Do transgenesis and marker-assisted backcross breeding produce substantially equivalent plants? - A comparative study of transgenic and backcross rice carrying bacterial blight resistant gene Xa21

The primary objective of breeding in agriculture is to develop plants of desired genotypes or traits, such as high yields and resistance to adverse environmental impact. Marker-assisted backcrossing (MAB) and transgenesis (aka genetic modification or GM) are two widely adopted plant breeding techniques. As a conventional technique, MAB breeding has been used to develop new crop cultivars of, e.g., baley, maize and rice [1–4]. The basis of MAB breeding [5] is to transfer a specific allele at the target locus in a donor line to a recipient line while selecting against donor introgression across the rest of the genome. In most cases, the recipient line used for backcrossing has a large number of favorable attributes but is deficient in a few characteristics. Since MAB breeding to certain degree mimics or replicates natural selection, novel cultivars produced through MAB breeding have been regarded as genetically safe. However, MAB breeding is laborious, requiring several backcrossings and a large number of individual plant screenings, typically on the order of thousands. It typically takes a great deal of luck to produce a product of a desirable trait.

Transgenesis is an effective means for improving crop genetic makeup for deriving favorable traits. Breeding by transgenesis has several advantages over breeding by MAB. Transgenesis is a direct means for introducing a gene or genotype to a genome in order to produce a target trait. As a result, the potential to produce plants with favorable traits increases dramatically. For example, by transfering a gene that encodes a plastidial adenyalte kinase into a potato variety, the transgenic potato displays wild-type growth and developmental phenotype but also has an increased yield and starch concentration [6]. By introducing the hybrid cellulase gene cel-hyb1 into a spring barley variety through Agrobacterium-mediated transformation, the selected marker-free transgenic barley produces a high level of cellulase (1,4-β-glucanase) in developing grains, suggesting that the transgenic barley has the potential for producing a large quantity of cellulase for commercial use [7]. The nutritional value of Golden Rice is improved with increased pro-vitamin A content by introducing genes encoding phytoene synthase (psy) in combination with the Erwinia uredovora carotene desaturase (crtI) into rice [8]. Through an Agrobacterium-mediated genetic transformation system, Xa21, a rice bacterial blight resistance gene, has been introduced into five Chinese rice varieties and as a result, the transgenic rice plants exhibit a high resistance to bacterial blight [9].

While transgenesis offers immense opportunities to curtail the severe threat of food shortage the expanding world population is facing, there are considerable public concerns over the use of transgenesis for crop improvement. Indeed, it remains extremely controversial whether or not transgenic crops have an adverse impact on human health. At the center of this controversy is the issue of whether or not insertion of a transgene into the host plant genome or manipulation of an allele in the host genome may affect the expression of other genes and ultimately lead to unintended phenotypes. Unfortunately, it is technically challenging to address this issue because accurate prediction of phenotypes based on genotypic variation and/or gene expression alteration remains a research topic.

Due to the difficulty, efforts of evaluating the safety of transgenic crops has been geared toward assessing the substantial equivalence between transgenic plants and wildtype or conventionally bred plants, like plants from MAB breeding [10]. Substantial equivalence has been introduced as a standard by the Organization for Economic Cooperation and Development (OECD) and has been endorsed by the Food and Agriculture Organization of the United Nations/World Health Organization (FAO/WHO) [11]. However, the standard is based on comparative analysis and offers only a general principle. No specific molecular, biological, chemical or nutritional basis has been established to precisely specify the degree of substantial equivalence [12, 13]. It thus leaves widely open the study of various aspects of equivalence, ranging from molecular, biological, and chemical to nutritional equivalence, between a transgenic plant and a wildtype or plant produced by MAB breeding. Nevertheless, it has been agreed that to be considered substantially equivalent, the characteristics of a transgenic plant must be within a natural range of variation [14], a guideline we follow in our study.

Rice is an essential staple crop for the world population and a model plant for basic and applied research. Rice bacterial leaf blight (BLB), caused by bacteria Xanthomonas oryzae pv. oryzae (Xoo), is one of the most devastating rice diseases throughout the world. Utilization of BLB resistant genes in breeding is the most effective and economical strategy for controlling BLB [15]. Xa21, the first-cloned BLB resistant gene from Oryza longistaminata, has received much attention because of its broad spectrum of resistance to BLB [16]. The gene has also been widely used in BLB resistance breeding through both the transgenic and MAB strategies [17–21]. Therefore, transgenic rice and MAB rice carrying Xa21 offer an excellent opportunity to assess the possible substantial equivalence of transgenic and MAB rice as well as rice in natural environments.

