Mechanisms of feedback inhibition and sequential firing of active sites in plant aspartate transcarbamoylase

ARTICLE Mechanisms of feedback inhibition and sequential firing of active sites in plant aspartate transcarbamoylase Leo Bellin 1, Francisco Del Caño-Ochoa 2, Adrián Velázquez-Campoy 3,4,5,6,7, Torsten Möhlmann 1✉& Santiago Ramón-Maiques 2,8✉ Aspartate transcarbamoylase (ATC), an essential enzyme for de novo pyrimidine biosynthesis, is uniquely regulated in plants by feedback inhibition of uridine 5-monophosphate (UMP). Despite its importance in plant growth, the structure of this UMP-controlled ATC and the regulatory mechanism remain unknown. Here, we report the crystal structures of Arabidopsis ATC trimer free and bound to UMP, complexed to a transition-state analog or bearing a mutation that turns the enzyme insensitive to UMP. We found that UMP binds and blocks the ATC active site, directly competing with the binding of the substrates. We also prove that UMP recognition relies on a loop exclusively conserved in plants that is also responsible for the sequential firing of the active sites. In this work, we describe unique regulatory and catalytic properties of plant ATCs that could be exploited to modulate de novo pyrimidine synthesis and plant growth. https://doi.org/10.1038/s41467-021-21165-9 OPEN 1Pflanzenphysiologie, Fachbereich Biologie, Universität Kaiserslautern, Erwin-Schrödinger-Strasse, D-67663 Kaiserslautern, Germany. 2Instituto de Biomedicina de Valencia (IBV-CSIC), Jaime Roig 11, 46010 Valencia, Spain. 3Institute for Biocomputation and Physics of Complex Systems (BIFI), Joint Units IQFR-CSIC-BIFI, and GBsC-CSIC-BIFI, Universidad de Zaragoza, 50018 Zaragoza, Spain. 4Departamento de Bioquímica y Biología Molecular y Celular, Universidad de Zaragoza, 50009 Zaragoza, Spain. 5Instituto de Investigación Sanitaria de Aragón (IIS Aragón), 50009 Zaragoza, Spain. 6Centro de Investigación Biomédica en Red en el Área Temática de Enfermedades Hepáticas Digestivas (CIBERehd), 28029 Madrid, Spain. 7Fundación ARAID, Gobierno de Aragón, 50018 Zaragoza, Spain. 8Group 739, Centro de Investigación Biomédica en Red de Enfermedades Raras (CIBERER)- Instituto de Salud Carlos III, Valencia, Spain. ✉email: [email protected];[email protected] NATURE COMMUNICATIONS | (2021)12:947 | https://doi.org/10.1038/s41467-021-21165-9 | www.nature.com/naturecommunications 1 1234567890():,; Pyrimidine nucleotides are crucial to all living organisms as components of nucleic acids as well as cofactors in the synthesis of sugars, polysaccharides, glycoproteins, and phospholipids1,2. However, much remains unknown in plants about the unique organization, regulation, and localization of the enzymes required for de novo biosynthesis of pyrimidines1, which are potential targets for crop improvement and weed control. This metabolic pathway starts in the chloroplast, where aspartate transcarbamoylase (ATC) catalyzes the condensation of carbamoyl aspartate from aspartate (Asp) and carbamoyl phosphate (CP)1,3(Fig. 1a). ATC can be anchored to the inner plastid membrane4, which might facilitate the channeling of carbamoyl aspartate to a cytosolic dihydroorotase (DHO) potentially associated to the outer plastid membrane1. The dihydroorotate produced by DHO diffuses to the mitochondrial intermembrane space and is oxidized to orotate by dihydroorotate dehydrogenase (DHODH), a membrane flavoprotein coupled to the respiratory chain3,5. Orotate returns to the cytosol and is transformed to uridine 5-monophosphate (UMP), the precursor of all pyrimidine nucleotides, in a two-step reaction catalyzed by UMP synthetase (UMPS). Importantly, only in plants, UMP inhibits the activity of ATC by a yet unknown mechanism, creating a feedback loop that ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21165-9 2NATURE COMMUNICATIONS | (2021)12:947 | https://doi.org/10.1038/s41467-021-21165-9 | www.nature.com/naturecommunications controls the flux through the pathway6–9(Fig. 1a). UMP also inhibits carbamoyl phosphate synthetase (CPS), the chloroplast enzyme producing CP for both the synthesis of pyrimidines and arginine10, although this effect