To identify potential transporter genes encoding ABC and multidrug and toxic compound extrusion (MATE) proteins, contig sequences from the NTSR blackgrass transcriptome were assembled (Tétard-Jones et al., 2018) and compared to transporter sequences deposited in the National Centre for Biotechnology Information (NCBI) database (https://www.ncbi.nlm.nih.gov/) using the online protein Blast (BlastP) analysis tool (https://blast.ncbi.nlm.nih.gov/Blast.cgi). The sequences of ABCC transporters were confirmed by amplifying the respective open reading frames (ORFs) by polymerase chain reaction (PCR) using cDNA prepared from NTSR blackgrass using specific primers from untranslated regions (UTR). The amplified fragments were cloned in pGEM-T Easy Vector (Promega) and the ORFs were sequenced using specific primers ( Supplementary Table S1 ). A topology model of the AmABC1-3 membrane proteins was based on the TOPCONS consensus sequence (Tsirigos et al., 2015) and models were visualised with Protter (Omasits et al., 2014), with NBD alignments generated with SEAVIEW software version 4.6.4 (Gouy et al., 2010) using Muscle (Edgar, 2004).

For phylogenetic analysis, alignments of novel ABC sequences from blackgrass and homologues from other species were generated within SEAVIEW software version 4.6.4 (Gouy et al., 2010) using Muscle (Edgar, 2004) and subsequently trimmed (trimAI software v.1.3) using the settings ‘gappyout’ (Capella-Gutíerrez et al., 2009), as accessed through the webserver Phylemon 2 (Sánchez et al., 2011). Trimmed alignments were used for infer maximum likelihood phylogenies with IQ-TREE software (Nguyen et al., 2015). Automatic model selection mode was used (Bayesian information Criterion selecting the model LG+G4+I+F) and branch support values calculated by Ultrafast bootstrap approximation. The phylogenetic tree was edited with iTOL tool (https://itol.embl.de/) as described in Letunic and Bork, 2021.

The herbicide resistance profiles of blackgrass populations used in this study were as previously described (Moss, 1990; Moss et al., 2007; Marshall et al., 2013; Sabbadin et al., 2017). The sources and resistance profile of each population are summarized in Table S3 . Blackgrass seeds were germinated and grown as described previously (Franco-Ortega et al., 2021).

Nicotiana benthamiana seeds were surfaced sterile with 5% sodium hypochlorite containing 0.1% (v/v) Tween 20 for 5 min. Seeds were washed (5x) with sterile deionised water and placed on MS media containing 3% (w/v) sucrose. Plates were kept in the dark at 4°C for 2 d before being transferred to a growth cabinet at 21 ± 1°C; 120 μmol m^–2 s^–1 light intensity, and a 16 h:8 h (light: dark) photoperiod for 14 d. Seedlings were transferred to plastic pots (10 cm diameter) containing John Innes Number 2 compost and maintained under the same environmental conditions.

A solution of aqueous monochlorobimane (MCB; 100 µM) and propidium iodide (PI; 50 µM) was pipetted on a microscope glass slide, with 5µM sodium azide (NaN₃) added to provide a negative control. Blackgrass roots from 5 d old seedlings were placed in the solutions under coverslips. Images were acquired by SP8 invert confocal microscope (Leica microsystem) using 20x lens at 5-, 10-, 15- and 20-minutes intervals with excitation at 442 nm for the glutathione-bimane (GSB) conjugate and 480 nm for the PI. Fluorescence intensity of the GSB in individual cells was determined using LASX software (Leica microsystem) with 20 cells (n = 20) individually quantified and experiments repeated twice with independent samples. Data were analysed by Student’s t-test, using SPSS 27 (IBM, Chicago, IL, USA).

Herbicides were obtained from Sigma Aldrich (Gillingham, UK) and stock solutions prepared in dimethyl sulfoxide (DMSO). Four- week-old (3-5 leaf) HS or NTSR blackgrass plants were treated with 40 μM mesosulfuron-methyl, or 85 μM clodinafop-propargyl, with 0.1% (v/v) DMSO used as a solvent control. After a 24 h treatment, meristem and leaf tissue were separately harvested from individual plants (n = 7), then flash-frozen in liquid nitrogen and stored at -80°C.

