Molecular switching in transcription through splicing and proline-isomerization regulates stress responses in plants

Theisen, Frederik Friis; Prestel, Andreas; Elkjær, Steffie; Leurs, Yannick H. A.; Morffy, Nicholas; Strader, Lucia C.; O’Shea, Charlotte; Teilum, Kaare; Kragelund, Birthe B.; Skriver, Karen

doi:10.1038/s41467-024-44859-2

Download PDF

Article
Open access
Published: 18 January 2024

Molecular switching in transcription through splicing and proline-isomerization regulates stress responses in plants

Nature Communications volume 15, Article number: 592 (2024) Cite this article

2235 Accesses
1 Citations
9 Altmetric
Metrics details

Subjects

Abstract

The Arabidopsis thaliana DREB2A transcription factor interacts with the negative regulator RCD1 and the ACID domain of subunit 25 of the transcriptional co-regulator mediator (Med25) to integrate stress signals for gene expression, with elusive molecular interplay. Using biophysical and structural analyses together with high-throughput screening, we reveal a bivalent binding switch in DREB2A containing an ACID-binding motif (ABS) and the known RCD1-binding motif (RIM). The RIM is lacking in a stress-induced DREB2A splice variant with retained transcriptional activity. ABS and RIM bind to separate sites on Med25-ACID, and NMR analyses show a structurally heterogeneous complex deriving from a DREB2A-ABS proline residue populating cis- and trans-isomers with remote impact on the RIM. The cis-isomer stabilizes an α-helix, while the trans-isomer may introduce energetic frustration facilitating rapid exchange between activators and repressors. Thus, DREB2A uses a post-transcriptionally and post-translationally modulated switch for transcriptional regulation.

DHH1/DDX6-like RNA helicases maintain ephemeral half-lives of stress-response mRNAs

Article 01 June 2020

Poly(ADP-ribose)-binding protein RCD1 is a plant PARylation reader regulated by Photoregulatory Protein Kinases

Article Open access 19 April 2023

Overlapping roles of spliceosomal components SF3B1 and PHF5A in rice splicing regulation

Article Open access 05 May 2021

Introduction

Transcriptional pathways regulate diverse biological processes such as stress responses and development, and they culminate in regulation of gene expression by transcription factors (TFs)¹. For this, interactions between activation domains (ADs) of TFs and co-regulators are essential by ensuring the correct location of RNA polymerase II². ADs are interchangeable and can bind unrelated co-regulators³, and most ADs are intrinsically disordered with low sequence conservation^4,5. This has spurred the idea that interactions between ADs and co-regulators are short-lived and non-specific with stochastic burial of hydrophobic residues^4,6,7. The mediator multi-protein complex is an important co-regulator, responsible for facilitating interactions between the basal RNA polymerase II machinery and TFs. Studies of the interactions between mediator subunits and TFs contribute to unraveling the complex nature of TF:co-regulator interactions, and have revealed emerging themes of multivalency, coupled folding and binding, and dynamic interfaces. Thus, dynamic and multivalent interactions were demonstrated for the interaction between ETV4 and the activator interaction domain (ACID) of mediator complex subunit 25 (Med25)⁸, bivalency was shown for the interactions between Med25-ACID and VP16^9,10 and p53¹¹, and the interface between Med25-ACID and ETV5 was dynamic with the AD of ETV5 undergoing coupled folding and binding upon complex formation¹².

Plant Med25 is implicated in a range of biological processes spanning from plant development to hormone signaling and stress responses¹³. Arabidopsis thaliana Med25 physically interacts with abiotic-stress-associated TFs, including Drought Response Element Binding protein 2 A (DREB2A), through its ACID domain¹⁴. The Arabidopsis dreb2a and med25 mutants both displayed increased sensitivity to salt stress. However, Med25 and DREB2A had opposite functions in response to drought with DREB2A and Med25 increasing and decreasing drought resistance, respectively^14,15. Together, Med25 and DREB2A also function in the repression of PhyB-mediated light signaling and thus integrate signals from different important regulatory pathways in plants¹⁴.

The RCD1-SRO-TAF4 (RST) αα-hub domain^16,17,18 is responsible for negative regulatory interactions of Arabidopsis Radical-induced Cell Death1 (RCD1) with DREB2A^19,20. Downregulation of RCD1 or loss of the RCD1-interacting region of DREB2A, as in a DREB2A splice variant (Fig. 1A), was required for DREB2A accumulation during heat stress and senescence and thus for DREB2A function under stress conditions²¹. DREB2A interacts with RCD1-RST using a short linear motif (SLiM), the RCD1-interaction motif (RIM)²¹, which undergoes coupled folding and binding²⁰. Recently, the SLiM context was shown to increase affinity by stabilizing the complex²². The same SLiM has previously been hypothesized to facilitate the interaction with AtMed25-ACID²³, however, full-length DREB2A was necessary to achieve strong binding²⁴. Together, this suggests that the disordered region surrounding the RIM is a functional hot spot (Fig. 1).

**Fig. 1: Identification of the Med25-ACID binding region of DREB2A with a primary SLiM.**

In this work, we explore features of the functional hot spot to obtain mechanistic insight into molecular switching in transcriptional regulation. We identified a SLiM, the ACID-binding motif (ABS), which in conjunction with the RIM forms a bivalent Med25-ACID-interacting region that is functionally regulated by alternative splicing. Complementary structural analyses revealed that DREB2A bound to a groove and a hydrophobic surface on the ACID domain, and chemical exchange saturation transfer (CEST) NMR showed that the Med25-ACID:DREB2A complex was structurally heterogenous due to a DREB2A proline residue occupying highly populated cis and trans isomers. This introduces proline isomerization as a player in mediator:TF complexes, facilitating the energetic frustration needed for regulator exchange, and adding an additional layer of regulatory functionality to a transcription factor switch unit.

Results

The Med25-ACID-binding region of DREB2A contains two SLiMs

To define a DREB2A core region needed for Med25-ACID binding, we initially produced a long²³ fragment spanning from Val151 to the C-terminus, constituting the entire C-terminal intrinsically disordered region (IDR) of DREB2A (Fig. 1A). The interaction was characterized using isothermal titration calorimetry (ITC) (Supplementary Fig. 1) and for DREB2A_151-335 we obtained a K_d of 540 ± 40 nM (Table 1, Fig. 1A), with thermodynamics indicating an enthalpically driven interaction (ΔH = −52.3 ± 0.8 kJ/mol) with an entropic penalty (-TΔS = 16.5 ± 1.0 kJ/mol).

Table 1 Thermodynamic parameters of DREB2A interaction with Med25-ACID at 25 °C

Full size table

To narrow down the Med25-ACID-binding region, DREB2A was truncated from both termini. Based on considerations of the ID-profile and predicted secondary structure, three DREB2A fragments (DREB2A_195-276, DREB2A_234-335, and DREB2A_234-276) were produced, all showing the similar binding affinity and thermodynamic parameters (Table 1 and Fig. 1A). Further truncation of the C-terminus (DREB2A_234-272 and DREB2A_234-256) resulted in a significant reduction of binding affinity, increasing K_d to 1.8 ± 0.4 µM and 6.0 ± 0.2 µM, respectively. The RIM (residues 259-269, Fig. 1) was thus involved, but not necessary for binding. Although the RIM contributed substantial interaction energy, it was not possible to measure the affinity of the RIM by itself (DREB2A_255-276) using ITC. We then produced the DREB2A_{195-335,Δ244-276} variant, which had an internal deletion of a region (Trp244-Gly276), absent in a DREB2A splice variant unable to bind RCD1²¹ but still with transcriptional activity²⁵. This variant bound Med25-ACID with reduced affinity (K_d = 5.4 ± 0.6 µM). Comparison of DREB2A sequences from phylogenetically representative plant species revealed that DREB2A_234-276 contains two separate conserved regions, one corresponding to the RIM present in DREB2A from land plants²⁶, and a putative motif with the consensus sequence LxVxD[YFL][GS]W[PI] (Fig. 1B). This region was designated the ACID-binding SLiM (ABS).

Since aromatic and acidic residues are important for binding of ADs to co-regulators²⁷, we analyzed how their mutations affected Med25-ACID binding (Table 1, Supplementary Fig. 2). Changing Asp241, Tyr242, or Trp244 of the ABS to alanine resulted in either a four-fold decrease in affinity (D241 A and W244 A) or no detectable binding (Y242A). For Phe259, which is essential for DREB2A binding to RCD1-RST²⁰, substitution to alanine hardly affected Med25-ACID binding affinity. Substitution of Phe274 with alanine also had only a minor effect on binding, although Phe274 belongs to the Val273-Gly276 region which affected the affinity considerably when removed (Table 1). In conclusion, the ABS serves as the primary Med25-ACID binding motif, although the RIM region also contributes. Hence, the RIM was defined as the secondary Med25-ACID binding motif.

