Revealing eEF-2 kinase: Recent structural insights into function

Abstract

The α–kinase eukaryotic elongation factor 2 kinase (eEF-2K) regulates translational elongation by phosphorylating its ribosome-associated substrate, a guanosine triphosphatase, eEF-2. eEF-2K is activated by calmodulin through a distinctive mechanism unlike that in other calmodulin-dependent kinases. Here, we describe recent structural insights into this unique activation process and examine the effects of specific regulatory signals on this mechanism. Additionally, we highlight critical unanswered questions to guide future structure-function studies. These include structural mechanisms which enable eEF-2K to interact with upstream/downstream partners and facilitate its integration of diverse inputs, including Ca²⁺ transients, phosphorylation mediated by energy/nutrient-sensing pathways, pH changes, and metabolites. Answering these questions is key to establishing how eEF-2K harmonizes translation with cellular requirements within the boundaries of its molecular landscape.

Whence came eEF-2K?

In 1983, End et al. [1] discovered the phosphorylation of a 100-kDa protein in PC-12 cell extracts and noted that this covalent modification was sensitive to nerve growth factor stimulation. Subsequent work by Togari and Guroff led to the partial purification and characterization of the kinase responsible [2]. This kinase was subsequently purified to homogeneity from rat pancreatic cells by Nairn et al., who demonstrated its reliance on calmodulin (CaM) for activity [3]. Christened CAMKIII (calmodulin-dependent kinase III, based on the order of fractionation from rat brain extracts [4]), it was found to be ubiquitous in mammalian cells, and its 100-kDa substrate was identified as the ribosome-associated guanosine triphosphatase (GTPase), eukaryotic elongation factor 2 (eEF-2) [3, 5]. Now named eukaryotic elongation factor 2 kinase (eEF-2K), this kinase was shown to lack [6] several conserved sequence features [7] characteristic of “conventional” eukaryotic protein kinases (cPKs; see Glossary), positioning it as a member of a novel grouping of atypical protein kinases, the “α–kinase” superfamily [8]. Though pervasive in eukaryotes, vertebrate and invertebrate genomes encode distinct α–kinases [9]; eEF-2K orthologs, however, are found in both groups (Box 1). This suggests a relatively ancient origin of eEF-2K, perhaps serving as a hub for differentiation within the α–kinase superfamily. Notably, eEF-2K is the only known CaM-dependent α–kinase.

BOX 1. The omnipresent eEF-2K.

eEF-2K orthologs are found in most eukaryotes, encoded in both invertebrate and vertebrate genomes (Figure IA), contrasting other members of the α–kinase superfamily that are phylum/subphylum-specific [9]. Curiously, eEF-2K orthologs are absent in insects and fungi [75], suggesting a somewhat unclear evolutionary history.

eEF-2K consists of an N-terminal calmodulin targeting motif (CTM), connected through a regulatory element (RE) to an α-kinase domain (KD). The C-terminal domain (CTD), comprising multiple helical repeats, houses the binding site for the substrate eEF-2 [68, 76]. Despite significant species-specific sequence variability, this domain structure appears to be conserved in all eEF-2K orthologs. Not surprisingly, the KD, which encodes the catalytic elements, is the least variable. The most significant variability is contained within the disordered loops on the extreme N-terminus and that bridging the KD and the CTD (Figure IB). The latter, termed the regulatory loop (R-loop), contains several sites of regulatory phosphorylation, including the primary site of activating autophosphorylation (T348) [47, 67]. Additional R-loop phosphorylation sites include S359, targeted by p38δ [52], S366 targeted by the p70 S6 kinase and p90^RSK1 [53], S398 targeted by AMPK [51], and S500, an autophosphorylation site, also targeted by PKA [66, 67], to name a few. The N-terminus also contains phosphorylation sites, including S78, reliant on the nutrient-sensing mTOR pathway [29]. Some of these sites are universally conserved, e.g., T348 (a serine in rare cases, e.g., in the nematode Seteria digitata); others, e.g., S500, are lost in some invertebrates. Yet, others, e.g., S78 or S398, show significant variability. Indeed, more than two dozen sites of post-translational modification, including those proposed to be ubiquitinated [77], SUMOylated [78], or hydroxylated [49], distributed throughout the sequence, have been suggested from detailed studies or high-throughput analyses. The biological significance of many of these sites remains to be determined. Figure IB shows a few of these sites in human eEF-2K. Phosphorylation sites that activate kinase function are shown (and labeled) in blue, those that lead to deactivation are in red, and those undiscernible/unknown effects are in purple. P98 is a site of deactivating hydroxylation.

Figure I

The absence of structural information hindered efforts to obtain mechanistic insights into eEF-2K function. Identification of the minimal CaM-activatable core of eEF-2K (eEF-2K_TR, Box 2) [10] and the recent solving of a series of structures of its complex with CaM in various functional states [11–13] provided, for the first time, a structural framework for the exploration of functional mechanisms.

BOX 2. Not a typical eukaryotic kinase.