is counteracted by ornithine to sustain the production of arginine11,12 (Fig. 1a). Thus, ATC is the major regulated enzyme for de novo synthesis of UMP in plants13, but yet the feedback mechanism that makes this enzyme different from any other ATC remains uncharacterized. In general, ATCs consist of a catalytic homotrimer with three active sites in between the subunits that can be allosterically regulated by association with other proteins. The Escherichia coli ATC, for instance, is formed by association of two catalytic trimers with three dimers of regulatory subunits14 responsible for the binding of nucleotides that diminish (UTP and CTP) or enhance (ATP) the activity15,16. Other prokaryotic ATCs lack regulatory subunits and thus, are insensitive to nucleotides17,18. In eukaryotes other than plants, ATC is fused together with CPS into a single multienzymatic protein named CAD that also contains an active DHO (animals) or an inactive DHO-like domain (fungi)10,19–21. The ATC domain of CAD also forms homotrimers, favoring the assembly of the protein into large hexameric particles22–24, and its activity is modulated by the binding of UTP to an allosteric region within CPS25,26. In striking contrast, plant ATCs consist of a UMP-inhibitable homotrimer with no associated subunits, meaning that both the catalytic and regulatory sites must reside within the same polypeptide chain27. However, despite the wealth of biochemical and structural knowledge on ATCs from prokaryotes, fungi, and animals, there is no structural information of any plant ATC so far. Thus, the putative binding site for UMP and the catalytic and regulatory mechanisms of ATC in plants remain unknown. Here we show that the development of Arabidopsis thaliana (Arabidopsis) can be severely impaired or enhanced by the expression level of ATC. To understand this fundamental activity, we determined the crystal structure of Arabidopsis ATC free and bound to UMP, in complex with a transition-state analog, with CP or bearing a site-specific mutation that turns the enzyme insensitive to UMP. The structural, mutagenesis, and biochemical analyses reveal unique catalytic and regulatory properties of plant ATCs, suggesting new strategies to control de novo pyrimidine synthesis and plant growth. Results ATC is key for plant growth and efficient photosynthesis.To explore the importance of ATC for plant growth we used artificial microRNA (amiRNA) to knockdown ATC (At3g20330) in Arabidopsis. Two selected lines, atc-1 and atc-2, exhibited 16% and 10% residual ATC transcriptanda3-foldor20-folddropinproteinlevels compared to wild-type (WT; Col-0) controls (Fig. 1b–d). Conversely, we constitutively overexpressed ATC in two Arabidopsis lines, ATCOx1 and ATC-Ox2, which showed 13and 16-fold increase in ATC transcript and a 2.9-fold increase in protein levels (Fig. 1b–d). After4weeksonsoil,atc-1 and atc-2 downregulated lines showed a strong reduction of growth, with fresh weights of 19% (0.53 ± 0.09 g plant−1) and 6% (0.16 ± 0.03 g plant−1)oftheCol-0 control plants (2.73 ± 0.21 g plant−1)(Fig.1b, e). In contrast, ATC-Ox1 and ATC-Ox2 showed increased growth with fresh weights of 119% (3.26 ± 0.09 g plant−1) and 126% (3.45 ± 0.09 g plant−1)comparedto Col-0, respectively (Fig. 1b, e). ATC downregulated lines also showed pale leaves, suggesting lower chlorophyll levels, and presumably impaired photosynthesis, whereas ATC-Ox lines exhibited no phenotypic alterations other than the bigger size (Fig. 1b). Because of the pale leaf coloration, 4 week old atc plants were subjected to pulseamplitude-modulation (PAM) fluorometry, which measures chlorophyll fluorescence as an indicator of the photosynthetic capacity. The electron transport rate (ETR) of atc lines was less than 20% of Col-0, whereas ATC-Ox lines showed no significant differences (Fig. 1f). The pronounced ETR decrease was in line with a reduction in the maximal photosynthetic efficiency (Fv/ Fm), with values of 0.53 ± 0.03 and 0.50 ± 0.02 for atc-1 and -2, respectively, which are markedly lower than the 0.79 ± 0.01 measured in controls (Fig. 1g). Again, Fv/Fm values in ATCOx lines were similar to Col-0. These