The AD of DREB2A was previously mapped to the C-terminal residues 254-335 in Arabidopsis protoplasts²⁵. To examine whether the RIM and ABS are both active in transcription, we analyzed the AD pattern using a high-throughput yeast-based assay based on synthetic TFs containing a DNA-binding domain, targeting a GFP reporter gene, an mCherry fluorescence tag and a variable putative AD (Supplementary Fig. 3). The screen was performed using 40-residue overlapping tiles spanning the entire DREB2A sequence with the AD score of each tile calculated from the GFP:mCherry fluorescence intensity ratio, thus reflecting the transcriptional activity²⁸. This revealed that the C-terminal AD²⁵ contains two distinct ADs, AD1 (app. residues 250-270) and AD2 (app. residues 320-335) (Fig. 1A), defined as residues for which all tiles displayed an AD score above the threshold (Supplementary Fig. 3E). The RIM overlapped with AD1 whereas the ABS, although enhancing the effect of the RIM, did not by itself correlate with transcriptional activity.

AtMed25-ACID forms a β-barrel with two accessory helices and dynamic regions

The structure of the human hMed25-ACID domain was previously determined using NMR spectroscopy^9,10. However, sequence alignments of At- and hMed25-ACID showed low sequence identity²³ (Supplementary Fig. 4A) and the AlphaFold2 (AF2) prediction of the AtMed25-ACID structure (Supplementary Fig. 4B)^29,30 differed from the published hMed25-ACID structure. Both domains form a core β-barrel with accessory α-helices of various lengths, but hMed25-ACID featured an additional helix and a considerably longer loop 1 (L1). To validate the predicted structure of the AtMed25-ACID domain, the backbone NMR chemical shifts were assigned (Supplementary Fig. 5A). Secondary chemical shifts of ¹³C^β nuclei showed positive signals suggesting a structure composed mostly of β-strands (Supplementary Fig. 5B). Secondary structure propensities (SSPs) were then quantified based on multiple nuclei using the motif identification from chemical shifts (MICS) server³¹ (Fig. 2A) and mapped onto the AF2 structure model (Fig. 2B). This showed an excellent agreement between prediction and data which was further supported by small-angle X-ray scattering (SAXS) data validating the overall dimensions of the AtMed25-ACID domain (χ² = 1.008, Supplementary Fig. 5C).

**Fig. 2: Characterization of the AtMed25-ACID domain structure using NMR, MD, and AlphaFold2.**

Using molecular dynamics (MD), the predicted structure was simulated for 1.7 µs to examine the dynamics of the Med25-ACID domain structure. Several regions exhibited larger fluctuations on the ns timescale (Fig. 2C and D). We therefore recorded backbone {¹H}-¹⁵N heteronuclear NOEs (hetNOEs), as well as ¹⁵N transverse (R₂) and longitudinal (R₁) relaxation rates of AtMed25-ACID (Fig. 2C), which report on the ps to ns timescale dynamics of the backbone amides. These data provided some indication of increased dynamics of the loop regions, although chemical exchange contributions made definite conclusions difficult. The hetNOEs, which are not affected by chemical exchange, revealed a clearer pattern with increased dynamics (lower hetNOE) of the backbone amides in L1 and H1L3, and to a smaller degree in L2 and L4.

Thus, the analyses of secondary structure, overall shape, and dynamics of the AtMed25-ACID domain were in good agreement with the AF2 prediction, suggesting that the AF2 structure is a suitable model of the AtMed25-ACID domain.

DREB2A binds a groove in AtMed25-ACID formed by H2 and L1

To identify the DREB2A binding sites in Med25-ACID, we performed ¹⁵N-HSQC titration experiments using ¹⁵N-labeled Med25-ACID. Mapping of ¹⁵N,H^N chemical shift perturbations (CSPs), induced upon addition of ABS (DREB2A_234-256, Fig. 1A) (Fig. 3A), revealed CSPs localized around a groove defined by the dynamic L1 and the H2 α-helix (Figs. 2D and 3B-D). The largest CSPs were seen around L1 while the effects in H2 were smaller, indicating that L1 undergoes structural changes, while H2 remains relatively unaffected. To examine the bound state dynamics of L1 we recorded {¹H}-¹⁵N hetNOEs of Med25-ACID in complex with the ABS (Supplementary Fig. 6A). Comparison of the hetNOEs of the folded ACID domain in the free and ABS bound states revealed two regions with significant differences, one of which mapped to L1 and the other to the C-terminus of H2.

**Fig. 3: Mapping of DREB2A binding sites in Med25-ACID using NMR.**

To validate the proposed binding groove, we introduced an alanine in L1, replacing Arg568, which experienced a large CSP upon DREB2A binding, and determined the binding affinity using ITC with the high-affinity ABS-RIM fragment (DREB2A_234-276, Fig. 1A). This gave a K_d of 3.1 ± 0.7 µM, more than five-fold higher than that of the wild type (Table 1, Supplementary Fig. 7), confirming that L1 is important for binding. Thus, both NMR and thermodynamics supported the importance of the groove for the interaction with DREB2A.

Since inclusion of the RIM resulted in a higher affinity (Table 1), we performed an NMR titration using the bivalent ABS-RIM to determine the interaction site of the RIM. Similar CSPs, both in magnitude and direction, were observed for residues in the binding groove, suggesting similar binding of the ABS with and without the RIM. However, additional CSPs were observed on a surface of Med25-ACID extending from the ABS-binding groove (Fig. 3E). Furthermore, the addition of ABS-RIM resulted in a reduction of the spectral quality with many peaks disappearing at stoichiometric and higher DREB2A concentrations. Analysis of NMR peak intensity changes identified several regions affected by addition of ABS-RIM (Fig. 3C) and mapping the largest effects (Supplementary Fig. 6B) highlighted the same area as the additional CSPs (Fig. 3E), further supporting this region as the Med25-ACID binding surface of the RIM.

While ITC experiments using the RIM fragment (DREB2A_255-276, Fig. 1A) did not show any binding to Med25-ACID by itself, NMR titrations showed large CSPs upon its addition. Unexpectedly, both the binding groove and the binding surface of Med25-ACID were affected, suggesting an ability of the RIM to bind both sites (Fig. 3B). Analysis of the RIM-induced CSPs as a function of concentration identified two different affinity regimes, with a K_d of 200-400 µM for the groove and a K_d of 3–6 mM for the surface (Supplementary Fig. 6C). While this indicated that the RIM interaction with the binding surface was relatively weak compared to that of the groove, the RIM interaction of the bivalent DREB2A fragments will be influenced by the effective concentration imposed by high-affinity ABS binding. Extrapolation to a linker length of 13 residues, as in the bivalent ABS-RIM, gave an effective RIM concentration of around 9 mM near Med25-ACID with the ABS bound³², resulting in an expected 65 ± 10% saturation of the RIM towards the binding surface. Comparison of ABS, ABS-RIM, and RIM bound state chemical shifts showed that residues located in the ABS-binding groove were affected differently by the RIM, compared to the ABS and ABS-RIM fragments, which showed similar CSPs in the groove, while residues of the RIM-binding surface showed similar perturbations for the RIM and ABS-RIM fragments (Fig. 3A). This suggested that, for their cognate sites, the bound state of the isolated SLiMs resembles that seen in the bivalent fragment, but that using the isolated motifs resulted in incorrect binding modes. Together, these results suggested that the ABS binds the groove, while the RIM targets the binding surface.

AF2 modeling supports bivalent binding

We next exploited the AF2 heterodimer modeling option on the ColabFold server³³ to predict the structure of the Med25-ACID:ABS-RIM complex. Out of three different complex structure configurations, two predicted the ABS to bind the groove and the RIM to bind the surface, in agreement with the NMR data (Fig. 3D, E). The third model (model 3) swapped the positions of the ABS and RIM and was therefore discarded based on the NMR data (Fig. 3A). The two remaining models differed by the direction of DREB2A wrapping around the ACID domain, with model 1 having DREB2A wrapping along β2, while model 2 had DREB2A wrapping around the H1L3 side. In addition, model 1 had Trp244 buried in the binding groove, while model 2 had Tyr242 buried in the groove (Fig. 4B). Based on the mutational analysis, which showed Tyr242 to be crucial for binding while Trp244 was less important (Table 1), model 2 was considered the more likely model.

