Gemering a weed seed

Combined ionic liquid and supercritical carbon dioxide based dynamic extraction of six cannabinoids from Cannabis sativa L. †

The potential of supercritical CO2 and ionic liquids (ILs) as alternatives to traditional extraction of natural compounds from plant material is of increasing importance. Both techniques offer several advantages over conventional extraction methods. These two alternatives have been separately employed on numerous ocassions, however, until now, they have never been combined for the extraction of secondary metabolites from natural sources, despite properties that complement each other perfectly. Herein, we present the first application of an IL-based dynamic supercritical CO2 extraction of six cannabinoids (CBD, CBDA, Δ 9 -THC, THCA, CBG and CBGA) from industrial hemp (Cannabis sativa L.). Various process parameters were optimized, i.e., IL-based pre-treatment time and pre-treatment temperature, as well as pressure and temperature during supercritical fluid extraction. In addition, the impact of different ILs on cannabinoid extraction yield was evaluated, namely, 1-ethyl-3-methylimidazolium acetate, choline acetate and 1-ethyl-3-methylimidazolium dimethylphosphate. This novel technique exhibits a synergistic effect that allows the solvent-free acquisition of cannabinoids from industrial hemp, avoiding further processing steps and the additional use of resources. The newly developed IL-based supercritical CO2 extraction results in high yields of the investigated cannabinoids, thus, demonstrating an effective and reliable alternative to established extraction methods. Ultimately, the ILs can be recycled to reduce costs and to improve the sustainability of the developed extraction process.


Cannabis sativa L. is an annual herbaceous blossoming plant that has been used throughout history in the textile industry, for recreational purposes and in medical applications. It is regarded as one of the oldest cultivated plants, and one of the most essential crops for the progress of humankind. Although native to Eastern Asia, its extensive applications led to its global spread. 1

The medicinal properties of Cannabis sativa L. can be attributed to the many bioactive compounds present in the plant, such as terpenes, polyphenols, phytosterols, tocopherols, fatty acids, and, specifically, cannabinoids, which are terpenophenolic secondary metabolites. 2,3 It is important to mention that cannabinoids are not equally distributed in the plant. They are mainly found in the trichomes and in smaller to negligible amounts in the seeds, while roots contain none. 4

Presently, over 100 cannabinoids have been identified. 5 They are primarily encountered in their carboxylated form in the plant which constitutes a structure of 22 carbon atoms. So far, cannabinoids have been categorized into 11 subclasses: (1) (−)-Δ 9 -tetrahydrocannabinol (Δ 9 -THC), (2) (−)-Δ 8 -tetrahydrocannabinol (Δ 8 -THC), (3) cannabidiol (CBD), (4) cannabigerol (CBG), (5) cannabichromene (CBC), (6) cannabinol (CBN), (7) cannabinodiol (CBND), (8) cannabicyclol (CBL), (9) cannabielsoin (CBE), (10) cannabitriol (CBT) and (11) miscellaneous. The structures of cannabinoids from hemp investigated in this study are depicted in Fig. 1 . 6

In terms of the biosynthesis of cannabinoids, CBGA is the main precursor for THCA and CBDA. 7 However, under high temperatures, both acids are prone to degrade into their respective decarboxylated analogues, Δ 9 -THC and CBD. 8

Δ 9 -THC and CBD are the most abundant cannabinoids present in cannabis plants. Δ 9 -THC is well-known as a psychoactive compound, which influences the central nervous and cardiovascular systems. Contrarily, CBD is non-psychoactive, but is regarded as a compound of enormous medical interest, as it has demonstrated numerous health benefits. It has been reported to have anti-inflammatory, antiepileptic and anticonvulsive properties, among many others. 9–11 Excellent medicinal potential have been attributed to cannabinoids; thus, significant effort has been made in the past decades towards the research of the functions and mechanisms of cannabis-derived secondary metabolites in the human body.

Due to the growing medicinal interest in cannabinoids over the years, scientists have undertaken efforts in the development of extraction methods for these valuable bioactive compounds. Traditionally, Δ 9 -THC and other cannabinoids have been isolated by solvent-based extractions, with hydrocarbons and alcohols delivering the highest yields. 12,13 Soxhlet extraction (SE) is also a commonly used technique, 14,15 which is characterized by shortcomings, namely, long extraction times and high temperature that may promote thermal degradation of the target compounds. 16

Other advanced extraction techniques, such as microwave-assisted extraction (MAE) allow higher yields, shorter extraction times, less solvent and reduced energy consumption. 14,17 Nevertheless, uneven heating and/or overheating may cause thermal degradation, and thus negatively impact the extraction efficiency. 18 Alternatively, the use of ultrasound-assisted extraction (UAE) achieves high yields in short times; 19 however, the distribution of ultrasound energy lacks uniformity and over time the power decreases, which can lead to inefficient use of the ultrasound-generated energy. 20

Supercritical fluid extraction (SFE) is an innovative separation technique, which has thus far been employed for extractions of valuable constituents from over 300 plant species. 21 Carbon dioxide is a widespread choice for SFEs due to its several advantageous properties, such as low reactivity, non-toxicity, non-flammability, affordability, availability, and recyclability. Additionally, its selectivity can be adjusted by modification of pressure and temperature, while product fractionation and recovery with high purity is feasible. Nevertheless, due to its low polarity, addition of small quantities of organic solvents (co-solvents or modifiers) is necessary to access more polar compounds, thereby expanding its extraction range. 22 The selection of an appropriate co-solvent is key for achieving optimum solubility of the bioactive compounds present in the plant. 23 Supercritical carbon dioxide has previously been used to assess the solubility of individual cannabinoids, for example, Δ 9 -THC, 24 CBD 25 and CBG. 25 Moreover, several extractions of cannabinoids from different parts of the cannabis plant, for instance, leaves, trimmings, buds, flowers and threshing residues, have been performed using ethanol as a co-solvent. 26–29

Within the past years, ionic liquids have also emerged as alternative reaction media for the extraction of biomass that is regarded as a source of natural medicinally relevant complex compounds. Many different properties are attributed to ionic liquids, such as exceptional dissolution properties, high thermal stability and broad liquid range, to name a few. Furthermore, ILs display high tuneability, as the combination of different cations and anions leads to hydrophilicity or hydrophobicity and different polarity. 30

The dissolution and processing of lignocellulosic biomass is a particularly interesting application of ionic liquids (ILs), as they can directly dissolve and fractionate (ligno-)cellulose in an overall less energy intensive process. 31,32 The biomass dissolution capability of ILs is impacted by both their cation and anion, however, current publications suggest that anions have a more significant impact, since they play a role in breaking the many intermolecular hydrogen bonds. 30 Regarding the cation, imidazolium-based ILs were the most successful for the direct dissolution of cellulose, followed by pyridinium- and ammonium-based ones. 33 In addition, increasing the chain length of the cation had a negative influence on the dissolving capabilities of the ILs, as the viscosity increased, and the H-bond acidity decreased. As far as the anion is concerned, dissolving efficiency seems to be determined by the H-acceptor properties of the anion. In general, anions with weak H-bond basicity, for instance, [BF4] − and [PF6] − , could not successfully dissolve cellulose, while ionic liquids based on halide or acetate anions are typically the candidates of choice. 30,34 The growing research on ILs as solvents for lignocellulosic biorefinery also prompted innovations for the extraction of valuable ingredients from plant materials. 35 There are several aspects of ILs that are potentially advantageous for the extraction of high-value compounds: apart from their unique solvent properties and potential environmental benefits, the ability of ILs to dissolve biomass can lead to a better, and higher, yielding access to valuable ingredients embedded in the biopolymers and contribute to a value-added biorefinery. 36,37 However, the recovery of natural products from ionic liquids is often more demanding than the mere extraction: many studies require extensive back-extraction with volatile solvents to actually isolate the valuable ingredients from ILs, thereby rendering the original solvent reduction less significant or even negating it altogether.

The combination of non-volatile polar ILs with volatile non-polar scCO2 has several advantages for extractions, as well as for catalysis. Since scCO2 is highly soluble in ILs, but ILs cannot dissolve in scCO2, it can easily penetrate the IL-phase. This allows the extraction of compounds from the IL-phase into the scCO2 phase, taken into account that the organic compound of interest is soluble in scCO2. Ultimately they are transported into an extraction vessel in a pure, solvent-free and solid form. 38

Furthermore, ILs in the presence of CO2 expand their applicability, as their melting point and viscosity decrease, thus, promoting mass transportation. 39 Consequently, the combination of ionic liquids with scCO2 has found application in several catalytic processes, such as hydroformylations, hydrogenations or carboxylations of alkenes in IL-scCO2 biphasic reaction media. 40–43 In the IL-scCO2 reaction systems, the reactants and products are carried by the scCO2 and IL is used as a reaction media. 44,45 Additionally, it is demonstrated that IL-scCO2 biphasic systems avoid cross-contamination of the extracted solute. 38,46

Until now, IL-based pre-treatment and subsequent SFE (IL-SFE) for natural products has not been described, although ideal conditions arise from the unique properties of both media. Hence, by comparing IL-scCO2 extraction with the utilization of both applications individually or to traditional solvent-extraction, the IL-scCO2 approach is preferable. To begin with, less additional preparation, e.g., filtration of the raw material and consequent evaporation of solvents or separation of IL from the organic solvent is required to obtain a solvent-free and solid extract ( Fig. 2 ). Consequently, there is a lower chance of loss of product or impurities, due to less post processing steps. On the other hand, IL-SFE is performed without additional co-solvents, therefore it reduces further solvent consumption and leads to lower expenses. Ultimately, if chosen appropriately, the ionic liquid can be recovered and re-used to improve the sustainability of the extraction process.

Conceptualization for the comparison of work up steps and yields of cannabinoids extraction techniques.

Recently, an investigation of the extraction of cannabidiol with the aid of ILs has been published; however, isolation of cannabidiol required tedious back-extraction with organic solvents or with an aqueous AgNO3 solution. 47 To the best of our knowledge, no data has been reported thus far regarding a combined extraction process that takes advantage of the complementing properties.

Herein, we present the first application of IL-SFE from industrial hemp of six cannabinoids (Δ 9 -THC, THCA, CBD, CBDA, CBG and CBGA). Several parameters during the IL-assisted pre-treatment, such as time, temperature and dilution with H2O, were investigated. In addition, pressure and temperature during SFE were evaluated. Ultimately, the optimized process for 1-ethyl-3-methylimidazolium acetate ([C2mim][OAc]) was additionally performed with choline acetate ([Ch][OAc]) and 1-ethyl-3-methylimidazolium dimethyl phosphate ([C2mim][DMP]) to compare the extraction efficiency of the investigated cannabinoids. In addition, the developed extraction process is complemented by a simple ionic liquid recovering process without the usage of additional organic solvents.

