Co-regulation of the transcription controlling ATF2 phosphoswitch by JNK and p38

Kirsch, Klára; Zeke, András; Tőke, Orsolya; Sok, Péter; Sethi, Ashish; Sebő, Anna; Kumar, Ganesan Senthil; Egri, Péter; Póti, Ádám L.; Gooley, Paul; Peti, Wolfgang; Bento, Isabel; Alexa, Anita; Reményi, Attila

doi:10.1038/s41467-020-19582-3

Download PDF

Article
Open access
Published: 13 November 2020

Co-regulation of the transcription controlling ATF2 phosphoswitch by JNK and p38

Nature Communications volume 11, Article number: 5769 (2020) Cite this article

5045 Accesses
28 Citations
3 Altmetric
Metrics details

Subjects

Abstract

Transcription factor phosphorylation at specific sites often activates gene expression, but how environmental cues quantitatively control transcription is not well-understood. Activating protein 1 transcription factors are phosphorylated by mitogen-activated protein kinases (MAPK) in their transactivation domains (TAD) at so-called phosphoswitches, which are a hallmark in response to growth factors, cytokines or stress. We show that the ATF2 TAD is controlled by functionally distinct signaling pathways (JNK and p38) through structurally different MAPK binding sites. Moreover, JNK mediated phosphorylation at an evolutionarily more recent site diminishes p38 binding and made the phosphoswitch differently sensitive to JNK and p38 in vertebrates. Structures of MAPK-TAD complexes and mechanistic modeling of ATF2 TAD phosphorylation in cells suggest that kinase binding motifs and phosphorylation sites line up to maximize MAPK based co-regulation. This study shows how the activity of an ancient transcription controlling phosphoswitch became dependent on the relative flux of upstream signals.

The JAK/STAT signaling pathway: from bench to clinic

Article Open access 26 November 2021

Xiaoyi Hu, Jing li, … Wei Wang

Transcription bodies regulate gene expression by sequestering CDK9

Article Open access 08 April 2024

Martino Ugolini, Maciej A. Kerlin, … Nadine L. Vastenhouw

N6-methyladenine in DNA antagonizes SATB1 in early development

Article 15 July 2020

Zheng Li, Shuai Zhao, … Andrew Z. Xiao

Introduction

Receptor activation by extracellular cues at the cell membrane and gene expression in the nucleus are linked through signaling pathways involving protein kinases. To reveal the molecular basis underlying the transcriptional response of cells to complex biological inputs, it is important to explore how protein kinases control transcription factors (TF). TFs bind to DNA sites using their DNA-binding domain (DBD) and to cofactors using their transcription activating domain (TAD). TADs are often unstructured and enable regulated binding to cofactors by using phosphoswitches: regions functionally affected by protein phosphorylation and evolved to respond to post-translational modification^1,2. TAD phosphorylation is a major mechanism through which environmental cues affect cell physiology³.

Mitogen-activated protein kinases (MAPK) control transcription by phosphorylating TADs. For example, ERK1/2 phosphorylates Elk1—an immediate-early (IE) transcription factor—on its TAD promoting its interaction with a specific subunit of the RNA II polymerase binding Mediator complex involved in transcription initiation^4,5. Similarly, activator protein 1 (AP-1) transcription factors, such as c-Fos, c-Jun or ATFs, contain MAPK regulated phosphosites in their TAD^6,7,8. Ser63/Thr73 from human c-Jun or Thr69/71 from ATF2 are known to be phosphorylated by MAPKs leading to upregulation of AP-1 transcriptional activity^9,10. The ATF2 TAD was analyzed by NMR spectroscopy, and apart from a small C2H2-type Zn-finger, it was found to be disordered¹¹. Phosphosites in AP-1 TFs are essential in making phosphorylation-dependent contacts with gene expression controlling cofactors/effectors (e.g., p300/CBP—a coactivator/adaptor protein with histone acetyltransferase activity—is an important ATF2 cofactor¹²), but how AP-1 TAD phosphorylation leads to increased transcription is not yet understood.

Activating transcription factor 2 (ATF2) is ubiquitously expressed in most cell types and regulates the transcription of various genes involved in a broad spectrum of cellular functions: cell growth, development and stress¹³. JNK and p38 were shown to control ATF2 function via the 69-TPTP-72 phosphoswitch^10,14. MAPKs phosphorylate their substrates on Ser/Thr-Pro target motifs (S/TP) but substrate phosphorylation requires binding to at least one of two distinct protein–protein interaction hot-spots on the MAPK: the D- or the F-groove¹⁵. The MAPK D-groove (also referred to as the D-motif recruitment site, DRS) binds to D(ocking)-motifs comprising a positively charged and a hydrophobic region interacting with the negatively charged common docking (CD) groove and the hydrophobic pockets, respectively¹⁶. The F-groove (also referred to as the F-motif recruitment site, FRS) binds F(xFP)-type motifs and is located at the opposite side of the kinase compared to the D-groove. F-peptides bind to the activated, double-phosphorylated ppMAPK with increased affinity^17,18. D-motifs and F-motifs are differently located compared to the MAPK phosphorylation target site (S/TP): 10-50 amino acids N-terminal or 10-20 amino acids C-terminal, respectively¹⁹.

MAPKs play central roles in many signaling pathways and all three major groups of them (ERK1/2, JNK, and p38) are ubiquitously expressed. Their activation depends on the cell-type and the nature of the input signal, however, most physiological stimuli activate more than one MAPK, which brings about systems level opportunities for more complex regulation. What is the mechanistic basis of this regulation operating at the MAPK-TF level? How do TADs contribute to quantitative gene expression depending on the strength of upstream signals?

In the present study, we explore the structural and mechanistic basis of ATF2 regulation by two MAPKs: JNK and p38. These two MAPKs have distinct physiological roles and their activity is often altered in human tumors and cancer cell lines²⁰. We started by mapping out the ATF2 TAD region that binds JNK. We find that a Zn-finger located next to a D-motif is required to mediate MAPK-specific binding. Next, we identify a region that binds the phosphorylated form of p38 selectively. MAPK-ATF2 TAD complexes are structurally elucidated by X-ray crystallography/NMR and monitored by protein–protein interaction assays in cells. These data are used to create a mechanistic model of cellular TAD phosphorylation that reveals how two distinct signaling pathways converge on the same transcription controlling phosphoswitch and enable ATF2 TAD mediated quantitative gene expression.

Results

JNK binds to ATF2 using a linear D-motif and a Zn-finger

The transactivating region of both c-Jun and ATF2 contains a MAPK binding D-motif. The structure of the JNK1-ATF2 D-motif peptide complex was earlier determined by X-ray crystallography²¹. In contrast to c-Jun, which binds JNK using a D-motif only, ATF proteins (e.g., ATF2, ATF7, and CREB5) contain a C2H2 Zn-finger which partially overlaps with the adjacent D-motif (Supplementary Fig. 1a–c). This proximal arrangement between a structured domain and the D-motif consensus sequence seems to be an evolutionarily conserved trait in ATFs, suggesting that JNK-ATF binding needs to be studied using protein constructs including both regions.

In order to determine the minimal binding region, we probed the full transactivation region (ATF2 residues 19–100) by GST pull-down assays using purified JNK1 (Fig. 1a). This showed that neither the Zn-finger (19–50) nor the D-motif (42–55) alone was sufficient for binding and that the minimal region (19–58) contained both the Zn-finger and the D-motif. This construct bound JNK with micromolar affinity (~5 µM) and the interaction required an intact Zn-finger, as a chelating agent (EDTA) diminished binding (Fig. 1b,c). The interaction also depended on the basic region of the D-motif, as a charge reversal mutation (K48E) also diminished binding (Fig. 1b). In contrast, the JNK-c-Jun interaction was not sensitive to the presence of EDTA, but a charge reversal mutation in ATF2 or in c-Jun (K48E or K35E, respectively) decreased binding similarly (~10 fold) (Fig. 1d). Isothermal titration calorimetry (ITC) also showed an ~10 fold decrease in JNK-ATF2 binding in the presence of EDTA, due to an increase in the entropic cost of binding as the enthalpic contribution stayed nearly the same (Fig. 1e). This suggests that an intact Zn-finger is necessary for binding by possibly constraining the flexibility of critical D-motif residues.

**Fig. 1: JNK-ATF2 binding depends on an intact Zn-finger + D-motif module.**

Next, we assessed the role of JNK binding on MAPK target site phosphorylation. The Zn-finger + D-motif module (19–58) was left intact or mutated (K48E) or replaced by c-Jun or its mutated version (K35E) and phosphorylation of the 69-TPTP-72 target site by activated JNK was monitored by anti-phospho immunoblots (Fig. 1f). This analysis confirmed the importance of key lysine residues in ATF2 Zn-finger + D-motif or c-Jun D-motif mediated phosphorylation.

The importance of the Zn-finger + D-motif module in JNK binding, full-length ATF2 phosphorylation, and ATF2 TAD mediated transcriptional activity was also examined in HEK293T cells (Supplementary Fig. 2). The K48E charge reversal mutation decreased full-length ATF2-JNK1 binding in a yellow fluorescent protein (YFP) fragment complementation experiment and Zn-finger disrupting mutations (C27A and C32A; CACA) had the same effect (Supplementary Fig. 2a). The YFP signal was mostly localized in the cytosol, but co-expression with full-length c-Jun shifted the fluorescent signal from the cytosol to the nucleus (Supplementary Fig. 2b). This is in agreement with earlier findings demonstrating that nuclear translocation of AP-1 transcription factors may be controlled by c-Jun/ATF2 heterodimerization²². JNK-specific phosphorylation of full-length ATF2 was examined in an engineered HEK293T cell line (HT-MLK3-MKK7) in which MAPK activity was turned on by inducible expression of specific upstream activators of JNK. These experiments showed that full-length ATF2 is less phosphorylated upon JNK activation when the critical residues (K48, CACA) are mutated (Supplementary Fig. 2c). In addition, a GAL4 promoter/luciferase assay confirmed the role of these residues in ATF2 TAD mediated basal transcription (Supplementary Fig. 2d).