Total RNA was isolated from ~100 mg of finely ground plant tissue using a NucleoSpin RNA plant kit following the manufacturer’s protocol (Macherey-Nagel, Germany). cDNA was synthesized from 1 μg of RNA using an iScript cDNA synthesis kit (Bio-Rad, United Kingdom) in a 20 μl reaction volume. RT-qPCR was performed in Light Cycler 96 system (Roche, United Kingdom) in a total volume of 15 μl containing 3.75 μl of cDNA,7.5 μl of LightCycler FastStart DNA Master SYBR Green I (Roche, United Kingdom), and 0.6 μl of 10 μM forward and reverse gene-specific primers. The primer sequences used in this study are listed in  Supplementary Table S1 . The reactions were run in a three-step program including melting curve; pre-incubation at 95°C for 3 minutes; amplification over 45 cycles (95°C for 10 s, 60°C for 10 s, and 72°C for 20 s; and melting analysis from 65°C to 95°C). For normalization, specific primers directed toward glyceraldehyde-3-phosphate dehydrogenase from blackgrass (AmGADPH; accession number: JN599100) were used, with relative gene expression (2^–ΔΔCt) then calculated (Livak and Schmittgen, 2001; Pfaffl, 2001).

Synthetic peptides were prepared from gene-specific coding sequences of AmABCC1, AmABCC2 and AmGSTU2a respectively ( Supplementary Table S2 ) and used to generate specific antisera in rabbits, that were quality tested by ELISA (Agrisera, Vännäs, Sweden). To test the specificity of the resulting antibodies, blackgrass shoots (2-3 tiller) were homogenized in 50 mM Tris-HCl buffer p H 8.0 containing 2 mM EDTA, 250 mM sorbitol, 5 mm DTT, 0.6% (w/v) PVP (Sigma), 500 µL mL^-1 PMSF, and 10 µL mL^-1 of Protease inhibitor Cocktail (Sigma 9599) before centrifuging at 10,000g, 5 min, 4°C. The extract was then re-centrifuged (100,000 x g, 60 min, 4°C) to separate supernatants from microsomal pellets. Microsomal pellets were resuspended in 5 mM potassium phosphate buffer (pH 7.8) containing 330 mM saccharose, 3 mM KCL and 0.5% n-dodecyl-ß-D-maltopyranoside. After determining the protein concentration by Bradford assay (Bio-Rad, UK), 40 µg protein of the microsomal pellet (P), or supernatant (Sup), were loaded onto a 4-20% gradient gel (SDS PAGE, Bio-Rad, UK). After immunoblotting, the membrane was incubated with primary rabbit antisera raised to AmABCC1, AmABCC2 (dilution 1:10), or AmGSTU2a (1:500) overnight at 4°C. The membrane was washed with buffer before incubating with anti-rabbit HRP-conjugated secondary antibodies (Sigma-Aldrich), with immunoreactive polypeptides visualized by chemiluminescence using a ChemiDoc MP+ imaging system (Bio-Rad, UK).

The full length ORFs of AmABCCs were codon optimised and synthesised with the restriction sites (PacI and AscI; GeneArt, Thermo Fisher Scientific) for expressing in N. benthamiana. The constructs were cloned into pMDC83 vector (Curtis and Grossniklaus, 2003) and the insertion confirmed by sequencing. The AmABCC1-GFP and AmABCC2-GFP constructs were then transformed into Agrobacterium tumefaciens (GV3101:PM90v).