Results and discussion

The focus of this research was the investigation and optimization of various parameters for the extraction of CBD, CBDA, Δ 9 -THC, THCA, CBG and CBGA from partially pre-dissolved hemp in various room-temperature ILs with supercritical CO2. The optimization was divided into three successive stages ( Scheme 1 ).

In the first stage, the pre-treatment conditions to digest and partially dissolve hemp using [C2mim][OAc] before SFE were investigated. The lignocellulosic composition of hemp hurds is reported to contain 43.0% cellulose, 24.4% lignin and 29.0% hemicellulose. 48 ILs are known to dissolve a variety of carbohydrates, e.g., cellulose, by combining strongly basic anions (e.g., Cl − or OAc − ) with various cations. 49–51 In particular, [C2mim][OAc] was selected in this study as it was used to pre-treat various lignocellulosic biomasses 52 and it is known to effectively dissolve, hemicellulose 53 and lignin. 54 Furthermore, [C2mim][OAc] is liquid at room temperature, non-halogenated and miscible with H2O.

Subsequently, the best extraction conditions of stage 1 were employed in determining the most effective ratio of [C2mim][OAc] : H2O during SFE. In the third stage, the previously optimized conditions from the first and second stage were utilized to investigate several combinations of pressure and temperature during SFE. Ultimately, the optimum parameters were employed with two additional ILs, namely [Ch][OAc] and [C2mim][DMP]. Both ILs are liquid at room temperature, non-halogenated and hydrophilic. Moreover, both ILs have been reported for pre-treatment of biomass. 55,56 In addition, the positive rating of choline-based ILs in terms of toxicity and biodegradation renders them ideally suited for natural product extractions. 57,58

Pre-treatment with ionic liquid (Stage 1)

Herein, the influence of temperature and time for the partial dissolution of Cannabis sativa L. in [C2mim][OAc] before the scCO2 extraction is evaluated.

Initially, the conditions to partially dissolve industrial hemp in [C2mim][OAc] were investigated in experiments 1–4 ( Table 1 ).

Yields of cannabinoids in mg g −1 for the optimization of pre-treatment with [C2mim][OAc] at different temperatures and time. SFE was performed at 20 MPa and 70 °C with a ratio of [C2mim][OAc] : H2O 1 : 2 (Stage 1)

Exp. t Pre/min T Pre/°C (CBD) (mg g −1 ) (THC) (mg g −1 ) (CBG) (mg g −1 )
1 60 25 13.1 ± 0.8 a 0.464 ± 0.008 b 0.229 ± 0.011 c
2 60 70 12.9 ± 0.3 a 0.471 ± 0.019 b 0.244 ± 0.014 c
3 15 25 13.0 ± 0.8 a 0.48 ± 0.03 b 0.221 ± 0.019 c
4 15 70 13.6 ± 0.6 a 0.513 ± 0.017 b 0.247 ± 0.017 c

Therefore, the pre-treatments were carried out at 25 and 70 °C, each 15 and 60 min, afterwards diluted with H2O to a ratio of 1 : 2 and subsequently subjected to SFE at 20 MPa and 70 °C. To evaluate the quality of the performed experiments during the development of IL-SFE for hemp the yields of cannabinoids are expressed as the sum of cannabinoid types e.g. CBD and CBDA are referred to as (CBD). Analogously (THC) and (CBG) are calculated. All experimental conditions and results for individual cannabinoid yields are shown in the ESI (Tables S1 and S2 † ).

The cannabinoids CBD and CBDA are predominantly accumulated in industrial hemp compared to THC, THCA, CBG and CBGA, which are considered minor compounds.

The pre-treatment with [C2mim][OAc] of industrial hemp at 25 °C and 70 °C indicated comparable cannabinoid yields. Increasing the time from 15 to 60 min at 70 °C in exp. 2 led to a small decrease of roughly 5% (CBD) and 8% (THC). However, similar (CBD), but significantly more CBD (6.58 mg g −1 ) and less CBDA (6.3 mg g −1 ) at 60 min, was yielded in exp. 2 compared with exp. 4 (15 min), which led to 5.29 mg g −1 CBD and 8.8 mg g −1 CBDA, respectively (p < 0.05, Fig. 3 , Table S2 † ). It was reported that an extraction process including [C6mim][NTf2] at 60 °C and 50 min leads to high amounts of CBD and that the IL preserves CBD, 47 which correlates with the observations herein. In addition, the decarboxylation of cannabinoids at higher temperatures for longer times has been described before. 8 The IL [C6mim][NTf2] was not utilized in this study, as the anion [NTf2] − renders it is less suitable to dissolve cellulose compared to the basic [OAc] − or [DMP] − and similarly, the longer alkyl side chain of the cation would be disadvantageous for this purpose. 59 Ultimately, [NTf2] − was not considered for the extraction process, as it is hydrophobic and not mixable with H2O and thus, not suitable for the IL recovering process shown in here.

Comparison of cannabinoid yields (mg g −1 ) at different pre-treatment temperatures and pre-treatment times with [C2mim][OAc] : H2O 1 : 2 and subsequent SFE at 20 MPa and 70 °C, (n = 3 ± SD). Experiments refer to Table 1 for Stage 1.

A total time of 15 min instead of 60 min seems to be sufficient to release the investigated cannabinoids from the plant tissue with [C2mim][OAc] and hence, allows a significantly shorter pre-treatment time The highest cannabinoids yields were obtained at 70 °C for 15 min in exp. 4, namely 13.6 mg g −1 (CBD), 0.513 mg g −1 (THC)and 0.247 mg g −1 (CBG) ( Table 1 ).

Ratio of ionic liquid to water (Stage 2)

Optimization of temperature and time during the pre-treatment was performed with a constant ratio of 1 : 2 [C2mim][OAc] : H2O. Here, the influence of several IL : H2O ratios was investigated and compared with the sole use of IL as well as pure H2O in the extraction vessel ( Table 2 and Fig. 4 ).

Cannabinoid yields (mg g −1 ) for IL-SFE with pure IL and different IL : H2O ratios, using 15 min of pre-treatment time at 70 °C and for SFE with H2O (pure H2O) as well as for scCO2 (no pre-treat). All extractions were performed at 70 °C and 20 MPa; IL = [C2mim][OAc], (n = 3 ± SD). Experiments refer to Table 2 for Stage 2.

Yields of cannabinoids in mg g −1 by investigating the influence of H2O and [C2mim][OAc] during SFE with a pre-treatment at 70 °C for 15 min and SFE at 20 MPa and 70 °C (Stage 2)

Exp. m IL/g m H2O/g (CBD) (mg g −1 ) (THC) (mg g −1 ) (CBG) (mg g −1 )
4 3 6 13.6 ± 0.6 b 0.513 ± 0.017 a 0.247 ± 0.017 b
5 3 3 8.53 ± 0.19 e 0.330 ± 0.014 c 0.226 ± 0.007 bc
6 3 9 15.6 ± 0.7 a 0.542 ± 0.016 a 0.335 ± 0.016 a
7 3 0.322 ± 0.022 f 0.033 ± 0.006 d n.d.
8 9 12.0 ± 0.6 c 0.375 ± 0.022 b 0.260 ± 0.008 b
9 10.1 ± 0.5 d 0.355 ± 0.009 bc 0.196 ± 0.019 c

A decrease of water in the IL : H2O ratio from 1 : 2 in exp. 4 to 1 : 1 in exp. 5 led to a significant reduction of (CBD) as well as (THC) yield ( Table 2 ) at 20 MPa and 70 °C. However, the significantly highest yield of CBD (7.45 mg g −1 ) of all performed IL-SFE was obtained under these conditions in exp. 5 (p < 0.05) and additionally, low yields of CBDA (1.09 mg g −1 ) and no CBGA were extracted (Fig. 4 , Table S2 † ). Therefore, a ratio of 1 : 1 [C2mim][OAc] : H2O during SFE seems to favour the extraction of neutral CBD and CBG. Recently, it has been discovered that high yields of CBD are extracted by pre-heating hemp and subsequent extraction with supercritical CO2 combined with EtOH as a modifier. 29 Similar behaviour can be observed under the previously mentioned IL-SFE conditions, without addition of co-solvents.

On the other hand, significantly more (CBD) and (CBG) (p < 0.05) were obtained in exp. 6 by addition of more H2O to increase the ratio of [C2mim][OAc] : H2O from 1 : 2 to 1 : 3. The (CBD) yield increased by 15% to 15.6 mg g −1 , (THC) by 6% to 0.542 mg g −1 and (CBG) by 36% to (0.335 mg g −1 ) ( Table 2 ). Adding more than 15 wt% H2O during [C2mim][OAc] pre-treatment does not allow complete cellulose dissolution, as reported by Le et al. in 2012. 60 Therefore, H2O was added to the IL after the initial pre-treatment. The addition of H2O resulted in a reduction of the mixture’s viscosity, and thus improved mass transport. 60 It is reported that the viscosity of [C2mim][OAc] is reduced by 50% when mixed with 10 wt% H2O and that the IL is less viscous at higher temperatures. 61 Lower viscosity of the IL : H2O mixture led to higher yields, possibly due to the higher mobility of dissolved cannabinoids and better penetration of scCO2. An increase in carboxylated cannabinoids was observed by adding more water ( Fig. 4 ). Furthermore, water is the only solvent without any negative impacts on the environment. Additionally, it is reported to have low solubility in scCO2 62 and therefore less potential contamination of the extract.

The absence of H2O during the extraction with scCO2 and [C2mim][OAc] (pure IL) led to the lowest yields of all SFE in exp. 7 ( Table 2 and Fig. 4 ). Low yields can be a result of the high viscosity of the IL, which leads to less permeability of scCO2 and subsequently lower mass transfer in the extraction. Therefore, dilution with H2O is essential during the extraction process.

However, the sole extraction with H2O (pure H2O) in the absence of [C2mim][OAc] in exp. 8 compared to exp. 6, leads to a significant reduction of (CBD) by 23% to 12.0 mg g −1 , (THC) by 31% to 0.375 mg g −1 and (CBG) by 22% to 0.260 mg g −1 (p < 0.05, Table 2 ). In particular, the use of H2O alone tends to yield fewer neutral cannabinoids ( Fig. 4 ), which verifies what has previously been reported; ILs preserve neutral CBD. 47 When comparing exp. 8 with exp. 4, even though the same total quantity of liquid was added in the high-pressure vessel, significantly less yields of (CBD) by 12% and (THC) by 27% are observed (p < 0.05, Table 2 , Fig. 4 ) in the sole water-based SFE extraction. Therefore, a pre-treatment with IL to liberate the cannabinoids from the plant tissue and subsequent dilution with H2O positively affects the yield.