Structure of the JNK1-ATF2 complex

We examined JNK1-ATF2 binding by NMR spectroscopy based on an existing NMR spectral assignment and structure for ATF2 and crystal structure of JNK-peptide interactions. Different amounts of JNK1 was added to ¹⁵N-labeled ATF2(19-78), and the intensity of HN cross-peaks in 2D [¹H,¹⁵N] HSQC spectra was monitored (Fig. 2a and Supplementary Fig. 3a). In addition to the expected changes in the ATF2 D-motif, the intensity of Zn-finger residues also similarly changed confirming that this small domain contributes to binding. Next, we created a structural model of the JNK-ATF2(19–58) complex using the High Ambiguity Driven protein Docking (HADDOCK) approach to see how the Zn-finger complements the D-motif in JNK binding²³ (Supplementary Fig. 3b,c). In addition, we set out to determine the crystal structure of the JNK-ATF2 complex using the ATF2(19–58)_Q34R/H47R construct that bound to JNK1 ~100-fold stronger compared to wild-type (Fig. 2b). The crystal structure showed that M51, L53, and F55 from the ATF2 D-motif bind in the hydrophobic pockets, while K48 binds in the negatively charged CD groove (Fig. 2c). Zn-finger residues also make contacts to JNK1 and mutations of these (K48, V44, H47 or Q34) greatly affected binding. Furthermore, arginine replacements at position 34 and 47 increased binding by creating additional H-bonds at the JNK:Zn-finger interface (Supplementary Figs. 3d and 4). The new crystallographic model provides the structure of the biologically relevant JNK1-ATF2 complex (Supplementary Fig. 3e).

p38 and ATF2 TAD binding

In addition to JNK, p38 also regulates ATF2 and thus we wanted to test if the interaction is similar or different. We used a luciferase complementation based dynamic protein–protein interaction (PPI) assay (NanoBit) to monitor both p38α and JNK1 binding to ATF2 TAD in HEK293T cells. Cells were treated with anisomycin, which activates both JNK and p38^24,25 (Fig. 3a). JNK-ATF2 binding was found to be the same if activated or not, but p38-ATF2 binding was elevated when p38 was activated. To map the p38 binding region of ATF2, phosphorylation of different ATF2 TAD constructs by activated p38α (pp-p38) was tested (Supplementary Fig. 5a). This analysis unexpectedly identified a critical short sequence (92-FENEF-96) C-terminal to the 69-TPTP-72 phosphoswitch. In ATF2 variants lacking the two phenylalanine residues (92-FENEF-96 → 92-AENEA-96; referred to as MUT4), phosphorylation of the TAD by pp-p38 was greatly reduced, but JNK mediated phosphorylation remained unaffected.

**Fig. 3: p38 mediated binding and phosphorylation of ATF2 TAD.**

The role of all evolutionarily conserved phenylalanine residues C-terminal to the phosphoswitch was then tested (MUT1, 2, 3, and 4) in a HEK293 cell line engineered for MAPK-specific activation (HT-M6) (Fig. 3b). Specific activation of p38, but not that of JNK or other MAPKs, was achieved by Tet-on inducible expression of the p38-specific MAPK kinase (MKK6) with doxycycline²⁶. These experiments confirmed the pivotal role of two phenylalanine residues (92-FENEF-96) in binding to pp-p38. To assess the importance of the Zn-finger + D-motif module and the newly identified FENEF region, we monitored the phosphorylation of mutated TAD constructs (K48E and MUT4) by pp-JNK1 and pp-p38 in vitro (Fig. 3c). These experiments demonstrated that both the Zn-finger+D-motif module and the FENEF motif are indispensable for p38 mediated TAD phosphorylation. In contrast, JNK mediated phosphorylation was unaffected by FENEF motif mutations but was greatly impaired by the K48E mutation.

Next, p38 binding of ATF2 TAD was explored by using NMR spectroscopy. ATF2 (19-106) was ¹⁵N-labeled, mixed with pp-p38 or nonphosphorylated (np) p38, and chemical shift perturbations (CSP) and intensity of HN cross-peaks in 2D [¹H,¹⁵N] HSQC spectra were compared (Fig. 3d and Supplementary Fig. 5b). Residues displaying line broadening or large CSPs mapped to the FENEF motif and also to the D-motif with pp-p38, while np-p38 affected only the D-motif region weakly. This analysis showed that the major interaction site for pp-p38 is at the FENEF motif but some binding to the D-motif region was also detected.

JNK phosphorylation attenuates p38-SPFENEF motif binding

The region between the phosphoswitch and the FENEF motif in ATF2 contains a putative S/TP phosphorylation site (S90). This SP site is only conserved in vertebrates, but not in other eukaryotes: the serine is replaced by asparagine in all invertebrate ATF2 orthologs (see Supplementary Fig. 1c). Interestingly, Ser90 was shown to be phosphorylated only by JNK but not by p38²⁷. Intrigued by these observations, we created a human ATF2 TAD construct in which Ser90 was changed to asparagine (S90N) and analyzed MAPK-ATF2 TAD binding in HEK293T cells using the NanoBit PPI assay. We used anisomycin treatment to activate JNK and p38 in parallel (Fig. 4a). ATF2 S90N showed increased binding over WT-ATF2 to pp-p38 and p38 binding was not affected by JNK phosphorylation. In contrast, JNK activation decreased binding of ATF2 TAD to pp-p38, since cells in the presence of a JNK-specific inhibitor (JNK-IN-8) displayed elevated pp-p38:WT TAD binding. Next, the effect of chemically synthesized peptides (83-GLFNELASPFENEFKKASED-102, referred to as SPFENEF peptide henceforth) on pp-p38 mediated ATF2 phosphorylation was tested in vitro. Three peptides (WT, S90N and pS90) added in trans inhibited phosphoswitch phosphorylation differently in vitro (K_i = 33.6 µM, 7.5 µM and 820 µM, respectively) (Fig. 4b). These results showed that N90 promotes pp-p38:ATF2-TAD binding, while Ser90 phosphorylation by JNK in vertebrate orthologs negatively affects it (Fig. 4c).

**Fig. 4: JNK mediated phosphorylation affects p38-ATF2 TAD binding.**

Structure of the p38-SPFENEF motif complex

After mapping out the p38 interacting regions of the ATF2 TAD, we wanted to explore the binding surface of the SPFENEF motif on pp-p38. First, we used solution state biomolecular NMR spectroscopy. We compared the 2D [¹H,¹⁵N] TROSY spectra of (²H,¹⁵N)-labeled np-p38 or pp-p38 when titrated with WT or the S90N SPFENEF peptides (ATF2 83–102). Since we previously determined the sequence-specific backbone assignment of np/pp-p38, we were able to identify those p38 resonances that showed significant intensity change upon binding the SPFENEF motif²⁸. Interestingly, the interaction is in slow exchange in the NMR timescale—where the chemical shifts for bound and free pp-p38 could be both observed. This is in contrast to other weak MAPK-protein complexes that normally display fast or intermediate exchange, where peaks shift or undergo large intensity changes, respectively. In brief, the doubling of peaks suggests that the ATF2 TAD-bound pp-p38 state is unique and differs form the unbound state. We found that a great number of pp-p38 peaks were affected, but strikingly np-p38 did not show any major changes (Supplementary Fig. 6a). This confirms that SPFENEF specifically binds to pp-p38. Notably, pp-p38 residues at the FRS displayed significant intensity changes in the presence of S/NPFENEF peptides. We also determined the crystal structure of the pp-p38:ATF2(83–102)_S90N complex (Fig. 5a and Supplementary Table 1). This structure revealed that the FENEF motif adopts an α-helix which is wedged in-between the N-terminal half of αG and one of the α-helices from the so-called MAPK insert (α2L14). The bulky hydrophobic side-chains of F92 and F96 are buried at the FRS pocket that forms only in pp-p38 when the activation loop is double-phosphorylated (Fig. 5b) and W197 (located in the αEF/αF loop) is flipped out from its original position (Fig. 5c). This explains why ATF2 binding to p38 depends on the activation state of the kinase.

**Fig. 5: Structure of the pp-p38:ATF2(SPFENEF) complex.**

Next, we created a HADDOCK model of the pp-p38:ATF2(TAD) complex that included the Zn-finger + D-motif region docked into the DRS, the phosphoswitch region docked at the MAPK active site, and the SPFENEF peptide docked to the FRS (Fig. 5d). This structure showed that phosphoswitch (69-TPTP-72) binding at the active site is compatible with the bipartite binding of the TAD at the DRS and the FRS, and explained why mutations at either of these regions (e.g., K48E and MUT4 constructs, see Fig. 3c) decreased phosphorylation. The model also hinted on the role of S90 (as well as on the impact of asparagine replacement or phosphorylation at this position) in the regulation of the pp-p38:ATF2 complex: the residue at 90 is placed into a key position where the chemical nature of its side chain affects the conformation of the SPFENEF motif by influencing intra-motif and intermolecular H-bond formation. Apart from direct contacts between pp-p38 and the SPFENEF motif, there are helix capping interactions that stabilize the α-helical conformation of α2L14, αG, and the ATF2 helix itself by involving glutamate (E93, E95) or the S/N90 side-chains. Phosphorylation at S90 is sterically detrimental (Fig. 5e). Moreover, the structural model explains why the asparagine in invertebrates enables higher affinity binding (Fig. 5f).

