The structure of B-ARR reveals the molecular basis of transcriptional activation by cytokinin

Significance

In plants, cytokinin signal transduction pathway resembles the two-component system in bacteria. While the structural basis of the two-component system in bacteria has been extensively studied, our understanding of how cytokinin rapidly triggers gene transcription in plants remains limited. Here, we revealed that the DBD is initially locked in an inactive state by the RD. The phosphoryl transfer from the histidine of AHPs to the aspartate of the RD of B-ARRs enables the DNA binding capacity of the DBD, thereby allowing a rapid transcriptional response to cytokinin. Our findings thus not only provide insights into the molecular mechanisms of cytokinin-induced transcriptional activation but also shed light on how the two-component system orchestrates gene transcription through phosphorylation relay in eukaryotes.

Abstract

The phytohormone cytokinin has various roles in plant development, including meristem maintenance, vascular differentiation, leaf senescence, and regeneration. Prior investigations have revealed that cytokinin acts via a phosphorelay similar to the two-component system by which bacteria sense and respond to external stimuli. The eventual targets of this phosphorelay are type-B ARABIDOPSIS RESPONSE REGULATORS (B-ARRs), containing the conserved N-terminal receiver domain (RD), middle DNA binding domain (DBD), and C-terminal transactivation domain. While it has been established for two decades that the phosphoryl transfer from a specific histidyl residue in ARABIDOPSIS HIS PHOSPHOTRANSFER PROTEINS (AHPs) to an aspartyl residue in the RD of B-ARRs results in a rapid transcriptional response to cytokinin, the underlying molecular basis remains unclear. In this work, we determine the crystal structures of the RD-DBD of ARR1 (ARR1^RD-DBD) as well as the ARR1^DBD–DNA complex from Arabidopsis. Analyses of the ARR1^DBD–DNA complex have revealed the structural basis for sequence-specific recognition of the GAT trinucleotide by ARR1. In particular, comparing the ARR1^RD-DBD and ARR1^DBD–DNA structures reveals that unphosphorylated ARR1^RD-DBD exists in a closed conformation with extensive contacts between the RD and DBD. In vitro and vivo functional assays have further suggested that phosphorylation of the RD weakens its interaction with DBD, subsequently permits the DNA binding capacity of DBD, and promotes the transcriptional activity of ARR1. Our findings thus provide mechanistic insights into phosphorelay activation of gene transcription in response to cytokinin.

The phytohormone cytokinin possesses pleiotropic effects on plant physiology and development, including roles in shoot and root architecture, vascular differentiation, leaf senescence, and shoot regeneration (1–8). A current model for cytokinin signal transduction has been put forth that is similar to the two-component system by which bacteria are able to sense and respond to external stimuli (9–14). The cytokinin receptors named ARABIDOPSIS HIS KINASE2 (AHK2), AHK3, and AHK4 (also known as WOODENLEG 1 and CYTOKININ RESPONSE1) are hybrid kinases akin to the histidine kinases within the two-component system (15–18). All these AHKs possess an extracellular cytokinin-binding CHASE (cyclase/His kinase-associated sensing extracellular) domain, a cytoplasmic histidine (His) transmitter, and receiver domains (RDs).

Downstream phosphorelay elements comprise five ARABIDOPSIS HIS PHOSPHO​TRANSFER PROTEINS (AHPs) containing conserved amino acids required for phosphotransfer through a conserved His (19, 20). AHPs mediate the transfer of a phosphoryl group from the AHKs to the ARABIDOPSIS RESPONSE REGULATORS (ARRs) (21–23), which are the eventual nuclear targets of the phosphorelay. All ARRs have conserved amino acids required for N-terminal RD phosphorylation. According to the C-terminal domain structure, the Arabidopsis cytokinin response-related ARR genes can be grouped into two classes, namely type-A ARRs (A-ARRs) and type-B ARRs (B-ARRs, e.g., ARR1, ARR10, and ARR12) (24). In addition to the RD located at the N terminus, the B-ARRs also contain a medial Myb-like DNA binding domain (DBD) and a C-terminal transactivating domain (2, 10). B-ARRs are critical for the early transcriptional responses of plants to cytokinin (25–28). In response to cytokinin treatment, the phosphoryl transfer from the histidine of AHPs to the aspartate of the RD of B-ARRs rapidly causes transcription of cytokinin-activated targets.

In the past decade, a surge in our understanding of the structural basis for cytokinin signal transduction has taken place (29). For instance, structural analyses of the sensor domain of AHK4 have provided mechanistic insights into how AHK4 is activated by cytokinin using its membrane-distal Per-Arnt-Sin (PAS) domain and how it discriminates between diverse cytokinin conjugates (30, 31). Analyses of the RDs of AHK4 and AHP2 have further revealed a high level of conformational conservation between prokaryotic and eukaryotic RDs (32–34). Moreover, structural elucidation of AHP1 in concert with the RD of AHK5 has elucidated the molecular recognition events controlling phosphorelay in plants (35).

