The MAP kinases, ERK1 and ERK2, are key effectors in the MAP kinase cascade, a signaling pathway downstream of RAS that is essential for cell proliferation, differentiation, motility, and survival.27, 55 ERKs are activated by dual phosphorylation of specific threonine and tyrosine residues on the activation loop (A-loop), both catalyzed by upstream MAP kinase kinases 1 and 2 (MKK1/2 aka MEK1/2). MKK1/2 in turn are activated by members of the RAF family of protein kinases in all cells, and by c-MOS in germ cells. The prevalence of oncogenic mutations in RAS and RAF has motivated the successful development of inhibitors towards B/C-RAF and MKK1/2 for the treatment of melanomas and other cancers. Preclinical outcomes show that ERK inhibitors are active towards cancer cells and tumors that are resistant to RAF or MKK inhibitors.17, 38 Therefore, ERK is an important target whose mechanisms of activation are important to understand.51

X-ray crystallographic studies of the phosphorylated (2P) and unphosphorylated (0P) states of ERK2 have provided a framework for understanding structural changes associated with kinase activation8, 69 (Figure 1). The largest conformational change occurs in the activation loop (A-loop), which contains the phosphorylation sites. Remodeling of the A-loop results in salt bridge interactions between pT183 and pY185 (rat ERK2 numbering throughout) and multiple Arg residues in the kinase N- and C-lobes (Figure 1A). The reorientation of pY185 opens a proposed recognition site for proline-directed sequence motifs in ERK substrates,8 and a rearrangement of residues N-terminal to the phosphorylation sites (F181, L182) exposes a C-lobe binding site for a hydrophobic docking motif (“DEF”) found in ERK substrates and effectors.29

Despite these changes in the A-loop, structural differences between the active sites in the crystal structures of 0P-ERK2 and 2P-ERK2 are minimal. This contrasts with other protein kinases, where X-ray structures reveal significant conformational shifts that commonly accompany the switch from active to inactive states.60 These include rotation of helix αC and consequent disruption of a critical Lys-Glu salt bridge (K52-E69 in ERK2) that coordinates phosphate oxygens in ATP; a “DFG flip” backbone rotation that buries a catalytic Asp residue (D165 in ERK2) needed for Mg2+ coordination; and disrupted alignments of regulatory-spine (R-spine) and catalytic-spine (C-spine) residues involved in nucleotide binding and phosphoryltransfer.12, 24, 34, 66 Oddly, the positions of these active site residues are largely invariant between the crystal structures of the active 2P and inactive 0P forms of ERK2. Thus, ERK2 can be considered a prototype to investigate regulatory mechanisms in kinases that do not display substantial conformational rearrangements at the active site.

Solution measurements have revealed changes in protein dynamics following ERK2 phosphorylation and activation. Studies using hydrogen–deuterium exchange mass spectrometry (HX-MS) showed that phosphorylation of ERK2 altered the rates of deuterium uptake in localized regions where X-ray structures were invariant.19 NMR Carr-Purcell-Meiboom-Gill (CPMG) relaxation dispersion measurements of [methyl-13C,1H]-Ile, Leu and Val residues in ERK2 revealed that activation by phosphorylation led to global exchange behavior within the N- and C-lobes and surrounding the active site. This exchange was modeled by an equilibrium between two energetically similar conformational states, named “L” and “R”, that interconvert on a millisecond timescale.64 Importantly, residues in the ERK2 A-loop were included in the global exchange, and mutations in the A-loop blocked formation of the R-state. These results suggest allosteric coupling between the A-loop and residues surrounding the active site.21 Furthermore, different ATP-competitive inhibitors of ERK2 displayed conformational selection for the L and R states, shifting the L⇌R equilibrium in opposite directions.43, 49 These inhibitors induced changes in HX protection within the P+1 segment adjoining the A-loop, confirming coupling from the active site to the A-loop.43 Together, the results revealed an allosteric mechanism in 2P-ERK2, where the A-loop is not found in a single state but instead interconverts between two or more discrete states that are in turn coupled to motions at the active site. The nature of these states and how they contribute to ERK2 activation are unknown.

Inspired by this solution-phase evidence for a role of dynamics in ERK2 activation, we applied long conventional molecular dynamics simulations to characterize potential motions in ERK2 and their structural framework. The results show multiple long-lived conformations of the A-loop that have not previously been observed by crystallography. Notably, simulations of 2P-ERK2 showed settled conformations of the A-loop that formed variable interactions with the N-lobe and C-lobe, and/or altered the salt-bridges formed by the phosphorylated residues. Simulations of 0P-ERK2 showed new settled states of the A-loop that exposed the Y185 phosphorylation site to solvent. Difference contact network analysis, principal component analysis, and RMSF calculations revealed that movements of the A-loop alter the dynamics of the kinase core and active site residues. The states of 2P-ERK2 were correlated with reduced dynamics and greater compactness of the N-lobe and active site, while states of 0P-ERK2 showed greater N-lobe motions and active site disorganization. The results reveal unexpected flexibility of the A-loop, and a role of the varying conformational states for regulating active site dynamics in a phosphorylation-dependent manner.