A novel mechanism of herbicide action through disruption of pyrimidine biosynthesis

Significance

A combination of adverse events, particularly the proliferation of herbicide-resistant weeds, is limiting growers’ ability to achieve the crop yields necessary to satisfy the ever-increasing demand for food, feed, and fiber. One practice that can reverse this trend is to increase the number of molecular targets available to the grower to provide better solutions to combat weed resistance. Here, we describe a mechanism of weed control through disruption of plant de novo pyrimidine biosynthesis. The aryl pyrrolidinone anilide class of chemistry inhibits dihydroorotate dehydrogenase in the pathway, and crystal structures of the inhibitor bound to the enzyme define the nature of the interaction and the basis of herbicidal action.

Abstract

A lead aryl pyrrolidinone anilide identified using high-throughput in vivo screening was optimized for efficacy, crop safety, and weed spectrum, resulting in tetflupyrolimet. Known modes of action were ruled out through in vitro enzyme and in vivo plant-based assays. Genomic sequencing of aryl pyrrolidinone anilide-resistant Arabidopsis thaliana progeny combined with nutrient reversal experiments and metabolomic analyses confirmed that the molecular target of the chemistry was dihydroorotate dehydrogenase (DHODH), the enzyme that catalyzes the fourth step in the de novo pyrimidine biosynthesis pathway. In vitro enzymatic and biophysical assays and a cocrystal structure with purified recombinant plant DHODH further confirmed this enzyme as the target site of this class of chemistry. Like known inhibitors of other DHODH orthologs, these molecules occupy the membrane-adjacent binding site of the electron acceptor ubiquinone. Identification of a new herbicidal chemical scaffold paired with a novel mode of action, the first such finding in over three decades, represents an important leap in combatting weed resistance and feeding a growing worldwide population.

Accumulating resistance to crop protection chemistries stands as one of the greatest threats to future global food security (1), creating a need for novel and efficacious molecules that act at new or underexploited target sites. In the area of weed control, decades have passed since the last novel mode of action herbicide was discovered, limiting the tools by which food producers combat an increasingly resistant weed landscape (2). While some cellular processes, such as light harvesting, have yielded multiple viable protein targets, much of the chemical space around known scaffolds has been well explored (e.g., ureas, auxinics, triketones). Discovery of novel mechanisms outside of these overexploited targets and chemotypes remains a challenge for the entire weed control industry (3, 4). While uncovering a previously unidentified mode of action for known chemistry (5) or developing newly efficacious molecules for a previously underutilized mode of action (6, 7) represent an important step toward this goal, the combination of novel herbicidal chemistry with a novel mode of action is of heightened need within the agrochemical sector.

One approach in the discovery arsenal for finding new active molecules has been to acquire many compounds of diverse structure and rapidly identify molecules with interesting activity using screening campaigns on suitable whole organisms in replicated assays. The progenitor of the candidate molecule (Fig. 1A) was identified from high-capacity screening efforts on whole plants which yielded a weakly active compound whose pyrrolidinone structure and phenotypic profile did not match known modes of action (8). Here, we demonstrate that the optimized lead of this structural type, tetflupyrolimet (9), inhibits dihydroorotate dehydrogenase (DHODH) of the de novo pyrimidine biosynthetic pathway, representing a novel herbicide mode of action with enough potency to be of commercial importance (10). Pyrimidine biosynthesis is an essential process for creation of important biological macromolecules, most directly DNA and RNA, but is also involved in production of glycoproteins and phospholipids (11). In plants, the first two steps of de novo pyrimidine biosynthesis occur in the chloroplast after which the product, N-carbamoyl aspartate, is shuttled to the cytoplasm where dihydroorotase resides, producing dihydroorotate, the DHODH substrate (12). Both in plants and mammals, this type II DHODH is a flavin-dependent enzyme residing on the outer surface of the inner mitochondrial membrane (13). Reducing equivalents pass from dihydroorotate via the tightly bound flavin cofactor to a ubiquinone acceptor molecule that subsequently exchanges with the ubiquinol pool of the membrane and ultimately links DHODH enzymatic activity with oxidative phosphorylation. The remaining steps of the pathway leading to pyrimidine nucleotides occur in the cytoplasm.

Fig. 1.

Structures and biological activity of tetflupyrolimet (II) and analogs on plants. (A) Optimization of compound (I) that ultimately produced the candidate molecule (II) was accompanied by a significant improvement in efficacy and weed control spectrum. (B) Structures of other analogs studied in this report—note that compounds (I), (III), and (VI) were unresolved enantiomers. (C) The effect on Arabidopsis (Col-0) and (D) foxtail millet growth treated with (II) at 7 DAT (days after treatment) with the doses indicated. (E) Dose–response curves of (II) inhibition of Arabidopsis (blue) and foxtail (red). Calculated IC₅₀ values based on the root length measured from five independent biological replicates. Values are given as mean ± SD. (F) Distinct morphological symptoms of tetflupyrolimet-treated Arabidopsis plants compared to plants treated with VLCFA elongase inhibitors at 7 DAT. The scale bar indicates 0.1 cm.

Inhibition of DHODH has long been explored for treatments of disease states in which cells are hijacked to quickly produce biomass, such as cancers and viral proliferation (including COVID-19), or for immunosuppressive properties (14, 15). Use of DHODH inhibitors for malaria (16) and human fungal disease has also been proposed in recent years (17). In agriculture, there is one fungicide targeting DHODH commercialized for use in crops (ipflufenoquin, FRAC code 52), and with the rise in field resistance to older chemistries, such as strobilurins and triazoles, there is much interest in discovering new chemotypes and modes of action like pyrimidine biosynthesis, specific for controlling relevant pathogenic species. In plants, validation that DHODH might be a potential weed control target was found using antisense studies in Arabidopsis where transformants showed clear growth inhibition when transcripts were reduced by about 90% (18). In a more recent study, DHODH knock-down Arabidopsis RNAi lines showed delayed development in early growth stages followed by further retardation in both vegetative and reproductive phases (19). Studies of DHODH expression in Arabidopsis thaliana find that the gene is highly expressed during seed germination and early seedling development with delayed germination in lines transformed with interference sequences compared to the wild-type (WT) plant. The study reported no detectable effects on the appearance and functionality of chloroplasts based on the size, chlorophyll content, or photosystem II efficiency but did find morphological alterations in mitochondria organellar substructure consistent with the localization of the enzyme (19). Identification of this class of chemistry and subsequent identification of DHODH as the target has now further validated the enzyme as being intimately involved in the growth and development of plants and has also established inhibition of pyrimidine biosynthesis as a novel and viable mechanism for weed control. This new chemotype not only broadens the scope for exploration of interference of pyrimidine production in multiple applications but also the potential to offer food producers additional efficacious options with which to protect their crops. Tetflupyrolimet will soon be the only commercial herbicide combining novel chemistry with a novel mode of action to be introduced to growers in three decades and has been recognized as the first entry into the Herbicide Resistance Action Committee Group 28 for inhibitors of pyrimidine biosynthesis (20).

