Introduction

Because phytic acid (PA) cannot be digested by monogastric animals, but can reduce the bioavailability of micronutritional elements like Zn and Fe, it is considered a major antinutrient in cereal and legume grains (Lott et al. 2000; Raboy 2001, 2009). To address PA-related nutritional and environmental implications, a new type of crop, low phytic acid (lpa) crops, featuring reduced PA content in seeds, has been advocated during the past two decades. A number of lpa lines have been generated through chemical or physical mutagenesis (reviewed by Raboy 2009) and recently by gene-silencing techniques (Kuwano et al. 2009; Ali et al. 2013a, b; Li et al. 2014a, b) in major crops.

Soybean meal is a very important protein source for humans and is commonly used in animal feed worldwide. In soybean seeds, 60–80 % of total P exists in the form of PA (Raboy and Dickinson 1984). Thus, a number of lpa soybean lines have been generated through mutagenesis as well as genetic engineering (Wilcox et al. 2000; Hitz et al. 2002; Shi et al. 2007; Yuan et al. 2007). In our laboratory, two lpa soybean mutants, GM-lpa-ZC-2 (ZC-lpa) and GM-lpa-TW-1 (TW-lpa), were developed through gamma irradiation and EMS treatment (Yuan et al. 2007). Subsequent investigations indicated that they not only have reduced PA content, but also other beneficial changes, such as reduced content of other antinutrients (e.g., galactinol, raffinose, stachyose) (Frank et al. 2009), and increased content of sugar in TW-lpa (Yuan et al. 2007) and isoflavone in ZC-lpa (Yuan et al. 2009). The two mutants have already been introduced into breeding programs in both China and abroad (USA and Canada) to develop value-added soybean cultivars. Molecular characterization revealed that ZC-lpa resulted from a G → A mutation in GmIPK1 (Yuan et al. 2012) and TW-lpa from a 2-bp deletion in GmMIPS1 (Yuan et al. 2007). These findings enabled the development of CAPS (cleaved amplified polymorphic sequences) markers for the two mutations (Yuan et al. 2007, 2012). Genotyping using CAPS markers depends on restriction enzymes and involves a number of laborious steps, i.e., cleavage of PCR fragments, electrophoresis of cleaved fragments and visualization of separated fragments by ethidium bromide or silver staining; hence, it is costly and time consuming and thus is not suitable for commercial breeding in which a large number of plants need to be genotyped within a short time period at a reasonable cost.

High-resolution melting (HRM) analysis is a method for differentiating genotypes based on fluorescence change of PCR amplicons saturated with double strand

DNA-intercalating dye over temperature increase; it is performed post-PCR without additional processing and data analysis (Montgomery et al. 2007; Studer et al. 2009) and can be used for genotyping and mutation screening (Gundry et al. 2003). Its simplicity, robustness and high throughput make it potentially useful for plant genotyping. HRM has been successfully used in soybean to identify SNPs in genes associated with fatty acid content (da Cruz et al. 2013) and in rice to genotype different lpa mutations (Tan et al. 2013). In breeding practice, genotyping that can be performed using DNA extracted with a simple protocol, particularly one without toxic chemicals, is highly desirable. In soybean, two protocols were recently reported for rapid DNA extraction; King et al. (2014) developed a DNA extraction method for soybean leaves using non-toxic reagents (but it involves many steps, and the whole process takes up to 100 min), and Ma et al. (2014) reported a rapid DNA extraction method that could be completed within 30 min, but it still used a toxic agent (2-mercaptoethanol).

The quality of DNA templates is a factor that affects the melting curves of DNA amplicons, and to our knowledge all HRM-based genotyping in plants has been performed with DNAs extracted using regular protocols. In the present work, we adopted a fast, simple and safe protocol (Xin et al. 2003) for DNA extraction from soybean leaf disks and developed a matching HRM-based genotyping method that directly uses the raw DNAs extracted.

Materials and methods

Low phytic acid mutants and breeding progenies

The two lpa soybean mutants, GM-lpa-ZC-2 (ZC-lpa) and Gm-lpa-TW-1 (TW-lpa), were previously developed using 60Co gamma rays and EMS mutagenesis (Yuan et al. 2007). Breeding populations were developed by crossing ZC-lpa to the wild-type (WT) cultivar “Cu” and TW-lpa to another WT line, “26001-8.” Segregating F5 populations were developed from heterozygous F4 plants, the genotypes of which were determined by high inorganic phosphorus (HIP) testing (Yuan et al. 2007, 2012). F5 plants were genotyped by regular HRM and CADMA-HRM analysis using DNA extracted by the CTAB method to produce F6 populations, the genotypes of which were determined by CADMA-HRM analysis using DNA extracted with a simple, fast protocol (Fig. S1). All soybean plants were grown in the experimental farm of the Zhejiang Academy of Agricultural Sciences in Hangzhou, Zhejiang Province, China.

