Grunewald & Bury (2014; in this issue of New Phytologist, pp. 367–369) criticize our recent peer-reviewed paper (Wang et al., 2014; in this issue of New Phytologist, pp. 679–683), stating that we ‘unnecessarily harm the sensitive debate on GM crops.’ We will not focus on this politically charged topic here, but we do want to address scientific questions about our study of transgenic crop–weed (Oryza sativa and O. sativa f. spontanea) hybrids of rice. Grunewald and Bury propose that an insertion effect that stimulated tiller formation in the EP3 crop parent of these hybrids offers a more convincing explanation of our results than direct effects of a transgene for over-expressing 5-enolpyruvoylshikimate-3-phosphate synthase (epsps) (our central hypothesis). In our study, the genetically engineered (GE) segregants had significantly greater expression of epsps and produced significantly greater amounts of the enzyme of EPSPS than the non-GE segregants, as expected (Wang et al., 2014). As discussed later, we doubt that an insertion effect could account for the clear and significant increases in our transgenic lines in terms of enhanced fecundity, greater tryptophan concentrations in leaves, and other traits, which are more parsimoniously explained by transgenic over-expression of epsps and its key role in the shikimic acid pathway (Wang et al., 2014). Tryptophan is produced by this pathway and is a precursor of growth hormones (auxin) and secondary metabolites that play a role in plant defense (e.g. Maeda & Dudareva, 2012). Tryptophan is just one of many products of the shikimic acid pathway, which can account for as much as 35% of a plant's biomass (e.g. Franz et al., 1997).
To review, our two crop parental rice lines (Fig. 1) were the inbred line Minghui-86 and a transgenic rice line (EP3), which was obtained by transforming Minghui-86 (Su et al., 2008; Lu et al., 2014 (this issue of New Phytologist, pp. 363–366); Wang et al., 2014). Thus, these lines differed only in the absence or presence of a single-copy insertion of the transgenic construct and possible unknown side-effects of transformation. Su et al. (2008) and our Supporting Information Table S1 (Wang et al., 2014) showed that EP3 produced significantly more tillers and panicles per plant than Minghui-86. This suggests a direct effect of the transgene on plant growth and reproduction in EP3. We do not understand why Grunewald and Bury do not even acknowledge this explanation under ‘option (2)’ of their letter. Instead, they assume that the superior performance of EP3 was due to a linked sequence that ‘putatively stimulates tiller formation’ and was ‘the result of the insertion’ rather than expression of the epsps transgene (see Grunewald & Bury, 2014, Fig. 1). All of their arguments against our hypothesis hinge on this assumption. Although we did not consider the point (similar to their assumption) to be the major cause of the enhanced fecundity, we included this caveat in our paper: ‘… we assume that the over-production of EPSPS and the downstream differences that we observed between GE plants and their non-GE counterparts were attributable to the over-expression of the modified transgene, epsps, rather than other tightly linked genes from the cultivated parent, although this possibility cannot be ruled out entirely.’ (Wang et al., 2014).
In a field experiment, we also found that EP3-derived F1 crop–weed progeny had significantly more tillers, panicles, and seeds than those derived from Minghui-86. With the exception of the transgene, which was deliberately engineered to over-express epsps, overall differences in the crop-specific alleles found in these two types of F1 progeny should be negligible, barring the type of major insertion effect postulated by Grunewald and Bury. Remarkably, the GE F1 plants produced up to 50% more seeds per plant than non-GE controls in the mixed competition treatment (Wang et al., 2014, Table S3).
Our evidence for an association between over-expressing epsps and strong, heritable increases in fecundity in both the EP3 parent and EP3-derived F1 progeny is consistent with other traits measured in subsequent generations (Wang et al., 2014). In GE F3 plants, we found greater leaf concentrations of tryptophan, an aromatic amino acid produced by the shikimic acid pathway downstream from EPSPS (Herrmann, 1995), compared to non-GE F3 plants. These GE plants also exhibited greater photosynthetic rates and greater percent seed germination than their non-GE counterparts. In the F2 generation, GE plants produced 48–125% more seeds than the non-GE controls. Theoretically, any of the fitness-related effects that we documented in the F2 and F3 generations could have been influenced by non-GE crop alleles that were linked on the same chromosome as the transgene insertion site. However, these types of alleles, if present, would not have differed between the GE and non-GE controls in the crop parents or the F1 generation (Fig. 1).
