Rice grain quality is a complex trait that involves grain physical appearance, milling quality, eating and cooking quality, and nutritional quality. Among them, nutritional quality as well as eating and cooking qualities (ECQs) play significant indices to consumer’s preference and market-place (Yang et al. 2019). Endosperm is the main edible part of cereal grains and largely plays a key role in assessing the nutritional value of rice (Kang et al. 2006; Kasem et al. 2011). Milled rice constitutes 80-82% starch, 12-13% moisture content, 6.6-7.1% protein, 0.3-0.5% crude fat, and 0.2-0.5% crude fiber (Kim et al. 2021). Starch, a main component of rice, is comprised of two glucose polymers: amylose and amylopectin. Amylose makes up 5-35% of the most natural starches and largely affects the starch properties in foods (Seung 2020). In rice, three physicoche-mical properties: amylose content (AC), gel consistency, and gelatinization temperature, largely control the ECQs which greatly influence the consumer’s preference (Zhang et al. 2019). Among the three physicochemical properties, AC acts as a key determinant in controlling the ECQs due to its influence on gelatinization, pasting, retrogradation, syneresis, and other functional properties (Wani et al. 2012). Thus, ECQs have been of important in rice breeding programs.

Association between starch properties and natural variation in starch synthesis–related genes had already been confirmed in various studies. The Waxy (Wx) gene encoding the granule bound starch synthase I (GBSSI) on chromosome 6 is primarily involved for the natural varia-tion in amylose content as several SNPs are identified in the 5’splice site of intron 1, exon 6, and exon 10 (Cai et al. 1998; Larkin and Park 2003; Chen et al. 2008; Kharabian- Masouleh et al. 2012). The amylose level is directly linked to GBSS1 level in the rice endosperm (Mikami et al. 2008; Ferdous et al. 2018). The natural variation in starch synthase IIa (SSIIa) has been charac-terized and associated to affect the cooking and starch properties of rice (Umemoto et al. 2004). Other than natural variations in starch synthesis-related genes, several induced mutations that result in loss of function or change starch biosynthesis have already been studied in rice. The CRISPR/Cas9 mutations in the waxy locus in the rice endosperm resulted in reduced GBSSI activity, loss of amylose content in rice endosperm, and alteration of expression of other starch biosynthetic genes (Perez et al. 2019). Yang et al. (2012) reported that genetic modification of starch branching enzyme 3 (SBE3) on chromosome 2 resulted in increased resistant starch (RS) content and AC in mutant Jiang-tangdao 1. Zhang et al. (2011) reported that double muta-tion lines of starch synthase genes (SSIIa/SSIIIa) displayed a chalky endosperm and influenced the ECQs that resulted in increased AC, increased pasting tempera-ture, and decreased viscosities. Shim et al. (2022) recently reported that sequence variations in SBE3 and GBSS1 led to changes in seed morphology, starch structure, starch crystal-linity, amylopectin chain length distribution, digestibility, AC, and RS. Thus, naturally occurring variations and induced mutants are of paramount important in breeding programs in regulating important agronomic traits, in-creasing grain yield, and improving grain quality traits in rice.

Asian cultivated rice is widely grown and is divided into two major sub-species, indica and japonica with indica type being genetically more diverse than japonica type (Caicedo et al. 2007; Mather et al. 2007; Huang et al. 2012; Campbell et al. 2020). A study revealed that both indica and japonica cultivars have significant variation in AC which is mainly determined by Wx gene, the major locus controlling AC in rice (Hirano et al. 1998). The waxy locus has two functional alleles, Wx^a and Wx^b. The Wx^a allele is mainly present in indica type that results in higher AC while Wx^b allele is predominant in japonica rice that results in fairly low AC (Juliano et al. 1993; Hori et al. 2021). Seve-ral quantitative trait loci (QTLs) associated with AC have already been identified on all of rice chromosomes (Zhang et al. 2021). Also, single nucleotide polymorphisms (SNPs) have been recognized in genes known to affect AC such as SBE3 (Yang et al. 2012; Adeva et al. 2020), SSIIa, and SSIIIa (Gurunathan et al. 2019). However, there are genes previously associated with starch biosynthesis but whose function in amylose synthesis had not yet been fully establ-ished like SSIVa (LOC_Os01g52250) (Wambugu et al. 2018), histone-fold domain-containing protein (LOC_ Os01g01290) (Wambugu et al. 2018), NAC (LOC_Os11g 31330) (Wambugu et al. 2018), and protein disulfide isomerase like 1-1 (PDIL1-1), also known as endosperm storage protein 2 (ESP2) (LOC_Os11g09280) (Kim et al. 2012). Thus, studying the underlying mechanism of these genes will be helpful in modifying AC as well as other physicochemical properties of rice.

