Introduction

Rice is grown on all the continents in the world except Antarctica and makes up the majority of the source of calories for half the world’s population. Different anthocyanin pigments appear in photosynthetic and reproductive tissues and organs in rice. Diverse grades of purple pigmentation mostly appear in the apiculus, stigma, leaf blade, root, leaf sheath, and root. Anthocyanin accumulation in various tissues plays a key role in the stimulation of hormonal responses, protection from ultraviolet radiation, and biotic stress resistance and abiotic stresses tolerance (Chalker-Scott 1999; Ithal and Reddy 2004; Lin-Wang et al. 2010; Reddy et al. 1995; Saitoh 2004). Purple apiculus trait is commonly seen not only in wild species but also frequently observed in landraces and cultivars of rice. Purple apiculus attracts insects and animals for pollination and seed dispersal (Chin et al. 2016). It is also an important morphological marker for varietal selection and purity determination (Saitoh 2004).

Anthocyanins are pigment-producing secondary metabolite molecules classified as part of flavonoid (Reddy et al. 1995). The genetic basis of anthocyanin pigmentation has been widely studied in the past and several genes have been identified in rice, maize, Petunia, and model plants like Arabidopsis (Dooner et al. 1991; Brenda 2001; Koes et al. 2005). The expression and function of genes encoding enzymes that directly catalyze the conversion of intermediates in the anthocyanin biosynthesis pathway is modulated by numerous activators and transcription factors. The transcriptional activator genes C1 and Pl that encode proteins with MYB DNA binding domains and B and R genes that code for proteins containing a basic helix-loop-helix (bHLH) domain are transcriptional activators and regulate downstream genes of the anthocyanin metabolism pathway in maize (Chandler et al. 1989). The OsC, OsB1, and OsB2 genes were identified from rice based on sequence similarity with maize C1 and B genes and regulated pigmentation pattern in different tissues (Reddy et al. 1995; Sakamoto et al. 2001). In addition to the C and B genes, the tissue-specific pigmentation patterns in rice were controlled by R, B, Lc, and Sn genes (Ludwig and Wessler 1990). Other genes like Pl, Rc, and Rd regulated purple leaf and red pericarp formation; whereas kala4 is responsible for black rice trait (Furukawa et al. 2007; Gao et al. 2011; Liu et al. 2012; Maeda et al. 2014; Oikawa et al. 2015; Sakamoto et al. 2001). In addition to Rc and Rd, Kala1, and Kala3 genes were responsible for the formation of the purple pericarp (Maeda et al. 2014; Reddy et al. 1995). A sequence rearrangement in the promoter of Kala4 resulted in for the formation of the purple pericarp trait in rice (Oikawa et al. 2015). The red pericarp trait was regulated by Rc gene that encodes bHLH together with functional Rd gene that encodes dihydroflavonol reductase (Furukawa et al. 2007; Sweeney et al. 2006), also, the pl gene which was responsible for the purple coloration of the leaf encodes bHLH protein (Sakamoto et al. 2001). The bHLH domains of regulatory genes make protein–protein interaction with OsC that has an R2R3 MYB domain and activated the expression of the structural anthocyanin biosynthesis genes that directly participate in the anthocyanin production pathway (Koes et al. 2005; Sakamoto et al. 2001).

Basically, anthocyanin pigmentation required network of C (chromogen), A (activator), and P (purple, tissue-specific regulator), and other genes that determine tissue-specific distribution of coloration (Fan et al. 2007; Mikami et al. 2000; Saitoh 2004; Sakamoto et al. 2001; Sun et al. 2018; Zhao et al. 2016). Anthocyanin pigmentation in the grain of rice is dependent on a single-copy gene, OsC (chromogen in rice) in the presence of other functional activators, tissue-specific regulators, and structural anthocyanin biosynthesis genes (Fan et al. 2007; Mikami et al. 2000; Saitoh 2004; Zhao et al. 2016). Functional bHLH was required for activation of the structural anthocyanin biosynthesis gene expression together with OsC in rice (Sun et al. 2018). In the past, the genetic basis of apiculus coloration was widely investigated (Fan et al. 2007; Mikami et al. 2000; Saitoh 2004; Sakamoto et al. 2001; Sun et al. 2018; Takahashi 1957, 1982; Zhao et al. 2016). Apart from the basic gene OsC, the responsible activator and tissue-specific regulator genes for purple apiculus trait were not fully identified. In the present study, a new purple apiculus mutant, Ospa, was generated from wild-type (WT), Zhenong 34 (O. sativa L. ssp. indica) by ethyl methanesulfonate (EMS) mutagenesis. The objectives of the study were (1) to analyze the phenotypic differences between Ospa mutant and WT, (2) to map the candidate gene (OsPA) using the map-based cloning strategy, and (3) to develop molecular markers for OsPA that will assist molecular breeding in rice.