A construct, eEF-2K_TR (Figure IIA) that lacks several regions predicted to be disordered, including the N-terminus, and a portion of the R-loop (replaced by a 6-glycine linker; ΔR-loop), represents the minimal functional core of eEF-2K. eEF-2K_TR is similarly activated by CaM as the full-length enzyme and efficiently phosphorylates eEF-2 in cells [10]. eEF-2K_TR contains several key regulatory phosphorylation sites, including the primary and secondary stimulatory (P↑) sites, T348 and S500, respectively, and the inhibitory hydroxylation site, P98 (OH↓). Given the absence of a substantial R-loop segment and several phosphorylation sites therein, eEF-2K_TR lacks much of the secondary regulatory capacity of the full-length enzyme.

Despite lacking many of the conserved sequence features of Hanks-type cPKs [7], the eEF-2K KD (Figure IIB) displays a characteristic dual-lobed structure [11–13]. While the N-terminal lobe of the eEF-2K KD somewhat matches those of cPKs, the C-terminal lobe diverges significantly, and has been suggested to resemble ATP-grasp domains [79]; the extent of this similarity [80], and the supposed evolutionary linkage is not immediately evident [81]. The “catalytic” residues are largely conserved (Figure IIC; shown with pink sidechains and labeled using corresponding colors in the top right inset of Figure IIB) in the other human α-kinases (except ALPK3, likely a pseudo-kinase without catalytic activity [82]) and in the related Dictyostelium discoideum myosin heavy chain kinase A (dsMHCK-A) [48, 83]. The putative catalytic base, D274 (D166 in PKA), and the essential Mg²⁺-coordinating, Q276 (N171 in PKA), are structurally equivalent to conserved cPK residues. K170 (K72 in PKA) and D184 (E91 in PKA) are in similar spatial positions as their conventional counterparts where a salt bridge linking them stabilizes the “αC-in” configuration in the active state [84]. These residues appear to play different role in α–kinases including eEF-2K [48, 83]. In addition to a phosphate-binding loop (P-loop), eEF-2K (like other α–kinases) contains an “N/D-loop” suggested to facilitate substrate recognition [85]. The N/D-loop, like the “activation-loop” of cPKs [84], carries a DFG motif that is highly conserved in vertebrate eEF-2Ks. However, the role of this motif remains unclear at present.

Residues equivalent to D284 (and D184) are phosphorylated in MHCK-A (Figure IIC) [48, 83]; the biological implications of these modifications, if any, remain unknown. The KDs of α–kinases, including eEF-2K, contain a characteristic structural (Cys)₂-(His)₂ Zn²⁺-finger, representing another point of divergence from the cPKs.

Figure II

Many excellent reviews highlight the multi-faceted cellular regulation of eEF-2K, e.g., [14, 15] and its dysregulation in human disease, e.g., [16–18]; our goal is not to revisit these. Rather, in this review, we utilize recent structural studies to infer the unique mechanism of the CaM-mediated activation of eEF-2K that distinguishes it from other CaM-dependent kinases (CAMKs). While eEF-2K activity is regulated through numerous inputs (e.g., reviewed in [14, 15]), we highlight only those for which insights can be obtained using current structures. Further, we identify critical mechanistic questions that existing structures are yet to address. We expect these to form the conceptual foundation for future integrated explorations, encompassing structural, biophysical, computational, and biochemical analyses, to unravel the mysteries of eEF-2K function.

A master regulator of translational elongation

Currently, eEF-2 represents the sole confirmed cellular target of eEF-2K. The eEF-2K/eEF-2 nexus is critical in regulating the elongation stage of protein synthesis [19]. Briefly, after binding the 80S ribosome, eEF-2 hydrolyzes GTP to induce translocation of the nascent chain from the ribosomal aminoacyl-site (A-site) to the peptidyl-site (P-site), creating space for an incoming aminoacyl-tRNA [20]. eEF-2K phosphorylates eEF-2 at Thr56, compromising its ability to engage the ribosome, thereby suppressing elongation [21, 22]. However, despite this global suppression of elongation, the translation of some proteins appears to be selectively enhanced [23]. This differential protein expression induced by eEF-2K activity has been linked to physiological processes such as modifications in synaptic strength or dendritic spine morphology [24, 25]. Recently, it has been suggested that pThr56-EF-2 can bind to and stabilize inactive ribosomes, thereby hindering the recycling process [26]. Therefore, there appears to be a versatile and multi-faceted eEF-2K-centric strategy to regulate protein synthesis through eEF-2.

Protein translation is a highly energy- [27] and nutrient-consumptive [28] process that demands tight coupling to the availability of these vital life elements to maintain optimal cellular function and adaptability. Unsurprisingly, eEF-2K is pivotal in this delicate balance by integrating with key signaling pathways, like mTOR [29] and AMPK [30] responsible for sensing nutrient and energy fluctuations. This regulation is also vital in maintaining cellular homeostasis and minimizing undesirable consequences like cotranslational misfolding and aggregation, which can arise when translation rates deviate from their optimal levels [31]. Indeed, harmonizing translational elongation through precisely regulated eEF-2K activity is essential for overall cellular health, ensuring its ability to respond to environmental cues and avoid potential protein-related problems.

