the rod-like lemma was not developed and did not intersect with the palea primordium ( Fig.3-J). During the Sp7 stage, the pistil primordium was initiated and the lemma and palea primordia were normally developed and semi-closed, enclosing the inner whorl organ primordia in the wild-type oret ( Fig.3-C). In the del mutant, the slightly degenerated lemma primordium was noticeably shorter and narrower than the lemma primordium in the wild type ( Fig.3-G), cell differentiation on each side of the rod-like lemma was slow, and development into a rod-like structure occurred gradually (Fig.3-K). During the Sp8 stage, the single oret of the wild-type spikelet underwent a further stage of development ( Fig.3-D). In the del mutant, the slightly degenerated lemma was signi cantly narrowed, and the palea and lemma were unable to close to cover the inner oral organs (Fig.3-H). The rod-like lemma primordium was further elongated, but the anks remained undeveloped, and the inner oral organs were completely bare. However, the inner oral organs were normal (Fig.3-L). Collectively, these observations reveal that the defects of the lemma in the del mutant arose during the early stages of spikelet development.

Map-based cloning of DEL
The F 1 progeny was derived from the cross between 56S and the del mutant, and showed a wild-type phenotype. The segregation ratio of the normal to the mutant phenotype was 3:1 (1091 wild-type-like plants and 358 mutant-like plants; χ 2 = 0.26 < χ 2 0.05,1 = 3.84) in the F 2 population. This result indicates that the del mutant phenotype was controlled by a single recessive gene.
A map-based cloning approach was used to ne-map the DEL gene. The 358 recessive individuals in the F 2 population were used as a mapping population to localize the DEL gene. To screen for polymorphism between the parents, a total of 420 pairs of simple sequence repeat (SSR) and InDel primers evenly distributed in the rice genome were used (Table S1). Ninety InDel markers were selected and used to further screen two DNA pools prepared by mixing equal amounts of genomic DNA from either 10 wildtype-like F 2 plants or 10 mutant-like F 2 plants (Table S2). The DEL gene was localized between the InDel markers In1-18.68 and In1-20.376 on chromosome 1 (Fig. 4-A and B). To ne-map DEL, 30 pairs of InDel primers located between In1-18.68 and In1-20.376 were developed, of which In1-18.68, In1-18.79, In1-18.94, In1-19.08, and In1-20.38 exhibited polymorphism between the parents (Table S3). Among all the F 2 individuals, these ve markers detected separately 14, 8, 4, 12, and 23 recombinants (Fig. 4-C). These results show that DEL was located between the InDel markers of In1-18.94 and In1-19.08. The estimated physical distance between In1-18.94 and In1-19.08 was approximately 140 kb, and 15 annotated genes (RAPdb annotation) were included within this interval: four enzymes (Os01g0527600, Os01g0527700, Os01g0528800, Os01g0529800), ve expressed protein (Os01g0528300, Os01g0528700, Os01g0529700, Os01g0530100, Os01g0530200), a transcription factor (Os01g0528000), two transposon protein (Os01g0527900, Os01g0529400), a non-protein coding transcript (Os01g0527801), a hypothetical protein (Os01g0530000), and a gene with no annotated information (Os01g0529101) (Fig. 4-D and Table S6).
Sequencing analysis showed that Os01g0527600 had a single-nucleotide substitution from T to A, which was an allele of OsRDR6. The substitution of the nucleotide substitution in the del mutant caused the amino acid mutation of Leu-34 to His-34.In order to con rm that the mutation of Os01g0527600caused the del mutant phenotypes, we cloned a fragment consisting of a 2144-bp upstream sequence from the start codon and 4890-bp coding region sequence of the Os01g0527600gene in the wild type into the pCAMBIA1301 vector with the green uorescent protein (GUS). The recombinant plasmid was then introduced into the del mutant ( Fig.4-F). Subsequently, a total of 13 complementary transgenic lines were obtained, and GUS staining showed that 7 lines of them appeared indigo blue. This indicated that these lines had correctly transformed and successfully transferred into exogenous target vectors. All the T 0 positive plants were planted in the elds, and the spikelet phenotype of these 7 lines was restored to the wild-type phenotype (Fig.4-G). These results con rm that the Os01g0527600 gene is the DEL gene, and the phenotype of del is indeed caused by mutation of the Os01g0527600 gene.

