Detection of 30 bp DNA fragments with a sensitive modified Southern blot analysis

ABSTRACT

To evaluate crops generated by new breeding techniques, it is important to confirm the removal of recombinant DNAs (rDNAs) derived from foreign genes including unintentionally introduced short rDNA(s). We attempted to develop a sensitive detection method for such short rDNAs using Southern blot analysis and performed a model study targeting single-copy endogenous genes in plants. To increase the detection sensitivity, the general protocol for Southern blot analysis was modified. In the model study, we used endogenous-gene-targeting probes in which complementary sequences were serially replaced by dummy sequences, and detected complementary sequences as well as 30 bp. We further evaluated the sensitivity using short rDNAs derived from GM sequences as pseudoinsertions, and the results demonstrated that rDNA-insertions as small as 30 bp could be detected. The results suggested that unintentionally introduced rDNA-insertions were 30 bp or more in length could be detected by the Southern blot analysis.

Graphical abstract

Short sequences, equal to or longer than 30 bp, could be detected by a modified Southern blot analysis. The method would be used for an evaluation of crops generated by new breeding techniques.

KEYWORDS:

• Southern blot analysis
• recombinant DNA
• detection
• short sequence
• plant genome

Plant breeding is a general term for techniques for improving the traits of plants in order to produce desired characteristics. Technological innovation in plant breeding is necessary to address population growth and climate change on a global scale. Conventional mutation breeding using chemical mutagens or ionizing radiation has been used to create new genetic variations, but screening for the desired mutants is time-consuming, expensive, and labor-intensive. Thus, more effective technology is required, and plant breeding techniques have been steadily evolving. Genetically modified (GM) technology has been applied to improve the production of crops. Many GM crops have been developed, and the global area of GM crops has been continuously increased [1]. The utilization of GM crops has generated substantial economic benefits, but, nevertheless, has been subjected to rigid control. Many countries including Japan legislate labeling systems of authorized GM crops and/or their derived foods. To enforce the labeling system, practical detection methods are indispensable. For the detection of genetically modified organisms (GMO)s, polymerase chain reaction (PCR) has been commonly used worldwide [2–5].

Several new techniques were defined as New Plant-Breeding Techniques (NPBTs) by the European Commission Joint Research Centre, which were methods allowing the development of new plant varieties with desired traits using recombinant nucleotide(s) in plant cells [6,7]. Among these techniques, genome editing that enables the precise and efficient targeted modification of an organism’s genome using sequence-specific nucleases (SSNs) is considered to be a powerful and promising tool for crop improvement [8,9]. Many kinds of genome-edited agricultural products have been developed for practical use [10–13]. In the NPBTs, recombinant nucleotides, including recombinant DNA (rDNA), are utilized during the development of crops with desired traits, and in many cases, these intentionally introduced recombinant nucleotides are later removable. Intentionally introduced rDNA(s) such as SSN genes can be removed as negative (null) segregant plants in which the rDNA(s) have been segregated out through conventional crossing. For practical use, it is important to confirm the elimination of not only intentionally introduced rDNA(s) but also unintentional insertion of small exogenous DNA fragments derived from recombinant nucleotides.

However, it is difficult to detect such unintentional insertion(s). PCR-based methods rely on the availability of the target DNA sequences because forward and reverse primers have to be designed for PCR amplification; in other words, PCR is not applicable for the detection of unintentionally introduced rDNA(s). In addition, the two primers are usually designed to be 20–25 nucleotides in length [14], implying that the detection of rDNA(s) of less than 40 bp is difficult with general PCR techniques.

Southern blot analysis is a classic technique used to detect a specific sequence within genomes [15]. In brief, the protocol of Southern blotting is as follows: genomic DNA is first digested with restriction enzymes, and the resulting fragments are separated by electrophoresis. The DNA is then denatured and transferred to a nylon or nitrocellulose membrane. The membrane is exposed to a hybridization probe, which is a DNA or RNA fragment complementary to a specific sequence in the target. The probe is normally labeled by incorporating radioactivity or tagging the molecule with chemiluminescent-, or chromogenic dye depending on the detection system. After the removal of unbound probes, bands complementary to the probe appear and indicate the presence of the target sequence in the analyzed genomic DNA [15,16]. Southern blot analysis is sometimes utilized to estimate the copy number of the target sequence by analyzing the size and number of the bands with different restriction enzymes. For the commercialization of GM crops, rigorous safety assessment of these crops is required, and their molecular characterization is indispensable as part of that assessment. Southern blot analysis is one of the standard methods of obtaining basic information from GMOs for the molecular characterization [17,18].