Not just another CAMK

As noted above, eEF-2K is reliant on CaM for activity. However, in a mechanistic sense, it is not a “conventional” CAMK. The activation of conventional CAMKs occurs through variations on a “release-of-inhibition” theme (reviewed in Swulius and Waxham [32]) that may be summarized as follows: under basal conditions, an autoinhibitory segment containing an overlapping calmodulin-targeting motif (CTM) restricts substrate access and/or maintains the catalytic site in an inactive conformation; following increased Ca²⁺ levels, efficient engagement of the CTM by CaM enables kinase activation by releasing these inhibitory constraints. Deleting the CTM in a conventional CAMK yields a constitutively active kinase that no longer relies on CaM [33]. While eEF-2K contains an N-terminal CTM (Box 1, Figure IB), its removal leads to an inactive enzyme [34] suggesting a unique CaM-mediated activation mechanism (described below) distinct from conventional CAMKs.

An intimate association

A peptide, eEF-2K_CTM, encompassing the eEF-2K CTM, engages CaM with only ~15-fold less affinity than the full-length kinase, suggesting that this segment encodes a major CaM recognition element. The NMR structure of the CaM•eEF-2K_CTM complex [35] highlights a helical CTM bound to the hydrophobic face of the C-terminal lobe of CaM (CaM_C). W85, an invariant CTM residue that serves as the primary anchor for this interaction, is buried deep within the hydrophobic face of CaM_C. The importance of this residue in CaM recognition is evident from observations that its mutation substantially reduces CaM binding and diminishes eEF-2K activity in vitro [11]. Further, cells expressing W85 variants show reduced eEF-2 phosphorylation under conditions known to stimulate eEF-2K activity compared to those expressing wild-type eEF-2K [35]. Additional interactions involving I89 (a long-chain hydrophobic residue in vertebrates) and A92 (invariant in vertebrates) complete a 1–5-8 mode of engagement with CaM_C (Figure 1A, top left). This binding mode is maintained in the CaM•eEF-2K_TR complex, with little, if any, changes in the interactions of W85. However, a small shift of the CTM helix by about half a turn slightly alters its orientation within CaM_C (Figure 1A, top right). In the context of eEF-2K_TR, this CTM-driven interaction is reinforced by a second interaction interface involving the hydrophilic face of CaM_C and the N-lobe (KD_N) of the eEF-2K kinase domain (KD, Figure 1A, middle). These two sets of complementary interactions create an intimate association between CaM_C and eEF-2K_TR (Figure 1B). This binding mode is fully consistent with protection data obtained from hydrogen/deuterium exchange mass spectrometry (HXMS) measurements [36].

Figure 1. Calmodulin recognition by eEF-2K.

(A) 1-5-8 mode of hydrophobic interaction between the calmodulin-targeting motif (CTM) of eEF-2K (eEF-2K_CTM) and the calmodulin (CaM) C-lobe (CaM_C) in the NMR structure of the CaM•eEF-2K_CTM complex (top left; PDB:5J8H). W85 on the CTM represents the principal anchor (1-position); I89 (5-position) and A92 (8-position) are the other key residues. This binding mode is retained in the CaM•eEF-2K_TR complex (top right; PDB:8FNY), though with a slight realignment of the CTM within CaM_C. This interaction is supplemented by the formation of a second interface (middle panel) involving mostly hydrophilic interactions between CaM_C (using the face opposite that hosting the CTM) and the N-lobe of the kinase domain (KD) of eEF-2K. (B) The two distinct interfaces through which eEF-2K engages CaM are shown schematically. The interaction with the CTM occurs through the hydrophobic face of CaM_C (green). The second interface involves mostly hydrophilic interactions (magenta) between CaM_C and KD_N. (C) Schematic illustration of a possible two-step kinetic mechanism of CaM recognition by eEF-2K. In Step 1, the CTM binds CaM_C through hydrophobic interactions, resulting in a tethered complex. The enhanced local concentration of CaM near the KD facilitates additional interactions with CaM_C, leading to a collapsed complex (Step 2; a change in KD conformation is suggested) encoding global conformational features of the active state. Dynamics within the loops and the CaM lobes (CaM_N is colored light yellow; other elements are colored as in A) are indicated by the dotted lines and bent arrows, respectively.

