The Triple Amino Acid Substitution TAP-IVS in the EPSPS Gene Confers High Glyphosate Resistance to the Superweed Amaranthus hybridus

The introduction of glyphosate-resistant (GR) crops revolutionized weed management; however, the improper use of this technology has selected for a wide range of weeds resistant to glyphosate, referred to as superweeds. We characterized the high glyphosate resistance level of an Amaranthus hybridus population (GRH)—a superweed collected in a GR-soybean field from Cordoba, Argentina—as well as the resistance mechanisms that govern it in comparison to a susceptible population (GSH). The GRH population was 100.6 times more resistant than the GSH population. Reduced absorption and metabolism of glyphosate, as well as gene duplication of 5-enolpyruvylshikimate-3-phosphate synthase (EPSPS) or its overexpression did not contribute to this resistance. However, GSH plants translocated at least 10% more 14C-glyphosate to the rest of the plant and roots than GRH plants at 9 h after treatment. In addition, a novel triple amino acid substitution from TAP (wild type, GSH) to IVS (triple mutant, GRH) was identified in the EPSPS gene of the GRH. The nucleotide substitutions consisted of ATA102, GTC103 and TCA106 instead of ACA102, GCG103, and CCA106, respectively. The hydrogen bond distances between Gly-101 and Arg-105 positions increased from 2.89 Å (wild type) to 2.93 Å (triple-mutant) according to the EPSPS structural modeling. These results support that the high level of glyphosate resistance of the GRH A. hybridus population was mainly governed by the triple mutation TAP-IVS found of the EPSPS target site, but the impaired translocation of herbicide also contributed in this resistance.


Introduction
Plant breeding methods have delivered herbicide resistant crops that offer advantages for weed control [1]. The introduction of glyphosate-resistant (GR) crops in 1996 revolutionized weed management practices [2]. Agricultural areas occupied by these crops, mainly GR-soybean and GR-corn, increased considerably in Argentina, Brazil and USA [2,3]. Inadequate adoption of GR-crops, i.e., higher doses and more applications of glyphosate than recommended by the manufacturer, has

Shikimic Acid Accumulation and Genotyping
Shikimic acid accumulation differentiated the susceptibility to glyphosate of the GSH and GRH populations. GSH plants presented different shikimate accumulation patterns, while the GRH population seems to be more homogeneous ( Figure 1A). Thus, GSH (eight) and GRH (nine) plants, with similar shikimate accumulation profiles, were selected to obtain homogeneous lines for future experiments. Ten F 1 plants from each population were confirmed as being A. hybridus ( Figure 1B).

Dose-response Assays
Plant survival and fresh weight decreased as the dose of glyphosate increased ( Figure 2). The GSH population died at lower glyphosate doses than the doses used by Argentinian farmers (960 g ae ha −1 ). The GR50 and LD50 values estimated for the GRH population were 1395.2 and 3503.4 g ae ha −1 glyphosate, respectively. Looking at the LD50 parameter, the RI for GRH was 100.6 times more resistant than the GRH population (Table 1).

Dose-response Assays
Plant survival and fresh weight decreased as the dose of glyphosate increased ( Figure 2). The GSH population died at lower glyphosate doses than the doses used by Argentinian farmers (960 g ae ha −1 ). The GR 50 and LD 50 values estimated for the GRH population were 1395.2 and 3503.4 g ae ha −1 glyphosate, respectively. Looking at the LD 50 parameter, the RI for GRH was 100.6 times more resistant than the GRH population (Table 1).

Dose-response Assays
Plant survival and fresh weight decreased as the dose of glyphosate increased ( Figure 2). The GSH population died at lower glyphosate doses than the doses used by Argentinian farmers (960 g ae ha −1 ). The GR50 and LD50 values estimated for the GRH population were 1395.2 and 3503.4 g ae ha −1 glyphosate, respectively. Looking at the LD50 parameter, the RI for GRH was 100.6 times more resistant than the GRH population (Table 1).
where Y = response in relation to the control, c = lower limit, d = upper limit, b = slope of the curve, g = herbicide concentration at the inflection point (i.e., LD 50 , GR 50 , or I 50 ), and x = herbicide concentration. The model of three-parameters assumes that the lower limit is zero. RI= Resistance indexes (R/S) computed as R-to-S LD 50 , GR 50 or I 50 ratios. ± Standard error of the mean (n = 10 for LD 50 and GR 50 , and n = 5 for I 50 ).