Overnight cultures of A. tumefaciens containing AmABCCs-GFP as prepared from single colony were diluted (1:25) in growth media containing 20 μM of acetosyringone and grown overnight. Pelleted bacteria were re-suspended in infiltration buffer (10 mM MgCL₂, 100 μM acetosyringone) to a final OD₆₀₀ = 0.5 and incubated at room temperature for 3 h. Cultures were mixed (1:1) with Agrobacterium containing the 35S:p19 plasmid to suppress gene silencing (Voinnet et al., 2003). For co-localisation studies, agrobacterium transformed with a plasmid encoding the specific vacuole membrane marker m-cherry (vac-rk; Nelson et al., 2007) were co-infiltrated with the AmABCC-GFP constructs. The vacuole marker plasmids were obtained from the Arabidopsis Biological Resource Center (Ohio, USA). Agrobacterium were infiltrated with a needleless syringe into N. benthamiana leaves. Plants were kept in a growth cabinet at 21 ± 1°C at a light intensity of 120 μmol m^–2 s^–1, and a 16:8 h (light:dark) photoperiod for 7d. The localisation of AmABCCs-GFP and marker proteins were visualised on a SP8 invert confocal microscope (Leica microsystem), with images processed and analysed using LASX software (Leica microsystem). GFP was excited at 458 nm and the signals were acquired between 468 and 558 nm. m-cherry was excited at 561 nm and acquired between 567 and 665 nm.

AmABCC1, AmABCC2 and AmGSTU2a coding sequences were synthesised after being codon-optimised for expression in Saccharomyces cerevisae (GeneArt, Thermo Fisher Scientific). For AmABCC1 and AmABCC2, the synthetic sequences were cloned into the NotI site of the pNEV plasmid. For AmGSTU2a, the sequence was cloned into the BamHI and XhoI sites of the pYES3 expression vector (Invitrogen). The yeast cadmium factor protein 1 mutant (Δycf1, MATα ura3-52 his6 leu2-3,-112 his3-Δ200 trp1-901 lys2-801 suc2-Δ, ycf1::hisG), lacking the respective ABCC transporter, was transformed with pNEV or pYES3 vectors containing either AmABCC1, AmABCC2 or AmGSTU2a coding sequences to generate pNEV-AmABCC1, pNEV-AmABCC2 and pYES3-AmGSTU2a respectively. Plasmids without insertions were transformed into Δycf1 and used as vector controls for all experiments.

For co- expression studies with the AmABCCs and AmGSTU2a, yeast cells were transformed with pNEV and pYES3 empty vectors (pNEV + pYES3), or vectors containing AmGSTU2a in combination with either AmABCC (pNEV-AmABCC1 + pYES3-AmGSTU2a, or pNEV-AmABCC2 + pYES3-AmGSTU2). Transformed cells were selected on minimal synthetic dropout medium lacking uracil (pNEV) or tryptophan (pYES3), supplemented with 2% (w/v) glucose at 30° on a shaker (190 rpm) for 16 h. Cells were diluted into fresh media (OD₆₀₀ = 0.15) before treatment with 40 µM CDNB, or 20 µM mesosulfuron-methyl. Culture OD₆₀₀ was measured at 4, 6 and 24 h following treatment. The % relative growth was determined from the OD₆₀₀ relative to cells grown in media alone. The ycf1 yeast mutant and pNEV vector were kindly provided by Francisco RM, University of Zurich, Institute of Plant Biology, Switzerland.

For transcript expression in Figures 2A–D , the data were analysed by analysis of variance (ANOVA) followed by Tukey HSD posthoc test using SPSS 27 (IBM, Chicago, IL, USA). For transcript expression in Figures 2E–H , the data were analyzed by analysis of variance (ANOVA) followed by Tukey HSD posthoc using R Studio. For Figure 3 , the spearman correlations of transcript expression were analysed by R packages (corrplot and ComplexHeatmap). For Figures 4 , 5 , 6 , the data were analyzed by one-way ANOVA followed by Tukey’s honestly significant difference (HSD) post-hoc test or Dunn test rstatix, FSA, multcom and rcompanion using R-Studio.

Enhanced translocation of conjugated xenobiotics into the vacuole by ABC and other transporter proteins is known to protect plant cells from the over-accumulation of toxic compounds in the cytosol (Hwang et al., 2016). As a primary investigation, we therefore investigated the vacuolar import of the model xenobiotic monochlorobimane (MCB) in NTSR, as compared with HS blackgrass. MCB is commonly used for studying the translocation and compartmentalisation of its fluorescent glutathionylated bimane (GS-B) conjugate in vivo (Coleman et al., 1997; Meyer and Fricker, 2002). GS-B is generated through the action of GSTs, with the conjugate serving as a ligand for ABC transporter proteins for vacuolar import (Coleman et al., 1997). As the accumulation of GS-B in the cytosol reduces the formation of further conjugates due to product inhibition of the GSTs, the formation and active translocation of GS-B can be monitored dynamically in vivo by fluorescence microscopy (Meyer and Fricker, 2002).