Results

Plant Growth Inhibition.

Fig. 1A shows the structure of the weakly active compound (I) that initially emerged from in vivo herbicide screening campaigns that, through optimization, including resolving all four enantiomers, ultimately produced the candidate tetflupyrolimet (II). Since a significant portion of the work was completed before the candidate molecule was optimized and identified, other analogs were also studied with respect to mode of action, especially compounds III-VI, including unresolved enantiomeric mixtures (Fig. 1B).

The effects of the chemistry on plant growth were elucidated by performing pregermination analyses with A. thaliana and Setaria italica (foxtail) seeds treated with tetflupyrolimet and related analogs. Clear growth inhibition of shoot and root length was observed during early seedling development (Fig. 1 C and D). Both Arabidopsis and Setaria seedlings showed dose-dependent responses to these treatments, and IC₅₀ values were generated based on the measured root length from Arabidopsis (17 ± 0.07 nM) and foxtail (3.3 ± 0.10 nM), respectively (Fig. 1E).

Based on symptoms of growth inhibition observed previously from many known herbicides, there was consideration that this class of chemistry that arrests the emergence of young germinating seedlings might be acting as an inhibitor of very long chain fatty acid (VLCFA) elongases. This proposal was discounted based on additional distinct morphological symptoms on shoot apical meristem and roots differentiated from VLCFA inhibitors (Fig. 1F), and, unlike elongase inhibitors, this chemistry was particularly safe to direct-seeded rice. Furthermore, testing a series of close analogs in an internal panel of in vitro enzymatic and cell-based assays designed to identify known sites of action produced no inhibition. These data were suggestive of the chemistry acting via a novel mechanism.

Identification of the Target of Pyrrolidinone Analogs through Forward Chemical Screening.

Screening ~24,000 ethyl methanesulfonate (EMS)-mutagenized Arabidopsis seeds with pyrrolidinone selection (SI Appendix, Fig. S1A) revealed two resistant variants, 45R1 and 60R1 (Fig. 2A and SI Appendix, Fig. S1B, photos for 45R1 shown). The first identified resistant biotype, 45R1, was subjected to a combination of PCR-based mapping and SNP (single nucleotide polymorphism) analyses following two rounds of backcrossing and self-crossing. The PCR mapping approach localized the resistance gene to a region of chromosome 5 (Fig. 2B) encompassing over 3,300 genes. We narrowed down these candidates to 41 genes by only considering nonsynonymous amino acid changes by transition mutation (G/C to A/T substitutions) because EMS, an alkylating agent of guanine, dominantly induces transition mutations in more than 99% of SNPs in Arabidopsis (21). After 41 genes were reduced to 8 candidates by considering their functionality and morphological symptomology (SI Appendix, Table S1), SNPs were confirmed by Sanger sequencing. Among this list, one gene encoded an antiproliferative enzyme, DHODH (At5g23300, DHODH) with a G198E (nucleotide G593A) mutation located in exon 4 of the DHODH gene (Fig. 2C). The second independent resistant line, designated 60R1, was found to also carry a nonsynonymous mutation, an A141T change (nucleotide G421A) located in exon 3 of the DHODH gene (Fig. 2C), thus providing strong evidence that DHODH was indeed the target of the chemistry. Both A141 and G198 are conserved among many plant species (SI Appendix, Fig. S2). The 45R1 line was cross-resistant to numerous active analogs including compound IV (SI Appendix, Fig. S1C). Interestingly, when homozygous 45R1 resistant mutant seedlings were grown on plain Murashige and Skoog salts (MS) media without compound treatments, the growth of mutant seedlings was relatively slower than Col-0 WT Arabidopsis seedlings (SI Appendix, Fig. S3). Such a fitness cost is possible for herbicide resistance mutations relative to the susceptible WT in untreated conditions (22). Herbicide resistance resulting from a mutation is occasionally offset by a high fitness-cost penalty and the fate of the herbicide resistance allele can be ultimately affected by the balance between herbicide resistance and fitness cost (23). Resistance management through exploitation of fitness cost might be one factor against the emergence of future tetflupyrolimet-induced resistant mutations in the field.

Fig. 2.

Identification of the target of (III) through forward chemical genetic screening. (A) Variant 45R1, the first resistant biotype selected in the presence of compound III (5 μM) compared to the response of Arabidopsis WT (Ler-0) plant to the same molecule 7 DAT. The scale bar indicates 0.1 cm. (B) A 45R1 mutation was mapped to the region highlighted in green between two PCR markers, ctr1 (At5g03730) and phyc (At5g35840) on chromosome 5. (C) The position of the nonsynonymous SNPs in the DHODH gene of the two independent resistant lines 45R1 and 60R1 resulting in the G198E and A141T amino acid replacements, respectively.

Rescue of Compound-Induced Growth Inhibition with Pyrimidine Biosynthesis Intermediates.

The ability to reverse the herbicidal effect of potent pyrrolidinone analogs on Arabidopsis growth by inclusion of pyrimidine pathway intermediates in minimal medium provided good evidence pyrimidine biosynthesis interference is the mode of action of this class of chemistry. When supplying two products of the pathway downstream of the DHODH reaction, uridine monophosphate (UMP) and orotate, seedling growth was rescued 4.8-fold with UMP and 4.3-fold with orotate compared to seedlings treated with only 5 µM tetflupyrolimet based on measured root length (Fig. 3 A and B). Inosine monophosphate (IMP), involved in de novo purine biosynthesis and used as a negative control, was ineffective in reversing the phenotype.

Fig. 3.