NMR spectroscopy has been employed to determine how the DBD of B-ARRs binds to DNA (36). However, due to the dynamism of this system and the partial disorder observed in the solution by NMR, residues responsible for recognizing the consensus DNA sequence have not yet been resolved. Particularly, how the RD is able to function as a phosphorylation-regulated switch to tune the transcriptional activities of the B-ARRs remains unclear. In this study, by solving crystal structures of the ARR1^DBD–DNA complex, we uncovered the mechanism by which the DBD recognizes the GAT trinucleotide in a sequence-specific manner. A comparison of the ARR1^RD-DBD and ARR1^DBD–DNA structures further revealed that the DBD is initially locked in an inactive state by the RD. The phosphoryl transfer from the histidine of AHPs to the aspartate of the RD of B-ARRs enables the DNA binding capacity of the DBD, thereby allowing a rapid transcriptional response to cytokinin.

Results

Crystal Structure of ARR1^DBD.

B-ARRs exhibit a preference for recognition of the DNA sequences (A/G)GAT(T/C), and these GAT motifs are often identified in the upstream regions of many cytokinin-regulated genes (25, 36–38). Based on these achievements, a universal cytokinin reporter known as TCS (two-component-output-sensor) was uncovered (39), where the concatemerized B-ARR-binding motif fused to a minimal 35S promoter and used to drive reporter expression. This synthetic reporter has been broadly employed to visualize cytokinin output in vivo.

To examine the molecular mechanism by which the DBD of ARR1 (ARR1^DBD) is able to recognize GAT motifs in the TCS, we determined the crystal structure of ARR1^DBD (residues 238–297) in complex with a modified 25-bp TCS DNA fragment containing multiple GAT motifs (DNA^TCS, 5′-AATCCAGATTAATCTAATCTAATCC-3′, and 5′-GGATTAGATTAGATTAATCTGGATT-3′ as its complementary strand), at 2.35-Å resolution utilizing the single-wavelength anomalous diffraction method (hereafter known as ARR1^DBD–DNA, Fig. 1 and SI Appendix, Table S1). The asymmetrical unit consists of six ARR1^DBD molecules as well as two DNA molecules, with three ARR1^DBD molecules inserted into the major groove of each DNA molecule (SI Appendix, Fig. S1A). The six ARR1^DBD molecules adopt similar conformations, with RMSD values of 0.15 to 0.43 Å for the Cα atoms (SI Appendix, Fig. S1 A–D). ARR1^DBD consists of three consecutive α-helices, adopting a classic MYB domain topology similar to those identified previously across several MYB domain proteins (40–42). As illustrated in SI Appendix, Fig. S1 E and F, the MYB domains of rice phosphate starvation response 2 (OsPHR2) (42), Arabidopsis phosphate starvation response 1 (AtPHR1) (40), and Arabidopsis LUX ARRHYTHMO (AtLUX) (41) can be superimposed onto ARR1^DBD with RMSDs of 0.434 Å, 0.388 Å, and 0.537 Å, respectively.

Fig. 1.

The crystal structure of ARR1^DBD complexed with DNA^TCS. (A) An electrostatic surface view of ARR1^DBD indicates that ARR1^DBD inserts into the major groove of DNA^TCS, forming extensive interactions. (B) A schematic view of the overall interactions between ARR1^DBD and DNA^TCS. The canonical GAT motif in the TCS is highlighted in yellow. The interaction residues are identified as sticks, with hydrogen bonds indicated as dashed silver lines. (C) Details of the interactions between ARR1^DBD and DNA^TCS. The DNA contacts mediated by Ser283, Gln286, and Lys287, as well as other essential residues of ARR1^DBD, are identified. Hydrogen bonds (H-bonds) between protein side chains and DNA bases are shown in green, whereas those between main chains and phosphates of DNA are highlighted in blue. (D) The amino acid sequences of the Arabidopsis thaliana (NP_850600.2), Oryza sativa (NP_001389022.1), Populus trichocarpa (XP_052312522.1), Ostreococcus tauri (OUS42701.1), Volvox carteri (XP_002958010.1), Selaginella moellendorffii (XP_024534381.1), Physcomitrella patens (XP_024388633.1), and Marchantia Polymorpha (PTQ32201.1) were aligned, and conserved amino acids are outlined by black triangles. (E) Binding affinities of DBD with DNA^TCS. The equilibrium dissociation constant (K_D) was determined using the ForteBio Octet data analysis software. (F) Functional analyses of critical residues responsible for the binding of ARR1^DBD with DNA^TCS in vivo. The wild-type (WT) and transgenic plants were grown for 8 d in LDs. (Scale bar, 1 cm.) (G) Quantification of primary root length. Data are presented in violin plots showing the median (dashed lines) and quartiles (dotted lines), n ≥ 12. Letters indicate significant differences determined through one-way ANOVA, P < 0.01, n ≥ 20. (H) Phenotype of 21-d-old WT and transgenic plants grown in LDs. Data for two representative plants in each genotype (F–H) are shown. (Scale bar, 1 cm.)