**Fig. 4: Modeling of the Med25-ACID:DREB2A complex using AF2.**

The model indicated that Leu237 and Val239 are important for binding as they were buried in the binding groove. To test this, two mutant DREB2A fragments were produced, individually substituting each residue with alanine, and the affinities were measured using ITC (Table 1). The V239A and L237A variants both exhibited slightly reduced affinities with K_d values of 2.0 ± 0.2 µM and 2.4 ± 0.3 µM, respectively. However, the L237A variant exhibited distinct thermodynamics, featuring a favorable entropic contribution. This indicated structural effects of mutations of ABS residues buried in the binding groove, consistent with the ITC results of the Y242 A variant. Together the ITC and NMR (Fig. 2A) data support model 2, although alternative configurations may be populated under certain conditions.

Based on the AF2 model, binding of the RIM involves a hydrophobic surface of Med25-ACID (Fig. 4C), which was in line with the limited effect of mutating single phenylalanine residues (Table 1). All AF2 models demonstrated α-helical structure in the RIM, yet a helix-disrupting DREB2A E263P variant showed unchanged affinity (K_d = 580 ± 40 nM; Table 1), indicating that helix formation is not a prerequisite for binding and that multiple binding modes may exist. In agreement with the H1L3 wrapping model (Fig. 4A, model 2), evolutionary analysis of plant Med25-ACID revealed that the opposite side, in particular β2, was less conserved than the rest of the ACID domain (Supplementary Fig. 8), suggesting reduced requirements to support specific interactions. Overall, the complexes predicted using AF2 agreed well with the experimental data (Fig. 4).

Multiple bound states of DREB2A uncovered using CEST

Previous characterization of free DREB2A revealed that the region around Pro245 exists in two states, likely resulting from Trp244 stabilizing the cis peptidyl-prolyl Trp244-Pro245 isomer³⁴. Quantification of HSQC peak volumes of free DREB2A (Supplementary Fig. 9 and Supplementary Table 1) showed that 30 ± 1% was in the cis proline form, resulting in an equilibrium constant (K_{eq,trans→cis}) of 0.42 ± 0.02. Attempts to use NMR to study DREB2A in the Med25-ACID bound state were hindered by line broadening. Instead, we exploited the CEST-type NMR experiments^34,35, which can detect lowly populated states by observing the modulation of the unbound state. Initial analysis of all non-overlapping peaks (Supplementary Fig. 10A) revealed two-state behavior of both the cis and trans proline isomers of the ABS region, indicating that both isomers bind Med25-ACID (Fig. 5A), although differences in the bound populations suggested different affinities of cis and trans DREB2A towards Med25-ACID (Supplementary Fig. 10B). Many peaks from the RIM displayed three-state behavior (Fig. 5A) with large CSP differences between the two bound states, suggesting structural heterogeneity. The three-state behavior may be caused by different structural constraints imposed by the two proline isomers. We therefore produced a DREB2A variant with a substitution of Pro245 with alanine (ABS-RIM_P245A), which eliminated the three-state behavior of the RIM (Supplementary Fig. 10C). Structural heterogeneity of the bound RIM, driven by Pro245 isomerization, may explain the HSQC peak intensity loss of Med25-ACID RIM-binding surface residues upon binding of ABS-RIM (Fig. 3C and Supplementary Fig. 6B).

**Fig. 5: CEST and stopped-flow spectroscopy analyses of DREB2A showing two bound states.**

To enable global analysis of the CEST data, we grouped the CEST profiles into trans ABS, cis ABS, and apparent three-state RIM peaks (Supplementary Figs. 10A–C). Analysis of the ABS confirmed the distinct behavior of cis and trans isomer peaks, with the cis isomer having a larger bound fraction of 6.8 ± 0.1% compared to the 3.6 ± 0.1% of the trans isomer, supporting a difference in affinity. Based on the overall affinity of 500 ± 20 nM, determined using ITC, we estimated the K_d of the trans and cis isomers to 640 ± 30 nM and 330 ± 20 nM, respectively (Fig. 5B and Supplementary Method 1). The differences in bound fractions were accompanied by different exchange rates, indicating that the dissociation rate constants differed for the two isomers, similar to the proline isomerization effects described for the ACTR:NCBD interaction³⁶. Global analysis of three-state ABS-RIM CEST profiles suggested that approximately 4.7% was in one of the two bound states, indicating that the RIM of the bivalent DREB2A fragment was found exclusively in a bound state (Supplementary Method 1). This was more than expected based on the NMR-derived affinity of the RIM alone and the linker-length-based estimates of the effective concentration, suggesting either that the linker also contributes to binding or that binding of ABS results in cooperative binding of the RIM, as observed for other Med25-ACID interactions³⁷.

We then extracted ¹⁵N CSPs for ABS-RIM and mapped them onto the AF2 complex model (Fig. 5C, D). Large CSPs were observed for both motifs while the linker region showed only minimal perturbations, indicating weak contact and binding effects. In agreement with the complex model, hydrophobic residues in the interface (Fig. 4C) generally experienced larger CSPs (Fig. 5C, D).

To further explore the observed proline-dependent effect on the exchange rate, we determined the association (k_on) and dissociation (k_off) rate constants with stopped-flow fluorescence spectroscopy using FITC-labeled DREB2A and unlabeled Med25-ACID (Fig. 5E and Supplementary Fig. 11). A k_on of 29 ± 1 µM⁻¹ s⁻¹ was obtained for the ABS; however, ABS-RIM showed a non-linear concentration dependence of the observed rate constant (k_obs), indicating a complex binding model with potential contributions from non-specific binding. Consequently, a k_on could not be obtained for the ABS-RIM. The dissociation was analyzed using competitive displacement experiments. Time-dependent fluorescence traces of the ABS dissociation could be adequately fitted using a single exponential decay function, whereas a double exponential function was necessary to obtain a good fit of the ABS-RIM dissociation (Fig. 5E and Supplementary Fig. 12). This was in accordance with the two distinct k_ex values obtained from the CEST analysis (Fig. 5B) and suggested that the effect of proline isomerization was insignificant in the absence of the RIM. The k_off values obtained for the ABS-RIM (Fig. 5E) were similar to the cis and trans proline state k_ex values obtained from CEST experiments conducted at 10 °C (Fig. 5B), confirming that the biphasic behavior of the dissociation kinetics was the result of proline isomer dependent structural differences.

DREB2A undergoes coupled folding and binding forming two short α-helices

To analyze the secondary structure of the DREB2A bound state, we determined the ¹³C^α and ¹³C’ chemical shifts using triple-resonance ¹³C-CEST experiments^38,39. Due to the reduced sensitivity of the experiments, it was not possible to analyze the relatively weak cis proline-associated peaks. However, the ¹³C-CEST profiles of the RIM (Fig. 6A and Supplementary Fig. 13) supported the three-state behavior seen in the ¹⁵N-CEST profiles (Fig. 5A), not only confirming the existence of two bound states, but also indicating that the structure of the two bound states differed. CEST-derived CSPs of the DREB2A_P245A variant showed that whereas the ABS was not greatly affected by the proline substitution, the RIM displayed reduced CSPs, resembling the less affected of the two bound states seen in the WT (Supplementary Figs. 10C and 14). This suggested that the large RIM CSPs seen in the WT belong to the cis proline bound state. To quantify the secondary structure of the two bound states, we compiled two sets of ¹⁵N, ¹³C^α, and ¹³C’ chemical shifts, extracted from the minimum (trans) and maximum (cis) CEST CSPs and submitted these to the δ2D SSP webserver⁴⁰ (Fig. 6B). For the cis state, predictions indicated formation of well-folded helical structure in the RIM, matching the AF2 complex model (Fig. 6C). Conversely, the trans state showed reduced helical structure in the RIM, suggesting that it can bind to Med25-ACID in both a helical and a non-helical structure, depending on the isomeric state of Pro245.