Materials and Methods

Plant Materials and Growth Condition

Rice (Oryza sativa L.) purple apiculus mutant was obtained from an M2 population of the indica cultivar Zhenong 34 after EMS mutagenesis. For phenotypic segregation analysis and genetic mapping, F1 plants were developed by crossing the Ospa mutant (female parent) and green apiculus Zhenongda 104 (male parent). Then, the F1 plants were self-crossed to generate F2 population. All the parents and 2302 F2 individuals were cultivated in the paddy field at Zhejiang University in Hangzhou, China (N 30° 15′ 49″,E 120 7′15″) during the natural growing season.

Measurement of Phenotypic Traits

The wild-type (Zhenong 34) and Ospa mutant were planted in the paddy field at the same growing season. Agronomic traits such as plant height (PH), panicle number per plant (PN), grain number per panicle (GN), seed setting rate (SR), thousand-grain weight (TGW), grain length (GL), grain width (GW), grain thickness (GT), brown rice length (BRL), brown rice width (BRW), and brown rice thickness (BRT) were measured during maturity.

Physiological Analysis

The grain anthocyanin content was recorded in the 5th, 10th, 15th, 20th, 25th, 30th, and 35th day after heading (DAH). Total grain anthocyanin contents were extracted from fresh matured hull samples following the procedures described by Yang et al. (2018) with minor modification. In brief, anthocyanin was isolated using 1 mL of 1% acidic methanol (HCl/methanol, 1:99) solution. 1 g of finely ground mature hull tissue was incubated at 25 °C in the dark. The extracted solution was centrifuged at 12,000×g rpm for 20 min. The absorbance values of the supernatant were measured against the reagent blank at 530 and 657 nm with spectrophotometer UV-1800 (Shanghai, China), and total anthocyanins were quantified based on the absorption of the extracts using the equation:QAnthocyanins = (A530 − 0.25 × A657) × FW−1, where QAnthocyanins is the quantity of total anthocyanins; A530 and A657 are the absorptions levels at 530 nm and 657 nm, respectively, and FW is the weight of hull tissue (g). Anthocyanins were quantified as absorption levels in three independent biological replicates.

Total flavonoids were measured in the 5th, 10th, 15th, 20th, 25th, 30th and 35th DAH by the aluminum method previously reported by Shen et al. (2009) with slight modification. Mature hulls (1 g) ground with liquid nitrogen were mixed with 25 mL of methanol containing 1% HCl at 25 °C in the dark. The methanolic extracts were centrifuged at 4,000×g rpm for 5 min, and the supernatant was separated. Aliquots (0.5 mL) of extracts or standard solutions were transferred into 15 mL conical centrifuge tubes, and 2 mL double distilled H2O and 0.15 mL 5% NaNO2 were added. Then, 0.15 mL 10% AlCl3.6H2O and 1 mL 1 M NaOH were added in 5 min interval respectively. After 15 min, total flavonoids content was recorded in the form of absorbance values by spectrophotometer UV-1800 (Shanghai, China) at the wavelength of 415 nm, and the measurement was repeated three times. Total flavonoid content was calculated using the standard quercetin curve prepared by serial dilution of quercetin and expressed as μg quercetin equivalent (μg QE) per g of fresh weight.

Histological Analysis

The cross-sectional analysis was carried out to investigate the cellular anatomy of the spikelet hulls. The double fixation was carried out by excising tissue samples from the young spikelet hulls. The samples were fixed with 2.5% glutaraldehyde in phosphate buffer (0.1 M, pH 7.0) for 4 h and washed three times in the phosphate buffer for 15 min at each step followed by post-fixation with 1% OsO4 in phosphate buffer for 2 h. Consecutively, the specimen was first dehydrated with 30%, 50%, 70%, 80%, 90%, 95%, and 100% ethanol for 20 min at each step and transferred to absolute acetone for 20 min. The infiltration process was carried out by placing the specimen in 1:1 and 1:3 mixture of absolute acetone and Spurr resin mixture for 1 h and 3 h respectively at 25 °C. Then, the samples were placed in Spurr resin mixture overnight. Embedding was done by placing the specimen in Eppendorf contained Spurr resin and heating at 70 °C for 9 h. The specimen was sectioned in LEICA EM UC 7 Ultratome and stained by uranyl acetate alkaline lead citrate for 5 and 10 min respectively. The cross section was observed in transmission electron microscope Hitachi Model H-7650 TEM (Hitachi, Japan). The cell number in the outer parenchyma layer was analyzed using the multipoint tool of the ImageJ program version 1.51(National Institute of Health, USA).

