Transgenerational memory of drought stress and low rates of glyphosate reduce the sensitivity of Eragrostis plana to the herbicide

: Background: Eragrostis plana is the main invasive plant in the Pampa Biome of southern Brazil. This plant is highly competitive, tolerant to abiotic stresses, and very difficult to manage. Plants exposed to abiotic stresses and herbicides develop mechanisms that help alleviate or reduce damage caused by stressors and transmit this ability to the progenies. Objective : The study’s objectives were to ascertain whether acclimatization to stress due to drought and sub-lethal doses of glyphosate may decrease plant sensitivity to glyphosate and investigate the possible memory mechanisms involved in this process. Methods : A population of E. plana was submitted to drought, glyphosate, or a combination of drought plus glyphosate for two generations. The progenies were analyzed for sensitivity to the herbicide glyphosate and for biochemical, metabolites, and molecular responses. Results : When testing sensitivity to glyphosate in the G 2 generations, the DRYxGLY (plants stressed with drought plus glyphosate) was the least sensitive (ED 50 443.0 g a.e. ha -1 ), while CHK (without stresses) was the most sensitive (ED 50 278.0 g a.e. ha -1 ). Evaluating the effects of glyphosate, the CHK population showed a greater accumulation of oxidative damage and shikimic acid. While DRYxGLY had greater antioxidant activity and higher expressions of the EPSPS and ABC-carrier MRP10 genes. Conclusions : Recurrent selection with drought stress and sub-lethal rate of glyphosate (DRYxGLY) showed reduced sensitivity to glyphosate in the second generation ( G 2 ). The conjunction of factors, including the upregulation of EPSPS and the ABC MRP10 transporter, antioxidant enzymes seem to decrease the sensitivity of the DRYxGLY population to glyphosate.


Introduction
Eragrostis plana (Ness), known as South African lovegrass or Tough lovegrass, is the primary invasive weed in over two million hectares of the Pampa Biome (Medeiros, Focht, 2007) in southern Brazil. South African lovegrass is a slow-growing C 4 grass with high seed production and dormancy, allelopathic activity, and tolerance to abiotic stresses (Bastiani et al., 2021).
The most used herbicide to control E. plana in the grassland is glyphosate (N-phosphonomethyl glycine). Glyphosate inhibits the enzyme 5-enolpyruvylshikimate-3-phosphate synthase (EPSPS), reducing the synthesis of aromatic amino acids, which are intermediate compounds in the shikimate pathway (glycolytic and phosphate pathways), proteins, and secondary metabolism (Maroli et al., 2015). The shikimic acid pathway route expends around 30% of the fixed carbon, which is used by plants to synthesize intermediate compounds, auxins, lignins, secondary metabolism, and aromatic amino acids (Maroli et al., 2015). Therefore, interruption of the shikimate pathway by glyphosate triggers several metabolic and physiological processes, leading to plant death (Gaines et al., 2020). In addition to the direct effects of EPSPS inhibition, glyphosate indirectly induces some ROS production, leading to lipid peroxidation, DNA, RNA, and protein oxidation, and slow cell death (Gomes et al., 2017;Maroli et al., 2015).
Despite the efficient action of glyphosate, plants have evolved multiple mechanisms to mitigate the herbicide effects (Gaines et al., 2020). The known resistance mechanisms are classified into target-site resistance (TSR) and non-target-site resistance (NTSR) mechanisms, being the latter the most complex, involving several genes and metabolic routes (Gaines et al., 2020). TSR mechanisms include mutation and upregulation of the target gene (Gaines et al., 2020). The main known NTSR mechanisms for glyphosate are reduction of absorption and translocation of the herbicide, metabolism of the molecule (forming mainly aminomethylphosphonic acid -AMPA), and exclusion of the herbicide (mainly in vacuole) (Gaines et al., 2020).
Abiotic stress and herbicide sub-lethal doses lead to several plant responses, such as NTSR regulation and other pleiotropic effects (Dyer, 2018). These defense responses are similar and interconnected (Burns et al., 2018). Therefore, it is expected that plants adapted to abiotic stresses may have greater tolerance to herbicide stresses and vice-versa (Burns et al., 2018;Fipke et al., 2022). An increase in ROS-detoxifying enzymes promotes acclimation to drought and heat stress in Conyza bonariensis and Abutilon theophrasti, increasing the tolerance to paraquat and glyphosate (Dyer, 2018;Zhou et al., 2007).
As transcriptome, proteome, and plant metabolism changes caused by herbicide stress are similar to abiotic stresses, plants adapted to abiotic stresses are likely more tolerant to herbicides (Yuan et al., 2010). In this context, the study's objectives were to ascertain whether the transgenerational effect of acclimatization to stress due to water deficit and sub-lethal doses of glyphosate may decrease herbicide sensitivity and investigate whether antioxidant enzymes and other resistance mechanisms are involved in this process.