A tether and a collapse

The recognition of eEF-2K by CaM may be imagined as comprising two distinct steps (Figure 1C). In Step 1, the CTM captures CaM_C, resulting in an intermediate where CaM_C is “tethered” to the CTM through mostly hydrophobic interactions without other stable contacts with the rest of eEF-2K. Structural features of this tethered state are captured in the NMR structure of the CaM•eEF-2K_CTM complex [35]. This tethering enhances the local concentration of CaM near the KD_N, allowing the efficient engagement of the second interface through mostly hydrophilic interactions between CaM_C and KD_N. The resulting “collapse” (Step 2) yields a collapsed state with global features characteristic of the active, chemistry-compatible species. These features are illustrated by the various crystal structures of the CaM•eEF-2K_TR complex [11–13]. Additional conformational changes involving critical active site elements (and perhaps those at remote sites as in cPKs [37–39]) that facilitate and follow the binding of ATP, the catalytically essential Mg²⁺ ion, and substrate (the T348 phospho-receiver) are necessary to enable phospho-transfer chemistry.

Within the structures of the collapsed complex (CaM•eEF-2K_TR), the largely acidic face of CaM_C lies proximal to several basic residues on KD_N. It is possible that electrostatic steering [40] enabled by these complimentary features facilitates initial bimolecular encounters. Two sets of salt-bridges consistently present in the NMR ensemble of the tethered (CaM•eEF-2K_CTM) complex are fully/partially lost in the collapsed complex. The first, K91-E114’ (CaM residues primed), is retained in some of the CaM•eEF-2K_TR structures. In contrast, the second, which is somewhat more heterogeneous within the NMR ensemble, with K93 variously partnering with E84’ or E87’, is lost in all CaM•eEF-2K_TR structures obtained thus far. These apparently non-native salt-bridges may stabilize initial “encounter complexes” [41] facilitating CTM capture and/or formation of the final collapsed state. Electrostatic steering and the resulting rapid (~10⁸ M⁻¹s⁻¹) on-rates [40], measurable in pre-steady-state binding measurements, would enable eEF-2K to successfully compete for free cellular CaM on fast timescales relevant to many physiological processes, e.g., modulation of synaptic structures involved in learning and memory [25]. While this two-step kinetic sequence is supported by NMR [35, 42] and crystallographic studies [11], it requires validation through precise kinetics analyses.

A-spine in the active state

It has been shown that cPKs encode two conserved networks of hydrophobic residues (“spines”), bridging the N- and C-lobes of their KDs. The assembly of these regulatory (R-spine) and catalytic (C-spine) spines is generally indicative of the active state of the kinase [43, 44]. Equivalent spines are not evident in eEF-2K (or other α–kinases). However, an inspection of the structure of the CaM•eEF-2K_TR complex bound to ATP reveals a set of highly conserved hydrophobic residues extending from CaM_C through a regulatory element (RE) and the KD_N to the active site (Figure 2). This “activation” spine (A-spine) linking CaM_C to the kinase active site through the bound ATP can only form in the collapsed complex. The use of the bound nucleotide as a spine component is somewhat reminiscent of the C-spine in cPKs, though the latter is spatially and geometrically distinct. The A-spine appears to represent the key structural feature through which CaM, specifically CaM_C, communicates with the eEF-2K active site, making its formation a critical feature for kinase activation. The A-spine is buttressed by two conserved tryptophan residues, the CTM anchor W85, and W99 on the RE (Figure 2, top right). The role of these residues in the activation mechanism is confirmed by the fact that mutation of either residue suppresses catalytic ability (reduced catalytic constant, k_cat) [11]. Similarly, a conservative H107’K mutation on CaM also predicted to distort the spine (Figure 2), is compromised in its ability to activate eEF-2K [36, 45].

Figure 2. Characteristics of the A-spine.

A set of conserved hydrophobic residues form the activation-spine (A-spine; PDB:8FNY), a spatially connected network that includes the C-lobe of calmodulin (CaM_C), the calmodulin-targeting motif (CTM), the regulatory element (RE), and the N-lobe of the kinase domain (KD) of eEF-2K. The A-spine extends to the eEF-2K active site through the bound ATP. The A-spine is anchored by two invariant tryptophan residues, W85 (on the CTM) and W99 (on the RE). Surface, space-filling, and schematic representations of the A-spine are shown on the top left, top right, and bottom panels. Key residues that form the A-spine are indicated and labeled using the colors of the structural units they belong to; W85 and W99 are labeled with larger fonts. Residues of the eEF-2K active site are italicized, and calmodulin residues are primed.

Our proposed mechanism for the CaM-mediated activation of eEF-2K is CaM_Ccentric. The role of the N-lobe of CaM (CaM_N), if any, in influencing the process remains unclear. Indeed, the available crystal structures of the CaM•eEF-2K_TR complex illustrate a disordered CaM_N [12, 13] or one stabilized through non-physiological contacts in crystallo [11]. One may wonder if CaM_C alone would be sufficient to activate eEF-2K, a scenario that needs examination through functional analysis. Such lobe-specific activation of a CAMK by CaM is not unprecedented [46]. It is, however, possible that CaM_N, while not a primary driver of activation, can modulate eEF-2K function by interacting with elements absent in the minimal eEF-2K_TR construct (see Box 2). Further structural/biochemical analyses, not resolvable through eEF-2K_TR, are necessary to address this question.

