Recurrent Selection by Herbicide Sublethal Dose and Drought Stress Results in Rapid Reduction of Herbicide Sensitivity in Junglerice

Echinochloa colona (junglerice) is a problematic global weed for many crops, primarily controlled with herbicides. Drought stress alters the overall plant physiology and reduces herbicide efficacy. This research aimed to study the joint effect of drought stress (DS) and recurrent selection with sublethal dose of herbicide on adaptive gene expression and herbicide efficacy on E. colona. Three factors were evaluated: (A) E. colona generation (G0, original population from susceptible standard; G1 and G2, progenies of recurrent selection); (B) herbicide treatment (florpyrauxifen-benzyl, 0.25×; glyphosate, 0.125×; quinclorac, 0.125× the recommended dose; and nontreated check); (C) DS (50% and 100% field capacity). Recurrent exposure to sublethal herbicide dose, combined with drought stress, favors the selection of plants less susceptible to the herbicide. Upregulation of defense (antioxidant) genes (APX: ascorbate peroxidase), herbicide detoxification genes (CYP450 family: cytochrome P450), stress acclimation genes (HSP: heat-shock protein, TPP: trehalose phosphate phosphatase, and TPS: trehalose phosphate synthase), and genes related to herbicide conjugation (UGT: UDP glucosyltransferase) in the G2 population was significant. Recurrent exposure to sublethal herbicide dose under drought stress reduces junglerice sensitivity to herbicide, seemingly due to “imprinted” upregulation of metabolic and protection genes in response to these stresses.


Introduction
Attaining global food security entails increasing food production without increasing crop production area. This is a seemingly an insurmountable challenge considering our changing climate and steady population growth [1][2][3]. Beyond having improved crop varieties, weeds remain the primary biological constraint in irrigated and non-irrigated crop production areas [4].
Echinochloa colona (L.) Link, commonly known as junglerice, a self-pollinating species, is among the top five worst weeds in the world, being a persistent threat in 35 cropping systems in more than 60 countries [5][6][7][8]. To prevent yield losses, weed control has been almost exclusively done with herbicides [9][10][11]. However, the extensive use of herbicides, repeated application of the same herbicide or of herbicides with the same mechanism of action (MOA), and repeated exposure to inadvertent

Plant Material
This study was conducted using seeds of susceptible E. colona (referred to hereafter as G0), collected in 2011 from a field in Prairie County, Arkansas, USA. G0 was then exposed to three successive cycles of recurrent selection with sublethal doses of herbicides under well-watered or drought stress conditions to produce G1 and G2. This procedure is described in the next section.

General Procedure for Population Feneration
Seeds (G0) were planted into 50-cell trays containing commercial potting soil (Sun Gro Horticulture Canada Ltd., Vancouver, Canada). At the one-leaf stage, seedlings were transplanted into square pots (7.6 cm wide, 10.2 cm tall) containing a 1:3 mixture by volume of commercial potting soil and field soil (Captina silt loam-fine-silty, siliceous, active, mesic typic Fragiudults). The experiment was performed in a randomized complete design with six replications in each cycle. The experimental unit was one pot containing one plant. The experiment with populations G1 and G2 followed the same methodology described for G0 ( Figure 1).