**Fig. 6: ¹³C-CEST and thermodynamic analyses show coupled binding and folding of DREB2A.**

To further explore the degree of structuring, we estimated the number of residues folding upon binding (R_th,ID) using the ID-adapted thermodynamic Spolar-Record method^22,41. For this, we used three DREB2A fragments; the monovalent ABS fragment, the bivalent ABS-RIM fragment, and the splice variant fragment (Fig. 6D). The R_th,ID value of the ABS (17 ± 3) was larger than the number of residues in the DREB2A fragment which suggested structuring of the ACID domain. Comparison of the R_th,ID values of ABS and ABS-RIM (28 ± 5) suggested that the RIM undergoes coupled folding and binding, matching the δ2D chemical shift analysis (Fig. 6B). The reduced entropic penalty of the longer bivalent fragment suggested dehydration of hydrophobic surface upon binding of the RIM. In agreement with this, substitution of large hydrophobic sidechains with alanine resulted in an increased entropic penalty, which was compensated by favorable enthalpy (Table 1), suggestive of a highly adaptable interface enabled by enthalpy-entropy compensation. Binding of the splice variant, lacking the RIM, but containing long flanking regions, also resulted in folding of more residues (26 ± 7) than for the ABS fragment. In this case, thermodynamic parameters suggest some association of the C-terminal context with the unfavorable change in entropy possibly due to conformational restrictions of ABS residues. Thus, as for the RCD1:DREB2A interaction²², the motif context ensures coupled folding and binding through intricate enthalpy-entropy compensation.

Discussion

In this study, we explored how a co-regulator interacts with a TF through a disordered functional hot spot. Focusing on the DREB2A interaction with Med25 enabled comparison to previous studies with the negative regulator RCD1 and the co-activator TAF4, both of which bind DREB2A through their RST domain⁴². DREB2A interacts with Med25-ACID using a 40-residue region containing two SLiMs. The primary Med25-ACID binding SLiM, the ABS, is characterized by key aromatic, acidic and proline residues, but did not correlate with transcriptional activity in yeast (Fig. 1A and Supplementary Fig. 3B) and plants²⁵. The other SLiM, the RIM, constitutes a secondary binding motif for Med25-ACID, but is the primary interaction motif for the RST domains of RCD1 and TAF4 (Fig. 7A)²¹, and overlaps with AD1. Thus, the interaction regions of co-regulator and repressor overlap, but each interaction is driven by distinct motifs with both interactions showing a 10-fold higher affinity in the bivalent configuration (Table 1)²².

**Fig. 7: Stress dependent positive and negative regulation of DREB2A.**

We also demonstrated that a DREB2A splice variant, expressed under stress and lacking most of the bivalent interaction region including the RIM and AD1, retained the ability to bind Med25-ACID. This suggests that the Med25 interaction may serve as a scaffold for the transcriptional machinery, allowing for DREB2A transcriptional activation through either AD1 or AD2. Although DREB2A has a higher affinity for RCD1-RST (K_d = 12 ± 2 nM)²² than for Med25 (520 ± 20 nM, Table 1) the in vivo competition scenario between e.g. RCD1, TAF4, and Med25 for DREB2A will depend on many other factors, including concentrations. However, alternative splicing of DREB2A will shortcut such regulatory fine-tuning of competition by removing the RIM, thereby eliminating RCD1 as a repressor while leaving AD2 available (Fig. 7B). This provides an obvious mechanism for functional switching in a transcriptional pathway.

As a hub protein, RCD1 binds many stress-related TFs with low nanomolar affinity²⁰ and could act as a leaky expression repressor of e.g. DREB2A. Stress responses may involve rapid degradation of RCD1²¹, thus releasing bound DREB2A to interact with Med25 using its bivalent high-affinity region. This would provide a strong initial response, while the prolonged response would be of a lower potency due to stress-induced expression of the splice variant²¹ lacking the ABS context.

Biophysical and structural characterization together with AF2 modeling uncovered a groove and a surface on AtMed25-ACID capable of binding the ABS and RIM, respectively, and CEST NMR confirmed involvement of both motifs in binding. Although the RIM-binding surface (Fig. 3E) corresponds to the hMed25-ACID:VP16-H2 interaction interface, the ABS-binding groove (Fig. 3D) does not match the VP16-H1 binding site¹⁰. Similarly, a RIM binding surface equivalent was reported for p53-TAD1¹¹ and ERM¹². However, neither ERM nor p53-TAD2 bound to a groove corresponding to that identified for the ABS and although a 40-residue ETV4 fragment induced CSPs in L1, the effect was generally dispersed across the ACID domain⁸ (Supplementary Fig. 15). Thus, the AtMed25-ACID:DREB2A interaction contributes to the diversity in Med25:TF complexes and highlights the theme of multivalency in Med25-ACID:TF interactions^8,9,10,11 conserved across kingdoms. The most striking observation is the identification of the ABS-binding groove. Although this appears unique for the interaction between Med25-ACID and DREB2A, this may be due to experimental challenges of NMR studies using bivalent interaction partners and could be a general feature. Furthermore, the NMR titrations employing the individual motifs and the bivalent association kinetics suggested that alternative binding modes may be populated under certain conditions depending on the cellular context.

Structural heterogeneity and dynamics characterize mediator:TF interactions^43,44, making structural analysis challenging. Here we show that DREB2A exists in highly populated cis and trans Pro245 states. According to CEST analysis, both isomers bound Med25-ACID, with the RIM showing two distinct bound states, displaying considerable α-helical propensity in the cis bound state. This both supported the AF2 complex model and the Spolar-Record analysis by confirming the presence of helical structure in the RIM, and explained why the helix-breaking E263P mutation had no significant effect on affinity. The structural heterogeneity is unlikely to be caused by proline-dependent conformational changes in the ACID domain, but rather by restraints imposed on the disordered DREB2A region linking the ABS and the RIM. This is also in accordance with the kinetic analysis revealing no effects of proline isomerization in the interaction of the isolated ABS. Thus, our characterization of the AtMed25-ACID:DREB2A complex demonstrates that proline isomerization contributes to defining the mediator:TF interactome by providing the characteristic heterogeneity. Proline isomerization is an emerging property in ID-based protein-protein interactions^36,45. Thus, proline isomerization has recently been shown to enable differential binding of transcription factors to the NCBD αα-hub domain of CBP^36,46 and to modulate the liquid-liquid phase separation of IDPs⁴⁷. Proline isomerization can also function as a molecular timer controlling the amplitude and duration of cellular processes⁴⁵. Molecular timing appears highly relevant for DREB2A in regulation of stress-responses through its molecular switch region binding transcriptional activators and repressors. Whether the DREB2A proline is a substrate for one of the stress-associated Arabidopsis cyclophilins cis-trans isomerases⁴⁸ remains an open question.

Interactions in structurally heterogeneous complexes may have a high degree of energetic frustration^49,50, which is also likely for the bivalent interaction of DREB2A with Med25-ACID (Fig. 7A). Bivalency allows high-affinity binding and thereby specificity in interactions. However, for the DREB2A:Med25-ACID interaction, the weak secondary binding site and energetic frustration, particularly of the trans bound state, may allow other co-regulators, such as TAF4⁴² and RCD1²⁰, to bind the RIM and either displace DREB2A from Med25 or participate in ternary complexes (Fig. 7A). ID-based molecular competition has been demonstrated for the HIF1α and CITED2 transcription factors in their binding to the TAZ1 domain of CBP as part of the hypoxic response⁵¹. In summary, the results suggests mechanisms for the complex interplay between co-regulators and transcription factors, highlighting the importance of SLiM context, proline isomerization, and alternative splicing in molecular switching enabling fine-tuning of transcriptional regulation.

Methods

Protein expression and purification

All protein fragments were produced with an N-terminal Glutathione-S-transferase (GST) tag which was removed using Tobacco Etch Virus (TEV) protease and expressed from E.coli BL21 DE3 cells. DREB2A (AT5G05410) variants were expressed at 37 °C for three hours. The DREB2A splice variant (AT5G05410.2) was based on GenBank ID CD530293²¹. Due to stability and aggregation considerations, Med25-ACID (AT1G25540) was expressed with additional N-terminal residues⁵². Expression of Med25-ACID_532-680 was induced at high OD (0.8-1.0) and cultured overnight at 18 °C. Cells were resuspended in 20 mM Tris-HCl pH 8.0, 100 mM NaCl lysis buffer and sonicated. The lysate was incubated with glutathione Sepharose for one hour and washed using the lysis buffer. Aliquots of 5 mL resin were resuspended in 20 mL of lysis buffer with 5 mM dithiothreitol (DTT) and 1 mM EDTA and the protein was cleaved overnight at 4 °C using TEV protease. For DREB2A, cleaved protein was freeze-dried and buffer-changed to experimental buffer using size exclusion chromatography. Med25-ACID was further purified using cation exchange (Source 15 S) at pH 8.5 and the buffer was changed using size exclusion chromatography (Superdex S75 10/300 GL). Isotope-labeled proteins were produced by growing the initial culture in lysogenic broth and transferring to half volume M9 minimal media 30 min before induction.

