Design and application of the HbGBTS80K liquid chip in rubber tree

Molecular markers, also known as genetic markers, refer to specific DNA fragments that can reflect the differences among biological individuals. Developing molecular markers closely related to the target traits is the prerequisite for molecular marker-assisted breeding. Traditional molecular markers such as restriction fragment length polymorphism (RFLP)^[1], random amplified polymorphic DNA (RAPD)^[2], amplified fragment length polymorphism (AFLP)^[3], and simple sequence repeats (SSR)^[4] are limited in their application in modern breeding due to their relatively low coverage of the genome. Single nucleotide polymorphisms (SNPs) refer to the DNA sequence polymorphisms caused by the variation of a single nucleotide at the genomic level. As SNP is the most common type of genetic variation, it has evolved into one of the most important molecular markers in the study of plant genetic variation in recent years^[5−7].

SNP-based gene chips are a useful platform for genetic diversity, evolutionary analysis, target gene identification, and more. Traditional gene chips mainly refer to solid-phase chips, which work by in-situ synthesizing oligonucleotides in an ordered manner on a solid support, and then hybridizing with labeled samples to be tested. A fluorescence detection system scans the chip and genetic information of the sample to be tested is obtained through software analysis. Solid-phase chips have achieved significant development in the field of molecular breeding in the past few decades^[6,8−10]. However, due to their own shortcomings, such as the high cost of developing and applying loci and the physical fixation model of target loci, their application is increasingly limited. Recently, researchers have developed liquid-phase chips represented by SNP-based genotyping by targeted sequencing (GBTS)^[11−13]. Compared with solid-phase chips, liquid-phase chips are more flexible, more cost-effective, and have lower requirements on the detection platform, and have been applied in cotton^[7,12], soybeans^[14], cucumbers^[6], rice^[15,16], etc.

Natural rubber (NR) is an essential industrial raw material. Over 2,000 plant species worldwide can produce natural rubber^[17]. Among them, the rubber tree (Hevea brasiliensis) is the primary source of NR, supplying more than 98% of the global demand^[18]. The laticifer located in the bark of the rubber tree is the site for NR biosynthesis and storage^[19]. The rubber tree is a perennial crop of the Euphorbiaceae family and is native to the Amazon Basin in South America^[20]. Unlike other crops (such as rice, wheat, and soybean) that have undergone domestication for hundreds or thousands of years, the domestication of the rubber tree is less than 150 years^[21,22]. As a perennial tree with a long juvenile period, traditional rubber tree breeding is very time-consuming, usually taking 30-40 years. There is an urgent need to develop a breeding chip to accelerate the efficiency of rubber tree breeding.

Based on the high-quality rubber tree genomic data and the advantages of the GBTS approach, we developed the first rubber tree liquid SNP chip, termed 'HbGBTS80K'. Research on genetic diversity and target gene identification demonstrated the high efficiency and accuracy of the HbGBTS80K chip, indicating its vast potential in rubber tree molecular breeding.

In this study, a total of 256 wild germplasms and 148 cultivars planted in the field of the National Tropical Plants Germplasm Resource Center-Rubber Tree (Danzhou, Hainan Province) were utilized (Supplemental Table S1). Among them, 335 accessions (including 208 wild germplasms and 148 cultivars) were re-sequenced previously and utilized for the design of the liquid-phase chip, while the remaining 69 accessions (including 48 wild germplasms and 21 cultivars) were used for the quality control of the liquid-phase chip.

DNA isolation

The genomic DNA of 69 accessions was extracted using the DNAsecure Plant Kit (Tiangen Biotech, China). The integrity and quality of the total DNA were evaluated in 1.5% agarose gel and by NanoDrop 2000 (Thermo Fisher Scientific, USA), respectively.

