Amaranthus palmeri GS and GR biotypes were originally collected from North Carolina (United States) (Chandi et al., 2012; Fernández-Escalada et al., 2016). The resistance mechanism of the GR biotype was described to be EPSPS gene amplification (Chandi et al., 2012), with 47.5 more gene copies in GR plants than in GS plants (Fernández-Escalada et al., 2016). Seeds were surface-sterilized and germinated. The seeds were then transferred to 2.7-L tanks in a phytotron and grown in aerated hydroponic culture under controlled conditions, as described previously (Fernández-Escalada et al., 2016).

Before performing the incubation, a leaf of each plant of the A. palmeri GR population was harvested and immediately frozen in liquid nitrogen to determine the EPSPS relative genomic copy number of those individuals. After evaluating the results, 30 out of 48 evaluated plants were selected to obtain a homogeneous population with a similar relative EPSPS genomic copy number (between 60 and 100) to perform the experiments.

When the GS and selected GR plants were 21 days old, leaf disks were excised from the leaves using a Harris Uni-Core puncher (4-mm diameter) (Healthcore, Bucks, United States), avoiding the leaf nerves. From each plant, the two youngest fully expanded leaves were used: one leaf to determine the shikimate content and another leaf to determine transcript levels and enzyme content measurements.

The same treatments and doses were applied in both populations (Table 1). Leaf disks were incubated for 24 h under continuous light (300 μmol m^–2 s^–1 photosynthetic photon flux) at 24°C. Solutions were freshly prepared, and the pH was adjusted to 7.0 with NaOH in all the treatments. Technical glyphosate (glyphosate, isopropylamine salt, 61%; Dr. Ehrenstorfer GmbH, Augsburg, Germany) was used. All other reagents were purchased from Sigma–Aldrich Chemical, Co. (St. Louis, MO, United States).

The incubation system is shown in Supplementary Figure 2. For shikimate content determination, individual disks were placed in individual wells of a 96-well microtiter plate containing 100 μL of treatment solutions. To obtain one disk per treatment from the same plant, as many disks as treatments were obtained from the youngest leaf of the plant. For enzyme content and transcript level determinations, 25 or 45 disks were placed, respectively, in the wells of 6-well microplates containing 2.5 mL of each treatment. In each well, disks from different leaves were incubated, but the same proportion of original plants was maintained in all treatments tested. In both incubation options, each well was considered a biological replicate.

After incubation, disks incubated in 96-well plants were washed thoroughly before freezing, and the plates were placed in a freezer (−20°C). Disks incubated in 6-well microplates were removed from the incubating medium, immediately frozen in liquid nitrogen and stored at −80°C. The experiment was repeated twice.

Analysis was performed using 10 biological replicates for shikimate content determination and three or four biological replicates for enzyme content or nucleic acid determination, respectively. Replicates from both experiments were used. In the study of glyphosate and AAAs, differences between treatments for each population were evaluated by one-way ANOVA with a multiple-comparison adjustment (Tukey) at p ≤ 0.05. In the study using the intermediates, the difference between each parameter of untreated disks and disks of each treatment was evaluated using Student’s t-test (p ≤ 0.05). Statistical analyses were performed using IBM SPSS statistics 24.0 (IBM, Corp., Armonk, NY, United States).

Shikimate content was low and similar (0.4 μg shikimate leaf disk^–1) among untreated plants in both populations (Figure 1A). After glyphosate treatment, the shikimate content increased 11-fold in the GS population and 1.7-fold in the GR population. AAA applied alone did not modify the shikimate content in any of the populations, and when they were applied in combination with glyphosate (AAA+G treatment), the shikimate content was similar that accumulated by the herbicide-alone treatment (around 5 or 0.6 μg shikimate leaf disk^–1 in GS and in GR, respectively).