The addition of EtOH as a co-solvent to IL-SFE would lead to the extraction of both IL and cannabinoids, thus, leading to impurities in the extract. In particular, IL-SFE does not require the use of a co-solvent to obtain cannabinoids in high yields, avoiding further solvent consumption.

Hence, the highest extraction yields were obtained with a IL : H2O ratio of 1 : 3 in exp. 6, which achieved 15.6 mg g −1 (CBD), 0.542 mg g −1 (THC) and 0.335 mg g −1 (CBG) ( Table 2 ).

SFE extraction parameters – pressure and temperature (Stage 3)

Apart from the optimization of pre-treatment conditions and the ratio of [C2mim][OAc] to H2O, temperature and pressure during SFE were investigated ( Table 3 ).

Yields of cannabinoids (mg g −1 ) for different temperatures and pressures during SFE. Pre-treatment with [C2mim][OAc] was carried out at 70 °C for 15 min and extracted with a IL : H2O ratio of 1 : 3 (Stage 3)

Exp. P SFE/MPa T SFE/°C (CBD) (mg g −1 ) (THC) (mg g −1 ) (CBG) (mg g −1 )
6 20 70 15.6 ± 0.7 a 0.542 ± 0.016 a 0.335 ± 0.016 a
10 10 70 3.66 ± 0.06 d 0.0885 ± 0.0026 d 0.045 ± 0.008 c
11 15 70 13.00 ± 0.19 c 0.457 ± 0.005 c 0.257 ± 0.009 b
12 30 70 14.7 ± 0.7 ab 0.500 ± 0.024 ab 0.36 ± 0.04 a
13 20 35 14.9 ± 0.7 ab 0.493 ± 0.014 bc 0.323 ± 0.009 a
14 10 35 13.7 ± 0.8 bc 0.468 ± 0.019 bc 0.248 ± 0.010 b

Initially, the pressure was reduced from 20 MPa in exp. 6 to 15 MPa in exp. 11 at 70 °C and led to a significant reduction by 17% (CBD) to 13.00 mg g −1 , 16% (THC) to 0.457 mg g −1 and 23% (CBG) to 0.245 mg g −1 (p < 0.05, Table 3 ). After further decreasing the pressure to 10 MPa in exp. 10, a significantly diminished yield of 3.66 mg g −1 (CBD), 0.0885 mg g −1 (THC), and 0.045 mg g −1 (CBG) was observed ( Table 3 ). Even though lower cannabinoid yields were obtained at 10 MPa and 70 °C in exp. 10, the extraction of neutral cannabinoids was favoured ( Fig. 5 ). In literature, sole scCO2 extractions yield neither CBD nor CBDA at 10 MPa at 70 °C for 120 min, 63 but SFE can be improved upon by adding EtOH 26 or by the pre-treatment with IL, as herein reported. In addition, the pressure was increased to 30 MPa at 70 °C in exp. 12, which resulted in comparable yields of (CBD, THC, CBG) as IL-SFE at 20 MPa in exp. 6 ( Table 3 ). It can be assumed that 20 MPa at 70 °C are sufficient to extract cannabinoids during IL-SFE.

Cannabinoid yields (mg g −1 ) at 10, 15, 20 and 30 MPa at 70 °C by scCO2 extraction combined with [C2mim][OAc] : H2O 1 : 3 and a pre-treatments at 70 °C for 15 min, (n = 3 ± SD). Experiments refer to Table 3 for Stage 3.

Furthermore, the temperature was lowered to 35 °C at 20 MPa during SFE in exp. 13. This led to comparable yields of (CBD) and (CBG), but significantly lower (THC) yields (0.493 mg g −1 ) compared to 70 °C in exp. 6 (p < 0.05, Table 3 ). This corresponds to literature data, where similar yields of (CBD) were extracted during SFE at 35 °C and 70 °C at 50 MPa. 63 Lower temperatures are known to reduce the viscosity of H2O and additionally, have been reported to decrease the viscosity of [C2mim][OAc]. 61 Hence, the mixture is less penetrable for scCO2 to extract the target cannabinoids. In comparison of exp. 13 and exp. 6, the yields of decarboxylated cannabinoids decreased significantly (CBD by 28%; Δ 9 -THC by 16%; CBG by 33%) and similar yields of THCA and CBDA, but significantly more CBGA by 23% was obtained in exp. 13 (p < 0.05, Table S3 † ).

Consequently, the optimum cannabinoid yields were obtained at 20 MPa and 70 °C in exp. 6 during supercritical CO2 extraction.

Type of ionic liquid

Two additional ILs, namely, [Ch][OAc] and [C2mim][DMP], were selected for evaluation alongside [C2mim][OAc]. The optimized extraction conditions, with a pre-treatment at 70 °C for 15 min and subsequent SFE at 70 °C and 20 MPa with a IL : H2O ratio of 1 : 3, were additionally applied to these two ILs to observe differences in cannabinoid yields ( Table 4 and Fig. 6 ).

Cannabinoids yields (mg g −1 ) for IL-SFE with [C2mim][OAc], [Ch][OAc] as well as [C2mim][DMP] (IL : H2O 1 : 3), for SFE with H2O (pure H2O) and for scCO2 (no pre-treat). All extractions were performed at 70 °C and 20 MPa, (n = 3 ± SD). Pure H2O (exp. 8) and scCO2 (no pre-treat) (exp. 9) refer to Table 2 . [C2mim][OAc] (exp. 6), [Ch][OAc] (exp. 15) and [C2mim][DMP] (exp. 16) refer to Table 4 .

Yields of cannabinoids (mg g −1 ) by comparing different ILs and reference extractions in EtOH and H2O. IL-SFE was performed with a pre-treatment at 70 °C for 15 min, a ratio of 1 : 3 IL : H2O at 70 °C and 20 MPa during SFE

Exp. IL or solvent (CBD) (mg g −1 ) (THC) (mg g −1 ) (CBG) (mg g −1 )
6 [C2mim][OAc] 15.6 ± 0.7 a 0.542 ± 0.016 a 0.335 ± 0.016 c
15 [Ch][OAc] 15.4 ± 0.5 a 0.535 ± 0.010 ab 0.401 ± 0.024 b
16 [C2mim][DMP] 11.8 ± 0.9 b 0.449 ± 0.025 c 0.292 ± 0.028 c
17 a EtOH for 2 h 15.4 ± 0.4 a 0.498 ± 0.018 b 0.452 ± 0.019 a
18 a EtOH for 24 h 14.84 ± 0.15 a 0.447 ± 0.005 c 0.440 ± 0.004 ab
19 a H2O for 2 h 1.6 ± 0.3 c 0.057 ± 0.012 d 0.031 ± 0.007 d

To investigate the influence of the anion in IL-SFE of cannabinoids from industrial hemp, the imidazolium-based IL [C2mim][DMP] was used in exp. 16. This resulted in a significant reduction of (CBD) to 11.8 mg g −1 and total THC to 0.449 mg g −1 compared with the acetate-based ILs in exp. 6 and exp. 15 (p < 0.05, Table 4 ). [C2mim][DMP] is described as effectively dissolving biomass, but has a high viscosity, 56,65 which could affect the extraction at supercritical conditions, due to the weaker penetration of scCO2. Nonetheless, phosphate based and acetate based IL-SFE yielded higher amounts of (CBD, THC, CBG) compared with sole supercritical CO2 extraction without IL pre-treatment ( Fig. 6 ).

The following mechanism can be proposed for IL-SFE. Firstly, the biomass is partially dissolved by breaking down the lignocellulose structure of the industrial hemp powder. This depends on the anion and cation of the ILs. 34,64 The cannabinoids are released from the plant tissues and the IL possibly stabilizes them. 47 Secondly, the water is added, which reduces the viscosity of the mixture 61 and lowers the solubility of the target cannabinoids. Due to the lower surface tension and higher mobility of cannabinoids, a higher mass transfer between the scCO2 phase and the IL : H2O phase is generated. As reported the scCO2 dissolves in ILs, however, neither the IL nor the H2O does dissolve in scCO2. 38,62 Finally, these synergic effects allow the scCO2 to extract the targeted cannabinoids, due to better solubility in the supercritical phase without contaminating it with IL or H2O. Thus, no further organic solvents are necessary to purify the compounds from the IL phase and consequently, no additional work up is needed to obtain a solid and solvent free product ( Fig. 2 ).

Ultimately, IL-SFE was compared with reference solvent extraction (exp. 17–19). Ethanol is one of the most commonly used solvents to extract cannabinoids. 66 Herein, a conventional extraction for 2 h, at 70 °C, with EtOH in exp. 17, sufficiently extracted the investigated cannabinoids; however, employing H2O in exp. 19 alone under the same conditions, low yields of cannabinoids were obtained ( Table 4 ). A control extraction in EtOH for 24 h was carried out in exp. 18 to investigate the influence of longer extraction times. Longer times at high temperatures seem to degrade carboxylated cannabinoids significantly, reducing CBDA by 52%, THCA by 65% and CBGA by 53% (p < 0.05, Table S2 † ). The decarboxylation of cannabinoic acids at high temperatures for longer times is described in literature. 8 However, it can be reported that the degradation over time does not affect the overall cannabinoid yields.


Herein, we report a novel IL-based dynamic supercritical CO2 extraction process for the isolation of cannabinoids from Cannabis sativa L. The investigation showed that 15 min at 70 °C pre-treatment of hemp with [C2mim][OAc] and [Ch][OAc], dilution of IL with H2O (1 : 3) and ultimately, scCO2 extraction at 20 MPa and 70 °C for 2 h, led to high yields of the investigated cannabinoids. Acetate-based ILs resulted in higher yields of cannabinoids compared to phosphate-based ILs. In addition, IL-SFE with [C2mim][OAc] yielded significantly more (THC) than conventional extraction with EtOH. Hence, the type of IL is of great importance and affects the cannabinoid yield significantly. However, not only the type of IL needs to be selected carefully, also the SFE parameters. In dependence of various parameters, e.g. IL pre-treatment temperature or the ratio of IL : H2O during SFE, it is possible to adjust the proportion of carboxylated and decarboxylated cannabinoids in the extracts. In addition, IL-SFE allows extracting cannabinoids in highest yields and, therefore, it can be reported as a novel competitive alternative to traditional extraction techniques or supercritical fluid extraction with co-solvents. Ultimately, the ILs can be recycled without additional usage of further organic solvents to reduce costs and improve the sustainability of the process. IL-SFE offers the opportunity to extract secondary metabolites from different natural sources without volatile organic solvents and the presented process has great potential for future industrial applications.