Liquid-phase chip design and quality control

The VCF files containing variation information of 335 re-sequenced accessions were used for the design of the liquid-phase chip. Briefly, 96,044 SNPs were selected based on the criteria of minor allele frequency > 0.05, deletion rate < 0.2, and heterozygosity rate < 0.3. The probes were designed using GenoBaits Probe Designer (Molbreeding Biotech., Shijiazhuang, China) and used for hybridization capture sequencing of 69 test accessions according to the protocol of the GenoBaits DNA seq Library Prep kit (Molbreeding Biotech., Shijiazhuang, China). The captured sequencing results were sorted in the order of priority of gene region > gene promoter region > gene downstream region > gene intergenic region. Finally, 80,080 high-confidence SNPs were retained to obtain the HbGBST80K liquid-phase chip (Supplemental Table S2).

Bioinformatics analysis

A phylogenetic tree with 100 bootstrap replicates was constructed using RAxML (version 8.2.12), PCA was performed using SNPRelate (version 1.16.0), and the population structure of the rubber tree accessions was determined using fastStructure (version 1.0). The genotypic data of NLR reported previously was used herein. For GWAS analysis, the Q-matrix corrected linear regression model (GLM-Q) of the PLINK software was applied. The genome-wide significance threshold of the GWAS corresponding to the raw P values was set to −log10P > 4.

We previously assembled a high-quality rubber tree cultivar 'CATAS8-79' genome and re-sequenced 335 accessions with an average depth of ~20×^[19]. After aligning the re-sequencing data with the 'CATAS8-79' genome, a total of 5,323,701 SNPs were generated and used as a panel for the design of the liquid chip (Fig. 1). Based on the criteria of minor allele frequency > 0.05, deletion rate < 0.2, and heterozygosity < 0.3, 96,044 SNPs were selected for the design of capture probes. All probes were evaluated using an additional 69 accessions. Finally, 80,080 high-confidence SNP sites were retained (Supplemental Table S2), referred to as the HbGBST80K liquid-phase chip.

Figure 1. 

Design pipeline of the rubber tree HbGBTS80K chip.

Characteristics of the HbGBTS80K chip

All 80,080 SNPs were distributed across the genome, with an average of 4,449 SNPs per chromosome (Fig. 2a). Among them, Chr6 had the maximum number of SNPs (7,128), while Chr1 had the minimum number of SNPs (2,642). Gene annotation revealed that 35.20% (28,191) of the SNPs were located in the intergenic region, 30.80% (24,667) were in the exonic region, 20.46% (16,386) were in the intronic region, 6.93% (5,550) were in the upstream region, and 6.60% (5,286) were in the downstream region (Fig. 2b).

Figure 2. 

Characteristics of the HbGBTS80K chip. (a) SNP marker density throughout the genome. (b) SNP locations within the genome.

HbGBTS80K chip application to population genetic diversity analysis

ADMIXTURE analysis showed that all accessions could be clearly classified into four groups by 80,080 SNPs (Fig. 3). Group I was mainly composed of cultivars, while Groups II, III, and IV were mainly composed of the wild germplasms from the Acre (WAC), Rondonia (WRO), and Mato Grosso (WMG) States of Brazil, respectively. In addition, the principal component analysis and the structure analysis obtained similar results, suggesting that the HbGBTS80K chip exhibited high accuracy in the genotyping of rubber tree germplasms.

Figure 3. 

Application in population genetic diversity analysis. (a) ADMIXTURE analysis. (b) Principal component analysis. (c) Structure analysis.

HbGBTS80K chip application to genome-wide association study

The number of laticifer rings (NLR) is the major trait for NR yield. Using re-sequencing data, a major gene related to NLR was identified by GWAS on 208 individuals^[19]. To evaluate the accuracy of the HbGBTS80K chip, the genotypic data of NLR reported previously was reanalyzed for GWAS analysis. The results showed that a strong association signal was detected in Chr15 (Fig. 4). Around the signal peak, the major gene controlling NLR, HbPSK5 was also detected, suggesting high consistency with that from resequencing.

Figure 4. 

Application in GWAS. (a) Manhattan plots and Q-Q plots for NLR. (b) Significant markers around the peak signal.

