Identification and analysis of novel recessive alleles for Tan1 and Tan2 in sorghum

Plant materials

Sorghum accessions include wild sorghum resources, landraces and cultivars from all over the world, were collected from the Sorghum Research Institute, Liaoning Academy of Agricultural Sciences, China. Plants were grown at the experimental site of Liaoning Academy of Agricultural Sciences (Shenyang (41.8°N, 123.4°E)). Each sorghum accession was planted in a single 3-m-long plot with 0.6-m row spacing. Leaf tissue was collected, frozen in liquid nitrogen and stored at −80 °C for DNA extracting. Grains were harvested to determine the tannin contents.

DNA extraction

Leaves from each sorghum accession were sampled for genomic DNA extraction by the cetyltrimethylammonium bromide (CTAB) method as previously described with minor modifications (Allen et al., 2006). Add 2.5 volume of ethanol and incubate at room temperature for 10 min to precipitate DNAs. The mixtures were centrifuged under 13,000 g at room temperature for 10 min to pellet the DNAs. The DNA pellets were directly air-dried at room temperature for 5 min, and then dissolved in nuclease-free water.

PCR, DNA sequencing, and sequence analysis

To genotype Tan1 and Tan2 alleles in different sorghum accessions, primers were designed (Table S1). The PCR products were sequenced by Beijing Tsingke Biological Technology Co., Ltd. (Beijing, China). The DNAMAN program (version 5.2.2) was used for sequence alignment and translation of nucleotides into amino acids. Tan1-1F/ Tan1-1R primers were designed for detecting tan1-a and tan1-b (Table S1). To develop the CAPS (cleaved amplified polymorphic sequence) marker to detect tan1-c allele, Tan1-2F/ Tan1-2R combined with Dde I were designed for tan1-c (Table S1). PCR products were digested with Dde I and analyzed by 8% polyacrylamide gel electrophoresis. Because tan1-c lost a Dde I restriction enzyme site as a result of A-to-T transversion at position 1054 in the coding sequence, PCR amplification with Tan1-2F/ Tan1-2R resulted in a 164 bp product (G deletion at position 1057); whereas dominant Tan1 contained a single Dde I site in the corresponding PCR product (165 bp) and was cut into 109 bp and 56 bp fragments. Primers of Tan2-1F/Tan2-1R to Tan2-6F/ Tan2-6R were designed for detecting Tan2 alleles (Table S1).

Determination of tannin content by reagent test kit

Tannin was determined according to the Tannin Microplate Assay Kit (Cohesion Biosciences, CAK1060). Five grams of grains were crushed into powder in a grinder. Tissue samples (0.1 g) were homogenized with 1 ml distilled water, placed in a water bath at 80 °C for 30 min, and centrifuged at 8,000 g at 4 °C for 10 min. The supernatant was placed into a new centrifuge tube for detection. 10 μl sample supernatant, 160 μl distilled water and 20 μl reaction buffer were mixed and incubated for 5 min at room temperature. Then, 10 μl dye reagent was mixed for 10 min, and the absorbance was measured and recorded at 650 nm to calculate the tannin content. The tannin contents were scored as low (≤0.5%), medium (0.5%> tannin content <1.0%) and high (≥1.0%). In this study, medium tannin content accessions were not exhibited.

Mapping population

8R191 was a Chinese landrace without tannins. 8R306 was a landrace from Africa and had tannins in the grains. The RIL population “8R306 × 8R191” was obtained by advancing random individual F₂ plants to the F₆ generation by single-seed descent with 557 lines.

Chlorox bleach test

Chlorox bleach test was performed previously described with minor modifications (Dykes, 2019). A total of 100 sorghum grains was placed into a 100 ml beaker, and 15 ml 6% NaClO was added to fully immerse sorghum grains. The beaker was left to sit for 20 min at room temperature, and the contents were swirled in the beaker every 5 min. The reaction solution, was discarded then it was rinsed with distilled water 2–3 times, and poured on filter paper to remove excess water. All bleach tests were repeated three times. The presence or absence of tannins in sorghum grains was evaluated based on grain color after dyeing. Sorghum grains were divided into three types: Type I grains were completely black and had tannins, Type II grains were lighter brown black or had small black spots, and Type III grains were white or lightly colored and had no tannins. In 557 RILs, Type I had 269 lines, Type II had 179 lines and Type III had 109 lines. For the accuracy of phenotypic identification, Type I and Type III lines were used to map the new Tannin locus. After dyeing, one grain was selected from one line to use for germinating and the seedling was sampled to extract the genomic DNAs.