In addition, we also analyzed the NMR structure of WT, S90N and pS90 peptides in solution. This analysis suggested that this ATF2 TAD region has α-helical propensity (up to ~20% in aqueous buffer) around the critical SP site, but depending on the residue at 90, the conformation of the peptide varied (Supplementary Fig. 6b). In summary, these structural data show how the amino acid at 90 affects the conformation and p38-binding capacity of the SPFENEF motif.

Transcriptional activation by two MAPKs

We showed that JNK and p38 bind to the ATF2 TAD through distinct regions. To specifically address the role of these two MAPKs in ATF2 TAD mediated transcription, wild-type TAD and different binding site mutants thereof were expressed as fusion proteins with the GAL4 DNA-binding domain and their capacity to activate GAL4 DNA-binding based transcription were tested. In agreement to the results of in vitro or cell-based binding experiments, all mutations (both in the Zn-finger + D-motif or in the SPFENEF motif), apart from the S90N mutation, showed decreased basal transcription from a luciferase expression based reporter vector (Fig. 6a). Basal transcription from the reporter vector was decreased in the presence of JNK- or p38-specific inhibitors (JNK-IN-8 or SB202190, respectively) or increased by heterologous expression of JNK1 or p38α, but the impact on p38 was only apparent in the case of the S90N construct (Fig. 6b, c).

**Fig. 6: Co-regulation of ATF2 TAD mediated transcription by JNK and p38.**

Next, the ATF2 TAD was turned on by artificial and highly specific induction of JNK or p38. HEK293T cells were transfected with an expression plasmid allowing doxycycline (DOX) inducible (Tet-on) expression of an MLK3-MKK7 fusion construct or a constitutive active version of MKK6 (MKK6) to ensure JNK- or p38-specific activation, respectively. The HT-MLK3-MKK7 and HT-MKK6EE stable cells lines (HT: HEK293T Tet-on) allowed single MAPK activation, but the HT-MKK7-MLK3/MKK6EE cell line allowed controlled double activation of JNK and p38 at the same time (Fig. 6d). These experiments, which lacked the pitfalls of cross-activation of both MAPK pathways by unspecific stimuli (e.g., anisomycin or UV radiation), demonstrated that under these conditions JNK and p38 mediated activation of the ATF2 TAD can indeed be synergistic.

Quantitative modeling of MAPK mediated ATF2 TAD phosphorylation

To simulate the level of phosphorylated ATF2 in response to different patterns of JNK/p38 pathway, we created a rule-based mechanistic model (Fig. 7a and Supplementary Table 2). The MAPKs bound to their respective TAD regions, their activated form phosphorylated the T69/T71 phosphoswitch, and pp-JNK also phosphorylated S90. Binding affinities and phosphorylation rates were determined in vitro (Supplementary Fig. 7). Proteins are dynamically phosphorylated by protein kinases and dephosphorylated by phosphatases in the cell, and this reversible modification plays a key role in biological regulation. Therefore, the model was extended to include the action of upstream kinases and deactivating phosphatases for MAPKs, as well as phosphatases that counteracted MAPK mediated phosphorylation of ATF2. In order to focus on MAPK → substrate phosphorylation and keep the number of parameters low, we used a simplified model to implement MAPK activation and dephosphorylation where MAPKs had only two states: an unphosphorylated/inactive and a phosphorylated/active. MAPKs were activated by upstream kinases (k6 or k7) and deactivated by phosphatases (dp1 and dp2) at a single site, and the action of these enzymes on MAPKs were simply modeled as first order reactions. These enzymes are known to bind to the same surface on MAPKs as the substrates (e.g., ATF2), therefore, their binding is mutually exclusive²⁹. ATF2 phosphorylation was counteracted by independent phosphatases acted on T69/T71 or S90 phosphorylation sites (dp3 and dp4). Parameters were determined based on the response of HEK293T cells to anisomycin treatment (Fig. 7b). Anisomycin stimulation experiments were carried out with wild-type, S90N, and MUT4 ATF2 constructs in the absence or presence of a JNK-specific inhibitor and p38-ATF2 binding (i.e. NanoBit signal in live cells), MAPK and ATF2 probe phosphorylation (by Western-blots) were concomitantly monitored in time up to 40 min.

**Fig. 7: Modeling the impact of JNK and/or p38 MAPK pathway flux on ATF2 TAD phosphorylation.**

The mechanistic model was then used to calculate the in-cell phosphorylation level of the ATF2 T69/T71 phosphoswitch (pp-ATF2) under varying amounts of activated p38 and JNK levels (Fig. 7c). For ATF2-S90N, the JNK regulated S90 is OFF as this site cannot be modified and MAPK effects are additive towards building up high levels of pp-ATF2, and notably the inflection point is ~5% of the total flux for both MAPKs (see the left panels on Fig. 7c). In contrast, the 2D response surface is more complex with the ATF2 WT protein (where the JNK regulated S90 is ON and may get phosphorylated): (i) pp-ATF2 level stays low in response to low-to-medium p38 flux even under small amounts of JNK pathway flux, which increases the 2D dynamic range (i.e., the inflection point is ~25% of the maximum; see right panels on Fig. 7c), (ii) in general pp-ATF2 levels are more sensitive to JNK compared to p38 pathway flux, and (iii) increasing JNK flux under high pp-p38 levels can bring about maximum pp-ATF2 output only after going through a local minimum or threshold (see Fig. 7c, right panel, 3D plot).

The in silico model on MAPK controlled ATF2 phosphorylation was based on experimentally determined parameters regarding MAPK and ATF2 TAD binding. Next, we were curious about how these binding parameters affect ATF2 phosphorylation. The parameters for JNK(DRS)-, p38(FRS)-, and p38(DRS)-ATF2 TAD binding were arbitrarily changed to simulate three distinct scenarios: (i) JNK would bind to Zn-finger + D-motif module 100-fold stronger (which could be achieved by increasing the k_on of the interaction); (ii) p38 would bind to the SPFENEF motif 100-fold stronger, and (iii) p38 would bind with 10-fold stronger—both to the Zn-finger + D-motif and the SPFENEF motif—and JNK with 10-fold weaker affinity (Fig. 7d). In the first and second case, pp-ATF2 levels were less responsive or oversensitive to p38 activation, respectively, and in the third case, pp-ATF2 was less responsive to JNK activation.

In conclusion, the model showed that the MAPK binding affinity and the specificity of ATF2 TAD regions are set in vertebrates so that to allow graded phosphoswitch activation under broad p38/JNK pathway activity in the cell.

Discussion

We found that an intact Zn-finger partially overlapping with the D-motif is the minimal JNK binding domain of ATF2. Interestingly, similar synergism between a small structured domain (PB1) and the MKK5 D-motif was described in the unrelated MKK5-ERK5 complex³⁰. In addition to making extra contacts, the Zn-finger in the ATF2 TAD contributes to the binding energy by decreasing the entropic cost of binding in the MAPK D-groove. The ATF2 D-motif confers to a classical NFAT4-type MAPK binding consensus, and a linear D-motif would be sufficient to promote the phosphorylation of the T69/71 phosphoswitch³¹. However, we had showed earlier that D-motifs have only limited MAPK binding specificity¹⁶. Interestingly, the ATF2_CACA construct (in which the Zn-finger structure is disrupted) bound p38 better compared to wild-type while JNK binding decreased (Supplementary Fig. 5c). This suggests that the Zn-finger limits the capacity of the partially overlapping D-motif to bind to p38 and favors JNK binding. The Zn-finger+D-motif module is more JNK-specific compared to an NFAT4-type D-motif, and specificity is achieved through both positive and negative factors that promote JNK- and limit p38-binding, respectively.

Double-phosphorylated p38 bound ATF2 stronger than non-phoshorylated p38. This is similar to how ERK2 binds F-site peptides (e.g., the FxFP motif from ELK1)³². Although ATF2 does not have an FxFP motif, the SPFENEF motif found to be critical for pp-p38 binding also has two phenylalanine residues. Therefore, we posited that the SPFENEF region binds to the FRS, which was then confirmed by NMR based interface mapping and X-ray crystallography. How the residue at position 90 in the sequence affects pp-p38 binding is intriguing. An asparagine at this position mediates stronger binding (by about 5-fold) compared to a serine, while phosphorylation at Ser90 abolishes binding. The former may be explained by a more optimal side-chain at a critical position of the protein–protein binding interface. Conversely, a far less favorable chemical group (i.e. a highly charged and bulky phosphate) in the same position explains diminished binding.

The SPFENEF motif specifically binds to p38 and the structural basis of FRS binding specificity can be explained by the crystal structure of the pp-p38:ATF2(83–102) complex. The MAPK FRS pockets are overall similar but residues corresponding to W197 of p38 are different in ERK2 (serine) and JNK1 (methionine), thus ERK2 or JNK1 cannot form the FRS pocket required for ATF2 SPFENEF motif binding.

The JNK binding Zn-finger + D-motif module and FENEF motif-like regions are both present in most extant organisms even in sponges (Supplementary Fig. 1d). However, the JNK-binding module is absent while the p38-binding linear motif is clearly present in Drosophila (Supplementary Fig. 1e). In agreement with this, ATF2 is phosphorylated and regulated by p38 but not by JNK in this organism³³. Furthermore, the fission yeast (Schizosaccharomyces pombe) ATF2 homolog (Atf1) is also known to be regulated by the mammalian p38 homolog (Spc1/Hog1)³⁴. In conclusion, comparative sequence analysis of ATF2 homologs suggests that co-regulation of the ATF2 phosphoswitch by two distinct MAPKs is an ancient evolutionary trait, but the architecture of the ATF2 TAD have gone through important evolutionary changes regarding phosphoregulation.