Liquid-phase chip based on GBTS represents the next-generation biochip. Compared with the solid-phase chip, the liquid-phase chip has more advantages^{[13, 23]}. (1) Higher accuracy: GBTS can perform multiple amplifications and sequencing of the target sites to ensure accuracy. (2) Lower price: Without the complicated chip hybridization work, the price is only one-fifth that of the solid-phase chip. (3) More convenient writing loci: Due to the use of PCR amplification technology, the targeted sites can be added or removed conveniently. In the past few years, the liquid-phase chip has been widely used in the biological breeding of more than 100 animals and plants^[24,25].

Due to the lack of genomic data, the development of molecular markers for rubber trees has been slow for a long time, seriously restricting breeding innovation. With the development of next-generation sequencing technology, several rubber tree genomes have been published^[19,26−30], and hundreds of germplasms from different sources have also been re-sequenced, providing sufficient data for the design of liquid-phase chips. Using the high-quality CATAS8-79 genome and the high-depth re-sequencing data of 335 accessions^[19], we designed the first rubber tree liquid-phase chip, HbGBTS80K. The selection of loci avoids highly repetitive sequences and leans more towards gene-rich regions. As evidence, 64.80% of the loci of HbGBTS80K are located in the gene body region (exon, intron, up-/down-stream), which will help to identify functional genes in subsequent use.

Population diversity analysis is a prerequisite step for evolutionary analysis and the identification of trait-related genes. By using RAPD markers and SSR markers, researchers have shown that wild germplasms have diverged to some extent^[31]. Using 336,200 SNPs, Chao et al. demonstrated that wild germplasms can be classified into three groups, and the cultivars are derived from the wild germplasms in the Mato Grosso State^[19]. In the present study, HbGBTS80K can also clearly divide wild germplasms into three groups, suggesting that the chip achieves the accuracy of high-depth resequencing in genetic diversity, while greatly reducing the number of SNPs.

Unlike other crops such as rice, soybean, and wheat, the mechanism of rubber tree yield formation is very complex. Tapping, which involves cutting laticifer rings to obtain latex, is the only way to harvest NR^[19]. This process is not only influenced by genetic factors such as laticifer differentiation and NR biosynthesis, but also by environmental factors such as temperature and tapping method. Although attempts have been made to use GWAS to identify NR yield-related genes, the results have not been satisfactory. Recently, we have shown that NLR is the major trait of NR yield and identified a major gene HbPSK5^[19]. In the present study, performing GWAS on the HbGBTS80K platform can also identify the HbPSK5 gene, reinforcing the fact that HbPSK5 is the major gene for NLR and that the HbGBTS80K chip has high sensitivity in target gene identification by GWAS. Additionally, because most SNPs of HbGBTS80K are located in the gene body, it is easier to obtain functional genes by GWAS compared to high-depth re-sequencing.

In this study, a liquid SNP chip named 'HbGBTS80K' was developed. This SNP chip comprised 80,080 SNPs evenly distributed across 18 chromosomes. The application of HbGBTS80K in population genetic diversity analysis effectively distinguished 404 rubber accessions into four groups. GWAS analysis of NLR could also detect the major gene HbPSK5 by using the HbGBTS80K platform. This newly developed HbGBTS80K liquid SNP chip is a valuable tool that will facilitate functional studies and molecular breeding of rubber tree.

The authors confirm contribution to the paper as follows: study conception and design: Chao J, Zhang J, Tian W; data analysis: Li Y, Chen X, He Y, Yue Z; sample collection: Li Y, Yang S; draft manuscript preparation: Chao J, Li Y. All authors reviewed the results and approved the final version of the manuscript.

The genome assembly that support the findings of this study are available in the Zenodo repository: https://doi.org/10.5281/zenodo.7123623. The raw sequencing data used in this study are available in the National Genomics Data Center (NGDC, https://ngdc.cncb.ac.cn/) under project number PRJCA004986.