results, together with previous studies13,28, demonstrate a regulatory role of pyrimidine de novo synthesis in plant growth and a key function of ATC herein. Crystal structure of Arabidopsis ATC bound to UMP.To investigate this central enzymatic activity, we attempted to produce the ATC from Arabidopsis, a 390 amino acid (aa) precursor protein with an N-terminal chloroplast transit peptide (Fig. 2a and Supplementary Fig. 1). Having difficulties to express the fulllength protein in E. coli, we tested different N-terminal truncated forms. One construct spanning aa 82–390 (named atATC) was purified as a stable homotrimer (Fig. 2b), and matched in size (excluding the 2.6 kDa fusion tag) the 36 kDa mature enzyme in whole leaf extracts (Fig. 2c). atATC produced diffraction quality crystals readily and the structure was determined at 1.7 Å resolution (Supplementary Table 1 and Supplementary Fig. 2). The structure of the atATC trimer resembles a three-bladed propeller with a concave face holding three active sites in between subunits, thus preserving the overall architecture of the transcarbamoylase family29 (Fig. 2d). Each subunit folds into two subdomains of similar size: an N-domain (aa 82–221 and 374–390) occupying the center of the trimer and holding the binding site for CP, and a C-domain (aa 222–373) bearing the Asp binding site (Fig. 2a, d and Supplementary Figs. 1 and 3). The relative orientation of the Fig. 1 ATC central activity in de novo pyrimidine synthesis and plant growth. a Scheme of de novo pyrimidine biosynthesis pathway (highlighted in cyan with enzymatic activities in yellow background) and arginine synthesis (in orange) in plants. ATC, aspartate transcarbamoylase; DHO, dihydroorotase; DHODH, dihydroorotate dehydrogenase; UMPS, UMP synthetase; CPS, carbamoyl phosphate synthetase; OTC, ornithine transcarbamoylase. Allosteric inhibition by UMP and activation by ornithine are indicated by red and blue lines, respectively. bArabidopsis Col-0 and ATC downregulated (atc-1 and -2) or overexpressing lines (ATC-Ox1 and 2) grown for 4 weeks in a 14 h light and 10 h dark regime. cATC transcript levels in knockdown and overexpressing lines relative to Col-0 (n=9). ATC transcript levels relative to actin: 9.5 × 10−3(Col-0), 1.5 × 10−3(atc-1), 0.93 × 10−4(atc-2), 0.13 (ATC-Ox1), and 0.15 (and ATC-Ox2). p-values are 5.722 × 10−14 for atc-1, 5.93 × 10−15 for atc-2, 1.056 × 10−10 for ATC-Ox1, and 1.016 × 10−10 for ATC-Ox2.dImmunoblot with anti-ATC antibody on whole leaf extracts; Coomassie Brilliant Blue (CBB) stained SDS-PAGE was used as loading control. The ATC protein levels relative to Col-0 quantified from n-different experiments are: atc-1, 0.34 ± 0.17 (n=5); atc-2, 0.05 ± 0.024 (n=5); ATC-Ox1, 2.9 ± 0.44 (n=4); and ATC-Ox2, 2.5 ± 0.87 (n=4). eFresh weight quantification (n=10). fElectron transport rate (ETR) determined by PAM in standard light curve setting (n=12 for Col-0, n=8 for atc-1,n=11 for atc-2,n=12 for ATC-Ox1,n=11 for ATC-Ox2). p-values are 2.879 × 10−18 for atc-1, 4.563 × 10−20 for atc-2, 3.692 × 10−5for ATCOx1, 1.142 × 10−6for ATC-Ox2.gFalse color presentation of maximal photosynthesis yield monitored by PAM of lines shown in f. Error bars indicate standard error of the mean. Asterisks depict significant changes between the different lines referring to Col-0 control according to one-way ANOVA followed by Tukey’s multiple comparison test (*p< 0.05, **p< 0.01, ***p< 0.001). NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21165-9 ARTICLE NATURE COMMUNICATIONS | (2021)12:947 | https://doi.org/10.1038/s41467-021-21165-9 | www.nature.com/naturecommunications 3 domains is similar to the open conformation observed in other ATCs crystallized without ligands (Supplementary Fig. 3)23,24,30. In addition, atATC has the CP-loop (aa 156–169) and Asp-loop (aa 309–332) (Fig. 2a, d) that, as in other ATCs16,24, undergo large conformational movements upon substrate binding (see below). Unexpectedly, additional electron density in each active site indicated the presence of a molecule of UMP captured during protein expression and kept throughout the purification and crystallization process (Fig. 2d, e