14 C-Glyphosate Absorption and Translocation
The 14 C-glyphosate absorption ranged between 12 and 76% from 12 to 96 HAT, but there were no differences between GRH and GSH plants ( Figure 3A). Already, the GSH plants moved more 14 C (glyphosate or metabolites) from the treated leaf to the remainder of plant and roots than GRH plants ( Figure 3B). Quantitative translocation results showed that GRH plants retained~68% of the glyphosate in the treated leaf while the GSH plants only retained~53%. Thus, the 14 C-herbicide translocated to the rest of the plant and roots was up to~18 and~28%, respectively, in GSH plants, while in GRH plants it was~14 and~18% ( Figure 3C).
} (four-parameters) ‡ : where Y = response in relation to the control, c = lower limit, d = upper limit, b = slope of the curve, g = herbicide concentration at the inflection point (i.e., LD50, GR50, or I50), and x = herbicide concentration. The model of three-parameters assumes that the lower limit is zero. RI= Resistance indexes (R/S) computed as Rto-S LD50, GR50 or I50 ratios. ± Standard error of the mean (n = 10 for LD50 and GR50, and n = 5 for I50).

14 C-Glyphosate Absorption and Translocation
The 14 C-glyphosate absorption ranged between 12 and 76% from 12 to 96 HAT, but there were no differences between GRH and GSH plants ( Figure 3A). Already, the GSH plants moved more 14 C (glyphosate or metabolites) from the treated leaf to the remainder of plant and roots than GRH plants ( Figure 3B). Quantitative translocation results showed that GRH plants retained ~68% of the glyphosate in the treated leaf while the GSH plants only retained ~53%. Thus, the 14 C-herbicide translocated to the rest of the plant and roots was up to ~18 and ~28%, respectively, in GSH plants, while in GRH plants it was ~14 and ~18% ( Figure 3C).

Glyphosate Metabolism
Glyphosate was not metabolized by A. hybridus plants. At 48 and 96 HAT, GRH plants translocated less herbicide towards roots than GSH ones, observing the biggest differences at 96 HAT. The amounts of glyphosate found in the A. hybridus plants ranged from 437 to 605 nmols g −1 fresh weight in the aerial part, and from 48 to 238 nmols g −1 fresh weight in roots ( Table 2). Small amounts of glyoxylate were detected at 96 HAT in both GSH and GRH plants ( Figure 4), but this compound cannot be considered a metabolite because it does not originate solely from glyphosate. This data confirmed the results obtained in the absorption and translocation assays, i.e., the 14 C detected by LSS corresponded to glyphosate molecules and not a possible metabolite.

Glyphosate Metabolism
Glyphosate was not metabolized by A. hybridus plants. At 48 and 96 HAT, GRH plants translocated less herbicide towards roots than GSH ones, observing the biggest differences at 96 HAT. The amounts of glyphosate found in the A. hybridus plants ranged from 437 to 605 nmols g −1 fresh weight in the aerial part, and from 48 to 238 nmols g −1 fresh weight in roots ( Table 2). Small amounts of glyoxylate were detected at 96 HAT in both GSH and GRH plants ( Figure 4), but this compound cannot be considered a metabolite because it does not originate solely from glyphosate. This data confirmed the results obtained in the absorption and translocation assays, i.e., the 14 C detected by LSS corresponded to glyphosate molecules and not a possible metabolite.

EPSPS Enzyme Activity Assays
Amaranthus hybridus plants (GSH and GRH) presented similar EPSPS basal activity profiles ( Figure 5A). The EPSPS was inhibited in both GSH and GRH plants as glyphosate concentrations increased, but inhibition occurred at different herbicide concentrations ( Figure 5B). These concentrations were 0.52 and 52.8 µM glyphosate for the GSH and GRH populations, respectively, resulting in an RI (R-to-S ratio) of 100.9.