Fine-tuning activity through post-translational modifications

The formation of the collapsed complex and the assembly of the A-spine represent the primary drivers for the activation of eEF-2K. This scenario aligns with the fact that CaM binding alone generates a >1000-fold enhancement in kinase activity [47]. This activity is further regulated by additional inputs, notably phosphorylation events, including several centered on the regulatory-loop (R-loop, see Box 1). Principal among these is the rapid (~3 sec⁻¹) autophosphorylation on the conserved T348 following CaM binding, which leads to a ~6-fold enhancement in k_cat and a similar decrease in the Michaelis constant (K_M) for a peptide substrate [47], together generating a ~30-fold enhancement in catalytic efficiency and indicating a further conformational change upon phosphorylation. The nature of this change can be inferred from the structures of the CaM•eEF-2K_TR complex that show pT348 bound to a phosphate-binding pocket (PBP) on the C-lobe of the KD (Figure 3, top right). This pocket and its engagement by a phosphorylated residue appear to be a conserved regulatory event in α–kinases [48]. T348 phosphorylation correlates with a reduction of conformational dynamics, mainly within the kinase C-lobe and CaM_C, as indicated by HXMS and NMR measurements [36]. These changes appear to stabilize the active complex, thereby enhancing catalytic efficiency.

Figure 3. Regulation of eEF-2K activity.

Three distinct stimuli that regulate eEF-2K activity are indicated. ADP that stabilizes the active complex by binding at the basic pocket (details of the interactions, labeled in scarlet, are shown in the left inset; PDB:8FNY) at the interface between the calmodulin C-lobe (colored jasmine) and the kinase domain (green) of eEF-2K. Phosphorylated T348 (pT348) engages a phosphate-binding pocket on the C-lobe of the kinase domain (key interactions, labeled in cyan, are shown on the top right inset). S500, another site of activating autophosphorylation, rests in a surface pocket in the unphosphorylated state (key interactions, labeled in black, are shown in the bottom right panel) in some structures of the CaM•eEF-2K_TR complex (PDB:8SHQ). This residue is expected to have an altered binding mode upon S500 phosphorylation due to enhanced electrostatic repulsion with E264.

eEF-2K activity may also be modulated through irreversible post-translational modifications that cannot be enzymatically “erased,” e.g., the proposed inhibitory hydroxylation of P98 on the RE [49]. The CaM•eEF-2K_TR structure provides a rationale for such an effect. P98 and F102 bracket W99, a critical component of the A-spine (mentioned above). Given the closed-packed nature of this spine (see Figure 2, top right), any modification, especially an increase in the volume of P98, would misalign/disrupt the spine, impairing catalytic function. This outcome, in essence, can be predicted to be similar to altering W99 (discussed above) [11].

An unexpected allosteric site

The structures of the CaM•eEF-2K_TR complex in the presence of nucleotides revealed the surprising presence of ADP bound in a basic pocket (BP) at the interface between KD_N and CaM_C (Figure 3, left) away from the catalytic site. The fact that ADP engages both KD_N and CaM_C [13] indicates its stabilizing influence on CaM within the complex. Indeed, eEF-2K is activated by lower concentrations of CaM (reduced K_CaM) in the presence of ADP, suggesting that it acts as a non-essential allosteric activator [50] of the kinase. This observation suggests a direct energy-sensing role for eEF-2K and enhanced activation under energy deprivation with elevated ADP and depressed ATP levels, likely supplementing an indirect effect mediated by the master energy-sensor, AMP-dependent kinase (AMPK) [30, 51]. Indeed, the significant enhancement of eEF-2 phosphorylation upon stimulation with the glycolysis inhibitor 2-deoxyglucose is no longer seen in cells expressing BP mutants compromised in the ability to bind ADP. However, whether the BP is specific to ADP or can also accommodate other agonistic/antagonistic metabolites remains an open question.

It is also possible that the BP may host phosphorylated R-loop residues, demonstrating responsiveness to the cellular energy state. Such interactions can potentially impact the ADP content within the BP dynamically. Potential candidates for this regulatory role include S359 [52] and S366 [53], recognized as inhibitory sites targeted by the mTOR pathway. Phosphorylated forms of S359 or S366 could effectively access the BP and directly compete with ADP. Notably, AMPK signaling, activated in response to energy stress, coordinates mTOR downregulation [54]. Consequently, the ensuing reduction in phosphorylation levels at S359 and S366 would alleviate their inhibitory constraints within the BP. Alternatively, the inhibition at the BP might take an indirect route, potentially involving a conformational change induced by inhibitory R-loop phosphorylation, altering the BP geometry, and diminishing its affinity for ADP.

