Introduction

Transgenesis provides an important experimental strategy for generating knowledge on the functions and potential uses of cloned plant genes for agriculture. Growing pressures to create new crops that are better suited for sustainable production (Park et al. 2011) demands an examination of the regulatory frameworks that govern the uptake of transgenic crops into commercial production (Smyth and McHughen 2008; McHughen and Smyth 2008). A significant barrier to the commercialization of transgenic crops is the diversity of regulatory policies and the limited extent of scientific data supporting them (McHughen 2007). For example, induced mutations have generally been used in plant breeding without the same scrutiny imposed on transgenic crops. Consequently mutation strategies are being encouraged and the use of transgenics discouraged (Waugh et al. 2006; Parry et al. 2009). Yet, recent studies reveal that mutagenesis can create more transcriptional changes in rice than transgenesis (Batista et al. 2008) and that variation in transcriptomes, proteomes or metabolomes of many crops is lower in transgenic crops than among conventionally bred varieties already in production (Kogel et al. 2010; Baudo et al. 2006; Lehesranta et al. 2005). Transgenesis also offers a wider range of options in the modification of a trait as regulatory and/or functional domains within genes can be separately modified with a greater variety of sequences. Avoidance of transgenesis may undermine the full potential of cloned genes to provide the novel traits needed for the production of foods, feeds and agricultural products within the more restrictive environments of the future.

The regulatory system in Canada is unique in that it examines the novelty of traits in new crop varieties and not the process used to generate them (Smyth and McHughen 2008). A crop variety with a novel trait, such as herbicide resistance, would undergo the same regulatory process whether generated by mutagenesis or transgenesis (Smyth and McHughen 2008). In other countries the use of transgenesis would initiate a more rigorous regulatory process than for mutagenesis. The rationale for the Canadian approach is to capture a scientific evaluation of the biosafety of a novel trait without bias towards or against new methodologies. Comprehensive data on the effects of trangenesis on compositional variation and risk is being generated using non-targeted profiling technologies (Kuiper et al. 2003; Davies 2010). This information complements but does not replace the traditional targeted approaches for examining the performance and composition of new crop varieties and provides perspective on the impact of transgenesis on biosafety and risk (Cellini et al. 2004; Parrott et al. 2010).

Here, we compare Arabidopsis plants with the same novel trait, i.e. imidazolinone resistance, produced by different genetic processes, i.e. mutagenesis and transgenesis. The target gene, CSR1, codes for the catalytic subunit of acetohydroxyacid synthase [AHAS, EC 2.2.1.6; a. k. a. acetolactate synthase (ALS)], a holoenzyme complex consisting of tetrameric subunits that catalyzes the first common step of the branched chain amino acid (BCAA) pathway (McCourt et al. 2006). AHAS inhibitors include slow-acting, systemic herbicides that can be used as soil-applied or post-emergent herbicides because they are absorbed by roots and foliage. Currently, five structurally distinct classes of herbicide are known to inhibit AHAS: imidazolinones, sulfonylureas, triazolopyridines, sulfonylaminocarbonyltriazolinones and pyrimidyl (oxy/thio) benzoates (McCourt and Duggleby 2006). The imidazolinone herbicides are used extensively in agriculture in combination with imidazolinone-tolerant Clearfield© crops (Tan et al. 2005), which exhibit resistance due to specific mutations in the gene coding for the AHAS catalytic subunit (Duggleby et al. 2008). AHAS mutations responsible for selective resistance to the imidazolinones have been mapped and include csr1-2 (Ser-653-Asn), csr1-5 (Ala-122-Thr) and csr1-6 (Ala-205-Val) (Duggleby et al. 2008). The point mutations responsible for herbicide resistance function by creating amino acid substitutions that decrease the binding affinity of the herbicide to the substrate-access channel of the enzyme so that the herbicides can no longer block the enzyme’s active site (Duggleby et al. 2008). Catalytic activity is not altered in the herbicide-resistant mutants, as the herbicide-binding domain is distinct from the active site (McCourt et al. 2006).

A comparison of the Arabidopsis substitution mutant csr1-2 and wild-type plants using microarray analysis has revealed the complete absence of pleiotropic effects attributable to the mutation at the transcriptional level revealing that imidazolinones act specifically on AHAS as their sole target (Manabe et al. 2007). We now extend this study to transgenic plants in which the CSR1-2 gene functionally replaces the native CSR1 gene and show once again the absence of significant unanticipated pleiotropic effects under conditions that elevate the level of herbicide resistance and biomass accumulation. The study illustrates the high similarity of a mutant and transgenic plant with the same novel trait.