According to the quantitative rheostat-like switch concept, the rate of transcription of a specific gene is continuously variable and coupled directly to the strength of intracellular signaling events³⁵. Compared to the stable “on” and “off” modes of gene regulation, this model can be better applied to describe transcription initiation by immediate-early (IE) genes (e.g., AP-1 factors) that do not depend on new transcription/translation and are directly linked to signaling cascades. The primary response to an extracellular cue is triggered by TF TAD phosphorylation. In our study we found that the ATF2 TAD architecture is key in determining how the transcription factor will respond to JNK, p38 or to the concomitant activation of these two. Unexpectedly, ATF2 TAD does not simply integrate the signal coming from the two MAPKs, but due to its unique architecture may sense the relative strength of these two MAPK pathways. This latter capacity of the ATF2 TAD goes beyond additive phosphorylation of the same phosphoswitch. We postulate that the basic integration capacity, emerging from an ancient pre-vertebrate ATF2 architecture comprised of JNK and p38 binding sites, became more complex due to an amino-acid change (Asn to Ser at 90) in the critical p38 binding region in vertebrates: one of the MAPKs (JNK) acquired the capacity to have a direct influence on how the other (p38) effects ATF2 mediated transcription. Unfortunately, the biological significance of this more complex regulation is not known yet. ATF2 is a constitutively expressed protein and plays a major role in controlling the transcription of other IE transcription factors and its knock-out in mice causes defects in skeletal and CNS development³⁶. Tentatively, JNK/p38 based co-regulation of ATF2 activity may contribute to primary IE gene response diversity that is characteristic to different cell-types or development processes³⁵.

We have a substantial understanding on how individual MAPKs bind to TFs, since most of the known D- or F-motifs were originally found in these proteins decades ago. Physiological responses to extracellular cues depend on a complex activation pattern of different MAPKs (e.g. ERK1/2, JNK and p38), which is not exclusive but combinatorial, and not binary (i.e., on or off) but quantitative. Moreover, deviation from a normal pattern—which may depend on the cell type—leads to signaling related diseases such as cancer or inflammation^20,37, and now we know that even to drug resistance³⁸. For ATF2 we showed how this MAPK substrate reads out different MAPK activation profiles. The mechanistic basis behind this seemingly complex process is simple: MAPK binding regions and their phosphorylation target sites are organized into a specific linear pattern. Ultimately, it is the TAD architecture around the transcription controlling phosphoswitch that is fine-tuned according to the signaling logic of the organism, but the building blocks—protein kinases, their binding and phosphorylation target motifs—are universal.

Methods

Protein expression and purification

All primers and oligonucleotides used in this study are listed in Supplementary Table 3. All protein constructs are based on the human sequences. Full-length ATF2 (1–505; Uniprot: P15336-1) was amplified from HEK293T cDNA and then various fragments of the TAD were cloned into a modified pET-15 vector (Novagen) including N-terminal GST/MBP tag followed by a tobacco etch virus (TEV) cleavage site and a C-terminal hexa-histidine tag (Supplementary Table 3). For structural studies ATF2 TAD fragments were inserted into a modified pBH4 vector. Proteins were expressed with an N-terminal hexa-histidine tag that was cleaved off by the TEV protease. Protein constructs had a tyrosine inserted at the ATF2 N-terminus to enable protein concentration determination by absorbance at 280 nm. Mutant constructs were generated by two-step PCR mutagenesis. DNA expression plasmids were transformed into BL21(DE3) cells and grown in LB supplemented with 200 µM Zn-acetate until OD₆₀₀ = 0.5, protein expression was induced with 200 µM isopropyl β-d-1-thiogalactopyranoside (IPTG) overnight at 18 °C. Cell pellets were lysed and the recombinant protein was purified using Ni-NTA agarose beads. Some ATF2 constructs were expressed as maltose-binding protein (MBP) fusions. A modified pET-15 vector (pET-MBP) was used to express MBP-ATF2-His6 constructs that were expressed similarly as above and purified on maltose resin after Ni-NTA purification. Protein samples were stored in MBP elution buffer (20 mM Tris pH 8, 150 mM NaCl, 10% glycerol, 2 mM TCEP, 0.1% BOG (Octyl-beta-Glucoside), 2 mM BA (benzamidine), 30 mM maltose) for fluorescence polarization (FP) based measurements or for kinase assays. For ITC measurements the MBP tag was removed by TEV protease and ATF2-His6 was further purified on a Resource S column in buffer containing 20 mM MES (pH 6.3), 25 mM NaCl, 10% glycerol and 2 mM βME and eluted with NaCl gradient. For crystallography His6-ATF2-(19–58)-WT or the Q34R/H47R mutant was captured on Ni-NTA agarose beads, the hexa-hisitidine tag was cleaved off by the TEV protease, the protein was purified on a Resource S column (pH 6.3) and gel-filtrated into storage buffer (20 mM Tris, pH 8, 150 mM NaCl, 2 mM TCEP) using a Superdex75 10/30 size-exclusion column (GE Healthcare). For NMR studies His6-ATF2-(19-78) or -(19-106) were transformed into BL21(DE3) and the culture (2 L) was grown in LB at 37 °C until OD₆₀₀ = 0.6–0.8. Cells were pelleted and transferred to 0.5 L M9 media containing 1 g ¹⁵NH₄Cl, 200 µM Zn-acetate, 1× vitamin mix (Sigma, #B6891), 1× MOPS micronutrient mix and 2 g glucose, then induced with 1 mM ITPG at 37 °C for 4 h. Labeled protein was captured on Ni-NTA resin; after cleaving the hexa-histidine tag by TEV protease, ¹⁵N-ATF2 was further purified on Resorce S (19–78, pH 6.3) or Q column (19–106, pH 9) and gel-filtrated into storage buffer (20 mM Tris, pH 8, 150 mM NaCl, 2 mM TCEP).

JNK1-dc20 (where the C-terminal 20 residues are truncated), p38α and ERK2 were expressed in E. coli using standard procedures. Phosphorylated MAPKs were produced by co-expressing them with constitutively active GST-tagged MAP2Ks¹⁶. The purity of the protein samples were analyzed in SDS-PAGE. All of the protein samples were flash-frozen in liquid nitrogen and stored at −80 °C in buffers containing 10% glycerol. None of the purification, storage and assay buffers contained EDTA nor DTT so that to maintain the structural integrity of the ATF2 zinc-finger domain, but samples contained 2 mM TCEP.

SPFENEF peptide (ATF2: 83–102), its S90N and phospho-S90 versions were all chemically synthesized on an automated PSE Peptide Synthesizer (Protein Technologies, Tucson) with Fmoc strategy.

Protein–protein binding assays

GST pull-down experiments with GST-ATF2 constructs and GST control as baits and JNK1 as prey were performed using purified proteins in GST-binding buffer (20 mM TRIS, pH 8.0, 100 mM NaCl, 0.1% IGEPAL, 2 mM TCEP)¹⁶. Fluorescence polarization (FP) based protein–protein binding experiments were carried out using 50 nM carboxyfluorescein(CF)-labeled CF-evJIP (CF-PPRRPKRPTSLDLPSTPSL) for JNK1, and the data was analyzed using OriginPro. For direct measurements the concentration of JNK1 was increased in a titration experiment, then the binding of the unlabeled peptide was determined by titrating the JNK1-labeled peptide complex sample—at a JNK1 concentration corresponding to ~60–80% complex formation—with increasing amounts of the unlabeled peptide¹⁶. All of the FP measurements were performed in 20 mM Tris (pH 8.0), 100 mM NaCl, 0.05% Brij35P, and 2 mM TCEP and measured using a Cytation^TM 3 (BioTek Instruments) plate reader in 384-well plates. ITC measurements were performed on MicroCal iTC200 (GE Healthcare) at 20 °C. Injection parameters were identical for all experiments (1.5 μL injections every 180 s, 20 in total at a stirring speed of 750 rpm). Before the experiments protein samples were dialyzed overnight in ITC buffer (50 mM HEPES, pH 8.0, 150 mM NaCl, 10% glycerol, 0.5 mM TCEP) in the absence or presence of 2 mM EDTA and subsequently concentrated. MAPK concentration in the measuring cell was 100 µM.

Kinase assays

MBP-ATF2-His6 protein constructs were double affinity purified and used as MAPK substrates. For ATF2 substrates with c-Jun and NFAT4 docking motifs, the ATF2 Zn-finger+docking motif region was replaced by the following docking peptide sequences: SNPKILKQSMTLNLADP and LERPSRDHLYLPLEP, respectively), which were followed by the ATF2 59-100 region. Substrates were dotted on nitrocellulose membrane and the assay was performed by incubating the membranes with the MAPK in kinase buffer (50 mM HEPES, pH 7.4, 100 mM NaCl, 5% glycerol, 0.05% NP-40, 2 mM TCEP, 2 mM MgCl₂) supplemented with 1 mM ATP. Kinase assays were performed in kinase buffer containing 2 μM substrate and 2 nM enzyme. Reactions were stopped by adding SDS-loading buffer containing 20 mM EDTA and samples were subjected to western-blots. Enzyme kinetics measurements on ATF2 phosphorylation were performed in solution using 5 nM enzyme, MBP-ATF2(19–100) substrate (100, 50, 25, 10, 5, 2.5, 1, and 0.5 μM) and the reaction was started by adding 5 mM ATP. Competition assays with synthetic SPFENEF peptides were performed using 2 nM pp-p38, 2 μM MBP-ATF2 substrates and 3-fold dilution series of peptides starting from 1 mM. Reactions were stopped by adding kinase buffer containing 50 mM EDTA and diluted to 0.05 µM MBP-ATF2 in the same buffer. 1 µl of the diluted reaction mixture was spotted and dried on a nitrocellulose membrane in the presence of a calibration series. Calibration series were made from 2× dilution of fully phosphorylated MBP-ATF2 mixed with dephosphorylated MBP-ATF2 (0.05 µM in total) to confirm the linear range of the detection, and the rate of product formation was calculated. Membranes were developed using anti-phospho-T71 ATF2 antibody (Cell Signaling Technology, #9221) at 1:3000 dilution or anti-phospho-T69/T71 ATF2 (Merck, #05-891) at 1:5000 dilution or anti-HIS (Sigma, #H1029) at 1:10,000 dilution and a secondary anti-rabbit antibody (IRDye® 800CW Goat anti-Rabbit IgG, #926-32211) or secondary anti-mouse antibody (IRDye® 800CW Goat anti-Mouse IgG, #926-32210), respectively, at 1:10,000 dilution with LI-COR Odyssey® Clx infrared imaging system and Image Studio Lite. For K_i measurements, the maximal velocity was determined from a reaction containing no peptide, and IC₅₀ values were obtained by using the logistic dose–response equation in QtiPlot. K_i was calculated as K_i = IC₅₀ /(1 + S/K_M) where the K_M was 17.3 μM (pp-p38–MBP-ATF2(19–100)-WT).