eEF-2K as a therapeutic target

eEF-2K has been suggested to be a therapeutic target for several human diseases, including those of the cardiovascular system [55], neuro-pathologies [18], and several cancers [56]. The most common mode of targeting kinases is through small molecule inhibitors [57, 58], and indeed, eEF-2K inhibition may be beneficial in some instances, e.g., in Alzheimer’s disease [59]. However, an activator, rather than an inhibitor, may be warranted in some other cases, e.g., to promote cell death and halt cancer progression [60]. The BP possesses characteristics that are amenable to targeting by rationally designed pharmacological agents that allow both outcomes to be achieved by enhancing or inhibiting CaM/eEF-2K interactions, thereby, eEF-2K activation and activity. Since eEF-2K is the only CaM-activated α–kinase, the BP is unique to eEF-2K. This contrasts with the highly conserved nature of the catalytic site in the other human α–kinases (Box 2). Thus, targeting the BP may be of interest in future efforts to design eEF-2K activators and inhibitors, the latter perhaps in tandem with ATP-competitive active site binders [12] for enhanced efficacy.

The known unknowns

While the CaM•eEF-2K_CTM and CaM•eEF-2K_TR structures provide considerable insights into the mechanism by which CaM recognizes and activates eEF-2K, many questions about regulation, an essential feature that allows eEF-2K to fine-tune its activity in response to diverse cellular cues remain unanswered. The effects of these cues on structure and dynamics, individually and in combination, are needed to understand how these many inputs are integrated by eEF-2K into a single self-consistent response, the regulation of translational elongation through eEF-2 (Figure 4A). While some unresolved questions, e.g., the role of CaM_N or regulation through the BP, have been mentioned above, below we discuss a few additional incomplete, occasionally contradictory results that need resolution through future studies.

Figure 4. eEF-2K is a signal integrator.

(A) Multiple signals integrate at eEF-2K to regulate its activity. These are translated into a single coherent response by eEF-2K through eEF-2 phosphorylation to modulate translational elongation. Stimulatory and inhibitory inputs are indicated in blue and red, respectively. The effects of phosphatases (in cyan) may either be stimulatory (dephosphorylation of an inhibitory site) or inhibitory (dephosphorylation of a stimulatory site). While various inputs are represented as independent, they may be highly correlated, leading to complexity like feedback or feedforward loops. For example, crosstalk between H⁺ (pH changes) and Ca²⁺ signals, perhaps through CaM [65] could be expected in eEF-2K regulation. Further, specific post-translational modifications, e.g., S500 phosphorylation, may alter the propensity of the enzyme to be degraded [61]. (B) eEF-2K activity is proportional to the population (P_C) of the collapsed complex at a fixed CaM concentration. Various stimuli modulate activity by altering P_C rather than by significantly changing the characteristics of the active state. Stimulatory cues enhance activity by stabilizing the collapsed state or, perhaps, destabilizing the inactive state, while inhibitory cues have the opposite effect. There are likely many inactive states (e.g., free eEF-2K and the tethered state are inactive with no, or extremely low, basal activity) distributed over a broad range of conformations. The active state may also comprise an ensemble of structures, but these would be distributed within a narrower range of collapsed conformations with similar catalytic ability (k_cat) but slightly different CaM affinities (K_CaM).

The cationic modulators – Ca²⁺ and H⁺

Since CaM is an essential activator of eEF-2K, it is unsurprising that its kinase activity is Ca²⁺-responsive. Indeed, all CaM•eEF-2K_TR structures contain metal ions assigned to Ca²⁺ at the CaM_C sites. However, solution NMR studies suggest the absence of Ca²⁺ at the CaM_C sites in the CaM•eEF-2K_CTM [35] and CaM•eEF-2K_TR complexes [42]. The reason for this divergence is unclear, though plausible explanations can be conceived. First, the solution studies were carried out with eEF-2K_TR that was not phosphorylated on T348. It is possible that while Ca²⁺ is dispensable within the CaM_C sites to form the tethered and collapsed complexes, it stabilizes the specific state of the latter in which T348 is phosphorylated and engaged at the PBP. It is also possible that increased stability (and reduced dynamics) of the Ca²⁺-bound species makes it amenable to crystallization. It is worth mentioning that CaM, when bound, can fully activate eEF-2K in the absence of Ca²⁺, as reflected by the essentially unchanged k_cat values towards a peptide substrate. It appears that Ca²⁺ enhances the CaM sensitivity of eEF-2K (Ca²⁺ induces a ~800-fold reduction in K_CaM) [61] rather than influence the nature of the active state. This ability of Ca²⁺ to modulate the CaM dependence of eEF-2K activity likely facilitates its tunability to Ca²⁺ transients [62]. Analyses, both experimental [63] and theoretical [64], like those available for CAMKII, are needed for a more complete understanding.

eEF-2K has been suggested to be activated by reduced pH through an unknown mechanism [45]. This may be attributed to the protonation states of histidine residues, which generally have pKa values near physiological pH, in the active CaM•eEF-2K complex (eEF-2K has 30, CaM has 1). While contributing eEF-2K residues have not yet been identified, H107’ on CaM has been suggested to play a role in the pH sensitivity [45]. Indeed, the H107’-bearing CaM F-helix at the CaM_C/eEF-2K_TR interface shows enhanced protection from solvent exchange upon lowering the pH in HXMS measurements [36]. This suggests a stabilization of the collapsed complex at reduced pH, thereby enhancing the CaM sensitivity (reduced K_CaM) of eEF-2K.