EPSPS Enzyme Activity Assays
Amaranthus hybridus plants (GSH and GRH) presented similar EPSPS basal activity profiles ( Figure 5A). The EPSPS was inhibited in both GSH and GRH plants as glyphosate concentrations increased, but inhibition occurred at different herbicide concentrations ( Figure 5B). These concentrations were 0.52 and 52.8 µM glyphosate for the GSH and GRH populations, respectively, resulting in an RI (R-to-S ratio) of 100.9.

EPSPS Enzyme Activity Assays
Amaranthus hybridus plants (GSH and GRH) presented similar EPSPS basal activity profiles ( Figure 5A). The EPSPS was inhibited in both GSH and GRH plants as glyphosate concentrations increased, but inhibition occurred at different herbicide concentrations ( Figure 5B). These concentrations were 0.52 and 52.8 µM glyphosate for the GSH and GRH populations, respectively, resulting in an RI (R-to-S ratio) of 100.9.

EPSPS Copy Number and Gene Expression
EPSPS copy gene number relative to ALS gene showed that there were no differences between GSH and GRH biotypes. However, a slight increase in the EPSPS gene expression was observed ( Figure 6). The EPSPS and expression copy number averages were 0.99 ± 0.03 and 0.82 ± 0.04, respectively, for the GSH population, and 1.36 ± 0.05 and 1.45 ± 0.10 for the GRH population.

EPSPS Copy Number and Gene Expression
EPSPS copy gene number relative to ALS gene showed that there were no differences between GSH and GRH biotypes. However, a slight increase in the EPSPS gene expression was observed ( Figure 6). The EPSPS and expression copy number averages were 0.99 ± 0.03 and 0.82 ± 0.04, respectively, for the GSH population, and 1.36 ± 0.05 and 1.45 ± 0.10 for the GRH population.

Mutations in the EPSPS Coding Sequence
EPSPS gene sequencing revealed a novel amino acid substitution from Ala to Val at the 103 position in the GRH population, in addition to those already described at Thr-102-Ile and Pro-106-Ser, i.e., a triple mutation was found. This triple amino acid substitution consisted of TAP (wild type, GSH) to IVS (triple mutant, GRH). The nucleotide substitutions were ATA 102 , GTC 103 , and TCA 106 instead of ACA 102 , GCG 103 , and CCA 106 , respectively ( Figure 7A). The A. hybridus EPSPS cDNA sequences can be found in GenBank (accession numbers MG595170-MG595171).

Mutations in the EPSPS Coding Sequence
EPSPS gene sequencing revealed a novel amino acid substitution from Ala to Val at the 103 position in the GRH population, in addition to those already described at Thr-102-Ile and Pro-106-Ser, i.e., a triple mutation was found. This triple amino acid substitution consisted of TAP (wild type, GSH) to IVS (triple mutant, GRH). The nucleotide substitutions were ATA 102 , GTC 103 , and TCA 106 instead of ACA 102 , GCG 103 , and CCA 106 , respectively ( Figure 7A). The A. hybridus EPSPS cDNA sequences can be found in GenBank (accession numbers MG595170-MG595171).
EPSPS gene sequencing revealed a novel amino acid substitution from Ala to Val at the 103 position in the GRH population, in addition to those already described at Thr-102-Ile and Pro-106-Ser, i.e., a triple mutation was found. This triple amino acid substitution consisted of TAP (wild type, GSH) to IVS (triple mutant, GRH). The nucleotide substitutions were ATA 102 , GTC 103 , and TCA 106 instead of ACA 102 , GCG 103 , and CCA 106 , respectively ( Figure 7A). The A. hybridus EPSPS cDNA sequences can be found in GenBank (accession numbers MG595170-MG595171).

Structural Modeling
EPSPS structural modeling revealed increases in the hydrogen bond (H bond) distances between the Gly-101 and Arg-105 positions of the EPSPS gene from GRH plants. In EPSPS target site of GSH plants, the H bond distance between these two amino acids was 2.89 Å, while in GRH plants this distance was 2.93 Å ( Figure 7B,C).