For investigating the inner parenchyma layers cells and tubercles in the outer layer, the tissue samples from the young spikelet hull were fixed with 2.5% glutaraldehyde in phosphate buffer (0.1 M, pH 7.0) for 4 h and washed three times in the phosphate buffer for 15 min at each step. Subsequently, the samples were post-fixed with 1% OsO4 in the phosphate buffer for 2 h and washed three times in the phosphate buffer for 15 min at each step. Following post-fixation, the samples were dried by 30%, 50%, 70%, 80%, 90%, 95%, and 100% ethanol for 20 min at each step and transferred to the mixture of alcohol and iso-amyl acetate (1:1, v:v) for 30 min followed by pure iso-amyl acetate overnight. Finally, the samples were dried in Model HCP-2 critical point dryer (Hitachi, Japan) with liquid CO2, coated with gold-palladium and observed in scanning electron microscope TM-1000 SEM (Hitachi, Japan). Cell area in the inner parenchyma layer was analyzed using the ImageJ program version 1.51.

Mapping and Cloning of the OsPA Gene

The DNA isolation was carried out from fresh rice leaves (Murray and Thompson 1980). The candidate gene for Ospa mutant was located by fine mapping after bulked segregant analysis (BSA) (Michelmore et al. 1991). BSA was carried out using the DNA isolated from the female parent (Ospa mutant) and male parent (Zhenongda 104 with WT trait) and DNA pooled from 20 individuals with purple apiculus and 20 individuals with green apiculus respectively. A total of 400 SSR markers covering all the 12 chromosomes of rice were employed for mapping the candidate region. The candidate markers were selected based on the polymorphism of the parents, F2 mutant pool, and F2 wild-type pool taking in to account the similarity of F2 mutant pool to the female parent (Ospa mutant). For genetic mapping, 10 InDel markers were developed by comparing the diversity of indica and japonica rice DNA sequences from the gramene database (Gupta et al. 2016; Tello-Ruiz et al. 2018). Analyzing recombination events in the respective positions of the newly designed InDel markers (Table S1) between the WT and Ospa mutant individuals of F2 together with the parental lines, the mapping region was further delimited. For each marker designed, the candidate region was amplified with PCR cycle of denaturation at 95 °C for 5 min, followed by 35 cycles of 94 °C for 30 s; annealing at 55 °C for 30 s, followed by 72 °C for 30 s, and elongation step at 72 °C for 10 min. The PCR results were analyzed by polyacrylamide gel (8%) electrophoresis.

Analyzing the predicted function of the candidate ORFs in the delimited region (Table S2) using the Rice Genome Annotation Project (Kawahara et al. 2013), primers were designed for the candidate gene PCR amplification and sequencing (Table S1). The candidate gene was PCR amplified with cycles of denaturation at 95 °C for 5 min, followed by 35 cycles of 98 °C for 15 s; annealing at 55 °C for 30 s followed by 68 °C for 30 s, and elongation at 72 °C for 10 min. Agarose gel (1%) was used to visualize amplification result.

Structural and Functional Prediction of OsPA Protein Sequence

The secondary structure of the encoded protein was predicted based on the ResQ approach (Yang et al. 2016), and the function prediction of the OsPA protein was carried out based COFACTOR (Zhang et al. 2017) and COACH prediction (Yang et al. 2013a, b). The peptide model for the conserved domain where the mutation occurred was produced using the SWISS-MODEL program (Waterhouse et al. 2018). For sequence alignment, the protein sequence of OsPA was obtained from the rice genome annotation project database (Kawahara et al. 2013), and orthologous proteins were obtained using the protein basic local alignment search tool (Altschul et al. 1990). Protein family database (El-Gebali et al. 2019) and the conserved domain database (Sayers et al. 2019) programs were used to perform a domain analysis. The obtained sequences were compared with rice OsPA protein sequence using the Clustal Omega program (Sievers et al. 2011) with default settings and were viewed in the Jalview program version 2 (Waterhouse et al. 2009).

Phylogenetic Analysis

To analyze the evolutionary relationship and functional association between orthologous proteins of OsPA, additional sequences were obtained by the protein basic local alignment search tool (Altschul et al. 1990). The alignment of protein sequences was carried out and the phylogenetic tree was created using MEGA X program (Kumar et al. 2018). Neighbor-joining (NJ) method and a bootstrap test were performed with 1000 replicates to evaluate the support of clusters and nodes in the phylogenetic tree.

RNA isolation and real-time quantitative PCR analysis.