Given that both H⁺ and Ca²⁺ affect eEF-2K activity, and their compensatory effect on calmodulin conformation has been noted [65], the crosstalk between these two regulatory ions on eEF-2K activity would not be unexpected.

Additional phosphorylation events

As was mentioned earlier, a substantial segment of the R-loop was removed to generate the eEF-2K_TR construct (see Box 2). This deletion resulted in the loss of several regulatory phosphorylation sites, including S359 and S366 (mentioned above). However, the eEF-2K_TR construct does retain a key activating autophosphorylation site, S500 (invariant in vertebrates), that is also a target of PKA [66, 67]. Phosphorylation on S500 (validated through a phospho-mimetic S500D mutation) has been shown to enhance autophosphorylation on T348 and result in full activation of eEF-2K at a ~20-fold lower K_CaM in addition to making the kinase more resistant to dephosphorylation in cells [61]. These observations suggest a conformation distinct (locally, at the very least) from those captured by the available structures of the CaM•eEF-2K_TR complex. Further, a pS500-generated conformation makes eEF-2K susceptible to proteasomal degradation [61]. Notably, regions proximal to S500 show significant conformational variability across the various CaM•eEF-2K_TR structures [12, 13] determined thus far. In a subset of these structures, the unphosphorylated S500 docks into a shallow pocket and is stabilized by interacting with E264 (Figure 3, bottom right) [11]. This conformation is highly unlikely for a pS500 residue due to enhanced electrostatic repulsion with E264.

Substrates one, maybe more, and many modulators

To understand its function in atomic detail, it is necessary to define the structural characteristics of eEF-2K beyond its binary complex with CaM. For example, while the spatial location of the eEF-2-binding site on eEF-2K has been determined [68], the structural determinants of their specific interaction remain unknown. Indeed, how these features are influenced by eEF-2 phosphorylation (on Ser595 by CDK2) [69] which has been suggested to drive eEF-2K recruitment, remains unclear. While eEF-2 remains the only fully validated cellular target of eEF-2K, recent studies have suggested that there may be others [70]. Among the most intriguing of the predicted eEF-2K substrates is the α4 subunit of PP2A phosphatase that is known to dephosphorylate eEF-2 [71] and is, therefore, a participant in the eEF-2K/eEF-2 nexus. Even less is known about the association of eEF-2K with regulatory kinases like PKA [66] or recently discovered binding partners such as Group I metabotropic glutamate receptors and the associated scaffolding protein, Homer [24]. Validating, constituting, and characterizing these complexes and those with other predicted interaction partners [72] represent a future challenge for eEF-2K structural biology.

eEF-2K at rest

In the sections above, we infer atomic details of the activation and regulation of eEF-2K (using eEF-2K_TR as a proxy) from the structures of its complexes with CaM. However, the structure (and conformational dynamics) of eEF-2K in its resting, CaM-free state remains unknown. Defining key structural characteristics of this state is also necessary to fully understand the nature of the inactive, CaM-free state of the kinase. This is challenging given the enhanced heterogeneity and dynamics expected in this state without the stabilizing influence of CaM [36].

Concluding Remarks

The current state of knowledge based on structural, biophysical, and biochemical studies suggests an underlying theme in the activation and regulation of eEF-2K. As noted above, the singular event that activates eEF-2K is its interaction with CaM. All other signals (at least those characterized thus far) appear to modulate the ability of CaM to engage eEF-2K, i.e., they alter K_CaM while minimally affecting k_cat. Thus, the interaction with CaM may represent an on/off switch, with the additional regulatory signals emulating a rheostat to fine-tune kinase activity. Thus, this simple model suggests a correlation between the population (P_C, Figure 4B) of the collapsed complex and the activity of eEF-2K at a given CaM concentration. Each regulatory signal changes P_C by altering the fraction of eEF-2K in the active collapsed complex. These inputs may elicit substantial responses, e.g., variations in Ca²⁺ levels. The effects of others, e.g., individual phosphorylation events, could be more subtle. These may function in unison or opposition, thereby reinforcing or dampening, respectively, the influence of individual effects to precisely synchronize the eEF-2K/eEF-2 nexus with cellular requirements. The structural changes driven by many, if not most, of these effects, individually or in combination, to influence eEF-2K activity remain unknown. It is worth noting that eEF-2K is also modulated through spatiotemporal changes in expression [73]. Discussion of these latter aspects is beyond the scope of this review.

The apportioning of the cellular pool of eEF-2K (estimated to be ~70 nM in HeLa cells [74]) between free, CaM-bound, and in various states of post-translational modification at basal conditions or under the influence of specific stimuli remains unknown. Therefore, precise in cellulo studies would be needed to establish the physiological importance of the specific eEF-2K states characterized through current and future structural studies.