Confirmation of DHODH as a target of tetflupyrolimet (II) by nutrient reversal and transgenic approaches. (A) Exogenously supplied UMP (30 μM) and orotate (30 μM), two downstream products of DHODH in the de novo pyrimidine biosynthesis pathway, successfully rescued the inhibitory effect of the chemistry (5 μM) on plant growth at 7 DAT. (B) A bar graph using measured primary root length of seedlings shown in A with and without supplementary nutrients. (C) Increased resistance to tetflupyrolimet (5 μM) with representative transgenic Col-0 homozygous plants for WT-10 and MT-8, that express a WT or 45R1 DHODH variant, respectively. Values are given as mean ± SD. (D) Quantitation of the measured primary root length of seedlings shown in C and nontransgenic WT and 45R1 mutant plants used as references for comparison purposes in the presence of tetflupyrolimet (5 μM). Bars are shown for nontransgenic WT Col-0, 45R1 homozygotes (45R1), and Col-0 transgenic homozygotes for WT version (WT-10) and G198E mutant version of DHODH (MT-8) respectively. All experiments were performed with three independent biological replicates. The scale bar indicates 0.1 cm.

Decreased Susceptibility of Transgenic Arabidopsis Plants to Compound Treatments.

In order to confirm that growth inhibition symptoms upon compound treatment are attributable to interference in pyrimidine biosynthesis through DHODH inhibition, transgenic plants were generated by introducing the mutated DHODH gene with a G198E SNP from 45R1 into WT Arabidopsis (ecotype Col-0) under the control of CaMV 35S promoter. Ten T2 transgenic lines [designated as AtDHODH (A.thaliana DHODH) MT] were selected from the transgenic screen and AtDHODH MT-8 line was selected as representative by carrying out qRT-PCR experiments to detect the transcripts of the DHODH gene (44.1-fold more transcript than WT, SI Appendix, Fig. S4). The MT-8 line was studied for their ability to be grown in the presence of 5 µM of tetflupyrolimet in MS agar medium. AtDHODH MT-8 was indeed more resistant compared to nontransgenic WT (~5.4-fold) and had a similar level of resistance with 45R1 (~1.2-fold) based on measured root length (Fig. 3 C and D). Transgenic lines overexpressing the WT version of the gene encoding DHODH were also generated and a WT-10 line was analyzed as a representative of this version of transgenic plant that had about 40-fold more transcript than in nontransgenic WT plants (SI Appendix, Fig. S4). Based on measured root length, WT-10 was ~2.8-fold more tolerant to tetflupyrolimet treatment than the nontransgenic WT, yet less so when compared to 45R1 itself or the MT-8 transgenic line (Fig. 3 C and D). These results demonstrate that while WT gene overexpression can partially rescue the plant from compound treatment, the identified DHODH mutation in the 45R1 line was necessary for the strongest observed resistance in transgenic lines, again validating the site of action of this class of chemistry as DHODH.

Metabolite Studies Confirm that Active Analogs Impair De Novo Pyrimidine Biosynthesis.

Both a broad and focused study of the metabolic response of treated plants was used to determine the effect of aryl pyrrolidinone anilide analogs on de novo pyrimidine biosynthetic pathway intermediates in greater detail. For a non-target-based metabolic profiling approach, Arabidopsis plant samples were treated with three active pyrrolidinone analogs (II, IV, V) for 24 h along with inactive analogs at two rates (1 and 5 µM) and an untreated control. Total metabolic features were acquired from TOF-LC/MS (Time-of-Flight - Liquid Chromatography/Mass Spectrometry) analysis and further filtered into 3,644 metabolites based on the peak intensity and presence within biological replicates. Approximately 341 metabolites were assigned to specific chemical formulae and compound name based on the KEGG (Kyoto Encyclopedia of Genes and Genomes) database. Clustering analysis including principal component analysis (PCA) and heatmap revealed that metabolic data from samples treated with active analogs were readily differentiated from those with inactive analogs and controls (Fig. 4 A and B). A scatter plot was generated between active and inactive groups with fold-change and P values determined by an ANOVA (Fig. 4C). Significantly differentiated metabolites are listed in SI Appendix, Table S2 where several intermediates involved in the pyrimidine metabolism were found. Of note, N-carbamoyl-D,L-aspartic acid and 4,5-dihydroorotic acid, the substrates of dihydroorotase and DHODH, respectively, accumulated significantly in the treated samples (SI Appendix, Table S2). Additionally, two precursor amino acids of pyrimidine biosynthesis, L-glutamic and L-aspartic acid accumulated only in the treated samples. Another metabolic intermediate of uracil catabolism, 5-6-dihydrouracil, accumulated in plants treated with tetflupyrolimet, whereas it was not detected in the control samples. Accumulation of this intermediate suggests that the pyrimidine salvage pathway might have been activated in response to the impairment of the de novo pathway.

Fig. 4.

Metabolomic analysis of Arabidopsis seedlings treated with aryl pyrrolidinone anilide analogs. (A) Non-target-based metabolic analysis was conducted with three active analogs (blue circle) and two inactive analogs (red triangle) at two doses (1 µM; empty and 5 µM; colored) along with DMSO control (black square). The PCA plot was generated from all metabolic features across 12 treatments. (B) Heatmap with hierarchical phylogenetic trees: each line in the heat map represents an individual metabolic feature. The red color represents a decrease, and the green color represents an increase. (C) A scatter plot was generated with fold-change and P values between active and inactive groups with axes as mean group intensities. The blue and red color represents increased and decreased metabolic features respectively. Two notable de novo pyrimidine biosynthesis intermediates N-carbamoyl-D,L-aspartic acid and 4,5-dihydoroorotic acid are highlighted. (D) Target metabolite analysis: Absolute quantitation of dihydroorotate (orange) and N-carbamoyl-L-aspartic acid (blue) were determined in the samples treated with tetflupyrolimet (II). Values are given as mean ± SD.

With the subsequent emergence of tetflupyrolimet as the candidate molecule, a narrower target-based metabolomics approach was implemented, focusing solely on the pyrimidine de novo synthesis intermediates N-carbamoyl-aspartic acid and dihydroorotic acid. In normal conditions, these intermediates were barely detectable. However, consistent with earlier experiments, significant accumulations were detected in samples treated with tetflupyrolimet in a dose-dependent manner (Fig. 4D and SI Appendix, Fig. S5). The N-carbamoyl-aspartic acid accumulated to 16.70 ± 2.02 and 24.23 ± 12.15 ng/g fresh weight (FW) at 1 and 5 μM tetflupyrolimet, respectively, and dihydroorotic acid accumulated to 6.45 ± 13.19 and 22.02 ± 4.54 ng/g FW at these rates.