Discussion
Argentinian farmers quickly adopted the technology packages for cultivation of GR-crops, mainly soybean and corn [4]. In the present study, the degree of glyphosate resistance in a population of the superweed A. hybridus, occurring in a GR-soybean field from Cordoba, was characterized, as well as the resistance mechanisms that govern it. The accumulation of shikimic acid is a biochemical indicator of glyphosate effects on susceptible plants [11]. Once the glyphosate susceptibility profile of the GSH and GRH populations via shikimate accumulation were differentiated, F 1 individuals from these populations were identified as being A. hybridus subsp. hybridus, based on the length of intron 1 of the EPSPS gene [27,28].
The dose-response assays showed that the GRH population had a high level of resistance to glyphosate. The GR 50 estimated for the GRH populations was higher than the glyphosate dose of 960 g ae ha −1 used in the field by Argentinian farmers. However, this GR 50 (1395 g ae ha −1 ) was 15 times less than the GR 50 value determined for another R population of A. hybridus carrying the triple substitution TAP-IVS [29]. These differences were mainly due to the phenological stage to which the plants were treated in both experiment (4-true leaves vs 20 cm in height (8-12 true leaves)), since older plants are less susceptible to herbicides [30]. In addition, the volume applied and the equipment used for glyphosate applications (200 L ha −1 in a treatment chamber vs 75 L ha −1 with a backpack sprayer) influenced in the differential response observed in both resistant populations. However, the high level of glyphosate resistance selected by the two resistant populations of A. hybridus, which were collected in different regions of the province of Cordoba, Argentina, was consistent. The impact on crop yield of superweeds is strong due to the wide use of glyphosate and because GR crops are designed to tolerate this herbicide [4]. In GR crop systems from Argentina, Shorgum halepense was confirmed as being glyphosate-resistant [31], and other common weeds as tolerant [32].
Reduced absorption and/or metabolism of glyphosate were not relevant in the high resistance of the GRH A. hybridus population; however; the translocation was significantly lower than the GSH population. Previously, an unknown barrier in the phloem system or in the mesophyll cells was suggested as limiting glyphosate translocation [33]. Studies with 31 P nuclear magnetic resonance have shown that glyphosate is compartmentalized and deactivated into the vacuole [14,34,35], a process that is regulated by tonoplast-active transporters [35]. This NTSR mechanism endows moderate-level glyphosate resistance [36]; therefore, the high resistance level of the GRH population (100.6-times in relation to the GSH) cannot be explained only by this mechanism alone, suggesting that TSR mechanism may be preventing the interaction of glyphosate with the EPSPS.
The GSH and GRH A. hybridus populations presented similar EPSPS gene copy numbers, and although there was a slight increase in the expression of this enzyme in the GRH population, such expression did not result in greater EPSPS basal activity. EPSPS gene amplification was reported in several GR Amaranthus species. For example, A. palmeri had from 5 to 160 copies of the EPSPS gene [10], A. tuberculatus had from 32 to 59 copies [37], and A. spinosus presented 33-37 copies [38]. Given that the GRH A. hybridus plants did not show an increased EPSPS gene copy number, the slight increase in EPSPS gene expression did not explain the differences observed in the EPSPS activity.
Interestingly, for the first time, an EPSPS triple mutation (Thr-Ile 102, Ala-Val 103, and Pro-Ser 106) was found in the GRH A. hybridus population [25], which could explain the high levels of glyphosate resistance observed in these plants. Several mutations in the EPSPS gene have been suggested as contributing to glyphosate resistance [8,39]. However, the mutations conferring resistance to this herbicide must occur in the conserved region of this gene ( 95 LFLGNAGTAMRPL 107 ), that interact directly with glyphosate, as demonstrated in Escherichia coli [40,41]. Single amino acid substitutions occurring at the 106 position from Pro to Ser, Ala, Thr, and/or Leu, endowing total or partial resistance to glyphosate, has been widely reported in mono-and dicotyledonous weeds [8]. These single mutations provide low levels of resistance (2-3-fold), but they are sufficient for weeds to survive at field doses of glyphosate [36]. Single Pro-106 substitutions cause a slight narrowing of the glyphosate/phosphoenolpyruvate (PEP) binding site cavity, endowing glyphosate resistance but preserving EPSPS functionality [40]. Single substitutions at Gly-101 and Thr-102 confer high-level glyphosate resistance, but decrease the volume of the glyphosate/PEP binding site, reducing affinity for PEP [41,42]. Sammons and Gaines [8] warned that Thr-102 mutations would be unlikely to occur first or independently of Pro-106 mutations due to their significant reduction in PEP Km; however, Tridax procumbens presented a novel single mutation from Thr to Ser at the EPSPS genomic position 102 [43]. The double mutation known as TIPS (Thr-102-Ile + Pro-106-Ser), used in some GR-crops, was reported in Eleusine indica [36,44] and Bidens pilosa [13], endowing them high glyphosate resistance levels (GR 50 values ranging from 1055 to 2050 g ae ha −1 ). The GRH A. hybridus population showed an additional mutation, occurring at the 103 position, associated with the TIPS mutation, which could be responsible for further increasing the level of glyphosate resistance.
How the Ala-103-Val substitution can affect the protein function is still unknown, but based on the amino acid properties, the effect that this mutation could produce on the EPSPS activity can be predicted. The side chain of alanine is poorly reactive and rarely participates in protein function, but may play a role in the recognition or specificity of the substrate [45]. Being hydrophobic, valine prefers to be buried in hydrophobic protein cores; however, this amino acid has an often overlooked property, it is branched into Cβ like isoleucine and threonine [45]. While most amino acids contain only one nonhydrogen substituent on their Cβ carbon, these three amino acids contain two. This means that there is much more volume in the main chain of the protein, such that these amino acids are more restricted in the conformations of the main chain that they can adopt [45]. The substitution Val-103-Ala found in the GRH A. hybridus plants increased the distance between the H bonds in the α-helix of EPSPS target site, i.e., the alanine side chain occupies more space than the valine one. Therefore, the occurrence of the triple substitution in the conserved region of the EPSPS encoding gene supports that TAP-IVS mutation contributed to the high level of glyphosate resistance of the GRH A. hybridus population. In addition, in silico conformational studies in another glyphosate-resistant population of A. hybridus, also collected in the province of Cordoba Argentina, showed that the TAP-IVS mutation restricts the binding of glyphosate with the EPSPS enzyme [29].

