Abstract:Gressel et al. (2014; in this issue of New Phytologist, pp. 360–362) disapprove of media coverage of our recent peer-reviewed paper (Wang et al., 2014; this issue of New Phytologist, pp. 679–683), and they question whether the paper can be seen as ‘deserving publication’. In this study, our main hypothesis was that over-production of a key metabolic enzyme (EPSPS) may have the unanticipated effect of stimulating plant growth and fecundity in crop–weed hybrids of rice (Oryza sativa). This enzyme is the target of glyphosate-based herbicides, which are widely used on transgenic glyphosate-tolerant crops. We found that transgenic overexpression of an endogenous epsps gene from rice – which was developed to confer glyphosate resistance – was associated with increases in EPSPS protein levels, tryptophan concentrations, photosynthetic rates, seed germination, plant growth and fecundity of crop-weed rice progeny, relative to their nontransgenic counterparts. This transgenic event was crossed into four weedy rice accessions and tested under field conditions in China. Here, we address several of our critics’ questions and concerns, while noting that others fall outside the scope of our research and therefore are not relevant to the publication. Given the inherent constraints of conducting ecological field studies with strictly regulated, experimental transgenes, we argue that our paper is similar in scope and depth to many other publications in this field. However, in an effort to provide a concise summary of the work, we left out some important details that are included below. We appreciate the opportunity to address these gaps in the following section. Gressel et al. argued that we did not provide sufficient details about the transgenic construct in our parental EP3 transgenic rice line, making it difficult to evaluate or reproduce our study, nor did we explain how its glyphosate resistance was documented. This was an oversight on our part. Su et al. (2008), which is published in Chinese and which describes several experimental transgenic events, showed that the EP3 transgenic event has one copy of the epsps transgene driven by an ubiquitin promoter from maize and is resistant to glyphosate, as intended. Fig. 1 shows the major components of this construct. Briefly, Su et al. obtained the EP3 line using an endogenous epsps gene from Minghui-63 rice that was described in Xu et al. (2002). Without introns, the endogenous epsps gene is 1585 bp (NCBI, http://www.ncbi.nlm.nih.gov/nuccore/15778435), while the full length of the epsps sequence with introns is 3661 bp (NCBI, http://www.ncbi.nlm.nih.gov/nuccore/15724391). The transgenic EP3 used the epsps sequence without introns (indicated in Su et al., 2008). Others could repeat our experiment by creating rice lines similar to EP3, using sequence data in NCBI and crossing these rice lines with accessions of weedy rice. In Wang et al., we used specific PCR primers for discriminating between sequences from the endogenous epsps and the smaller inserted transgenic epsps, with fragments that were c. 1000 and 700 bp, respectively. Su et al. (2008) produced 84 independent clones and used PCR amplification to confirm the presence of the target epsps transgene in the T1 EP3 generation. T1 and T2 progeny obtained from selfing were grown under field conditions and sprayed with glyphosate (Roundup) at commercially recommended dosages to confirm glyphosate resistance (note: the English abstract was translated incorrectly, using the word glufosinate in place of glyphosate in the third sentence). This dosage of Roundup was described as 3 l ha−1. The commercial Roundup formulation was 41% glyphosate isopropylamine salt (molecular formula: C3H9N·C3H8NO5P; molecular weight: 228.2), which is equal to 30% glyphosate (molecular formula: C3H8NO5P; molecular weight: 169.1). For the dilution of Roundup before spraying, it is recommended to use 3–6 l ha−1 of Roundup solution diluted with water (450 l ha−1) in rice fields, which is equivalent to 11.83–23.66 mM of glyphosate for the concentration. T3 plants were used for Southern blot analysis and further glyphosate screening (Su et al., 2008). When the T3 EP3 progeny were tested at the three-leaf stage and compared to nontransgenic, Minghui-86 controls, 100% of the EP3 plants survived from treatments with 15 mM glyphosate (Sigma), which killed all of the control plants (Table 1). The T3 EP3 plants were not tested for glyphosate resistance under field conditions by Su et al. (2008), as was the case for the T1 and T2 generations, but these plants were resistant to concentrations that were lethal to the nontransgenic controls (Table 1; Fig. 2) and were selfed to obtain the T4 and T5 generations. This implies that it is accurate to describe the T5 EP3 plants as being resistant to glyphosate in Wang et al. Before accepting Gressel et al.'s critique, we encourage readers to study our publication and reflect on the content, nuances and caveats that cannot be captured in media reports or letters to the Editor. Also, it is important to distinguish between our interpretation of the data and the interpretation of our data by others, including the opinions mentioned in Nature (Qiu, 2013) and those of Gressel et al. Below, we discuss four major issues related to Gressel et al.'s interpretation of our original paper. As with any study of a single transgenic event, we noted that genetically engineered (GE) plants and their nonGE counterparts differed in the presence or absence of the inserted construct, ‘as well as the selectable marker gene and any crop-specific genes that may be linked to the transgene insertion site’. Position effects and other artifacts associated with tissue culture can alter a plant's phenotype (e.g. Matzke & Matzke, 1998; Bhat & Srinivasan, 2002; Zeller et al., 2010). Therefore, it is well known that studying a single transgenic event does not allow one to generalize about all