Figure 1. Schematic diagram of the progression of experiments on
Echinochloa colona submitted to selection by low herbicide rates and water regimes starting with the parental population (G0) and selected (G1 and G2) progenies. The letters WW and DS indicate well-watered and drought stress treatments, respectively. Herbicide treatments were: nontreated (no herbicide), florpyrauxifen-benzyl at 0.25×, glyphosate at 0.125×, and quinclorac at 0.125× the recommended dose. Time t0 (prior to herbicide application) and t1 (12 h after herbicide application) were the timings of leaf tissue collection for RNA extraction and subsequent gene expression.
The experiment was conducted in the greenhouse at the University of Arkansas, Fayetteville, USA with 14-h photoperiod under a day/night temperature regime of 30 • C/21 • C. The pots were placed in trays and sub-irrigated until the seedlings reached the 2-3-leaf stage for treatment. The plants were then submitted to two water regimes: well-watered, which corresponds to 100% of the total water holding capacity of the pot (Cw), and drought stress, which corresponds to 50% of Cw in the respective treatment for 7 d, when the plants were sprayed with the respective herbicide treatments (Table 1). a Weed Science Society of America Mechanism of Action (MOA) classification system; b Application rates were a fraction of the recommended dose: florpyrauxifen-benzyl at 0.25×; glyphosate and quinclorac at 0.125×. Doses were based on preliminary experiments to allow survival and seed production (data not shown).
Cw was determined using the fresh mass of the soil after water saturation (Cfm) and the dry mass (Cdm) after soil drying for 24 h at 105 • C and used in the equation Cw = (Cfm − Cdm)/Cfm × 100 as described by Santos et al. [48]. At the beginning of water treatment, all pots were saturated with water, drained, and weighed. The pots were weighed every day to determine the volume of water lost by evapotranspiration, which was then replenished, to maintain Cw at 50% and 100% and assuming 1 mL = 1 g.
The water-stress treatment continued for seven more days (total of 14 days under drought), then all plants were flooded. Preliminary dose-response experiments with and without drought stress were conducted to determine the sublethal dose that would allow plants to survive and produce seeds. The herbicides were applied in a spray chamber, at a pressure of 221 kPa and a volume of 187 L·ha −1 . Watering by sub-irrigation was resumed after 24 h. Three weeks after herbicide application, the herbicide effect was evaluated visually. Injury was evaluated visually on a scale of 0% (no symptoms) to 100% (dead).
Prior to flowering, the plants were separated spatially to prevent cross-pollination. At maturity, seeds were harvested and bulked for each treatment. During each selection cycle, a separate group of six plants, without herbicide treatment, was grown to produce generations of plants exposed only to well-watered or drought stress conditions. This procedure was repeated over three cycles.

Determination of Sensitivity Level to Herbicides
The original population (G0) and selected progenies (G1 and G2) were subjected to their respective water regime assignment (100% and 50% of Cw) for 7 days, as implemented in the previous experiments, and sprayed with a range of herbicide rates when the plants were at the 2-3-leaf stage. The herbicide rates were 0×, 0.0625×, 0.125×, 0.25×, 0.5×, 0.75×, 1.0×, 1.5×, 2.0×, and 4.0× the recommended rate. The herbicide treatments were applied as described previously. After herbicide application, the plants were returned to the greenhouse and the water regimes maintained for seven more days. At 3 weeks after herbicide application, the herbicide effect was evaluated visually. Dose-response data were analyzed using the drc package in R v. 3.1.2 [49,50]. The three-parameter log-logistic model in Equation (1) was used.
where Y is the response (% control), expressed as a percentage of the nontreated check, d is the asymptotic value of Y at the upper limit, b is the slope of the curve around e (ED 50 : the herbicide rate giving response halfway between d and the lower asymptotic limit, which was set to 0), and x is the herbicide rate.

Plant Material
The control sample was composed of leaf tissues of G0 plants, well-watered and without herbicide treatment, cultivated for three cycles, at the same period as the other plants, to produce G2. In other words, the control samples were G0 plants that were cultured at 100% of Cw without herbicide treatment. For all treatments, leaf tissues were collected before herbicide application (t0) and 12 h after herbicide application (t1). The tissues were flash-frozen in liquid nitrogen and stored at −80 • C until processed.