Spatiotemporal expression pattern of DEL
In order to clarify the functions of the DEL gene, we used RT-qPCR and in situ hybridization to test for DEL expression in the wild type. RT-qPCR analysis showed that DEL was constitutively expressed in whole organs, including root, stem, blade, sheath, and panicles. The lowest expression was found in the stem, and the highest expression was found in the blade. The expression level of the sheath was slightly higher than that of the stem, and the expression levels of the root and panicle were similar (Fig.5-A). In the spikelet, DEL was expressed in the lemma, palea, pistil, stamen, and lodicule ( Fig.5-B). Later, in situ hybridization was used to detect the expression pattern of DEL, and a strong signal was shown in the rice spikelet ( Fig.5-C-F). At the Sp3 stage, the DEL gene was highly expressed in the rudimentary glume, the sterile lemma, and the oral meristem ( Fig.5-C). At the Sp4 stage, a strong DEL signal was detected in the lemma and palea primordia and oral meristem (Fig.5-D). At Sp5-Sp8, DEL was expressed in the lemma, palea, stamen, pistil, and lodicule (Fig.5-E-F). These results indicate that DEL was mainly expressed in the spikelet and orets, and was highly expressed throughout the period of plant growth, which implies its role in regulating the development of the lemma. Because del showed a phenotype of lemma degeneration to differing degrees, we also explored the expression of DEL in the wild-type glume. The results show that DEL was highly expressed in the whole of the lemma and palea, especially in their eight vascular bundles (Fig. 5-F).

Expression analysis of oral organ identity genes in the del mutant and wild type
Given that the del mutant showed an abnormal phenotype of oral organs, the expression patterns of known genes associated with oral organ identity were measured. DL (which is expressed in the middle of lemma)expression was 150% higher in the slightly degenerated lemma of the del mutant than in the lemma of the wild type, but the expression level of DL in the rod-like lemma was greatly reduced, with only a basal level similar to that of the wild-type palea (Fig.6-A). Moreover, the expression level of OsMADS1 in both the slightly degenerated and the rod-like lemma was higher than that of the wild-type lemma, by a factor of 2 and 12, respectively. The expression level of OsMADS1 in the del palea was only twice as high as that of the wild type (Fig.6-B). OsMADS14 expression was similar to OsMADS1, and the expression of the slightly degenerated and the rod-like lemma was much higher than that of the wild-type lemma. There was no signi cant difference between the expression of the palea in the wild type and del ( Fig.6-C). For the expression level of OsMADS15, there was little difference between the rod-like and the wild-type lemma, but the expression level of the slightly degenerated lemma was 2.5 times that of the wild-type lemma, and the expression level of the palea in del was 1.5 times that of the wild-type palea (Fig.6-D). There was no difference in expression between the two types of del lemma and the wild-type lemma, and the expression of the palea in the del palea was about 70% higher than that of the wild-type palea (Fig.6-E). These results suggest that the slightly degenerated lemma retained the identity of the lemma, and degeneration of the rod-like lemma in the del mutant may be caused by reduced expressionof DL.OsMADS6 (which is an mrp identity gene) was expressed in the palea of the wild type and del mutant, which indicates that the palea identity in the del mutant was normal (Fig.6-E). In addition, we also assessed the expression of OsMADS2, OsMADS3, and OsMADS4 (the stamen and pistil genes). Compared with the wild type, the expression of OsMADS2 and OsMADS3 was reduced by a factor of 2 and 4 in the pistil of del, and was increased by a factor of 111 and 59 in the stamens of del (Fig.6-F and G). The expression level of OsMADS4 was slightly higher in the del than in the wild-type pistil, but was 50 times higher in the del than in the wild-type stamen, which indicates that the stamen and pistil of the del mutant showed normal identities (Fig.6-H). Collectively, these results indicate that the del mutation plays a critical role in the regulation of DL for lemma identity.