Establishment of the populations
Seeds of E. plana were collected in an experimental field (bulk of seed) in the Embrapa located in Bagé (RS Brazil) and denominated G 0 . In the greenhouse facility at UFPel, the populations G 1 and G 2 were generated in 2017 and 2018 ( Figure 1). In order to build the populations, the initial G 0 population seeds were divided into four groups (each group with eight plants) and exposed to the treatments: untreated Check (CHK -without stress); drought treatment (DRYwater deficit until 35% of stomatal conductance in relation to unstressed plants); glyphosate treatment (GLY -120 g a.e. ha -1 of glyphosate); DRY×GLY (water deficit until 35% of stomatal conductance and after recovery, sprayed with 120 g a.e. ha -1 of glyphosate). The dose of glyphosate (dose to caused 30% to injury, data are not shown) or level of drought were defined in the pre-experiment (data are not shown), aiming at plant stress. The drought stress was performed according to the methodology described by Bastiani et al. (2021). In each treatment group, eight plants were used. As E. plana is probably predominantly allogamous, the plants from each treatment were isolated (in different greenhouse, avoiding any crossing among treatments) and subsequently seeds were collected (in bulk), generating G 1 population. G 1 populations were planted and exposed to the same treatments to obtain G 2 generation.

Dose-response curves with glyphosate
After obtaining all the population's offspring, a growth chamber experiment was conducted, with experimental units consisting of 1 L pots filled with sieved soil (Haplosol) organized in a randomized block design with six replications. Factor A included the different populations G 0 , G 1 (CHK, DRY, GLY, and DRY×GLY) and G 2 (CHK, DRY, GLY, and DRY×GLY), and factor B was of glyphosate doses (0,90,120,180,250,360,540 and 720 a.e. g ha -1 ). All treatments were performed when plants reached an average of eight tillers, using glyphosate (Monsanto, Roundup Original™) applied with a CO 2 sprayer calibrated to deliver 150 L ha -1 of spray solution. At 35 days after application (DAA), the visual injury (%) and shoot dry weight (SDW) ratings were analyzed. The SDW was transformed as the percentage compared to the plants without herbicide (%) of each population and generation. The experiment was repeated twice.

Physiological, biochemical, and profiling analysis of G 2 populations
The experiments were carried out in a growth chamber (temperature 32/28 °C and photoperiod of 12 h). The experiments had three replications and consisted of 8 L Figure 1 -Design of CHK (check), DRY (drought), GLY (glyphosate), and DRY×GLY (drought plus glyphosate) populations in two generations (G 1 and G 2 ) from this G 0 E. plana plants surface. Herbicide extraction was performed according to the methodology described by Gomes et al. (2015).
To quantify glyphosate and metabolites, HPLC-MS/MS system equipped with a mass spectrometer (Quadrupole-Orbitrap, Q-Exactive Focus, Thermo Scientific TM ). For separation, a Hypercarb (5 µm 250 Å, 50 mm x 3 mm) column was used, and as mobile phase: (A) 5 mM of ammonium acetate in water (pH 7.0) and (B) 5 mM of ammonium acetate in methanol (pH 7.0), in gradient condition. Calibration curves were prepared in water employing analytical standards with purity ≥99% (Sigma-Aldrich, St. Louis, MO, USA). Method linear range (R 2 ≥ 0.99) was 1 to 500 µg L -1 for all compounds, and samples were diluted when necessary.