Primary roots of HS and NTSR blackgrass seedlings were exposed to 100 µM MCB and the formation of fluorescent GS-B metabolites spatially monitored and quantified using a confocal microscope over a 20 min time course ( Figures 7A–F ). The average fluorescent intensity in NTSR roots were significantly higher than those seen in the respective HS cells at all time points ( Figure 7G ; Student’s t-test, P _(5minutes) = 0.001, P _(10minutes) = 0.001, P _(15minutes) = 0.001, P _(20minutes) = 0.002). To investigate if the formation of the GS-B fluorescent signal was energy-dependent as would be anticipated if the action of ABC transporters were involved, sodium azide (NaN₃), an inhibitor of cytochrome c oxidase that suppresses ATP generation, was added in combination with MCB to HS and NTSR roots. A strong reduction of fluorescent signal was observed in both HS and NTSR roots in the presence of NaN₃, suggesting that active transport was indeed involved in GS-B accumulation ( Figure 7G ). Analysis of the images showed the greatest intensity of the signal in the NTSR root cells was associated with highly fluorescent bodies within the cells, suggestive of compartmentalisation. In contrast, the GS-B fluorescence in the HS root cells was lower at every time point than those in NTSR cells suggesting the feedback inhibition preventing further conjugation occurring, presumably due to the over accumulation of the conjugate in the cytosol. Overall, these results were consistent with NTSR blackgrass having an enhanced capacity to metabolise MCB to its fluorescent conjugate GS-B and accumulate the metabolite in vivo as compared with HS plants and that this process was energy dependent as a consequence of the associated ABC transporter activity.

Global transcript expression analyses of NTSR and HS blackgrass populations obtained previously (Tétard-Jones et al., 2018), were analysed for sequences derived from ABC proteins, as well as from other transporter families linked to xenobiotic transport. On the basis of their relative abundance, three contigs derived from ABC transporter subfamily C genes were found to be constitutively up-regulated in NTSR, as compared with HS, blackgrass populations ( Table 1 ). In addition, two members of ABC subfamily B (AmABCB1 and AmABCB2) and two (MATE) transporters, termed AmMATE1 and AmMATE2 were also identified ( Table 1 ).

Based on their relative enhanced expression in NTSR plants and sequence coverage, it was possible to assemble full-length DNA sequences representing distinct ABCC transporter proteins. The sequences of the assembled ORFs were then confirmed following their RT-PCR amplification from NTSR plants. To further confirm their identity, the full-length sequences were subjected to a web server analysis (TOPCONS) for topology prediction. All three AmABCC sequences had an orientation TMD-NBD-TMD-NBD with an additional N terminal transmembrane domain (TMD0) and a cytosolic linker region (L0) ( Figure 8 ), typical of the ABC-C subfamily (Biemans-Oldehinkel et al., 2006). The sequences of NBDs exhibited the typical motif for ATP binding and hydrolysis, termed the Walker A and B motifs respectively (Walker et al., 1982; Supplementary Figure S1 ). With this topology analysis confirming the three AmABCs belonged to subfamily C, they were subsequently named as AmABCC1, AmABCC2 and AmABCC3 respectively. It is noteworthy that the in-silico analysis using WolF PSORT and SignalP tools could not detect the sequence that could pinpoint the localization of the three transporters.