RNA Extraction and Complementary DNA (cDNA) Synthesis
Total RNA was extracted from 2 g of shoot tissue using the reagent PureLinK TM (Plant RNA Reagent-Invitrogen TM , Carlsbad, CA, USA), following the manufacturer's recommendations. The quantity and quality were assessed by agarose gel electrophoresis (1% w/v). The amount and purity of RNA were determined using a NanoDrop™ 2000 spectrophotometer (Thermo Scientific, Waltham, MA, USA). The RNA (2 µg) was treated with DNase, and cDNA was obtained using the RevertAid First Strand cDNA Synthesis Kit (Thermo Scientific, Waltham, MA, USA) using oligo(dT) according to the manufacturer's instructions.

qRT-PCR Assay
The qRT-PCR was conducted to determine changes in expression of NTSR candidate genes relative to the reference gene. The qRT-PCR was performed following MIQE guidelines [51], using specific oligonucleotides for target genes and ACT1, UBQ5, and EF1-α housekeeping genes ( Table 2). The qRT-PCR experiments were conducted in a total volume of 10 µL containing 5.0 µL of iTaq™ Universal SYBR ® Green Supermix (Bio-Rad, Hercules, CA, USA), 0.5 µL of forward primer (10 mM), 0.5 µL of reverse primer (10 mM), 1 µL of cDNA (1:5 dilution), and 3.0 µL of water. Two biological replicates and two technical replicates were used for each treatment and primer pair. The amplification efficiency was determined for each primer pair and melt-curve analyses were performed. The cycle conditions were as follows: one cycle of 95 • C for 5 min, followed by 39 cycles of denaturation at 95 • C for 15 s, annealing at 60 • C for 1 min, and extension at 72 • C for 30 s, and a final dissociation curve at 95 • C for 5 s, followed by cooling to 70 • C for 1 min and gradual heating at 0.11 • C steps to 95 • C and cooling to 40 • C for 30 s. For each gene analyzed, UBQ5 was used as an endogenous control to quantify cDNA abundance, after the stability analysis of the expression data using DataAssist ™ v3.0 software (Applied biosystems, Life Technologies, Carlsbad, CA, USA).
Cycle threshold (Ct) was obtained during the reaction cycles, and relative gene expression values were calculated using the 2 −∆∆Ct method [57]. Gene expression data were analyzed using the Multi Experiment Viewer (TIGR MeV) v3.0 software [58] and presented as a heat map diagram using the harvest stage as a baseline. The messenger RNA (mRNA) abundance of each gene from nontreated G0 population served as the baseline for determining relative RNA levels in each treatment from the G2 population.

Sensitivity to Herbicides
The sensitivity of E. colona to herbicides was generally reduced under drought stress compared to the well-watered condition, after three cycles of selection with sublethal herbicide dose under drought (Table 3). Under well-watered conditions, the ED 50 increased minimally in general between G0 and G2 with all herbicides tested. Specifically, the ED 50 increased 0.12 points with florpyrauxifen-benzyl ( 50 ), and tolerance index (TI)) for the control of Echinochloa colona with various herbicides across cycles of exposure to drought stress and sublethal herbicide dose, 3 weeks after herbicide treatment. The data were fitted with a three-parameter log-logistic regression model in Equation (1). Under drought stress (50% of Cw), the ED 50 increased significantly between G0 and G2 with all herbicides (Table 3). These increases were orders of magnitude higher compared to the ED 50 increases under well-watered conditions. The ED 50  The magnitude of increase in tolerance was 6-25 times higher under drought stress compared to well-watered conditions.

Log-Logistic Regression Estimates
These data indicate that recurrent exposure to sublethal herbicide dose can accelerate sensitivity reduction of E. colona to some herbicides, even without added abiotic stress. In this case, E. colona became less susceptible to glyphosate and quinclorac after three cycles of recurrent selection. Rapid reductions in herbicide sensitivity across generations occurred under drought stress in all herbicides tested.