Gene expression analysis
In this study we evaluated the expression of the target genes 5-Enolpyruvylshikimate-3-phosphate synthase gene (EcEPSPS) Aldo-keto reductase (EcAKR), ABC-C Family MRP10 (EcM10), and ABC-C Family MRP8 (EcM11). EPSPS was chosen because it is directly involved in synthesizing the enzyme inhibited by glyphosate (Gaines et al., 2010). The AKR, M10, M11 genes are known to be involved in the process of glyphosate metabolism (Pan et al., 2019;Piasecki et al., 2019a).
The total RNA was extracted from leaves (three biological replicates) using PureLink™ (Plant RNA Reagent-Invitrogen TM , Carlsbas, USA) following the manufacturer's instructions. RNA concentration, quality and integrity were assessed using NanoVue™ (GE Healthcare TM , Buckinghamshire, UK) and agarose gel electrophoresis, respectively. RNA (1 µg) of each sample was treated with DNase I (Invitrogen) according to the manufacturer's recommendations and converted into pots filled with the same soil described in the 2.2 section. Factor A consisted of populations of G 2 generation: CHK and DRY×GLY (more and less sensitive glyphosate, respectively). Factor B: doses of 0 and 360 g a.e. ha -1 of glyphosate. The glyphosate dose used was the one that previously demonstrated plant injury without plant death, allowing the study of the physiological state of responses to the herbicide. The herbicide was applied when plants reached an average of 10 tillers. In experiment 1, plants were collected at 24, 48, 96, and 144 h after glyphosate application (HAA), and in experiment 2 were collected at 48, 96, and 192 HAA. Samples were kept in an ultra-freezer (-80 °C). Leaf samples (second and third fully expanded leaves) were collected and immediately frozen in liquid nitrogen and stored in an ultra-freezer (-80 °C).

Measurements
Stomatal conductance was measured with an LI-1600 steady-state porometer (Li-Cor Biosciences™ -Lincoln, NE USA). All plants were evaluated, taking measures in the adaxial leaf face.

Quantification of glyphosate, AMPA, and aromatic amino acids
The concentrations of glyphosate, AMPA, shikimic acid, phenylalanine, and tyrosine were determined in duplicate using High-Performance Liquid Chromatography (HPLC) coupled with Tandem Mass Spectrometry (MS/MS). First, the treated leaves were immersed in a 10-mL wash solution (water in pH 2.5 with phosphoric acid 6 M) and the unabsorbed glyphosate was removed from the leaf In G 0 plants (Figure 2A), the dose of glyphosate required to cause 50% plant injury (ED 50 ) was 173 g a.e. ha -1 (Table 2). Regarding G 1 and G 2 generation, CHK populations showed cDNA using oligo(dT) and the SuperScript™ III First-Strand Synthesis System kit (Invitrogen).
RT-qPCR experiments were performed in LightCycler™ 480 thermocycler with three biological and three technical replicates using oligonucleotides for the target and reference genes (Table 1), according to the MIQE Guidelines (Bustin et al., 2009). The amplification efficiency and specificity of each oligonucleotide were determined in validation experiments using four dilutions of cDNA. The reactions were performed following the protocol described by Fipke et al. (2022). Gene expression was calculated following the 2^(-delta-delta CT) method (Livak, Schmittgen, 2001), using CHK population without glyphosate (0 g a.e. ha -1 ) at each time of collection as a baseline and normalized with OsACT1, Os18S, and OsEF1α reference genes.

Statistics analysis
Data were tested for homogeneity of variance and normality. Visual injury and SDW data were fitted with a nonlinear sigmoid (y = a/(1+exp(-(x-ED 50 )/b))) and  an increased dose of glyphosate compared to G 0 , with ED 50 229 and 278 g a.e. ha -1 , respectively. In G 1 there were no differences in the ED 50 of the DRY, GLY, and DRY×GLY populations compared to the CHK population. When analyzing the G 2 generation, the DRY×GLY population presented the highest ED 50 , 443 g a.e. ha -1 , among populations. In G 2 generation, the sensitivity reduction rate (Table 2) showed 1.59 in the DRY×GLY population.
In the dose of glyphosate required to cause 50% of shoot dry weight reduction (GR 50 ), there were no differences between the CHK populations of the G 0 , G 1 , and G 2 generations, with 122, 199, and 152 g a.e. ha -1 , respectively. In G 1 generations, there were no differences in GR 50 between populations. However, in the G 2 generation, the DRY, GLY, and DRY×GLY populations presented higher GR 50 (284, 308, and 436 g a.e. ha -1 , respectively) than the CHK population. In G 2 generation, the sensitivity reduction rate demonstrated 1.85, 2.02, and 2.85 for DRY, GLY, and DRY×GLY populations, respectively.
These results indicated that the offspring of plants exposed to water stress decreased sensitivity to glyphosate, mainly in G 2 . Plants from the DRY×GLY population, which received both stresses (drought followed by a sub-lethal dose of glyphosate), were less sensitive to glyphosate. It is well known that plants exposed to previous stresses may have improved tolerance to other stresses, a phenomenon related to memory mechanisms in plants (Galviz et al., 2020). For instance, in another study, E. plana submitted to drought, quizalofop, and drought plus quizalofop had a lesser sensitivity to quizalofop than non-stressed plants (Fipke et al., 2022). Consequently, the current study also sought to determine the possible biochemical mechanisms involved in priming for glyphosate reduced sensitivity in E. plana.