In order to pave the way for future studies of the safety of transgenic crops, we focused on the molecular and biological aspects of substantial equivalence of transgenic rice. We adopted a systems-biology perspective and examined the transcriptome variation of transgenic rice. Specifically, we incorporated Xa21 into the three-line maintainer line D62B through transgenesis and MAB breeding. Adopting Next Generation (Next-Gen) sequencing, we profiled the transcriptomes of four rice plants: the Xa21 transgenic line (named as DXT), the Xa21 MAB breeding line (named as DXB), the untransformed recipient D62B, and another rice variety MH86 (restorer line). We then analyzed transcriptome variation of the two rice plants carrying Xa21 in reference to that of D62B and transcriptome change between D62B and MH86. This transcriptome analysis was further enhanced by a pathway analysis to understand the pathways that might be disturbed in the two rice plants carrying Xa21.

MAB breeding is a well accepted breeding technology for production of safe crops. Comparing crops produced from transgenesis and MAB breeding thus offers a viable approach to assessing the potential impact that transgenic crops might have on human health. As a general principle, substantial equivalence [11] can be followed to assess the safety of transgenic crops. Using this principle as a guideline and yardstick, various aspects of substantial equivalence, e.g., molecular, biological, chemical and nutritional equivalence, can be examined. Taking a systems-biology perspective, we focused on molecular and biological equivalences between transgenic crops and plants from MAB breeding in the current study. This was done by contrasting transcriptome variations of a transgenic rice and a rice by MAB breeding. As a caveat, any results from such a study should not be interpreted to directly address the safety of transgenic plants. But rather, in addition to gaining insights, at a genome scale, into the biological processes and pathways that might be perturbed by transgenesis, such results can be used to pave the way for further study of chemical and nutritional equivalence.

Adopting the above view point and taking Xa21, the most widely used resistance gene to rice bacterial leaf blight, as the target gene, we studied the phenotype and transcriptome differences between the transgenic and MAB rice carrying Xa21. Included in our expriment system were a transgenic rice (i.e., DXT) and a MAB rice (i.e., DXB) that were developed from the same parental line D62B. DXT was identical to D62B except the key Xa21 gene. The genetic makeup of DXB, through careful selection over six generations of backcrossing to D62B, was almost identical to that of D62B. Therefore, with exception of the specific breeding techniques used, DXT and DXB carried nearly identical genetic materials, making them ideal for the comparison of the two breeding techniques. A critical component of our experiments and analysis was another natural rice variety, MH86, which does not carry Xa21. Here, MH86 played the role of a“control”, when compared with D62B, to reveal the natural range of transcriptome variation between the two natural rice plants.

The comparison of transgenic and MAB rice was done at two levels. At the physiological level, we confirmed through extensive, although laborious, in-field screening of various morphological properties to ensure that DXT and DXB were morphologically equivalent or similar. Due to genetic diversity, the morphological difference between natural varieties are typically greater than that of transgenesis, as we saw in our comparison between MH86 vs. D62B and DXT vs. D62B. At the molecular and cellular level, we carefully studied transcriptome variations as well as function and pathway alterations in DXT and DXB. The scale of transcriptome variations caused by transgenesis was smaller than that by MAB breeding. More importantly, the molecular and cellular functions that may be affected by transgenesis were also affected by MAB breeding, and the majority of pathways perturbed in the transgenic rice were also distorted in the MAB rice. In stark contrast, transcriptome and function variations between MH86 and D62B, which can be regarded as natural variations, were substantially larger than that between DXT and D62B. In short, the variations caused by transgenesis were smaller than that of MAB breeding and were within the natural range of variations.