NMR spectroscopy

For the JNK:ATF2 complex the ¹H-¹⁵N heteronuclear single-quantum coherence (HSQC) spectra were obtained using 50 μM uniformly ¹⁵N-labeled ATF2-(19–78) in the absence or presence of JNK1 at molar ratios of 0.5:1 and 1:1 [JNK1]:[ATF2]. Experiments were carried out in 20 mM potassium phosphate (pH 6.3), 200 mM NaCl, 5% d-Glycerol and 2 mM TCEP containing 10% (v/v) D₂O at 25 °C on an AVANCE II Bruker spectrometer (800 MHz) equipped with a TCI-active HCN cooled z-gradient cryoprobe. Published ¹H, ¹⁵N resonance assignment of ATF2-(19–106) were used to analyze the data¹¹. Spectra were processed using NMRPipe³⁹ and NMR data were analyzed using NMRFAM-SPARKY⁴⁰. Peak heights were used to calculate intensity ratios of ATF2 in the presence versus in the absence of JNK1.

To map out the p38 binding region of the ATF2 TAD experiments were carried out on a four-channel 600-MHz Varian NMR SystemTM spectrometer equipped with a 5-mm indirect detection triple resonance (¹H¹³C¹⁵N) z-axis gradient probe. ¹H-¹⁵N HSQC spectra were performed on 100 μM uniformly ¹⁵N-enriched ATF2-(19–106) in the absence and presence of pp-p38 and np-p38 at molar ratios of 0.5:1 and 1:1 ([pp-p38]:[ATF2]) and 2:1 ([np-p38]:[ATF2]). Experiments were carried out in 20 mM potassium phosphate (pH 6.3), 200 mM NaCl, 2 mM TCEP containing 10% (v/v) D₂O at 27 °C. For the analysis of the data, previously published ¹H, ¹⁵N resonance assignment of ATF2-(19–106) was used¹¹. Spectral processing and data analysis were carried out using Felix 2004 (Accelrys, Inc.) and CCPNMR. To assess the combined (¹HN, ¹⁵N) chemical shift perturbations, values of Δδ(¹HN,¹⁵N) were calculated using the equation of Δδ_1HN,15N = √[(Δδ_HN)2 + (w1*Δδ_N)2], where w1 (= 0.154) is a weight factor determined using the BioMagResBank chemical shift database⁴¹.

Production of np-p38 and pp-p38 for NMR measurements was done similarly as described before²⁸. Briefly, expression of uniformly [²H,¹⁵N]-labeled p38 was achieved by growing cells in D₂O based M9 minimal media containing 1 g/L ¹⁵NH₄Cl, respectively, and D-glucose as the sole nitrogen and carbon sources. Multiple rounds (0, 30, 50, 70, and 100%) of D₂O adaptation was necessary for high-yield expression. The interaction between nonphosphorylated or phosphorylated p38 and the SPFENEF motif was studied by mixing the chemically synthesized peptide with 0.1 mM [²H,¹⁵N]-labeled np-/pp-p38 and the 2D [¹H,¹⁵N] TROSY spectra of free and peptide added samples were compared. Samples were in 10 mM HEPES pH 7.4, 0.15 M NaCl, 5 mM DTT and 90% H₂O/10% D₂O. The spectra were processed using Topspin 4.0.6 and analyzed using NMRFAM-SPARKY. The NMR spectra were acquired at 35 °C on a Bruker Avance NEO 600 MHz NMR spectrometer equipped with a TCI-active HCN cooled z-gradient cryoprobe.

Resonance assignment of the SPFENEF peptides (ATF2 83–102) was obtained using the combination of 2D TOCSY, 2D NOESY and natural abundance ¹H-¹³C HSQC experiments. ¹H chemical shifts were referenced to external 4,4-dimetyl-4-silapentane-1-sulfonate (DSS) and the ¹³C chemical shifts were referenced indirectly to DSS. Experiments were carried out on a four-channel 600-MHz Varian NMR SystemTM spectrometer equipped with a 5-mm indirect detection triple resonance (¹H¹³C¹⁵N) z-axis gradient probe. Secondary structure propensities (SSP) of the peptides were calculated on the basis of H_α, C_α, and C_β chemical shifts using the SSP program⁴². Briefly, the SSP score of a residue can be interpreted as the expected fraction of α-helical or β-strand/extended structure. In the lack of long-range distance restraints in NOESY spectra, dihedral angles derived from SSP scores were used as restraints in structure calculations with ARIA (Ambiguous Restraints for Iterative Assignment, version 2.2). Typically, in each of the seven iterations, the ten lowest energy structures were used as templates for the next iteration and the seven best structures were used for restraint violation analysis. The computational algorithm in the structure calculation employed torsional angle simulated annealing followed by torsional angle and then Cartesian molecular dynamics cooling stages. Structural refinement was completed in a water shell.

HADDOCK modeling

For the JNK-ATF2 complex, the input model ensemble for ATF2 was created manually based on the ATF2 TAD NMR structure (PDB ID: 1BHI) and D-motif containing JNK-peptide complexes in Coot (Supplementary Fig. 3b)⁴³. The input model for JNK was the JNK1-pepNFAT4 crystal structure (PDB ID: 2XS0) after deleting the NFAT4 peptide. Docking was performed with the Haddock2.2 webserver using the Expert interface and unambiguous distance restraints were derived from MAPK:D-motif crystal structures (Supplementary Fig. 3c)²³. For the first step of rigid-body energy minimization, 1000 structures were generated. 200 structures with the lowest energy from rigid-body docking were subjected to semi-flexible simulated annealing in torsion angle space followed by refinement in explicit water.

For the pp-p38:ATF2-WT(TAD) complex, the pp-p38 and ATF2 SPFENEF (83–102) structure was extracted from the crystal structure of the pp-p38-ATF2(S90N) complex, and N90 were mutated to serine in Coot. The ATF2 Zn-finger + D-motif was taken from the JNK-ATF2(Zn-finger + D-motif) crystal structure and the substrate motif (67-DQTPTP-72) was modeled as in the DYRK1A-substrate peptide complex (PDB ID: 2WO6)⁴⁴. Unambiguous distance restraint were used for docking the Zn-finger + D-motif to pp-38 similarly as described for the JNK-ATF2 complex (see Supplementary Fig. 3c). For the docking of the substrate motif (DQTPTP) to pp-p38 at the active site, unambiguous distance restraints were derived from the DYRK1A-peptide complex, where the T69 was treated as the phosphorylation target site. SPFENEF docking was driven by intermolecular (ambiguous) restraints based on perturbations on the pp-p38 2D [¹H,¹⁵N] TROSY spectra. Active residues in the FRS site of p38 were chosen based on the results of the pp-p38:SPFENEF peptide (1:4) sample. Residues with >50% intensity decrease and with >10% solvent accessibility (RSA) were defined as active. Passive residues were defined as solvent exposed (RSA values of 10% or above) and contiguous to active residues. Active residues were limited to the MAPK FRS (or also called DEF pocket), since the pp-p38-ATF2(S90N) crystal structure indicated this MAPK region to be the major contact surface, moreover, mutating this region in p38 had been shown to negatively affect p38-mediated ATF2 phosphorylation^45,46. In the SPFENEF peptide F92 and F96 were treated as active residues, and its neighbors as passive residues. 1000 docked models were created through a rigid body minimization protocol, then the top scoring 200 structures underwent a semi-flexible simulated annealing protocol. Finally, all the resulting structures underwent refinement in explicit solvent. The three ATF2 peptides were stitched together in MODELLER⁴⁷.

Protein crystallization and X-ray structure solution

JNK1-dC20 and ATF2-Q34R/H47R(19–58) were mixed at 1:1.5 ratio (240 µM JNK1 + 360 µM ATF2) in the presence of 5 mM TCEP and 1 mM adenosine-5’-[(β,γ)-imido]triphosphate (AMP-PNP). Crystals grew in 8% w/v PEG 20 000, 10% v/v PEG MME 550, 0.025 M MgCl2, 0.025 M CaCl₂, 0.08 M bicine/Trizma base (pH = 8.8) in hanging drops at 4 °C and the reservoir solution was 1.25 M NaCl. Crystals were flash cooled in liquid nitrogen after adding 2 µl crystallization solution with increased PEG concentration (11% w/v PEG 20,000 and 22% v/v PEG MME 550) to the drop. Data were collected at PETRA III beamlines, Hamburg, and processed by XDS and CCP4i. The structure was solved by molecular replacement using PHASER⁴⁸. The JNK1 search model contained the full polypeptide chain of the MAPK from the JNK1-NFAT4 protein-peptide structure (PDB ID: 2XS0), while the ATF2 search model was from the solution structure of the zinc-finger domain (PDB ID: 1BHI, model 1). The MR solution contains two JNK-ATF2 complexes. The ATF2 C-terminal docking motif region, the ATF2 N-terminal residues, the JNK activation loop (chain B) and AMP-PNP were manually built in Coot. Structure refinement (2.7 Å resolution) was done using PHENIX with NCS restraints for corresponding polypeptides⁴⁹. Structural figures were prepared using PyMol.