Protein disulfide isomerase (PDI) is a chaperone protein that plays a significant function in protein folding in endoplasmic reticulum (ER) (Yagi-Utsumi et al. 2015). Rice has 19 PDI-like genes as revealed by genome database but whose functions are remarkably unknown (Houston et al. 2005; Kim et al. 2012). In plants, PDI has been involved in various physiological processes. In Arabidopsis, loss of PDI5 resulted in premature initiation of programmed cell death during embryo development indicating that cysteine proteases are prevented during the vacuolar trafficking (Andeme-Ondzighi et al. 2008). In rice, the esp2 mutants lacked PDI and accumulated proglutelins showing the formation of glutelin-prolamin aggregates through inter-chain disulfide bonds within the ER cisternal space (Takemoto et al. 2002). Han et al. (2012) revealed that absence of PDIL1-1 is linked to ER stress in the endosperm which might be the major cause of the formation of a floury endosperm in the T3612 mutant. Recently, Hori et al. (2022) introduced a deficient mutant allele of PDIL1-1 to two genetically different cultivars, Koshihikari and Oonari. The results revealed that the genetic background of rice cultivars made ineffective the favorable effect of a PDIL1-1 mutant allele in accumulation of glutelin in the endosperm and improvement of flour characteristics and food processing properties of rice.

In this study, we investigated the association of PDIL1-1 on AC using F₂ plants and Korean japonica cultivars. In addition, we developed gene-based molecular markers to discriminate the genes of various japonica rice cultivars.

Plant materials and field trials

In our previous study, we established a RIL population by crossing two closely related japonica cultivars Dodamssal and Hwayeong (a non-glutinous cultivar), with large dif-ferences in their AC and RS content (Supplementary Fig. S1). A total of 92 RILs were used in QTL mapping and whole genome sequencing (WGS) was done between the parental lines (Adeva et al. 2020). Previously, QTLs related to amylose content were identified on chromo-somes 2, 6, and 11. The effect of QTLs found on chro-mosomes 2 and 6 on amylose content and physico-chemical characteristics in rice was previously studied (Adeva et al. 2020; Shim et al. 2022).

Here, our study mainly focused on QTL found on chro-mosome 11. All genes covering the qAC11 on chromosome 11 was listed (Supplementary Table S1). Among the genes, we selected the PDIL1-1 gene based on the previous report that PDIL1-1 was associated with AC (Hori et al. 2022). To characterize PDIL1-1 based on the sequence variants between Dodamssal and Hwayeong, haplotype analysis was carried out using all nonsynonymous SNPs located inside of the DNA coding region and one InDel in the non-coding region. As test materials for haplotype analysis, a total of 79 Kongju National University accessions from the KRICE_CORE set were used to determine the SNPs/ InDels linked to PDIL1-1 (Supplementary Table S2; Kim et al. 2016). To confirm the association of PDIL1-1 on AC, a total of 216 F₂ plants derived from a cross between Hwayeong and CR2121 (a RIL derived from a cross between Hwayeong and Dodamssal) were developed and generated. The F₂ plants and parental lines were grown in the experimental field at Chungnam National University in Daejeon, South Korea. Also, a total of 164 Korean japonica cultivars were used as plant materials. In 2020, cultivars were grown in Department of Southern Area Crop Science (Miryang, Gyeongsangnam-do) of the National Institute of Crop Science (Supplementary Table S3) with four rows of each cultivar.

DNA extraction and genotyping

Total genomic DNA was extracted from fresh leaves of each cultivar using the cetyltrimethylammonium bromide (CTAB) method as described by Murray and Thompson (1980). To determine whether plants are possessing the sbe3 (C SNP) and Wx^a (G SNP) on chromosomes 2 and 6, respectively, we used gene specific CAPs markers that distinguish the SBE3 T/C SNP and GBSS1 T/G SNP following the protocol as described by Adeva et al. (2020). The list of primer sequences used in this study was provided in Supplementary Table S4. All F₂ plants and Korean japonica cultivars possessing the Wx^b (T SNP) and SBE3 (T SNP) were selected and further used to investigate the association of PDIL1-1 on AC. Since sbe3 (C SNP) is a mutant allele, genotyping of Korean japonica cultivars at this locus was not carried out. Out of 164, two japonica cultivars, Goami2 and Dodamssal, are known cultivars possessing the sbe3 (C SNP). Dodamssal was developed from a cross between Goamybyeo (developed from a Milyang95//Gimcheonaengmi/Ilpum cross) and Goami2 (an N-nitroso-N-methylurea-induced mutant) (Cho et al. 2019).