and Supplementary Fig. 2). The nucleotide fills the active site, with the ribose in C3′endo pucker and the base in anti conformation (Fig. 2e). The phosphate binds near the N-end of helix H2 and interacts with R136, T137, R187, and H215 (at the N-domain), whereas the ribose 2′- and 3′-OH bind to the side chain of R248 (C-domain). The 4-O atom of the pyrimidine ring interacts with R310 (Asp-loop), and the 2-O and 3-NH bind through three waters to R310, R248, and V250 (Cdomain), whereas the C5 and C6 atoms make Van der Waals contacts with the 349PLP351 loop (C-domain). In addition, the CP-loop from the adjacent subunit interacts with the inhibitor through Van der Waals contacts of residues A164 and A165 and makes a H-bond between S162 and the phosphate (Fig. 2d, e). Next, we freed the enzyme of UMP by a gel filtration procedure (Supplementary Fig. 4) and determined the crystal structure of the apo form at 3.1 Å resolution (Supplementary Table 1). The structure turned to be similar to the UMP-inhibited conformation except for aa 160–166 of the CP-loop that appear disordered in the absence of the nucleotide (Supplementary Fig. 3). atATC only binds one molecule of PALA per trimer.We investigated the effect of PALA [N-(phosphonacetyl)-L-aspartate], a potent ATC inhibitor with structural features of both substrates that mimics the transition-state of the reaction31–33. Seedling assays in the presence of 0.2 mM or 0.4 mM PALA showed a decrease in fresh weight to 59% or 23%, respectively, compared to untreated seedlings (3 times 10 seedlings were weighted per treatment, n=3). Root length in untreated seedlings was 2.68 ± 0.49 cm and was reduced to 36% at 0.2 mM PALA (0.96 ± 0.22 cm) and to 8% at 0.4 mM PALA (0.21 ± 0.09 cm) (n=30) (Fig. 3a). These results support the reduced growth observed in atc downregulated lines (Fig. 1b–d) and agree with previous PALA-inhibition studies32. Chlorosis was also apparent in PALA-treated seedlings (Fig. 3a), further endorsing the effect of reduced ATC levels on chloroplast functionality (Fig. 1f, g). To gain further insight into the reaction mechanism, we determined the structure of atATC in complex with the transition-state analog at 1.6 Å resolution (Supplementary Table 1). Surprisingly, the structure showed the atATC trimer with PALA bound to only one of the subunits (Fig. 3b and Supplementary Fig. 5). This subunit undergoes a 10° hinge closure of the Nand C-domains and a 24° rigid body rotation of the Asp-loop (Fig. 3c), emulating the movement needed to bring CP and Asp in close contact to favor the reaction33,34. In contrast, the other two subunits exhibit an open conformation, similar to the apo or UMP-bound states, and have the active sites empty or with two sulfate ions and one glycerol molecule from the crystallization solution (Fig. 3b and Supplementary Fig. 5). Only Fig. 2 Structure of Arabidopsis ATC bound to UMP. a Scheme of Arabidopsis ATC protein. bSEC-MALS analysis of purified atATC proves the formation of a homotrimer. cImmunoblot of atATC and mature ATC from leaf extract. The experiment was performed three times with identical results. dCrystal structure of atATC trimer with each subunit bound to one molecule of UMP (shown as yellow spheres). One subunit is shown on gray background. Protein domains are colored as in (a). eDetail of the active site with UMP bound. Water molecules are shown as red spheres. Electrostatic interactions are indicated as dashed lines. ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21165-9 4NATURE COMMUNICATIONS | (2021)12:947 | https://doi.org/10.1038/s41467-021-21165-9 | www.nature.com/naturecommunications the CP-loop interacting with PALA is well-defined in the electron density map, whereas the other CP-loops are flexibly disordered (Fig. 3b). The substoichiometric binding of PALA is remarkable, since other ATCs bound three molecules of PALA per trimer24,34–37 and the interactions with the transition-state analog are virtually identical to those observed in atATC (Fig. 3d and Supplementary Fig. 6). The phosphonate group of PALA binds to the N-end of helix H2 (N-domain) and the O atom of the carbamate moiety interacts