For the p38-ATF2 complex, pp-p38 and pepATF2_S90N(83–102) were mixed together at 1:5 ratio (250 µM pp-p38 + 1250 µM peptide) and supplemented with 5 mM TCEP and with an ATP competitive inhibitor (375 µM, Skepinone-L). Crystals grew in 14% PEG3350, 0.1 M cacodylate pH 6.5 in hanging drops at 4 °C and the reservoir solution was 1.375 M NaCl. Crystals were flash cooled in liquid nitrogen in the crystallization solution containing 30% glycerol. Data were collected at PETRA III beamlines, Hamburg. The structure was solved similarly as described for the JNK-ATF2 complex. The search model for molecular replacement was phosphorylated p38 from the crystal structure of the pp-p38:MK2 binary complex (PDB ID: 6TCA). The final crystallographic model contains one pp-p38:ATF2(83–102):Skepinone-L complex where ATF2 residues 88-101 could be traced in the electron density map (1.95 Å resolution). Details of the data collection and structure refinement are found in Supplementary Table 1.

Cell culture

HEK293T cells (purchased from ATCC, CRL-3216) were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Gibco) containing 10% fetal bovine serum (FBS) and 50 μg/ml gentamicin at 37 °C in an atmosphere of 5% CO₂. For preparing the HT-M cell lines, HEK293T cells were transfected with pEBDTet vectors containing a MAP2K-MAP3K kinase domain fusion or constitutively activated versions of full-length MAP2Ks: MLK3-MKK7 and MKK6EE that turn on JNK and p38, respectively²⁶. HEK293T Tet-on (HT) stable cell lines (HT-MLK3-MKK7 and HT-MKK6EE) were established by keeping the cells under puromycine or hygromycin for at least one week, then the expression of the FLAG-tagged transgene and concomitant specific MAPK activation were monitored by western-blots after doxycycline (DOX) treatment (2 µg/mL) in DMEM without FBS. Cells were transfected with Lipofectamine 3000 reagent and were kept under puromycine (5 µg/mL) or hygromycin (100 µg/mL) for selection.

For analyzing the phosphorylation of full-length ATF2 mutants by endogenous JNK, HT-MLK3-MKK7 cells were seeded in 24-well plate and were transfected the next day with 250 ng F1-ATF2 (1–505) constructs with Lipofectamine 3000 in Optimem. The medium was changed to DMEM 4 h after transfection and cells were induced for 0, 8 or 24 h with 2 µg/mL doxycycline. Cells were harvested in SDS loading buffer and subjected to SDS-PAGE, then samples were transferred to nitrocellulose or PVDF membrane. Anti-phospho-ATF2-T69/T71 (Merck, #05-891; 1:5000 dilution) and anti-phospho-p38 (Cell Signaling, #9215; 1:3000 dilution) signals were detected in nitrocellulose membrane, while anti-phospho-JNK (Cell Signaling, #9251; 1:2000 dilution), anti-FLAG (Sigma, #F1804; 1:10,000 dilution) and anti-tubulin (Sigma, #T6199; 1:10000 dilution) were detected on PVDF membrane using appropriate secondary antibodies (IRDye® 800CW Goat anti-Rabbit IgG, Li-Cor, #926-32211; IRDye® 800CW Goat anti-Mouse IgG, Li-Cor, #926-32210; or IRDye® 680RD Goat anti-Mouse IgG, Li-Cor, #926-68070; 1:10,000 dilution).

Cell-based protein–protein interaction assays

Yellow fluorescence protein (YFP) based BiFC assay was performed with YFP constructs split at position 159, and fragments (F1 and F2) were pasted into pcDNA 3.1 vectors (Invitrogen) to monitor interaction and localization of the JNK-ATF2 complex in HEK293T cells⁵⁰. JNK1 was expressed as N-terminal F2 fusions and ATF2(1–505) as N-terminal F1 fusion. Full-length c-Jun (1–331, UniProt ID: P05412-1) was cloned into pcDNA 3.1 vector containing C-terminal FLAG-tag or N-terminal F1 fusion. For imaging, cells were seeded to 8-Well Glass Bottom µ-Slide plate (#80827, Ibidi) and fixed with 4% paraformaldehyde solution 24 h after transfection, stained with 0.1 μg/ml DAPI and imaged with EVOS M7000 Imaging System (Thermo Fischer Scientific). Images were processed by the ImageJ software and nuclear localization of the fluorescent signal was calculated as a percentage of the total fluorescence intensity.

For luciferase complementation NanoBiT (Promega) assays⁵¹, 2 × 10⁴ cells per well were seeded onto a white 96-well plate (Greiner, 657160) 24 h prior to transfection. Transient transfections were carried out with Lipofectamine 3000 reagent according to the NanoBiT system’s instructions. cDNAs were sub-cloned into Lbit and Sbit expression vectors: MAP kinases (JNK1 or p38α) were expressed as N-terminal Lbit fusions and ATF2 constructs as N-terminal Sbit fusion proteins. The FBS was removed 4 h after transfection and cells were starved for 20 h before assayed using the Nano-Glo reagent (Promega) according to the manufacturer’s instructions in a Cytation 3 plate reader (BioTek). MAPK stimulation was initiated by adding 10 μg/mL anisomycin (#A9789, Sigma-Aldrich) at 37 °C and 5% CO₂ level. Transfected and treated cells were subjected to western-blots using western-blots using anti-phospho-ATF2-T69/T71 (Merck, #05-891; 1:2000 dilution), anti-phospho-c-Jun-S63 (Santa-Cruz, #sc-822; 1:2000 dilution) and other phospho-specific antibodies mentioned above. In the case of HTM-6 cell line, 2 μg/ml doxycycline was added 20 h prior the measurement to induce MKK6-EE expression and the luciferase activity was measured (for 10–15 min) after addition of 10 μM luciferase substrate (Coelenterazine h).

Transcription activation assays

ATF2-TAD (19–100) constructs were subcloned into pBIND vector as N-terminal fusion of GAL4 DNA-binding domain (1–147). The pBIND vector also expresses Renilla luciferase as a transfection control. 2 × 10⁴ HEK293T cells per well were seeded onto a white 96-well plate 24 h prior to transfection and co-transfected with 50 ng GAL4-ATF2 and 50 ng 5×GAL4-Firefly-luciferase reporter plasmid (pGL4.31) using Lipofectamine 3000 in DMEM. 4 h after transfection the medium was changed to DMEM, and cells were serum starved for 20 h and assayed with Dual-Glo® Luciferase reagents (Promega, #E2920) according to manufacturers instruction. RLU (Relative Luminescence Unit) were determined as Firefly Luciferase activity/Renilla Luciferase activity. Inhibitor treatments were performed adding 1 µM JNK-IN-8 (Selleckchem, #S4901), 1 µM SB202190 (Sigma-Aldrich, #S7067) or 1 + 1 µM JNK-IN-8 + SB202190 8 h before measurement (final DMSO concentration was held at 0.1%). 10 ng Cerulean-empty-FLAG, Cerulean-JNK1-FLAG and Cerulean-p38α-FLAG were co-transfected with 50 + 50 ng GAL4-ATF2 and reporter vector where indicated. HT-MLK3-MKK7(Puro), HT-MKK6EE(Hygro) or HT-MLK3-MKK7/MKK6EE(Puro+Hygro) cell lines were used to activate endogenous MAPKs by adding 2 ug/mL doxycycline 16 h before the measurement. Samples were subjected to western-blot analysis and GAL4-ATF2 expression was detected using anti-GAL4 antibody (Santa-Cruz, #sc-510; 1:5000 dilution) or phospho-specific antibodies mentioned above. All experiments were performed independently at least 3 times and tested with paired Student’s t-test, two-sided. Visualization and statistical analysis was done using LibreOffice Calc.

Simulation of TAD phosphorylation

Rule-based network modeling was carried out with the software package BioNetGen with the ordinary differential equation solver running on a desktop PC (Supplementary Software 1)⁵². Pathway activation was initiated from a pre-equilibrated state with baseline k7 (k_eq7) and k6 (k_eq6) activity (see Fig. 7a). To mimic stimulation by anisomycin, signaling was started by increasing k7 (k_stim7) and k6 (k_stim6) activity. The BioNetGen model includes the rules that JNK-ATF2 binding is not affected by the phosphorylation state of proteins and ATF2 binds to the DRS of JNK. ATF2 binds to the DRS and FRS of pp-p38 in a bipartite manner (k_on2, k_off2), but only when p38 is phosphorylated and ATF2 is not phosphorylated on S90. p38 binds to the DRS with only low affinity (k_on3, k_off3, K_D ~1 mM) when bipartite binding is not possible. Kinetic parameters for ATF2-JNK (k_on1, k_off1) and ATF2-p38 (k_on2, k_off2) binding and rate constants for ATF2 T69 and T71 phosphorylation were determined based on in vitro kinase assay data (see Supplementary Fig. 7). Parameters were adjusted to the data using the differential evolution algorithm of PyBioNetFit⁵³. The rate of T69 and T71 phosphorylation by JNK was set to be equal (k1). In the case of p38 mediated phosphorylation events, the T71 phosphorylation rate of T69 phosphorylated ATF2 was set to be slower (k4), compared to the normal p38-mediated T69/T71 phosphorylation rates (k3)⁵⁴. The rate of S90 phosphorylation by JNK was derived from [γ-³²P]ATP based in vitro kinase assay data using the ATF2-TATA substrate (k2) (see Supplementary Fig. 7c).