To detect the PDIL1-1 G/A SNP at exon 5, a specific gene-based molecular marker was used for genotyping. The PCR reaction mixture was adjusted to a volume of 30 mL containing 3.0 mL 10× PCR buffer, 2.4 mL 2.5 mM dNTP, 1.0 mL 10 pmol forward primer, 1.0 mL 10 pmol reverse primer, 0.3 mL Taq polymerase, 1.0 mL template DNA, and 21.3 mL water. The PCR conditions were as follows: 95℃ for 5 minutes, followed by 35 cycles of 95℃ for 30 seconds, 58℃ for 40 seconds, and 72℃ for 1 minute, and 10 minutes at 72℃ of final extension. The PCR products were digested using a PvuII restriction enzyme. The digestion mixture was as follows: 8.75 mL of PCR product, 1.0 mL of 10× restriction buffer, and 0.25 mL of PvuII (10,000 U/mL). The digestion mixture was incubated at 37℃ for 3 hours, electrophoresed on a 3% agarose gel, and visualized by illuminating the gel on a standard UV transilluminator.

To detect the PDIL1-1 ATTCG/G at intron 1, two InDel markers were used for genotyping. The PCR reaction mixture was adjusted to a volume of 30 mL containing 3.0 mL 10x PCR buffer, 2.4 mL 2.5 mM dNTP, 1.0 mL 10 pmol forward primer, 1.0 mL 10 pmol reverse primer, 0.3 mL Taq polymerase, 1.0 mL template DNA, and 21.3 mL water. The PCR conditions were as follows: 95℃ for 5 minutes, followed by 35 cycles of 95℃ for 30 seconds, 60℃ for 40 seconds, and 72℃ for 1 minute, and 10 minutes at 72℃ of final extension. The PCR product was electrophoresed on a 1% and 3% agarose gel, and visualized by illuminating the gel on a standard UV transilluminator. Primer sequences are listed in Supplementary Table S4.

Evaluation of amylose content, crude protein content, and days to heading

The amylose content of F₃ seeds was measured using the Megazyme assay kit (Megazyme Ltd., Wicklow, Ireland), following the manufacturer’s instructions and was analyzed in Plant Breeding and Genetics laboratory, Chungnam National University. The amylose content of Korean japonica cultivars was analyzed in the laboratory of Southern Crop Department (Miryang, Gyeongsangnam-do) of the National Institute of Crop Science following the protocol as described by Juliano (1971). For crude protein content (PC) determination, rice flours at a measurement of 0.6 g were measured and determined using near-infrared (NIR) spectroscopy (XM-1100 series, FOSS NIR Systems INC., Laurel, MD, USA). Samples were placed in a micro insert ring attached to the mini sample cup and closed with a sample cup disposable back. Measurements were carried out in the near-infrared wavelength range of 400-2,500 nm at room temperature (25℃). The crude protein content was calculated using software ISI scan (version 4.5.0, InfraSoft International, Port Matilda, PA, USA) (Song et al. 2014; Oh et al. 2017). For heading date analysis, heading date was defined as the number of days from the date of transplanting to the date when 50% of individual plants reached the appearance of the first spikelet.

Statistical analysis

All analyses i.e. one-way analysis of variance and correlation analysis were carried out using MINITAB 16.2.4 software.