Materials and methods

Plant materials and selection conditions

Growth conditions and imidazolinone treatments of Arabidopsis thaliana (ecotype: Col-0) plants were conducted as previously described (Manabe et al. 2007). 0.8% (w/v) and 1.2% Agar (Sigma, MO, USA) are used in place of 0.6% and 1.2% PHYTOAGAR as production of PHYTOAGAR had been discontinued. The reverse genetic database of the Salk Institute Genomic Analysis Laboratory (http://signal.salk.edu/cgi-bin/tdnaexpress) was searched for putative loss-of-function mutants of CSR1 (At3g48560). Seeds of csr1-2 D and T-DNA insertional lines, csr1-7 (SAIL_910_E06; CS877955) and csr1-8 (GABI_562B05; N453873), were obtained from The Arabidopsis Biological Resource Center [(ABRC) Columbus, OH, USA]. T-DNA insertional mutants, csr1-7 and csr1-8, were isolated from SAIL (Sessions et al. 2002) and GABI-KAT (Rosso et al. 2003) collections, respectively. Transformation vectors used in generating csr1-7 and csr1-8 are pDAP101, which carry the glufosinate-resistant (BAR) gene, and pAC161, which carries sulfadiazine-resistant (SUL1) gene, respectively. For selection, csr1-7 and csr1-8 were sown on MS-Agar medium supplemented with 25 mg/L analytical standard glufosinate-ammonium (Riedel-de Haën, Seelze, Germany) or with 75 mg/mL sulfadiazine (Sigma), respectively.

Plants were grown to maturity in a Conviron growth cabinet set to a temperature of 21°C and a 16 h day length. One to two week old seedlings were transplanted to 5 inch pots at a density of 5–6 plants per pots. These were fertilized 1 time per week with a solution of 20-20-20 general purpose fertilizer.

Genetic analysis

Genomic DNA was isolated from leaf tissue using the REDExtract-N-Amp Plant PCR kit (Sigma). PCR was also done using the REDExtract-N-Amp Plant PCR kit with primers at a final concentration of 0.3 μM. PCR cycling conditions were one cycle at 95°C for 3 min, 40 cycles of 94°C for 1 min, 53°C for 1 min, and 72°C for 1 min, and then a final extension at 72°C for 10 min. For amplification with the 8409 and SAIL_LP primer pair, an annealing temperature of 59°C was used instead of 53°C. The primers used in this study are: SAIL_RP: 5′-GTGCATCAATGGAGATTCACC-3′; SAIL_LP: 5′-CTTCCCCAAACGTCTTAAAGC-3′; 8409: 5′-ATATTGACCATCATACTCATTGC-3′; pDAP101-LB: 5′-TTCATAACCAATCTCGATACAC-3′. The position of each primer is shown in Online Resource 1.

Transformation

The coding region of the CSR1-2 gene was synthesized (GenScript USA Inc.Piscataway, NJ, USA) and inserted into pCAMBIA1300 (http://www.cambia.org). The vector was transferred to Agrobacterium tumefaciens GV3101 and Agrobacterium-mediated transformation was accomplished using the floral dip method (Clough and Bent 1998). Heterozygous CSR1/CSR1-7 plants were grown to the budding/flowering stage (6–8 weeks old) and inoculated with 500 mL Agrobacterium cultures grown to a cell density of 1.2–1.5 at OD600. After centrifugation the bacteria were collected and resuspended in 300 mL of ½ strength MS media containing 5% sucrose. Silwet L-77 (“Vac-In-Stuff”, Lehle Seeds, P.O. Box 2366, Round Rock Tx. 78680 USA) was added just prior to dipping at a concentration of 0.02% w/v. Mature flowers were removed before dipping and the infiltration treatment was maintained for 30 s. Following the treatment, the pots were placed on their sides, covered loosely with plastic wrap and placed in the dark for 24 h. The plants were then returned to the growth cabinet set to a temperature of 21°C and a 16 h day length until seed set.

Heterozygous CSR1/CSR1-7 plants were essential for genetic transformation because the homozygous CSR1-7/CSR1-7 genotype was lethal. Complementation and screening for segregants using PRC primers (Online Resource 1) allowed the recovery of viable homozygous CSR1-7/CSR1-7 plants carrying the CSR1-2 transgene (Online Resource 1).