DAHPS and EPSPS enzymes were determined by immunoblotting of (40 μg of total soluble proteins for DAHPS and, 80 μg or 15 μg for EPSPS for GS GR, respectively) (Figure 1). In relation to the untreated control, the content of these enzymes was no altered by glyphosate or AAA treatments applied alone in any of the populations. The incubation with the combination of both compounds provoked different responses depending on the population. In the GS population, the combined treatment provoked a 4.7-fold increase in DAHPS content relative to the untreated plants, but did not modify the content of EPSPS enzyme. In contrast, in the GR population, DAHPS content was not affected by this treatment, but a two-fold increase in EPSPS was observed.

In the GS population, glyphosate produced an increase (of more than two-fold in all cases) in the relative expression level of all the genes of the prechorismate part of the AAA biosynthetic pathway (DAHPS, DHQS, DQSD, SK, EPSPS, and CS) and only an increase of ADHα transcript level in the postchorismate pathway (AS, CM2, CM1-3, ADHα, and ADHβ). In contrast, in the GR population, glyphosate did not induce the relative expression level of these genes in this pathway (Figure 2).

The only effect detected after AAA treatment was a five-fold decrease in the DQSD relative mRNA level to control plants in GS plants (Figure 2), and no effects were observed in the GR population. When AAA was applied in combination with glyphosate (AAA+G), a different response was observed in both populations. On the one hand, the GR population did not show changes in gene expression, with the exception of de DQSD gene that presented an increase of 2.3 times compared to the control (Figure 2). On the other hand, the presence of AAA in the combined treatment reversed the increase in transcript levels of the shikimate pathway detected after the herbicide alone in the GS population, as in this treatment, the effect of glyphosate on DQSD, SK, CS, or ADHα expression seemed to be abolished by the presence of AAA (Figure 2) and the relative mRNA level was not higher than control plants (Figure 2).

With leaf disk incubation, it was possible to reproduce the same physiological effects of glyphosate observed on whole plants. On the one hand, shikimate is a known marker of glyphosate activity (Dyer et al., 1988; Baerson et al., 2002; Zhu et al., 2008; Dillon et al., 2017), and it accumulated more in the GS than in the GR population. On the other hand, glyphosate upregulated the genes participating in the prechorismate pathway in the GS population (Figure 2), as previously described in this population (Fernández-Escalada et al., 2017) and in other species (Baerson et al., 2002; Garg et al., 2014). These results validate the incubation system used in this study to approach the regulation of the shikimate pathway by final products and intermediates.

In both populations, the accumulation of shikimate was similar in the two treatments with glyphosate, regardless of the presence of AAA (Figure 1A), suggesting that shikimate accumulation would be directly related to EPSPS inhibition and not to other physiological changes caused by the effect of the herbicide, such as a potential transitory modification of AAA content.

An increase in DAHPS or EPSPS enzyme content was not detected after incubation with glyphosate (Figure 1B,C), contrary to previous reports (Pinto et al., 1988; Gaines et al., 2010, 2011; Fernández-Escalada et al., 2017), which may be related to the short incubation period or the low dose applied. In previous experiments in which an increase in DAHPS and EPSPS content was observed, the time of treatment was 48 or 72 h (Baerson et al., 2002; Fernández-Escalada et al., 2017).

In microbes, it has been widely described that DAHPS activity is regulated in response to cellular levels of AAA (Herrmann, 1995; Tzin and Galili, 2010). However, most of the studies performed in plants suggest that DAHPS is not regulated by AAA, and only a few reports have described a regulatory effect of AAA levels on this enzyme activity in vitro (Graziana and Boudet, 1980; Suzich et al., 1985). Similarly, AAA treatment alone did not modify DAHPS content in this experiment (Figure 1). Indeed, the application of AAA and glyphosate together induced an increase in the DAHPS content in the GS population, supporting that in plants, the DAHPS content is not regulated by AAA levels, as it was not downregulated when AAA were externally supplied.