In this review, we have highlighted recent advances that enhanced our understanding of eEF-2K function in structural terms. However, as noted above, many questions remain open (see Outstanding Questions). In the coming years, we expect these will be resolved through the creative synergy between structural, biochemical, cell-biological, and computational approaches.

Outstanding questions.

• What are the structural and dynamic features of the resting, calmodulin-free state of eEF-2K?

• What are the mechanisms, in terms of structure and dynamics, through which Ca²⁺ ions, pH changes, post-translational modifications, and ligand binding regulate eEF-2K activity? How do intrinsically disordered regions of eEF-2K and the N-lobe of calmodulin contribute to these mechanisms?

• What are the effects of the crosstalk between different regulatory inputs, e.g., pH changes and Ca²⁺ transients, on the structure and conformational dynamics of eEF-2K and its activation and activity?

• How does eEF-2K recognize and phosphorylate its substrate eEF-2? Are there other cellular substrates of eEF-2K, and if so, what are the structural determinants of their interactions with eEF-2K?

• What are the structural elements on eEF-2K that enable the formation of higher-order complexes that are suggested to play critical roles in specific cellular processes?

• What is the relevance of distinct functional forms of eEF-2K and its complexes characterized in vitro within the cellular context?

• Can eEF-2K activity be modulated through small molecules targeting the basic pocket to alter its interactions with CaM?

Highlights.

• The α–kinase eEF-2K regulates translational elongation by phosphorylating the ribosome-associated GTPase eEF-2.

• eEF-2K is activated by calmodulin through a unique allosteric mechanism distinct from the release-of-inhibition mechanism operative in other calmodulin-dependent kinases.

• eEF-2K activity is synergistically regulated by multiple inputs, including Ca²⁺ transients, nucleotides, pH, and post-translational modifications mediated by energy and nutrient sensing signaling pathways.

• Recent studies have provided structural insight into the recognition and activation of eEF-2K by calmodulin and its regulation by a subset of cellular stimuli.

GLOSSARY

1–5-8 mode

Interaction mode in which the hydrophobic face of a calmodulin lobe engages hydrophobic residues at positions 1 (generally a bulky residue, e.g., FWLIV), 5, and 8 on an amphipathic helix on its target. This mode and variations, such as 1–8-14, 1–5-10, 1–5-8–14, 1–5-8–26, etc., have been noted in calmodulin/target interactions

α–kinase

Atypical kinases missing conserved sequence motifs characteristic of conventional protein kinases. Human α–kinases include ALPK1–3, TRPM6,7, and eEF-2K

A-spine

A network of conserved hydrophobic residues linking the C-lobe of calmodulin to the active site of eEF-2K through the bound nucleotide. The formation of this spine represents a key structural feature of the active kinase

Basic pocket (BP)

A pocket formed at the interface between the C-lobe of calmodulin and the N-lobe of the kinase domain that engages ADP and possibly other ligands

Calmodulin-dependent kinase (CAMK)

Here, we use this term to refer to those protein kinases that rely on calmodulin for activity. We do not imply members of the CAMK group of the kinome tree. Indeed, some calmodulin-dependent protein kinases, such as eEF-2K and the CAMKKs, are not members of the CAMK group

Calmodulin-targeting motif (CTM)

Short linear sequence on a target protein that is recognized by calmodulin. eEF-2K contains such a sequence on its N-terminus

Collapsed state/complex

A conformation in which the C-lobe of calmodulin interacts with both the calmodulin-targeting motif and the N-lobe of the kinase domain of eEF-2K. This conformation encodes the global features of the active state

Conventional protein kinase (cPK)

These are also called Hanks-type kinases and are found on the many branches (CMGC, TKL, AGC, etc.) of the main dendrogram defined by Manning et al

eEF-2K_CTM

A 27-residue peptide from the N-terminus (74–100) of eEF-2K encompassing its calmodulin-targeting motif

eEF-2K_TR

An eEF-2K construct missing 69 residues from its N-terminal region, and the 359–489 segment of the regulatory loop has been replaced by a 6-Gly linker. This construct represents a minimal calmodulin-activable core of eEF-2K

K _CaM

Concentration of calmodulin required for half-maximal activity of eEF-2K. K_CaM provides a measure of the sensitivity of eEF-2K to calmodulin

Phosphate-binding pocket (PBP)

A pocket on the C-terminal lobe of the kinase domain of eEF-2K that represents the docking site for pT348

Regulatory element (RE)

A loop that links the calmodulin-targeting motif to the kinase domain of eEF-2K. Several conserved residues within this element are part of the A-spine

Regulatory loop (R-loop)

A long loop, predicted to be disordered, that links the kinase and C-terminal domains of eEF-2K and contains several regulatory phosphorylation sites. A large R-loop segment is absent in the eEF-2K_TR construct

Tethered state/complex

In this intermediate conformation, the C-lobe of calmodulin is bound to eEF-2K only through its calmodulin-targeting motif