Mission CaMKIIγ: Shuttle Calmodulin from Membrane to Nucleus

Abstract

Neuronal plasticity depends on plasma membrane Ca²⁺ influx, resulting in activity-dependent gene transcription. Calmodulin (CaM) activated by Ca²⁺ initiates the nuclear events, but how CaM makes its way to the nucleus has remained elusive. Ma et al. now show that CaMKIIγ transports CaM from cell surface Ca²⁺ channels to the nucleus.

Learning and memory depend on long-term neuronal plasticity and the ability of neurons to weaken or strengthen synapses in response to changes in electrical activity. A long-lasting decrease in neuronal network activity leads to an overall increase in the average synaptic strength to maintain homeostasis of excitatory inputs into the neurons. This process is initiated at the cellular membrane, where Ca²⁺ influx through surface Ca_V1 (L-type) channels starts a series of events culminating in the activation of nuclear CREB and gene transcription. Though many of the players in these signaling events are known, surprisingly, the mechanism through which electrical activity at the cell surface gets transmitted to the nucleus has remained a mystery. Richard Tsien and colleagues now show that γCaMKII, a member of the CaMKII family of kinases, acts as a shuttle and, independent of its kinase activity, conveys the Ca-dependent events from the membrane to the nucleus.

The molecular details of this shuttle mission are as intriguing as the voyage of CaMKIIγ through the neuronal microcosm. In hippocampal neurons, L-type Ca²⁺ channels Ca_v1.2 and Ca_v1.3 couple neuronal excitation to Ca²⁺-controlled gene expression via the transcription factor NFAT (Murphy et al., 2014; see also Nystoriak et al., 2014) through the direct association of key signaling elements with Ca_v1.2 and Ca_v1.3. For instance, the anchor protein AKAP5 links PKA (Hall et al., 2007; Oliveria et al., 2007) and the Ca²⁺/CaM-activated phosphatase calcineurin to Ca_v1.2 for localized activation of NFAT (Figure 1) (Murphy et al., 2014). Moreover, CaMKII binds to a specific motif in Ca_v1.2 (called the IQ motif) and likely to the identical motif in Ca_v1.3, to Ca²⁺ channel β subunits, and to Ca_v1.3-associated densin-180 (Hell, 2014). Work in superior cervical ganglion neurons shows that CaMKIIβ is recruited to Ca_v1.3 upon Ca²⁺ influx through this channel and, less effectively, non-L-type channels (Wheeler et al., 2012). This recruitment turns out to be critical for activation of nuclear CaMKK and CaMKIV by Ca²⁺/CaM and for the ensuing phosphorylation of CREB. Ma et al. now reveal the missing link between these two events: CaMKIIγ, activated by CaMKIIβ upon Ca influx, carries Ca²⁺/CaM to the nucleus for the initiation of the CaMKK-CaMKIV-CREB cascade. Indeed, Ca²⁺ influx through Ca_v1.3 plasma membrane channels triggers CaMKIIγ accumulation at Ca_v1.3 clusters, where CaMKIIγ becomes loaded with the Ca²⁺/CaM cargo. Calcineurin-mediated dephosphorylation then launches CaMKIIγ from Ca_v1.3 to the nucleus.

Figure 1. Excitation-Transcription Coupling by Ca²⁺/CaM-CaMKIIγ Shuttling.

In hippocampal neurons (left), the depolarization-triggered Ca²⁺ influx via prevalent Ca_v1.2 leads to dephosphorylation and thereby activation of NFAT by AKAP5-anchored calcineurin, which is activated by Ca²⁺/CaM (Murphy et al., 2014). In SCG neurons, which are thought to lack Ca_v1.2, Ca²⁺ influx via Ca_v1.3 first results in clustering of Ca²⁺/CaM-CaMKIIγ and CaMKIIβ at or near clusters of Ca_v1.3 and calcineurin, which is probably linked to Ca_v1.3 by AKAP5 analogous to Ca_v1.2. The IQ motif (red segment) of Ca_v1.2 and Ca_v1.3 constitute docking sites for CaMKIIα and β, but not CaMKIIγ (Hudmon et al., 2005). The insert shows specific binding of increasing amounts of CaMKIIα to the IQ peptide TVGKFYATFLIQEYFR (red), but not to peptides upstream and downstream of IQ, in Ca_v1.2/Ca_v1.3 by fluorescence anisotropy assays (Z.A.M. and J.W.H., unpublished data), as seen earlier (Hudmon et al., 2005) (Kd here is 31 ± 7 nM). CaMKIIγ could bind to the Ca²⁺channel β subunit or Ca_v1.3-associated densin-180 (not depicted) (Hell, 2014). Effective delivery of Ca²⁺/CaM to the nucleus occurs upon phosphorylation of CaMKIIγT287 by CaMKIIβ for trapping of Ca²⁺/CaM and dephosphorylation of CaMKIIγS334 by calcineurin for nuclear import. Dephosphorylation of pT287 in CaMKIIγ by PP2A in the nucleus releases Ca²⁺/CaM for stimulation of CaMKK and CaMKIV. CaMKIV requires both Ca²⁺/CaM binding and phosphorylation of T196 by CaMKII before it can phosphorylate CREB on S133.