In this study, we attempted to develop an evaluation method for detecting unintentionally introduced rDNA by Southern blot analysis. We found that several modifications of the Southern blot analysis procedure enabled it to detect short sequences, i.e., sequences equal to or longer than 30 bp using a model study. Via these modifications, the old technique may be used for the characterization and regulation of crops that are generated by new breeding techniques.

Materials and methods

Plant materials

For Southern blotting analyses, plant genomic DNAs extracted from plant leaves were used. Rice leaves (Nipponbare) were kindly provided by the Institute of Agrobiological Sciences, NARO (Ibaraki, Japan). Tomato (Solanum Lycopersicum) and maize were grown in growth chambers at 27°C with a 16 h light/8 h dark cycle, and fully expanded leaves were used for the experiments. Tomato and maize seeds were purchased from local markets in Japan.

DNA extraction

Genomic DNAs were extracted using a Nucleon Phytopure extraction kit (GE Healthcare, UK). The concentration of the extracted DNA solutions was determined by measuring ultraviolet absorbance with an ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE), and the DNA purity was evaluated using the A260/280 and A260/A230 ratios.

Southern blot analysis

For Southern blot analyses, 5, 12, and 30 μg samples of genomic DNA were used for rice, tomato, and maize, respectively. Plant genomic DNAs were digested with restriction enzymes, separated by 1.0% agarose gel electrophoresis using the Mupid-exU system (Takara, Japan). The agarose gel is 1 X TAE (Tris-acetate-EDTA) 120 mL and 9 mm thick. The electrophoresis was performed at 50 V for 2.5 h at 25°C. After the electrophoresis, the gel was first soaked in the depurination solution (0.25 M HCl) for 12 min, although the depurination was only performed for Figure 1(a,b), and (c), and the solution was changed into the denaturation solution (0.5 M NaOH, 1.5 M NaCl) and soaked for 60 min, and then the solution was changed into the neutralization solution (0.5 M Tris-Cl, (pH7.5), and 1.5 M NaCl) and soaked for 60 min. The gel was blotted onto a nylon membrane, Hybond N+ (GE Healthcare). Transfer of DNA from agarose gel was performed with the paper capillary transfer method using the Pad type blotter A-set (Taitec, Japan). The transfer process was proceeded more than 16 h. These steps were performed at 25°C. Fixation of DNA to the membrane was performed by UV crosslinking using ultraviolet crosslinker, UVC500, (GE Healthcare). The amount of energy for UV crosslinking was set at 70,000 μJ/cm². For hybridization, the AlkPhos Direct Labeling and Detection Systems (GE Healthcare) was used. For probe preparation, 100 ng of 1.0 kb or 500 bp DNA fragments were labeled with alkaline phosphatase (AP) and used as probes. For hybridization buffer, 0.5 M NaCl and 4% blocking reagent was added into the AlkPhos Direct hybridization buffer and used. Hybridization was carried out at 55°C more than 16 h. After hybridization, the blotted membrane was transferred to the primary wash buffer (2 M Urea, 50 mM Na phosphate buffer (pH 7.0), 150 mM NaCl, 0.1% SDS, 1 mM MgCl₂, 0.2% blocking reagent), and washed for 10 min at 55°C. The primary wash was repeated in the fresh buffer at the same condition. The membrane was then transferred to the secondary wash buffer (50 mM Tris base (pH 10.0), 100 mM NaCl, 2 mM MgCl₂), and washed for 5 min at 25°C. The secondary wash was also repeated in the fresh buffer. For signal detection, CDP-Star Detection Reagents (GE Healthcare) which is a chemiluminescent substrate for AP was used. The blotted membrane was covered with the detection reagent, light-shielded and left for 2 h at 25°C. Signals were then detected with a chemiluminescence imager, Amersham Imager 600 (GE Healthcare). When several nonspecific signals appeared, Auto detect function with the Amersham Imager 600 software was used to automatically detect bands. If nonspecific signals were detected as bands, we calculated the signal-to-noise ratio (SNR) values. To calculate the SNR, signal intensities of each band were divided by those of background noise in the absence of signals on the membrane.