Microarray analysis

For microarray analysis seedlings were grown in culture as above in the presence of 100 μg/L imazapyr for 2 weeks. Three separate biological samples consisting of 50–100 mg of fresh tissue consisting of the shoots were collected as described by Manabe et al. (2007) and RNA was extracted using the RNeasy Plant Mini Kit (Qiagen, CA, USA). Probe preparation, hybridization and analysis were carried out as previously described (Abdeen et al. 2010) by Affymetrix for the ATH1 GeneChip (www.affymetrix.com) at the Winnipeg Research Centre labs of Agriculture and Agri-Food Canada.

Results

Loss-of-function mutants reveal that CSR1 is an essential housekeeping gene

To be sure that AHAS is an essential housekeeping gene we conducted a search for putative loss-of-function mutants of CSR1. Two independent and different T-DNA insertional alleles, csr1-7 and csr1-8, were obtained and used for further analysis. The csr1-7 and csr1-8 mutations are at different locations within the only exon of CSR1 (Online Resource 1). The csr1-7 and csr1-8 mutants were sown on MS-agar medium supplemented with 25 mg/L glufosinate-ammonium or with 75 mg/mL sulfadiazine, respectively, for selection of the mutant alleles. The genotypes were assessed by PCR using the primers shown in Online Resource 1. Plants that were susceptible on selection possessed the wild-type CSR1/CSR1 genotype and all resistant plants were heterozygous CSR1/CSR1-7 or CSR1/CSR1-8 genotypes with a wild-type and a mutant allele. There were no homozygous mutants with the CSR1-7/CSR1-7 or CSR1-8/CSR1-8 genotypes found among progeny derived from selfed, heterozygous mutant plants (Table 1) indicating that these were lethal genotypes. This finding is consistent with observations that incomplete silencing of CSR1 transciptional activity by either anti-sense suppression or herbicide treatment results in plants which are almost non-viable (Höfgen et al. 1995).

Table 1 Segregation of the csr 1-7 and csr 1-8 alleles
Full size table

Among the progeny of selfed heterozygous mutant plants the ratios of heterozygous mutant plants to wild-type plants were lower than expected. As shown in Table 1, 22% of the selected plants were CSR1/CSR1-7 and 42% were CSR1/CSR1-8. As both are below the expected 67% (Table 1) there appeared to be reduced fitness among the heterozygous plants; however, plants with the genotypes CSR1/CSR1-7 and CSR1/CSR1-8 were phenotypically indistinguishable from wild-type plants in germination and vegetative growth (data not shown). The frequencies of seed abortion (Fig. 1b, c) were elevated to about 9% for CSR1/CSR1-7 and about 19% for CSR1/CSR1-8 (Table 2). Morphologically, two kinds of seed abortion could be distinguished. The most prevalent type was characterized by the failure of both the embryonic and integumentary tissues to develop. This was termed type II (Fig. 1b, c). A less severe type of seed abortion was also observed and characterized by the development of the integumentary tissues but not the embryonic tissues (Fig. 1b, c). These occurred at a lower frequency and were termed type I (Table 2). Both types could be observed in the same siliques of the heterozygous mutants (Fig. 1b, c). In wild-type siliques (Fig. 1a) seed abortion was occasionally observed but it was always the less severe type I seed abortion. This finding could explain the unexpected segregation ratios that were observed in Table 1.

Fig. 1
figure 1

Seed abortion phenotypes observed in csr1-7 and csr1-8 insertional mutants. Seed abortion phenotypes for type I are indicated by the red arrows and type II by white arrows. a Silique of wild-type Arabidopsis, ecotype Col-0. b Silique of csr1-7. c Silique of csr1-8

Full size image
Table 2 Frequency of aborted seeds in heterozygous mutant lines of csr1-7 and csr1-8 and transgenic lines expressing CSR1-2 in a homozygous CSR1-7 background
Full size table

Complementation of a CSR1 loss-of-function mutant by a CSR1-2 transgene

Transgenic lines carrying the 35S/CSR1-2/NOS gene (Online Resource 1) were generated by transformation of Arabidopsis csr1-7 hemizygous plants (CSR1/CSR1-7). This strategy was necessary as homozygous plants (CSR1-7/CSR1-7) were lethal. Ten lines that carried the 35S/CSR1-2/NOS transgene and that segregated with the homozygous CSR1-7/CSR1-7 background were recovered by screening with PCR primers shown in Online Resource 1. None of these lines carried a wild-type CSR1 allele that could be vulnerable to gene silencing induced by transgenesis. Because the resident CSR1-7 allele is already silenced by the insertion mutation the recovery of viable plants would have to be due to complementation by an active CSR1-2 transgene (Online Resource 1). Any silencing of the CSR1-2 transgene that might occur would be under negative selection pressure in this experiment and would not be recovered.