ABCC transporters are known to transport plant metabolites, including conjugated xenobiotics in several plant species (Bartholomew et al., 2002; Goodman et al., 2004; Francisco et al., 2013; Behrens et al., 2019). To explore the prospective functions of the newly identified AmABCCs in blackgrass, a phylogenetic analysis was conducted with the ABC sequences identified in black-grass and ABC homologues. The tree was constructed with all the annotated ABCC transporter sequences present in rice (Verrier et al., 2008). The three AmABCCs shared between 37% and 41% similarity in amino acid sequences to each other, with each clustered in distinct clades ( Figure 9 )

AmABCC1 clustered with grape ABCC1 (Vitis vinifera, VvABCC1) and maize MRP3 (Zea may; ZmMRP3). Both transporters have been reported to function as anthocyanin transporters (Goodman et al., 2004; Francisco et al., 2013). AmABCC2 formed a distinct monophyletic group with ABC proteins collectively annotated as ABCC8 transporters, comprising EcABCC8 (Echinochloa colona), GmABCC8 (Glycine max; soybean), ZmABCC8 (Zea mays) and OsABCC8 (Oryza sativa; rice). While the role of these transporters in endogenous plant metabolism is unknown, EcABCC8 has been shown to be responsible for contributing to glyphosate resistance in jungle rice (Pan et al., 2021). The related ZmABCC8 shared 98% amino acid similarity with ZmMRP1, a transporter protein induced by safeners that enhance herbicide detoxification in maize (Pang et al., 2012). Based on the high similarity of amino acid sequences between ZmABCC8 and ZmMPR1, these two proteins mapped to the same position in the phylogenetic tree. Therefore, ZmMRP1 was not shown in the tree. With respect to AmABCC3, this protein clustered with a largely uncharacterised transporter from wheat (Triticum aestivum) termed TaMRP1, which is induced following herbicide safener treatment (Theodoulou et al., 2003). This phylogenetic analysis suggested the prospective biological functions of ABCCs. Whereas AmABCC1 probably functions in endogenous metabolite transportation, AmABCC2 and AmABCC3 could potentially be involved in EMR based on the sequence similarity to transporters in other plant species linked to tolerance to xenobiotics and herbicides.

Over the course of multiple studies, the molecular basis of the NTSR phenotype has been partially defined in multiple blackgrass populations (Tétard-Jones et al., 2018; Franco-Ortega et al., 2021). Based on this analysis, three major types of NTSR have been described; NTSR1 derived in the field following exposure to multiple classes of herbicides; NTSR2 which arises when HS plants are repeatedly selected for tolerance to the herbicide pendimethalin and NTSR3, arising from selection using fenoxaprop-ethyl (Tétard-Jones et al., 2018; Franco-Ortega et al., 2021). While NTSR1 and NTSR2 populations exhibit a classic upregulation of detoxification enzymes consistent with EMR, NTSR3 plants are resistant due to a distinct mechanism that does not involve enhanced herbicide metabolism. To test for the association between enhanced expression of the different ABCC genes and the different NTSR traits, the transcript expression of each set of plants were subjected to qRT-PCR analysis to quantify the relative abundance of each transporter relative to an HS population ( Figure 2 ). For reference, the gene encoding the tau (U) class glutathione transferase AmGSTU2a was included, as the respective enzyme is highly active in detoxifying chloroacetanilide herbicides and showed a high level of correlation with the degree of EMR found in different blackgrass populations (Nandula et al., 2019). The relative expressions of AmABCC1 and AmABCC2 were found to be comparable in the NTSR1 and NTSR2 populations ( Figures 2A, B ; one-way ANOVA; P_(AmABCC1) = 0.99; P_(AmABCC2) = 0.98) and significantly higher than those determined in HS or NTSR3 plants ( Figures 2A , 9B ; one-way ANOVA; P < 0.05). In contrast, AmABCC3 relative expression was similar in all the blackgrass populations tested ( Figure 2C , one-way ANOVA; p > 0.05). This finding suggested that AmABCC1 and AmABCC2 were more likely to link to EMR than AmABCC3. It is intriguing that while phylogenetic analysis showed that AmABCC3 was related to ABCCs in other species that function in xenobiotic detoxification, the transcript expression of AmABCC3 were unaltered in NTSR blackgrass populations. This discrepancy prompted us to further confirmed the expression of AmABCCs in wider blackgrass populations.

