Cd content in rice grain of the SSSLs
Fifty-six SSSLs derived from seven different donor parents were selected for phenotypic evaluation during both the early and the late seasons from 2016 to 2018 at two sites, Guangzhou city and Shaoguan city. Grain Cd accumulation in the recurrent parent, HJX74, was 0.232 ± 0.009 mg/kg and 0.296 ± 0.007 mg/kg in the two seasons of 2016, respectively. The grain Cd content of most of the SSSLs was similar to that of HJX74. During the first 2016 cropping season in the Guangzhou experimental field, the Cd accumulation of SSSLs was between 0.095 to 1.903 mg/kg (Fig. S1a), while the cadmium content ranged from 0.050 to 0.636 mg/kg during the second 2016 cropping season (Fig. S1b).
Four SSSLs, namely W04-1, W04-4, W04-5 and W04-7, derived from donor parent BG367, had grain Cd content that was significantly higher than that of HJX74 across all tested environments (Supplementary Table 1). Thirteen additional SSSLs showed lower Cd content when grown in the paddy field with low Cd levels in Guangzhou, but no additional SSSLs differed in Cd accumulation compared to HJX74 in the environment with high Cd content in Shaoguan. Measurable Cd accumulation in the SSSLs fluctuated from season to season, even in the same location (Supplementary Table 1). After six seasons of planting at the two sites between 2016 ~ 2018, a total of 17 SSSLs were identified that had significantly different Cd accumulation in rice grain compared to HJX74 in at least three growing seasons.
Substitution Mapping Of Qtls For Cd Accumulation
The location and length of substitution fragments in the collection of 56 SSSLs were analyzed using molecular markers (Supplementary Table 2). The substitution segments were distributed on 8 chromosomes and covered nearly half of the genome (Supplementary Table 3). The eight QTLs associated with Cd accumulation in rice grain were represented in 17 SSSLs and were distributed on chromosomes 2, 3, 5, 6, 7, 8 and 11 (Fig. 1, Fig. S2). The QTLs were delimited to intervals of 1.55 ~ 12.78 Mb based on substitution mapping in the 17 SSSLs (Supplementary Table 2). The additive effects of the QTLs ranged from − 0.061 to 0.397 (Table 1).
Table 1
QTLs and additive effects
QTL | Chr. | Interval (kb) | Estimated length (Mb) | Maximum length (Mb) | Additive effect |
qCd-2-1 | 2 | 19620.62–32400.01 | 12.78 | 13.29 | -0.061 ± 0.018 |
qCd-3-1 | 3 | 13486.88–17576.99 | 4.09 | 4.38 | -0.096 ± 0.004 |
qCd-3-2 | 3 | 9536.04–13486.88 | 3.95 | 4.35 | -0.075 ± 0.009 |
qCd-5-1 | 5 | 25970.94–29749.53 | 4.06 | 4.34 | -0.066 ± 0.029 |
qCd-6-1 | 6 | 4014.71–5693.28 | 1.68 | 1.99 | -0.105 ± 0.006 |
qCd-7-1H | 7 | 7419.84–18546.41 | 11.13 | 12.90 | 0.397 ± 0.034 |
qCd-7-1L | 7 | 7419.84–18546.41 | 11.13 | 12.90 | -0.090 ± 0.006 |
qCd-8-1 | 8 | 1787.03–3332.17 | 1.55 | 1.93 | -0.101 ± 0.001 |
qCd-11-1 | 11 | 18422.14–28957.56 | 10.54 | 10.68 | -0.086 ± 0.009 |
As summarized in Table S1 and Table S2, the QTL qCd-2-1, located between RM13353 and RM250 on chromosome 2, was identified using SSSLW11−6 which carried a substitution fragment derived from the donor parent Basmati 370. Two closely linked QTL were detected on chromosome 3; qCd-3-1 was located between RM5489 and RM2346, and qCd-3-2 was located between markers RM3278 and RM3292. Three substitution lines were used to identify the qCd-3-1 QTL; SSSLW13−1 and SSSLW13−6 with donor segments derived from Jiangxisimiao, and SSSLW17−10, whose substitution fragment derived from the donor Ganxiangnuo. A single line, SSSLW27−18, derived from the donor IAPAR9, carried qCd-3-2. The QTL qCd-5-1 was carried by SSSLW27−05 in a substitution fragment that was also derived from IAPAR9, located between markers RM6792 and RM5818 on chromosome 5. The QTL qCd-6-1 was mapped on chromosome 6 between markers RM6734 and RM2615 and was carried by SSSLW10–13 and SSSLW10–14, both derived from the donor Nanyangzhan (Supplementary Table 2). The grain Cd contents of these eight SSSLs were significantly lower than that of HJX74 in most of the tested seasons, especially in the first cropping season of Guangzhou (Supplementary Table 1).
The QTL qCd-7-1 was associated with two phenotypes. Lower Cd concentrations were detected in SSSLW06−4 and SSSLW17−2 compared to HJX74, while higher grain Cd concentrations were detected in SSSLW04−1, SSSLW04−4, SSSLW04–5 and SSSLW04–7. These six SSSLs carried overlapping substitution fragments and shared the same QTL qCd-7-1. Furtherly, the six SSSLs were derived from three different donors and they could be divided into two distinctly different groups; one group manifested a low-cadmium phenotype designated as qCd-7-1L and the other was a high-cadmium phenotype, qCd-7-1H (Supplementary Table 1).