with T137, R187, and H215 (N-domain), whereas the N atom binds to L350 (C-domain). Also, the α-carboxylate group binds to R248 (C-domain) and the β-carboxylate binds to R310 and Q312 (Asp-loop). In addition, the CP-loop from the adjacent subunit binds through S163 to the phosphonate moiety, and places K166 at interacting distance of the phosphonate and the αand β-carboxylates (Fig. 3d). Isothermal titration calorimetry (ITC) analyses confirmed that PALA binds to only one site per trimer (K D PALA =0.6 μM) and somehow blocks the entrance of subsequent PALA molecules to the other sites (Supplementary Table 2 and Supplementary Fig. 7). This negative cooperativity effect is specificforPALA, since the unoccupied subunits can still bind CP (K D CP = 140 μM) or UMP (K D UMP =1.2 μM). We also observed negative cooperativity, but to a lesser extent, in the titration with UMP, since the nucleotide binds with a K D UMP =0.2 μM to the first site and reduces 10-fold the affinity of the other subunits. In turn, CP showed equal affinity for the three active sites (K D CP = 77 μM). These results strongly suggested that despite the overall structural similarity with other ATCs, the atATC trimer uses a mechanism of communication between active sites that allows only one subunit to attain the closed catalytic conformation. The CP-loop blocks the simultaneous closure of the subunits. The explanation for the unusual binding of PALA to atATC was Fig. 3 Inhibition of atATC by the transition-state analog PALA. a Inhibition of Arabidopsis seedlings after 7 days treatment with PALA. Images are typical examples of 30 seedlings viewed per treatment. Scale bars indicate 0.5 cm. bCrystal structure of atATC trimer in complex with only one molecule of PALA (shown as orange spheres). Dashed lines indicate flexibly disordered CP-loops. cSuperposition of the three subunits in the PALA-bound trimer (colored as in b) and in the UMP-bound subunit (in black). The arrows indicate the closure of the subunit upon PALA binding. The Asp-loop undergoes a 29° rotation around an axis that has been represented and colored in red. dDetail of the binding of PALA to the active site. Water molecules are shown as red spheres. Electrostatic interactions are indicated as dashed lines. NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21165-9 ARTICLE NATURE COMMUNICATIONS | (2021)12:947 | https://doi.org/10.1038/s41467-021-21165-9 | www.nature.com/naturecommunications 5 likely at the CP-loop, as the most distinct element compared to other non-plant ATCs (Fig. 2a and Supplementary Fig. 1). This loop is flexible in the absence of ligands (Fig. 3b and Supplementary Fig. 3) but adopts two distinct conformations whether UMP or PALA are bound to the adjacent subunit (Figs. 2e and 3d). With UMP, the CP-loop folds in an extended “inhibited” conformation, with A164, A165 and S162 interacting with the nucleotide, S163 and K166 pointing outwards the active site, and the side chain of F161 inserted in between subunits (Fig. 4a). In turn, upon PALA binding, the CP-loop rearranges into two short and nearly perpendicular 3 10 α-helices, placing S163 and K166 to interact with the transition-state analog and moving A164, A165, and S162 outwards the active site (Fig. 4b). In this “active”conformation, F161 flips 180° compared to the position with UMP, and projects towards the trimer three-fold axis, where intersubunit distances are shortened by the interactions between neighbor E156 residues (Fig. 4b). These tight contacts at the center of the trimer are not observed in other ATCs bound to PALA (Supplementary Fig. 6), suggesting that the position of F161 may prevent other CP-loops from reaching a similar active conformation. Two additional atATC structures reinforced this hypothesis. One structure, obtained from crystals with PALA and soaked in CP (Supplementary Table 1), showed a trimer with one subunit bound to PALA, a second subunit with CP, and a third subunit with CP and with one glycerol and one sulfate ion filling the Asp binding site (Fig. 4c and Supplementary Fig. 5). The second structure, obtained by co-crystallization with CP, showed all three subunits in