To further assess quantitative links between levels of the expression of the AmABCC genes and the degree of resistance conferred by enhanced herbicide metabolism, a further seven field derived NTSR blackgrass populations were tested that had previously been characterised with respect to the relative degrees of EMR exhibited (Marshall et al., 2013; Sabbadin et al., 2017). While the ‘Kent’, ‘Velcourt’, ‘Suffolk’ and ‘Peldon’ populations show high levels of EMR, ‘LongC’ was classified as possessing medium EMR, with ‘Notts’ and ‘Hor’ primarily resistant due to TSR mechanisms alone (Sabbadin et al., 2017). Due to the genetic variation within these field-derived populations, the expression of each of the AmABCCs showed high degree of variation within population which lead to the non-normal distribution of transcript expression data. Using non-parametric analysis, levels of expression of all three AmABCC genes were not statistically different among the TSR populations (Nots, Hor) the HS and EMR populations (Kent, Velcourt, Suffolk, Peldon and LongC). However, the expression trend of AmABCC1 was shown to be elevated in the highly-EMR populations Velcourt, Suffolk and Peldon, while no enhancement was determined in Kent (high) or LongC (medium) relative to that determined in HS blackgrass ( Figure 2E ). In the case of AmABCC2, expression tended to increase in all EMR populations ( Figure 2F ), while the transcript abundance of AmABCC3 was comparable in all plants irrespective of their resistance status ( Figure 2G ).

As a point of reference, the studies of AmGSTU2a broadly demonstrated the validity of linking the transcript abundance of key detoxification genes with the degree of EMR exhibited ( Figures 2D, H ). With respect to the transporters, collectively these results were consistent with AmABCC1 and AmABCC2 being linked to EMR in herbicide resistant blackgrass. In contrast, the similar levels of expression of AmABCC3 observed in the blackgrass populations irrespective of their resistance status suggested that this transporter was not linked to NTSR or EMR. These results highlight the diversity of functions of ABCCs transporters among plant species that could not be predicted based on sequence similarity. Based on the results of these gene expression studies, the further analysis of the transporters and their role in EMR was focused on AmABCC1 and AmABCC2.

Our previous study reported the enhanced co-expression of genes encoding proteins with linked metabolic functions in NTSR blackgrass including those involved in herbicide detoxification (Tétard-Jones et al., 2018). As such, it was determined whether the transporters AmABCC1 and AmABCC2 were co-expressed with the genes encoding other transporters and detoxifying enzymes in the transcriptomes of different NTSR blackgrass populations (Tétard-Jones et al., 2018). The genes selected for correlative analysis included the ABC transporters from family B (AmABCB1, AmABCB2), the AmMATE1 and AmMATE2 transporters, the (phi F) glutathione transferases AmGSTF1, AmGSTF2 as well as AmGSTU2a.

Relative gene expression was determined by qRT-PCR, with AmABCC1 and AmBCC2 subjected to correlation analysis with the other six genes present in the meristem tissues of the seven field-derived herbicide resistant blackgrass populations. Analysis confirmed the positive correlations in the expression of AmGSTU2a with AmABCC1 (Spearman’s correlation, r₍₃₇₎ = 0.56, p < 0.05), and with AmGSTU2a and AmABCC2 (r₍₃₇₎ = 0.62; p < 0.05) ( Figure 3 ). Besides AmGSTU2a, the relative expressions of AmABCC1and AmABCC2 also positively correlated with AmGSTF2 and AmMATE2 ( Figure 3 ; Spearman’s correlation, r _{(A m ABCC1- Am GSTF2)} = 0.68, r _{( AmABCC2 -AmGSTF2)} = 0.77, r = _{( AmABCC1- AmMATE2 )} = 0.70, r = _{( AmABCC2 - AmMATE2 )} = 0.77; p = < 0.05). AmGSTF1 was also included in the correlation analysis as a previously defined biomarker of NTSR in blackgrass (Tétard-Jones et al., 2018; Comont et al., 2020). As anticipated, AmGSTF1 transcript expression was higher in NTSR/EMR populations ( Figure S2 ) and found to positively correlate with AmGSTU2a ( Figure 3 ; Spearman’s correlation; r₍₃₇₎ = 0.61, P < 0.05) and AmABCC2 ( Figure 2 ; Spearman’s correlation; r₍₃₇₎ = 0.38, P = 0.19). The positive correlation in the expression of the two AmABCCs and AmGSTU2a were particularly interesting, being consistent with coupled functions of a herbicide conjugating enzyme and transporters capable of rapidly removing conjugation products in EMR plants. In contrast, the correlation between AmGSTF1 and the AmABCC transporters was weaker than that determined with AmGSTU2a, suggesting any coupling in function was less direct.