Characterization of qAC11

In our previous work, we conducted QTL mapping and identified QTLs associated with AC in a recombinant inbred line (RIL) population derived from two japonica cultivars, Dodamssal and Hwayeong (Supplementary Fig. S1). The QTLs for AC were identified on chromosomes 2, 6, and 11. The qAC2 was allelic to SBE3 while qAC6 was near the Wx gene locus encoding GBSS1 (Supplementary Figs. S2A and S2B). The previous work revealed that SBE3 and GBSS1 were mainly accountable for large phenotypic variation in the RIL population. Compared to these two genes, qAC11 had a minor effect on AC as revealed by R² values (R² = 10.5% (2017); R² = 10.1% (2018)). To confirm the effect of qAC11 on AC and find the potential candidate gene for qAC11, we compared the whole genome sequencing data between Dodamssal and Hwayeong. Sequence variants in the DNA coding region of all genes nearest to qAC11 (within the marker interval KJ11_005 and InDel-2) were investigated (Supplementary Table S1) and obtained from the Rice Annotation Project Database (RAP-DB). Prior to analysis, all hypothetical proteins/genes and non-protein coding transcripts were omitted. The candidate gene was selected based on biological function and presence of at least one nonsynonymous SNP or InDel. As the only gene involved in starch biosynthesis and maturation of proglutelin in endosperm, the PDIL1-1 gene was selected as the potential candidate gene for qAC11.

Characterization, identification and verification of mutation site in PDIL1-1 locus

Whole genome sequencing of parental lines, Dodamssal and Hwayeong, was determined with reference to the Nipponbare reference genome. Sequencing results led to the identification of various amylose-linked SNPs/InDels in PDIL1-1. The DNA sequence of PDIL1-1 includes 3,817 nucleotides with 10 exons. The PDIL1-1 gene is comprised of 1539 bp encoding a predicted protein with 512 amino acid residues. The variation in the genomic region of PDIL1-1 between Dodamssal and Hwayeong displayed SNPs in the coding region that led to amino acid substitutions causing missense mutations (Fig. 1). From the identified sequence variants between Hwayeong and Dodamssal, we used eight non-synonymous SNPs in the DNA coding region and one InDel in the DNA non-coding region of PDIL1-1 causing splice variant and performed haplotype analysis using 79 KNU accessions (Fig. 2, Supplementary Table S2). The genotypes of 79 KNU accessions were analyzed to determine the effect of the nucleotide sequence variations found between Dodamssal and Hwayeong on AC. As a result, 79 KNU accessions were classified into three haplotype (HAP) groups. A total of 51, 14, and 14 accessions belonged to HAP1, HAP2, and HAP3, respectively, and there was no significant diffe-rence in AC between HAP1 and HAP3, but HAP2 had AC significantly higher than HAP1 and HAP3 by 9.3% and 9.5%, respectively. Hwayeong belonged to HAP1, however Dodamssal did not, showing that Dodamssal PDIL1-1 mutation caused it to diverge from other haplotypes.

Figure 1. The PDIL1-1 gene structure and sequence variants between Hwayeong and Dodamssal. Numbers on top indicate the physical SNP position in bp with the corresponding nucleotide in red. Gray boxes indicate the 5’ and 3’ UTR regions, black boxes and lines between black boxes indicate exons and introns, respectively. SNP variants of the accessions in the exons leading to changes in the amino acid compared to the Nipponbare as the reference are underlined with the corresponding amino acid in the parentheses. Ala: Alanine, Ile: Isoleucine, Val: Valine, Glu: Glutamic acid, Lys:Lysine, Pro: Proline, Met: Methionine, Leu: Leucine, Thr: Threonine, Asp: Aspartic acid, Gln: Glutamine.

Figure 2. Haplotype analysis of SNPs and an InDel associated with protein disulfide isomerase like 1-1 (PDIL1-1) using the 79 japonica accessions from the KRICE_CORE set and their amylose content (AC). Values are means ± standard deviation. Mean with different letter is significantly different from each other at P = 0.05 based on Tukey’s test. SNP or InDel: Single nucleotide polymorphism or Insertion-Deletion. ¹⁾polymorphism is located in the coding region, ²⁾polymorphism is located in the non-coding region. HAP: haplotype.

By excluding lines with a sbe3 (C SNP) or a Wx^a (G SNP), two important genes that controlled AC variation in the RIL population, we characterized the effect of PDIL1-1 on the same RIL population. A gene-specific marker located at position -4973180 of PDIL1-1 was utilized to genotype all RILs carrying SBE3 (T SNP) and Wx^b (T SNP) for the analysis. In Table 1, the Dodamssal allele (A SNP) greatly increased the AC in the RIL population indicating that PDIL1-1 plays a role in regulating AC in rice. Despite the fact that the amylose-linked SNP controlling the AC between haplotype and RILs was different, the PDIL1-1 still demonstrates its function in regulating the AC in rice.