Examination of the ARR1^DBD–DNA structure uncovered that the ARR1^DBD employs a highly electropositive protruding face for binding with DNA^TCS (Fig. 1A). These protein–DNA interactions can be separated into two groups, namely sequence-specific base recognition and sequence-nonspecific backbone recognition. Specifically, the residue Lys287 engages in two hydrogen bonds with O6 and N7 of guanine (G), the residue Ser283 forms two hydrogen bonds directly with N7 of adenine (A), and N6 via a water molecule, while residue Gln286 forms two hydrogen bonds with N6 and N7 of adenine (A) within the complementary chain (Fig. 1 B and C and SI Appendix, Fig. S1B). The three specific bases are in sequence, and the readout in the same chain of dsDNA is GAT, as previously identified (36, 37, 43–45).

Additionally, several hydrogen bonds were observed between DNA^TCS and residues Val240, Trp242, Pro264, Lys265, Arg278, Glu279, Asn280, Arg289, and Arg293 (Fig. 1C and SI Appendix, Fig. S1C). Notably, many of the above residues are evolutionarily conserved from bryophytes to dicotyledons (Fig. 1D) (46), suggesting that these residues are crucial in ARR1^DBD binding to DNA^TCS.

We next conducted mutational analyses of ARR1^DBD to assess the roles of these residues in DNA binding. The findings demonstrated that the S283A, Q286A, K287A, Q286A/K287A, and S283A/Q286A/K287A mutations impaired sequence-specific base recognition, causing a severe decrease in ARR1^DBD binding to DNA^TCS (Fig. 1E and SI Appendix, Fig. S2). In contrast, mutations to the residues involved in sequence-nonspecific backbone recognition, including R239A and E279A, had a moderate weakening effect on DNA binding or had no effect (SI Appendix, Fig. S2). Thus, these findings indicate that the sequence-specific base recognition residues are critical for DNA binding, while the sequence-nonspecific backbone recognition residues assist in further stabilization of the protein–DNA complex in preparation for the subsequent transcription process.

To further identify whether these residues are functionally relevant in vivo, we constructed a series of complementation assemblies, where the 3×FLAG-tagged wild-type or mutated ARR1 (ARR1^S283A, ARR1^Q286A, and ARR1^K287A) expression was driven by the 35S promoter. As outlined above, since B-ARRs are functionally redundant, their single mutants did not exhibit strong developmental defects with altered cytokinin responses (21, 22). As such, the complementation constructs were introduced into the arr1 arr10 arr12 triple mutant. Consistent with previous findings (22), compared to the wild-type, the primary root length and rosette size were greatly limited in the arr1 arr10 arr12 triple mutants (Fig. 1 F–H). While the expression of the wild-type version of ARR1 could fully complement root growth limitations, the ARR1 protein carrying an S283A, Q286A, or K287A mutation could not rescue the short root phenotype of the arr1 arr10 arr12 triple mutants (Fig. 1 F and G). Similarly, only the wild-type ARR1 could restore a small rosette phenotype found in the triple mutants (Fig. 1H) (21). Collectively, these findings reveal the structural basis for sequence-specific recognition of the GAT motif by ARR1^DBD and highlight the functional importance of the identified critical residues in vivo.

Crystal structure of ARR1^RD-DBD.

To search how the RD functions as a phosphorylation-regulated switch to tune the transcriptional activities of the B-ARRs, we sought to solve the crystal structure of ARR1^RD-DBD. To this end, the recombinant ARR1 protein (residues 1–301) was purified, crystallized, and its crystal structure was determined at a resolution of 2.2 Å using the molecular replacement method (Fig. 2 and SI Appendix, Table S1). There are two ARR1^RD-DBD molecules within one asymmetrical unit. The two ARR1^RD-DBD molecules contact face-to-face with one another via hydrophobic interactions through their loop regions (Fig. 2A). The structure contains the full-length ARR1^RD-DBD, aside from three regions (residues 1–30, 159–234, and 297–301) that are disordered (Fig. 2 A and B). The complete structure can be divided into two domains: the N-terminal RD (residues 31–158) and the C-terminal DBD (residues 235–296). The RD consists of six β-strands and five α-helices, while the five-stranded parallel β-sheet (β2-6) is surrounded by five α-helices (α1-5). The DBD contains three α-helices (α6-8), similar to the DBD in the ARR1^DBD–DNA complex structure. Between the two domains is a long linker, which is disordered in the current structure (residues 159–234).

Fig. 2.