Plant material

The type III chemovar Futura 75 was cultivated in Austria, in the fields of Biobloom (Apetlon, Austria, 7°41′23.4′′N 16°56′26.7′′E), in September 2020. After the harvest, the plants (flowers, leaves and stems) were stored under mild conditions at 40 °C for 14 h. The samples were milled with a Fritsch Universal Pulverisette 19 mill through a 2 mm sieve (Fritsch, Oberstein, Germany). The dry matter was 94.73 ± 0.05 wt% (n = 3). A second batch of the same industrial hemp harvested in 2019 was used for the preliminary experiments, mentioned in section Results and discussion. The dry matter was 93.68 ± 0.03 wt% (n = 3). The hemp raw material was stored in the dark, at −20 °C, between experiments.

Ionic liquid-supercritical fluid extraction

For pre-treatment, a high-pressure vessel of approximately 50 mL (EV-3), produced by Jasco (Jasco Corporation, Tokyo, Japan), containing one input and one output connections on the lid, was used. The batch reactor was charged with 0.20 g milled hemp and 3 g of IL. [C2mim][OAc] (≥90%) was purchased from BASF (Ludwigshafen am Rhein, Germany), [Ch][OAc] (98%) from IoLiTec (Heilbronn, Germany) and [C2mim][DMP] (98%) from ABCR (Karlsruhe, Germany). Pre-treatment optimization was performed for 15 min and 60 min, at 25 °C and 70 °C, respectively, with [C2mim][OAc]. Furthermore, [C2mim][OAc] was diluted with different amounts of H2O (filtered through a Milli-Q ion exchange system) after the pre-treatment to evaluate the effect on extraction efficiency. Therefore, 3 g of IL were mixed with 3 g, 6 g and 9 g of H2O and stirred for 10 min before SFE. In addition, extraction purely with [C2mim][OAc], without the addition of water, was tested.

The SFE setup is presented in Fig. 7 . All extractions were performed with a scCO2 device manufactured by Jasco (Jasco Corporation, Tokyo, Japan). Liquid CO2 (>99.995% purity; with ascension pipe; Messer GmbH, Vienna, Austria) was pressurized by two CO2-pumps (PU-2086, Jasco Corporation, Tokyo, Japan) with cooled heads (CF40, JULABO GmbH, Seelbach, Germany). An oven (CO-2060, Jasco Corporation, Tokyo, Japan) with a heating coil was used and was thermostated to the desired temperature. The vessel containing the IL pre-treated hemp was placed on a heating mantle set to a certain temperature and a stirring rate of 500 rpm and, subsequently, connected to the supercritical carbon dioxide (scCO2) device. A back-pressure regulator (BP-2080, Jasco Corporation, Tokyo, Japan), a gas/liquid separator (HC-2086-01, Jasco Corporation, Tokyo, Japan), and a product collector (SCF-Vch-Bp, Jasco Corporation, Tokyo, Japan) were used to obtain the extracts.

General setup for the dynamic extraction of cannabinoids using IL-SFE. (1) Liquid CO2 supply, (2) chiller/cooling system, (3) CO2 pump, (4) manually operated valve, (5) thermostated oven with preheating coil, (6) high pressure vessel placed on a thermostated stirrer, (7) back pressure regulator (BPR), (8) gas–liquid separator, and (9) fraction collector.

The conditions employed for the SFE of cannabinoids were based on literature data 63,69 and adapted for our purposes. The CO2 flow rate, the static extraction and the dynamic extraction were set to 5.0 mL min −1 , 30 min and 120 min, respectively. Different variables were evaluated during SFE, e.g., oven temperature, heating mantle temperature (35 °C and 70 °C, respectively) and pressure (10 MPa, 15 MPa, 20 MPa and 30 MPa), using [C2mim][OAc]. Ultimately, the optimized conditions were applied to [Ch][OAc] and [C2mim][DMP]. After each extraction, the extracts were collected and diluted to a defined volume with ethanol and prepared for analysis by HPLC. EtOH was purchased from Chem-Lab (Zedelgem, Belgium, abs.).

Solvent-based extraction

For comparison, conventional solvent extractions were performed in 30 mL Teflon screw cap vials. The hemp quantity used in each extraction was 0.2 g. Two extractions were performed in triplicate using 2 mL solvent, more precisely, H2O and EtOH, for 2 h at 70 °C and a third one, also in triplicate, using 10 mL EtOH for 24 h and 70 °C. 70

Ionic liquid recovering

After extraction, the scCO2 device was depressurized, the metallic extraction reactor was disconnected and brought to room temperature. The IL–water–hemp mixture (Fig. S3 † ) was filtered to remove hemp particles, the water was evaporated in vacuo and the remaining ionic liquid was dried under vacuum (0.65 mbar) for 24 h. Afterwards 20 mg of purified IL (Fig. S4 † ) were dissolved in chloroform-d3 (Sigma Aldrich, St Louis, USA) and a 1 H-NMR was recorded with a 400 MHz Bruker Advanced Ultra Shield 400 spectrometer (Bruker, Billerica, USA). Spectroscopic data and NMR spectra are given in the ESI (Table S4 and Fig. S5–7 † )

Cannabinoid quantification

The determination of CBDA, CBD, CBGA, CBG, THCA, Δ 9 -THC, was carried out on a High-Performance Liquid Chromatography (HPLC) in a Dionex UltiMate© RSLC System, with DAD-3000RS Photodiode Array Detector (Thermo Scientific, Germering, Germany), on a Dionex Acclaim™ RSLC 120 C18 (2.2 μm, 120 Å, 2.1 × 150 mm, Bonded Silica Products: no. 01425071, Thermo Scientific, Germering, Germany). A mobile phase flow rate of 0.2 mL min −1 was employed and the oven temperature was set to 25 °C. As a mobile phase, H2O with 0.1% formic acid (A) and acetonitrile with 0.1% formic acid (B) were used. The following gradient was carried out: 2 min of pre-equilibration at 70% B, 6 min hold at 70% B, 6 min from 70% B to 77% B, 18 min hold at 77% B, 0.5 min from 77% B to 95% B, 1.5 min at 95% B, 0.5 min from 95% B to 70% B, and 5 min at 70% B. 71 Acetonitrile was purchased from VWR Chemicals (Radnor, PA, USA) and formic acid from Merck (Darmstadt, Germany). All solvents for HPLC were of analytical grade.

The cannabinoid standards CBD, CBDA, THCA, Δ 9 -THC, CBG and CBGA were provided by Medical Cannabinoids Research and Analysis GmbH (Brunn am Gebirge, Austria) in the course of previous joint research. A mixed cannabinoid stock solution (1 mg mL −1 ) in MeOH of the investigated cannabinoids diluted for calibration.

Statistical analysis

Statistical data analysis was performed with Origin 2021. One-way ANOVA for multiple groups, followed by Tukey honestly significant difference (HSD) post hoc test at the 0.05 significance level, was carried out.


The authors would like to point out that the focus of this study was the extraction of cannabinoids as a class, not THC specifically. Any THC extraction is purely incidental, and bound to be negligible, given that industrial hemp was used, which in the EU must have a THC content not in excess of 0.2%.

In particular, we refer to Article 32, paragraph 6.

Additionally, the authors do hold a licence to for the purposes of research, in accordance with Austrian law, available under:

Author contributions

C.K. & A.S.M.: conceived the research, designed and performed the experiments, analysed the data, wrote the original draft, edited and reviewed the manuscript. M.S.: supervised the research, edited and reviewed the manuscript. K.S. & H.H.: conceived and supervised the research, designed the experiments, edited and reviewed the manuscript.

Conflicts of interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this article.

Supplementary Material



The authors acknowledge TU Wien for the Open Access Funding Programme of TU Wien Bibliothek for financial support and for the funding of the Doctoral College “Bioactive” (, Christian Löfke (Biobloom, Apetlon, Austria) for kindly providing the plant material, Renate Paltram for the technical assistance during cannabinoid quantification and Kristof Stagel for the support of recovering ionic liquids. This project has also received funding from the European Research Council (ERC) under the Horizon 2020 research and innovation programme (Grant agreement No. 864991)


† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1gc03516a

Gemering a weed seed

Our systems have detected unusual traffic activity from your network. Please complete this reCAPTCHA to demonstrate that it’s you making the requests and not a robot. If you are having trouble seeing or completing this challenge, this page may help. If you continue to experience issues, you can contact JSTOR support.

Block Reference: #ba45ddaa-e054-11ec-a20a-6f5952754f6f
VID: #
Date and time: Mon, 30 May 2022 20:11:44 GMT

©2000- ITHAKA. All Rights Reserved. JSTOR®, the JSTOR logo, JPASS®, and ITHAKA® are registered trademarks of ITHAKA.

Metabolic Pathway of Topramezone in Multiple-Resistant Waterhemp (Amaranthus tuberculatus) Differs From Naturally Tolerant Maize

This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.


Waterhemp [Amaranthus tuberculatus (Moq.) Sauer] is a problematic dicot weed in maize, soybean, and cotton production in the United States. Waterhemp has evolved resistance to several commercial herbicides that inhibit the 4-hydroxyphenylpyruvate-dioxygenase (HPPD) enzyme in sensitive dicots, and research to date has shown that HPPD-inhibitor resistance is conferred by rapid oxidative metabolism of the parent compound in resistant populations. Mesotrione and tembotrione (both triketones) have been used exclusively to study HPPD-inhibitor resistance mechanisms in waterhemp and a related species, A. palmeri (S. Wats.), but the commercial HPPD inhibitor topramezone (a pyrazolone) has not been investigated from a mechanistic standpoint despite numerous reports of cross-resistance in the field and greenhouse. The first objective of our research was to determine if two multiple herbicide-resistant (MHR) waterhemp populations (named NEB and SIR) metabolize topramezone more rapidly than two HPPD inhibitor-sensitive waterhemp populations (named SEN and ACR). Our second objective was to determine if initial topramezone metabolite(s) detected in MHR waterhemp are qualitatively different than those formed in maize. An excised leaf assay and whole-plant study investigated initial rates of topramezone metabolism (

Keywords: herbicide metabolism in plants, detoxification, triketone herbicides, pyrazolone, cytochrome P450, oxidative metabolism, herbicide resistance, HPPD inhibitor


Topramezone is a 4-hydroxyphenylpyruvate dioxygenase (HPPD)-inhibiting herbicide primarily used postemergence (POST) in maize (Zea mays L.) for broadleaf and grass weed control (Grossmann and Ehrhardt, 2007; Gitsopoulos et al., 2010). Herbicides that inhibit the HPPD enzyme cause sensitive plants to die by depleting plastoquinone, which in turn leads to depletion of tocopherols, carotenoids, and eventual bleaching of leaf tissues and cell membrane damage (Hess, 2000; Pallett et al., 2001; Ndikuryayo et al., 2017). Maize possesses natural tolerance to topramezone via rapid oxidative metabolism of the parent compound, specifically an N-demethylation reaction, which is presumably catalyzed by cytochrome P450 monooxygenase (P450) enzyme activity (Grossmann and Ehrhardt, 2007). Recent field and greenhouse studies reported resistance to several POST HPPD inhibitors in waterhemp (Amaranthus tuberculatus) and Palmer amaranth (A. palmeri), including topramezone (Hausman et al., 2011, 2016; Oliveira et al., 2017; Heap, 2018), as well as resistance to the photosystem II inhibitor atrazine by distinct metabolic mechanisms (Ma et al., 2013).