Figure 1. Southern blot analyses targeting the rice phospholipase D (PLD) gene using DNA probes of 1,000 bp in length. The following probes were used. (a) 100% match probe, (b) 20% match probe containing 200 bp of a complementary sequence of PLD and 800 bp of HPT, and (c) and (d) 10% match probe containing 100 bp of PLD and 900 bp of HPT. In (d), the depurination step was omitted. The predicted molecular sizes of positive bands were as follows; HindIII 5,890 bp, KpnI 7,629 bp, SphI 3,502 bp

For the preparation for probes of DNA labeling by PCR-, and for transcriptional RNA labeling, the DIG system (Roche) was used. The PCR probe was prepared using PCR DIG Probe Synthesis Kit (Roche), and the RNA probe was prepared using the DIG RNA Labeling Kit. The PCR and RNA probes were labeled with DIG-dUTP by the method of PCR and RNA polymerase, respectively. Probe preparation and hybridization were performed according to the DIG Application Manual (Roche). The DIG-labeled probes were detected by an AP conjugated anti-DIG antibody. For signal detection, CDP-Star and Amersham Imager 600 were used in a similar manner as above.

Evaluation of the detectable limit of DNA length for Southern blot analysis

The rice endogenous gene, phospholipase D (PLD) (GenBank accession number, AB001919) was used as a model to evaluate the detectable limit of DNA length for Southern blot analysis. As a 100% match probe, a sequence derived from PLD corresponding to nucleotides 3,521 to 4,520 was used. The hygromycin phosphotransferase (HPT) gene (GenBank accession number, KU561939) was used as a dummy sequence for the replacement within the PLD probes. The sequence in the PLD probe was serially replaced by the HPT sequence. For PLD C30 probe preparation, a 30-bp fragment of PLD corresponding to nucleotides 3,756 to 3,785 was sandwiched by 235-bp fragments of the HPT corresponding to nucleotides 21 to 255 and 286 to 520. For rice sucrose phosphate synthase (SPS) detection, the probe consisted of a sequence derived from SPS (GenBank accession number, D45890) corresponding to nucleotides 2,415 to 2,444 and a partial sequence of the cauliflower mosaic virus 35S promoter (P35S) derived from pBI121 (GenBank accession number, AF485783) that was used as a dummy sequence. For detection of the tomato endogenous gene, the sequence derived from LAT52 (GenBank accession number, X15855) was used. Partial sequences corresponding to nucleotides 1 to 50 or 1 to 30 of LAT52 were connected to the HPT and used as probes. For detection of the maize endogenous gene, the sequence derived from high mobility group A (HMG) (GenBank accession number, AJ131373) was used. As a 100% match probe, a partial sequence derived from HMG corresponding from nucleotide 1 to 500 was used. For a 6% (30 bp) match probe, a 30 bp fragment of HMG corresponding to nucleotides 236 to 265 was sandwiched by 235 bp of each partial sequence of the P35S.

Preparation of pseudoinsertions

An approximately 3.5 kb DNA fragment, which was derived from the rice PLD genomic sequence corresponding to nucleotides 23,929,495 to 23,932,989 in Oryza sativa Japonica, chromosome 6, cultivar: Nipponbare, complete sequence, was cloned (GenBank accession number, AP014962). The cloned DNA fragment was used as a negative control, PLD (NC). For the preparation of pseudoinsertion1 (PI)1, 30 bp of the HPT corresponding to nucleotides 296 to 325 and a 50 bp sequence of Agrobacterium nopaline synthase terminator (TNOS) derived from pBI221 (GenBank accession number, AF502128) corresponding from nucleotide 2,834 to 2,883 were cloned into the NarI and HindIII sites in the PLD (NC), respectively. For the preparation of PI2, a 20 bp sequence of the right border derived from pBI121 (GenBank accession number, AF485783) corresponding to nucleotides 8,626 to 8,645, and a 40 bp sequence of the P35S corresponding to nucleotides 5,727 to 5,766 were cloned into the NarI and HindIII sites in the PLD (NC), respectively. The three DNA fragments, PLD (NC), PI1, and PI2, were isolated and purified, and then 1.1 × 10⁷ copies of these fragments were respectively added to 5 μg of rice genomic DNA.

