The TAB1-p38α complex aggravates myocardial injury and can be targeted by small molecules

Abstract

Inhibiting MAPK14 (p38α) diminishes cardiac damage in myocardial ischemia. During myocardial ischemia, p38α interacts with TAB1, a scaffold protein, which promotes p38α autoactivation; active p38α (pp38α) then transphosphorylates TAB1. Previously, we solved the X-ray structure of the p38α-TAB1 (residues 384-412) complex. Here, we further characterize the interaction by solving the structure of the pp38α-TAB1 (residues 1-438) complex in the active state. Based on this information, we created a global knock-in (KI) mouse with substitution of 4 residues on TAB1 that we show are required for docking onto p38α. Whereas ablating p38α or TAB1 resulted in early embryonal lethality, the TAB1-KI mice were viable and had no appreciable alteration in their lymphocyte repertoire or myocardial transcriptional profile; nonetheless, following in vivo regional myocardial ischemia, infarction volume was significantly reduced and the transphosphorylation of TAB1 was disabled. Unexpectedly, the activation of myocardial p38α during ischemia was only mildly attenuated in TAB1-KI hearts. We also identified a group of fragments able to disrupt the interaction between p38α and TAB1. We conclude that the interaction between the 2 proteins can be targeted with small molecules. The data reveal that it is possible to selectively inhibit signaling downstream of p38α to attenuate ischemic injury.

Keywords: Cardiology; Pharmacology; Protein kinases; Structural biology; Therapeutics.

Figure 1. Thermodynamic characterization of the interaction between p38α and TAB1 in their phosphorylated and nonphosphorylated forms.

Top panels: ITC raw data of the interaction between dual (pThr180, pTyr182) phosphorylated p38α (pp38α) and TAB1 in mono (pSer423) phosphorylated (A) and native forms (B). Bottom panels: all but the first titration point were used for the curve fitting.

Figure 2. X-ray structure of the dual phosphorylated pp38α-TAB1 (residues 1–438) complex.

TAB1 is shown in ribbon format (cyan) and pp38α in ribbon format (magenta) with a transparent surface overlay. The TAB1 residues belonging to the pseudo PP2 domain of TAB1 (residues 15–370) and the residues interacting with pp38α are visible. The N- (residues 1–14) and C- (residues 413–438) terminal regions, the residues (residues 371–383) between the pseudo PP2 domain and the p38α binding region and the residues (residues 396–402) between the canonical and noncanonical site are disordered and not visible in the X-ray structure. The conclusion is that no other region beyond residues 384-412 of TAB1 interacts with p38α, confirming the ITC data of thermodynamically similar binding characteristics between TAB1 (residues 1–438) and TAB1 (residues 384–412).

Figure 3. Comparison of p38α bound to TAB1 peptide (residues 384–412, PDB 4LOO) and pp38α bound to TAB1 protein (residues 1–438, PDB 5NZZ).

Activation of p38α induces a major shift in the conformation of the activation loop and the N-terminal domain rotates approximately 11° toward the C-terminal domain creating a more closed active site typical of activated kinases. The rest of the C-terminal domain of p38α and the p38α interacting region of TAB1 align closely, which is consistent with the observation that the affinity of TAB1 for p38α and pp38α is virtually identical.

Figure 4. Comparison of p38α and pp38α activation loops in the complexes.

This comparison illustrates the reorientation of Y182 and movement of T180 away from the ATP binding site upon phosphorylation. For clarity, only the ATP-γ-S ligand of the pp38α structure is shown.

Figure 5. Characterizing the interaction between p38α and wild type (WT) and mutated forms of TAB1 by co-expression.

(A) Overexpression of p38α and TAB1 in E.Coli. Coexpression of TAB1 and p38α increases p38α and TAB1 phosphorylation. Mutation of each of the individual recognition sites in TAB1 diminishes these phosphorylations with the strongest effect apparent when both sites are mutated. (B) Similar results are obtained with overexpression of p38α and TAB1 in HEK293. (C) Overexpression in HEK293 cells of p38α, TAB1, and MKK3b in an attempt to normalize the levels of p38α activation. The data show that the recognition regions within TAB1 that are needed to autoactivate p38α are also those used to present TAB1 as a p38α substrate. (D) Stopping TAB1 from docking onto p38α does not affect the ability of TAB1 to activate TAK1. One-way ANOVA was used for the statistical analysis, *P < 0.05, **P < 0.01 vs. p38α/TAB1 WT.

Figure 6. Creation of the TAB1 (V390A,Y392A,V408G and M409G) KI mouse.