Differential Gene Expression
In order to understand the reduction in sensitivity to herbicides between generations submitted to a single stress factor (drought or herbicide) or a combination of these two stress factors, the differential gene expression response was studied for genes related to antioxidant defense system (APX: ascorbate peroxidase), herbicide detoxification (CYP450 family: cytochrome P450), stress acclimation (HSP family: heat-shock protein, TPP: trehalose phosphate phosphatase, and TPS: trehalose phosphate synthase), and a gene related to herbicide molecule conjugation (UGT: UDP glucosyltransferase). E. colona plants (G2) exposed to drought stress over three cycles showed increased expression of TPP (0.1-fold), CYP709B1 and CYP72A14 (0.2-fold), APX (1.2-fold), UGT (1.6-fold), HSP10 (1.7-fold), and TPS (1.9-fold), with the highest relative expression observed for HSP15 (2.4-fold) compared to G0 plants (Figure 3).  [58], and values in parentheses represent 95% confidence intervals. The mRNA abundance of each gene from G0 population (original susceptible standard) served as the baseline for determining relative RNA levels. The letters WW and DS indicate the water regimes of well-watered and drought stress, respectively; nontreated = G2 without herbicide; t0 = before and t1 = 12 h after herbicide application to G2 plants. The color scale above the heatmap shows the expression level, where red indicates high transcript abundance while green indicates low abundance.

Gene Expression Profile with Respect to Florpyrauxifen-Benzyl Treatment
The gene expression analysis of G1 and G2 populations in water regimes associated with the florpyrauxifen-benzyl herbicide were compared to G0, showing induced transcriptional activation of the analyzed genes in most cases. In well-watered conditions, all genes studied, except for APX and CYP72A15, were upregulated 12 h after florpyrauxifen-benzyl application (Figure 3). In drought stress conditions, all analyzed genes were upregulated after florpyrauxifen-benzyl treatment.
For the APX gene, transcript accumulation was similar before herbicide application (t0) in well-watered (0.2-fold) and in drought stress conditions (0.3-fold). However, when subjected to drought stress, APX expression was induced 12 h after florpyrauxifen-benzyl treatment to 1.8-fold under drought stress t1, but repressed −0.2-fold in well-watered t1 ( Figure 3).
CYP709B1 had a similar gene expression profile to APX. The application of florpyrauxifen-benzyl to well-watered plants induced CYP709B1 expression from 0.5-fold before herbicide treatment to 1.2-fold 12 h after treatment. Under drought stress, the expression of CYP709B1 was −0.2-fold before herbicide application and 2.4-fold after herbicide application ( Figure 3). The expression of CYP709B2 increased from −3.9-fold at t0 to 1.2-fold at t1 in well-watered plants treated with herbicide and from 2.0-to 2.1-fold after herbicide treatment of drought-stressed plants. The expression of CYP72A14 gene, which belongs to Clan 72 of the cytochrome P450 family, was also induced after the application of florpyrauxifen-benzyl, mainly in drought-stressed plants, increasing from 0.6-fold at t0 to 4.2-fold at t1 in drought stress alone, and drought stress plus herbicide treatment. However, as previously mentioned, the CYP72A15 gene was repressed after herbicide treatment of well-watered plants (1.3-fold in t0 to 1.0-fold in t1). Likewise, as observed with the other cytochrome P450 members studied, the CYP72A15 gene was induced upon florpyrauxifen-benzyl treatment in drought stress conditions (−0.1-fold at t0 to 2.2-fold at t1).
The expression of HSP10, HSP15, TPP, and UGT genes increased 12 h after florpyrauxifen-benzyl treatment, mainly in drought stress conditions (Figure 3). The expression of TPS increased minimally after herbicide treatment; even so, it was induced from 0.6-fold at t0 to 0.9-fold at t1 in well-watered plants and downregulated in drought-stressed plants (−1.6-fold to −1.5-fold).