Mapping and identification of the candidate gene

Genomic DNAs were extracted from 8R191, 8R306 and RIL population plants using the CTAB method. BSA (Bulked Segregant Analysis) was used to determine the new Tannin locus. Equal amounts of genomic DNAs from 50 tannin (Type I, 50/269) and 50 non-tannin (Type III, 50/109) plants were pooled to construct the tannin and non-tannin bulks, respectively. Whole-genome resequencing was performed using Illumina HiSeq2000 platform by Beijing PlantTech Biotechnology Co., Ltd (Beijing, China). The depths of two bulks and two parental lines were about 50× and 10×, respectively. The reads were aligned to the Sorghum bicolor v3.1.1 reference genome (https://phytozome-next.jgi.doe.gov/info/Sbicolor_v3_1_1) using Burrows-Wheeler Aligner (BWA) software bwa-0.7.10 (Li & Durbin, 2009), GATK toolkit used to detect and filter SNPs (McKenna et al., 2010). The SNPs were employed as input in the R package “QTLseqr” version 0.7.5.2 for subsequent analysis (Takagi et al., 2013). To perform the SNP-INDEX analysis, the window size was set at 1 Mb. The significance of ΔSNP-index was determined at a 99% confidence interval and at least 10 SNPs within the window size. Ten InDel markers within the region were used for fine mapping. Information on molecular markers for fine mapping was provided in Table S1.

Relationships between Tan1 and Tan2 genotypes and tannin contents

The presence of tannins in sorghum grains is regulated by a pair of genes (Tan1 and Tan2), and both genes have three recessive alleles (Wu et al., 2019, 2012). Twenty accessions were used to determine the relationship between Tan1 and Tan2 genotypes and tannin contents in sorghum grains. Primers of Tan1-1F/ Tan1-1R, Tan1-2F/ Tan1-2R and Tan2-1F/Tan2-1R to Tan2-6F/ Tan2-6R were used to detect Tan1 and Tan2 alleles in different sorghum accessions. Tannin presence was scored by the Tannin Microplate Assay Kit for sorghum grains. As shown in Table 1, homozygous recessive genotypes at one or both genes can cause low-tannin phenotypes, and two wild sorghum resources had high tannin contents because they carry dominant alleles, which was consistent with the reported data (Wu et al., 2019). However, six accessions carrying Tan1 and Tan2 dominant alleles had low tannin contents in sorghum grains, indicating that there may be variation in unknown genes contributing to tannin production or new alleles of the known regulatory genes (Table 1).

A novel recessive allele of Tan1 in sorghum landrace

To identify new tannin genes or new alleles, a recombinant inbred line (RIL) population was built from 8R191 and 8R306. Sequencing and digestion were used to determine the Tan1 and Tan2 alleles. Amplifying and sequencing by Tan1-1F/ Tan1-1R indicated that 8R191 and 8R306 didn’t have tan1-a and tan1-b (Table S1). A CAPS marker was designed to detect the tan1-c allele. 165 bp PCR products with Tan1-2F/2R primers were digested by Dde I, 109 bp and 56 bp DNA fragments of dominant Tan1. Because of A-to-T transversion at position 1054 in the coding sequence of tan1-c, the PCR products of Tx2752, OK11 and Tx631 remained uncleaved (Table S1, Fig. 1) (Wu et al., 2019). The PCR product of 8R191 and 8R306 can cleave indicating that two parents did not have tan1-c. Six primers were used to detect Tan2 genotypes in 8R191 and 8R306 (Table S1). Tannin presence was scored by the chlorox bleach test for 8R191 and 8R306 grains. 8R191 is a non-tannin landrace and 8R306 is a tannin landrace, both of them carrying Tan1 and Tan2 dominant alleles.