Plant Materials
Mature seeds of a resistant A. hybridus population (GRH) were harvested in no-till GR-soybean cropping systems from the Cordoba province (Argentina) (31.48 • S, 64.01 • W) in the crop season 2015/2016. This field had a 20-year history of glyphosate application (twice per crop season) at 960 g ae ha −1 . There was one application (glyphosate alone or in mixture with a residual pre-emergent herbicide) prior to crop sowing, and another 30 d after sowing. In addition, this field was cultivated with wheat Clearfield ® in the winter season. Susceptible (GSH) seeds were collected in a nearby area with no history of herbicide application in 2016.
The GRH and GSH populations were germinated in trays containing peat and sand, moistened with distilled water and placed in a growth chamber (28/18 • C day/night, 16-h photoperiod, 850 µmol m −2 s −1 photosynthetic photon flux, and 80% relative humidity). Seedlings were then transplanted individually into 250 mL pots containing sand/peat in a 1:2 (v/v) ratio, and kept in the growth chamber until the herbicide applications. Afterwards, treated plants were placed in a greenhouse under similar conditions to those of the growth chamber.

Fast Screening of Resistance via Shikimic Acid Accumulation
Ten plants from the A. hybridus population were used for a fast screening using shikimic acid accumulation as parameter to separate resistant plants from susceptible ones. Three disks (4 mm in diameter) of fresh tissue from the youngest expanded leaf per plant were transferred to 2-mL Eppendorf tubes. Shikimic acid accumulation was determined according to Shaner et al. [11] at a single concentration of glyphosate (1000 µM). Assays were repeated three times. Results were expressed in mg of shikimic acid g −1 fresh tissue. Plants were separated by high (GSH) or low (GRH) accumulation of shikimic acid, transplanted into pots (30 × 60 cm), and left to grow until maturity to produce new seeds (F 1 ) which were used for all future experiments.