Gene Expression Profile with Respect to Glyphosate Treatment
The transcript accumulation of the studied genes increased 12 h after glyphosate treatment in drought-stressed plants (Figure 3). In well-watered plants, all genes were induced by glyphosate, except for APX, CYP72A15, and TPS, which were repressed. While the APX expression was downregulated in well-watered plants treated with glyphosate (3.3-to 1.2-fold), gene expression increased from 1.3-fold at t0 to 3.0-fold at t1 in drought-stressed plants (Figure 3). After glyphosate treatment, the expression of CYP709B1, CYP709B2, CYP72A14, HSP10, and HSP15 increased, mainly in drought-stressed plants, reaching values of 5.2-, 6.9-, 5.2-, 7.0-, and 10.0-fold, respectively, being 1.1 to 3.3 points higher than glyphosate-treated plants in well-watered conditions.
CYP72A15 and TPS genes were downregulated after glyphosate treatment in well-watered plants, showing transcript accumulation from 2.1-to 1.2-and from 2.1-to −2.8-fold, respectively ( Figure 3). However, in drought-stressed plants, the expression of these genes ranged from 1.6-to 4.1-and from 2.2-to 3.3-fold, respectively, when treated with glyphosate.
TPP and UGT genes were upregulated after glyphosate treatment, and the transcript accumulation in drought-stressed plants was very close to that in well-watered plants sprayed with glyphosate ( Figure 3). TPP and UGT reached the same expression levels in well-watered and drought-stressed plants, 12 h after the glyphosate application (3.3-and 5.2-fold, respectively).

Gene Expression Profile with Respect to Quinclorac Treatment
The expression of genes studied changed in drought-stressed plants 12 h after quinclorac treatment ( Figure 3). APX, CYP709B2, CYP72A14, and CYP72A15 were repressed after quinclorac treatment of well-watered plants. On the other hand, APX, CYP72A14, and CYP72A15 were induced in drought-stressed plants treated with quinclorac, reaching values of 1.2-, 4.4-, and 0.8-fold, respectively. CYP709B2 remained repressed by quinclorac even in drought-stressed plants, showing a −1.9-fold change.
The expression profile of CYP709B1 and UGT genes was similar, whereby these genes were induced after quinclorac treatment; however, the induction was higher in well-watered plants (Figure 3). CYP709B1 increased from 0.7-fold at t0 to 1.6-fold after quinclorac application in well-watered plants and from −2.4to -0.7-fold in drought-stressed plants. Likewise, UGT transcripts increased from 1.4-fold at t0 to 2.7-fold after quinclorac application in well-watered plants and from −0.4 to 1.6-fold in drought-stressed plants.
Genes coding for heat-shock proteins (HSP10 and HSP15) were induced after quinclorac application ( Figure 3). The increased expression was detected in both water regimes, but with higher values in drought-stressed plants, reaching 5.0-and 7.3-fold, respectively, for HSP10 and HSP15. Quinclorac application also induced the TPP gene, which showed expression averages of 2.3-and 1.3-fold in well-watered and drought-stressed plants, respectively.
TPS was induced in well-watered and herbicide-treated plants and was repressed in plants subjected to quinclorac plus drought stress treatment (Figure 3). This indicates a possible deviation from the normal metabolic pathway of trehalose biosynthesis to provide substrates for UGT, displaying a role in conjugation reactions.