Stomatal conductance
At 24 and 48 HAA ( Figures 3A and 3B), all plants treated with glyphosate had lower stomatal conductance  Figures 3A and 3B) also presented reduced stomatal conductance compared with untreated plants. In our study, the reduction in stomatal conductance in all populations was a typical response to glyphosate (Figure 3). However, this response is not necessarily involved in harmful effects caused by the herbicide in plants. Lolium perenne resistant to glyphosate had a stomatal closure regulation; however, it was linked to carbon fixation optimization according to assimilated demands (Yanniccari et al., 2012). Stomatal conductance can be used as a biomarker in plants sensitive to glyphosate (Yanniccari et al., 2012).

H 2 O 2 content and lipid peroxidation
Glyphosate-treated CHK population produced 82 and 80% higher hydrogen peroxide than untreated plants at 96 and 144 HAA (Figure 4a). At 144 HAA, the glyphosate-treated DRY×GLY population (Figure 4b) also showed higher hydrogen peroxide than untreated plants. Depending on the efficiency of redox homeostasis, it can be related to apparent antagonistic functions such as, on the one hand, signaling to prevent stressful situations and, on the other hand, causing oxidative stresses (Moretti et al., 2017).
Inhibition of EPSPS by glyphosate indirectly induces ROS production (Maroli et al., 2015). This unregulated accumulation of H 2 O 2 and other ROS causes lipid peroxidation and loss of membrane integrity, and, ultimately, cell death (Foyer, Noctor, 2005). The glyphosate-treated CHK population showed the highest lipid peroxidation levels (Figure 4c), with an increase of 49,75,and 70% (48,96,and 144 HAA,respectively). On the other hand, primed DRY×GLY plants (Figure 4d) had no ROS increases in the evaluated periods.
Accumulation of H 2 O 2 and the concomitant increase of lipid peroxidation (Figure 4) could explain the susceptibility of the CHK population to glyphosate. At 96 and 144 HAA, the high production of H 2 O 2 likely triggered an increase in lipid peroxidation. At 48 HAA, although H 2 O 2 apparently did not induce lipid peroxidation, it could be reached by other types of ROS responsible for this process. In C. bonariensis, susceptible plants showed 5.2 and 3.2-fold higher in H 2 O 2 and lipid peroxidation than untreated plants (Piasecki et al., 2019a). Salix miyabeana plants subjected to glyphosate had an increase in the production of H 2 O 2 after 6 HAA, reaching a peak at 72 HAA. Likely, from 6 HAA, a gradual increase in lipid peroxidation has occurred (Gomes et al., 2017).