Isothermal titration calorimetry

ITC was performed on either a MicroCal ITC200 or a Malvern Pananalytical PEAQ-ITC. All ITC experiments were performed using a 50 mM HEPES pH 7.4, 100 mM NaCl and 0.5 mM tris(2-carboxyethyl)phosphine (TCEP) buffer. All experiments were performed using a ~ 1:10 cell to syringe protein concentration ratio. For high-affinity DREB2A fragments, the cell concentration was around 15 µM, while lower-affinity interaction experiments used higher concentrations. Samples were degassed by centrifuging at 20,000 x g for 20 minutes at experimental temperature. For three DREB2A fragments, experiments were performed over a range of temperatures to determine the ΔC_p. Reported thermodynamic parameters are given at 25 °C. Data were analyzed using Origin and inhouse scripts using the one-set-of-sites model as documented by MicroCal and Origin. Dilution heats were fitted using an offset parameter. Most experiments were performed using a 0.5 µl initial injection, which was discarded in the analysis, followed by 18 injections of 2.0 µl.

High-throughput screening for ADs in Arabidopsis TFs

The DREB2A protein sequence was broken up into forty amino acid tiles with a step size of ten amino acids. Yeast codon optimized sequences for each DREB2A tile were synthesized and cloned in bulk into pMVS1421²⁸ to create a synthetic TF consisting of an N-terminal mCherry tag, a mouse Zif269 DNA-binding domain, an estrogen binding domain, and the tested DREB2A tile expressed under the yeast ACT1 promoter. The expression cassette was cloned using homologous recombination into the mating type A strain DHY211 (courtesy of Angela Chu and Joe Horecka) at the URA3 locus as previously described²⁸. Positive strains carrying the synthetic transcription factor were mated to yeast carrying a GFP reporter driven under the P3 promoter as previously described²⁸. Yeast were sorted based on reporter to transcription factor ratio (GFP:mCherry), and each bin was sequenced to determine the abundance of each fragment tested. Reads were aligned and quantified using BWA aligner and SAMtools. Fragments counts were analyzed using custom scripts in python to generate the AD score. AD score was calculated by taking the dot product of fragment abundance in each bin by the median GFP:mCherry ratio of each bin and then Z-score normalized, as previously described²⁸. Activity of synthetic control TFs with known ADs is shown in Supplementary Fig. 3.

Bioinformatics

AtDREB2A orthologous sequences were obtained from the PLAZA database⁵³ and aligned using CLUSTAL Omega^54,55. The MEME suite⁵⁶ was then used to define a motif, which described the binding region. ID-profiles were predicted using DISOPRED⁵⁷, and α-helix propensity was analyzed using Agadir⁵⁸.

Small-angle X-ray scattering

SAXS data were acquired on the PETRA III P12 beamline using in-line size exclusion chromatography. A single sample of 5.8 mg/mL Med25-ACID_532-680 was prepared in 20 mM Na₂HPO₄/NaH₂PO₄ pH 6.5, 100 mM NaCl and 1 mM TCEP. Scattering data was collected from the single peak on the elution profile. Buffer reference scattering data was collected from areas surrounding the protein elution peak. Scattering data were recorded and processed by staff at the facility in Hamburg and the data were analyzed using the ensemble optimization method⁵⁹. Scattering data were analyzed using the ensemble optimization method (EOM)⁶⁰ with a structure pool extracted from a 1.7 µs molecular dynamics simulation of the AF2 Med25-ACID_540-680 structure. Missing N-terminal residues were generated using the EOM RANCH algorithm.

Molecular dynamics simulation

Molecular dynamics simulations were performed using GROMACS⁶¹ using single-precision floating point calculations. The MD simulations were initiated from the equilibrated AF2 model of Med25-ACID_540-680 using the AMBER99SB forcefield⁶² and TIP3P water model⁶³. Overall system charge was naturalized by adding Na⁺ and Cl^- ions with additional ions added for a final concentration of 100 mM NaCl.

NMR spectroscopy

NMR data were acquired on Bruker AVANCE 600, 750 or 800 MHz (¹H) spectrometers equipped with cryogenic probes. Free induction decays were transformed and visualized using NMRPipe⁶⁴ and analyzed using the CcpNmr Analysis software⁶⁵. All NMR samples were prepared using 10% (v/v) D₂O, 0.02% (w/v) NaN₃, 0.2 mM 4,4-dimethyl-4-silapen-tane-1-sulfonic acid (DSS), 20 mM Na₂HPO₄/NaH₂PO₄, and 100 mM NaCl. All non-CEST samples were prepared at pH 6.5. Backbone assignment of Med25-ACID_532-680 in the unbound state was done manually from analysis of ¹⁵N-HSQC, HNCACB, HNCOCACB, HNCO and HNCACO experiments (BMRB ID: 52040). Assignment of the DREB2A_234-256 bound state (BMRB ID: 52042) was done using BEST⁶⁶ TROSY (BT)-HSQC, BT-HNCA and ¹³C^β optimized BT-HNCACB experiments⁶⁷. All Med25-ACID assignment samples were recorded using 500 µM ¹³C,¹⁵N-labeled Med25-ACID_532-680 and, if relevant, 2 mM unlabeled DREB2A_234-256, Assignment of the unbound DREB2A_234-276 fragment was acquired from previous work (BMRB ID: 51055)²². SCSs were calculated using the SBiNLab⁶⁸ and PONTENCI⁶⁹ web tools and are reported as mean, with error bars representing the standard deviation. {¹H}-¹⁵N hetNOEs were recorded in triplicates, processed individually and reported as mean and standard deviation. Unbound DREB2A_234-276 cis- and trans-proline populations were quantified from HSQC spectra recorded with a recycle delay of 6 seconds by fitting peak volumes using the PINT software⁷⁰. Five non-overlapping peaks, visible in both isomeric states, were quantified and the populations were estimated as the volume of a given peak divided by sum of the cis and trans peak volumes.

NMR titration analyses

All NMR titration data were acquired on a Bruker AVANCE III HD 600 spectrometer using between 100 and 150 µM ¹⁵N-labeled Med25-ACID_532-680 (BMRB ID: 52041, 52042, 52043). The binding was monitored with ¹⁵N-HSQC spectra using constant Med25-ACID concentrations with increasing concentration of DREB2A ligand. All titration series included a spectrum with no ligand, which was used as the unbound state when calculating CSPs. ¹⁵N,H^N CSPs were calculated as the Euclidian distance using a 0.154 weight for the ¹⁵N chemical shift⁷¹.

Chemical exchange saturation transfer experiments

All CEST NMR data was acquired on a Bruker AVANCE Neo 800 spectrometer with a 5 mm CPTXO Cryoprobe. CEST samples were prepared at pH 6.1 using 500 µM ¹³C,¹⁵N -labeled DREB2A_234-276, or 1 mM DREB2A_{234-276,P245A}, with, if relevant, approximately 5% molar ratio of unlabeled Med25-ACID_532-680. ¹⁵N-CEST data was acquired using previously published pulse sequences³⁴ using three different B₁ field strengths. ¹⁵N-CEST data acquired at 25 °C was done using 25, 12.5, and 6.25 Hz B₁ fields while experiments at 10 °C were done using 40, 20, and 10 Hz B₁ fields. ¹³C-CEST data was acquired using pulse sequences^38,39 provided by prof. Lewis Kay with a B₁ field of 30 Hz. For all CEST data, free induction decays were transformed and visualized using NMRPipe⁶⁴ and peak intensities, I, were quantified as the height at the specific peak position. Overlapping peaks and peaks showing no exchange were discarded in global fitting procedures, but approximate CSP were extracted when possible. Analysis and fitting were done using ChemEx⁷². Cis- and trans-proline peaks were analyzed separately using a two-state model and C-terminal residues, which displayed three-state behavior, were analyzed using a three-state model. Peaks that could not be grouped were analyzed individually to determine CSPs. ¹⁵N-CEST global fitting groups are shown in Supplementary Fig. 10.

AF2 complex modeling

AF2 modeling of the Med25-ACID:DREB2A complex was done using the ColabFold webserver³³ (accessed June 2022) by providing the sequences of isolated Med25-ACID domain and the DREB2A_234-276 fragment using the heterodimer syntax. The server reports five models for each prediction. Two predictions were performed, from which three different configurations were observed (Fig. 4A), although the DREB2A chain was generally assigned a low confidence (Supplementary Fig. 4D). Models were filtered based on binding site data and mutational analyses resulting in one model with satisfaction of experimental observations.