Species Identification
Genomic DNA from ten individuals of each F 1 A. hybridus population was purified using the Qiagen DNeasy Plant mini kit (Qiagen, Valencia, CA, USA) according to manufacturer's instructions. Species identification was performed by PCR using the specific primers AW473/AW483 (1623 bp) developed by Wright et al. [27]. PCR products were analyzed by agarose gel electrophoresis.

14 C-Glyphosate Visualization
Plants of GRH and GSH populations at the 4-leaf growth stage were treated and handled at 96 HAT using the same media as those used in the previous section. After removing the nonabsorbed 14 C-glyphosate by washing the treated leaf three times with water-acetone (1:1, v/v), the whole plant was removed from the individual pot and its roots carefully washed with distilled water. Excess moisture was removed with paper towel and plants were fixed on filter paper (25 × 12.5 cm) and dried at room temperature (24 • C) for a week. Later, plant samples were placed on phosphor storage film for 4 h and scanned using a Cyclone (phosphor imager, Perkin-Elmer, Bioscience BV Packard, Groningen, Netherlands) to visualize the 14 C distribution (as glyphosate and/or its potential metabolites) within plants. The assay was carried out with three plants per A. hybridus population.

Glyphosate Metabolism
A set of five plants (3-4 leaves) per A. hybridus population were treated at 360 g ae ha −1 glyphosate as in dose-response assays, while another group of five plants were assigned as the control. Treated and untreated plants were harvested at 48 and 96 HAT and leaf tissue was washed with distilled water, flash-frozen in liquid nitrogen, and stored at −40 • C until use. Glyphosate and its metabolites (AMPA, glyoxylate, sarcosine and formaldehyde) were quantified using a 3D Capillary Electrophoresis Agilent G1600A instrument equipped with a diode array detector (DAD, wavelength range: 190 to 600 nm) [46]. Calibration curves were obtained by using known concentrations of standards (glyphosate and metabolites) supplied by Sigma Aldrich (St. Louis, MI, USA). The experiment had a completely randomized design and repeated three times. Data was expressed as percentages from the total of glyphosate plus metabolites recovered.

EPSPS Enzyme Activity Assays
Five grams (g) of leaf tissue from each A. hybridus population finely powdered with liquid nitrogen. EPSPS enzyme extraction was performed following the protocol described by Dayan et al. [47] The total soluble protein (TPS) in the extract (EPSPS basal activity in absence of glyphosate) was determined by the Bradford assay [48], using a Kit for Protein Determination (Sigma-Aldrich, Madrid, Spain) following manufacturer's instructions. The specific EPSPS activity was assayed in the presence of glyphosate (0, 0.1, 1, 10 100, and 1000 µM) using the EnzChek Phosphate Assay Kit (Invitrogen, Carlsbad, CA, USA) to determine the EPSPS inhibition by 50% (I 50 ). Five replications of each population per glyphosate concentration were assayed. EPSPS inhibition was expressed as a percentage relative to the control (absence of glyphosate)

EPSPS Copy Number and Gene Expression
Young leaf tissue of ten individuals from each A. hybridus population was collected by taking two samples per plant, one for the total RNA extraction and the other for the genomic DNA extraction. Total RNA was extracted using the Tri Reagent solution (Molecular Research Center, Inc. Cincinnati, OH, USA) according to the manufacturer's instructions. RNA was treated using the RNase free DNase set (Qiagen, Valencia, CA, USA). First strand complementary DNA (cDNA) synthesis was carried out with 3 µg of total RNA per sample using an M-MLV reverse transcriptase (Promega, Madison, WI, USA) and random hexamers as primers. Cycle conditions: 37 • C for 1 h, 42 • C for 30 min, 50 • C for 10 min, and 15 • C for 10 min. The gDNA was purified using the Qiagen DNeasy Plant mini kit (Qiagen, Valencia, CA, USA) according to manufacturer's instructions.
Copy number (from gDNA) and gene expression (from cDNA) assays were performed using EPSPS and ALS primer pairs developed by Gaines et al. [10]. Reactions were performed using a qRT-PCR Bio-Rad CFX connect thermal cycler and the following amplification profile; 50 • C for 2 min, 95 • C for 10 min, 40 cycles at 95 • C for 15 s and 60 • C for 1 min, and 95 • C for 15 s. PCR reactions were set up in 20 µL of SYBR Green PCR Master Mix (Bio-Rad, Hercules, CA, USA), following manufacturer's instructions. Controls containing water were included to check for contamination in the reaction components. The ALS gene was used as a reference gene to normalize qRT-PCR results. Reactions was carried out in triplicate per sample and standard curves were performed for each primer pair to confirm the amplification efficiency (E = 100 ± 10%). The relative expression levels were calculated from the threshold cycle (Ct) values and the primer efficiencies by the Pfaffl method [49]. The EPSPS gene copy number was determined [10]. Results were expressed as relative EPSPS gene copies in relation to the ALS gene [49].