Several published reports indicated enhanced oxidative metabolism of either mesotrione (Ma et al., 2013; Kaundun et al., 2017; Nakka et al., 2017) or tembotrione (Küpper et al., 2018) contributes significantly to whole-plant resistance levels relative to HPPD inhibitor-sensitive populations. Since these two herbicides belong to the triketone subfamily of HPPD-inhibiting herbicides (Figure ​ (Figure1) 1 ) (Lee et al., 1998; Ndikuryayo et al., 2017), it is not surprising that metabolic resistance in Amaranthus populations proceeds via 4-hydroxylation of the cyclohexanedione ring, which is the same mechanism underlying maize tolerance and selectivity (Hawkes et al., 2001). Mechanistic research investigating topramezone metabolism has not been reported in multiple herbicide-resistant (MHR) Amaranthus populations, yet studying topramezone detoxification in MHR plants is of great interest since topramezone belongs to the pyrazolone subfamily of HPPD inhibitors (Figure ​ (Figure1) 1 ) (Siddall et al., 2002; Grossmann and Ehrhardt, 2007; Ndikuryayo et al., 2017). It remains to be experimentally determined whether populations resistant to HPPD-inhibiting herbicides mimic maize by detoxifying topramezone by N-demethylation (Grossmann and Ehrhardt, 2007) or via ring/alkyl hydroxylation at a liable position, as is the case for mesotrione (Hawkes et al., 2001; Ma et al., 2013).

Representative structures from different subclasses of commercial HPPD-inhibiting herbicides. (A) Mesotrione, a triketone. (B) Tembotrione, a triketone. (C) Pyrasulfotole, a pyrazolone. (D) Topramezone, a pyrazolone.

The two MHR waterhemp populations studied to date (MCR/SIR from Illinois and NEB from Nebraska; both HPPD inhibitor and s-triazine resistant) utilized foliar-applied mesotrione to investigate degradation rates and identify metabolites compared with sensitive populations (Ma et al., 2013; Kaundun et al., 2017). However, although both populations also exhibit resistance to POST topramezone (Hausman et al., 2016; Kaundun et al., 2017), only the SIR population had prior exposure to the pyrazolone topramezone in the field (Hausman et al., 2011). These differences in field-use histories of HPPD inhibitors between populations previously led us to speculate that mesotrione and/or tembotrione may have selected for cross-resistance to topramezone via enhanced oxidative metabolism (Kaundun et al., 2017). However, since topramezone does not possess a cyclohexanedione ring as with the triketones (Figure ​ (Figure1), 1 ), we hypothesized that the same P450(s) catalyzing 4-hydroxylation of the dione ring of mesotrione and tembotrione in HPPD-resistant waterhemp might also catalyze N-demethylation of topramezone in MCR/SIR and NEB, similar to the initial detoxification reaction in tolerant maize (Grossmann and Ehrhardt, 2007).

As a result, the objectives of our current research toward further investigations of HPPD-inhibitor resistance mechanisms in waterhemp are two-fold: (1) determine if two MHR waterhemp populations metabolize topramezone faster than two HPPD inhibitor-sensitive populations, and (2) qualitatively determine if initial topramezone metabolite(s) formed in waterhemp are different than those in maize. Our results shed new light on the multigenic, complex inheritance patterns for HPPD-inhibitor resistance (studied with mesotrione only; Huffman et al., 2015; Kohlhase et al., 2018) by demonstrating that MHR waterhemp populations have the potential to evolve complex metabolic mechanisms leading to cross- and/or multiple resistance that might be herbicide-dependent, and may also differ from mechanisms in naturally tolerant cereal crops.

Materials and Methods

Plant Materials

The four waterhemp populations investigated in this research are the same as those described by O’Brien et al. (2018). Two are sensitive to HPPD-inhibiting herbicides (SEN and ACR) and the two others (SIR and NEB) are resistant to POST applications of mesotrione, tembotrione, topramezone, and atrazine (Hausman et al., 2011; Kaundun et al., 2017). The SIR population was sampled from the same field site as the MCR population described in Hausman et al. (2011). Hybrid corn (DKC 63-14 RR) was used for comparison with waterhemp.

Whole-Plant POST Herbicide Dose-Response Study

For conducting dose-response studies with topramezone in the greenhouse, approximately 30 seeds of each waterhemp population were directly sown in 10-cm diameter pots containing a commercial potting medium of sandy-loam soil. For corn, approximately 10 seeds were sown per pot. Each pot (comprising one replicate) was maintained in a greenhouse providing a 16/8 h photoperiod of 180 μmol m −2 s −1 with day/night temperatures of 24/18°C at constant 65% relative humidity.

When waterhemp plants were 7-cm tall and corn plants reached the 2–3 leaf stage, they were treated with topramezone (Armezon TM , BASF Corp., NC, United States) at 0, 0.1, 0.2, 0.39 0.78, 1.56, 3.13, 6.25, 12.5, 25, 50, or 100 g ai ha −1 using a track sprayer fitted with a Teejet nozzle calibrated to deliver 200 L ha −1 . All treatments included Agridex 1% (v/v) (Helena Chemical) as well as ammonium sulfate at 2.5% (w/v) as spray adjuvants. Following herbicide treatments, each pot (one replicate) was arranged within a randomized complete block design and maintained in the greenhouse as described above. Five replicate pots were used per herbicide treatment and population for SEN, SIR, NEB, and corn. Due to limited seed availability only three replicate pots were utilized for ACR. Pots were assessed for visual percent control compared to an untreated control at 21 days after treatment (21 DAT). Percent visual control of maize was zero across all rates of topramezone tested (data not shown) while typical dose responses were generated for each waterhemp population.

Topramezone Metabolism in Excised Waterhemp and Maize Leaves

Waterhemp seeds were suspended in 0.1 g L −1 agar:water solution at 4°C for at least 30 days to enhance germination. Seeds from each waterhemp population were germinated in 12 cm × 12 cm trays with a commercial potting medium (Sun Gro Horticulture, Bellevue, WA, United States) in the greenhouse. Emerged seedlings (2-cm tall) were then transplanted into 80 cm 3 pots in the greenhouse. When the seedlings were 4-cm tall they were transplanted into 950 cm 3 pots containing a 3:1:1:1 mixture of potting mix:soil:peat:sand. The soil component contained 3.5% organic matter with a pH of 6.8. Slow-release fertilizer (Osmocote, The Scotts Company, Marysville, OH, United States) was added to this mixture. Corn seeds were planted 2.5-cm deep in the same soil mixture. Plants at a height of 10–12 cm were transferred to a growth chamber for 24-h before conducting herbicide metabolism studies with either excised leaves or whole plants, as described below. Greenhouse and growth chamber (Controlled Environments Limited, Winnipeg, Canada) conditions were maintained at 28/22°C day/night with a 16/8 h photoperiod. Natural sunlight was supplemented with mercury halide lamps, providing a minimum of 500 μmol m −2 s −1 photon flux at plant canopy level in the greenhouse. Light in the growth chamber was provided by incandescent and fluorescent bulbs delivering 550 μmol m −2 s −1 photon flux at plant canopy level.

Excised leaves were prepared according to the protocol of Ma et al. (2015) with minor amendments as described below. On the day of the experiments, the fourth youngest leaves of waterhemp plants (one from each plant) were collected in a container with water, leaf petioles were cut again with razor blade under water and placed in 1.5 mL plastic tubes containing 200 μL of 0.1 M Tris-Cl buffer (pH 7.5) to equilibrate for an hour. Leaves were then transferred to new 1.5 mL tubes containing herbicide incubation solution [150 μM topramezone in 0.1 M Tris-Cl buffer (pH 7.5)]. The youngest corn leaf from 10–12 cm plants was processed as above with waterhemp leaves for comparison. Excised leaves were incubated in the 150 μM topramezone solution for 1 h to allow herbicide uptake, then washed with 0.1 M Tris-Cl buffer (pH 7.5) and placed in 1.5 mL tubes containing 500 μL of one quarter-strength MS salts liquid media. Leaves were harvested at 0 (immediately after the one hr incubation), 2, 5, 11, and 23 h after removal from the topramezone uptake solution [representing 1, 3, 6, 12, and 24 h after treatment (HAT)] by briefly rinsing in deionized water and drying with tissue paper. Tissue fresh weights were recorded, then leaves were placed in 2 mL tubes with screw caps and immediately frozen in liquid nitrogen. Frozen leaves were stored at −80°C until further analysis. Each plant population x time point consisted of four replicates (i.e., excised leaves from different plants) and three independent experiments were conducted. Only the relative concentrations of topramezone remaining at 6 and 24 HAT are reported, which were normalized to the average of the 1-h concentrations (separately for each population and repeated experiment) for the purposes of statistical analysis, as described in detail below.

Leaf samples were freeze-dried with a FlexyDry MP (FTS Systems, Stone Ridge, NY, United States) and ground to a powder with glass beads with a tissue grinder FastPrep FT120, (Savant Instruments Inc., Holbrook, NY, United States). The powder was extracted twice with 80% methanol (1 mL each time) on a rotary shaker at 23°C. The first extraction occurred overnight (at least 16 h) and the second extraction was for 4 h. After shaking, samples were centrifuged at 12000 × g for 10 min. Supernatants from the first and second extractions were combined and pellets resulting from each centrifugation step were discarded.