It has been suggested that a reduced level of AAA may act as a signal to induce the expression of the shikimate pathway and restore carbon flux through the pathway in plants (Maeda and Dudareva, 2012). In concordance with this hypothesis, a decrease in the relative expression level after the exogenous supply of AAA could have been expected. Nevertheless, this decrease did not occur, as no changes in the relative expression level were detected in the GS or GR population after AAA supply with the exception of the bifunctional enzyme DQSD (Figure 2). Similar to the results obtained in this study, the effect of the increase in Phe and Tyr in Arabidopsis transgenic plants on the transcriptome was low (Dubouzet et al., 2007; Tzin et al., 2009; Less et al., 2010).

The reversion of glyphosate effects on growth with exogenous AAA has been previously shown in fungi, bacteria and higher plant cell cultures (Jaworski, 1972; Gresshoff, 1979; Amrhein et al., 1983), while in higher plants, the reversion of the effect on growth effect has been detected in Arabidopsis (Gresshoff, 1979) but not in other plant species (Duke and Hoagland, 1978; Cole et al., 1980; Duke et al., 1980). Nevertheless, reversion of the physiological effects induced by glyphosate, such as anthocyanin (Hollander and Amrhein, 1980) and protein (Tymonko and Foy, 1978) synthesis, has been reported in several plant species. Similarly, in this study, supplemental AAA completely prevented one physiological effect of glyphosate: the upregulation of shikimate pathway genes.

It appears that in the GS population, the increase in gene expression of the enzymes in the shikimate pathway after glyphosate might be mediated by a transitory lack of AAA, as the exogenous supply of AAA can abolish the gene upregulation. This finding might appear contradictory to previous studies in which reduction of AAA was proposed to not elicit the increased expression of AAA pathway genes because the AAA concentrations increase with glyphosate dose (Vivancos et al., 2011; Maroli et al., 2015; Fernández-Escalada et al., 2017). In the same way, the previously reported DAHPS and EPSPS increases after glyphosate (Pinto et al., 1988; Fernández-Escalada et al., 2016) were not detected. Nevertheless, these results might be explained by the short incubation time used in this study: 24 h compared to 3 days. Indeed, previous studies have shown that in glyphosate-treated pea roots the relative content of AAA was transitorily decreased for 24 h and abolished after 3 days of treatment (Orcaray et al., 2010). The reversion effect of AAA on the upregulation of expression genes of the shikimate pathway enzymes indicated that at least in the 24 h following application of treatment, the effect was mediated by a lack of the end products of the pathway.

Shikimate content was determined after incubation with the intermediates, and the pattern observed in both populations was similar (Figure 3A). Shikimate content increased in leaf disks after incubation with shikimate (40 times and 27 times in relation to the untreated control in the GS and GR populations, respectively), confirming that the compound was absorbed (Figure 3A). Shikimate content also increased after quinate treatment also (3 times in GS and 1.4 times in GR). These results showed that both in shikimate and quinate treatments, the increase observed in GS population was two times the increase observed in the GR population. Interestingly, anthranilate induced an increase in shikimate content in both populations, similar to the increase detected after quinate incubation.

DAHPS and EPSPS contents were determined after disk incubation in both populations (Figure 3). The incubation with shikimate and quinate did not induce any changes in the content of these enzymes in relation to the control disks in any population. Incubation with anthranilate elicited different responses, depending on the population and on the specific enzyme: a 2.7-fold increase in EPSPS content in GS and a 5.8-fold decrease in DAHPS expression in the GR population. Interestingly, chorismate was the only intermediate that induced a general increase in both enzymes and in both populations (DAHPS was increased 2.4 and 2.8 times and EPSPS 1.6 and 2.4 times, in GS and GR, respectively).

The incubation with shikimate induced the upregulation of more than half of the genes of the AAA pathway in the GS population (DHQS, CS, AS, CM2, ADHα, and ADHβ), with the post-chorismate part being the more affected. Interestingly, this pattern was not observed in the GR population, where only DQSD and the ADHα isoform increased their relative mRNA levels after shikimate (Table 2).