Drought Stress Effect on E. colona Tolerance to Herbicides
Recurrent selection with low doses of herbicides was studied to understand the mechanisms of resistance evolution and to propose sustainable management strategies. The increased tolerance of G2 plants to florpyrauxifen-benzyl, glyphosate, and quinclorac compared to G0 plants supports our hypothesis. In this study, recurrent selection with sublethal herbicide dose under drought stress selected E. colona plants with greater adaptability to the submitted conditions, with a transgenerational effect, resulting in reduced sensitivity to chemical control. The change was significant after three cycles of selection, thus highlighting the importance of adhering to best management practices pertaining to herbicide application. In reality, several mitigating factors prevent producers from applying herbicides under optimum conditions all the time. The timing of herbicide application is not always ideal. Therefore, the next line of defense is preventing seed production of plants that survive herbicide application in the field to break the selection cycle. It is important to sustain the efficacy of herbicides currently available because the prospect of discovering new herbicidal chemistries with a different mechanism of action or not yet compromised by evolved weed resistance is low [31,35,37,59,60].
Although the increase in tolerance between G0 and G2 was significant in all cases, except with florpyrauxifen-benzyl under well-watered conditions, the tolerance level of G2 E. colona to florpyrauxifen-benzyl, glyphosate, and quinclorac was still below the recommended rates of each herbicide (Figure 2 and Table 3). For practical purposes, G2 E. colona would still be classified as susceptible. Nevertheless, the progression in population dynamics clearly pointed toward resistance evolution, regardless of abiotic stress, and accelerated resistance evolution under drought stress. E. colona is a self-pollinating species, and evolution may be slower due to the low crossing rate [15,16,38,61]. In the same way, the self-pollinated A. fatua responses to low-dose herbicide selection was marginal, and it was much lower than in cross-pollinated L. rigidum, where L. rigidum evolved 40-fold resistance to diclofop-methyl via progressive enrichment of quantitative resistance-endowing traits [34]. A high level of evolved resistance was observed in allogamous species as L. rigidum with low doses of the herbicide diclofop-methyl for two generations and R. raphanistrum following four generations of low doses of 2,4-D selection [31,33,35].
Abiotic stresses such as CO 2 limitation , drought, and heat also impart selection pressure on plants in addition to herbicides and, therefore, can accelerate the evolution of herbicide-resistant plants. Previous related research and our current study support this premise [7,55,62]. Drought stress is a good example of an environmental stress that easily compromises herbicide performance. Lack of soil moisture reduces the absorption, translocation, and metabolism of florpyrauxifen-benzyl in E. crus-galli, Sesbania herbacea, and Cyperus esculentus [45]. The obvious consequence is reduced herbicide efficacy on these species, just as drought stress reduced the efficacy of florpyrauxifen-benzyl on E. colona in this current study. Furthermore, we showed that tolerance to florpyrauxifen-benzyl increased with time, under drought stress (as it did with the other herbicides tested). The ability to adapt was carried to the successive generations and strengthened as the progeny was subjected to the same stress conditions. E. colona has already evolved resistance to glyphosate, quinclorac, and other herbicides. Resistance to glyphosate was reported starting in 2007 in Australia and then in succeeding years in California, Venezuela, and Argentina due to intensive selection pressure [5,63]. Resistance to quinclorac is widespread among weedy Echinochloa species (E. crus-galli, E. crus-pavonis, E. zelayensis, and E. colona) as quinclorac is one of the primary herbicides used to control these grasses in rice production. Although florpyrauxifen-benzyl and quinclorac have the same mode of action, they each belong to a different chemical family. These two auxinic herbicides have activity on annual grasses but differ in the level of activity across species. A better understanding of E. colona response to synthetic auxins under different water regimes in rice production would enable producers to better use this weed management tool. The same principle applies to the use of glyphosate for general weed control in upland crops.