events with a particular transgene, but this caveat is rarely explained when results are simplified in the news media. Unfortunately, options for obtaining more than one transgenic event for research purposes often are not available (e.g. Dalton, 2002; Burke & Rieseberg, 2003; Sasu et al., 2009). In our case, the cost of developing and breeding several transgenic events into multiple weedy rice accessions led us to focus on a single transgenic event. Contrary to the implications of Gressel et al., our methods and interpretations regarding fitness-related effects of the epsps transgene are similar in approach to many previous ecological publications on this topic (e.g. Stewart et al., 1997; Snow et al., 1999, 2003; Burke & Rieseberg, 2003; Vacher et al., 2004; Chen et al., 2006; 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). Gressel et al. argue that our fitness study was ‘flawed’ and ‘highly artificial’. We found that F2 GE crop–weed hybrids produced 48–125% more seeds per plant than their nonGE counterparts in common garden field experiments. This suggests that the GE trait could persist and increase in frequency over time, with the caveat that ‘we do not know how other stages of the life cycle, such as seed longevity or seedling establishment, are affected by the epsps transgene’. We also noted that ‘further life-cycle ecological research is needed to examine whether weedy rice biotypes that overproduce EPSPS could have accelerated population growth rates and enhanced competitive ability relative to existing biotypes in the absence of glyphosate, as we surmise’. Many studies showing fitness-related effects of specific transgenes have used similar experimental approaches with similar caveats (see examples cited above). Common garden field experiments are widely used in other areas of ecology to test for differences in fitness-related traits between groups of interest, such as hybrids vs parents, selfed vs outcrossed progeny, native vs introduced biotypes, or resistant vs susceptible biotypes. Although these conditions may be described as ‘artificial’, common garden experiments represent an important tool for evolutionary ecologists. Gressel et al. incorrectly imply that we intended to test for differences in ‘relative competitive ability’ between GE and nonGE crop–weed hybrids. This would have required a much more elaborate experimental design. Gressel et al. introduce the term ‘competitive fitness’ (not used in our paper) and proceed to discuss self-thinning and the design of competition experiments at length. However, our study was not intended to examine competitive ability per se. We explicitly stated that ‘the mixed planting treatment was included to increase the likelihood of detecting small fitness differences that might be evident only with direct competition between GE plants and their non-GE counterparts’. In fact, these differences were more pronounced in the mixed planting treatment, as expected. Contrary to what is implied by Gressel et al., our paper does not mention biosafety, nor do we recommend how the epsps trait or other transgenic traits should be regulated. A much broader topic that Gressel et al. bring up is whether fitness-related transgenic traits are sufficiently different from other crop traits to require any regulation at all. We did not discuss these topics in our paper. Instead, we noted that fitness-enhancing transgenes could allow novel traits to persist in wild or weedy populations that hybridize with the crop. This type of information about gene flow and its consequences is a standard requirement of many regulatory agencies dealing with GE crops (but see Ledford, 2013). Echoing a common theme of research articles about gene flow from GE crops, Stewart et al. (1997) argued that when a transgene is associated with increased fecundity and suitable habitat is available, ‘there is a likelihood of enhanced ecological risk associated with the release of certain transgene/crop combinations’. Regarding best practices for scientists who publish research on politically or ethically controversial topics, including genetically engineered crops, we agree that it is important to be as patient, accurate and strategically sensitive as possible. Peer-reviewed journals allow one to do this with careful editing and feedback from reviewers. By contrast, authors have very limited control, if any, over how such findings will be interpreted by science writers or the popular press. In the face of nearly instantaneous internet connectivity, there is no way for scientists to revise or correct inappropriate statements in the media, other than to follow up with further information and clarification. Does this mean that new and interesting findings should be left unpublished or, if published, that the authors should not participate in interviews? We think not. As scientists, we feel that we have an obligation to discuss our findings with both scientific and public audiences in a timely and responsible manner. We endeavored to write our paper as clearly as possible, without undue speculation, and fully expecting that we and others will pursue studies to confirm or reject our hypothesis in the future. A press release from Ohio State University included an interview with Allison Snow, who explained that: ‘We don't know yet if our findings are going to be generalizable…’ and ‘ecological studies such as ours can help inform risk assessment and biosafety oversight’. Gressel et al. give the impression that we misled journalists as to the rigor and implications of our study, but this is untrue. Our paper met standard criteria for peer-reviewed publications in evolutionary ecology, and it was subjected to peer review. The authors declare no financial interest in the use of herbicides nor in genetically engineered (GE) herbicide-resistant plants. We thank Professor Jun Su for clarifications regarding the EP3 transgenic line and the colour photo for Fig. 2.