The results from the current study were in agreement with that in the literature on transgenic wheat and maize. For example, Gregersen et al. compare the gene expression profiles of wildtype and transgenic wheat expressing an aspergillus fumigatus phytase [42]. The results show that the expression profiles of the two plants are not significantly different. A similar study of transgenic wheat, which carries additional High Molecular Weight subunit genes, reaches the same conclusion that transgenesis has less impact on the transcriptome of wheat grain than conventional breeding [10]. Further, Coll et al. uses microarrays to compare the expression profiles of commercial maize variety MON810 and near-isogenic varieties in leaves in vitro and also field cultured plants of AristisBt/Aristis and PR33P67/PR33P66 [43, 44]. The target gene is Bt, a gene non-native to maize encoding insecticidal crystal protein of the soil bacterium Bacillus thuringiensis (B.t.). The results show that gene expression profiles of MON810 and comparable non-GM maize varieties are more similar to that of conventional lines and natural variation.

On the other hand, the experiment designs of previous studies are far from perfect. In [10], comparative analysis of the transcriptomes of three wheat lines, including a parental line, a conventional breeding line and a transgenic line, reaches the conclusion that transgenesis has less impact on the transcriptome of wheat grain than conventional breeding. However, several additional genes or sequences, including the marker gene (Bar gene), the reporter gene (uidA gene), and sequences derived from the bacterial plasmid, are present in the transgenic line. The transgenic line, referred to as a “clean fragment” line, still contains the bar gene. In fact, selective markers, such as bar, are the main concern of biosafety of transgenic crops which induce specific pleiotropic effects [45]. In the studies reported in [42–44], foreign transgenes are used. The current study overcame these issues. First, the target gene Xa21 was cloned from Oryza longistaminata, which is native to rice. Second, other than Xa21 sequence, DXT did not carry any additional sequences such as that for marker genes and reporter genes. Third, unlike the previous studies where a target gene is under the control of a foreign promoter [42, 46], in the current study Xa21 in DXT carried its own promotor. As a result of these factors, the results from our analysis were specific to the functions of Xa21 rather than some artifacts of genes foreign to rice.

More importantly, our analysis expanded beyond transcriptome profiling to include analyses of gene functions and signaling pathways that might be altered by the introduction of Xa21. Our results showed that most of the molecular and cellular functions affected by transgenesis were influenced by MAB and functional categories that affected by MAB were more than those that affected by transgenesis. Analyzing the pathways perturbed by transgenesis and MAB showed that majority pathways altered in transgenic rice were also distorted in MAB rice. The bigger difference between the transcriptome variations of DXB and D62B could be attributed to two factors. Firstly, the transgenic rice analyzed in our study was carefully selected to have favorable properties for the purpose of the study. In particular, DXT was a single-copy, marker-free, and homozygous Xa21 transgenic line and also had consistent agronomic traits similar to that of the parental line D62B. The genetic makeup of DXT was identical to D62B with the exception of the Xa21 gene. Secondly, although six backcrosses have been used to eliminate linkage drag, a few unwanted donor chromosomal segments could still have been retained in DXB when Xa21 was introgressed into D62B. Therefore, the transcriptome of DXT was closer to that of D62B than DXB to D62B.

Thanks to Next-Gen sequencing, the gene profiling experiments of our study were genome wide and provided a high resolution [47]. In contrast, the previous studies focus on a limited number of genes using microarray profiling which is less accurate and restricted to annotated genes [10, 42–44, 46, 48]. Such a large scale functional analysis performed in the current study has never been attempted before and the functional analyses provided deep insight into the functional equivalence between the transgenesis and MAB breeding.

MAB breeding and transgenesis are two most popular breeding techniques for producing plants of favorable traits. As a newer biotechnology, transgenesis has made visible contributions to increase yield of staple crops. While plants produced by MAB breeding have already been widely accepted, transgenic plants are facing the challenge regarding safety on human health. In the current study, these two distinct breeding techniques was closely compared. The analysis focused on transcriptome variations in rice plants generated from these two breeding techniques and on that of two natural rice plants as a baseline of the comparison. The study combined careful assessment of agronomic traits, transcriptome profiling by Next-Gen sequencing, and functional and pathway analyses. Two important conclusions can be drawn from the results. First, transcriptome variation caused by transgenesis is significantly smaller than that by MAB breeding and is within the range of natural variation. Second, the functional categories of differentially expressed genes due to these two breeding techniques and the pathways perturbed by these techniques are not substantially different. These results suggest the transgenic rice and rice from MAB breeding that were compared in the current study are substantially equivalent at the molecular and biological levels. The data and results can be used to study chemical and nutritional equivalence of rice generated by transgenesis and MAB breeding.