Table 1 . Characteristic of the PDIL1-1 gene on amylose content in RILs in a two-year field experiment.

SNP ID	Chr	Physical position (bp)	Year	Genotypex)
				DD	HH
G/A	11	-4973180	2017	16.1 ± 2.60	14.3 ± 2.09*
			2018	16.2 ± 1.23	15.2 ± 1.45*

^x)DD: Dodamssal homozygous, HH: Hwayeong homozygous. * indicates significant difference at P < 0.05.

Variation in amylose content, protein content, and days to heading

Amylose content is the key factor that controls ECQs in rice while protein content plays a significant role in rice nutritional quality. In our present study, we determined the AC in 29 F₂ plants and AC, PC, and days to heading (DTH) in 160 Korean japonica cultivars. The AC in polished rice of F₂ plants ranged from 12.6 to 15.3% (Fig. 3A) in which all plants belonged to low AC class according to the five amylose content classes given by Kumar and Khush (1987): high (>25%); intermediate (20-25%); low (10-19%); very low (3-9%); and waxy (0-2%). In polished rice, the AC of Korean japonica cultivars ranged from 16.2 to 20.7% in which most of the japonica cultivars belonged to low AC class. Only 10 japonica cultivars had an intermediate AC (20-25%) (Fig. 3B). In our present study, grain PC in rice flour of Korean japonica cultivars ranged from 5.17 to 7.65% (Fig. 3C). The number of DTH in Korean japonica cultivars ranged from 51 to 80 days (Fig. 3D).

Figure 3. Frequency distribution of amylose content (AC), protein content (PC), and days to heading. (A) %AC in 29 F₂ plants, (B) %AC, (C) %PC, and (D) number of days to heading in 160 Korean japonica cultivars.

Correlation of amylose content, protein content, and days to heading

Previous studies have reported a correlation between AC and heading date and between AC and PC. Rice quality is affected by the mean temperature during the grain filling stage, which is determined by the heading date. High temperature (30℃) occurring at the early-filling stage hampered starch synthesis and accumulation that resulted in the formation of chalky seeds with low AC, while lower temperatures (22℃) occurring at the late-filling stage allowed starch synthesis and accumulation to revert to normal levels. In RILs possessing the Wx^b (T SNP) and SBE3 (T SNP), the correlation between DTH and AC was also analyzed in two years (Table 2). Significant positive correlation was observed between 2017 AC and 2017 DTH, between 2017 AC and 2018 DTH, and 2018 AC and 2017 DTH. However, no significant correlation was observed between 2018 AC and 2018 DTH. In Table 3, correlation analysis revealed that AC of the Korean japonica cultivars was highly significant and positively correlated with DTH while significant negative correlation was observed between PC and DTH. The results indicated that DTH affects the eating and cooking quality as well as the nutritional quality of rice. However, no correlation was observed between AC and PC among the Korean japonica cultivars.

Table 2 . Correlation analysis between amylose content (AC) and days to heading (DTH) of RILs in a two-year field experiment.

Trait	2017 AC	2018 AC	2017 DTH
2018 AC	0.376*
2017 DTH	0.432*	0.411*
2018 DTH	0.404*	0.221	0.769***

* and *** indicate significant differences at P < 0.05 and P < 0.001, respectively.

Table 3 . Correlation analysis between amylose content (AC), protein content (PC), and days to heading (DTH) of 160 Korean japonica cultivars.

Traits	AC	PC
PC	‒0.061
DTH	0.552***	‒0.198*

* and *** indicate significant differences at P < 0.05 and P < 0.001, respectively.

Development of gene-based molecular markers

Functional markers (FMs) were developed using the SNPs identified between Dodamssal and Hwayeong. Three gene-based molecular markers, namely one CAPs (Fig. 4A) and two InDel markers (Figs. 4B and 4C) were designed. The CAPs marker targeted the position of -4973180 (G/A SNP) of PDIL1-1 located in exon 5 while the InDel markers targeted the position of -4974161 (ATTCG/G) of PDIL1-1 gene located in intron 1. To detect the G/A SNP at exon 5 of PDIL1-1 among the KNU accessions, the CAPS marker was PCR amplified followed by digestion with PvuII. Amplified fragments having a G SNP were cleaved by PvuII, yielding 108 bp and 46 bp fragments, while amplified fragments having an A SNP were not cleaved, yielding the intact 154 bp fragment. The CAPS marker from Dodamssal was not digested by PvuII, but that from Hwayeong was cut.