Stopped-flow kinetics

Stopped-flow fluorescence experiments were performed using a sequential SX20 stopped-flow spectrometer (Applied Photophysics). DREB2A_234-276 was labeled using a fluorescein isothiocyanate (FITC) labeling kit (Pierce^TM) and purified with reversed phase chromatography (Zorbax 300SB C18, 70% ACN 0.1% TFA) to obtain single-labeled FITC-DREB2A. FITC-DREB2A_234-256 synthetic peptide was purchased from TAG Copenhagen A/S. Samples were prepared in 50 mM HEPES buffer, pH 7.4, 100 mM NaCl, 1 mM DTT. FITC was excited at 490 nm and fluorescence detected at 519 nm, with a long pass cutoff filter at 515 nm. All samples were filtered through a 0.22 μm PVDF syringe filter prior to experiments and FITC samples were handled with aluminum foil to limit fluorophore damage. At least 20 fluorescence traces were obtained for each condition. Association kinetics were conducted under pseudo-first-order conditions by mixing of equal volumes of FITC-DREB2A (final concentrations of 50 and 250 nM for DREB2A_234-276 and DREB2A_234-256, respectively) and increasing concentrations of Med25-ACID_532-680. Dissociation rate constants were determined using competitive displacement experiments, mixing preformed Med25-ACID_532-680:FITC-DREB2A complex (1:0.1 µM for Med25-ACID:FITC-DREB2A_234-276 and 4:0.1 µM for Med25-ACID:FITC-DREB2A_234-256) with increasing concentrations of unlabeled DREB2A. Association traces of both DREB2A fragments were fitted using a single exponential function. Dissociation traces of DREB2A_234-256 were fitted using a single exponential decay function, while DREB2A_234-276 traces were fitted using a double exponential function. Traces were fitted individually and observed rate constants and standard deviations were obtained by fitting a cumulative Gaussian function to the distribution of rates. k_on was obtained as the slope of the linear fit of the k_obs concentration dependence. k_off was determined from the exponential decay asymptote of the concentration dependence of the observed dissociation rate(s).

Effective concentration calculation

The bound fraction of C-terminal motif of the bivalent DREB2A fragments was calculated using an empirical correlation between disordered linker length and effective concentration³². Extrapolating to the DREB2A motif linker length of 12.5 ± 2.5 residues resulted in an effective concentration of 9 ± 3 mM. Given the affinity of 4.7 ± 1.8 mM extracted from NMR titrations we calculated an expected saturation level of 65 ± 10%.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

The bioinformatic, ITC, SAXS, stopped-flow and AF2 complex model data generated in this study have been deposited on GitHub [https://doi.org/10.5281/zenodo.10409674]⁷³. Processed NMR and MD data are also available on GitHub. The NMR chemical shift assignments for Med25-ACID in the free and DREB2A bound states have been deposited in the BioMagResBank (BMRB) under the following accession codes and DOIs: free Med25-ACID (52040 [https://doi.org/10.13018/BMR52040]), Med25-ACID with DREB2A ABS-RIM (52041 [https://doi.org/10.13018/BMR52041]), Med25-ACID with DREB2A ABS (52042 [https://doi.org/10.13018/BMR52042]), Med25-ACID with DREB2A RIM (52043 [https://doi.org/10.13018/BMR52043]). Source data for main and Supplementary items are provided with this paper. Source data are provided with this paper.