Structural analysis of ARR1^RD-DBD. (A) The overall structure of ARR1^RD-DBD is outlined by ribbon cartoon. The RD and DBD of ARR1^RD-DBD are in blue and green, respectively. (B) Zoomed-in view of the intermolecular interactions between the RD (blue) and DBD (green) of ARR1^RD-DBD. The interacting residues are identified in stick representations, with hydrogen bonding interactions shown by dashed lines. (C) Topology of the RD of ARR1^RD-DBD. Structural elements are labeled. (D) SDS-PAGE illustrating the effects of the mutations on interface formation between the RD and DBD determined through in vitro pull-down experiments. Wild-type (WT) and mutated versions of ARR1^RD and ARR1^DBD were fused to glutathione S-transferase (GST) and 6xHis, respectively. (E) Sequence alignment of B-ARRs from Arabidopsis. Accession numbers are ARR1 (NP_850600.2), ARR2 (NP_193346.5), ARR10 (NP_194920.1), ARR11 (NP_176938.1), ARR12 (NP_180090.6), ARR14 (NP_001324402.1), and ARR18 (NP_200616.4). The conserved amino acids on the interface (B) are outlined by black triangles. (F) Phylogenetic tree of ARRs in Arabidopsis. The A-ARRs and B-ARRs are highlighted in red and blue shading, respectively.

The interaction surface between the RD and DBD was assessed further in detail. As underscored in Fig. 2 B and C, the RD primarily binds to the DBD through hydrogen-bonding interactions. The interaction surface is primarily composed of α4/α5/β6 of the RD and α8 of the DBD, which buries 1,211 Ų of surface area (Fig. 2 B and C). Specifically, Asp109, Val124, Val128, Gly131, Asp134, and Tyr135 residues from the RD form hydrogen bonds with Lys287, Tyr291, His247, Tyr288, and Arg294, respectively, within the DBD. Additionally, the residues His151 and His284 form hydrogen bonds with residues Asp134 and His247, respectively, to dictate their correct positioning. The side chains of residues Val124 and Val128 form hydrophobic interactions with Tyr135 and Tyr291, respectively, to assist in orientating these residues correctly.

Consistent with these findings, in vitro pull-down experiments revealed that the interactions between the RD and the DBD were significantly impaired by the RD residue mutations D134A, Y135A, and H151A, or by the DBD residue mutations K287A, Y288A, Y291A, R294A, H247A/H284A, and Y288A/Y291A (Fig. 2D). Moreover, sequence alignment and comparisons revealed that most of these residues, including D134, Y135, H151, H284, K287, and Y288, are highly conserved in B-ARRs of A. thaliana (Fig. 2 E and F).

Members of the response regulator (RR) family share an approximately 130-residue RD domain containing conserved amino acids that are essential for phosphorylation (47, 48). As in bacterial RDs, three amino acids Asp43, Asp89, and Lys138 in the ARR1^RD, also known as the DDK motif, form an acidic pocket of phosphorylation with a negatively charged surface in ARR1^RD-DBD (SI Appendix, Figs. S3 and S4) (9, 46).

Inhibition of DNA Binding by the RD.

To elucidate how the phosphorylation of the RD impacts the DNA binding capacity of the DBD, we attempted to solve the structures of beryllium trifluoride (BeF₃)-bound ARR1^RD-DBD as well as the phosphorylation mimic versions of ARR1^RD-DBD or ARR1^RD (see below) (47, 48). Unfortunately, despite several attempts, we were unable to obtain these crystals. To overcome this difficulty, we carefully compared the structures of ARR1^DBD–DNA and ARR1^RD-DBD. Superposition analysis showed that the RD creates a severe steric hindrance for DNA binding, due to its tight interaction with the DBD in the structure of ARR1^RD-DBD (Fig. 3A and SI Appendix, Fig. S4D). The DBD in the structure of ARR1^DBD–DNA represents an active state. In contrast, the DBD in the structure of ARR1^RD-DBD represents an inactive state. In the active state, the DBD contacts DNA predominantly through the helix α8. However, in the presence of the RD, the RD–DBD interaction results in the curving of helix α8 of the DBD, causing a maximum movement of 8 Å at the terminus of helix α8.

Fig. 3.

ARR1^RD inhibits binding of ARR1^DBD to DNA^TCS. (A) Structural superpositions of ARR1^DBD–DNA^TCS and ARR1^RD-DBD. The RD in ARR1^RD-DBD, the DBD in ARR1^DBD–DNA^TCS, and ARR1^RD-DBD are identified in blue, magenta, and green, respectively. Major clashes of the RD with the DNA are indicated by dashed black circle. (B) Zoomed-in view of the active DBD (magenta) and inactive DBD (green) conformational change. Key residues are highlighted in stick representations. (C) Quantitative analyses of LUC activities in Nicotiana benthamiana leaves. 35S:REN was employed as an internal control. Quantification was conducted by normalizing LUC activity to that of REN. Results are presented as mean ± SE, n = 4. (D) Binding affinities of RD-DBD with DNA^TCS. The K_D was determined using ForteBio Octet data analysis software. (E–G) Expression of ARR5 (E), ARR7 (F), and ARR15 (G) in WT and transgenic seedlings. Different colors represent different genotypes. Letters indicate significant differences as determined by one-way ANOVA, P < 0.01. (H) Phenotype of 6-d-old seedlings. The WT, arr1 10 12, and transgenic plants were grown on vertical Murashige and Skoog (MS) agar plates in the presence or absence of 100 nM 6-BA. (Scale bar, 1 cm.) (I) Quantification of the primary root length of 6-d-old seedlings. Two different 6-BA concentrations (10 and 100 nM) were utilized. Results are presented as mean ± SE, n ≥ 14. Data for two representative plants for each genotype are shown. An identical color code for different genotypes is employed (E–G). (J) Proposed model for the transcriptional activation by cytokinin. Cytokinin is identified by AHKs at the cell membrane. AHPs mediate the phosphoryl group transfer from the AHKs to A-ARRs and B-ARRs. The monomeric unphosphorylated ARR1^RD-DBD adopts an inactive conformation, where the contacts surrounding the β6-α8 face hold the α8 helix of the DBD at a position unfavorable for DNA binding. Phosphorylation at D89 causes a conformational alternation of β6, weakening the interactions surrounding the β6-α8 face. The structural rearrangement of α8 facilitates binding with the DNA harboring the GAT motif. Note that the unphosphorylated ARR1^RD-DBD exists in equilibrium between inactive and active conformations, and phosphorylation shifts the equilibrium toward the active state.