(A) TAB1 protein sequence, the targeted residues are shown in red. (B) Targeting strategy to produce the TAB1 (V390A, Y392A, V408G, and M409A) KI mouse. Schematic representation of the WT allele, targeted allele, and the constitutive KI allele after Cre recombination. (C) Southern blot of WT and 2 targeted ES cell clones (denoted as 2C9 and 2E10) with 3 probes. The expected molecular weight band for WT and the targeted allele are shown. (D) PCR of the targeted allele using genomic DNA derived from tail snips. (E) Sequencing of the genomic DNA of the targeted allele confirming the 4 single point mutations.

Figure 7. (A) Growth curves comparing within colony WT and KI mice.

Mouse weights taken between 5 and 12 weeks of age from each genotype were analyzed using 2-way repeated measures ANOVA. Weights are not significantly different. (B) Immunoblot analysis of total TAB1 expressed in organs of WT vs. KI mice. (C) Volcano plot showing minimal changes of the transcriptome profile between WT and KI hearts. (D) Immune phenotype characterization. Each data point represents a cell subset. MLN, mesenteric lymph nodes; SPLN,spleen. The position of the data point on the x axis represents how the mean size of that subset changes in the KI relative to the WT. The position of each data point on the y axis represents the 1/P value. The dotted horizontal line represents the Bonferroni adjusted threshold for statistical significance. None of the populations are statistically different between KI and WT mice.

Figure 8. (A) Immunoblot analysis of heart homogenates taken from hearts subjected to regional myocardial ischemia at 5 and 10 minutes.

(B) Quantification of phosphorylated p38α normalized against total p38α. WT control vs. WT and control vs. KI at 5 minutes, ***P < 0.001; WT control vs. WT at 10 minutes, *P < 0.05. (C) Phosphorylated TAB1 normalized against total TAB1. WT control vs. WT 10 minutes, ***P < 0.001; WT vs. KI at 10 minutes, ††††P < 0.0001. As previously reported in the literature, ischemia causes p38α activation and subsequent TAB1 phosphorylation in the WT hearts. In the KI heart, TAB1 phosphorylation is almost completely abolished during ischemia, despite the nearly equivalent activation of p38α.

Figure 9. Myocardial infarction in TAB1 WT and KI mice, and competition between TAB1 and MKK3b on p38α.

(A) Myocardial infarction volume (INF) as a percentage of the risk zone (RZ) in the area at risk (AAR) of the left ventricle (LV) is significantly reduced in KI mice after 30 minutes of occlusion of the left anterior descending coronary artery and 120 minutes of reperfusion (29.4% ± 2.0% vs. 22.2% ± 1.4%, n = 12, **P < 0.01). (B) Infarction area plotted against RZ area measured on heart slides as described in Methods confirming the decreased susceptibility to ischemic injury in KI mice. P < 0.05 vs. WT. Fluorescence polarization titration revealing the displacement of fluorescein-labeled MKK3b by unlabeled MKK3b (C) or TAB1 (D). One-way ANOVA and 1-way ANCOVA used.

Figure 10. TAB1 and 3-amino-1-adamantanol compete for the same hydrophobic pocket on the noncanonical site of p38α.

(A) 3-Amino-1-adamantanol binding site. The hydrophobic pocket on p38α is shown as a semitransparent surface, the 3-amino-1-adamantanol molecule is shown in magenta. The hydrogen bonds between the 3-amino group and the hydroxyl group of the adamantane and the oxygens of the backbone carbonyl of Leu 222 and Leu 234 are shown as dotted lines. The van der Waals surface of the adamantanol is shown as a mesh in magenta. (B) Zoom-in of the noncanonical site in pp38α-TAB1 complex, the hydrophobic pocket used by the 3-amino-1-adamantanol is shown in identical representation and orientation as in A. TAB1 is shown in ribbon form (cyan) with the side chains of Arg 384, Val 385, and Tyr 386 shown. (C) Representative fluorescence based thermal shift assay of p38α (5 μM) with DMSO baseline control (green) vs. 12.5 mM 3-amino-1-adamantanol (orange) and DMSO + 50μM SB220025 (blue) vs. 12.5 mM 3-amino-1-adamantanol + 50μM SB220025 (red). Inset: Box and whisker representation indicating that 3-amino-1-adamantanol binding is not competitive to the high-affinity ATP competitive inhibitor SB220025. (D) Western blot analysis of pp38α-TAB1 IVKA in the presence and in the absence of 12.5 mM 3-amino-1-adamantanol. It shows that the ligand prevents pp38α phoshorylation of TAB1.