For analyzing the expression profile, total RNA was isolated from seedlings, culm, and root of Ospa mutant and WT. In addition, total RNA was extracted from the panicles at 2, 7, 12, 17, and 22 days after booting (DAB) for studying the time-course of gene expression. To evaluate the expression level of structural anthocyanin biosynthesis genes, total RNA was isolated from the panicle of Ospa mutant and WT at 17 DAB. Total RNA isolation was carried out using Trizol (Invitrogen, USA) following the manufacturer’s instruction. The Prime Script RT reagent Kit with gDNA Eraser and SYBR premix ExTaq™ (Takara, Japan) was used for quantitative real-time PCR (qRT-PCR) (Takara, Japan). One microgram RNase-free DNase I treated total RNA was utilized for the first-strand cDNA synthesis in a 20 μL reaction volume. Quantitative RT-PCR was performed using the SYBR Premix ExTaq™ (Tli RNaseH Plus). The OsACTIN1 was used as an internal control for the RT-PCR analysis. The RT-PCR was carried out with activation at 95 °C for 30 s, followed by 40 cycles of denaturation at 95 °C for 5 s, and annealing and extension at 60 °C for 30 s. The primers used in the qRT-PCR analysis were listed in Table S3.

Results

Phenotypic Characterization of Ospa Mutant

The morphological and agronomic traits of WT and Ospa mutant were measured at maturity under field conditions. The Ospa mutant plants exhibited purple coloration in the apiculus, while the WT has green apiculus (Fig. 1a, b). Traits difference between WT and Ospa mutant was observed only in the grains and a significant increase in GL, GW, and TGW in Ospa mutants compared with those of WT (Table 1). There was no difference in PH, PL, PN, GN, GT, BRL, BRW, BRT, and SR between WT and Ospa mutants (Table 1).

Fig. 1
figure 1

Comparison of different traits between WT and Ospa mutant. a Phenotype of WT and Ospa. b Spikelet of WT and Ospa mutant. Scale bar, 5 cm in (a) and 2 mm in (b).c Grain anthocyanin content of WT and Ospa mutant. d Total flavonoids content of WT and Ospa mutant. All data represented as mean ± SD. **represents significant difference at the 0.01 level by the Student t test (n = 3)

Full size image
Table 1 Agronomic traits of WT and Ospa
Full size table

Physiological Changes of Ospa Mutant

Related to the purple color formation in the apiculus, grain anthocyanin content was measured in WT and Ospa mutant. Significantly higher anthocyanin content was recorded in the Ospa mutant hulls than the WT (α = 0.01) (Fig. 1c). However, a significant difference (α = 0.05) was not seen for the grain flavonoid content between the Ospa mutant and WT (Fig. 1d). It was notable that the differences between WT and Ospa mutant might be limited only to the grains and grain composition. The anthocyanin and grain flavonoid contents were also analyzed during the grain development stages. Anthocyanin contents increased dramatically in the Ospa mutant from 5 to 15 days after heading (DAH) and gradually declined in the later stages, whereas there was no significant change in the hulls of WT (Fig. S1a). The grain flavonoids content increased from 5 to 20 DAH and maintained the same level in the later stages, but values were not significantly different between the Ospa mutant and WT (Fig. S1b). These results showed that the purple apiculus trait might be due to the increased anthocyanin accumulation in the Ospa mutant.

Changes in Cellular Anatomy of Ospa Mutant

In addition to pigmentation and increased anthocyanin content, increased GL, GW, and TGW were measured in the Ospa mutant. To find out the cellular anatomy in relation to the increased grain size, the spikelet hulls were investigated by transmission electron microscopy (TEM) in WT and Ospa mutant. The anthocyanin accumulation was evident in the purple color observed in the outer parenchyma layer of the spikelet hulls in the Ospa mutant (Fig. 2a–d). The number of parenchyma cells in the outer layer of the spikelet hull of Ospa mutant was significantly higher than that in WT (Fig. 2c–e). Likewise, while the cell area in the inner surface of the spikelet hulls was larger in Ospa mutant than that in WT (Fig. 2f–h). On the contrary, the number of tubercles per mm2 was significantly lower in Ospa mutant than the WT (Fig. 2i–k). Therefore, the cells in the mutant were bigger in size and higher in number in Ospa mutant than WT. These findings clearly showed that the increase in grain length and width might be from the expansion and proliferation of the spikelet hull cells.

Fig. 2
figure 2

Histological analysis of spikelet hulls. a, b Spikelets of WT and Ospa mutant. Scale bars, 2 mm. The red lines indicated the positions of tissue analyzed. c, d TEM analysis of parenchyma layers of spikelet in WT and Ospa mutant respectively. Scale bars, 100 μm. f, g SEM analysis of inner and outer layers of spikelet hulls in WT and Ospa mutant respectively. Scale bars, 100 μm. i, j SEM analysis of outer surface of spikelet hulls in the WT and Ospa mutant respectively. Scale bars, 100 μm. e Comparison of cell number of inner parenchyma layer in WT and Ospa mutant. h Comparison of the cell area of inner surfaces of spikelet hulls in WT and Ospa mutant. k Comparison of the number of tubercles/mm2 in the outer surfaces of spikelet hulls in WT and Ospa mutant. All data represent mean ± SD (n = 3). **represents significant difference at the 0.01 level by the Student t test