Abstract: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. The authors declare no financial interest in the use of herbicides nor in GE herbicide-resistant plants. Fudan University and Ohio State University are educational organizations that conduct research including GE plants for generating scientific knowledge. The authors thank several anonymous reviewers for helpful comments and discussion.

Scientific data published by a peer-reviewed journal should be properly interpreted: a reply to the letter by Gressel et al. (2014).

Using a single transgenic event to infer fitness effects in crop-weed hybrids: a reply to the Letter by Grunewald & Bury (2014).

Genetically engineered rice endogenous 5-enolpyruvoylshikimate-3-phosphate synthase (epsps) transgene alters phenology and fitness of crop-wild hybrid offspring

A Novel 5-Enolpyruvoylshikimate-3-phosphate (epsp) Synthase Transgene for Glyphosate Resistance Stimulates Growth and Fecundity in Weedy Rice (oryza Sativa) Without Herbicide

Overexpressing Exogenous 5-Enolpyruvylshikimate-3-Phosphate Synthase (EPSPS) Genes Increases Fecundity and Auxin Content of Transgenic Arabidopsis Plants

Development of Transgenic Glyphosate-Resistant Rice with G6 Gene Encoding 5-Enolpyruvylshikimate-3-phosphate Synthase

Effects of over-expressing a native gene encoding 5-enolpyruvylshikimate-3-phosphate synthase (EPSPS) on glyphosate resistance in Arabidopsis thaliana

Increased Longevity and Dormancy of Soil-Buried Seeds from Advanced Crop–Wild Rice Hybrids Overexpressing the EPSPS Transgene

Genetic Analysis of Transgenic Glyphosate-resistance Cotton with EPSPS Gene

Isolation of a glyphosate-resistant gene and functional analysis in Escherichia coli

Effect of Glyphosate on the Gamete Fertility of Transgenic Glyphosate-resistant Cotton Germplasms with EPSPS-G6

Contrasting plant ecological benefits endowed by naturally occurring EPSPS resistance mutations under glyphosate selection

The Naturally Evolved EPSPS from Goosegrass Confers High Glyphosate Resistance to Rice.

Physiological and molecular mechanisms of glyphosate tolerance in an in vitro selected cotton mutant

Impact of a G2-EPSPS & GAT Dual Transgenic Glyphosate-Resistant Soybean Line on the Soil Microbial Community under Field Conditions Affected by Glyphosate Application

Overexpression of improved EPSPS gene results in field level glyphosate tolerance and higher grain yield in rice

Mutations and amplification of EPSPS gene confer resistance to glyphosate in goosegrass (Eleusine indica)

The Effects of Commercial Glyphosate Herbicide on the Seedlings of Transgenic Cotton with Glyphosate Resistance

Impact of Glyphosate on the Rhizosphere Microbial Communities of an EPSPS-Transgenic Soybean Line ZUTS31 by Metagenome Sequencing.

Non-Random Transmission Of Parental Alleles Into Crop-Wild And Crop-Weed Hybrid Lineages Separated By A Transgene And Neutral Identifiers In Rice

Glyphosate resistance and no fitness cost in backcross offspring of wild soybean and transgenic soybean with epsps gene