Figure 4. Development of gene-specific markers for PDIL1-1. (A) Genotype data using a CAPs marker. PCR products were subjected to PvuII digestion and then separation on a 3% agarose gel. Two digested fragments (108 bp and 46 bp) were observed in KNU japonica rice accessions having a G SNP at position -4973180; whereas, only a single undigested band (154 bp) was present in KNU japonica rice accessions having an A SNP. Upper and lower figures showed the PCR products before and after digestion, respectively. Genotype data using InDel markers, namely (B) Dominant marker and (C) Co-dominant marker. For dominant marker, PCR amplification (429 bp) was scored as the presence (1) or absence (0) of band. Band presence (1) = ATTCG. Band absence (0) = G SNP. For co-dominant marker, PCR amplification was scored as 1 and 3 for ATTCG and G, respectively. Lanes: 1: RWG-140; 2: RWG-142; 3: RWG-148; 4: RWG-159; 5: RWG-143; 6: RWG-144; 7: RWG-146; 8: RWG-179; 9: RWG-150; 10: RWG-185; 11: RWG-188; 12: RWG-224; 13: RWG-247; 14: Hwayeong; 15: Dodamssal. M: 100 bp plus DNA ladder.

To detect the ATTCG/G at intron 1 of PDIL1-1 among the KNU accessions, two InDel markers, namely one domi-nant and one co-dominant markers were PCR amplified. For dominant marker, fragments having an ATTCG were amplified while fragments having a G SNP did not. For co-dominant marker, fragments having an ATTCG yielded 125 bp while fragments having a G SNP yielded 121 bp. The PCR amplification using the gene-based molecular markers confirmed the sequence data of the KNU accessions as well as Dodamssal and Hwayeong. Therefore, the genotyping results showed that the accuracy of the three gene-based molecular markers could serve as diagnostic molecular markers linked to PDIL1-1 gene and can be used to select the desired genotype with elevated AC level.

Genotypic evaluation of F₂ plants and Korean japonica cultivars

Prior to analysis, all plants/cultivars possessing the sbe3 (C SNP) and Wx^a (G SNP) on chromosomes 2 and 6, respectively, were excluded from the analysis as these genes tend to have larger effects resulting in increased AC. In the F₂ population, the detection of a T/G SNP located at 5’splice site of intron 1 of GBSS1 and a T/C SNP at exon 16 of SBE3 was performed using 216 F₂ plants. A total of 29 F₂ plants harbored the Wx^b (T SNP) and SBE3 (T SNP) alleles. All plant materials were also genotyped using the gene-specific markers for PDIL1-1. At position -4973180 of PDIL1-1, a total of ten and two F₂ plants harbored an A and G genotype, respectively (Table 4). At position -4974161 of PDIL1-1, a total of ten and two F₂ plants harbored a G and ATTCG genotype, respectively. When we combined the SNP at exon 5 and an InDel at intron 1, 12 F₂ plants possessing homozygous genotype were classified into two groups. The AC of Group II was significantly higher than the AC of Group I confirming the presence of qAC11 (Table 4).

Table 4 . Comparison of amylose content between two genotypes classified based on SNP/InDel at PDIL1-1.

Group	No. of F2 plants	Position (bp) and SNP (InDel)		% AC
		-4973180	-4974161
I	2	G	ATTCG	13.2 ± 0.07
II	10	A	G	14.2 ± 0.70
P-value		0.02*	0.03*	0.02*

Percentage of amylose content (%AC) was presented as mean ± standard deviation.

* indicates significant difference at P < 0.05 based on one-way ANOVA.

Out of 164, a total of 161 japonica cultivars harbored the Wx^b (T SNP) including Goami 2. However, Goami 2 was not included in the analysis since it harbors the sbe3 (C SNP) at chromosome 2. Thus, four japonica cultivars, namely Dodamssal, Goami, Saegoami, and Goami 2 were excluded from the analysis (Supplementary Table 2). To validate the developed gene-based molecular markers, we applied our newly developed markers to 160 Korean japonica cultivars. For the CAPs marker, all PCR products were subjected to PvuII digestion. From the analysis, the results revealed that Korean japonica cultivars had higher proportion of G SNP at position -4973180 of PDIL1-1 gene located in exon 5 and an ATTCG at position -4974161 of PDIL1-1 (Supplementary Table 3).