As illustrated in Fig. 3B and SI Appendix, Fig. S4E, residues Asp109, Asp134, and Arg154 in the RD can fix the location and orientation of residues Lys287, Arg294, and Gln286, respectively, from the helix α8 of the DBD through hydrogen bonding interactions. This RD–DBD interaction prevents DNA binding with the DBD through residues Lys287 and Gln286. Compared to the inactive state, a significant conformational change occurs at residue Arg294 in the terminal region of helix α8 of the DBD in the active state. The residues Asp134 and Tyr135 in β6 of the RD form three hydrogen bonds with Arg294 to lock its position. Notably, the critical residue Lys138 in the DDK motif protrudes to the phosphorylation pocket and is localized to the loop, closely connected to β6. Thus, these findings indicate that an allosteric conformational change at the phosphorylation site of the RD may propagate to the distal interaction surface between β6 of the RD and α8 of the DBD via Lys138.

To examine this hypothesis, we conducted tobacco transient assays with TCS:LUC as the reporter and various versions of ARR1 as effectors. Overexpression of 3×HA-tagged ARR1 with the 35S promoter (35S:ARR1-3×HA) led to a 2.5-fold enhancement in LUC activity (Fig. 3C). A similar effect was observed when ARR1^K138A-3×HA, where Lys138 was mutated to Ala (K138A), was overexpressed (SI Appendix, Fig. S4C). This result is consistent with the notion that Lys138 is located in a flexible region (Fig. 3B and SI Appendix, Fig. S4B), and a mutation at this residue is unlikely to cause an extensive conformational change of the whole structure. We then mutated Asp89 to either the negatively charged amino acid Glu (D89E) to mimic the negative charge of the phosphate group or to Ala (D89A) to mimic the non-phosphorylated form. Overexpression of ARR1^D89A-3×HA resulted in lower LUC activity compared to the control. In contrast, overexpression of ARR1^D89E-3×HA led to sixfold increase in transactivation activity (Fig. 3C). Similarly, the substitutions of Tyr288 or Arg294 with Ala (35S:ARR1^Y288A-3×HA or 35S:ARR1^R294A-3×HA), another two crucial residues involved in the interaction between the RD and DBD (Fig. 3B), caused enhanced transcriptional activity (Fig. 3C). Consistently, in vitro binding assays employing biolayer interferometry uncovered that both ARR1 and ARR1^D89A exhibited weak binding affinity to DNA^TCS. In contrast, ARR1^D89E and ARR1^R294A could bind to DNA^TCS strongly (Fig. 3D and SI Appendix, Fig. S5).

To gain deeper insight into the function of these residues in planta, we generated transgenic Arabidopsis lines that overexpressed ARR1-3×HA (ARR1-ox), ARR1^Y288A-3×HA (ARR1^Y288A-ox), and ARR1^R294A-3×HA (ARR1^R294A-ox) in a wild-type background. Using western blot, we selected the T₂ transgenic lines (two lines for each genotype) with comparable and moderate expression levels (SI Appendix, Fig. S6A). Quantitative real-time PCR (qRT-PCR) identified that transcript levels of ARR5, ARR7, and ARR15, three A-type ARR genes activated by ARR1, were elevated in ARR1^Y288A-ox and ARR1^R294A-ox plants compared to ARR1-ox plants (Fig. 3 E–G), suggesting an elevated cytokinin response. To further verify this finding, we assayed the root growth inhibition using 6-benzylaminopurine (6-BA), a synthetic cytokinin that inhibits the growth of the primary root in Arabidopsis (49, 50). As illustrated in Fig. 3 H and I, ARR1^Y288A-ox and ARR1^R294A-ox seedlings developed significantly a shorter primary root than ARR1-ox lines upon exposure to 100 nM 6-BA. Notably, compared to wild-type and ARR1-ox, ARR1^Y288A-ox, and ARR1^R294A-ox exhibited strong developmental defects when grown in soil: Leaves are small and curled from the third leaf onward, flowers were abnormal, and siliques were short and thick (SI Appendix, Fig. S6B). Collectively, these findings indicate that the mutations of Tyr288 or Arg294 in ARR1 results in enhanced cytokinin sensitivity in Arabidopsis and support the idea that the interaction between β6 of the RD and α8 of the DBD of ARR1 plays a crucial role in transcriptional activation by cytokinin.