The qCd-8-1 QTL was detected in SSSLW27−3 which carried a donor segment from IAPAR9 located between RM6356 and PSM154 on chromosome 8. Finally, qCd-11-1, located on chromosome 11 from marker RM209 to the end of the chromosome, was carried by both SSSLW27−1 and SSSLW27−7, both of which carried substitution segments derived from donor parent IAPAR9. Among these QTLs, the region containing qCd-8-1 was the shortest in length and showed larger additive effects for low Cd accumulation, while qCd-2-1 was mapped to a large region and made the smallest contribution to lowering Cd content in rice grain (Table 1).
Allelic Variation In The Likely Candidate Genes
Several genes have been reported to regulate Cd uptake and/or transport and eight of them were located in the same regions as the QTLs identified in this study (Fig. 1). Two reported genes, OsYSL2 (Koike et al. 2004; Li et al. 2022) and OsCAL1 (Luo et al. 2018) associated with Cd transport were located on chromosome 2 in the region of qCd-2-1. OsMSRMK2, which could be transiently regulated by Cd in vitro (Agrawal et al. 2002), was located in the qCd-3-2 region. OsNRAMP1 (Takahashi et al. 2011a, b), OsNRAMP5 (Ishimaru et al. 2012; Sasaki et al. 2012), and OsHMA3 (Tezuka et al. 2010; Ueno et al. 2010) were mapped to the interval of qCd-7-1L and qCd-7-1H. OsMAPK2 (Yeh et al. 2004) was located in the qCd-8-1 region and OsMT (Hsieh et al. 1995; Yu et al. 2018) was identified in the qCd-11-1 region (Fig. 1). There were no reported genes associated with Cd accumulation in the regions carrying qCd-3-1, qCd-5-1 or qCd-6-1.
To explore the natural variation in these likely candidate genes, resequencing was performed across both the promoter and the coding regions of each gene using Sanger sequencing, and alleles from HJX74 were compared with those found in the donor parents of the SSSLs (Fig. 2). Four haplotypes were observed for OsNRAMP5. Six SNPs were detected in OsNRAMP5BG367 compared with HJX74; three were in the promoter and three were in the coding region, with nonsynonymous SNPs located at 4890 bp and 6840 bp. Eight variants were detected in OsNRAMP5Katy; seven in the promoter region and one SNP at 6840 bp in the CDS that was shared with OsNRAMP5BG367. OsNRAMP5Ganxiangnuo differed from OsNRAMP5HJX74 by a single, nonsynonymous SNP at 4890 bp that was also shared with OsNRAMP5BG367 (Fig. 2a). Two promoter haplotypes that differed by nine variants were found in OsNRAMP1, with OsNRAMP1HJX74 and OsNRAMP1BG367 carrying one of the promoter haplotypes, and OsNRAMP1Katy and OsNRAMP1Ganxiangnuo carrying the other. In the coding region, a SNP in OsNRAMP1BG367 and 4 SNPs in OsNRAMP1Katy differentiated them from OsNRAMP1HJX74 (Fig. 2b). Three haplotypes of OsHMA3 were identified in the SSSLs; OsHMA3Ganxiangnuo and OsHMA3HJX74 carried the same type, while two SNPs differentiated OsHMA3 Katy, and one InDel in exon7 was found in OsHMA3BG367 (Fig. 2c). Two haplotypes were discovered for each of the other five genes, OsYSL2, OsCAL1, OsMT, OsMSRMK2 and OsMAPK2 (Fig. 2d-h).
In cases where sequence variation in the promoter regions of genes differentiated HJX74 alleles from those of at least one of the donor parents, qPCR assays were employed to examine the transcript levels of those genes in the roots of the SSSLs. The transcript levels of OsNRAMP5Katy were similar to those of OsNRAMP5HJX74, while the transcript levels of OsNRAMP5BG367 and OsNRAMP5Ganxiangnuo were notably higher than in HJX74 (Fig. 3a). The transcript levels of the other Cd relevant genes carrying HJX74 haplotypes were significantly higher than those associated with donor parent haplotypes (Fig. 3b-h).
Development Of Pyramided Lines With Lower Cd Accumulation In Grain
Four QTLs, including qCd-2-1, qCd-3-1, qCd-7-1L and qCd-8-1, were selected to construct pyramided lines and six pyramided lines were developed using conventional crossing and maker assisted selection (Fig. S3). The pyramided lines were planted in two seasons (18SCS and 19FCS) at two experimental sites, Guangzhou (low Cd contamination) and Shaoguan (heavily Cd contaminated). Lower levels of Cd were found in rice grain from the SSSLs and the derived pyramided lines than in grain from HJX74, the recurrent parent. The Cd content was lower in all the pyramided lines and the maximum reduction of Cd was detected during 2019 FCS in Shaoguan, though Cd content varied across environments in all lines. We further discovered that the lines carrying qCd-3-1 & qCd-7-1L accumulated less Cd than other lines in most environments (Fig. 4, Supplementary Table 4). This was particularly notable in the pyramided line qCd-3-1/ qCd-7-1L; the grain cadmium content in this line was reduced by nearly 50% during 2019 FCS in Shaoguan (Fig. 4d; Supplementary Table 4).
High yield is always an important target in rice breeding. Tiller number, number of kernels per plant and 1000-kernel weight are the most important yield components in determining final yield, and these components of yield were therefore evaluated in this study. There was no significant difference for these traits between HJX74 and the pyramided lines grown in Guangzhou (Supplementary Table 5). These results strongly suggest that combining low Cd QTLs is an efficient and effective way to minimize rice Cd accumulation in rice without jeopardizing grain yield.