At position -4973180, Korean japonica cultivars posses-sing the G SNP had significantly higher AC compared to cultivars possessing the A SNP (Table 5) while no signifi-cant difference in AC was observed between cultivars possessing the ATTCG or G SNP at position -4974161 (Table 5). In addition, when we combined the SNP at exon 5 and an InDel at intron 1, the 160 Korean japonica cultivars were categorized into three groups. The AC of Group III was significantly different from Groups I and II (Table 5). From the analysis (in both F₂ and Korean japonica cultivars), the results suggested that the G/A SNP at position -4973180 of PDIL1-1 might be associated with AC, whereas the ATTCG/G at intron 1 of PDIL1-1 might have only an ancillary effect. However, no significant difference in PC was observed among the Korean japonica cultivars at position -4973180 (Table 5) while a significant difference in PC was observed at position -4974161 of PDIL1-1 (Table 5). However, when we combined the SNP at exon 5 and an InDel at intron 1, significant difference among the groups was not observed (Table 5). The identified variable sites in PDIL1-1 might not be involved in the rice protein content.

Table 5 . Comparison of amylose and protein content among groups classified based on each SNP and their combination of InDel and SNP.

Group	No. of Korean japonica cultivars		-4973180		-4974161	% ACx)	% PCy)
	Nucleotide		G/A		ATTCG/G
Group I	100		G		ATTCG	18.4 ± 0.9a	6.3 ± 0.5a
Group II	35		G		G	18.8 ± 1.1a	6.5 ± 0.5a
Group III	25		A		G	17.5 ± 0.8b	6.4 ± 0.4a
P-value	% AC		0.000***		0.472ns	0.000***
	% PC		0.699ns		0.037*		0.074ns

Percentage of amylose content (%AC) and protein content (%) was presented as mean ± standard deviation. * and *** indicate significant differences at P < 0.05 and P < 0.001 based on one-way ANOVA, respectively. ns: not significant. ^x), ^y) Means followed by the same letter or letters in the column are not significantly different from each other (P > 0.05 ANOVA followed by Tukey’s test).

Rice is a widely grown crop and extensive varietal differences in eating quality characteristics had been discovered in worldwide rice germplasm accessions including indica and japonica (Nakamura et al. 2012; Zhao et al. 2020). Consumer preferences and how rice is pre-pared or consumed both have an impact on the quality of the grain. Amylose content greatly affects the milling, eating, and cooking quality of rice. Most of the factors affecting grain quality are quantitative traits controlled by multiple genes, making the development of high-quality varieties difficult and time-consuming. Earlier studies of the genetic basis of the apparent amylose content in rice endosperm mainly focused on the Wx gene, which codes for the granule-bound starch synthase. As previously reported, the Waxy gene post-transcriptional regulation, which is influenced by the intron 1 SNP, is mostly responsible for the amylose content in rice endosperm (Bligh et al. 1998).

In this study, with the aim of identifying other genes involved in starch biosynthesis that can affect the amylose content in japonica, we characterized the PDIL1-1 gene that helps in the folding of immature secretory proteins in the ER of rice endosperm cells. Whole genome sequencing of Dodamssal and Hwayeong led to the identification of eight SNPs in PDIL1-1 causing missense mutations (Fig. 1). When 79 KNU accessions were classified and analyzed based on the association between SNPs and amylose content in the PDIL1-1 gene, three haplotype groups were formed. Haplotype analysis revealed that the AC of HAP2 was significantly higher than HAP1 and HAP3 (Fig. 2) indicating that the nucleotide variations between genotypes are a significant source of heritable variation, thus, func-tional markers could be derived from them and could deliver a useful indicator of genotypic and phenotypic variations between varieties.