Genes involved in xenobiotic detoxification are often induced when plants are exposed to chemical injury (Xu et al., 2015). As such it was of interest to determine if the AmABCCs and the functionally linked AmGSTU2a which were all constitutively highly expressed in EMR blackgrass, were responsive to chemical treatments. In the first instance, HS and NTSR1 (Peldon) plants were exposed to two herbicides with differing chemistries and modes of action, namely the sulfonylurea mesosulfuron-methyl and the aryloxyphenoxypropionate clodinafop-propagyl. These two herbicides were selected since they had been widely used to control blackgrass and are both subject to evolved NTSR in the field. With each treatment, the relative expression of AmABCC1, AmABCC2 and AmGSTU2a were determined in the leaf and meristem tissues of HS and NTSR1 plants 24h after foliage application of herbicides, or solvent control.

AmABCC1, AmABC2 and AmGSTU2a expression were significantly higher in NTSR blackgrass compared to those determined in HS blackgrass at the constitutive level (untreated plants). The significantly higher expression of these genes in NTSR (Peldon) blackgrass were also observed in the solvent control treatment ( Figures 4A–C ). In the HS plants, the 24 h exposure to mesosulfuron resulted in a major induction of AmABCC1 and AmABCC2 in the meristems, comparable to that determined in NTSR plants, but not in the leaves ( Figures 4A, B ; one-way ANOVA; p < 0.05). This enhancement in response to herbicide treatment was not observed in any tissue in the HS plants with the AmGSTU2a gene ( Figure 4C ). It is noteworthy that while herbicide treatment did not enhance the expression of AmABCCs or AmGSTU2a in meristems of NTSR plants, clodinafop treatment significantly induced the expression of AmABCC1 transcripts in NTSR plants ( Figures 4A–C ).

To study the expression of AmABCC1 and AmABCC2 proteins in blackgrass plants, polyclonal antibodies were generated to peptide sequences specific to each transporter. In addition, an antibody was raised against native AmGSTU2a (Nandula et al., 2019). All three antisera were then used to probe a membrane protein (P) and a total soluble protein (Sup) fraction derived from meristem extracts from HS and NTSR1 blackgrass. After separating by SDS-PAGE, the polypeptides present in the two fractions were blotted and tested with the respective antisera. The anti-AmGSTU2a-sera readily detected polypeptides with a relative molecular mass (MW) 24.5 kDa, which is typical for GSTU subunits, in the soluble fraction (Sup) in NTSR1 plants ( Figure 5B ). A polypeptide with a slightly higher MW was also detected with this antisera in the membrane fraction (P). The peptide antisera raised against AmABCC1 detected a polypeptide with a relative molecular mass ~150kDa, consistent with it being derived from an ABC protein, in the membrane fraction (P) of HS and NTSR 1 blackgrass ( Figure 5A ). Densitometric analysis showed that the signal associated with the band from NTSR1 plants was double that determined in the extracts from HS plants. In contrast, it was not possible to detect membrane or soluble polypeptides using the AmABCC2 antisera in these preparations, even though the antibody had been demonstrated to recognise its targeted peptide sequence with equal affinity to that raised against AmABCC1 in earlier ELISA tests. From this it was concluded that at the level of expressed protein, AmABCC2 is less abundant than AmABCC1 in NTSR blackgrass.