Full size image

Genetic Analysis and Map-Based Cloning of OsPA Gene

The Ospa mutant showed purple coloration at the apiculus, while the WT exhibited green apiculus. All the F1 plants from the crossing of the Ospa mutant (female parent) and green apiculus Zhenongda104 (male parent) showed the green apiculus, and the phenotypic segregation between WT (green apiculus) and Ospa phenotypes (purple apiculus) was observed in the F2 population. There were 1,839 green apiculus and 576 purple apiculus phenotypes which fitted the 3:1 ratio of the Mendelian model (χ2 = 3.40 < χ20.05 = 3.84), indicating that the purple apiculus of Ospa mutant was regulated by a single recessive gene.

The bulked segregant analysis (BSA) was carried out using SSR and InDel markers spanning the 12 chromosomes in rice. Polymorphic marker between the parents and the F2 population derived from Ospa mutant/Zhenongda 104 was located on the long arm of chromosome 1. The OsPA gene was first delimited to a region between R1M30 and 1MYH01 (Fig. 3a). Subsequently, InDel markers were developed by comparing indica and japonica rice genomic sequences and used for fine mapping (Fig. 3b). Analyzing recombination events in the candidate region in 576 F2 individual plants with the newly designed InDel markers, the mapping region was delimited to a 60.74 kb between markers 1MYH03 and 1MYH06.

Fig. 3
figure 3

Schematic diagram of map-based cloning and expression profile of OsPA gene. a Bulked segregant analysis (BSA) located OsPA between markers R1M30 and 1MYH01 using 40 F2 individuals with WT and Ospa mutant trait. bOsPA was fine mapped to an interval of 60.74 kb region by analyzing the recombination events in 576 F2 individuals between markers 1MYH03 and 1MYH06. c Seven ORFs were annotated in the 60.74 kb region. d Schematic representation of LOC_01g39580. Filled boxes represent exons. e Site of mutation in LOC_01g39580 as revealed by genomic sequencing of WT and Ospa mutant. f Relative expression profile of OsPA in different tissues. P1, P2, P3, P4, and P5; expression level in the panicle in the 2, 7, 12, 17, and 22 DAB. All data represent mean ± SD (n = 3). **represents significant difference at the 0.01 level by the Student t test

Full size image

The target region contained seven open reading frames (ORFs) based on the functional prediction by Genome Annotation Project (Kawahara et al. 2013) (Fig. 3c). After cloning and sequencing the viable candidate genes, a single nucleotide substitution was detected only in LOC_01g39580, encoding anthocyanin regulatory Lc protein. In the Ospa mutant, T to C substitution of the +3730th nucleotide in the third exon was detected. The nucleotide substitution resulted in the replacement of cysteine with arginine in the encoded protein of OsPA gene (Fig. 3d, e). Hence, we thought the difference in the gene and the resultant protein was the cause of the observed phenotypic, cellular, and physiological changes in the Ospa mutant.

Predicted Structure of OsPA and Its Orthologous Proteins

The coding DNA sequence (CDS) of OsPA gene is 1,113-bp long with four exons and three introns. The CDS was predicted to encode a polypeptide of 371 amino acids with the calculated molecular mass of 40.29 kDa. The structural prediction showed the secondary structure of the encoded protein. The cysteine to arginine change was located on the helix motif of the peptide, which resulted in the formation of additional helix structure in the Ospa mutant (Fig. 4a, b; Fig. S2). The functional prediction of the protein was identified nucleic acid and peptide as ligands with several corresponding positions of binding sites, while there was no enzyme active site predicted that the gene might function as a transcription factor (TF) (Table S4).

Fig. 4
figure 4

Structure prediction of the WT and Ospa mutant amino acid sequence. a Secondary structure prediction of the WT and Ospa amino acid sequence. Normalized B-factor represents the level of accuracy in the prediction. b Ball and stick model prediction of the WT and Ospa mutant protein structure for the region indicated with the brace signs in (a) which represents HLH conserved domains. Doted-box represents region in the Ospa mutant indicates the extended helix motif

Full size image

Orthologous proteins were identified for structural similarity analysis. The protein basic alignment showed that the orthologous protein sequences from monocot plants such as Zea mays, Triticum urartu, Sorghum bicolor, Setaria italica, and Panicum hallii had high sequence similarity. The OsPA protein was predicted as TF, containing basic helix-loop-helix (bHLH) domain, in which the amino acid substitution was detected (Fig. S3). The results suggested that the mutation might have altered the peptide binding domain of the TF, which was conserved in the basic helix-loop-helix (bHLH) domain (Fig. 4b; Fig. S2 and S3).