Results

Selection of a suitable hybridization system for targeting short DNA fragments

For the Southern blot analyses, we used non-radioactive probes because of their versatility and ease of use. There are many types of labeling systems for probe preparation [19]. We first evaluated several probe-labeling systems; such as DNA labeling by enzyme (AP) -mediated labeling, PCR, and transcriptional RNA labeling. To evaluate the sensitivity of Southern hybridization systems, we performed a model study using the probes prepared by three labeling systems targeting a single-copy endogenous gene, the complementary sequences in these probes were serially replaced with a dummy sequence (Figure S1). A rice endogenous gene, PLD which has been utilized and characterized as an endogenous reference gene for GM rice detection [20,21], was used for the evaluation. When using the PLD probe whose sequence was completely identical to the target, i.e. a 100% match probe, PLD was detected with all of the labeling systems we used (Figure S2).

Next, the AlkPhos-labeled PLD probe sequence was serially replaced by the HPT gene, which was used as a dummy sequence. HPT is a typical selection marker gene for GM crops, and the sequence derived from HPT is absent in conventional plant genomes. We first used a 1,000 bp of DNA fragments as a probe. When probes containing shorter sequence with homology to the target were used, a weaker signal was detected (Figure 1(a-c)). When 90% of the probe sequence was replaced by the dummy sequence, containing a sequence with 10% (100 bp) homology to PLD, the signals were detected with the AlkPhos direct system (Figure 1(c)), but unable to be detected by either PCR-DNA or RNA probe (data not shown). From these results, the probes were prepared using AlkPhos direct system in the following experiments.

Modification of the Southern blot analysis protocol for highly sensitive detection

With the original Southern blot analysis protocol, homology sequences of up to 100 bp were detected. To increase the detection sensitivity, we modified the protocol. In conventional Southern blot analysis, after the electrophoresis, the agarose gel containing restriction enzyme-digested DNA fragments is treated with an acid, such as dilute hydrochloric acid, and this step is called the depurination step [16,22]. Generally, the rate of transfer of the DNA to a membrane depends on the size of the DNA fragment, with the transfer of a large DNA fragment (> 15 kb) being less efficient [16]. The depurination step breaks the DNA fragments into smaller pieces, thereby allowing more efficient transfer from the gel to the membrane. For the detection of short rDNA, we assumed that it is desirable to avoid extra fragmentation of the restriction enzyme-digested genomic DNA. We then omitted the depurination step, and the signals detected were clearer than those obtained with the conventional protocol (Figure 1(d)).

For subsequent analyses, the depurination step was omitted.

We further evaluated serially replaced probes, and homology sequences as short as 30 bp were found to be detectable with a 500-bp probe consisting of 30- and 470-bp sequences derived from PLD and HPT, respectively (Figure 2(b)). Further, we evaluated the effect of the locus by moving the homology sequence of PLD from the terminal to the center of the probe, and the predicted bands were also detected (Figure 2(c)). As shown in Figure 2(b), in the case of the PLD T30 probe, six major bands numbered 1 to 6 were detected in 4 lanes together, while two nonspecific bands, 1 and 6 also appeared. To eliminate these false signals, we tested several stringencies for hybridization and washing conditions. First, sodium chloride concentration in the hybridization buffer was varied from 0.25 to 1.0 M but no obvious differences were detected compared to the original protocol in which 0.5 M of sodium chloride was used. Next, stringencies were varied by adjusting the temperature of the primary washes. We tested higher temperatures up to 75°C. However, it was difficult to find a condition to detect true but not false signals. For our original purpose, unknown unintentionally introduced rDNA(s) were the target(s) for detection, and this result suggested the difficulty of distinguishing between specific bands and nonspecific bands in the actual evaluation of NPBT crops. To circumvent these difficulties, we used a 100% dummy sequence as a probe for a control experiment, and the two nonspecific bands appeared (Figure 2(d)) and confirmed that bands 1 and 6 were nonspecific signals. The result suggested that, in the actual evaluation, a negative control lane using genomic DNA derived from control non-GM plants is required to determine specific bands. We also calculated the SNR of the six bands. The SNR is the ratio of the strength of a signal to its background noise in the absence of signals on the membrane. The obtained SNRs are listed in Table 1 for the PLD T50, T30, and C30 probes, and the resulting SNRs of all of the specific bands, namely 2, 3, 4, and 5, were over 2.0. It would be helpful to expect specific bands using SNR values because nonspecific bands would tend to be weak.