Tetflupyrolimet Acts upon Purified DHODH Orthologs In Vitro.

The potency of tetflupyrolimet was assessed on plant DHODH more directly via enzymatic assay of recombinantly expressed DHODH from Escherichia coli. As with well-characterized recombinant human constructs, the N-terminal transmembrane domain was removed from the plant sequences along with the mitochondrial localization signal sequence (24). C-terminal polyhistidine tags facilitated affinity chromatography to obtain purified material. Dose–response IC₅₀ values were generated for tetflupyrolimet against multiple DHODH orthologs: 4.3 ± 0.9 nM for S. italica; 21 ± 3 nM for Oryza sativa (Fig. 5A); 36 ± 14 nM for Zea mays; 109 ± 51 nM for A. thaliana; and 380 ± 24 nM for Homo sapiens. The nearly 100-fold difference in IC₅₀ between S. italica and human DHODH represents a desirable safety threshold of this class of chemistry at the target level. Due to its favorable production in the bacterial host combined with high in vitro affinity for the pyrrolidinone chemistry, the crop DHODH ortholog from rice (Oryza sativa, OsDHODH) was selected for follow-up biochemical studies.

Fig. 5.

In vitro confirmation of the target site. (A) Dose–response curves of tetflupyrolimet (II) inhibition of recombinant S. italica (black) and O. sativa (red) DHODH. Values are given as mean ± SD. (B) Surface plasmon resonance of increasing concentrations of tetflupyrolimet flowing over immobilized O. sativa DHODH. (C) Steady-state kinetic study of reaction rate dependence on quinone concentration at fixed concentrations of inhibitor fit to a mixed-mode inhibition model (black lines). (D) 2D ligand interaction plot of tetflupyrolimet with O. sativa DHODH based on the determined crystal structure. Residues colored by reaction type indicating hydrophobic (green), polar (blue), charged (purple), or pi-pi stacking (green line). (E) Cartoon representation of the inhibitor binding site. Residues relevant to the binding interaction are displayed (sticks) along with waters (red spheres) and hydrogen bonds (yellow dashes). (F) Overlap of tetflupyrolimet (green sticks) and brequinar (cyan sticks) (25) in the quinone binding pocket of rice (green) and mammalian (cyan) DHODH. (G) The arrow highlights the closer orientation of the N-terminal helices in the O. sativa cocrystal with tetflupyrolimet compared to mammalian DHODH bound to brequinar.

We next asked whether we could confirm interaction between tetflupyrolimet and DHODH more directly via biophysical study. Surface plasmon resonance of increasing concentrations of tetflupyrolimet interacting with covalently immobilized OsDHODH was fit to a k_on of (616 ± 96) × 10⁴ M⁻¹s⁻¹ and k_off of (151 ± 24) × 10⁻³ s⁻¹ (Fig. 5B). The resulting K_D of 25 ± 8 nM is highly comparable to the IC₅₀ value calculated in the dose–response enzymatic assay experiments.

Tetflupyrolimet Shows Competitive Binding with the Ubiquinone Substrate.

Numerous crystal structures demonstrate that DHODH inhibitors bind within the membrane-adjacent ubiquinone binding pocket (26). We implemented steady-state kinetic analysis to determine whether tetflupyrolimet competes with the binding of the natural substrate ubiquinone. Velocity measurements of OsDHODH at varying ubiquinone concentrations were taken using the colorimetric assay at multiple fixed inhibitor concentrations. As judged by R-squared and sum of squares, global fit to the data preferred a mixed-mode inhibition model over purely competitive (Fig. 5C). However, when fitting K_i, the value for the α modifier of 9.5 ± 2.4 indicates that pyrrolidinone is more apt to bind in the absence of substrate, suggesting that the inhibition is more competitive with respect to ubiquinone than strictly mixed. These data suggest that tetflupyrolimet binds in the ubiquinone site, analogous to known DHODH inhibitors.

Cocrystal Structure of Plant DHODH with Bound Tetflupyrolimet.

We determined the crystal structure of OsDHODH in complex with tetflupyrolimet (Fig. 5 D and E), confirming that the molecule binds competitively with respect to ubiquinone in the same membrane-adjacent binding pocket as human DHODH inhibitors such as brequinar (25) (Fig. 5F). There are two regions of the protein that largely define the binding site of all structurally characterized DHODH inhibitors. The most extensive region is characterized by the internal face of a helix-turn-helix at the N terminus of the enzyme, comprising residues 97 to 121 (rice numbering). Many of these residues make primarily hydrophobic contacts with the aniline and phenyl rings of the compound (Fig. 5 D and E). In contrast to the human enzyme, these two helices are closer in their relative spacing by ~2 to 3 Å in the rice enzyme, providing structural justification as to why ortholog specificity among the various DHODH inhibitor chemotypes is strong (26) (Fig. 5G). The cap of a second helix-turn-helix near the C terminus comprising residues 430 to 438 is the second major structural motif that makes extensive hydrophobic contacts with the inhibitor and an edge-face aromatic interaction with F433 (Fig. 5D). A short segment of the protein that forms the side of the pocket closest to the aniline involves a β-strand encompassing residues 188 to 192. A conserved arginine side chain (R190, rice numbering) in this region, which in the human enzyme is involved in a charge-charge interaction with the carboxylate of the inhibitor brequinar (25), in the plant enzyme interacts with the lactam carbonyl oxygen via a network of water molecules (Fig. 5E). Finally, side chains of F152 and M165 interact with the amide oxygen and trifluorotoluene moieties, respectively.

Resistant Mutants Impair DHODH Function.