To bolster their argument that our results are simply an artifact of the insertion process and ‘unrelated to the transgene’, Grunewald and Bury claim that our paper contradicts 20 yr of experimental results involving comparisons between glyphosate-tolerant crops and their isogenic counterparts, citing a review paper in AgBioForum by Brookes & Barfoot (2006). Unfortunately, Brookes & Barfoot (2006) did not report empirical studies designed to test for an association between glyphosate resistance and increased yield in the absence of glyphosate, nor is it appropriate to group all types of genetic mechanisms for glyphosate resistance together. In soybean, we note that Owen et al. (2010) reported greater yields of glyphosate resistant (CP4) cultivars in comparison to non-GE cultivars in the absence of glyphosate, but they attributed this benefit to the improved genetic background used to develop GE cultivars. For our purposes, a better transgenic event and experimental design is that of Zhou et al. (2003), who studied a transgenic event in wheat with two aroA:CP4 expression cassettes, one driven by the CaMV enhanced 35S promoter and another by the rice actin1 promoter, in comparison to the non-GE parent cultivar. Zhou et al. (2003) found that the grain yield of GE wheat plants was greater than that of the non-GE counterparts in 2000, and non-significant in 1999 and 2001, based on field experiments without glyphosate applications (see Zhou et al., 2003, Table 5). It is possible that the two CP4 cassettes in GE wheat conferred over-production of EPSPS, although this was not addressed in their study. In the future, if data from sufficiently rigorous experiments of glyphosate-tolerant crops are available to public researchers, those involving transgenic events that confer over-production of EPSPS could be useful for testing our hypothesis.
In agreement with Grunewald and Bury, we note that the best way to test our hypothesis would have been to study crop–weed progeny derived from two or more insertion events for the transgene, but we were not able to carry out these studies due to various limitations (see Lu et al., 2014). For any study of a single transgenic event, a combination of insertion-site changes and tissue culture-induced changes could result in heritable phenotypic effects that are independent of the transgene and may exhibit abnormal phenotypes (e.g. Alonso et al., 2003; Filipecki & Malepszy, 2006; Neelakandan & Wang, 2012). Like many published studies of the fitness effects of particular transgenes (e.g. Stewart et al., 1997; Snow et al., 1999, 2003; Burke & Rieseberg, 2003; Guadagnuolo et al., 2006; Laughlin et al., 2009; Sasu et al., 2009; Londo et al., 2011; Xia et al., 2011; Yang et al., 2011, 2012), our research was based on a single transgenic event, which is far from ideal. Unfortunately, even single transgenic events that are under development for commercial applications are exceedingly difficult for ecologists to obtain (e.g. Dalton, 2002). Thus, although we agree that multiple events are needed to investigate fitness effects of transgenic traits, this is rarely feasible.
To conclude, we view our original publication as a key first step toward testing the novel hypothesis that over-production of EPSPS can stimulate growth and fecundity in crops and crop–weed progeny. To our knowledge, this hypothesis has not been addressed in the peer-reviewed literature or elsewhere. We have initiated further studies to test the generality of our findings by using multiple transgenic events in rice and Arabidopsis thaliana. Intriguingly, Klee et al. (1987) includes a photograph showing transgenic Arabidopsis seedlings that over-express endogenous epsps and are larger than wild-type seedlings. If our hypothesis is confirmed by further research, this would have broad implications for understanding and engineering a key enzyme (EPSPS) of the shikimic acid pathway. If we are not able to confirm this hypothesis, we plan to publish these findings accordingly, consistent with the scientific process.