For understanding the phylogenetic relationship and functional associations of related proteins, additional protein sequences were identified alignment of orthologs were performed. Only the orthologs from monocot plants were found to be and possessed bHLH domain. A phylogenetic tree was constructed for 87 orthologs based on the NJ method. The phylogenetic analysis showed that 87 orthologs were grouped into nine clades with effective bootstrap support (Fig. S4). The orthologs obtained from O. sativa ssp. japonica, O. sativa ssp. indica, O. brachntha, O. longistaminata, and A. tauschii were in the same clade. Therefore, it is noteworthy that the orthologs conserved in other monocots might be functionally related.

The Expression of Anthocyanin and Grain Size-Related Genes

The Ospa mutant was characterized and the candidate gene which might have resulted in the observed phenotypic, cellular, and biochemical changes was identified based on the map-based cloning strategy. The sequence analysis confirmed the genetic difference between the WT and Ospa mutant. Then, the mRNA transcript of the OsPA gene was analyzed in the different plant tissues and panicle during the growth stages using RT-qPCR. The results showed that the OsPA gene highly expressed in the young panicles, while lower expression was exhibited in leaf, stem, and root. When the expression pattern of OsPA gene was studied during the panicle growth stages in the WT and Ospa mutant, dramatic increase in expression was observed in the mutant, starting at 7 days after booting (DAB) and reaching the maximum expression level at the 17 DAB, while less variation of the expression level was found in the WT (Fig. 3f). Therefore, the results indicated that the tissue-specific significant increase of the OsPA mRNA transcript in the panicle of the Ospa mutant and could be linked to apiculus coloration and the increased accumulation of anthocyanin in the grains.

As the apiculus coloration phenotype is determined by metabolic pathways and their regulatory cascades, the expression profiling of related genes underlying purple apiculus were conducted. The expression level of the structural anthocyanin biosynthesis genes, such as OsF3’H, OsF3’5’H, OsDFR, and OsANS (Aizza and Dornelas 2011; Shih et al. 2008), significantly increased in the panicle of the Ospa mutant (Fig. 5a). In addition to structural enzymatic genes, the expression of color producing gene, MYB gene (OsC), was significantly higher in the Ospa mutant than WT. On the other hand, the expression of OsPAL, OsCHS, and OsCHI, which direct conversion of upstream intermediates in the biosynthesis pathway (Aizza and Dornelas 2011; Hichri et al. 2011a, b; Hossain et al. 2018; Shih et al. 2008), remained not different between the WT and Ospa mutant. In a similar fashion to the OsPA gene, OsC showed tissue-specific expression in the panicle of Ospa mutant (Fig. 5a-b). Despite polymorphic markers were obtained for the mapped loci and DNA, and polypeptide sequence changes were found only in the Ospa gene, the gene expression levels of OsF3’H, OsF3’5’H, OsDFR, and OsANS increased along with OsPA gene in the mutant. The results showed that the expression levels of the regulatory and structural anthocyanin biosynthesis genes were upregulated in the mutant which might have activated anthocyanin accumulation in apiculus of the Ospa mutant grains.

Fig. 5
figure 5

Gene expression analysis. a Relative expression of structural and regulatory genes of anthocyanin biosynthesis in spikelet hulls at 17 DAB (OsC, LOC_Os06g10350; OsANS, LOC_Os06g42130; OsPAL, LOC_Os02g41630; OsCHS, LOC_Os11g32650; OsDFR, LOC_Os01g44260; OsF3′H, LOC_Os04g56700; OsCHI, LOC_Os03g60509; OsF3′5′H, LOC_Os10g17260). b Relative expression profile of OsC in different tissues (P1, P2, P3, P4, and P5; expression level in the panicle in the 2, 7, 12, 17, and 22 DAB (d)). c Relative expression of cell size and proliferation related genes controlling grain size and weight in spikelet hulls at 17 DAB (GS3, Os03g0407400; GS5, Os05g0158500; GW2, Os02g0244100; GW5, Os05g0187500; GW8, Os08g0531600; OsAGSW1; Os05g0323800; qGL3, Os03g0646900; SMG1, Os02g0787300; TGW6, Os06g0623700; SG3, Os03g0388800) d Relative expression of cell expansion related genes controlling grain size and weight in spikelet hulls at 17 DAB (WTG1, Os08g0537800; D11, Os04g0469800; SLG, Os08g0562500; OsBUL1, Os02g0747900; PGL1, Os03g0171300; APG, Os05g0139100; SG1, Os09g0459200; SMOS1, Os05g0389000; GLW7, Os07g0505200; GS2, Os02g0701300). OsACTIN1, LOC_Os03g50885 was used as internal control. All data represent mean ± SD (n = 3). **represents significant difference at the 0.01 level by the Student t test