Discussion
In our study, we identi ed a mutant del, which contains a mutant allele of OsRDR6, in order to gain a better understanding of the functions of OsRDR6. The del mutant had the same phenotypes as the reported osrdr6 mutants, such as cracked glumes, caused by the different degrees of degradation of the lemma (needle/rod/awe-like structure). Moreover, the del mutant also exhibited an altered grain size. We suggest that the different backgrounds and mutation sites of the del mutant and the osrdr6 mutants could result in additional phenotypes. SHL2 encodes OsRDR6, eight shl2 alleles (shl2-1 through shl2-8) from O. sativa cultivar Taichung 65 have been identi ed, and most alleles have a nonsense mutation, frameshift mutation, or amino acid substitution in the conserved RdRP domain. The shl2 mutants reported to date are embryonic or seedling-lethal, caused by a failure in the formation of the SAM in the embryo. For the osrdr6-1 mutant from japonica cultivar Zhonghua11, a single nucleotide transition from G to T in osrdr6-1 leads to substitution of a highly conserved tryptophan to cysteine. Most spikelets in osrdr6-1 showed needle-shaped, awn-like lemma and altered stamen number. The shl2-rol mutant from the japonica cultivar Nipponbare has an amino acid substitution in the N-terminal region far from the RdRP domain. In the shl2-rol spikelets, morphological defects were observed in the lemma (needle-like structure), palea, and stamen, resulting in a lack of organs or a reduction in the number of them. By comparison, we considered the del from the indica restorer line, Xinong 1B (a local variety of Chongqing). Substitution of the nucleotide in the del mutant caused amino acid mutation of Leu-34 to His-34 in the Nterminal region far from the RdRP domain. The del mutant displayed degenerated lemma and abnormal grains, whereas the palea and all the inner whorl organs (stamen, pistil, lodicule) developed in normal shapes and numbers. This study provides some new insights into the role of OsRDR6, allowing a better understanding of the development and formation of the lemma and grain.
Improved grain yield is an important goal for basic and applied scienti c research in plants (Ren et al., 2019). Although several genes that affect grain size in rice have been identi ed, the molecular mechanisms remain unclear. Thus, it is still di cult to improve grain yield using these reported genes in rice. Normal development of oral organs, speci cally the glumes (i.e., lemma and palea) is the basis for ensuring the normal development of the grain, and is also closely related to the yield (Ren et al., 2019;Shomura et al., 2008;Xing and Zhang, 2010). Grain size is also an important factor in determining rice yield. In our study, compared with the wild type, the grain length and the brown-rice grain length of the del were reduced by 7.7% and 27.0%, respectively. The grain width of the del was increased by 16.3%, but the brown-rice grain width of the del was reduced by 26.8%. The 1000-grain weight and the 1000-grain brown rice weight were also reduced by 71.3% and 83.0%, respectively. The reduced grain or brown-rice grain length and width in del may be due to the degenerated lemma, which provides photosynthetic products for the oral organs in the early stages of development, to protect the grain from attack by diseases and insects after maturity, and determines the length and width of the grain. Taken together, the normal function of the DEL gene is the basis for the normal development of lemma morphology and size, and has an important in uence on rice grain yield. DEL is thus a possible target for efforts to improve rice grain yield.
The lemma and palea are specialized oral organs that are unique to grasses. The development of the lemma has been an important focus of research on oral organ development in rice. In previous research, LHS1 (OsMADS1) was shown to regulate the formation and development of the lemma, predominantly by controlling the differentiation of speci c cells of the lemma . The complete loss-offunction of OsMADS1 results in a homologous transformation of the three internal whorls of oral organs (lodicules, stamens, and pistil) into a lemma and palea-like structure (Agrawal et al., 2005;. The DL gene is a member of the YABBY family. The phenotype of the dl-sup6 mutant differs from that of other mutants in that a DNA fragment is inserted into the second intron of DL, resulting in loss of expression of DL and a change in the number of vascular bundles of the lemma (Ohmori et al., 2011).