In the GS population, quinate treatment also increased the relative expression level in five out of the genes, although the level of increase was milder than that detected after shikimate, as the medium relative increase detected after shikimate was 3 times and 2.28 times after quinate (Table 2). In the GR population, four genes of the pathway (DAHPS, DQSD, EPSPS, and ADHα) were upregulated after quinate incubation. The incorporation of quinate in the shikimate pathway can occur through two different pathways: through the reversible quinate dehydrogenase to 3-dehydroquinate and through the quinate hydrolyase to shikimate (Ossipov et al., 2000). The 2.6-fold reduction in the relative gene expression in the DQSD gene complex in the GS population after quinate would be related to the main incorporation of quinate into the pathway after DQSD, and the increase in the flux only at that point would act as a signal to reduce the relative gene expression of the enzyme to regulate the pathway. The increase in the relative transcript level of one of the isoforms of CM (3.6-fold) and ADH (1.6-fold) genes in GS and ADH (9.4-fold) in GR after quinate would confirm that quinate would direct the flux to the synthesis of Tyr and Phe, as has been described recently (Zulet-González et al., 2019).

The incubation with chorismate did not alter the transcription level of DAHPS, DQSD, SK, EPSPS, CM2, CM1-3, ADHα, ADHβ in any population (Table 2). In the GR population, no changes in relative gene expression were observed. In GS, only upregulation was observed in DHQS, CS, and AS genes, with the highest increase being observed for the AS gene (4.3-fold), the enzyme that uses chorismate as a substrate. It seems that as happens with glyphosate (Fernández-Escalada et al., 2017), the flux would be redirected toward the Trp biosynthetic pathway when chorismate is added exogenously. In addition, chorismate increased the content of the enzymes DAHPS and EPSPS (Figure 3), which does not match the results observed in the relative transcription level of the genes (Table 2), in which no changes were detected. It seems that the increase in the content of both enzymes would be due to the posttranscriptional regulation process elicited by the presence of chorismate. Nevertheless, it has to be noted that plants regulate carbon flux toward AAA biosynthesis at the transcriptional and post-transcriptional levels (Maeda and Dudareva, 2012).

Anthranilate was the intermediate that less affected the expression level of the shikimate pathway. In the GR population, only SK expression was 1.75-fold upregulated. In the GS population, a different response was observed on the two isoenzymes of CM: the plastidics 1 and 3 isoenzymes but not the cytosolic isoform 2 would be downregulated by anthranilate. In the same way, the two isoenzymes of ADH also showed different responses, and ADHα expression was the only downregulated 1.5-fold by anthranilate. In the same way, differences in the regulation of the isoforms by other metabolites of the pathway has been reported in Beta vulgaris, where only the activity of ADHα exhibited relaxed sensitivity to Tyr and ADHβ was strongly inhibited (Lopez-Nieves et al., 2017).

AAA biosynthesis is subjected to complex posttranscriptional and allosteric regulations (Mir et al., 2015). How carbon flow into the shikimate pathway is regulated in plants and the specific role of each intermediate remain ambiguous Siehl (1997) suggested that the inhibition of DAHPS activity by arogenate, a metabolite of the post-chorismate part of the pathway, was the key regulatory process in the shikimate pathway. In the case of glyphosate exposure, this regulatory pathway cannot occur, since chorismate and all its byproducts are not synthesized, resulting in an increase in the flux through this pathway and the accumulation of sikimate-3-phosphate (Gomes et al., 2014) and shikimate. This hypothesis was not confirmed in this study as, contrary to expectations, chorismate incubation increased the content of DAHPS and EPSPS in both populations.

The results showed that in sensitive plants glyphosate affects the shikimate pathway, and no such changes were observed in the EPSPS overexpressing population, consistent with its resistance to glyphosate, as has been reported before in susceptible and resistant soybeans (Marchiosi et al., 2009). Beside this, less effect on the pathway was detected on the GR than on the GS population after of the shikimate pathway intermediaries. Indeed, each intermediate induced higher changes in gene expression in the GS population than in the GR population, suggesting that the overexpression of EPSPS would have an effect in the regulation of the pathway, buffering or attenuating the transcriptional changes induced by the intermediates.