the trimer bound to CP (Supplementary Table 1 and Supplementary Fig. 5). In both structures, the three CP-loops in the trimer fold in an active conformation but show poor electron density compared to the rest of the protein. In fact, E160 and F161 were traced in only one subunit, and modeling in similar conformation in the other subunits caused steric clash and charge repulsion (Fig. 4d). Mutant F161A is not inhibited by UMP and binds three PALAs. To further test the role of the CP-loop, we replaced F161 with Ala (F161A). The mutation did not affect the solubility nor the oligomeric state of the protein, but changed the susceptibility of the enzyme to UMP, PALA, or to high Fig. 4 The CP-loop folds in two different conformations for UMP or PALA binding. a,bCartoon representation of the CP-loop (in magenta) over the active site of the adjacent subunit (shown in surface representation) bound to UMP (a) or PALA (b). cStructure of atATC with one subunit bound to PALA, a second one with CP, and the third having CP, glycerol, and one sulfate ion. A UMP molecule is shown in semitransparent gray to compare the position relative to the ligands. dDetail view along the three-fold axis. The side chains of E160 and F161 are only seen in one subunit. Modeling of these two residues in the other two subunits (shown in yellow semitransparent representation) causes steric clash and charge repulsion. ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21165-9 6NATURE COMMUNICATIONS | (2021)12:947 | https://doi.org/10.1038/s41467-021-21165-9 | www.nature.com/naturecommunications concentrations of the substrates. Initial-rate plots of WT with CP as variable ligand are hyperbolic in the absence of UMP (V max =93.23 ± 2.53 nmol min−1μg−1,K 0.5 CP =0.46 ± 0.04 mM, K 0.5 Asp =0.94 ± 0.10 mM), but turn sigmoidal in the presence of UMP, with a Hill-coefficient h=2.2, indicating positive cooperativity for CP binding (Fig. 5a and Supplementary Fig. 8), as previously described for wheat-germ ATC7. In contrast, parallel assays with F161A proved that although the catalytic activity is highly similar to the WT (V max =115.0 ± 14.62 nmol min−1μg−1,K 0.5 CP =0.62 ± 0.13 mM, K 0.5 Asp = 3.20 ± 1.03 mM), the enzyme is not inhibited by UMP and also becomes more sensitive to the presence of PALA (Fig. 5aand Supplementary Fig. 8). In addition, F161A showed decreased activity at high substrate concentrations, with an inhibition Fig. 5 Activity and crystal structure of atATC mutant F161A. a,bCP saturation curves of atATC WT (a) and F161A (b) in the absence and presence of UMP or PALA. cAsp saturation curves for WT and F161A. Equations to fit the kinetic data are detailed in Methods. d,eCrystal structure of atATC-F161A trimer complexed with UMP (d), and detail of the interactions of the CP-loop (e), showing a glycerol molecule replacing the missing F161 side chain. f,gCrystal structure of F161A with PALA bound to the three active sites (f), and detail of the interactions around the molecular axis (g). NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21165-9 ARTICLE NATURE COMMUNICATIONS | (2021)12:947 | https://doi.org/10.1038/s41467-021-21165-9 | www.nature.com/naturecommunications 7 constant (K i ) of 4.52 mM, whereas this substrate inhibition effect was not apparent in the WT (Fig. 5b, c and Supplementary Fig. 8). ITC analysis failed to detect the binding of UMP to F161A, supporting the loss of inhibition by the nucleotide (Supplementary Table 2 and Supplementary Fig. 7). We also found that F161A binds one molecule of PALA (K D PALA =0.12 μM) with 5-fold higher affinity than WT, in agreement with the enhanced inhibition, and also shows ~100-fold higher affinity for CP (K D CP =0.7 μM) (Supplementary Table 2 and Supplementary Fig. 7). It seems likely that removal of the F161 side chain destabilizes the inhibited conformation of the CP-loop, reducing the affinity for UMP, and thus, favoring the alternate CPor PALA-bound conformation (Fig. 4a, b). To better understand the effect of the mutation, we determined the structure of F161A with UMP (Supplementary Table 1). In apparent contradiction with the activity and ITC results, the structure