Mutations in the EPSPS Coding Sequence
Sanger sequencing was used to detect target site mutations in the A. hybridus EPSPS gene sequence using the DNA purified previously. For the sequencing reactions, a 196 bp long DNA fragment was amplified using the primers described in Lorentz et al. [25]. The PCR conditions were 15 min of preincubation at 95 • C, followed by 45 cycles at 94 • C denaturating step (30 s), 55 • C annealing temperature (40 s), 70 • C elongation (40 s) and a final extension step for 10 min at 70 • C (Bio-Rad T100 thermal cycler, Hercules, CA, USA). Each plant (ten per population) was sequenced in triplicate.

Structural Modeling
The spatial structure of the wild type (WT; the GSH population) isoform of the EPSPS was reconstructed by a homology modeling approach using the swiss-Model software [50]. The previously obtained spatial structure EPSPS from Vibrio cholerae (PDB code: 3nvs) was used as a template for A. hybridus EPSPS reconstruction. To generate the promodels of mutant A. hybridus EPSPS isoforms, the appropriate amino acids at the positions 102 (Thr-Ile), 103 (Ala-Val) and 106 (Pro-Ser) were changed using the DeepView software version 4.1.0. This same software was used to analyze structural differences between the WT and mutant isoform.

Statistical Analysis
The amount of glyphosate causing the fresh weight reduction (GR 50 ), plant mortality (LD 50 ) and EPSPS inhibition (I 50 ) by 50% of each A. hybridus population were determined using the following log-logistic equations [51]: Y = d/1 + (x/g) b (three-parameters for GR 50 and LD 50 ) and Y = c + {(d-c)/[1 + (x/g) b ]} (four-parameters for I 50 ): where Y is the percentage of fresh weight, mortality and/or EPSPS inhibition relative to the control; c and d are the lower and upper limits, respectively, of the curve; b is the slope of the curve; g = herbicide concentration at the inflection point (i.e., LD 50 , GR 50 , or I 50 ), and x is the glyphosate dose. The three-parameters model assumes that the lower limit is zero. Regression analyses were conducted using the drc package with program R 3.2.5, and plotted using SigmaPlot 11.0 (Systat Software, Inc., San Jose, CA. USA).The resistance indexes (RI = R/S) were computed as R-to-S ratios.
Data regarding basal EPSPS activity, absorption, translocation, and metabolism were subjected to ANOVA using Statistix 9.0 (Analytical Software, Tallahassee, FL, USA). Model assumptions of normal distribution of errors and homogeneous variance were graphically inspected. Differences of p < 0.05 were considered significant and Tukey's test was conducted for means comparison.

Conclusions
The GRH A. hybridus population presented reduced translocation of glyphosate and signals of EPSPS gene amplification; however, the contribution of these mechanisms in the high resistance level showed by this population was minimal. That resistance was due to the triple amino acid substitution from TAP (wild type) to IVS (triple mutant) (Thr102Ile + Ala103Val + Pro106Ser), occurring in the conserved region of the EPSPS gene, found in the GRH A. hybridus population. This novel triple substitution increased the hydrogen-bond distance between the Gly-101 and Arg-105 positions restricting the binding of glyphosate with the EPSPS enzyme.