Antioxidant enzymes
SOD activity differed only at 24 and 96 HAA in the glyphosate-treated DRY×GLY population compared to the untreated plants (Figure 5b). In offspring of plants DRY×GLY population, it was observed to have a higher baseline of SOD activity than untreated CHK population. All plants presented increased SOD activity following glyphosate application, with 16 and 56% for CHK and DRY×GLY, respectively, demonstrating that plants submitted to stresses in previous generations abet stimulated SOD activity. SOD is a crucial enzyme in regulating oxidative stress. For instance, C. bonariensis GLY resistant had 1.6-fold SOD activity than the susceptible biotype (Piasecki et al., 2019a).
CHK population exhibited differences in CAT activity at 24, 48, and 144 HAA (Figure 5c), showing increases of 81, 60, and 49% in glyphosate-treated plants than the untreated. However, this increase in CAT activity in the CHK population was like or less than the glyphosatetreated DRY×GLY population. In 144 HAA (Figure 5d), the glyphosate-treated DRY×GLY population presented the highest CAT activity. APX activity (Figure 5f) exhibited differences for the DRY×GLY population (24 and 96 HAA).
CHK population did not differ from APX activity with or without quizalofop-treated (Figure 5e). SOD, CAT, and APX enzymes play an essential role in ROS detoxification. In C. bonariensis, a glyphosateresistant biotype demonstrated higher CAT and APX activity than a susceptible biotype when the herbicide was applied (Piasecki et al., 2019a). Application of glyphosate in S. miyabeana stimulated ROS production and, consequently, increased enzymatic activities of CAT and APX (Gomes et al., 2017). Antioxidant enzymes SOD, CAT, and APX are responsible for reducing oxidative stress and complement important resistance mechanisms enabling plants to tolerate herbicide glyphosate (Maroli et al., 2015).

Glyphosate, AMPA, shikimic acid, and amino acids concentration in plants
Glyphosate concentration in CHK and DRYxGLY populations (Figure 6a) reveals that in the CHK population, there was 72% higher glyphosate concentration at 48 HAA compared to the DRYxGLY population. There were no differences in glyphosate concentration regarding sampling times in CHK and DRYxGLY populations. In CHK and DRYxGLY populations, glyphosate concentration remained similar in the leaf tissues until 144 HAA, indicating low translocation and/or metabolization in the leaf tissues.
Glyphosate degradation in plants generates metabolites AMPA and glyoxylate (Rojano-Delgado et al., 2012). Herein, AMPA metabolite was found in both populations CHK and DRYxGLY (Figure 6b). In the CHK population, AMPA was detected only at 24 and 48 HAA, with a concentration of 4.4 and 2.5 µg g -1 . In the DRYxGLY population, AMPA was found in all evaluation periods, with the highest concentration at 24 HAA and the lowest at 144 HAA (3.1 and 0.8 µg g -1 , respectively).
These results indicated that there was detoxification of glyphosate forming AMPA and that it was present in both more sensitive and less sensitive populations to glyphosate (CHK and DRYxGLY, respectively). However, AMPA levels represent 13.6 and 4.9% of glyphosate concentration at 24 and 48 HAA in the CHK population, while in the DRYxGLY population, AMPA levels were equivalent to 10.2, 8.8, 6.9, and 1.6% of the glyphosate concentration at 24, 48, 96 and 144 HAA, respectively.
EPSPS inhibition by glyphosate leads to rapid accumulation of shikimic acid and reduced synthesis of aromatic amino acids (Rojano-Delgado et al., 2012). All glyphosate-treated plants had a greater accumulation of shikimic acid than plants without glyphosate (Figure 7a). At all evaluated periods, the CHK population showed a higher concentration of shikimic acid than DRYxGLY populations, exhibiting an increase of 291, 97, 35, and 195% at 24, 48, 96, and 144 HAA, respectively. These results explain the lower sensitivity of DRYxGLY compared to CHK population. Comparing both populations, when the plants were not treated with glyphosate, there were no differences in shikimic acid concentration.
In the glyphosate-treated CHK population, an increase in shikimic acid was observed after 24 HAA and later remained constant at 48 and 144 HAA. At 96 HAA, a reduction in shikimic acid was detected, consistent with the lower glyphosate concentration (Figure 6a) in the CHK population. In the glyphosate-treated DRYxGLY population, shikimic acid accumulation initiates at 24 HAA, reaching the highest concentration at 48 and 96 HAA and reducing to 144 HAA. Plants with faster and higher shikimic acid accumulation over time demonstrate substantial inhibition of EPSPS and, therefore, greater sensitivity to glyphosate (Tani et al., 2015). Conyza canadensis plants susceptible to glyphosate showed higher shikimic acid accumulation than resistant plants (Tani et al., 2015).
Tyrosine and phenylalanine are aromatic amino acids from the shikimic acid pathway (Fernández-Escalada et al., 2017). Between CHK and DRYxGLY, there was a difference in L-tyrosine concentration only for untreated plants (Figure 7b). In glyphosate-treated CHK population had increased L-tyrosine concentration compared to untreated plants at 48 and 96 HAA. However, at 144 HAA, there was  an inversion, and glyphosate-treated plants presented a reduction of 11%. Regarding the DRYxGLY population, contrasts were found at 24 and 96 HAA, with a difference of -25% and 16% in glyphosate-treated plants compared to untreated plants.
The L-phenylalanine concentration (Figure 7c) varied between populations at different times and glyphosate application. In the glyphosate-treated CHK population, an increased L-phenylalanine concentration was observed from the initial evaluation at 24 HAA until 96 HAA and a reduction at 144 HAA ( Figure 7C). From the glyphosatetreated DRYxGLY population, differences of -9, -42, -18, and 92% occurred compared to untreated plants (24,48,96,and 144 HAA,respectively).
Amino acid concentrations usually vary among plants; however, the increase in phenylalanine and tyrosine concentrations may be due to increased protein turnover following glyphosate treatment (Fernández-Escalada et al., 2017). In Amaranthus palmeri, susceptible to glyphosate, an increase in phenylalanine and tyrosine concentration at 72 HAA was reported. In contrast, in an A. palmeri population resistant to glyphosate, differences in concentration of these amino acids were not detected (Fernández-Escalada et al., 2017). Also, Lolium multiflorum plants susceptible to glyphosate showed a higher concentration of tyrosine than plants resistant to 72 HAA, while no changes in phenylalanine concentrations were observed (Barroso et al., 2018).