Substitution of CSR1 with CSR1-2 by transgenesis did not generate obvious differences in germination, growth and development of seedlings (Fig. 2a; data not shown). Interestingly, the seed abortion phenotype was highly variable among the 10 transgenic lines particularly for type I seed abortions (Table 2). Among the transgenic lines type I seed abortions ranged from 2.5 to 68%. The type II seed abortions were dramatically reduced or eliminated in the transgenic lines and ranged from 0 to 4.7%. Individual transgenic lines, such as line T1 (Fig. 2a), were found that were similar to wild-type plants in both vegetative and reproductive growth (i.e. 2.5% type I and 0% type II seed abortion) revealing that complementation of the loss-of-function mutation could be achieved by insertion of the 35S/CSR1-2/NOS gene without the introduction of significant pleiotropic effects on growth and development. This also suggested that the seed abortion phenotype in the heterozygous (CSR1/CSR1-7) mutant line described earlier resulted directly from the csr1-7 mutation (Fig. 1) and not from pleiotropic effects on other genes.

Fig. 2
figure 2

Growth of the csr1-2 mutant and transgenic lines expressing CSR1-2 in a homozygous CSR1-7 background in the presence of imazapyr. a The germination and growth of seedlings after 14 days for the csr1-2 mutant and transgenic lines T1, T2, and T3 in the presence of 0, 102 and 103 μg/L imazapyr. b Increases in fresh weights were measured after 14 days of growth on media with varying concentrations of imazapyr. The fresh weights are expressed as a percentage of the fresh weight on growth without imazapyr. The values for the transgenic lines were calculated as the average ± SE for three independent transgenic lines, T1, T2, and T3, shown in Fig. 1a

Full size image

Enhanced herbicide-resistance conferred by transgenes

No differences in vegetative growth between the csr1-2 substitution mutant and three randomly-selected transgenic lines carrying the 35S/CSR1-2/NOS transgene, including T1, were observed (Fig. 2a); however, differences were discovered when they were grown in the presence of imazapyr. All three transgenic lines were about an order of magnitude more resistant to imazapyr than the csr1-2 substitution mutant (Fig. 2a, b). Furthermore, the enhanced resistance was accompanied by more vigorous vegetative growth in the presence of imazapyr (Fig. 2b). The increases in fresh weight accumulation were associated with an average increase in dry matter accumulation of 14.9 ± 6.2% measured at 14 days of growth.

Pleiotropic effects on the transcriptome

Microarray analysis was performed on 14 day-old seedlings grown in the presence of 100 μg/L imazapyr. Of the approximately 24,000 genes represented on the Affymetrix ATH1 GeneChip microarray, 68, 159, and 52 were found to be up- or down-regulated by greater than twofold (P < 0.05) in lines T1, T2, and T3, respectively (Online Resource 2). Of these, only five transcripts were commonly differentially expressed in all three transgenic lines (Table 3). The most altered corresponded to the CSR1-2 transgene. It was elevated by approximately 7.9-, 7.1-, and 9.7-fold in T1, T2 and T3, respectively, relative to the levels in the csr1-2 mutant. The data indicated that the elevated CSR1-2 transcript expressed from the 35S regulatory sequences was likely responsible for the elevated resistance to imazapyr and the elevated biomass accumulation in the presence of imazapyr. The other genes included a putative peroxidase and an expressed protein, which were upregulated and the iron-responsive transporter IRT1 and a putative 2-oxoglutarate-dependent dioxygenase, which were downregulated.

Table 3 Genes up- or down-regulated greater than twofold (P < 0.05) in all three transgenic lines expressing CSR1-2 in a homozygous CSR1-7 background as compared to the csr1-2 mutant
Full size table