The results of in-cell NanoBit measurements with MAPK and ATF2 PPI probes and related Western-blot data were used to determine k7 or k6 mediated phosphorylation and phosphatase mediated dephosphorylation rates upon anisomycin treatment (see Fig. 4). The NanoBit assay data were simulated in a simplified system consisting of ATF2, JNK and p38 (1 µM), and additional exogenous p38 (1.65 µM). The level of activated JNK after 40 min of anisomycin treatment was set to be 10% of the total. Western-blot data of pp-JNK were used in PyBioNetFit to determine kinase (k_stim7) and phosphatase (dp1) rates acting on JNK. Basal activity of JNK was introduced with the k_eq7 rate constant. For p38 involving reactions, two sets of NanoBit data on the ATF2-p38 complex—CTR and JNK-IN-8 treated—were used for the fitting of k_stim6, dp2 and dp4 with PyBioNetFit. Basal activity of p38 was introduced with the k_eq6 rate constant. These model parameters predicted that 5% of p38 gets activated after 40 min of anisomycin treatment. The phosphatase activity on T69/T71 (dp3) was determined using the pT69/pT71 Western-blot signal, assuming that 80% of ATF2 are phosphorylated after 40 min in the control experiment (CTR). In the case of JNK-IN-8 treatment, k1 and k2 were set to 0, and in the case of the S90N mutant, k_on2 was set 4.5x fold higher compared to WT, and k2 was set to 0 (see Supplementary Table 2). Parameter scanning with various k_stim6 and k_stim7 rates was performed using the final model and endogenous protein phosphorylation levels were calculated for 40 min after stimulation.

Reporting summary

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

Data availability

The data that support this study are available from the corresponding author upon reasonable request. The crystal structure of the JNK-ATF2(19–58) and the pp-p38α:ATF2(83–102) complex were deposited in the Protein Data Bank (PDB) with the accession code 6ZR5 and 6ZQS, respectively. The following X-ray structures are available from the PDB: 1BHI, 2XS0, 2WO6, 6TCA. Source data are provided with this paper.

Code availability

Computer-readable model implementation written in BNGL is provided as Supplementary Software 1.

References

Betts, M. J. et al. Systematic identification of phosphorylation-mediated protein interaction switches. PLoS Comput. Biol. 13, 1–20 (2017).
Article CAS Google Scholar
Lynch, V. J., May, G. & Wagner, G. P. Regulatory evolution through divergence of a phosphoswitch in the transcription factor CEBPB. Nature 480, 383–386 (2011).
Article ADS CAS PubMed Google Scholar
Whitmarsh, A. J. & Davis, R. J. Regulation of transcription factor function by phosphorylation. Cell. Mol. Life Sci. 57, 1172–1183 (2000).
Article CAS PubMed Google Scholar
Stevens, J. L. et al. Transcription control by E1A and MAP kinase pathway via Sur2 Mediator subunit. Science 296, 755–758 (2002).
Article ADS CAS PubMed Google Scholar
Mylona, A. et al. Opposing effects of Elk-1 multisite phosphorylation shape its response to ERK activation. Science 354, 233–237 (2016).
Article ADS CAS PubMed PubMed Central Google Scholar
Chen, R. H., Abate, C. & Blenis, J. Phosphorylation of the c-Fos transrepression domain by mitogen-activated protein kinase and 90-kDa ribosomal S6 kinase. Proc. Natl Acad. Sci. U. S. A. 90, 10952–10956 (1993).
Article ADS CAS PubMed PubMed Central Google Scholar
Smeal, T., Binetruy, B., Mercola, D. A., Birrer, M. & Karin, M. Oncogenic and transcriptional cooperation with Ha-Ras requires phosphorylation of c-Jun on serines 63 and 73. Nature 354, 56–58 (1991).
Article Google Scholar
Livingstone, C., Patel, G. & Jones, N. ATF-2 contains a phosphorylation-dependent transcriptional activation domain. EMBO J. 14, 1785–1797 (1995).
Article CAS PubMed PubMed Central Google Scholar
Hibi, M., Lin, A., Smeal, T., Minden, A. & Karin, M. Identification of an oncoprotein- and UV-responsive protein kinase that binds and potentiates the c-Jun activation domain. Genes Dev. 7, 2135–2148 (1993).
Article CAS PubMed Google Scholar
Gupta, S., Campbell, D., Dérijard, B. & Davis, R. J. Transcription factor ATF2 regulation by the JNK signal transduction pathway. Science 267, 389–393 (1995).
Article ADS CAS PubMed Google Scholar
Nagadoi, A. et al. Solution structure of the transactivation domain of ATF-2 comprising a zinc finger-like subdomain and a flexible subdomain. J. Mol. Biol. 287, 593–607 (1999).
Article CAS PubMed Google Scholar
Duyndam, M. C. A. et al. The N-terminal transactivation domain of ATF2 is a target for the co-operative activation of the c-jun promoter by p300 and 12S E1A. Oncogene 18, 2311–2321 (1999).
Article CAS PubMed Google Scholar
Watson, G., Ronai, Z. & Lau, E. ATF2, a paradigm of the multifaceted regulation of transcription factors in biology and disease. Pharmacol. Res. 119, 347–357 (2017).
Article CAS PubMed PubMed Central Google Scholar
Raingeaud, J., Whitmarsh, A. J., Barrett, T., Dérijard, B. & Davis, R. J. MKK3- and MKK6-regulated gene expression is mediated by the p38 mitogen-activated protein kinase signal transduction pathway. Mol. Cell. Biol. 16, 1247–1255 (1996).
Article CAS PubMed PubMed Central Google Scholar
Jacobs, D., Glossip, D., Xing, H., Muslin, A. J. & Kornfeld, K. Multiple docking sites on substrate proteins form a modular system that mediates recognition by ERK MAP kinase. Genes Dev. 13, 163–175 (1999).
Article CAS PubMed PubMed Central Google Scholar
Garai, Á. et al. Specificity of linear motifs that bind to a common mitogen-activated protein kinase docking groove. Sci. Signal. 5, ra74 (2012).
Lee, T. et al. Docking motif interactions in Map kinases revealed by hydrogen exchange mass spectrometry. Mol. Cell 14, 43–55 (2004).
Article CAS PubMed Google Scholar
Sheridan, D. L., Kong, Y., Parker, S. A., Dalby, K. N. & Turk, B. E. Substrate discrimination among mitogen-activated protein kinases through distinct docking sequence motifs. J. Biol. Chem. 283, 19511–19520 (2008).
Article CAS PubMed PubMed Central Google Scholar
Sharrocks, A. D., Yang, S. H. & Galanis, A. Docking domains and substrate-specificity determination for MAP kinases. Trends Biochem. Sci. 25, 448–453 (2000).
Article CAS PubMed Google Scholar
Wagner, E. F. & Nebreda, Á. R. Signal integration by JNK and p38 MAPK pathways in cancer development. Nat. Rev. Cancer 9, 537–549 (2009).
Article CAS PubMed Google Scholar
Laughlin, J. D. et al. Structural mechanisms of allostery and autoinhibition in JNK family kinases. Structure 20, 2174–2184 (2012).
Article CAS PubMed PubMed Central Google Scholar
Liu, H. et al. Mutual regulation of c-Jun and ATF2 by transcriptional activation and subcellular localization. EMBO J. 25, 1058–1069 (2006).
Article CAS PubMed PubMed Central Google Scholar
van Zundert, G. C. P. et al. The HADDOCK2.2 Web server: user-friendly integrative modeling of biomolecular complexes. J. Mol. Biol. 428, 720–725 (2015).
Article CAS PubMed Google Scholar
Edwards, D. R. & Mahadevan, L. C. Protein synthesis inhibitors differentially superinduce c-fos and c-jun by three distinct mechanisms: lack of evidence for labile repressors. EMBO J. 11, 2415–2424 (1992).
Article CAS PubMed PubMed Central Google Scholar
Rosser, E. M., Morton, S., Ashton, K. S., Cohen, P. & Hulme, A. N. Synthetic anisomycin analogues activating the JNK/SAPK1 and p38/SAPK2 pathways. Org. Biomol. Chem. 2, 142–149 (2004).
Article CAS PubMed Google Scholar
Bach, M. et al. Fast set-up of doxycycline-inducible protein expression in human cell lines with a single plasmid based on Epstein-Barr virus replication and the simple tetracycline repressor. FEBS J. 274, 783–790 (2007).
Article CAS PubMed Google Scholar
Morton, S., Davis, R. J. & Cohen, P. Signalling pathways involved in multisite phosphorylation of the transcription factor ATF-2. FEBS Lett. 572, 177–183 (2004).
Article CAS PubMed Google Scholar
Kumar, G. S. et al. Dynamic activation and regulation of the mitogen-activated protein kinase p38. Proc. Natl Acad. Sci. USA 115, 4655–4660 (2018).
Article CAS PubMed PubMed Central Google Scholar
Liu, X. et al. A conserved motif in JNK/p38-specific MAPK phosphatases as a determinant for JNK1 recognition and inactivation. Nat. Commun. 7, 10879 (2016).
Glatz, G., Gógl, G., Alexa, A. & Reményi, A. Structural mechanism for the specific assembly and activation of the extracellular signal regulated kinase 5 (ERK5) module. J. Biol. Chem. 288, 8596–8609 (2013).
Article CAS PubMed PubMed Central Google Scholar
Zeke, A., Misheva, M., Reményi, A. & Bogoyevitch, M. A. JNK Signaling: regulation and functions based on complex protein-protein partnerships. Microbiol. Mol. Biol. Rev. 80, 793–835 (2016).
Article CAS PubMed PubMed Central Google Scholar
Piserchio, A. et al. Structural and dynamic features of F-recruitment site driven substrate phosphorylation by ERK2. Sci. Rep. 5, 11127 (2015).
Sano, Y. et al. Drosophila activating transcription factor-2 is involved in stress response via activation by p38, but not c-Jun NH(2)-terminal kinase. Mol. Biol. Cell 16, 2934–2946 (2005).
Article CAS PubMed PubMed Central Google Scholar
Shiozaki, K. & Russell, P. Conjugation, meiosis, and the osmotic stress response are regulated by Spc1 kinase through Atf1 transcription factor in fission yeast. Genes Dev. 10, 2276–2288 (1996).
Article CAS PubMed Google Scholar
Hazzalin, C. A. & Mahadevan, L. C. MAPK-Regulated transcription: a continuously variable gene switch? Nat. Rev. Mol. Cell Biol. 3, 30–40 (2002).
Article CAS PubMed Google Scholar
Reimold, A. M. et al. Chondrodysplasia and neurological abnormalities in ATF-2-deficient mice. Nature 379, 262–265 (1996).
Article ADS CAS PubMed Google Scholar
Kim, E. K. & Choi, E. J. Compromised MAPK signaling in human diseases: an update. Arch. Toxicol. 89, 867–882 (2015).
Article CAS PubMed Google Scholar
Lee, S., Rauch, J. & Kolch, W. Targeting MAPK signaling in cancer: mechanisms of drug resistance and sensitivity. Int. J. Mol. Sci. 21, 1102 (2020).
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
Lee, W., Tonelli, M. & Markley, J. L. NMRFAM-SPARKY: Enhanced software for biomolecular NMR spectroscopy. Bioinformatics 31, 1325–1327 (2015).
Article PubMed Google Scholar
Evenäs, J. et al. Ligand-induced structural changes to maltodextrin-binding protein as studied by solution NMR spectroscopy. J. Mol. Biol. 309, 961–974 (2001).
Article CAS PubMed Google Scholar
Marsh, J. A., Singh, V. K., Jia, Z. & Forman-Kay, J. D. Sensitivity of secondary structure propensities to sequence differences between α- and γ-synuclein: Implications for fibrillation. Protein Sci. 15, 2795–2804 (2006).
Article CAS PubMed PubMed Central Google Scholar
Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development of Coot. Acta Crystallogr. Sect. D. Biol. Crystallogr. 66, 486–501 (2010).
Article CAS Google Scholar
Soundararajan, M. et al. Structures of down syndrome kinases, DYRKs, reveal mechanisms of kinase activation and substrate recognition. Structure 21, 986–996 (2013).
Article CAS PubMed PubMed Central Google Scholar
Tzarum, N., Eisenberg-Domovich, Y., Gills, J. J., Dennis, P. A. & Livnah, O. Lipid molecules induce p38α activation via a novel molecular switch. J. Mol. Biol. 424, 339–353 (2012).
Article CAS PubMed Google Scholar
Tzarum, N., Komornik, N., Chetrit, D., Ben, Engelberg, D. & Livnah, O. DEF pocket in p38α facilitates substrate selectivity and mediates autophosphorylation. J. Biol. Chem. 288, 19537–19547 (2013).
Article CAS PubMed PubMed Central Google Scholar
Sali, A., Blundell, T., Sali, A. & Blundell, T. L. Comparative modelling by satisfaction of spatial restraints. J. Mol. Biol. 234, 779–815 (1994).
Article Google Scholar
McCoy, A. J. et al. Phaser crystallographic software. J. Appl. Crystallogr. 40, 658–674 (2007).
CAS PubMed PubMed Central Google Scholar
Adams, P. D. et al. The Phenix software for automated determination of macromolecular structures. Methods 55, 94–106 (2011).
Article CAS PubMed PubMed Central Google Scholar
Zeke, A. et al. Systematic discovery of linear binding motifs targeting an ancient protein interaction surface on MAP kinases. Mol. Syst. Biol. 11, 837 (2015).
Article CAS PubMed PubMed Central Google Scholar
Dixon, A. S. et al. NanoLuc complementation reporter optimized for accurate measurement of protein interactions in cells. ACS Chem. Biol. 11, 400–408 (2016).
Article CAS PubMed Google Scholar
Harris, L. A. et al. BioNetGen 2.2: Advances in rule-based modeling. Bioinformatics 32, 3366–3368 (2016).
Article CAS PubMed PubMed Central Google Scholar
Mitra, E. D. et al. PyBioNetFit and the biological property specification language. iScience 19, 1012–1036 (2019).
Article ADS PubMed PubMed Central Google Scholar
Waas, W. F., Lo, H. H. & Dalby, K. N. The kinetic mechanism of the dual phosphorylation of the ATF2 transcription factor by p38 mitogen-activated protein (MAP) kinase alpha. Implications for signal/response profiles of MAP kinase pathways. J. Biol. Chem. 276, 5676–5684 (2001).
Article CAS PubMed Google Scholar