We next sought to understand how the mutations at A141 and G198, used to identify DHODH as the target of the pyrrolidinone chemistry in A. thaliana, confer resistance. Based on an AlphaFold model of the AtDHODH enzyme (27) that aligns closely with our internally determined rice structure, A141 maps onto an FMN-adjacent loop region (SI Appendix, Fig. S6A). G198 initiates a helix and abuts loop regions that contact both FMN and the DHO substrate binding site. Intrigued that the FMN cofactor separates these mutant positions from the active site of DHODH where target-site resistance mutants might be expected to occur, we generated the A141T and G198E point mutants in both the A. thaliana and the equivalent positions in the O. sativa recombinant enzyme to study further. Expression levels were comparable to WT as judged by SDS-PAGE (sodium dodecylsulphate-polyacrylamide gel electrophoresis). Activity assessments indicate that these mutants display severely reduced dehydrogenase activity. Specific activity for AtDHODH of 3.8 ± 0.3 mmol/min/mg was reduced 140- and 290-fold in the A141T and G198E mutants, respectively. WT OsDHODH baseline activity was 33 ± 1 mmol/min/mg, but reduced 100- and 80-fold, respectively, for mutants at the equivalent positions. Gel filtration analysis of the affinity-purified material for OsDHODH (SI Appendix, Fig. S6B) indicates that the WT enzyme (predicted MW of 45.1 kDa) elutes as a monomer at 15.6 mL just before the 44 kDa standard, and coelutes with a strong FMN signal (A₄₅₀ trace). For the mutant proteins, a total or partial loss of FMN (A₄₅₀ trace) for the A141T and G198E mutants, respectively, (A. thaliana numbering) as well as apparent aggregation due to the appearance of a significant A₂₈₀ void peak indicates instability of the expressed protein. Early elution of the “monomer” peak also suggests that this material may not be properly folded. Loss of FMN binding and increased instability of DHODH rationalizes the impaired activity seen with the recombinant mutant proteins, and apparent fitness cost of the 45R1 resistant line compared to Col-0 (SI Appendix, Fig. S3).

Discussion

Identification that the 4-aryl pyrrolidinone anilides disrupt pyrimidine biosynthesis through inhibition of DHODH at commercial use rates will provide growers with a mechanism to control weeds, particularly in rice. This new chemotype with novel mode of action the first to be identified in three decades will combat emerging resistance to current commercial herbicides and has established the pyrimidine biosynthetic pathway as a bona fide herbicidal target (3, 12). The phenotypic response of Arabidopsis and Setaria to this class of chemistry is most apparent in the early stages of growth and development and thus the most effective application for weed control is pre-emergence (8). Presumably, the weaker performance through posttreatments might occur due to plants requiring less pyrimidine nucleotides from the de novo pathway but instead finding adequate supplies through the pyrimidine salvage pathway.

DHODH has thus proved to be attractive as a site of action not only for therapeutic applications in the pharmaceutical industry and pathogen control in agriculture, but now is also validated as an effective mechanism for weed control. Serious concern has been articulated recently about target site resistance that has emerged in human fungal pathogens due to agricultural use of pesticides with a common mode of action to that of clinical chemistry (28). Such considerations are relevant for tetflupyrolimet as DHODH inhibition has recently emerged as a potential remedy for pathogenic aspergillus species (17). Fortunately, overall homology between the plant enzyme and orthologs from fungal pathogens is ~40% sequence identity or less compared to the model weed S. italica, generally mirroring sequence conservation at the binding site as well (SI Appendix, Table S3). Tetflupyrolimet shows no activity against various Aspergillus species in vitro (29), and applied at field use rates displays no utility against fungal crop diseases. Likewise, fungal DHODH inhibitors like ipflufenoquin are ineffective as herbicides.

DHODH is the only enzyme of the pyrimidine biosynthetic pathway residing in mitochondria, and additional functions outside of nucleotide biosynthesis may make DHODH a particularly efficacious target for weed control. The intimate link between DHODH and oxidative phosphorylation has been well established through shuttling of the common electron carrier, ubiquinone, even going so far as demonstration of a physical interaction between DHODH and Complexes II and III in mammalian systems (30). Tetflupyrolimet, like well-characterized mammalian DHODH inhibitors, may exploit this additional function, slowing oxidative phosphorylation to accentuate its effect on plant growth.

Although the chemistry is particularly potent in disrupting monocot growth and development, there was enough efficacy on A. thaliana that forward screening approaches with mutated seed collections successfully identified two independent resistant lines showing significant lack of sensitivity to the chemistry. The combination of PCR-mapping analysis, subsequent SNP detection by next-generation sequencing of RNA transcripts (RNA-Seq), and metabolomics ultimately pointed to class II DHODH (At5g23300) as a candidate worth detailed investigation. Transgenic experiments with resistant DHODH further confirmed DHODH as the target of pyrrolidinone chemistry. Production of the functional enzyme by recombinant means following isolation of the gene from numerous plant species and determination of the potency of inhibition was further confirmation of the target. Attempts to reconstitute the resistant SNPs in recombinant enzyme indicates that these changes are not well tolerated and yield poorly functioning DHODH in vitro, arguing that the pyrimidine salvage pathway may play a significant role in growth of these resistant lines or that additional oxidative phosphorylation functions of DHODH are disrupted by these same changes (30). Indeed, an apparent fitness cost of the G198E substitution is clearly present (SI Appendix, Fig. S3).

The crystal structure of rice DHODH with tetflupyrolimet provides structural insights into binding of another chemical scaffold to DHODH. To date, class II DHODH-inhibiting chemotypes display a remarkable degree of specificity given that chemotypes targeting human, plasmodial, fungal, and plant orthologs bind the same ubiquinone pocket (16, 26). The variability in both sequence (SI Appendix, Fig. S7) and conformation of helix α1 between weed (foxtail) and human enzyme (Fig. 5G) appears to result in a 100-fold IC₅₀ difference. Tetflupyrolimet is most effective against grasses over broadleaf plants. Within the binding pocket, the most intriguing sequence differences exemplified by the soy sequence appear to be F121W and M165L, for which the meta-trifluoromethyl group on the phenyl ring may not be optimized (Fig. 5B and SI Appendix, Fig. S7B). Binding of the inhibitor would be disrupted by a tryptophan sidechain occupying this particular position in the soy DHODH sequence. The F121-inward rotameric conformation appears to provide a unique interaction compared with the human enzyme (26) where the equivalent-positioned leucine residue orients away from the binding pocket. Among the grasses in the in vitro assay, tetflupyrolimet displays a ~fivefold IC₅₀ difference between foxtail and rice DHODH, likely contributing to an acceptable intrinsic safety factor preventing crop damage at typical use rates (9). While there are only three conservative changes in the inhibitor binding pocket between these two orthologs, L433 may more favorably interact with the phenyl moiety for this grass weed than F433 in rice (Fig. 5E and SI Appendix, Fig. S7). Global protein dynamics due to distal sequence differences among orthologs that are not captured in a single static structure may afford the greater than fivefold selectivity toward the intended weed species, along with metabolic and compound translocation differences between weed and crop (9). In all cases, an absolutely conserved feature of all determined class II DHODH structures is the location of the essential and tightly bound FMN cofactor that mediates the transfer of reducing equivalents from the dihydroorotate substrate to the ubiquinone acceptor and is disrupted by displacement of the quinone by the inhibitor (26).