Discussion

AHAS is a critical and essential enzyme in the branched-chain amino acids (BCAA: valine, leucine and isoleucine) biosynthetic pathway. It catalyzes the first common step in the pathway and it is the target for end-product inhibition. The catalytic subunit, encoded by homologues of the Arabidopsis CSR1 gene, carries substitution mutations for resistance to the AHAS-inhibitor group of systemic herbicides in all plants examined (Duggleby et al. 2008). Plant death resulting from inhibition of AHAS by herbicides or by CSR1 silencing of the AHAS-encoding genes appears to occur through amino acid starvation (Höfgen et al. 1995). Herbicides such as the sulfonylureas and imidazolinones appear to bind to specific overlapping sites in the substrate-access channel of the catalytic subunit thus preventing access to the active site. In the csr1-2 substitution mutant (Ser-653-Asn) AHAS cannot bind imidazolinone herbicides, such as imazapyr, but the active site is unaltered (McCourt et al. 2006). There appear to be no other enzymatic targets of significance in plants as the presence of imidazolinones in csr1-2 mutant plants does not induce physiological alterations or changes in transcription profiles that can be detected by microarray (Manabe et al. 2007). These findings indicate that the mutant enzyme activity is unaffected by the mutation and that other plant processes are not affected by imidazolinones; therefore, the mutant plants are equivalent to wild-type plants in all other aspects.

Although tissue-specific variants of AHAS are found in some species, such as Brassica napus (Rutledge et al. 1991), the essential housekeeping genes are expressed constitutively throughout the plant (Ouellet et al. 1992). The lethal phenotype of the Arabidopsis deletion mutants csr1-7 and csr1-8 confirm that AHAS encoded by CSR1 is an essential enzyme. The recovery of 35S/CSR1-2/NOS transgenic lines that are homozygous for CSR1-7 indicate that the constitutively-expressed transgene can fully complement the null mutant in the transgenic state.

In heterozygous csr1-7 or csr1-8 plants we found that a single wild-type allele of CSR1 was sufficient for normal seed germination and vegetative plant development but defects in development appeared during the reproductive phase, specifically in seed development. Recent studies have shown that Arabidopsis embryos require an exogenous supply of amino acids for growth and development (Sanders et al. 2009). There is no evidence for transcriptional feedback regulation of AHAS levels in plants (Ouellet et al. 1994) and it is known that the CSR1 promoter is a weak promoter in transgenic plants (Charest et al. 1990). In other species, such as B. napus, the levels of transcripts of the housekeeping genes homologous to CSR1 are also low in the reproductive organs but not substantially lower than in the vegetative tissues (Ouellet et al. 1992). It is therefore likely that AHAS levels are not adequately regulated at the transcriptional or post-transcriptional levels to compensate for the reduction in gene dosage in the heterozygous csr1-7 and csr1-8 plants. It might be expected that BCAA supply to these organs was insufficient under conditions of reduced gene dosage through reduced levels of gene expression and thus AHAS activity. The use of the 35S promoter to elevate the level of CSR1-2 transcript by sevenfold to tenfold appeared to overcome the limitation and restored seed production levels to those of wild-type plants in some transgenic but not all lines. The embryo abortion phenotype, in particular type I, was highly variable in frequency among the transgenic lines. The transgenic lines analyzed were homozygous for the csr1-7 T-DNA insertion, so that expression controlled by the native promoter was eliminated and CSR1-2 expression is completely dependent on the 35S promoter, which is known to be weak during early embryogenesis in some species (Sunilkumar et al. 2002). Variable expression of the inserted CSR1-2 gene in the different transgenic lines during this developmental stage likely accounts for the observed variability in the seed abortion phenotype.

The introduction of CSR1-2 transcripts in transgenic plants using the 35S regulatory sequences was accompanied by a small number of changes in the transcriptome. In addition to CSR1-2, four genes were differentially expressed in all three transgenic lines. These small changes in gene transcription may be related to the higher levels of expression of the CSR1-2 gene in the transgenic plants, although the genes do not have any known connection with AHAS or amino acid biosynthesis. Alternatively, they may be the result of the more vigorous vegetative growth observed in the presence of imazapyr.

A search for changes in stress-related genes induced by imazapyr (Das et al. 2010) failed to indicate that any of the plants were under physiological stress. The transgenic plants were therefore very similar transcriptionally to the mutant csr1-2 plants. They had the added advantage of more robust growth in the presence of herbicide and a greater level of herbicide resistance.

In summary, this study provides insight into the differences and similarities of mutants and transgenic plants with the same novel trait, herbicide resistance, at the physiological and transcriptional levels. Furthermore it illustrates some of the advantages of transgenesis over mutagenesis by allowing for the manipulation of multiple regulatory domains that can affect gene expression and enzyme properties simultaneously. The data extends previous studies (El Ouakfaoui and Miki 2005; Manabe et al. 2007) by showing that transgenesis may be employed to generate plants that are similar transcriptionally to both wild-type and mutant plants, except for the gene encoding the novel trait. This kind of data is needed to develop scientifically sound regulatory policies for the integration of new crops to meet the unique challenges of future agriculture.