Table 1. Signal-to-noise ratio of the bands in Figure 2

S/N	1	2	3	4	5	6
PLD T50	1.6122	2.0687	2.2041	2.6571	2.0566	1.3356
PLD T30	2.2195	2.0583	2.1907	2.0583	2.1541	1.5945
PLD C30	2.5289	2.0923	2.5846	2.3454	2.1051	1.828
Underlined numbers indicate positive bands.

Figure 2. Sensitivity evaluation of Southern blot analysis using DNA probes 500 bp in length. PLD T50 containing 50 bp of PLD and 450 bp of HPT, PLD T30 containing 30 bp of PLD and 470 bp of HPT, and PLD C30 containing 30 bp of PLD sandwiched between 235 bp of HPT sequences were used as probes in (a), (b), and (c), respectively. As a negative control, 500 bp of HPT was used in (d). The predicted molecular sizes are as follows; SacI 1,970 bp, KpnI 7,629 bp, EcoRI 4,286 bp, SphI 3,502 bp. Black arrows indicate nonspecific bands

To confirm the versatility of this method, we evaluated the sensitivity of another rice endogenous target sequence derived from the SPS gene which is also well studied as an endogenous reference gene for GM rice detection [21,23]. For SPS detection, P35S was used instead of HPT as a dummy sequence. As shown in Figure 3(a), SPS was detected using the SPS T30 probe containing a 30 bp of homology sequences of SPS. The whole genome sequence of the Nipponbare cultivar was already available [24,25], and the sizes of the positive bands in the Southern blot analysis were predictable. The predicted sizes of the positive bands were 7,835, 22,788, 6,117, and 1,622 bp when rice genomic DNA was digested with restriction enzymes SacI, KpnI, EcoRI, and SphI, respectively. All the expected bands were detected except for the band for KpnI digestion, indicating that it is difficult to detect larger bands, probably longer than 15 kb, using Southern blot analysis. We recommend that more than one restriction enzyme should be used for the detection of unintended short rDNA insertion with the Southern method. To decrease the theoretically expected molecular sizes of positive bands, it was effective to use a double digest reaction (double digestion), i.e., digesting DNA with two restriction enzymes simultaneously. If restriction enzymes recognizing six base pairs such as SacI, KpnI, EcoRI, and SphI were used, a restriction sequence would be found once every 4⁶ (= 4,096) base pairs, on average, and the average length could be shortened by half with double digestions. We then used double digestions such as SacI + KpnI or EcoRI + SphI in the following analyses.

Figure 3. Sensitivity evaluation of the Southern blot analysis targeting (a) rice sucrose phosphate synthase (SPS), (b) tomato LAT52, (c) maize HMG. For SPS detection, a probe containing complementary sequences of 30 bp of SPS and 470 bp of P35S was used. The predicted molecular sizes are as follows; SacI 7,835 bp, KpnI 22,788 bp, SacI + KpnI 7,835 bp, EcoRI 6,117 bp, SphI 1,622 bp, EcoRI + SphI 1,622 bp. For LAT52 detection, probes containing complementary sequences of 50 or 30 bp of LAT52, and 450 or 470 bp of HPT were used. For HMG detection, a 100% match probe containing complementary sequences of 500 bp of HMG, and a 6% (30 bp) match probe containing 30 bp of HMG and 470 bp of P35S were used

In this study, we used Amersham Imager 600 as a charge coupled device (CCD)-based chemiluminescence imager. To evaluate the reproducibility of the Southern blot analysis, FUSION-SOLO6S.EDGE was used as an alternative CCD-based chemiluminescence imager (Vilber. Lourmat Deutschland GmbH, Eberhardzell, Germany). Under the same conditions of membrane preparation, the 30-bp homology sequence derived from PLD was detected by the Fusion system (Figure S3).

