When Ca2+ ions flow through the pore of an individual voltage-gated Ca2+ channel, they act back on the channel they've passed through and alter subsequent Ca2+ flow. Such local, almost instantaneous regulation involves both positive and negative feedback mechanisms: Ca2+-dependent facilitation (CDF) and Ca2+-dependent inactivation (CDI), respectively. Indeed, some types of Ca2+ channel are capable of undergoing both CDF and CDI, with each form of modulation following a different time course and having a different dependency on the rate, extent, and spatial localization of Ca2+ entry. These channels are equipped with a special Ca2+-sensitive toolkit, which they use to exquisitely manipulate their own Ca2+ influx, and thereby adjust the many effector responses that lie downstream of the Ca2+ entry. Although such self-regulation has been recognized at the cellular level for decades (Brehm and Eckert, 1978; Marban and Tsien, 1982), recent work has begun to illuminate the underlying molecular mechanisms and their inherent complexity. Chaudhuri et al. (see p. 385 of this issue) have added several essential links to this chain of discovery.
CDF and CDI are present in most, but not all, members of the high voltage–activated Cav1 and Cav2 channels, but are entirely absent in the low voltage–activated Cav3 channels. Curiously, despite their diametrically opposite effects on channel gating and consequent Ca2+ influx, both CDF and CDI are mediated by calmodulin (CaM). Site-directed mutagenesis of recombinant Cav1 and Cav2 channels demonstrated that CaM interacts with a canonical “IQ” domain (a highly conserved region in the cytoplasmic C-terminal tail of all Cav1 and Cav2 pore-forming subunits). Removal of the IQ domain and mutations in the IQ domain that prevent CaM binding eliminate both CDF and CDI (Zühlke and Reuter, 1998; Qin et al., 1999; Zühlke et al., 1999). The high selectivity of CaM for Ca2+ over Ba2+ (Chao et al., 1984) explains the long-recognized inability of Ba2+ to evoke CDF and CDI. Although at first glance this might seem like a limited toolkit, CaM operates a bit like a fully loaded Swiss Army knife (Fig. 1).
How Can CaM Produce Two Opposing Actions on the Same Channel?
This dumbbell-shaped molecule coordinates two calcium ions (via a pair of EF hands) in both its N- and C-terminal lobes (Babu et al., 1988; see Fig. 2). In previous work, the Yue lab introduced selective mutations into CaM to obliterate Ca2+ binding by the N lobe (CaM12), the C lobe (CaM34), or both (CaM1234), and these recombinant probes have identified key features of CaM-Cav channel signaling. Overexpressed CaM1234 acts in a dominant-negative fashion to eliminate both CDF and CDI in Cav2.1 channels, effectively competing with endogenous wild-type CaM (DeMaria et al., 2001). This result argues (a) that CaM can bind directly to the channel even in its Ca2+-free form and (b) that Ca2+ binding to CaM is essential for both CDF and CDI. In fact, the Yue lab has further demonstrated that preventing N lobe Ca2+ binding (with CaM12) selectively eliminates CDI in Cav2.1 channels, whereas preventing C lobe Ca2+ binding (with CaM34) selectively eliminates CDF (DeMaria et al., 2001).
CDF and CDI operate in distinctly different time domains as well. During a train of action potentials, for example, CDF develops over the first few milliseconds of train initiation, while CDI requires many tens or hundreds of milliseconds (which is nicely illustrated in Fig. 1 A of Chaudhuri et al., 2007). To explain this kinetic difference, investigators have speculated that CDF mediated by the C lobe may be most effectively activated by rapid spikes of calcium (such as those that occur locally at the inner mouth of an open calcium channel), whereas CDI mediated by the N lobe may result from more slowly developing Ca2+ elevations (such as those produced globally through contributions from many channels opening in concert) (DeMaria et al., 2001). In other words, CaM relies on spatiotemporal aspects of the Ca2+ signal in choosing which knife blade (or set of blades) to use on the channel. Support for such speculation has been indirect, based primarily on two experiments that have analyzed macroscopic currents through Cav2.1. First, Ca2+ chelators, capable of buffering Ca2+ globally in the cytoplasm but not locally near the channel's mouth, eliminate CDI without altering the development of CDF in Cav2.1 (Liang et al., 2003). Second, CDI (but not CDF) depends upon the amplitude of the macroscopic current, which affects global more than local Ca2+ (Soong et al., 2002).
Direct Test of the Local–Global Hypothesis
In the current article, Chaudhuri et al. provide a more direct test of the local–global hypothesis for CaM signaling by studying CDF and CDI at the level of single Cav2.1 channels in cell-attached patches. Because Ca2+ ions flowing through a single open channel will change the Ca2+ concentration in the immediate vicinity of the channel pore without significantly affecting the global Ca2+ concentration in the cytoplasm at large, Chaudhuri et al. reasoned that a single active Cav2.1 channel should undergo CDF but not CDI. Channel activity was recorded using 100 mM Ca2+ (or 90 mM Ba2+) as charge carrier and depolarizing test pulses of sufficient amplitude (20 mV) and duration (250 ms) for both CDF (rapid) and CDI (slow) to be observed, if present. Immediately following the onset of depolarization, Chaudhuri et al. found that the single channel open probabilities (po) were comparable in Ca2+ and Ba2+. Over the ensuing 50 ms, however, po in Ca2+ rose monotonically to a new level and stayed there throughout the remainder of the pulse (indicating CDF had occurred), while Ba2+ currents remained constant at the initial po for the entire length of the depolarization (consistent with the observation that CDF, and CaM, are not activated by Ba2+). In contrast, CDI is all but absent from the single channel records; at the end of the 250-ms-long test pulse, a period over which macroscopic Cav2.1 currents decline as much as 30% (Lee et al., 2000; DeMaria et al., 2001), the single channel po decreased no more than ∼5% in Ca2+. These data uphold the authors' prediction that CDF, but not CDI, would be manifest at the single channel level.