A Model for the Activation of B-ARRs through Phosphorylation of the RD.

According to the above structure analyses, we propose a model for the activation of B-ARRs through phosphorylation of the RD (Fig. 3J). In this model, ARR1^RD-DBD can adopt different allosteric conformations. As the DBD-containing RRs in prokaryotes (51), the unphosphorylated ARR1^RD-DBD may exist in equilibrium between inactive and active conformations, while phosphorylation tunes the equilibrium toward the active state. The monomeric unphosphorylated ARR1^RD-DBD adopts a closed conformation containing extensive contacts between the RD and DBD, holding the DNA-recognition helix (α8) in a position that is unfavorable for DNA binding. Phosphorylation of the RD at D89 results in a conformational alternation of β6, weakening the interactions surrounding the β6-α8 face. The structural rearrangement of α8 enables binding with the DNA harboring the GAT motif. Remarkably, prior observations support our model that the transactivating function of ARR1 was hidden by its RD (37) and transgenic Arabidopsis plants overexpressing ARR1 lacking the RD, but not the full-length ARR1, exhibited a robust phenotypic change (52). Similarly, deletions or impairment of the RDs in Arabidopsis ARR11, ARR14, ARR18, ARR20, and ARR21 cause a strong cytokinin phenotype (43, 44, 53).

Discussion

DBD-containing RRs make up the largest class of prokaryotic RRs. An increasing number of X-ray structures in this subfamily reveal allosteric conformational features and prevalent regulatory strategies (47). The RD often facilitates the function of the effector domain through both positive and negative regulations, as identified in Staphylococcus aureus VraR (54). The ARR1^RD showed high structural similarity to the RD of RRs in prokaryotes (SI Appendix, Fig. S7). Phosphorylation of the RR can alleviate RD inhibition, promote RR dimerization for DNA binding, or both. While our structural analyses of ARR1^BD-DBD support a relief inhibition model for the activation of B-ARRs, we cannot exclude the possibility that the RD of plant B-ARRs also dimerizes upon phosphorylation and dimerized RDs can promote DNA binding and transcription. Therefore, determining the structure of phosphorylated or BeF₃-bound ARR1^RD-DBD will assist us in developing this hypothesis in the future.

Our ARR1^RD-DBD structure did not resolve the conformation of the loop region (residues 170–238) between the RD and DBD. Sequence alignment across Arabidopsis B-ARRs reveals little conservation of these regions in both sequence and length (SI Appendix, Fig. S8). Given the importance of the β6-α8 face (i.e., the face between the RD and DBD), this segment may play a regulatory function. For example, a protein may interfere with the interaction between the RD and DBD via binding to the loop region. Similarly, reducing the length of the loop region may result in weakened contact between β6 and α8. Thus, the identification and functional characterization of proteins associated with this region represent another interesting future research topic.

A-ARRs repress cytokinin signaling through a negative feedback loop (55). The arr3 arr4 arr5 arr6 arr7 arr8 arr9 septuple mutants are viable but possess phyllotaxis and organ initiation defects, while the arr7 arr15 double mutant exhibits female gametophytic lethality (56). As B-ARR, A-ARR function requires phosphorylation (57), the mechanism whereby A-ARRs negatively regulate cytokinin signaling is poorly understood. It has been proposed that A-ARRs compete with B-ARRs for phosphoryl groups. Alternatively, the negative regulation by A-ARRs may be facilitated through phosphospecific interactions with target proteins (57, 58). To distinguish between these two alternatives, we must probe phosphorylation-dependent conformational changes of A-ARRs and determine the structure of the A-ARR-AHP complex. Notably, the comparison between A-ARR and B-ARR structures may identify whether phosphorylated A-ARRs could interact with phosphorylated B-ARRs via their RDs, forming inactive heterodimers.

Materials and Methods

Constructs and Purification of Recombination Proteins.