Applicability of the Southern blot analysis to other crops

We further evaluated the applicability of our modified Southern blot analysis to other crops. Tomato is one of the most promising candidates for the commercialization of genome-edited crops [10]. The LAT52 gene has been utilized as a control reference sequence for GM tomato detection as a single copy endogenous gene [26]. We evaluated the detection sensitivity to a sequence derived from LAT52 in a manner similar to that used for rice. When we applied a probe containing 30 bp or 50 bp of LAT52 complementary sequences to 5 μg of tomato genomic DNA as in the rice analyses, no signals appeared (data not shown). The tomato genome size is approximately 950 Mbp, about 2.2 times larger than the rice genome [27]. We then increased the amount of tomato genomic DNA to 12 μg for the Southern blot analyses, and the predicted bands were detected (Figure 3(b)). To further expand the applicability of the Southern method, we evaluated the detection of a maize endogenous gene, HMG, which has been used as a reference sequence for GM maize detection [28,29]. As shown in Figure 3(c), a 30-bp homology sequence derived from HMG was detected using 30 μg of maize genomic DNA. Since the reported size of the maize genome is approximately 2.3 Gbp [30], 30 μg of maize genomic DNA contained almost the same copy number as 5 μg of rice genomic DNA. These results suggested that 30-bp homology sequences could be detected in crops other than rice if sufficient amount of genomic DNA is provided with respect to the genome size of each crop.

Sensitivity evaluation of the Southern blot analysis using common GM sequences

So far, we have evaluated the sensitivity of the Southern method using plant endogenous sequences. However, to achieve our original purpose, the detection of recombinant sequences such as GM common sequences should be evaluated. DNA fragments containing short recombinant sequences from HPT, P35S, TNOS and the right border [31] were used as pseudoinsertions (Figure 4). Approximately 3.5 kb of DNA fragments derived from PLD containing pseudoinsertions ranging from 20 to 50 bp of the recombinant sequences were synthesized. Pseudoinsertion1 (PI1) contained a 50-bp sequence derived from TNOS and 30 bp derived from HPT, and pseudoinsertion2 (PI2) contained a 40-bp sequence derived from P35S and 20 bp derived from the right border. The PLD 3.5-kb fragment which did not contain GM sequences was used as a negative control, PLD (NC) (Figure 4(b)). We separately added equal copy numbers of PLD (NC), PI1, and PI2 DNA fragments to 5 μg of rice genomic DNA and attempted detection using specific probes for sensitivity testing.

Figure 4. Sensitivity evaluation of the Southern blot analysis targeting common GM sequences. The diagram shows (a) the simulated rDNA construction, and (b) pseudoinsertions. Pseudoinsertion1 (PI1) contains 50 bp derived from TNOS and 30 bp derived from HPT, and pseudoinsertion2 (PI2) contains 40 bp derived from P35S and 20 bp derived from the right border (RB). PLD (NC) is a negative control which does not contain any GM sequences. The positions of the probes used for the Southern blot analysis are indicated by the numbers 1 ~ 8 in (a). Red arrows indicate positive bands. A black arrow indicates nonspecific bands (c)

In actual evaluations of the presence of unintended short rDNA insertion(s), it will often be necessary to use many probes to cover the entire rDNA sequence used for the development of new cultivars by NPBTs. To simulate a target rDNA containing the left border, P35S, HPT, cauliflower mosaic virus 35S terminator, TNOS, and the right border, eight probes were required to cover the whole rDNA (Figure 4(a)).

To speed-up and simplify the process, we mixed two probes targeting different sequences. As shown in Figure 4(c), pseudoinsertions derived from TNOS, P35S, and HPT but not the right border were detected. These results suggested that recombinant sequences equal to or longer than 30 bp would be detectable. When probes 1 and 2 were used, PI1 and PI2 did not contain any of the complementary sequences found in these two probes, and a band detected with probes 1 and 2 was easily identified as a nonspecific signal because the band was detected with PLD (NC) (Figure 4(c)).

When we analyzed PI1 and PI2 with six individual probes, the obtained bands were completely consistent with expectations, e.g., the P35S, HPT, and TNOS sequence used as pseudoinsertions were detected with probe 4, probe 5, and probe 7, respectively (Figure S4).

Discussion

Southern blot analysis has been widely used as one of the basic techniques in molecular biology. Although Southern blot analysis has been utilized worldwide for many applications, the technical limit of detection for short DNA fragments has not been evaluated. In this study, we succeeded in enhancing the detection sensitivity of Southern blot analysis with several modifications and detected 30 bp fragments derived from rice, tomato and maize endogenous single copy genes. Using the modified Southern blot analysis, it would be able to detect 30 bp sequences with GC content variation at least between 33.3 to 60.0% (Figures 2(b) and 3(a)). Next, to evaluate the detectable limit of DNA length, we used pseudoinsertions which contained short homological sequences of typical recombinant GM sequences. These evaluations demonstrated that 30-bp homology sequences were long enough to be detected by Southern blot analysis.