The subcellular localisation of the AmABCC proteins in planta, was studied by transiently-expressing each transporter C-terminally fused with green fluorescent protein (GFP) under the control of the cauliflower mosaic virus 35S promoter in the leaves of Nicotiana benthamiana. Both AmABCC-GFPs were co-expressed with the aquaporin tonoplast marker γ-TIP fused with m-cherry (Nelson et al., 2007). The resulting localisation of the fluorescent fusion proteins was then monitored by confocal microscopy. The signal from AmABCC2-GFP co-localised with the tonoplast marker in small vesiculated bodies ( Figure 5C ). However, the fluorescent signal from AmABCC1-GFP could not be detected, even though the signal from the tonoplast marker was clearly visible in the sample ( Figure S3 ). The localization of AmABCC2 in the tonoplasts of blackgrass were consistent with a role of actively importing transport toxic compounds into vacuoles as determined for related ABC transporters in other plant species (Liu et al., 2001; Jaquinod et al., 2007). It is noteworthy that while we could not confirm the localization of AmABCC1 in the tonoplast using a transient expression approach, the subcellular fractionation of the protein in blackgrass plants confirmed its association with membranes.

To investigate any cytoprotective effect of these transporters towards xenobiotics and herbicides, AmABCC1 and AmABCC2 were expressed in Saccharomyces cerevisiae Δycf1strain defective in the cadmium factor protein, an ABCC transporter. These Δycf1 cells are susceptible to xenobiotics including the model GST substrate 1-chloro-2,4-dinitrobenzene (CDNB), in a dose responsive manner (Li et al., 1996). As such, the protective effect of plant ABC transporters in removing toxic xenobiotics from the yeast cytosol for deposition in the vacuole, or extracellular transport can be functionally assessed following their heterologous expression in the Δycf1 mutants through relative growth assays.

Yeast Δycf1 cells were transformed with either AmABCC1 or AmABCC2 and along with vector only controls (pNEV), exposed to 40 µM CDNB and the effect on growth determined by densitometry over 24 h as compared with untreated cultures, whose cell density were taken as being 100% at each time point. The percentage OD600 in cultures expressing either AmABCC1, or AmABCC2, were significantly higher than the pNEV controls at 4 h and 6 h after treatment (one-way ANOVA, p < 0.05; Figure 6A ). By 24 h this protective was no longer seen with all the cultures showing around 30% growth inhibition.

CDNB undergoes S-glutathionylation in cells containing the tripeptide glutathione through a combination of spontaneous conjugation and GST-catalysed reactions. The positive correlation in the co-expression of both AmABCCs with AmGSTU2a in NTSR blackgrass ( Figure 3 ), and the inherently high activity of AmGSTU2a toward CDNB (Nandula et al., 2019), prompted us to test whether co-expression of this GST with the transporters could enhance the protection to CDNB by providing a catalyst for detoxifying this xenobiotic. It was then postulated that the resulting CDNB conjugates would be removed from the cytosol by the AmABCCs thereby accelerating the overall rates of detoxification. The CDNB toxicity trial was therefore re-run with AmGSTU2 co-expressed with the AmABCC transporters along with the vector only control and expression of the respective proteins (AmABCC1 and AmGSTU2) confirmed by immunoblotting with the respective antisera ( Figure 6E ). The results obtained showed no additive protective effect of co-expressing the GST with either transporter ( Figure 6C ). In addition, expression of AmGSTU2a alone also showed no protective effect toward CDNB ( Figure 6B ), though it was not possible to confirm that the transformed cells were accumulating the respective conjugate. Taken together these results suggested that the activity of the transporter alone was able to protect the Δycf1 cells, though it was unproven as to whether the AmABCCs were translocating the native xenobiotic, or its detoxification products.

To examine the protective effect of the transporters in the yeast assays in further detail the growth of the AmABCC-transformed Δycf1 cells were determined in the presence of 8 different herbicide chemistries used to control blackgrass, including pendimethalin, fenoxaprop-ethyl, mesosulfuron-methyl and glyphosate. Of the compounds tested only the sulphonyl urea mesosulfuron-methyl, inhibited the growth of Δycf1 cell growth ( Figure S4 ). This inhibitory effect was markedly reduced in the cells expressing AmABCC1 and AmABCC2 at the 4 h and 6 h timepoints, though it was not apparent at 24 h, potentially as a consequence of the cultures entering the stationary phase ( Figure 6D ).