Full size image

In addition to the apiculus coloration, the Ospa mutant exhibited a significant increase in the grain size and thousand-grain weight. In relation to this change, the mRNA transcript levels of genes regulating grain size and weight were analyzed between the WT and Ospa mutant. The results exhibited that the expression level of OsAGSW1 and SG3, which regulated grain size and weight by modulating cell size and proliferation in the spikelet hulls (Li et al. 2015; Wang et al. 2018) significantly increased in the Ospa mutant. The results indicated the mutation had an effect on grain size and weight as well (Fig. 5c). The other genes, such as GS3, GS5, GW5, GW8, qGL3, SMG1, and TGW6, which showed similar function (Duan et al. 2014; Fan et al. 2006; Ishimaru et al. 2013; Kang et al. 2018; Li et al. 2011; Liu et al. 2017; Zhang et al. 2016) did not show a significant difference in the expression level between the WT and Ospa mutant (α = 0.05) (Fig. 5c). Also, WTG1, D11, SLG, OsBUL1, PGLI, APG, SG1, SMOS1, GLW7, and GS2, which regulated grain size and weight by modulating only cell expansion (Aya et al. 2014; Feng et al. 2016; Heang and Sassa 2012a, b; Huang et al. 2017; Jang et al. 2017; Si et al. 2016; Tong et al. 2018; Wuhan et al. 2013; You et al. 2007) did not show a significant difference in the expression level between WT and Ospa mutant (α = 0.05) (Fig. 5d).

Discussion

Apiculus color is one of the easily observed traits and regarded as a vital morphological marker in rice breeding. In the past, apiculus color has attracted the interest of a lot of researchers (Fan et al. 2007; Mikami et al. 2000; Reddy et al. 1995; Saitoh 2004; Sun et al. 2018; Zhao et al. 2016). Genes coding for bHLH proteins were involved in activation OsC expression which is considered as the basic gene for color production and regulated expression of structural anthocyanin biosynthesis genes to determine pigment distribution in the in rice (Sun et al. 2018). The bHLH domains make protein–protein interaction with the R2R3 MYB domain and activate downstream genes of the structural anthocyanin biosynthesis pathway (Koes et al. 2005; Sakamoto et al. 2001). Mutations altering the HLH domain can alter protein–protein interaction between HLH protein and any other protein and enhance or diminish the activities of the bHLH proteins. For instance, a single amino acid substitution in a bHLH transcription factor led to alteration of the interaction with partner proteins in Arabidopsis (Zhao et al. 2012). In the present study, a purple apiculus mutant, Ospa, was developed from wild-type with a green apiculus with EMS treatment and the phenotypic differences between the WT and Ospa mutant were observed in the grains. OsPA was identified by map-based cloning on chromosome 1. The OsPA protein was predicted as TF containing coil, strand, and helix motifs; however, the mutation was located in the helix motif which resulted in formation continuation of the helix motif in the predicted OsPA protein which is similar to the annotation by rice genome project which predicted as anthocyanin regulatory Lc protein (Kawahara et al. 2013) and annotation by gramene database as transcriptional activator protein (Gupta et al. 2016). The conserved domain analysis showed that OsPA protein was predicted to possess HLH domain which was highly conserved in the analyzed monocot species. The T to C nucleotide substitution that changed cysteine to arginine in the OsPA polypeptide was identified in the HLH domain. The differences in the predicted peptide structure and mRNA transcript levels between the WT and Ospa mutant have implied that the mutation could have enhanced the expression of the OsPA and related genes to form purple apiculus trait and increased the grain size.

Lc-like proteins have been shown to affect the growth and development in addition to the anthocyanin biosynthesis. Overexpression of Lc gene in an Arabidopsis ttg1 mutant complemented growth-related traits beyond anthocyanins content, indicating that genes coding for bHLH proteins with similarity to Lc can be involved in the regulation of epidermal cell growth (Lloyd et al. 1994). The bHLH AtPIF4 played a key role in regulating morphogenesis (Leivar and Quail 2011). Consistent with these observations, the OsPA gene possessed the bHLH domain and analysis of cellular structure between WT and Ospa mutant revealed increased cell number and size in the parenchyma layers of spikelet hulls which, in turn, might have resulted in increased grain size and weight in our study. The significant increase of grain size and weight was supported by the mRNA transcript level analysis of genes related to grain size using qRT-PCR. Expression of OsAGSW1 and SG3 were significantly upregulated in the Ospa mutant. Previously, OsAGSW1 increased grain size and weight by regulating the number of external parenchyma cells in the spikelet hulls of OsAGSW1-overexpressed rice and RNAi lines (Li et al. 2015). Similarly, EMS-generated sg3 mutant showed an increase in grain size and TGW by increasing cell proliferation in the spikelet hulls (Wang et al. 2018). In our study, the Ospa mutant showed upregulated expression of OsAGSW1 and SG3, and it might have regulated the size and number of spikelet hull cells. Grain size in rice is highly controlled by the size of the spikelet hull (Li et al. 2011; Song et al. 2007). The results suggested that the OsPA gene might have affected cell size in spikelet hulls in addition to its effect on apiculus coloration synergistically with OsAGSW1 and SG3.