The regulatory classifications and guidelines for modern breeding techniques such as NPBTs are still under debate and entail uncertainty [32–35]. To establish a new guideline, process- and product-based approaches could be used [36]. In the latter approach, it is important to demonstrate the absence of foreign sequences in plant genomes, although global criteria regarding, for example, which technology should be used for the evaluation and how many bases derived from recombinant nucleotide(s) should be detected have not been determined. The minimal length for a DNA fragment to be unique in a crop such as maize is approximately 17 to 20 bp, considering the genome size [35]. Below the minimal length, it is not possible to attribute the fragment to a specific organism [35]. On the other hand, in theory, an identical small mutation may be induced by both genome-editing and traditional mutagenesis, and thus it is difficult to determine whether or not a particular mutation was caused by genome editing. In addition, negative (null) segregant plants are predicted not to contain any alterations of their genomic DNA and seem to be no longer detectable. Under these circumstances, a new, science-based regulatory framework for NPBT crops is required.

Recently, the progress of high-throughput DNA sequencing technologies, so-called “next generation sequencing (NGS)” has been remarkable. Whole genome sequences of many organisms, including crops, have been revealed by NGS-mediated analyses [37], and NGS offers new technical options in molecular biology. Several reports indicate that NGS is useful for the molecular characterization of GMOs [38,39]. It was reported that Southern-by-sequencing (SBS) using targeted sequence capture coupled with NGS technology [40] could replace Southern blot analysis for the molecular characterization of GMOs.

In the report, to detect sequences derived from transformation plasmid to generate GM plants, the plasmid sequences were used as capture probes. The rDNAs were enriched by hybridizing whole genome libraries from GM plants, and sequences were generated. The sensitivity and reproducibility of SBS was evaluated using pseudoinsertions which contained from 35 to 100 bp of the plasmid backbone sequences and were 250 bp in total length, and it was concluded that 50 bp was reliably detected in the sensitivity testing [40].

NGS has great potential and is expected to be used for the molecular characterization of NPBT crops. Indeed, it has been reported that several genome-edited plants were evaluated with NGS analysis for detection of unintentionally introduced sequences such as Agrobacterium-derived DNA fragments or off-target mutations [41–44].

However, the high-throughput sequence technologies are still undergoing technical evolution [45], the sequencing analyses cannot be easily conducted due to their complexity, and for whole genome sequence analyses, it is occasionally required that an available and adequate reference genome sequence.

Southern blot analysis is simple and has high sensitivity, and its analytical scheme is well established. The developed modified Southern blot analysis would be utilized to screen NPBT crops to detect unintentionally introduced rDNA(s). We summarized a screening strategy in Figure 5. To detect very short insertions, nonspecific bands might appear. Negative control lanes derived from wild type plants should be used to distinguish between specific and nonspecific bands. For the Southern blot analyses, probes covering whole rDNA are required, and several, not one, restriction enzymes should be tested. To pursue rapid screening, the utilization of a mixture of probes and double digestion would be time- and cost-effective.

Figure 5. Schematic diagram showing a screening strategy to detect unintentionally introduced rDNA using the modified Southern blot analysis. Bars 1–22 indicate probe positions covering a recombinant DNA to generate GM plants. Lines 1–4 are segregants derived from NPBT crops. NC is a negative control such as wild type plants. Black and gray bands indicate specific and nonspecific bands, respectively

We conclude that the Southern method is useful as a method of choice for the evaluation of NPBT crops.

Author contributions

RT and KK designed the project. RT performed data analysis and drafted the manuscript. KK participated in preparing the manuscript. MK and MY performed experiments. All authors read and approved the final manuscript.

Acknowledgments

We would like to thank Dr. R. Ohsawa (University of Tsukuba) and Dr. Y. Tabei (NARO) for their helpful suggestions. We would also like to thank Dr. I. Mitsuhara (NARO) for supplying Nipponbare leaves.

Disclosure statement

No potential conflict of interest was reported by the authors.

Supplementary material

Supplemental data for this article can be accessed here.

Additional information

Funding

This work was supported by the Cabinet Office, Government of Japan, Cross-ministerial Strategic Innovation Promotion Program (SIP), “Technologies for creating next-generation agriculture, forestry and fisheries.”