Memory of Gene Expression and Sensitivity Reduction in E. colona to Herbicide
For the plant to function normally and cope with the ebb and flow of various environmental stresses, several biochemical and physiological processes have to work in seamless coordination. Under stress conditions, as in herbicide treatment, the obligatory processes are modified quickly to mitigate the stress or cope with the stress. These modifications can be inferred from altered gene expression profiles, for example, between quinclorac-susceptible and -resistant E. colona [55]. The complexity of auxin perception signaling and diversity of plant response networks present a myriad of possibilities of how the plant can respond to auxinic herbicides metabolically or to other pathways [11,64,65].
Gene reprogramming and epigenetic regulation occur during plant development, being constantly provoked by environmental stimuli, eventually being manifested in a mechanism of plant adaptation and evolution [66]. Here, we studied the expression of genes involved in herbicide signaling, detoxification, and conjugation. Enzymes involved in antioxidative defense system, such as APX (ascorbate peroxidase), play an important role in the elimination of reactive oxygen species (ROS) by having a higher affinity for H 2 O 2 than catalase and peroxidases, protecting plant cells during oxidative stress [67]. Not surprisingly, the APX gene was upregulated in drought-stressed and herbicide-treated plants. Increased APX enzyme activity has been related to glyphosate damage protection in glyphosate-resistant A. palmeri [68]. In rice, the expression of antioxidant genes (SOD, CAT, and APX) increased after bentazon, penoxsulam, and cyhalofop-butyl application [69]. We also now know that the APX enzyme family is involved in a variety of plant processes, in addition to mitigating oxidative stress. Eight members of the APX family in Arabidopsis participate in growth and development processes, as well as responses to drought, heat, and salt stress [70]. Each enzyme certainly has a specific function, but multi-functionality of one enzyme is possible; one APX enzyme may protect the plant from multiple abiotic stresses. This indicates the potential role of antioxidant systems in the evolution of weed resistance to herbicides, in the process of adapting to recurring abiotic stresses.
The cytochrome P450 monooxygenase superfamily is frequently associated with metabolic resistance to herbicides; it is responsible for the phase I metabolism of xenobiotics, including herbicides [21,28,[71][72][73][74][75][76]. The upregulation of CytP450 genes identified here suggests their involvement in the evolution of resistance to florpyrauxifen-benzyl, glyphosate, or quinclorac and adaptation to drought stress. The large number of cytochrome P450 genes testify to the complexity and biodiversity of the function of this enzyme family [24,77,78]. The differential induction of CytP450 genes tested here by the synthetic auxins (florpyrauxifen-benzyl and quinclorac) indicates that, although they are from the same MOA group, they are significantly structurally different, such that they are possibly metabolized by different CytP450 genes [72]. Structural similarity primarily dictates cross-reactivity of CytP450 with different herbicides. A comprehensive study by Iwakami et al. [21] demonstrated that CYP81A P450s are involved in concomitant cross-resistance to ALS and ACCase herbicides on E. phyllopogon, and Dimaano et al. [78] conducted a functional characterization of cytochrome P450 CYP81A subfamily to disclose the pattern of cross-resistance to the same species.
Heat-shock proteins (HSPs) are molecular chaperones that ensure the correct folding of other proteins, throughout the life cycle of plants, to maintain growth and acclimatization. In addition, HSPs are known as stress-responsive proteins, as their concentration increases rapidly in plant cells in response to environmental stresses [79,80]. Currently, many research groups are investigating the mechanisms of action or function of HSPs in plants under various abiotic stress-causing conditions such as drought, salt, cold, heat, and even herbicides [81][82][83][84][85][86]. As expected, we determined that HSP transcript levels are higher under combined drought and herbicide stresses. Published information about HSPs pertain only to drought and other environmental stresses, not herbicides. The combined effect of drought and herbicide stress was not previously investigated. With respect to drought, Xiang and coworkers [87] also obtained upregulation of HSP50.2 in drought-stressed rice. The same authors reported that the overexpression of this gene in rice confers drought tolerance, probably through the modulation of ROS homeostasis and osmotic adjustment of the accumulated content of proline, which contributes to the improved protection ability from drought stress damage.