Gene expression
In this study, EPSPS expression was relative to the untreated CHK population ( Figure 8A). At 48 HAA, EPSPS was downregulated in glyphosate-treated CHK and DRYxGLY populations. However, EPSPS was upregulated relative to the untreated CHK population in all other treatments and evaluation periods. At 96 HAA, the glyphosate-treated DRYxGLY population demonstrated the highest EPSPS expression, with 4-fold more transcripts than the untreated CHK population and 1.92-fold more than the untreated DRYxGLY population. At 192 HAA, glyphosatetreated CHK and untreated DRYxGLY populations showed the higher EPSPS expressions, with 2.3-fold and 2.6-fold, respectively, compared to the untreated CHK population.
A mechanism of resistance to glyphosate that has been widely studied is the upregulation of the EPSPS gene. Increased copies of the EPSPS gene allows for a higher synthesis of the EPSPS enzyme, as there is more enzyme than glyphosate molecules for inhibition, making the plant resistant (Gaines et al., 2010). Plants C. bonariensis and C. canadensis, both glyphosate-resistant, EPSPS expression was observed two times compared to susceptible plants. However, the authors report other NTSR mechanisms involved (Dinelli et al., 2008). In A. palmeri, it was found that increased expression of EPSPS was responsible for the resistance (Gaines et al., 2010). The increase in EPSPS expression was explained by the amplification of genes, where several copies of the gene are present in the genome (Gaines et al., 2010).
The AKR gene at 48 HAA was downregulated in all populations compared to the CHK population without herbicide ( Figure 8B). The CHK population with herbicide did not differ in expression levels at any time. In contrast, the DRYxGLY population showed a relative increase of 1.1 and 0.6-fold at 96 HAA and 2.6 and 1.6-fold at 192 HAA compared to the CHK population without herbicide. DRYxGLY population had no difference in glyphosate application at any evaluation time. Aldo-keto reductase is a superfamily of enzymes responsible for catalyzing the reduction of NAD(P) H-dependent aldehydes and ketones (Pan et al., 2019). However, this superfamily is also involved in the metabolism of xenobiotics, secondary metabolism, and protection of osmolytes. It was recently reported that EcAKR4-1 is involved in the metabolization of glyphosate (Pan et al., 2019). However, it is reported that Echinochloa colona had resistance to glyphosate by Pro-106-Thr EPSPS target-site mutation (McElroy, Hall, 2020).
These data suggest that the DRYxGLY population that has undergone two generations of drought stress and a sublethal dose of glyphosate was able to positively regulate AKR gene expression compared to the CHK population. A previous report demonstrates that E. colona could detoxify glyphosate, forming AMPA, by the AKR enzyme. The EcAKR4-1 gene was upregulated in this species (Pan et al., 2019).
ABC-transporter MRP8 (M11) was differentially expressed among treatments (Fig 8C)   ABC-transporter MRP10 (M10) was upregulated in almost all populations and times analyzed ( Figure 8D). In the glyphosate-treated CHK population, M10 was upregulated in all periods analyzed, with 4.8-fold, 1.4fold, and 2.8-fold at 48, 96, and 192 HAA, respectively. M10 was upregulated in the glyphosate-treated DRYxGLY population at 48 and 96 HAA and downregulation at 192 HAA. The higher difference between populations was detected at 96 HAA, where glyphosate-treated DRYxGLY showed the highest expression of M10, 6.3-fold, thus being upregulated in this period. The untreated DRYxGLY population also showed upregulation of the M10 gene, demonstrating a greater baseline expression than the untreated CHK population.
M10 and M11 genes are ABC-transporters located in the tonoplast (Moretti et al., 2017). These transporters are related to weed resistance to glyphosate, acting on the active transport of this molecule to the vacuole (Moretti et al., 2017;Piasecki et al., 2019b). In C. canadensis and C. bonariensis, the ABC-transporters M10 and M11 are involved in resistance to glyphosate (Piasecki et al., 2019b;Tani et al., 2015). In resistant C. canadensis, an upregulation of M10 and M11 was reported upon glyphosate treatment (Nol et al., 2012). The ABC-transporter M10 was 2.3-fold overexpressed in the resistant C. canadensis than the sensitive biotype (Tani et al., 2015). The authors suggest that the ABC-transporters resistance in glyphosate-treated plants is mainly due to the exclusion of glyphosate in the vacuole and reduced herbicide translocation to the other plant tissues (Nol et al., 2012;Tani et al., 2015).