To determine whether depolarization-induced Ca²⁺ influx via Ca_v1.3 results in nuclear accumulation of CaMKIIγ, the authors analyzed the effects of a mutation mimicking constitutive phosphorylation of the residue S334. Not only did this mutation block nuclear import as expected, it also unmasked activity-driven accumulation of CaMKIIγ at Ca_v1.3 clusters, which apparently is only fleeting for wild-type CaMKIIγ. Inhibition of calcineurin (but not PP1/PP2A) had the same effects, indicating that calcineurin dephosphorylates S334 for both release of CaMKIIγ from Ca_v1.3 and nuclear translocation. Furthermore, mutating A303 to arginine and application of KN93 (both of which block Ca²⁺/CaM binding to CaMKII) inhibited CaMKIIγ clustering at Ca_v1.3. Accordingly, CaMKIIγ accumulation at Ca_v1.3 requires Ca²⁺/CaM binding to its A303/T306/T307 segment (Ma et al., 2014), which activates CaMKII through displacement of the partially overlapping pseudo-substrate segment from the catalytic center (Hell, 2014). Such a requirement also explains why CaMKIIγT287E and CaMKIIγT287E/S334E did not accumulate at Ca_v1.3 clusters; analogous to the autophosphorylation of CaMKIIα(T286D) on T305/T306 that prevents Ca²⁺/CaM binding (Hell, 2014; Pi et al., 2010), the T287E mutation in CaMKIIγ likely abrogates Ca²⁺/CaM binding by corresponding autophosphorylation at T306/T307.

Going back to the membrane effects, the authors found that, upon excitation, CaMKIIβ accumulates in parallel with CaMKIIγ at Ca_v1.3 clusters (Wheeler et al., 2012) for trans-phosphorylation of CaMKIIγ at T287. This phosphorylation causes CaM trapping by dramatically increasing CaMKIIγ affinity for Ca²⁺/CaM (Hell, 2014). Following knockdown of CaMKIIβ, CaMKIIγ shuttles to the nucleus but without Ca²⁺/CaM on board, failing to activate CREB. These observations suggest that CaMKIIβ phosphorylates CaMKIIγ on T287 upon their encounter at Ca_v1.3, but CaMKIIβ knockdown could also act by, e.g., prohibiting CaMKIIβ from facilitating Ca²⁺ influx via Ca_v1.3, as previously seen during repetitive, frequent depolarization (Jenkins et al., 2010). This effect could limit Ca²⁺ influx and thereby CaMKIIγ autophosphorylation on T287 and consequently loading of its cargo. Nevertheless, replacing endogenous CaMKIIγ with a catalytically inactive CaM KIIγK43R mutant rescued nuclear translocation of both CaMKIIγ and Ca²⁺/CaM. These findings suggest that CaMKIIβ has at least the potential to phosphorylate CaMKIIγ, although CaMKIIγ autophosphorylation could become necessary as the one remaining mechanism of T287 phosphorylation when CaMKIIβ is absent.

Accumulation of CaMKIIγS334E at Ca_v1.3 clusters indicates that CaMKIIγ obtains temporarily heightened affinity for Ca_v1.3, which likely requires some sort of release mechanism for sending CaMKIIγ on its journey to the nucleus. In fact, inhibition of calcineurin prevents the CaMKIIγ release. Analogous to Ca_v1.2, calcineurin could be associated with Ca_v1.3 via AKAP5 for localized, selective, and effective signal transduction (Figure 1). Perhaps CaMKIIγS334 dephosphorylation by Ca_v1.3-associated calcineurin fulfills dual function by facilitating release from Ca_v1.3 as well as enabling nuclear import. However, accumulation of CaMKIIγS334E at Ca_v1.3 clusters could also simply be because abrogation of nuclear import increases Ca²⁺/CaM-loaded CaMKIIγ S334E in the cytosol, thereby shifting the equilibrium toward Ca_v1.3 binding. Dephosphorylation of T287 would not be useful for release of CaMKIIγ from Ca_v1.3 because it would result in loss of the Ca²⁺/CaM cargo. In fact, CaMKIIγT287A accumulates in the nucleus but without Ca²⁺/CaM. Although this finding implies that CaMKIIγT287A can transiently associate with Ca_v1.3, as required for S334 dephosphorylation, it also suggests that T287 phosphorylation is not required for Ca_v1.3 association.

How is Ca²⁺/CaM released from CaMKIIγ once it arrives in the nucleus? Inhibition of the serine/threonine phosphatase PP2A, which can dephosphorylate CaMKII, increased T287 phosphorylation and decreased CREB phosphorylation without affecting total CaMKIIγ or Ca²⁺/CaM inside the nucleus. It thus appears that activation of the CaMKK-CaMKIV-CREB cascade requires nuclear PP2A dephosphorylation of T287 to trigger Ca²⁺/CaM release from CaMKIIγ (Figure 1).

Important questions now arise from the current work. For instance, it still remains to be understood how Ca²⁺/CaM can remain associated with CaMKIIγ after release from Ca_v1.3 when CaM usually dissociates rather quickly from CaMKIIs upon falling Ca²⁺ levels. Perhaps prolonged stimulation by elevated K⁺ or high-frequency field stimulations as used by Ma et al. maintains high enough [Ca²⁺]_i to sustain Ca²/CaM-CaMKIIγ association (especially as CaM binding to targets typically increases affinity of CaM for Ca²⁺), but this issue warrants further detailed investigation. Moreover, an intriguing point to consider is how does the large CaMKIIγ-Ca²/CaM complex enter the nucleus? Is this translocation facilitated through interactions with the nuclear pore complex? Along these lines, how is dephosphorylation of T287 by PP2A regulated so that it occurs in the nucleus, but not cytosol? And finally, are different transcriptional pathways selectively activated and how? This is an especially fascinating question, as Ca_v1-associated calcineurin plays a central role in activation of both NFAT and CREB upon Ca²⁺ influx.

In summary, this study identifies a long sought-after mechanism of communication between the membrane and the nucleus upon excitation, which mediates the ability of neurons to sustain long-term plasticity. Future studies will undoubtedly expand on these findings to reveal additional mechanisms that mediate selective activation of transcriptional pathways in health and disease.