Some sugar compounds also play an important role in abiotic stress tolerance. Trehalose is a nonreducing disaccharide and a unique chemical compound, which is a key organic osmolyte involved in plant abiotic stress tolerance [88,89]. The pathway for trehalose biosynthesis contains two enzymatic steps: (1) trehalose-6-phosphate synthase (TPS) catalyzes the transfer of glucose from UDP-glucose to glucose 6-phosphate, forming trehalose 6-phosphate (T6P) and UDP; (2) trehalose-6-phosphate phosphatase (TPP) dephosphorylates T6P to trehalose and inorganic phosphate [90]. Recent findings associate the expression of carbohydrate metabolism-related genes with stress tolerance, such as the observed increase in drought tolerance in maize due to trehalose accumulation [91]. The overexpression of TPP increases drought tolerance in rice and Arabidopsis [92], the overexpression of TPS induces cold tolerance in Arabidopsis [93], and the overexpression of TPP and TPS increases tolerance to heat stress in transgenic tomato [94]. Our research showed the induction of TPP by florpyrauxifen-benzyl, glyphosate, and quinclorac under well-watered conditions. In addition, positive regulation of TPP and TPS was observed when glyphosate was applied to drought-stressed plants, suggesting the role of these genes in the adaptation of E. colona to simultaneous drought and glyphosate stress. On the other hand, TPS gene expression was downregulated after application of florpyrauxifen-benzyl or quinclorac to drought-stressed plants. We propose that, in water deficit, the action of TPS is inhibited by the auxinic herbicides, because there is interaction between auxin and the signaling sugar T6P synthetized by TPS [95]. In high auxin concentrations, TPS is inhibited, thereby affecting the sucrose status and normal plant growth [96][97][98][99].
Scientific advances in NTSR research have helped in the identification and understanding of resistance mechanisms in herbicide metabolism and in the network of detoxification processes [22,23,65,75,[100][101][102][103][104]. Plants contain large numbers of UDP-glucose-dependent glycosyltransferases (UGTs), responsible for conjugating metabolites with acceptor molecules in phase II of xenobiotic metabolism [105,106]. Increases in UGT expression after florpyrauxifen-benzyl, glyphosate, and quinclorac treatments indicate the involvement of this gene in the adaptation response to these herbicides. In rice, the glucosyltransferase gene was reported to be involved in the glucose conjugation of phenolics and possible quinclorac detoxification [107,108]. A glucosyltransferase was identified and enabled as a potential transcriptional marker for quick diagnosis of metabolic resistance to mesosulfuron in Alopecurus aequalis [18].
Rouse and collaborators [55] presented a model of the biological pathway for quinclorac conjugation via UGT. The quinclorac molecule contains an exposed OH − side group with which UGT can interact, suggesting the phase I step would not be necessary. Ljung et al. [109] and Jin et al. [110] described that the major route for deactivating endogenous auxin is through conjugation with sugars via glucosyltransferases, corroborating our findings. Furthermore, the upregulation of glycosyltranferase genes in glyphosate-resistant Conyza bonariensis after glyphosate treatment might indicate the increase in activity of these enzymes to cope with glyphosate [22]. Here, we detected a similar profile of UGT expression following glyphosate treatment.
Overall, our data support the involvement of several classes of candidate genes in E. colona adaptation to florpyrauxifen-benzyl, glyphosate, and quinclorac under drought stress ( Figure 4). The NTSR mechanism is complex and its occurrence is increasing, which means that weed control may become even more difficult in times of climate change, by accelerating resistance evolution. While this research provides information on the increase in E. colona tolerance to sublethal doses of florpyrauxifen-benzyl, glyphosate, and quinclorac under drought stress, we cannot extrapolate this response, with certainty, to other herbicide mechanisms of action. Further research is required to clarify our understanding of the underlying mechanisms which allow E. colona to adapt to various herbicide MOAs under different climate change conditions.

Conclusions
Drought stress and recurrent sublethal dose application reduce the susceptibility of E. colona to florpyrauxifen-benzyl, glyphosate, and quinclorac herbicides, thus facilitating the adaptation of E. colona to climate change. The upregulation of all genes studied in selected progeny plants (G2) of E. colona with drought stress may be involved in the reduction of sensitivity to glyphosate. Drought stress combined with a low dose of florpyrauxifen-benzyl induced the expression of all genes studied, except TPS. A low dose of quinclorac plus drought stress increased the expression of most candidate genes, except for CYP709B1, CYP709B2, and TPS. This differential gene regulation possibly facilitates the reduction in susceptibility of E. colona to these stresses. Overall, recurrent selection with sublethal herbicide dose under drought stress reduced E. colona sensitivity to herbicide, seemingly due to "imprinted" upregulation of metabolic and protection genes in response to these stresses.