Sample Preparation and HPLC Analysis

For analysis of topramezone and its metabolites via reverse-phase (RP)-HPLC, a published protocol utilizing UPLC-MS/MS for detecting topramezone in soil, water and plant samples (Li et al., 2011) was modified and optimized for compatibility with waterhemp and maize leaves. Briefly, a 1 mL aliquot of plant extract was placed in a 1.5 mL plastic tube along with 100 μL of 100 μM pyrasulfotole in methanol as an internal standard. Organic solvent was evaporated to incipient dryness with a rotary evaporator (SpeedVac, Farmingdale, NY, United States) and 0.5 mL of 1N HCl (saturated with NaCl) was added followed by 1 mL of methylene chloride. Residue remaining after evaporation was re-dissolved by vortexing and samples were centrifuged at 12000 × g for 10 min. An aliquot (800 μL) of the lower methylene chloride layer was removed and placed in a new 1.5 mL tube, then 300 μL of 0.05% NH4OH was added and the sample was vortexed to extract topramezone and pyrasulfotole. Samples were centrifuged at 10000 × g for 10 min and the upper aqueous layer was carefully collected, stored overnight at 4°C, and subsequently used for RP-HPLC analysis as described below.

The HPLC system consisted of a Waters Alliance separations module (model 2695) equipped with a Waters 996 photodiode array (PDA) detector. Absorbances from 200–400 nm were initially measured with the PDA and 312 nm was selected to quantify parent topramezone and pyrasulfotole, the internal standard. Topramezone was resolved with a Brownlee SPP HPLC column (C18, particle size 2.7 μm, 4.6 mm × 100 mm; PerkinElmer). RP-HPLC was performed with binary mobile phases consisting of 5 mM ammonium formate in water:methanol (90:10) as mobile phase A and 5 mM ammonium formate in methanol:water (90:10) as mobile phase B at 30°C and a flow rate of 0.5 mL min −1 . Samples were loaded in an injection volume of 20 μL and analytes eluted with a gradient of 0–10% B in 10 min, 10–20% B in 5 min, 20–95% B in 4 min, and 95% B for 3 min (isocratic) to wash the column before returning to 0% B for 7 min to re-equilibrate the column prior to analyzing the next sample. For calculation of relative topramezone concentrations, a calibration curve was generated based on the ratio of topramezone to pyrasulfotole peak areas. The calibration curve had an R 2 value of 0.99.

Topramezone Metabolism in Waterhemp and Maize Plants

The third- and fourth-youngest leaves from waterhemp plants (or youngest leaf from maize plants) were treated with 1.5 mM topramezone [in 0.1 M Tris-Cl buffer (pH 7.5) containing 0.1% (v/v) Tween 20]. Each treated leaf received a total of 20 μL of 1.5 mM topramezone solution applied as ∼0.3 μL droplets with a Hamilton glass syringe. The total amount of topramezone applied corresponded with the amount of topramezone supplied in the incubation solution for the excised leaf experiment. At 24 and 48 HAT, treated leaves (two from each plant) were harvested (including the petioles), washed in 20% methanol to remove unabsorbed topramezone, fresh weights recorded, and leaves frozen in liquid nitrogen. Leaves were stored at −80°C until extraction and further analysis. Two independent experiments were conducted with either two or three replicates of each population × time point after treatment.

Treated leaves were pulverized in liquid nitrogen with a mortar and pestle, and topramezone and its metabolites were extracted with 80% methanol twice (5 mL each time) on a rotary shaker at 23°C. The first extraction occurred overnight (at least 16 h) and the second extraction was for 4 h. After shaking, samples were centrifuged at 12000 × g for 10 min. Supernatants from the first and second extractions were combined and stored at 4°C before HPLC analysis, and pellets from each centrifugation step were discarded. The internal standard (pyrasulfotole) was added to each experimental sample, concentrated, re-dissolved and partitioned as previously described. The upper (aqueous) layer was discarded and the lower (organic) layer was carefully collected. A silica solid-phase extraction (SPE) column (500 mg/3 mL loading capacity) was conditioned with 3 mL methylene chloride and the sample was applied to the SPE column, then sequentially washed with 3 mL of methylene chloride followed by 2 mL of methylene chloride:ethyl acetate (1:3) to remove phenolic acids, chlorophylls and pigments. Analytes were eluted from the column with 3 mL of 100% methanol. Methanol was removed with a rotary evaporator, and the flask was washed twice with methanol (0.5 mL each time). The solution was placed in a 1.5 mL plastic tube and methanol removed under a stream of nitrogen gas to dryness. The residue was re-dissolved in 300 μL of methanol, samples were centrifuged at 12,000 × g for 10 min, stored overnight at 4°C, and subsequently used for low and high resolution LC-MS analysis as described below. Relative concentrations of topramezone metabolites were determined as described previously for parent topramezone since authentic metabolite standards were either not known or commercially available.

LC-MS Analyses

Samples for low-resolution LC-MS were analyzed with an Agilent LC-MS (1100 HPLC with XCT Plus Trap mass spectrometer) in the Metabolomics Laboratory of the Roy J. Carver Biotechnology Center, University of Illinois at Urbana-Champaign. LC separation was performed with the same column, mobile phases, and gradient conditions as described above for RP-HPLC analysis of excised leaf extracts, except the flow rate was 0.4 mL min −1 . The autosampler was set to 15°C and the injection volume was 10 μL. Mass spectra were acquired under negative electrospray ionization (ESI) with dry temperature of 350°C, dry gas flow of 8.5 L min −1 , and nebulizer gas was set to 35 psi. Mass scan range was 120–900 m/z. For MS/MS detection, m/z 362 was selected as the precursor ion.

Samples for high-resolution LC-MS were analyzed using a Dionex Ultimate 3000 series HPLC system (Thermo, Germering, Germany) and Q-Exactive MS system (Thermo, Bremen, Germany) in the Metabolomics Laboratory of the Roy J. Carver Biotechnology Center, University of Illinois at Urbana-Champaign. Software Xcalibur version 3.0.63 was used for data acquisition and analysis. LC separation was performed with the same column described previously but with different mobile phases and separation conditions. The binary mobile phases consisted of 0.1% formic acid in water as mobile phase A or 0.1% formic acid in acetonitrile as mobile phase B. The flow rate was 0.5 mL min −1 and a linear binary gradient was utilized for analyte elution as follows: 0% B for 1 min; 0–60% B in 14 min; 60–100% B in 4 min, 100% B isocratic for 3 min, then the column was returned to 0% B for 8 min before loading the next sample. The autosampler was set to 10°C and the injection volume was 10 μL. Mass spectra were acquired under both positive (sheath gas flow rate, 50; aux gas flow rate: 13; sweep gas flow rate, 3; spray voltage, 3.5 kV; capillary temp, 263°C; aux gas heater temp, 425°C) and negative ESI (sheath gas flow rate, 50; aux gas flow rate, 13; sweep gas flow rate, 3; spray voltage, −2.5 kV; capillary temp, 263°C; aux gas heater temp, 425°C). The full scan mass spectrum resolution was set to 70,000 with the scan range of m/z 50 ∼ m/z 750, and the AGC target was 1E6 with a maximum injection time of 200 ms. For MS/MS scanning the mass spectrum resolution was set to 17,500. AGC target was 5E4 with a maximum injection time of 50 ms. Loop count was 2 and the isolation window was 1.0 m/z with NCE of 25 and 30 eV.

Statistical Analyses

Whole plant dose-response data for the waterhemp populations were analyzed by straight line regression analysis of logit-transformed visual percent weed control on the logarithm of the rate applied, with the slope of the fitted regression lines being identical for each of the populations (Streibig and Kudsk, 1993). GR50s and 95% confidence limits for each population were estimated from the fitted lines. Resistance indices relative to SEN were estimated as the ratio of the respective GR50s.

Relative concentrations of topramezone from both the excised leaf and whole plant studies were analyzed by analysis of variance using the linear model:

where yijkl denotes the measured (relative) concentration in replicate l of experiment i for population j at time k, μ is the overall true mean response, βi is the effect of experiment 1, πj is the effect of population j, τk is the effect of time k, (πτ)jk is the true effect of the population × time interaction and εijkl is the random ‘error’ associated with each individual response. Populations were compared separately at each time point using t-tests (α = 0.05) based on the error variance from this model.


Whole-Plant Dose-Responses to Topramezone Applied POST in the Greenhouse

Four waterhemp populations were subjected to dose-response analysis with topramezone in the greenhouse. Two are sensitive to foliar HPPD-inhibiting herbicides (SEN and ACR) and two (SIR and NEB) are resistant (Hausman et al., 2011; Kaundun et al., 2017; O’Brien et al., 2018). Maize hybrid DKC 63-14 RR was also included for comparison, but data are not shown since this hybrid did not exhibit visual injury symptoms at any topramezone rate tested (24 g ha −1 is a field-use rate in maize). SEN, ACR, and NEB were completely controlled at 25 g ha −1 , while the SIR population exhibited an approximate level of control of 20% (Figure ​ (Figure2). 2 ). At the lower rates of topramezone examined, ACR was the most sensitive population and NEB was less resistant than SIR. Although both SEN and ACR are sensitive to HPPD-inhibiting herbicides (O’Brien et al., 2018), only the SEN population was utilized to generate resistance indices (RIs) for NEB and SIR (Table ​ (Table1). 1 ). Each population displayed GR50 values well below the field-use rate of topramezone in maize, ranging from 7.3 g ha −1 for SIR to 0.2 g ha −1 for ACR (Table ​ (Table1). 1 ). In relation to the SEN population, calculated RIs were 9.9 for SIR and 3.1 for NEB.

Dose-response analysis of topramezone in four waterhemp populations in the greenhouse. Waterhemp plants (7-cm) from each population were treated with various rates of topramezone, including Agridex at 1% (v/v) (Helena Chemical) and ammonium sulfate at 2.5% (w/v) as spray adjuvants. Plants were assessed for visual percent control compared to an untreated control for each corresponding population at 21 days after treatment. Dose-response curves were generated as described in Section “Materials and Methods” and used to determine the 50% growth reduction and resistance index values listed in Table ​ Table1. 1 . SIR and NEB are HPPD inhibitor-resistant populations while ACR and SEN are sensitive to HPPD inhibitors (O’Brien et al., 2018).

Table 1

Quantitative dose-response analysis of four waterhemp populations based on topramezone rates that cause 50% reductions in plant growth (GR50) and resulting resistance indices.