To construct plasmids for prokaryotic expression of ARR1^RD-DBD (residues 1–301) and ARR1^DBD (residues 221–301) in Escherichia coli, the coding sequence of ARR1 was PCR-amplified and cloned into pRSF-Duet (https://www.snapgene.com/plasmids/pet_and_duet_vectors_(novagen)/pRSFDuet-1). The plasmids were transformed into chemically competent E. coli BL21 (DE3) or Rosette (DE3) cells. The transformed bacterial cells were grown in Luria-Bertani (LB) medium supplemented with kanamycin at 37 °C to an OD₆₀₀ of 0.8 and then induced with 0.2 mM isopropyl β-D-thiogalactopyranoside (IPTG, Sangon Biotech, Cat No./ID: 367931). The cells were harvested and resuspended in buffer A (20 mM Tris-HCl, pH 8.0, 100 mM NaCl) supplemented with 1 mM phenylmethanesulfonyl fluoride (PMSF; Sigma-Aldrich, Cat No./ID: 93482). Cells were lysed by a high-pressure cell disruptor at 15,000 p.s.i. (pounds per square inch). The lysate was centrifuged at 20,000 × g for 45 min and the supernatant was loaded onto a Ni²⁺-NTA affinity column (Qiagen, Cat No./ID: 30230) and washed with buffer A with 20 mM imidazole. Proteins were eluted with buffer A supplemented with 250 mM imidazole and purified with Source Q, followed by gel filtration with a Superdex 200 column (GE Healthcare, Cat No./ID: 28-9909-44) in buffer A. Peak fractions were collected and concentrated for subsequent structural and biochemical studies. For selenomethionine (SeMet)-derived protein expression, the constructs were transformed into E. coli B834 (DE3) cells, and the cells were cultured in M9 medium containing 50 mg/L SeMet. SeMet-labeled protein expression was performed as previously described (59).

Protein Crystallization.

Purified protein was buffer exchanged into 20 mM Tris-HCl and 100 mM NaCl (pH 8.0) and concentrated to 10 mg/mL. Crystals of the ARR1^DBD–DNA and ARR1^RD-DBD were grown using the vapor diffusion method at 20 °C for 1 wk. The best ARR1^RD-DBD crystals were grown in 2.0 M ammonium sulfate, 2% (v/v) Polyethylene Glycol (PEG) 400, 0.1 M HEPES, pH 7.5. To obtain the crystals of the ARR1^DBD in a complex with DNA^TCS, SeMet-derived protein was incubated with 1.2-fold molar amounts of DNA^TCS on ice for 30 min before crystallization. Crystals of the ARR1^DBD–DNA complex were obtained under the following condition: 0.2 M ammonium sulfate, 0.1 M sodium acetate trihydrate pH 4.6, 30% (w/v) PEG monomethyl ether 2000. All crystals were transferred into cryoprotectant solution containing their respective mother liquor plus 30% (v/v) glycerol, before being flash-frozen in liquid nitrogen for storage.

Data Collection and Structure Determination.

Data collection was performed at BL19U1/ BL18U/ BL17U beamline of the Shanghai Synchrotron Radiation Facility under 100 K liquid nitrogen stream (wavelength = 0.9798 Å). The data were processed with HKL3000 (60). All of the models were manually built in Coot (61) and were refined by iterative rounds of manual adjustment with Coot and refinement with Phenix (62). Data collection and structure refinement statistics are shown in SI Appendix, Table S1.

DNA Binding Kinetics Measurement.

We employed bio-layer interferometry (BLI) to measure the binding kinetics between DBD and DNA^TCS. For the binding assays, purified proteins were prepared, except that buffer A was replaced by PBS buffer (10 mM phosphate buffer pH 7.4, 2.7 mM KCl, 137 mM NaCl). The 40-bp DNA^TCS fragments (5′ ggattagattagattaatctggattagattaatctagatt 3′ and 5′ aatctagattaatctaatccagattaatctaatctaatcc 3′) were prepared from synthetic oligonucleotides by an annealing procedure to form double-stranded DNA molecules in PBS buffer. Biotinylated ARR1^DBD or ARR1^RD-DBD-coated sensors were then incubated with DNA fragments at 0.5 μM for 30 s to record the signal resulted from association and then incubated with PBS buffer for 30 s to record the signal resulted from dissociation. The background signal was recorded using a reference sensor with ARR1^DBD loaded, but no analyte DNA during the association phase. The mean on-rate (K_on), off-rate (K_off), dissociation constant (K_D) values, and curve fitting were performed using the ForteBio data analysis software.

In Vitro Pull-down Assays.

The mutated versions of ARR1^RD and ARR1^DBD were generated by PCR-based mutagenesis. The oligonucleotide primers are given in Dataset S1. The 6xHis-tagged wild-type and mutated ARR1^RD proteins were coexpressed with Glutathione S-transferase (GST)-tagged wild-type and mutated ARR1^DBD in E. coli Rosette (DE3). After cell disruption and centrifugation, the supernatants were loaded to Ni²⁺-NTA affinity resin (Qiagen, Cat No./ID: 30230). The nonspecific bound protein was washed off. The resin was boiled, examined with Sodium Dodecyl Sulphate-Polyacrylamide Gel Electrophoresis (SDS-PAGE), and visualized by Coomassie Blue staining.

Tobacco Transient Assays.