Genomic DNA extraction

Genomic DNA was extracted from leaf tissues using either a modified CTAB method (Keim et al. 1988) (hereafter “CTAB-extracted DNA” or “CTAB extraction”) or a simple, safe and fast protocol adopted from Xin et al. (2003) (hereafter “SSF-extracted DNA” or “SSF extraction”). For SSF DNA extraction, leaf disks (Φ ~ 2 mm) were cut from seedling leaves about 30 days after germination using a hole puncher and placed in 96-well PCR plates. Fifteen microliters of solution A (100 mM NaOH, 2 % Tween 20, freshly prepared from 5 M NaOH and 20 % Tween 20 stock solutions) were added to each sample well, and the plate was frozen in a −70°C freezer for 10 min. The plate was then incubated at 95°C for 10 min, and 45 µl Buffer B (100 mM Tris–HCl, 2 mM EDTA) was added and mixed well by vortexing. The plate was centrifuged at 1500×g for 1 min; the supernatant was directly used for PCR. For CTAB extraction, DNAs were extracted from leaves of the same stage as for SSF, and if needed, adjusted to a final concentration of ~50 ng/µl using Nanodrop 2000 (Gene Company Ltd., Hong Kong, China).

PCR and genotyping by CAPS marker and HRM analysis

The position and sequence information for primers used for various PCRs are shown in Fig. 1. All primers were designed using the Primer Premier 5 software and synthesized by Shanghai Sangon Biological Engineer Technology and Services Co., Ltd. (Shanghai, China). Genotyping by CAPS markers was performed according to Yuan et al. (2010, 2012). Briefly, fragments of GmIPK1 and GmMIPS1 were first amplified using the primers IPK1-F1/R and MIPS1-F1/R (with expected sizes of 350 and 430 bp, respectively). The PCRs were performed in 20 µl volumes with 2 µl genomic DNA, 10 µl 10× master mix [containing 10× PCR buffer, 4 mM MgCl2, 0.4 mM dNTPs, 50 U/ml Taq DNA polymerase)] and 0.4 µl of each 10 µM primer. The amplicons were subsequently digested by Csp6I and HinfI, respectively. The cleaved amplicons were separated on 2 % agarose gels, visualized with ethidium bromide and documented using the VersaDoc Imaging System Model 3000 (Bio-Rad Laboratories, Inc., California, USA).

Fig. 1
figure 1

Schematics of GmIPK1 and GmMIPS1 showing the positions of lpa mutations (empty triangles) and primers (solid triangles) used for genotyping. Exons and introns are to scale and depicted as boxes and solid lines. Filled boxes are coding sequences and empty boxes are untranslated regions. The red letters “t” and “c” in IPK1-F2 and MIPS1-F2 are additional mutations introduced deliberately. (Color figure online)

Full size image

Several types of HRM analysis were performed to genotype the two lpa mutation loci. Regular HRM analysis was performed using fragments amplified with IPK1-F1/R and MIPS-F1/R. For the competitive amplification of differentially melting amplicons analysis (CADMA-HRM), three primers were designed to amplify amplicons for each mutation (Fig. S2) according to the method proposed by Kristensen et al. (2012). In detail, for ZC-lpa, a mutation-specific forward primer was designed to amplify a fragment containing the mutation with a common reverse primer (IPK1-R). The bold letter “t” with double underlines in the mutation-specific primer IPK1-F2 (ctctgaatttgcttcaaggagaga) is a second mismatching nucleotide, which was introduced deliberately (the underlined bold “a” is the mutated sequence). In addition, a second forward primer, IPK1-OLP, which overlaps the mutation-specific primer was designed to amplify both wild-type and mutated sequences (Fig. 1). Likewise, for TW-lpa, a WT-specific forward primer (MIPS1-F2) with an introduced mismatching nucleotide was designed to amplify the WT-specific fragment. MIPS1-OLP and MIPS1-R were the other two primers needed for CADMA-HRM analysis. Theoretically, in CADMA analysis, IPK1-OLP/F/R produced fragments with expected sizes of 191 and 183 bp, while MIPS1-OLP/F/R produced fragments of 188 and 180 bp. Practically, PCR products with a difference of 8 bp could not be separated on an agar gel for either mutation.