Compared with the aforementioned mutants, the del mutant displays a lemma degenerated to different degrees, including a rod-like lemma, and the surface of the lemma is similar to that of the awn. However, the palea and inner oral organs are normally developed. DEL is the allele of OsRDR6, which encodes an RNA-dependent RNA polymerase, and is essential for ta-siRNA synthesis in rice. Genes associated with the ta-siRNA pathway are likely to be involved in the establishment of adaxial-abaxial polarity. Rice SHO1 and SHO2 encode proteins closely related to Arabidopsis DCL4 and AGO7, respectively. Leaf growth of the sho1 mutant is accelerated and the leaves are deformed, showing short and narrow or lamentous shapes, and lemma development is hindered by the absence of adaxial surfaces (Nagasaki et al., 2007). The sho2 mutant exhibits leaf phenotypes similar to those of sho1 (Song et al., 2012). OsDCL4 is responsible for the formation of the 21-nucleotide-long siRNA. In the osdcl4-1 mutant, the lemma shows degeneration and widespread conversion to an awn (Liu et al., 2007). Similar to the above mutations in ta-siRNA synthesis pathway, the lemma of the del mutant was also degenerated to varying degrees, being partially or completely degenerated to the awn. Therefore, the function of the ta-siRNA synthesis pathway is conserved, in regulating lemma and palea development by affecting the establishment of polarity.

Conclusion
In this study, we identi ed a mutant, named del, related to the spikelet development with degenerated lemma. In addition, del also showed a signi cant reduction in grain length and width, seed setting rate, and 1000-grain weight, which led to a reduction in yield. Mapping, cloning, and sequencing revealed that DEL is an allele of OsRDR6, encoding the RNA-dependent RNA polymerase 6, and is highly expressed in the spikelet. The expression analysis indicated that the DEL gene regulates lemma development by modulating the expression of some oral organ identity genes. The present results suggest that DEL plays an important role in lemma development and rice grain yield.
Wenwen Xiao and Li Ye. Phenotypic experiment was performed by Qiannan Duan, Xinfang Zhang and Jun Zhang. Wenqiang Shen and Zhifeng He made the creation of new software used in the work. Data collection and analysis were performed by Jing You, Yan Xiang and Shunyi Song. The rst draft of the manuscript was written by Jing You and Ting Zhang and all authors commented on previous versions of the manuscript. All authors read and approved the nal manuscript. Guanghua He and Yunfeng Li agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.   Spikelet phenotypes of the wild type and the del mutant. A-D, Spikelet of the wild type at Sp4-8, The del mutant spikelet at ve developmental stages: E and I, Sp4; F and J, Sp5-Sp6; G and K, Sp7; H and L, Sp8. le, lemma; pa, palea; fm, oral meristem; st, stamen; rg, rudimentary glume; sl, sterile lemma; lo, lodicule; sdle: slightly degenerated lemma organ; rlle: rod-like lemma organ. Bars = 1 mm.  Spatiotemporal expression pattern of the DEL gene.
A and B, RT-qPCR of DEL. ACTIN was used as a control. Young panicles < 0.5 cm, 0.5-1.0 cm, 1.0-2.0 cm, vegetative organ, and oral organ of the wild type were used. Data are Mean ± SD (n = 3). C-F, Expression pattern of DEL in spikelets of the wild type. In situ hybridization in the spikelets of the wild type during stages Sp3 (C), Sp4 (D), Sp5-7 (E) and Sp8 (F). fm, oral meristem; le, lemma; pa, palea; st, stamen; pi, pistil; rg, rudimentary glume; sl, sterile lemma. The black arrows indicate the vascular bundles. Bars=100um.