Waterhemp population GR50 values a,b g ai ha −1 Resistance index c (relative to SEN)
SEN 0.74 (0.56–0.98) 1.0
SIR 7.34 (5.49–9.77) 9.86 (6.66–14.76)
NEB 2.28 (1.69–3.04) 3.06 (2.06–4.55)
ACR 0.17 (0.11–0.24) 0.22 (0.14–0.35)

Time-Course Analysis of Topramezone Metabolism in Excised Leaves From Four Waterhemp Populations and Maize

Previous studies of initial mesotrione metabolism rates using excised leaves with the HPPD-resistant MCR population (sampled from the same field site as the SIR population) determined rapid mesotrione metabolism within the initial 24 HAT, and a median 50% time for herbicide degradation (DT50) of 12-h was calculated for MCR (Ma et al., 2013). Interestingly, the DT50 calculated for maize (11.9-h) was almost identical to MCR. By contrast, the SIR population in the current research barely reached 50% topramezone degradation after 24-h while maize displayed a typical degradation curve expected during the time-course analysis (data not shown), achieving approximately 70% topramezone degradation at 24 HAT. As a result of the relatively slower rates of topramezone metabolism in each waterhemp population, only topramezone levels quantified from each population at 6 and 24 HAT are shown in Figure ​ Figure3 3 .

Metabolism of topramezone in four waterhemp populations and maize 6 and 24 hours after treatment (HAT) using an excised leaf assay. Waterhemp and maize seedlings (10–12 cm tall) were grown in the greenhouse and transferred to a growth chamber 24-h before conducting the excised leaf assays, as described previously for mesotrione by Ma et al. (2013, 2015). Excised leaves were incubated for 1-h in a 150 μM topramezone solution, then either harvested immediately or transferred to a dilute MS salts solution for the remainder of the time-course study. Relative concentrations of topramezone remaining in each excised leaf at 6 and 24 HAT (normalized to the average of the 1-h concentrations per population) are plotted on the Y-axis, which were determined with reverse-phase HPLC using pyrasulfotole as an internal standard as described in Section “Materials and Methods.” Treatment means significantly different (α = 0.05) than the SEN population mean are marked with asterisks.

As expected, topramezone levels in the two HPPD inhibitor-sensitive populations (SEN and ACR) were relatively high, ranging from approximately 80–90% at 6 and 24 HAT (Figure ​ (Figure3). 3 ). Since SEN and ACR were not different at either time point, only SEN was used as the sensitive population for statistical comparisons with SIR, NEB, and maize. Significant reductions in topramezone were determined for SIR, NEB, and maize at 6 HAT relative to SEN, while only SIR and maize displayed significant reductions at 24 HAT (Figure ​ (Figure3). 3 ). Lower levels of topramezone in SIR at both time points is consistent with the dose-response analyses (Table ​ (Table1 1 and Figure ​ Figure2), 2 ), indicating that rapid metabolism contributes to whole-plant resistance to topramezone in SIR. A significant reduction in topramezone levels in NEB at 6 HAT but not at 24 HAT is consistent with the intermediate level of whole-plant resistance to topramezone (relative to SIR and SEN) reported in Table ​ Table1 1 .

Quantification of Topramezone and Its Metabolites Formed 48 HAT in Treated Leaves of Waterhemp and Maize Whole Plants

Whole-plant studies were conducted to corroborate results from the initial experiments with excised leaves (Figure ​ (Figure3) 3 ) and to further investigate metabolism in treated leaves at later time points after topramezone application (both 24 and 48 HAT), as well as attempt to identify the nature of metabolite(s) formed in MHR waterhemp leaves. Topramezone levels in treated leaves did not differ significantly among waterhemp populations and maize at either time point (P = 0.30 for the overall population effect averaged across time; P = 0.36 for the population × time interaction), although the effect of time on topramezone metabolism was significant (P =

By contrast, lower amounts of topramezone remaining in excised leaves of SIR and maize at 24 HAT (Figure ​ (Figure3) 3 ) is supported by the greater abundance of several topramezone metabolites in SIR and maize described below and shown in Figure ​ Figure4. 4 . Initial topramezone metabolism via N-demethylation was reported in maize (Grossmann and Ehrhardt, 2007) and the long-term metabolic fate of topramezone has been determined in maize, wheat, and mustard greens (United States Environmental Protection Agency, 2005), but topramezone metabolism has not been reported in weedy species to date. Since radiolabeled topramezone and authentic metabolite standards were not available for our research, relative metabolite quantification and identification was determined via LC-MS using unlabeled topramezone. The pattern of metabolite abundances was not different between 24 and 48 HAT among waterhemp populations and maize treated leaves (data not shown), so only metabolites quantified and identified at 48 HAT are shown in Figure ​ Figure4 4 .

Quantification of topramezone metabolites in treated leaves from four waterhemp populations and maize 48 HAT using whole plants. The third- and fourth-youngest leaves from waterhemp and maize plants (10–12 cm tall) were treated with 60 × 0.33 μL droplets (20 μL total) of a 1.5 mM topramezone solution, including 0.1% (v/v) Tween 20 as a leaf wetting agent, using a glass syringe. Only the treated leaves were harvested from each plant at 24 and 48 HAT (only 48 HAT data are shown), extracted and partially purified by SPE chromatography, then analyzed by LC-MS (low resolution) as described in Section “Materials and Methods.” Relative concentrations of each topramezone metabolite extracted from the treated leaves are plotted on the Y-axis (authentic standards were not available), using pyrasulfotole as an internal standard as described in Section “Materials and Methods.” (A) Putative hydroxytopramezone-1. (B) Desmethyl-topramezone. (C) Putative hydroxytopramezone-2. (D) Putative benzoic acid metabolite of topramezone (only detected in maize). Vertical bars represent the standard error of the treatment mean.

Two major metabolites were identified and quantified in maize: N-demethylated (desmethyl) topramezone as previously reported (Grossmann and Ehrhardt, 2007) and a benzoic acid derivative presumably formed following cleavage of topramezone, which has also been reported previously in maize, wheat, and mustard greens (United States Environmental Protection Agency, 2005). The benzoic acid metabolite was not detected in any waterhemp samples while minor levels of desmethyl-topramezone were detected in NEB, SIR, and ACR (Figure ​ (Figure4). 4 ). Interestingly, two different putative hydroxylated forms of topramezone (hydroxytopramezone-1 and hydroxytopramezone-2) were identified and quantified in each waterhemp population and maize (Figures ​ (Figures4, 4 , ​ ,5); 5 ); in particular, hydroxytopramezone-1 was more abundant in SIR treated leaves relative to other populations.

Identification of hydroxytopramezone-1 and hydroxytopramezone-2 metabolites formed in treated leaves of multiple herbicide-resistant waterhemp (SIR) 48 HAT using low-resolution LC-MS. The third- and fourth-youngest leaves from waterhemp and maize plants (10–12 cm tall) were treated with 60 × 0.33 μL droplets (20 μL total) of a 1.5 mM topramezone solution, including 0.1% (v/v) Tween 20 as a leaf wetting agent, using a glass syringe. Only the treated leaves were harvested from each plant (48 HAT), extracted and partially purified by SPE chromatography, then analyzed by LC-MS (low resolution) as described in Section “Materials and Methods.” (A) LC analysis of compounds with m/z of 378 (M + 16). (B) LC-MS/MS spectrum of putative hydroxytopramezone-1 (RT = 6.8 min in A). (C) LC-MS/MS spectrum of putative hydroxytopramezone-2 (RT = 7.3 min in A).

Identification and Structural Analysis of Topramezone Metabolites Formed in MHR Waterhemp (SIR Population) and Maize 48 HAT

LC analysis of SIR extracts at 48 HAT showed that the two putative hydroxylated compounds derived from parent topramezone (m/z of 378) had similar retention times in the gradient utilized for metabolite separation. The compound eluting first (RT = 6.8 min) was tentatively labeled hydroxytopramezone-1 while the compound eluting later (RT = 7.3 min) was labeled hydroxytopramezone-2 (Figure ​ (Figure5A). 5A ). Subsequent LC-MS analysis (with relatively lower resolution; see high resolution below) of each compound revealed that hydroxytopramezone-1 displayed a distinctive fragmentation pattern yielding several informative ions [M – H] − to assist in determining its structure (Figure ​ (Figure5B), 5B ), while the fragmentation pattern of hydroxytopramezone-2 primarily yielded a major ion at m/z of 318 with limited further fragmentation (Figure ​ (Figure5C). 5C ). As a result, an additional LC-MS/MS analysis was performed with an instrument possessing higher resolution capability to provide additional structural information for metabolite identification.

The fragmentation pattern of hydroxytopramezone-1 with lower resolution LC-MS/MS had yielded ions at m/z of 360, 298, 236, 208, and 174.1 (Figure ​ (Figure5B), 5B ), which were present along with several additional ions at m/z of 208.042 and 78.984 using higher resolution LC-MS/MS (Figure ​ (Figure6A). 6A ). The fragmentation pattern shown in Figure ​ Figure6B 6B is proposed to account for each major m/z peak in Figures ​ Figures5B, 5B , ​ ,6A. 6A . If the molecular ion at m/z of 378.076 represents a hydroxylation of the isoxazoline ring, then loss of a water molecule leads to the fragment at m/z of 360.066. In the lower path shown for the loss of water from hydroxytopramezone-1 (Figure ​ (Figure6B), 6B ), subsequent loss of the sulfone-methyl group [M – SO2CH3] − (corresponding m/z of 78.984) leads to the fragment ion at m/z of 298.083 (Figure ​ (Figure6B). 6B ). In the upper pathway shown for loss of water from hydroxytopramezone-1 (Figure ​ (Figure6B), 6B ), loss of the pyrazolyl ring and carbonyl group leads to the fragment ion at m/z of 236.038 while the corresponding loss of carbon monoxide [M – CO] − from the isoxazole ring (Bouchoux and Hoppilliard, 1981) leads to the ion at m/z of 208.042. Alternatively, intramolecular sulfur dioxide elimination leads to the fragment ion at m/z of 174.054 (Figure ​ (Figure6B), 6B ), which was previously reported when analyzing photochemical degradation products of mesotrione (Chahboune and Sarakha, 2018).

Identification of hydroxytopramezone-1 metabolite formed in treated leaves of multiple herbicide-resistant waterhemp (SIR) 48 HAT using high-resolution LC-MS. The third- and fourth-youngest leaves from waterhemp and maize plants (10–12 cm tall) were treated with 60 × 0.33 μL droplets (20 μL total) of a 1.5 mM topramezone solution, including 0.1% (v/v) Tween 20 as a leaf wetting agent, using a glass syringe. Only the treated leaves were harvested from each plant (48 HAT), extracted and partially purified by SPE chromatography, then analyzed by LC-MS (high resolution) as described in Section “Materials and Methods.” Proposed fragment ions following the loss of –CO (Bouchoux and Hoppilliard, 1981), –SO2CH3 and intramolecular SO2 elimination (Chahboune and Sarakha, 2018) are supported by previous research investigating the fragmentation of isoxazole and mesotrione, respectively. (A) LC-MS/MS spectrum of hydroxytopramezone-1. (B) LC-MS/MS fragmentation pattern and proposed daughter ion structures.