showed a molecule of UMP in the active site (Fig. 5d), the CP-loop in the inhibited conformation, and the missing F161 side chain being replaced by a glycerol molecule (Fig. 5e). It is probable that the low-affinity binding of the nucleotide to the active site of the mutated protein is favored by the higher concentrations of nucleotide (5 mM) and protein (135 μM) used in the crystallization condition compared to those in the ITC experiments (70 μM UMP at most and 30-40 μM protein). We also determined the structure of F161A crystallized with PALA (Supplementary Table 1). Interestingly, the structure showed a trimer bound to three molecules of PALA rather than one as in the WT (Fig. 5f and Supplementary Fig. 5). Although ITC indicated that PALA binds with high affinity to only one site per trimer (Supplementary Table 2), the high concentrations of PALA (2 mM) and protein in the crystallization condition must favor a low-affinity binding to the other subunits. Importantly, the three CP-loops fold in an active conformation, have well-defined electron density, and show no steric clashes around the molecular three-fold axis, since the bulky F161 side chain is missing and E160 adopts alternate conformations (Fig. 5g). Discussion The pathway for de novo synthesis of UMP is evolutionary conserved in all plants examined so far, and loss of function of any of the enzymes involved is presumably lethal. However, whereas downregulation of CPS, DHO, DHODH, or UMPS (Fig. 1a) had little or no effect38,39, we showed that ATC downregulation strongly inhibits plant growth (Fig. 1b–e), as reported in previous studies38,40, and causes a severe decrease in photosynthetic efficiency (Fig. 1f, g). Conversely, we also proved that plant growth can be enhanced by ATC overexpression (Fig. 1b–f). These results support the notion that ATC is not produced in large excess in the cell38, and thus, those plants are especially sensitive to ATC levels that exert highest control over pyrimidine de novo synthesis. Indeed, the production of ATC is under transcriptional regulation in response to tissue pyrimidine availability38,40 and to growth signals mediated by the TOR pathway41. However, transcription, synthesis and translocation of ATC into the chloroplast are slow and energetically costly processes that do not correct for rapid fluctuations needed to maintain nucleotide homeostasis. For this, allosteric regulation by UMP is the major mechanism controlling ATC activity in plants7, but until now, we lack detailed information of how this feedback loop occurs. Now, the structures of atATC reveal the mechanism of inhibition and explain the unsolved problem of why plant ATCs are inhibited by UMP and not by UTP as in other organisms8,31 (Fig. 6). Rather than occupying an allosteric pocket, UMP binds and blocks the active site (Figs. 2and 6a), directly competing with CP, the substrate binding in first place18,42. UMP binds to the subunit in a wide-open conformation (Supplementary Fig. 3), where the Nand C-domains cannot move further apart to accommodate a dior tri-phosphorylated nucleotide, thus explaining why UDP or UTP are not inhibitors. On the other hand, the pocket for the nitrogenous base is too small for the double ring of a purine and highly selective for uracil, since the methyl group of thymine would clash with the 349PLP351 loop, whereas the cytidine amino group would distort the interaction with R310 (Fig. 2e). Also, one would expect the binding of deoxyUMP to be weak based on the interaction between the ribose OH groups and the side chain of R248, which mimic the recognition of the Asp α-COOH group (Fig. 2e). However, these UMPinteracting elements are common to other ATCs that do not bind the nucleotide at the active site (Supplementary Fig. 1). Thus, we propose that the capacity of plant ATCs to be selectively inhibited by UMP relies on few small changes in the CP-loop (Figs. 2e and 4a). Indeed, a single point mutation in the CP-loop, F161A, is sufficient to turn atATC insensitive to UMP without affecting the catalytic efficiency of the enzyme nor its inhibition by PALA (Fig. 5, Supplementary Table 2, and Supplementary Fig. 