We performed BSA using the 8R191/8R306 RIL population lines. Equal amounts of genomic DNAs from 50 tannin and 50 non-tannin plants were pooled to construct the tannin and non-tannin bulks, respectively. The 8R191(non-tannin parent), 8R306(tannin parent), and non-tannin, tannin bulks were subjected to Illumina high-throughput sequencing, from which, 70.99, 72.90, 290.71 and 280.82 million paired-end reads were produced, representing 14×, 14×, 54×, and 53× genome coverage, respectively (Table S2). Among them, 98.23%, 97.88%, 94.35% and 95.91% reads could be mapped to the Sorghum bicolor v3.1.1 reference genome, respectively, indicating good quality of the sequencing data (Table S2). Using the BSA-Seq method, we obtained only one region spanning 6.60 Mb on Chr4 between 57,400,000 to 64,000,000 was strongly associated with the tannin phenotype (Fig. 2A). Within this region, we developed 10 available InDel markers for fine mapping (Fig. 2B). Using a RIL population of 378 plants (Type I: 269 lines and Type III: 109 lines), the new Tannin locus was finally narrowed down to a 25.8 kb region defined by markers InDel-7 and InDel-8. We identified four candidate genes in this region, including Sobic.004G280700, Sobic.004G280800, Sobic.004G280900 and Sobic.004G281000 (Fig. 2C, Table S3). Sobic.004G280800 is Tan1, suggesting that Sobic.004G280800 is likely to be the causal gene. Primers (Tan1-F/Tan1-R; Table S1) were used to detect the sequence polymorphisms in Tan1 between 8R191 and 8R306. 8R306 had dominant Tan1, however, 8R191 had deletion and substitution in coding region of Tan1, and named as tan1-d. Therefore, Tan1 has another new genotype that affects its function.

Identification and distribution of Tan1 and Tan2 other new alleles

To identify more different Tan1 and Tan2 alleles, we collected 396 sorghum accessions, including wild sorghum resources, landraces and cultivars (Table S4). DNA was extracted using leaves by CTAB method. The PCR primers Tan1-F/Tan1-R were designed for detecting Tan1 alleles. Primers from Tan2-1F/Tan2-1R to Tan2-6F/ Tan2-6R were designed for detecting Tan2 alleles (Table S1). Tannin content was determined according to the Tannin Microplate Assay Kit. According to the coding sequence variation and low-tannin content, the new variation types were determined for Tan1 and Tan2.

Another novel Tan1 recessive allele was found, named as tan1-e. Among 396 sorghum accessions, 10 wild sorghum resources and 200 high-tannin-producing accessions carried dominant Tan1, and 89 low-tannin-producing accessions carried tan1-a, tan1-b, tan1-c, tan1-d, and tan1-e alleles. A total of 97 accessions carried the Tan1 allele, but had low tannin contents, accounting for 24.5% (97/396) of the total sample size (Table 2, Table S4). Maybe there was allelic variation of Tan2 in low-tannin-producing accessions with dominant Tan1. Then seventy-two accessions carrying the Tan1 allele were used to detect different Tan2 alleles, including 38 low-tannin-producing accessions, 24 high-tannin-producing accessions and 10 wild sorghum resources. Ten wild sorghum resources and 24 high-tannin-producing accessions had dominant Tan1 and Tan2 (Table 3, Table S5). Four different recessive alleles of Tan2 were identified, named as tan2-d, tan2-e, tan2-f, and tan2-g (Table 3, Table S5). 24 sorghum accessions carried the dominant Tan1 and Tan2 alleles, but had low tannin contents, accounting for 63.2% of 38 low-tannin sorghum accessions (Table 3). Our analysis indicates that there may be existed other unknown loci involved in tannin production.

More importantly, Tan1 and Tan2 recessive alleles had obvious regional distribution characteristics. Novel tan1-d, tan2-d, tan2-f and tan2-g alleles were distributed worldwide, including Ghana, France, Mexico, India, China and so on, but tan1-e and tan2-e were only found in Chinese landraces (Tables S4 and S5). The results implied that some Tan1 and Tan2 recessive alleles had different geographical distributions.