By contrast, hydroxytopramezone-2 yielded only one major fragment ion at m/z of 318 (Figure ​ (Figure5C) 5C ) and 318.055 (Figure ​ (Figure7A) 7A ) via low and high-resolution LC-MS/MS, respectively, which was also present as a minor fragment ion in the hydroxytopramezone-1 high-resolution spectrum (Figure ​ (Figure6A). 6A ). The lack of further fragmentation indicates this fragment ion is unusually stable during the LC-MS/MS conditions employed, which is supported by the highly conjugated structures proposed in Figures 7B,C . In either scenario, the daughter ion at m/z of 318.055 would result from loss of an exact mass of 60.021 from the parent ion at m/z of 378.076. Based on the relative high stability of hydroxytopramezone-2 compared to hydroxytopramezone-1 during our LC-MS/MS conditions, we propose that hydroxylation occurs β to the oxygen in the isoxazoline ring in hydroxytopramezone-2 (Figure ​ (Figure7B; 7B ; leading to the stable fragment ion at m/z of 318.055) whereas hydroxylation occurs α to the oxygen in the isoxazoline ring (thus relatively more electrophilic carbon) in hydroxytopramezone-1 (Figure ​ (Figure6B). 6B ). However, the existence of the putative hemi-aminal (N-alkyl hydroxylation) metabolite (Figure ​ (Figure7C) 7C ) cannot be excluded at this point without further structural analyses and information.

Identification of hydroxytopramezone-2 metabolite formed in treated leaves of multiple herbicide-resistant waterhemp (SIR) 48 HAT using high-resolution LC-MS. The third- and fourth-youngest leaves from waterhemp and maize plants (10–12 cm tall) were treated with 60 × 0.33 μL droplets (20 μL total) of a 1.5 mM topramezone solution, including 0.1% (v/v) Tween 20 as a leaf wetting agent, using a glass syringe. Only the treated leaves were harvested from each plant (48 HAT), extracted and partially purified by SPE chromatography, then analyzed by LC-MS (high resolution) as described in Section “Materials and Methods.” (A) LC-MS/MS spectrum of hydroxytopramezone-2, indicating the stable fragment ion with m/z of 318.055. (B) LC-MS/MS fragmentation pattern and proposed daughter ion structures derived from the putative isoxazole ring hydroxylation metabolite. (C) LC-MS/MS fragmentation pattern and proposed daughter ion structures derived from the putative hemi-aminal (i.e., N-alkyl hydroxylation) metabolite.

While alkyl hydroxylation of organic substrates by P450 enzymes is a common reaction (Siminszky, 2006; Mizutani and Ohta, 2010; Urlacher, 2012), it is relatively uncommon for the intermediate hydroxylation product that occurs during heteroatom release (e.g., O,N-dealkylation reactions) to accumulate without proceeding further to loss of formaldehyde (Kreuz et al., 1996). For example, the presence of desmethyl-topramezone as a major metabolite in maize leaves (Figure ​ (Figure5) 5 ) is consistent with P450-catalyzed N-demethylation of the pyrazole ring (Grossmann and Ehrhardt, 2007). Formation of the putative hemi-aminal metabolite of topramezone in SIR leaves is thus biochemically less favorable than hydroxylation of the isoxazoline ring at either position (esp. in Figure ​ Figure7B), 7B ), but as mentioned previously the hemi-aminal metabolite cannot be excluded without utilizing additional structural analyses such as 1 H- 13 C-HSQC NMR.


Dose-response analysis demonstrated that the SIR population is more resistant to POST topramezone than the NEB population, which is supported by enhanced topramezone metabolism (Figure ​ (Figure3) 3 ) and metabolite formation (Figure ​ (Figure4). 4 ). The greater fold-resistance of SIR may be related to prior usage of topramezone to control the SIR population (along with mesotrione and tembotrione; Hausman et al., 2011), in contrast with NEB (Kaundun et al., 2017), and is also consistent with higher fold-resistance levels of SIR to mesotrione and isoxaflutole applied POST relative to NEB (O’Brien et al., 2018). Moreover, NEB was never pressured with topramezone in the field yet significant levels of resistance were observed, suggesting that the gene(s) selected by mesotrione and/or tembotrione confer cross-resistance to topramezone in the NEB population (Kaundun et al., 2017).

One particularly troublesome aspect of metabolism-based resistance in weeds (Yu and Powles, 2014) is the potential for developing cross-resistance to herbicides from unrelated site-of-action families (Preston, 2004). In the case of HPPD-inhibitor resistance in waterhemp, it is not yet known precisely how many genes govern multigenic resistance (Huffman et al., 2015; Kohlhase et al., 2018), or if one or several P450s contribute to resistance to mesotrione, tembotrione, topramezone, and isoxaflutole (Ma et al., 2013; Kaundun et al., 2017; Nakka et al., 2017; Küpper et al., 2018; O’Brien et al., 2018). In most reported cases of metabolism-based resistance in weeds, mechanisms for metabolic detoxification appear to mimic natural mechanisms for tolerance in crops, leading to the formation of identical metabolite(s) between resistant weed populations and tolerant crops (Holtum et al., 1991; Ma et al., 2013; Yu et al., 2013). Although the precise molecular mechanisms behind enhanced herbicide metabolism in resistant weeds remain unknown, a prevailing theory is that constitutively expressed genes encoding detoxification enzymes [such as glutathione S-transferases (GSTs) or P450s] are expressed at higher levels in foliar tissues (Yasuor et al., 2010; Iwakami et al., 2014; Evans et al., 2017; Dyer, 2018). Alternatively, enhanced GST activity (with atrazine as substrate) resulting from an increase in Vmax (Anderson and Gronwald, 1991) and kcat (Plaisance and Gronwald, 1999) was documented in metabolic atrazine-resistant Abutilon theophrasti. Our results demonstrate that resistant weed populations possess the potential to metabolize herbicide substrates in a different manner than tolerant crops, which further complicates studies aimed at unraveling biochemical and genetic mechanisms that confer metabolism-based weed resistance.

Proposed initial routes of topramezone metabolism, based on our current results, in MHR waterhemp (SIR population) and tolerant maize are depicted in Figure ​ Figure8. 8 . The basis for topramezone selectivity in maize is primarily via N-demethylation of the pyrazole ring (Grossmann and Ehrhardt, 2007), while mesotrione selectivity in maize (and resistance in waterhemp) proceeds via 4-hydroxylation of the cyclohexanedione ring (Hawkes et al., 2001; Ma et al., 2013). Our findings demonstrate that the SIR population initially metabolizes topramezone by isoxazoline ring/N-alkyl hydroxylation at a liable position, indicating a different route for initial topramezone metabolism than tolerant maize (Figure ​ (Figure8), 8 ), although it is not known if both N-demethylation and hydroxylation reactions are catalyzed by the same or different P450 enzymes (Frear et al., 1969; Siminszky, 2006; Grossmann et al., 2011; Hamberger and Bak, 2013; Munro et al., 2013). It is of great interest to determine whether one or multiple P450(s) detoxify the three main commercial HPPD inhibitors applied POST, leading to cross- or multiple-resistance, respectively. In the case of tolerant maize, a single P450 gene located on chromosome 5 (named Nsf1 for nicosulfuron tolerance-1; Williams et al., 2006) confers cross tolerance to multiple herbicides within the HPPD-inhibitor family, as well as herbicides from other site-of-action groups (Nordby et al., 2008; Williams and Pataky, 2010).

Proposed initial routes of topramezone metabolism in multiple herbicide-resistant waterhemp (SIR population) and tolerant maize. The benzoic acid metabolite of topramezone was only detected in maize, desmethyl-topramezone was a major metabolite formed in maize but minor in SIR, and both hydroxytopramezone metabolites were major metabolites detected in SIR but very low in maize. Dotted lines indicate two possible structures of the hydroxytopramezone-2 metabolite from Figure ​ Figure7A: 7A : the hemiaminal metabolite that likely forms transiently in maize (Kreuz et al., 1996; Grossmann and Ehrhardt, 2007) but possibly accumulates in the SIR population, and a putative isoxazole ring hydroxylation metabolite (positional isomer) that has not been previously reported in plants or animals (United States Environmental Protection Agency, 2005).


Our findings indicate that MHR waterhemp populations possess multiple genes encoding diverse metabolic enzymes that confer complex, herbicide-dependent, cross- or multiple resistance patterns, which may be influenced significantly by prior field-use histories. Potential linkages among the genes conferring HPPD-inhibitor resistance in waterhemp can be explored with segregating F2 lines (Huffman et al., 2015) to determine if resistances to mesotrione, tembotrione, and topramezone are actually examples of cross-resistance or multiple resistance (Yu and Powles, 2014). Further mechanistic studies are required to determine whether additional non-target-site resistance mechanisms might be involved in conferring HPPD-inhibitor resistance, in particular reduced cellular transport or whole-plant translocation, since mesotrione is systemic and resistance in waterhemp is a quantitative trait (Huffman et al., 2015; Kohlhase et al., 2018). This can be accomplished by investigating the relative movement of putative metabolically blocked, experimental triketones that are systemic in nature (Beaudegnies et al., 2009). There remains great interest in discovering new chemistries for commercial HPPD-inhibiting herbicides (van Almsick, 2009; Wang et al., 2015; Ndikuryayo et al., 2017; Li et al., 2018). As a result, selection pressures for HPPD-inhibitor resistance will continue to increase in natural weed populations, particularly with the impending commercialization of HPPD-resistant soybean varieties (Siehl et al., 2014), which necessitates novel, integrated management strategies to combat resistance due to metabolic detoxification mechanisms. Given the dissimilar structures and maize selectivity basis between mesotrione or tembotrione and topramezone, further research by our group will continue to investigate physiological mechanisms by which SIR is resistant to the pyrazolone herbicide topramezone relative to triketone chemistry.

Author Contributions

DR and SK conceived the work. DR designed and supervised the work and wrote the manuscript. AL and AH planned and performed the experiments and generated the data. EM performed the statistical analysis and provided advice on experimental design. DR, JM, and SK analyzed the data and interpreted the results. SK provided the seed. SK and JM revised the manuscript critically.

Conflict of Interest Statement

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.


We thank Dr. Rong Ma, Sarah O’Brien, and Olivia Obenland for assistance with growing plants in the greenhouse and growth chamber, Dr. Alexander Ulanov for his assistance with LC-MS at the UIUC Metabolomics Laboratory, and Seth Strom for help in preparing the final figures.


Funding. Syngenta Ltd. provided funding for this research and well as seeds from various waterhemp populations.