Overall, this work highlights the power of forward genetic and metabolomic approaches for uncovering novel herbicide modes of action from whole-plant high-throughput screening efforts (9) and identifies the target for an example of herbicidal chemistry coupled with a new mode of action in three decades. Providing growers alternatives to overexploited molecular targets will help combat resistance and ensure that farmers can continue to sustain a growing world. Advances in screening methodology coupled with the increasing accessibility of -omic technologies provide the greatest promise for discovery of additional efficacious herbicide modes of action.

Materials and Methods

Plant Materials and Activity Tests.

A. thaliana ecotype Landsberg erecta (Ler-0) and foxtail millet (S. italica L.) plants were used for this study. Arabidopsis seeds were sterilized in chlorine gas and plated on 0.5× MS plant growth media, containing 0.05% (w/v) 2-(N-morpholino)-ethanesulfonic acid (MES), 0.5% (w/v) sucrose, and 0.8% (w/v) phytoagar. For foxtail, seeds were treated with 20% (v/v) commercial bleach solution containing 0.1% (v/v) of Tween-20 for 20 min followed by washing 3 to 4 times with sterile distilled water and germinated on the solid 0.8% (w/v) phytoagar plate after sterilization.

Activity tests with either plant involved direct application to the respective media described above. Seeds were grown in 24-well plates containing a series of compound concentrations (0.01, 0.1, 1, 5, 10, and 100 μM) in a 16-h/8-h day/night photoperiod with a light intensity of 120 to 150 µmol/m², 22 °C and 50% relative humidity in a growth chamber. At 7 d after treatment, the root length of five replicate seedlings in each treatment was measured from fixed orientation pictures using Image J software (Version 1.53j, NIH, USA), and IC₅₀ was determined using GraphPad Prism 7 software (GraphPad Software, San Diego, CA).

Screening for Resistant Mutants.

EMS-mutagenized M₂ seeds of Arabidopsis Ler-0 were purchased from Lehle Seeds (Round Rock, TX). About 12,000 sterilized M₂ seeds from the mutagenized stock were distributed in order to screen the resistant mutants as follows: seeds were sterilized in chlorine gas, sown on 0.5× MS containing 5 μM of (III), one of the potent aryl pyrrolidinone anilides, and stratified at 4 °C. After stratification for 48 h, the plates were transferred to a plant growth chamber and grown under the same conditions described for the activity tests above. Individual plants showing reasonably normal development after 7 to 10 d were isolated and grown in 4-inch pots containing Metromix360 potting soil. Herbicide resistance was confirmed in the next generation (M3) using the same concentration of compound as in the original selection. The originally identified 45R1 mutant is presented along with a schematic screening procedure (SI Appendix, Fig. S1A).

PCR-Based Mapping and SNP Analysis.

The mutated gene linked to the resistant trait in isolated 45R1 mutant was determined by conducting SNP analysis through next-generation sequencing (RNA-Seq) after PCR-based genetic analyses. In the first step, several rounds of crosses of the 45R1 mutant in Landsberg erecta (Ler-0, NW20) background to a Columbia (Col-0, N1092) accession were performed in order to reduce unlinked genetic background within the selected resistant mutant. The 45R1 F1 seedlings were generated by pollinating emasculated flowers of WT plants (Col-0) with pollen from the originally selected 45R1 mutant plants (Ler-0) in order to avoid accidental self-crossed resistant progenies. The resistant 45R1 F1 seedlings were identified by screening them on 5 µM of (III). In the same manner, the resistant 45R1 F2 population was created and identified from 45R1 F1 plants. The resistant trait for F1 and F2 population was confirmed by planting the seeds on MS media containing 5 µM of (III). Using standard protocols, genomic DNA was isolated from a single leaf of an F2 plant showing the resistant phenotype. The pooled DNAs were used as template for rough genetic mapping between Col-0 and Ler-0 Arabidopsis accessions and the PCR-based mapping analyses were generously carried out by the Genetic Discovery Group (DuPont Pioneer, Wilmington, DE). In this study, the most commonly used polymorphisms between Col-0 and Ler-0 were applied, and detailed information for the PCR-based genetic markers, including primer sequences, was obtained from ‘The Arabidopsis Information Resource’ (TAIR; www.arabidopsis.org).

For SNP analysis, lab-grown Arabidopsis seedlings of Ler-0 WT, Col-0 WT, isolated and selected resistant 45R1 F0, F1, and F2 population were collected at 2 wk after plating on 0.5× MS media and immediately frozen in liquid nitrogen. The frozen samples were used for RNA isolation, library construction, and next-generation sequencing (RNA-Seq) which was carried out by Genomics Lab at DuPont Pioneer (Johnston, IA). SNP detection analysis was carried out by the DuPont Bioinformatics Group (Wilmington, Delaware). In brief, total RNA from the collected samples was isolated and used to create Illumina TruSeq cDNA libraries. After sequencing using an Illumina HiSeq2500, images from sequencing run were analyzed via Illumina analysis pipeline and resulting sequences were filtered for quality control. High-quality sequences were aligned to TAIR10_cDNA reference database and SNPs were identified by using SAMtools (http://samtools.sourceforge.net/) packages. Among identified SNPs, we considered only SNPs satisfying three factors: the region on chromosome 5 from PCR-based mapping, transition mutations as an EMS signature, and nonsynonymous amino acid changes.

Nutrient Reversal Experiments.

Symptoms caused by tetflupyrolimet treatments in Arabidopsis were reversed by supplying specific intermediates of the biochemical pathway to the growth medium. Biosynthetic intermediates included UMP, OMP, and Orotate (5, 10, and 30 µM) which were applied to the wells of a 24-well plate filled with 0.5× MS media along with 5 µM tetflupyrolimet. IMP, a metabolite of purine biosynthesis, was also applied as a negative control. After 7 d of treatment, total growth of seedlings from three replicates was imaged and analyzed.

Generation and Analysis of Arabidopsis Transgenic Lines Overexpressing DHODH.