References

Ma, J. & Ptashne, M. A new class of yeast transcriptional activators. Cell 51, 113–119 (1987).
Article CAS PubMed Google Scholar
Plaschka, C., Nozawa, K. & Cramer, P. Mediator architecture and RNA polymerase II interaction. J. Mol. Biol. 428, 2569–2574 (2016).
Article CAS PubMed Google Scholar
Reeves, W. M. & Hahn, S. Targets of the Gal4 transcription activator in functional transcription complexes. Mol. Cell. Biol. 25, 9092–9102 (2005).
Article CAS PubMed PubMed Central Google Scholar
Sigler, P. B. Acid blobs and negative noodles. Nature 333, 210–212 (1988).
Article ADS CAS PubMed Google Scholar
Staby, L. et al. Eukaryotic transcription factors: paradigms of protein intrinsic disorder. Biochem. J. 474, 2509–2532 (2017).
Article CAS PubMed Google Scholar
Warfield, L., Tuttle, L. M., Pacheco, D., Klevit, R. E. & Hahn, S. A sequence-specific transcription activator motif and powerful synthetic variants that bind Mediator using a fuzzy protein interface. Proc. Natl Acad. Sci. USA 111, E3506–E3513 (2014).
Article ADS CAS PubMed PubMed Central Google Scholar
Ravarani, C. N. et al. High‐throughput discovery of functional disordered regions: investigation of transactivation domains. Mol. Syst. Biol. 14, e8190 (2018).
Article PubMed PubMed Central Google Scholar
Currie, S. L. et al. ETV4 and AP1 transcription factors form multivalent interactions with three sites on the MED25 activator-interacting domain. J. Mol. Biol. 429, 2975–2995 (2017).
Article CAS PubMed PubMed Central Google Scholar
Milbradt, A. G. et al. Structure of the VP16 transactivator target in the Mediator. Nat. Struct. Mol. Biol. 18, 410–416 (2011).
Article CAS PubMed PubMed Central Google Scholar
Vojnic, E. et al. Structure and VP16 binding of the Mediator Med25 activator interaction domain. Nat. Struct. Mol. Biol. 18, 404–410 (2011).
Article CAS PubMed Google Scholar
Lee, M. S., Lim, K., Lee, M. K. & Chi, S. W. Structural basis for the interaction between p53 transactivation domain and the mediator subunit MED25. Molecules 23, 2726 (2018).
Article PubMed PubMed Central Google Scholar
Landrieu, I. et al. Characterization of ERM transactivation domain binding to the ACID/PTOV domain of the Mediator subunit MED25. Nucleic Acids Res. 43, 7110–7121 (2015).
Article CAS PubMed PubMed Central Google Scholar
Kazan, K. The multitalented MEDIATOR25. Front. Plant Sci. 8, 1–11 (2017).
Article Google Scholar
Elfving, N. et al. The Arabidopsis thaliana Med25 mediator subunit integrates environmental cues to control plant development. Proc. Natl Acad. Sci. USA 108, 8245–8250 (2011).
Article ADS CAS PubMed PubMed Central Google Scholar
Sakuma, Y. et al. Dual function of an Arabidopsis transcription factor DREB2A in water-stress-responsive and heat-stress-responsive gene expression. Proc. Natl Acad. Sci. USA 103, 18822–18827 (2006).
Article ADS CAS PubMed PubMed Central Google Scholar
Bugge, K. et al. Structure of radical-induced Cell Death1 hub domain reveals a common αα-Scaffold for disorder in transcriptional networks. Structure 26, 734–746.e7 (2018).
Article CAS PubMed Google Scholar
Bugge, K. et al. αα-Hub domains and intrinsically disordered proteins: a decisive combo. J. Biol. Chem. 296, 100226 (2021).
Article CAS PubMed Google Scholar
Shapiguzov, A. et al. Arabidopsis RCD1 coordinates chloroplast and mitochondrial functions through interaction with ANAC transcription factors. Elife 8, 1–35 (2019).
Article Google Scholar
Jaspers, P. et al. Unequally redundant RCD1 and SRO1 mediate stress and developmental responses and interact with transcription factors. Plant J. 60, 268–279 (2009).
Article CAS PubMed Google Scholar
O’Shea, C. et al. Structures and short linear motif of disordered transcription factor regions provide clues to the interactome of the cellular hub protein radical-induced cell death1. J. Biol. Chem. 292, 512–527 (2017).
Article PubMed Google Scholar
Vainonen, J. P. et al. RCD1-DREB2A interaction in leaf senescence and stress responses in Arabidopsis thaliana. Biochem. J. 442, 573–581 (2012).
Article CAS PubMed Google Scholar
Theisen, F. F. et al. Quantification of conformational entropy unravels effect of disordered flanking region in coupled folding and binding. J. Am. Chem. Soc. 143, 14540–14550 (2021).
Article CAS PubMed Google Scholar
Aguilar, X. et al. Interaction studies of the human and Arabidopsis thaliana Med25-ACID proteins with the herpes simplex virus VP16- and plant-specific Dreb2a transcription factors. PLoS ONE 9, e98575 (2014).
Article ADS PubMed PubMed Central Google Scholar
Blomberg, J. et al. Interactions between DNA, transcriptional regulator Dreb2a and the Med25 mediator subunit from Arabidopsis thaliana involve conformational changes. Nucleic Acids Res. 40, 5938–5950 (2012).
Article CAS PubMed PubMed Central Google Scholar
Sakuma, Y. et al. Functional analysis of an arabidopsis transcription factor, DREB2A, involved in drought-responsive gene expression. Plant Cell 18, 1292–1309 (2006).
Article CAS PubMed PubMed Central Google Scholar
Christensen, L. F. et al. Evolutionary conservation of the intrinsic disorder-based Radical-Induced Cell Death1 hub interactome. Sci. Rep. 9, 18927 (2019).
Article ADS CAS PubMed PubMed Central Google Scholar
Staller, M. V. et al. Directed mutational scanning reveals a balance between acidic and hydrophobic residues in strong human activation domains. Cell Syst. 13, 334–345.e5 (2022).
Article CAS PubMed PubMed Central Google Scholar
Staller, M. V. et al. A high-throughput mutational scan of an intrinsically disordered acidic transcriptional activation domain. Cell Syst. 6, 444–455.e6 (2018).
Article CAS PubMed PubMed Central Google Scholar
Jumper, J. et al. Highly accurate protein structure prediction with AlphaFold. Nature 596, 583–589 (2021).
Article ADS CAS PubMed PubMed Central Google Scholar
Varadi, M. et al. AlphaFold Protein Structure Database: massively expanding the structural coverage of protein-sequence space with high-accuracy models. Nucleic Acids Res. 50, D439–D444 (2022).
Article CAS PubMed Google Scholar
Shen, Y. & Bax, A. Identification of helix capping and β-turn motifs from NMR chemical shifts. J. Biomol. NMR 52, 211–232 (2012).
Article CAS PubMed PubMed Central Google Scholar
Sørensen, C. S. & Kjaergaard, M. Effective concentrations enforced by intrinsically disordered linkers are governed by polymer physics. Proc. Natl Acad. Sci. USA 116, 23124–23131 (2019).
Article ADS PubMed PubMed Central Google Scholar
Mirdita, M. et al. ColabFold: making protein folding accessible to all. Nat. Methods 19, 679–682 (2022).
Article CAS PubMed PubMed Central Google Scholar
Vallurupalli, P., Bouvignies, G. & Kay, L. E. Studying ‘invisible’ excited protein states in slow exchange with a major state conformation. J. Am. Chem. Soc. 134, 8148–8161 (2012).
Article CAS PubMed Google Scholar
Delaforge, E. et al. Investigating the role of large-scale domain dynamics in protein-protein interactions. Front. Mol. Biosci. 3, 54 (2016).
Article PubMed PubMed Central Google Scholar
Zosel, F., Mercadante, D., Nettels, D. & Schuler, B. A proline switch explains kinetic heterogeneity in a coupled folding and binding reaction. Nat. Commun. 9, 3332 (2018).
Article ADS PubMed PubMed Central Google Scholar
Henderson, A. R. et al. Conservation of coactivator engagement mechanism enables small-molecule allosteric modulators. Proc. Natl Acad. Sci. USA 115, 8960–8965 (2018).
Article ADS CAS PubMed PubMed Central Google Scholar
Vallurupalli, P. & Kay, L. E. Probing slow chemical exchange at carbonyl sites in proteins by chemical exchange saturation transfer NMR spectroscopy. Angew. Chem. Int. Ed. 52, 4156–4159 (2013).
Article CAS Google Scholar
Long, D., Sekhar, A. & Kay, L. E. Triple resonance-based 13Cα and 13Cβ CEST experiments for studies of ms timescale dynamics in proteins. J. Biomol. NMR 60, 203–208 (2014).
Article CAS PubMed Google Scholar
Camilloni, C., de Simone, A., Vranken, W. F. & Vendruscolo, M. Determination of secondary structure populations in disordered states of proteins using nuclear magnetic resonance chemical shifts. Biochemistry 51, 2224–2231 (2012).
Article CAS PubMed Google Scholar
Spolar, R. S. & Record, M. T. Coupling of local folding to site-specific binding of proteins to DNA. Science (1979) 263, 777–784 (1994).
CAS Google Scholar
Theisen, F. F. et al. αα-hub coregulator structure and flexibility determine transcription factor binding and selection in regulatory interactomes. J. Biol. Chem. 298, 101963 (2022).
Article Google Scholar
Tuttle, L. M. et al. Mediator subunit Med15 dictates the conserved “fuzzy” binding mechanism of yeast transcription activators Gal4 and Gcn4. Nat. Commun. 12, 2220 (2021).
Article ADS CAS PubMed PubMed Central Google Scholar
Henley, M. J. et al. Unexpected specificity within dynamic transcriptional protein–protein complexes. Proc. Natl Acad. Sci. USA 117, 27346–37353 (2020).
Article ADS CAS PubMed PubMed Central Google Scholar
Gustafson, C. L. et al. A slow conformational switch in the BMAL1 transactivation domain modulates circadian rhythms. Mol. Cell 66, 447–457.e7 (2017).
Article CAS PubMed PubMed Central Google Scholar
Buholzer, K. J. et al. Multilayered allosteric modulation of coupled folding and binding by phosphorylation, peptidyl-prolyl cis/trans isomerization, and diversity of interaction partners. J. Chem. Phys. 157, 235102 (2022).
Article ADS CAS PubMed Google Scholar
Babu, M., Favretto, F., Rankovic, M. & Zweckstetter, M. Peptidyl prolyl isomerase A modulates the liquid-liquid phase separation of proline-rich IDPs. J. Am. Chem. Soc. 144, 16157–16163 (2022).
Article CAS PubMed PubMed Central Google Scholar
Singh, H., Kaur, K., Singh, M., Kaur, G. & Singh, P. Plant cyclophilins: multifaceted proteins with versatile roles. Front. Plant Sci. 11, 585212 (2020).
Article PubMed PubMed Central Google Scholar
Freiberger, M. I., Wolynes, P. G., Ferreiro, D. U. & Fuxreiter, M. Frustration in fuzzy protein complexes leads to interaction versatility. J. Phys. Chem. B 125, 2513–2520 (2021).
Article CAS PubMed PubMed Central Google Scholar
Gianni, S. et al. Fuzziness and frustration in the energy landscape of protein folding, function, and assembly. Acc. Chem. Res. 54, 1251–1259 (2021).
Article CAS PubMed PubMed Central Google Scholar
Berlow, R. B., Dyson, H. J. & Wright, P. E. Hypersensitive termination of the hypoxic response by a disordered protein switch. Nature 543, 447–451 (2017).
Article ADS CAS PubMed PubMed Central Google Scholar
Zhang, F. et al. Structural basis of JAZ repression of MYC transcription factors in jasmonate signalling. Nature 525, 269–273 (2015).
Article ADS CAS PubMed PubMed Central Google Scholar
Van Bel, M. et al. PLAZA 5.0: extending the scope and power of comparative and functional genomics in plants. Nucleic Acids Res. 50, D1468–D1474 (2022).
Article PubMed Google Scholar
Madeira, F. et al. Search and sequence analysis tools services from EMBL-EBI in 2022. Nucleic Acids Res. 50, W276–W279 (2022).
Article CAS PubMed PubMed Central Google Scholar
Sievers, F. et al. Fast, scalable generation of high-quality protein multiple sequence alignments using Clustal Omega. Mol. Syst. Biol. 7, 539 (2011).
Article PubMed PubMed Central Google Scholar
Bailey, T. L., Johnson, J., Grant, C. E. & Noble, W. S. The MEME suite. Nucleic Acids Res. 43, W39–W49 (2015).
Article CAS PubMed PubMed Central Google Scholar
Ward, J. J., McGuffin, L. J., Bryson, K., Buxton, B. F. & Jones, D. T. The DISOPRED server for the prediction of protein disorder. Bioinformatics 20, 2138–2139 (2004).
Article CAS PubMed Google Scholar
Muñoz, V. & Serrano, L. Elucidating the folding problem of helical peptides using empirical parameters. Nat. Struct. Biol. 1, 399–409 (1994).
Article PubMed Google Scholar
Tria, G., Mertens, H. D. T., Kachala, M. & Svergun, D. I. Advanced ensemble modelling of flexible macromolecules using X-ray solution scattering. IUCrJ 2, 207–217 (2015).
Article CAS PubMed PubMed Central Google Scholar
Franke, D. et al. ATSAS 2.8: a comprehensive data analysis suite for small-angle scattering from macromolecular solutions. J. Appl. Crystallogr. 50, 1212–1225 (2017).
Article ADS CAS PubMed PubMed Central Google Scholar
Berendsen, H. J. C., van der Spoel, D. & van Drunen, R. GROMACS: a message-passing parallel molecular dynamics implementation. Comput. Phys. Commun. 91, 43–56 (1995).
Article ADS CAS Google Scholar
Showalter, S. A. & Brüschweiler, R. Validation of molecular dynamics simulations of biomolecules using NMR spin relaxation as benchmarks: application to the AMBER99SB force field. J. Chem. Theory Comput. 3, 961–975 (2007).
Article CAS PubMed Google Scholar
Jorgensen, W. L., Chandrasekhar, J., Madura, J. D., Impey, R. W. & Klein, M. L. Comparison of simple potential functions for simulating liquid water. J. Chem. Phys. 79, 926–935 (1983).
Article ADS CAS Google Scholar
Delaglio, F. et al. NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J. Biomol. NMR 6, 277–293 (1995).
Article CAS PubMed Google Scholar
Vranken, W. F. et al. The CCPN data model for NMR spectroscopy: development of a software pipeline. Proteins: Struct. Funct. Bioinf. 59, 687–696 (2005).
Article CAS Google Scholar
Schanda, P., van Melckebeke, H. & Brutscher, B. Speeding up three-dimensional protein NMR experiments to a few minutes. J. Am. Chem. Soc. 128, 9042–9043 (2006).
Article CAS PubMed Google Scholar
Favier, A. & Brutscher, B. NMRlib: user-friendly pulse sequence tools for Bruker NMR spectrometers. J. Biomol. NMR 73, 199–211 (2019).
Article CAS PubMed Google Scholar
Kjaergaard, M. & Poulsen, F. M. Sequence correction of random coil chemical shifts: correlation between neighbor correction factors and changes in the Ramachandran distribution. J. Biomol. NMR 50, 157–165 (2011).
Article CAS PubMed Google Scholar
Nielsen, J. T. & Mulder, F. A. A. POTENCI: prediction of temperature, neighbor and pH-corrected chemical shifts for intrinsically disordered proteins. J. Biomol. NMR 70, 141–165 (2018).
Article CAS PubMed Google Scholar
Niklasson, M. et al. Comprehensive analysis of NMR data using advanced line shape fitting. J. Biomol. NMR 69, 93–99 (2017).
Article CAS PubMed PubMed Central Google Scholar
Mulder, F. A., Schipper, D., Bott, R. & Boelens, R. Altered flexibility in the substrate-binding site of related native and engineered high-alkaline Bacillus subtilisins. J. Mol. Biol. 292, 111–123 (1999).
Article CAS PubMed Google Scholar
Bouvignies, G. ChemEx: NMR Chemical Exchange Analysis Tool. www.github.com/gbouvignies/ChemEx (2022).
Theisen, F. F. et al. Data sets for ‘Molecular switching in transcription through splicing and proline-isomerization regulates stress responses in plants’. GitHub https://doi.org/10.5281/zenodo.10409674 (2023).