8). Since the sequence of the CP-loop appears invariant in all plant ATCs known up to date (Supplementary Fig. 1), we propose that the UMP-inhibition mechanism described here for Arabidopsis ATC must be conserved across the plant kingdom. Thus, unlike other bacterial or eukaryotic ATCs that rely on complex associations with regulatory proteins (Fig. 6b, c), the current structures explain how plant ATCs have evolved to maintain a simple organization with both catalytic and regulatory capacities within a single protein chain (Fig. 6a). We hypothesize that this simplicity is a convenient solution for the transcriptional regulation and translocation of a single gene product into the chloroplast. atATC has a surprising affinity for UMP, ∼400-fold higher than for CP (Supplementary Table 2), which is explained by the extensive contacts of the nucleotide with the CP and Asp binding sites, similar to what PALA does. However, whereas binding of PALA involves large conformational changes (Fig. 3c), UMP binds to the more energetically favorable open state, and this might explain the 3-fold higher affinity for UMP than for the transition-state analog (Supplementary Table 2). It is uncertain whether such affinity for the natural inhibitor is retained in vivo or if it could be further increased by interaction with phospholipids, as reported for wheat-germ ATC43. In any case, based on these results, one would expect that ATC is constitutively repressed under steady-state conditions if UMP in the plastid reaches sub-millimolar concentrations similar to those estimated at the cytoplasm38,39,44. However, the concentration of UMP in the plastid is unknown and other features found in atATC suggest that this important activity is fine-tuned by the balance of UMP and CP contents in the cell. Indeed, we showed that the affinity for UMP is modulated by the communication between subunits, so that activity is diminished by high-affinity binding of the inhibitor to one subunit, whereas complete inactivation requires a 10-fold increase in the UMP pool (Supplementary Table 2). On the other hand, the response to UMP also varies with the concentration of CP, as shown by the change in the kinetic curves from a hyperbola to a sigmoid (Fig. 5a). This behavior, which was first described for wheat-germ ATC7, could be relevant for the coordination of de novo pyrimidine and arginine synthesis, two pathways that depend on CP availability11,12 (Fig. 1a). For instance, when pyrimidine pools are low, the activity of UMP-free ATC responds linearly with the concentration of CP (Fig. 5a, hyperbola), competing with arginine synthesis to ensure the production of nucleotides. Then, as UMP pool builds up, the synthesis of pyrimidines is reduced by the end product inhibition of ATC and CPS. However, upon CPS inhibition, ornithine ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21165-9 8NATURE COMMUNICATIONS | (2021)12:947 | https://doi.org/10.1038/s41467-021-21165-9 | www.nature.com/naturecommunications Fig. 6 Different regulatory mechanisms in ATC. a Plant ATCs present a unique mechanism of regulation, where UMP binds and blocks the active site. The CP-loop (represented in magenta) alternates between an UMP-bound inhibited conformation and an active conformation that ensures the sequential and perhaps ordered firing of the active sites. bIn bacteria, isolated ATC catalytic trimers are unregulated. The association of two catalytic trimers with three dimers of regulatory subunits results in a holoenzyme that undergoes large conformational changes upon binding of UTP (inhibitor) or ATP (activator)to allosteric sites in the regulatory subunits. cIn eukaryotes other than plants, ATC is fused together with CPS and DHO into a single multienzymatic protein named CAD that oligomerizes into hexamers where ATC trimers are proposed to occupy apical positions. The CPS and ATC activities are regulated by the binding of UTP to a regulatory region within the CPS domain. NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21165-9 ARTICLE NATURE COMMUNICATIONS | (2021)12:947 | https://doi.org/10.1038/s41467-021-21165-9 | www.nature.com/naturecommunications 9