Amplification of cDNA encoding WT and resistant mutant of AtDHODH (At5g23300), was performed by RT-PCR on RNA extracted from leaves of WT and 45R1 resistant mutant plants, respectively. RNA was extracted using the Qiagen RNeasy kit following the manufacturer’s instructions (Qiagen). Aliquots of RNA (1 µg) were reverse transcribed using the SuperScript III First-Strand Synthesis System for RT-PCR (Invitrogen) following the manufacturer’s instructions. PCR for target cDNA fragments (1,383 bp) of AtDHODH was performed using the designed primers (5′-GGGGCAACTTTGTACAAAAAAGTTGATGGCCGG AAGGGCTGCGA -3′ and 5′- GGGGCAACTTTGTA CAACAAAGTTGTCATCTGTGATCAGCACCA -3′). GATEWAY entry clones were created using BP Clonase™II enzyme-mediated recombination (Invitrogen). Target gene fragments were further cloned into pCD3-857 (from Arabidopsis Biological Resource Center) binary vectors using LR Clonase™II enzyme mix (Invitrogen) according to the manufacturer’s protocol. The construct was verified by Sanger sequencing and introduced into Agrobacterium strain C58 (GV3101) by electroporation.

Arabidopsis plants (Col-0) were transformed using a modified floral dip procedure (31). After harvesting seeds from transformed plants, the surface-sterilized T1 seeds were plated and germinated on 0.5× MS medium containing hygromycin (20 µg/mL) to select transformed progenies. At least 10 independent T2 transgenic lines with a single T-DNA (transfer-DNA) insertion were isolated, and T3 homozygotes of three lines were selected and plated on MS medium with 5 μM of tetflupyrolimet in order to determine whether AtDHODH-overexpressing transgenic Arabidopsis lines were resistant to the chemistry. MT-8 transgenic line as a representative was selected and used for subsequent experiments.

In order to determine expression levels of transgenic AtDHODH in the transgenic lines, qRT-PCR experiments were performed in an Applied Biosystems QuantStudio 6 and 7 Pro real-time PCR System, and all reactions were done in three technical replicates. The transcript of DHODH of nontransgenic Arabidopsis WT (Col-0) plant was considered as the basal level, and ACTIN2 (At3g18780) was used as a reference gene for normalization. A transgenic line with only vector was also analyzed as a negative control.

Metabolite Profile of Treated Arabidopsis Seedlings.

Various active and inactive analogs of the chemistry were applied to 10-d-old Arabidopsis seedlings at two rates, 1 and 5 μM. Treated seedling samples were collected at 24 h after treatment and duplicate samples were analyzed per each condition. Approximately 100 mg of frozen samples were homogenized in a BeadBeater (Biospec, Mini-BeadBeater) with glass beads and extracted in 1 mL of a mixture of chloroform/methanol/water (1:2.5:1) for 30 min at room temperature. After spinning down the insoluble material, supernatants were transferred to fresh tubes and dried under nitrogen gas. Dried samples were dissolved either in 100 μL of 50% (v/v) acetonitrile−0.1% (v/v) formic acid for positive ion mode (ESI+) or 70% (v/v) acetonitrile−10 mM ammonium acetate for negative ion mode (ESI-) as described in ref. 32.

Mass spectroscopic analysis was performed using an Agilent 6520 Q-TOF LC/MS system using a Zorbax C18 column (4.6 × 150 mm) at a temperature of 40 °C. The mobile phases consisted of 0.1% (v/v) formic acid in water (solvent A) and 0.1% (v/v) formic acid in acetonitrile (solvent B). The flow rate of the mobile phase was 400 µL/min, and 10 µL samples were loaded per injection. Gradient conditions were 0 to 2 min hold 0% B, 2 to 7 min linear gradient 0 to 20% B, 7 to 10 min linear gradient 20 to 100%, 10 to 12 min hold 100% B, 12 to 13 min linear gradient to 100-0% B, and 13 to 15 min hold 0% B. Mass spectra were acquired from m/z 50 to 1,000 at an acquisition rate of 1.67 spectra/s in the Agilent Mass Hunter software and converted to mzXML format using MSConvert (ProteoWizard 3.0.18285). A nontarget metabolomics study was conducted to uncover primarily affected metabolites in the treated samples using Maven, a mass spectrometry data analysis software package (33). Significantly changed molecular features were profiled and searched against KEGG (https://www.genome.jp/kegg/) and METLIN (http://metlin.scripps.edu) with a 10-ppm mass accuracy threshold. Target metabolite analysis was conducted with standards purchased from Sigma (Sigma, Washington, MO). Peak areas were calculated using the Agilent Mass Hunter Quantitative software (Agilent Technologies) and manually adjusted if necessary. All subsequent calculations were performed with MS Office Excel software (Microsoft, Seattle, WA).

Expression, Purification, and Gel Filtration of Recombinant DHODH.

Codon-optimized coding sequences for E. coli expression (GenScript) were cloned into Isopropyl ß-D-1-thiogalactopyranoside (IPTG)-inducible bacterial expression vectors internally constructed by FMC. Mitochondrial signal sequences and the membrane-embedded N-terminal helices of DHODH were omitted from the coding sequences, and a C-terminal TEV-cleavable decahistidine affinity tag was included. Coding sequences for H. sapiens (Uniprot accession Q02127), Z. mays (B6U892), A. thaliana (P32746), S. italica (A0A368QCC1), and O. sativa (Q7XKC8) DHODH were initiated at sequence positions M29, D82, A72, E82, and G88, respectively. For the O. sativa sequence, the hexapeptide MADEAN precedes the indicated starting residue. Site-specific mutants were generated in the same manner.

DHODH expression plasmids were transformed into chemically competent BL21(DE3) pLysS cells (Invitrogen). A bacterial lawn on carbenicillin/agar plates was used to inoculate 3 L of LB, which was grown at 37 °C until the OD₆₀₀ reached 0.6, at which point the cultures were cooled and treated with 200 μM IPTG for overnight 16 °C induction. Cultures were pelleted by centrifugation and frozen at −80 °C for short-term storage. Pellets were thawed in 25 mL lysis buffer consisting of 25 mM potassium phosphate pH 7.4, 300 mM KCl, 100 μM FMN, 5% (v/v) glycerol, and 1× protease inhibitor cocktail (Roche). Following sonication on ice, subsequent purification steps were kept at 4 °C. Cell lysate was centrifuged 30,000 × g for 25 min. The resulting supernatant was filtered through a standard coffee filter and batch absorbed onto 0.5 mL of HisPur Cobalt Resin (Thermo Scientific). Resin was washed 3× with 3 resin volumes of wash buffer (lysis buffer in which protease inhibitor was replaced by 35 mM imidazole) and eluted with 3 resin volumes of elution buffer (wash buffer with imidazole increased to 500 mM). Buffer was exchanged over a PD10 desalting column (Cytiva) into storage buffer (identical to elution buffer but without FMN or imidazole) as instructed by the manufacturer. Protein concentration was assessed by the Bradford method (Bio-Rad), and aliquots were flash-frozen in liquid nitrogen for storage at −80 °C.