Download references

Acknowledgements

This work was supported by the Novo Nordisk Foundation challenge grant REPIN, rethinking protein interactions (grant no.: NNF18OC0033926; to B.B.K. and K.S.) and by the Danish Research Council (grant no.: 9040-00164B to B.B.K. and K.S.). Activation domain assay was done under the National Science Foundation (NSF) (PGRP BIO-2112056 to L.C.S.; Postdoctoral Research Program IOS-1907098 to N.M.). NMR spectra were recorded at cOpenNMR, an infrastructure supported by the Novo Nordisk Foundation (grant no.: NNF18OC0032996). The authors would like to thank Dr. Anders Lønstrup Hansen for many important discussions regarding analysis and representation of data. We also thank Prof. Lewis E. Kay for sharing the ¹³C-CEST pulse sequences necessary for this work, Dr. Elise Delaforge for discussions and suggestions regarding analysis of CEST data and Stase Bielskute for suggesting the δ2D tool for analysis of IDP secondary structure. We wish to thank Cy Jeffries at the EMBL Hamburg beamline for recording, processing and providing initial analysis of SAXS data. We thank Profs. Ben Schuler and Per Jemth for valuable discussions of kinetics data leading to further exploration.

Author information

Authors and Affiliations

The REPIN and The Linderstrøm-Lang Centre for Protein Science, Department of Biology, University of Copenhagen, Copenhagen, Denmark
Frederik Friis Theisen, Steffie Elkjær, Yannick H. A. Leurs, Charlotte O’Shea, Birthe B. Kragelund & Karen Skriver
Structural Biology and NMR Laboratory, Department of Biology, University of Copenhagen, Copenhagen, Denmark
Frederik Friis Theisen, Andreas Prestel, Kaare Teilum & Birthe B. Kragelund
Department of Biology, Duke University, Durham, NC, USA
Nicholas Morffy & Lucia C. Strader

Authors

Frederik Friis Theisen
View author publications
You can also search for this author in PubMed Google Scholar
Andreas Prestel
View author publications
You can also search for this author in PubMed Google Scholar
Steffie Elkjær
View author publications
You can also search for this author in PubMed Google Scholar
Yannick H. A. Leurs
View author publications
You can also search for this author in PubMed Google Scholar
Nicholas Morffy
View author publications
You can also search for this author in PubMed Google Scholar
Lucia C. Strader
View author publications
You can also search for this author in PubMed Google Scholar
Charlotte O’Shea
View author publications
You can also search for this author in PubMed Google Scholar
Kaare Teilum
View author publications
You can also search for this author in PubMed Google Scholar
Birthe B. Kragelund
View author publications
You can also search for this author in PubMed Google Scholar
Karen Skriver
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

F.F.T., A.P., S.E., Y.H.A.L., N.M., L.C.S, C.O.S, K.T., B.B.K., and K.S. designed the experiments. F.F.T., S.E., Y.H.A.L., N.M., and C.O.S. performed the experiments, F.F.T., A.P., S.E., Y.H.A.L., N.M., B.B.K., and K.S. analyzed the data. F.F.T., B.B.K., and K.S. wrote the manuscript. B.B.K. and K.S. supervised the study.

Corresponding authors

Correspondence to Birthe B. Kragelund or Karen Skriver.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Simon Currie, Stefan Björklund and the other anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Peer Review File

Reporting Summary

Source data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Theisen, F.F., Prestel, A., Elkjær, S. et al. Molecular switching in transcription through splicing and proline-isomerization regulates stress responses in plants. Nat Commun 15, 592 (2024). https://doi.org/10.1038/s41467-024-44859-2

Download citation

Received: 26 June 2023
Accepted: 09 January 2024
Published: 18 January 2024
DOI: https://doi.org/10.1038/s41467-024-44859-2

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.

Subjects

Abstract

Similar content being viewed by others

DHH1/DDX6-like RNA helicases maintain ephemeral half-lives of stress-response mRNAs

Poly(ADP-ribose)-binding protein RCD1 is a plant PARylation reader regulated by Photoregulatory Protein Kinases

Overlapping roles of spliceosomal components SF3B1 and PHF5A in rice splicing regulation

Introduction

Results

The Med25-ACID-binding region of DREB2A contains two SLiMs

AtMed25-ACID forms a β-barrel with two accessory helices and dynamic regions

DREB2A binds a groove in AtMed25-ACID formed by H2 and L1

AF2 modeling supports bivalent binding

Multiple bound states of DREB2A uncovered using CEST

DREB2A undergoes coupled folding and binding forming two short α-helices

Discussion

Methods

Protein expression and purification

Isothermal titration calorimetry

High-throughput screening for ADs in Arabidopsis TFs

Bioinformatics

Small-angle X-ray scattering

Molecular dynamics simulation

NMR spectroscopy

NMR titration analyses

Chemical exchange saturation transfer experiments

AF2 complex modeling

Stopped-flow kinetics

Effective concentration calculation

Reporting summary

Data availability

References

Acknowledgements

Author information

Authors and Affiliations

Contributions

Corresponding authors

Ethics declarations

Competing interests

Peer review

Peer review information

Additional information

Supplementary information

Supplementary Information

Peer Review File

Reporting Summary

Source data

Source data

Rights and permissions

About this article

Cite this article

Share this article

Comments

Search

Quick links