Anthocyanins formation was catalyzed by enzymes encoded by structural genes, which utilized phenylalanine as a common substrate (Winkel-Shirley 2001). Chalcone synthase, chalcone isomerase, and flavanone 3-hydroxylase are encoded by the early biosynthesis genes are and are involved in the production of common precursors, dihydroflavonols, in the anthocyanin biosynthesis pathway whereas dihydroflavonol 4-reductase and anthocyanidin synthase are encoded by late biosynthesis genes (Ma et al. 2009). The bHLH proteins were involved in the regulation of the anthocyanin biosynthesis synergistically with MYB-protein, and WD-repeat proteins and many other proteins (Goff et al. 1992; Grotewold 2006; Hernandez et al. 2007; Nair and Burley 2000; Ramsay and Glover 2005; Sun et al. 2018). The bHLH protein Lc was specifically involved in tissue-specific anthocyanin pigmentation (Ludwig and Wessler 1990). This is a common theme in other plant species, in which an Lc-like bHLH protein requires an MYB protein partner to activate tissue-specific anthocyanin biosynthesis (Bernhardt et al. 2003; Quattrocchio et al. 1998; Spelt et al. 2000; Sun et al. 2018; Zhang et al. 2003). In the present study, the gene expression analysis showed that the OsPA gene expressed specifically in the panicle of the Ospa mutant and the expression dramatically increased during panicle developmental stages. In the past, the OsC, which is and MYB TF, and different from OsPA locus, was linked with purple apiculus trait (Liu et al. 2012). Recently, OsC was reported as a regulator of the red apiculus trait indicating that it is the basic gene for color formation in rice (Zhao et al. 2016). In the present study, the expression of OsC significantly increased in the immature panicles of Ospa mutant along OsPA gene. Therefore, OsPA and OsC might have regulated purple apiculus trait formation synergistically. Furthermore, significantly upregulated the expression of OsF3′H, OsF3′5′H, OsDFR, and OsANS specifically in the Ospa mutant was observed in our study. The enzyme product of OsF3′5′H directs the conversion of dihydrokaempferol to dihydromyricetin whereas the enzyme product of OsDFR directs conversion of dihydromyricetin, dihydroquercetin, and dihydrokaempferol into leucoanthocyanidins. Similarly, the enzyme product of OsANS directs the conversion of leucoanthocyanidins into anthocyanidins (Aizza and Dornelas 2011; Hichri et al. 2011a). The upregulated expression of OsPA and OsC along structural anthocyanin biosynthesis genes such as OsF3′H, OsF3′5′H, OsDFR, and OsANS suggested that the OsPA gene might have involved in positive regulation of the conversion of unique intermediates in the anthocyanin pathway that produce purple color in the apiculus of Ospa mutant. The observed changes in the grain anthocyanin content; expression of OsC, OsF3′H, OsF3′5′H, OsDFR, and OsANS; the size of parenchyma cells of spikelet hulls, size, and weight of grains could be linked to the single nucleotide change and the upregulated OsPA gene expression detected in the Ospa mutant.

Conclusion

Apiculus color of grain is one of the easily observed traits in rice. In this study, we generated a new purple apiculus mutant, Ospa, using ethyl methanesulfonate (EMS) mutagenesis. We analyzed the phenotypic differences and isolated the candidate gene, OsPA, employing the map-based cloning strategy. In addition, we have shown that OsPA has regulated anthocyanin level quantifying the anthocyanin content in the grains and studying the expression of OsPA and structural anthocyanin biosynthesis genes using RT-qPCR. Investigating the cellular anatomy using histological analysis, we have demonstrated that parenchyma cells of spikelet hulls were more in number and enlarged size in the Ospa mutant which might implicate the pleiotropic nature of the OsPA gene. Since the gene has not been reported before, it could provide new insight into the network of genes involved in tissue-specific pigmentation and the formation of purple apiculus trait as well as expansion and proliferation of spikelet hull parenchyma cells in rice. The Ospa mutant developed for this study can be utilized in rice selective breeding, and the developed gene-based markers would be applied to rice molecular breeding. Studying the function of OsPA by gene transformation in focus on its sub-cellular expression and interaction with anthocyanin regulatory and structural biosynthesis genes as well as genes related to grain size and weight will be worthwhile in the future.