Gel filtration analysis was performed on an AKTA Pure FPLC system (Cytiva). High-molecular-weight standards obtained from a Gel Filtration Calibration Kit (Cytiva) were run on a pre-equilibrated Superdex 200 Increase 10/300 GL column (Cytiva) according to the manufacturer’s protocol using DHODH storage buffer as the running buffer. 0.5 mL of affinity-purified O. sativa DHODH (2.3 mg/mL), A141T mutant (A. thaliana numbering, 1.0 mg/mL), or G198E mutant (A. thaliana numbering, 1.8 mg/mL) were analyzed while monitoring simultaneously both A₂₈₀ and A₄₅₀.

Enzymatic Assay and Inhibition Kinetics.

Decylubiquinone solubilized in methanol (Santa Cruz Biotechnology, Inc.) was dried under nitrogen and reconstituted as a 2 mM stock solution in 5% (v/v) Brij 35 (Thermo Scientific) by sonication. This material was diluted to 0.1 mM into 195 μL of an assay solution consisting of 50 mM Tris-HCl pH 8.0, 150 mM KCl, 0.12 mM 2,6-dichlorophenolindophenol (DCPIP), 0.2% (v/v) DMSO (with or without tetflupyrolimet), and typically ~1 μg/mL of enzyme. After a 5-min preincubation at 25 °C, the reactions were initiated with 5 μL of a 40 mM dihydroorotate substrate in 50 mM Tris-HCl pH 8.0 solution, and A₆₀₀ was monitored for an additional 15 min in a Spectramax 384 Plus plate reader (Molecular Devices). Rates from 200 to 600 s in duplicate or triplicate were recorded. An extinction coefficient of 22,000 cm⁻¹ M⁻¹ for DCPIP was used to calculate specific activities. IC₅₀ and error values calculated from dose–response experiments were fit in Prism (GraphPad).

To determine the mode of inhibition of tetflupyrolimet via steady-state kinetics, reaction velocity was assessed while varying the concentration of decylubiquinone (Q) from 0 to 200 μM at multiple fixed inhibitor concentrations of 0, 3, 10, and 30 μM. The data were fit to the following equation for mixed-mode inhibition in Prism (GraphPad) where Vmax is the maximal reaction velocity, Km is the Michaelis constant of decylubiquinone, Ki is the inhibition constant, and α is a modifier of the inhibition constant specific to mixed mode inhibition.

$$V=\frac{Vmax\left[Q\right]}{Km(1+\frac{\left[II\right]}{\mathit{Ki}})+\left[Q\right](1+\frac{\left[II\right]}{\alpha Ki})}.$$

Rice DHODH Cocrystal Structure Determination.

Partially purified rice DHODH following the initial affinity chromatography step (see above) was buffer exchanged into 25 mM MES pH 6, 5% (v/v) glycerol, 200 mM KCl, and 5 mM 2-mercaptoethanol (ion exchange buffer). The sample was loaded onto a HiTrap Capto S column and eluted with a gradient from 0.1 to 1 M KCl while preserving the other ion exchange buffer components on an AKTA Pure FPLC system (Cytiva). Peak fractions were concentrated and subjected to gel filtration chromatography on a Superdex 200 column in a running buffer of 25 mM Hepes pH 7.5, 5% (v/v) glycerol, and 0.3 M KCl. Peak fractions were concentrated to 7.4 mg/mL and incubated with 100 μM tetflupyrolimet prior to setting trays. A sparse matrix screen was implemented which yielded crystals in a 0.1 M Bis-Tris pH 6.6, 20% (v/v) PEG 300 condition which were further optimized to obtain diffracting crystals. Prior to data collection, crystals were transferred to a reservoir solution containing 15% (v/v) glycerol and subsequently flash-frozen in a nitrogen cryostream. Data were collected at the BL17U beamline of the Shanghai Synchrotron Radiation Facility. Initial processing and scaling was performed with HKL3000. The structure of rat DHODH (PDB: 1UUM) provided a molecular replacement model for PHENIX. Iterative rounds of building from the initial model and refinement were performed using COOT, PHENIX, and Refmac5. A 2D ligand map and images of the final 3D model were generated with Maestro and Pymol (Schrodinger, LLC). Structure-based sequence alignment of DHODH was performed using the ESPript server (34).

Surface Plasmon Resonance.

A Series S Sensor Chip CM5 (Cytiva) was immobilized covalently using amine coupling with recombinant O. sativa DHODH on a Biacore T200 instrument (Cytiva). Protein sample was purified with the same method as the material used for crystallography but diluted to 25 μg/mL in 10% (v/v) DHODH storage buffer 90% (v/v) 10 mM acetate pH 5.5. In a running buffer of PBS-P⁺ (Cytiva), the amine coupling wizard in the Biacore software was used to execute a method in which protein flowed over the chip surface for 3 min at a flow rate of 5 μL/min resulting in a blank-subtracted final immobilized signal of ~5,100 RU. After priming and 6 blank injections performed for equilibration in a running buffer of PBS-P⁺ 2% (v/v) DMSO, twofold serial dilutions from 250 to 7.8 nM tetflupyrolimet in running buffer were injected onto the chip surface in increasing concentration at a flow rate of 20 μL/min. After each 30 s injection, the surface was re-equilibrated in running buffer for 3 min showing no significant carry-over between samples, and a needle wash in 50% (v/v) DMSO was performed. Blank-subtracted traces were fit to the 1:1 binding kinetics model using the Biacore Evaluation software package (Cytiva).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

Study data, Tetflupyrolimet, other materials and the underlying data to support the findings in this study (“Materials”) are available on request to the corresponding author (S.G.), such request being granted subject to 1) the recipient disclosing the use of the Materials, 2) the provider’s ability to provide such Materials; and 3) the recipient signing a confidentiality agreement and/or material transfer agreement, as applicable).