Download references

Acknowledgements

The authors thank Marie Bogoyevitch for her support and valuable discussions. This work was supported by the National Research Development and Innovation Office (NKFIH) grants (NN 114309, KKP 126963 awarded to A.R.), the Hungarian of Academy of Sciences (KEP-10/2019), and by a National Institute of Health (NIH) grant 1R01GM100910 to W.P.

Author information

Authors and Affiliations

Biomolecular Interactions Research Group, Institute of Organic Chemistry, Research Center for Natural Sciences, H-1117, Budapest, Hungary
Klára Kirsch, András Zeke, Péter Sok, Anna Sebő, Péter Egri, Ádám L. Póti, Anita Alexa & Attila Reményi
Laboratory for NMR Spectroscopy, Research Center for Natural Sciences, H-1117, Budapest, Hungary
Orsolya Tőke
Department of Biochemistry and Molecular Biology and Bio21 Molecular Science and Biotechnology Institute, University of Melbourne, Parkville, VIC, 3010, Australia
Ashish Sethi & Paul Gooley
Department of Chemistry and Biochemistry, University of Arizona, Tucson, AZ, USA
Ganesan Senthil Kumar & Wolfgang Peti
European Molecular Biology Laboratory, Hamburg, Germany
Isabel Bento

Authors

Klára Kirsch
View author publications
You can also search for this author in PubMed Google Scholar
András Zeke
View author publications
You can also search for this author in PubMed Google Scholar
Orsolya Tőke
View author publications
You can also search for this author in PubMed Google Scholar
Péter Sok
View author publications
You can also search for this author in PubMed Google Scholar
Ashish Sethi
View author publications
You can also search for this author in PubMed Google Scholar
Anna Sebő
View author publications
You can also search for this author in PubMed Google Scholar
Ganesan Senthil Kumar
View author publications
You can also search for this author in PubMed Google Scholar
Péter Egri
View author publications
You can also search for this author in PubMed Google Scholar
Ádám L. Póti
View author publications
You can also search for this author in PubMed Google Scholar
Paul Gooley
View author publications
You can also search for this author in PubMed Google Scholar
Wolfgang Peti
View author publications
You can also search for this author in PubMed Google Scholar
Isabel Bento
View author publications
You can also search for this author in PubMed Google Scholar
Anita Alexa
View author publications
You can also search for this author in PubMed Google Scholar
Attila Reményi
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

K.K. designed and carried out the experiments, analyzed data and, wrote the paper. A.R. conceived the project, analyzed data, supervised research and wrote the paper. A.Z. mapped out the JNK binding region of ATF2. O.T. analyzed the structure of ATF2 TAD fragments, Ashish S./P.G. characterized JNK-ATF2 binding, and O.T./G.S.K./W.P. elucidated pp-p38-ATF2 binding by NMR spectroscopy. I.B. collected X-ray data. P.S., Anna S., P.E., Á.L.P., and A.A. contributed by carrying out some of the biochemical or cell-based experiments.

Corresponding author

Correspondence to Attila Reményi.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Peer review information Nature Communications thanks Elizabeth Goldsmith and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

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

Description of Additional Supplementary Files

Supplementary Software 1

Reporting Summary

Source data

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

Kirsch, K., Zeke, A., Tőke, O. et al. Co-regulation of the transcription controlling ATF2 phosphoswitch by JNK and p38. Nat Commun 11, 5769 (2020). https://doi.org/10.1038/s41467-020-19582-3

Download citation

Received: 30 July 2020
Accepted: 21 October 2020
Published: 13 November 2020
DOI: https://doi.org/10.1038/s41467-020-19582-3

This article is cited by

The recombinant l-lysine α-oxidase from the fungus Trichoderma harzianum promotes apoptosis and necrosis of leukemia CD34 + hematopoietic cells
- Mariana do Nascimento Costa
- Thiago Aparecido Silva
- Roberto N. Silva
Microbial Cell Factories (2024)
Mycobacterium tuberculosis inhibits METTL14-mediated m6A methylation of Nox2 mRNA and suppresses anti-TB immunity
- Mingtong Ma
- Yongjia Duan
- Baoxue Ge
Cell Discovery (2024)
Investigation of the phytochemical composition, antioxidant, antibacterial, anti-osteoarthritis, and wound healing activities of selected vegetable waste
- Mohamed A. Salem
- Osama G. Mohamed
- Hend E. Abo Mansour
Scientific Reports (2023)
The two-faced role of ATF2 on cisplatin response in gastric cancer depends on p53 context
- Lingxue Xu
- Jingjing Wang
- Zhenguo Cheng
Cell & Bioscience (2022)
Decisive role of water and protein dynamics in residence time of p38α MAP kinase inhibitors
- Tatu Pantsar
- Philipp D. Kaiser
- Stefan A. Laufer
Nature Communications (2022)

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.