Functional variation of the newly identified Tan1 and Tan2 recessive alleles

Tan1 and Tan2 are conserved regulatory factors in the plant tannin synthesis pathway and have higher nucleotide similarity within major cereal crops other than rice, wheat and maize, which do not produce tannins in their grains. Tan1 encodes a WD-40 repeat protein and has four WD-40 repeat domains. Deletion, substitution and insertion mutations in tan1-a, tan1-b, and tan1-c have caused frame shifts and premature stop codons, leading to disruption of the highly conserved region of WD-40 domain and C-terminus and resulting in the absence or low level of tannins in sorghum grains. Seven independent mutations of the TTG1 gene reveal that the truncation of the C-terminal region and WD-40 domain produced nonfunctional alleles in Arabidopsis, indicating that the C-terminal region and WD-40 domain are vital for the structure and function of WD-40 protein (Wu et al., 2012). In tan1-d, A-to-T transversion at position 1054, GT deletion at positions 1057 and 1058, and C-to-T transition at position 1059 in the coding sequence affected TGA (at positions 1060, 1061 and 1062) stop codon frameshift and led to nonfunctional protein. Because of mutation and deletion, tan1-d had 1086 bp coding sequence and 361 aa protein sequence. Sequence variation of tan1-d was similar to tan1-c and four WD-40 domains presented, but the C-terminal sequence had changed greatly in tan1-d. Compared to dominant Tan1, tan1-e had a 10-bp deletion (CGACATACGT) in the coding sequence between positions 771 and 780. The 10-bp deletion caused a frameshift mutation and resulted in a truncated protein with a length of only 295 aa. The fourth WD-40 domain and C-terminal region were dramatically changed in tan1-e (Figs. 3 and 4, Fig. S1).

Tan2 encodes a bHLH transcription factor with 10 exons and nine introns in BTx623. tan2-a has a 5-bp (CTCCC) insertion in the 8^th exon, tan2-b has a 7-bp (CCACAGA) insertion in the 8^th exon and tan2-c has a 95-bp deletion removing the entire 8^th intron (Fig. 5 and Fig. S2). Three mutations lead to frame shift, disrupt the bHLH domain and result in non-tannin-producing or low-tannin-producing phenotype (Wu et al., 2019). tan2-d, with the causal polymorphism of a 1-bp C deletion at position 563 in the coding region, led to a truncated protein with a length of only 210 aa. Because of the C-to-T transition at position 1366 (CAG to TAG) in the coding sequence, tan2-e resulted in premature termination and had a 455 aa protein. tan2-f contained a frameshift mutation and an early termination site because of an 8-bp (AGCTGATC) insertion between positions 1375 and 1376 in the coding region, resulting in a 462 aa protein sequence. tan2-d, tan2-e and tan2-f disrupted the bHLH domain structure and lost function. tan2-g, a null allele, had multiple substitutions and insertions from position 1579 to 1607 in the coding region and didn’t disrupt the bHLH domain structure (Figs. 5 and 6, Fig. S2).

Tan1 and Tan2 alleles utilization in breeding programs

By investigating the distribution of Tan1 and Tan2 alleles in sorghum cultivars, we can determine which alleles have been used in breeding. 87 cultivars (from China and foreign countries), as well as 34 sterile lines and 43 restorer lines, were used to detect the different alleles of Tan1 and Tan2 (Tables S4 and S6). For Tan1, only tan1-a, tan1-b and tan1-c alleles were detected in Chinese cultivars, tan1-d and tan1-e may not have been inherited in low-tannin-producing cultivars (Tables S4 and S6). For Tan2, tan2-a, tan2-b and tan2-c were detected in cultivars in the reported data (Wu et al., 2019). In our study, tan2-f and tan2-g alleles were detected in cultivars, and tan2-d and tan2-e alleles were not (Tables S4 and S5). More importantly, tan1-e and tan2-e were only detected in Chinese landraces (Table S5). The results showed that only some Tan1 and Tan2 alleles were applied in breeding, leading to a decrease in the diversity of breeding resources.