For tobacco transient assay, we used TCS:LUC as the reporter as previously described (63). The 35S:ARR1-3xHA series constructs were generated by cloning the 3xHA-tagged wild-type or mutated ARR1 into the binary vector JW807 behind the 35S promoter. The oligonucleotide primers are given in Dataset S1. The binary constructs were delivered into Agrobacterium tumefaciens strain GV3101 (pMP90) by the freeze–thaw method. Tobacco transient assays were performed as described (63). Briefly, Agrobacteria were resuspended in infiltration buffer (10 mM methylester sulfonate, 10 mM MgCl₂, and 150 μM acetosyringone, pH 5.7) at OD₆₀₀ = 0.8. Agrobacteria harboring 35S:P19-HA (OD₆₀₀ = 0.8) were coinfiltrated to inhibit gene silencing (64). The ratio of TCS:LUC, 35S:ARR1, and 35S:P19-HA was 1:1:1. Luciferin (1 mM) was infiltrated before LUC activity was measured after three days. To quantify LUC activity, we used a dual-LUC reporter system in which 35S:RENILLA (REN) was used as an internal control (65). The LUC activity was quenched before REN activity was measured with a luminometer (Promega GloMax 20/20, Cat No./ID: E5311). Three independent experiments (biological replicates) were performed.

Plant Materials and Growth Conditions.

The A. thaliana and N. benthamiana plants were grown at 22 °C in growth chambers under long day (LD, 16-h light/8-h dark) conditions. The A. thaliana ecotype Columbia-0 (Col-0) was used as wild type. The arr1 arr10 arr12 mutant was described before (66).

Constructs and Generation of Transgenic Plants.

We used the 35S promoter to drive the expression of 3xFLAG- or 3xHA-tagged wild-type and mutated versions of ARR1. Site mutagenesis was performed using a PCR-based approach. The oligonucleotide primers are given in Dataset S1. The binary constructs were delivered into A. tumefaciens strain GV3101 (pMP90) by the freeze–thaw method. Transgenic plants were generated by the floral dipping method (67) and screened with 0.05% glufosinate (Basta) on soil.

Primary Root Length Measurement.

For all the measurements, we used two independent T₂ transgenic lines with comparable expression level for each genotype. Briefly, seeds were sown on Murashige and Skoog (MS) agar plates, stratified at 4 °C in darkness for 3 d, and grown in a vertical position under LD conditions in a Percival chamber for 6 d. The root phenotype of the wild-type and transgenic plants (n ≥ 14) was measured by Image J.

Gene Expression Analysis.

Total RNA was extracted with Trizol reagent (Invitrogen, Cat No./ID: 217084). Total RNA (1 µg) was treated with 1 µL of DNase I (1.0 unit/μL; Thermo Fisher, Cat No./ID: EN0521), and complementary DNAs (cDNAs) were synthesized with the All-in-one Kit (Vazyme, Cat No./ID: R333-01). Quantitative real-time PCR (qRT-PCR) was performed using a mix containing 10 μL of SYBR^® Premix Ex Taq™ (TaKaRa, Cat No./ID: RR820B), 2 μL of forward and reverse primer mix, 0.2 μL of ROX reference dye II, 1 μL of cDNA, and 6.8 μL of deionized water. The relative gene expression levels were calculated by 2^−ΔΔCt values and normalized against Ubiquitin-conjugating enzyme 21 (UBC21, AT5G25760) (68). Three independent experiments (biological replicates) were performed. The oligonucleotide primers for all genes are given in Dataset S1.

Western Blot.

Proteins were extracted with lysis buffer [50 mM Tris-HCl, 100 mM NaCl, 10% glycerol, 5 mM EDTA, 0.1% Triton X-100, 0.2% Nonidet P-40, 50 mM MG132, and complete protease inhibitor cocktail tablet (Roche, Cat No./ID:11697498001), pH 7.5] and centrifuged twice at 13,000g for 5 min. The fusion proteins were detected by immunoblotting using monoclonal anti-HA-HRP (Roche, Cat No./ID: 12013819001) or anti-TUBULIN (Sigma-Aldrich, Cat No./ID: T5168) antibodies.

Phylogenetic Tree.

BLASTp was performed using the deduced amino acid sequence from ARRs from different species. The accession numbers are given in the figure legend. Multiple sequence alignments and phylogenetic tree were generated by MEGA11 (69). Maximum Likelihood method was used to draw phylogenetic trees (70). Bootstrap analysis was performed to estimate nodal support on the basis of 1,000 replicates.

Statistics and Reproducibility.

Two-tailed Student’s t test and ordinary one-way ANOVA with Tukey test were performed to determine the statistical significance between different samples in quantification and phenotypes. For t test, P value is given in figure if there is statistical significance (P < 0.05), otherwise it will be noted as “ns” for no significant difference. For ANOVA, different letters indicate significant differences. All statistic results and graphs were generated by GraphPad Prism 8 (www.graphpad.com). The detailed descriptive statistics and results are given in Dataset S2.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Data, Materials, and Software Availability

The atomic coordinates of the ARR1^DBD–DNA (2.35 Å) and ARR1^RD-DBD (2.2 Å) structures have been deposited in the Protein Data Bank with accession codes 8XAS (71) and 8XAT (72), respectively.