To date, identification and characterization of new molecular markers associated with amylose content regulation has been wide-ranging. There have been several reported molecular markers that can distinguish rice cultivars based on their amylose content. Bligh et al. (1995) identified the RM190 microsatellite (CT)n alleles in the Wx gene. Shao et al. (2020) reported the Wx gene-specific KASP markers to explain the variation in AC that microsatellite markers were unable to explain. Similar to this, additional marker sets derived from storage protein genes (glutelin and prolamin) and starch-synthesizing genes were also examined to assess AC, PC, and the adhesiveness of cooked rice (Nakamura et al. 2004; Bao et al. 2006a; Bao et al. 2006b; Lestari et al. 2009). In our present study, sequence variants between Dodamssal and Hwayeong were applied to develop gene-based molecular markers. One non-synonymous SNP and one InDel generated by WGS were transformed to PCR-based markers using CAPS and InDel approaches, respectively. We developed functional markers that can detect the PDIL1-1 gene during the seedling stage that can help in improving the operability and predictability of the breeding programs. In the F₂ population possessing the A SNP at position -4973180 and G SNP at position -4974161 of PDIL1-1 had significantly higher AC than plants posses-sing the G SNP at position -4973180 and ATTCG at position -4974161 of PDIL1-1 (Table 4). A set of 160 Korean japonica cultivars possessing the Wx^b alelle were genotyped using gene-based molecular markers to investi-gate and determine the association of PDIL1-1 on AC and PC variation. Despite Korean japonica cultivars are gene-tically related to each other with narrow genetic diversity, still, phenotypic variation among the Korean japonica cultivars was observed in AC, PC, and DTH. After screening, it was observed that Korean japonica cultivars possessing the G SNP at position -4973180 of PDIL1-1 had significantly higher AC than Korean japonica cultivars possessing the A SNP (Table 5). No significant difference on AC was observed between Korean japonica cultivars possessing the ATTCG or G SNP at intron 1 of PDIL1-1 (Table 5). We also found out that the combination of a G SNP at position -4973180 of PDIL1-1 gene located at exon 5 and a G SNP at position -4974161 of PDIL1-1 gene located at intron 1 was always associated to an elevated AC level (Table 5). Although the association of PDIL1-1 on AC between Korean japonica cultivars and F₂ population were different based on the amylose-linked SNP and InDel, we suggest that our results are still in accordance with the G/A SNP at exon 5 of PDIL1-1 as being the key factor affecting amylose content in rice endosperm, whereas, an ATTCG/G at intron 1 of PDIL1-1 might have only an auxiliary effect. It is also possible that the sample size of F₂ plants was too small and insufficient to elucidate an association between the amylose-linked SNPs and InDels in PDIL1-1 and amylose synthesis. Hori et al. (2022) reported that the genetic background of rice cultivars made ineffective the favorable effect of a PDIL1-1 mutant allele in accumulation of glutelin in the endosperm and improve-ment of rice flour characteristics and processing properties.

The newly developed markers in our present study could benefit breeding programs by introducing the PDIL1-1 gene into elite japonica cultivars to regulate the AC since the G SNP is prevalent and maintained in temperate japonica accessions. Therefore, it is more effective to identify and choose genes using FMs developed from polymorphic regions within gene sequences that affect phenotypic variance. Previous studies have implicated the importance of PDIL1-1 to starch biosynthesis in rice endo-sperm as well as in improving the rice flour characteristics and food processing properties (Kim et al. 2012; Hori et al. 2022). Here, our results suggested that PDIL1-1 has an impact on regulating AC in japonica accessions, though the mechanism remains to be explored.

In rice, PC is one of the key factors affecting the eating and cooking qualities. Previous study reported that in common rice grains, there is highly significant negative correlation between amylose content and protein content (Li et al. 2009). Proteins tend to increase the hardness of cooked rice by inhibiting the absorption of water and swelling of starch granules (Martin and Fitzgerald 2002; Saleh and Meullenet 2007; Zhu et al. 2021). During recrystallization, proteins tend to combine with amylose by increasing gel hardness and lessening water holding capacity. In this study, PDIL1-1 showed no association on PC. No correlation was observed between AC and PC among 160 Korean japonica cultivars (Table 3). Liu et al. (2020) reported that AC and PC variation ranges may affect how AC, PC, and eating qualities are related.

This study investigated the association of PDIL1-1 gene on AC using gene-based molecular markers. The sequence variants in PDIL1-1 gene can be used to regulate the AC of japonica accessions and that PDIL1-1 would be valuable for fine-tuning the rice AC. Our findings can help further research in varietal development with desired traits activated by PDIL1-1 gene. Developing methods to fine-tune rice grain AC to satisfy different consumers is therefore of great commercial value. To our knowledge, there are no reports worldwide that the PDIL1-1 gene regulates AC. In addition, there are no studies reporting the development of PDIL1-1 gene-based molecular markers.

This work was carried out with the support of “Coo-perative Research Program for Agriculture Science and Technology Development (Project No. PJ015757)” Rural Development Administration, Republic of Korea.