For HRM analysis, PCRs were performed in 10 µl volumes with 1 µl DNA templates and 0.5 µl 20× EvaGreen (Biotium, Hayward, CA, USA) in Gene Solution master mix (containing PCR buffer, MgCl2, dNTPs and Taq DNA polymerase). Three types of DNA templates were used: DNAs extracted using CTAB extraction with or without concentration pre-adjustment (to 50 ng/µl) and the supernatant solution from SSF extraction. A drop of mineral oil was added to each PCR reaction to prevent solution evaporation. The PCR conditions were optimized by gradient PCR, and the following generalized PCR profile was used: 5 min at 94°C, followed by 40 cycles of 30 s at 94°C, 30 s at 50°C (for regular HRM) or 52°C (for CADMA-HRM) and 30 s at 72°C, with a final extension at 72°C for 8 min.

After PCR amplification, the PCR plates were briefly centrifuged at 2000 rpm for 1 min and used directly for HRM with a LightScanner (Idaho Technology Inc., USA), by ramping the temperature from 55 to 95°C at 0.1°C per second. The data were analyzed using the LightScanner analytical software Call IT™ 2.0 (Idaho Technology Inc.) after normalization and temperature shifting of the melting curves according to the LightScanner Operator’s Manual (Idaho Technology Inc.). The fluorescence difference curves, which show the relative difference in fluorescence (ΔFluorescence, ΔF) of a test sample relative to a reference, were analyzed by unbiased clustering for differentiating samples according to the manufacturer’s instructions. Fluorescence curves with a peak ΔF ≥ 0.05 were considered to be significantly different (Hofinger et al. 2009).

Results

Genotyping with regular HRM analysis

To assess whether HRM analysis could be directly used to genotype the ZC-lpa and TW-lpa loci, fragments were amplified using CTAB-extracted DNAs and subjected to melting analysis. For the TW-lpa locus, the curves of homozygous WT and mutant individuals clustered together and were mutually indistinguishable, but both could be easily differentiated from heterozygous plants, regardless of whether the DNA concentration was pre-adjusted (Fig. 2a and Fig. S3). For the ZC-lpa locus, the differentiation was not clear cut for samples without DNA concentration pre-adjustment (Fig. S4), and the classification was inconsistent with the results of CAPS marker-based genotyping (Fig. S5 and Table S1). However, when the DNA concentration of samples was pre-adjusted to a common level before PCR, the melting curves were grouped into three distinguishable classes, representing heterozygous (F1 plant), homozygous WT (Cu) and mutant (ZC-lpa) genotypes, with ΔFs all greater than 0.05 (Fig. 2b).

Fig. 2
figure 2

HRM analysis for two soybean lpa mutations using DNAs extracted from leaf tissues with a simplified CTAB method. The DNA concentration is adjusted before PCR. H, M and W represent heterozygous, mutant and wild-type genotypes, respectively

Full size image

Genotyping by CADMA-HRM with CTAB-extracted DNA

Our previous study demonstrated that CADMA is a method of choice for differentiating genotypes that cannot be discriminated by regular HRM analysis (Zhang et al. 2014). Therefore, we subjected the same set of samples used in regular HRM analysis (Fig. 1a and Fig. S4) to CADMA-HRM analysis using CTAB-extracted DNA without concentration pre-adjustment. After optimizing the PCR setup, mainly the ratio of the primers (Fig. 1), three distinct types of melting curve were observed, all with ΔFs greater than 0.05, for samples of the TW-lpa × 26001-8 population (Fig. 3a). Likewise, samples of the ZC-lpa × Cu population were also differentiated into three groups with not only peak ΔFs greater than 0.05, but also quite different shapes (Fig. 3b). These plants were further genotyped using the respective CAPS marker; the results showed the two genotyping systems were 100 % consistent (Fig. S5 and Table S1).

Fig. 3
figure 3

Genotyping ZC-lpa and TW-lpa mutations by CADMA-HRM analysis using CTAB-extracted DNA without concentration adjustment. H, M and W represent heterozygous, mutant and wild-type genotypes, respectively

Full size image

Genotyping by CADMA-HRM with SSF-extracted DNA

A pretest showed that with ratios of ratio of absorbance at 260/280 nm of SSF-extracted DNAs (the supernatant) ranging from 1.04 to 1.41 (Table S2), indicating their quality was not as good as that of CTAB-extracted ones (with ratio of absorbance at 260/280 nm ~2.0). However, the concentration of SSF-extracted DNAs reached 106.0–485.2 ng/μl (Table S2), suggesting that 1 μl supernatant would provide sufficient DNA templates for PCR analysis. Hence, CADMA-HRM was performed using SSF-extracted DNAs, (1 μl supernatant) without concentration adjustment.