Enhanced mechanisms of response to glyphosate in E. plana
The ability of plants to adapt to environmental stresses allows regulating their metabolism to increase their tolerance (Délye et al., 2013). Mechanisms of resistance to herbicides that do not involve a site of action are the same as those used by plants in responding to abiotic stresses (Délye et al., 2013;Fipke et al., 2022). According to our results, E. plana G 1 and G 2 generations, when exposed to combined stress (drought plus a sub-lethal dose of glyphosate), allowed a decrease in sensitivity compared with plants that did not undergo any stress. This transgenerational effect can be caused by epigenetic processes, allowing differentiated regulation of the expression of stress tolerance mechanisms, such as antioxidant machinery (Galviz et al., 2020). However, there is a need for further studies to explain the epigenetic role in this observed effect.
In addition to the antioxidant processes presented, it is suggested that the reduced sensitivity of the DRYxGLY population to glyphosate is due to multiple mechanisms. It is important to note that a high accumulation of shikimic acid was detected in both populations. However, the DRYxGLY population had a lower shikimate acid concentration than the CHK population in all evaluation periods (Figure 7a). The increased expression of EPSPS in DRYxGLY at 96 HAA ( Figure 8a) may be one factor in reducing shikimic acid concentration at 144 HAA (Figure 7a). The tyrosine and phenylalanine amino acids pool showed contrasting responses (Figures 7b and 7c).
Regarding the involved NTSR mechanisms, it was found that the concentration of glyphosate in the leaf tissue remained constant over time, which may demonstrate a reduction in translocation to other plant tissues ( Figure 6A). Also, AMPA was detected due to glyphosate detoxification ( Figure 6B), together with the increase in AKR expression ( Figure 6B). However, AMPA concentration was relatively low to explain the difference in sensitivity to glyphosate. Also, an increased expression of M10 in DRYxGLY at 96 HAA ( Figure 8D) may help exclude glyphosate in the vacuole and, consequently, reduce its translocation to other plant tissues.

Conclusions
Eragrostis plana exposed to combined stress (drought followed by a sub-lethal dose of glyphosate) showed induced transgenerational memory allowing acclimation to stressful conditions and decreasing sensitivity to glyphosate. The increased activity of the enzimes SOD, CAT, and APX following homeostatic adjustment in the DRYxGLY population, relieves glyphosate's oxidative stress. Additionally, the set of mechanisms such as the possible metabolization of glyphosate in AMPA and the upregulation of AKR, and the upregulation of MRP10 (exclusion of glyphosate in the vacuole) seem to be involved in decreasing glyphosate sensitivity in Eragrostis plana DRYxGLY population.