The results clearly showed that the samples of both F6 populations (ZC-lpa × Cu and TW-lpa × 26001-8) were grouped into three types, i.e., WT parent, mutant parent and heterozygote, with the ΔF between any two types in each population greater than the threshold value 0.05 (Fig. 4). To assess whether the results reflected real genotypes, all samples were subjected to CAPS marker analysis, and the results of which demonstrated that the two different genotyping analyses were completely consistent (data not shown).

Fig. 4
figure 4

Genotyping of TW-lpa and ZC-lpa mutations by CADMA-HRM analysis using SSF-extracted raw DNAs. H, M and W represent heterozygous, mutant and wild-type genotypes, respectively

Full size image

Discussion

Increasing the efficiency, cost-effectiveness and throughput of genotyping are a demanding task in molecular breeding. In the present study, we adopted a simple, fast and safe protocol for DNA extraction and a powerful HRM method for soybean genotyping and developed an HRM-based, fast and high-throughput genotyping system for two lpa mutations. Because the CADMA method can be applied to any gene of interest by designing gene-specific primers and optimizing the PCR setup, the system developed in the present study could also be applicable to other genes in soybean.

In breeding practice, it is highly desirable that genotyping be performed using DNA extracted with a simple and safe protocol, particularly one involving no toxic chemicals. Compared with the two DNA extraction protocols recently reported for rapid DNA extraction in soybean (King et al. 2014; Ma et al. 2014), the protocol adopted in the present study is either safer or simpler, because there is no toxic reagents in genotyping process and can be completed in less than 30 min involving only two steps.

HRM analysis has become one of the preferred choices for high-throughput genotyping because of its outstanding features including simplicity, low cost, ease of use and high sensitivity/specificity (Vossen et al. 2009). However, studies have shown that HRM analysis is less efficient for differentiating homozygous genotypes (Hofinger et al. 2009; Li et al. 2010; Tan et al. 2013), which was also observed in the present study, i.e., TW-lpa had HRM curves very similar to those of the WT (Fig. 2). Therefore, a number of variant HRM methods have been developed to overcome this weakness, e.g., pre-PCR mixing of DNA templates with a known WT or mutant type (Tan et al. 2013) or the use of small amplicons or unlabeled probe (Tan et al. 2013; Liew et al. 2004). While such amendments increase the differentiating power, additional work or costs are introduced, which automatically reduces the attractiveness. Although it was reported that HRM analysis could tolerate concentration variations of DNA templates up to tenfold (Montgomery et al. 2007), all studies reported so far have used DNA templates after concentration adjustment. In the present study, although the DNA concentrations were less than tenfold different (Table S1), the samples were not clearly differentiated using regular HRM analysis (Fig. S4), indicating that DNA concentration variation may play a role in regular HRM analysis.

CADMA was first developed by Kristensen et al. (2012) for screening and genotyping mutations in human genetic studies, but it was recently demonstrated to be equally useful for rice genotyping (Zhang et al. 2014). The advantage of CADMA rests in its increased power to differentiate genotypes, particularly homozygous genotypes. For example, TW-lpa and WT were not distinguished by regular HRM (Fig. 2a), but they were distinguishable in CADMA analysis (Fig. 3a). CADMA involves asymmetric PCRs with three primers at optimized ratios (here 1:10:10 for MIPS1-OLP:F2:R and 1:2:2 for IPK1-OLP:F2:R). This novel system generates two amplicons for one parent and one amplicon for another, significantly increasing their difference in melting (Fig. 1c). Therefore, not only the ΔF is greatly increased but the shape of the melting curves is also changed in CADMA analysis compared with regular HRM. Because the amplicon composition changes with a change of primer ratio, the difference between any two genotypes can be increased by optimizing the primer ratio. The third advantage of CADMA resides in its insensitivity to DNA quality and concentration variation. DNAs are often extracted using established protocols such as CTAB, and the concentrations are adjusted to similar levels before PCR in HRM analyses including CADMA analysis (Zhang et al. 2014). In the present study, the SSF-extracted DNAs not only had the quality inferior to that of CTAB-extracted ones, but also had concentration variations among samples; however, when the CADMA method was used, this problem was nicely solved (Fig. 3b), which further demonstrated the powerfulness of the CADMA method.

In conclusion, our present study introduced for the first time two important techniques into soybean genetic studies—a rapid, safe and simple DNA extraction protocol and the CADMA method—and thus developed a simple, fast and high-throughput system for genotyping two lpa mutations in soybean.