Additional experiments performed by Chaudhuri et al. (2007) further strengthen the linkage between CDF measured at the single channel and macroscopic levels. First, when a conditioning depolarization was delivered immediately before the test pulse, the channel opened directly to the higher probability facilitated state, mirroring previous results on macroscopic currents (Lee et al., 2000; DeMaria et al., 2001). Second, a splice variant of Cav2.1 that was shown previously to be incapable of undergoing CaM-induced CDF, measured macroscopically (Chaudhuri et al., 2004), also lacked CDF measured on single channels. The new publication from the Yue lab thus provides strong support for the “local calcium preference” of C lobe–induced CDF and the “global Ca2+ preference” of N lobe–induced CDI in Cav2.1 channels.
What Is the Mechanistic Basis for the Local/Global Preference?
CaM is a small molecule; in its most extended form, the N and C lobes are separated by no more than ∼65 Å (Babu et al., 1988). By contrast, the rapid Ca2+ transients that activate local C lobe–mediated responses of CaM extend over at least tens of nanometers (Naraghi and Neher, 1997), making it unlikely (based solely on a spatial argument) that the two ends of CaM are surrounded by different concentrations of Ca2+. Thus, the apparent local/global preference of CaM's lobes must result from the differential kinetics with which the two lobes of CaM bind Ca2+ under physiological conditions (Pitt et al., 2001; Tadross, M., H. Liang, and D.T. Yue. 2005. Biophys. Soc. Abstr. 439). That is, the local C lobe Ca2+ sensor may respond better to the rapid, local Ca2+ signal because it binds Ca2+ more quickly and/or loses it more slowly than does the N lobe, perhaps involving differential allosteric effects of the channel on the two lobes of CaM.
Proving that differences in the kinetics of Ca2+ binding to the two lobes of CaM actually underlie their local/global preferences is a formidable challenge, as methods have yet to be developed that will allow quantification of Ca2+ binding kinetics in the native cellular environment. As a first step, however, Hamilton and colleagues have edged closer to the in vivo situation by using stopped-flow fluorescence methods to study the interaction dynamics between Ca2+, CaM, and Cav channel IQ peptides in vitro (Black et al., 2005). They find that IQ peptides differentially alter the rates of Ca2+ binding to (and dissociation from) the N and C lobes of CaM. On average, the channel peptides slow Ca2+ dissociation from the C lobe more than from the N lobe, while speeding Ca2+ association to the N lobe more than to the C lobe. Furthermore, some Cav IQ domain peptides are more effective than others at regulating Ca2+ binding to CaM. Although it is too early to quantitatively extrapolate the in vitro results of the Hamilton lab to the native channel (in which multiple regions outside of the IQ domain are also known to alter CaM's effects on the channel), it is clear that variation in the two-way conversation between CaM and Cav channels offers abundant opportunities for fine tuning Ca2+ influx.
What Is the Biophysical Mechanism Underlying CDF?
In addition to providing direct support for the differential detection of Ca2+ transients by the two lobes of CaM, Chaudhuri et al. also used their single channel approach to discriminate between two separate mechanisms that could, in theory, give rise to the CDF detected macroscopically. They reason that facilitation might result from a Ca2+-CaM–mediated transition to a unique gating state with a po higher than that of the normal open state (the “enhanced opening model”). Alternatively, facilitation might encourage an otherwise sluggish channel to open more quickly, speeding the transition to the normal open state (the “accelerated activation model”). Chaudhuri et al. clearly summarize how these two models predict very different outcomes of experiments to quantify the latency to first channel opening, conditional open probability, and voltage dependence of po. Experiments on both single channel and macroscopic currents explore each of these in turn and lead the reader to the unequivocal conclusion that a unique open state (with a po significantly higher than that of the normal open state) is responsible for CDF in Cav2.1 channels.
Variable Actions of CaM on Cav1 and Cav2 Channels
Previous work has demonstrated that Cav1 and Cav2 channels are regulated unequally by CaM (Liang et al., 2003); in fact, given the high level of conservation of IQ domains in Cav channels, there is an amazing variation in CaM action (Fig. 2). CDF appears to be the more variable of the two Ca2+-dependent regulatory mechanisms. The C lobe of CaM promotes CDF in Cav2.1, for example, but not in Cav2.2 and Cav2.3, the two other members of the Cav2 family; in fact, the C lobe mutant CaM34 appears functionally silent on Cav2.2 and Cav2.3 (Liang et al., 2003). Even more surprising, alternative splicing of Cav2.1 in regions outside the IQ domain is capable of abrogating CDF (Chaudhuri et al., 2004), leading to the conclusion that although the IQ domain is necessary for CDF, it is not sufficient. Studies on Cav1.2 channels have further underscored the range of variation; CDF is absent from wild-type Cav1.2 channels but emerges following mutations in single hydrophobic residues in the IQ domain (Zühlke et al., 1999; Van Petegem et al., 2005).
Unlike CDF, CDI is found in all Cav1 and Cav2 channels (Budde et al., 2002; Liang et al., 2003), but the lobe of CaM responsible for CDI varies. For example, the C lobe of CaM mediates CDI in Cav1.2, not CDF as it does in Cav2.1, described above (Peterson et al., 1999). Interestingly, whether the C lobe is mediating CDF or CDI, its effects are refractory to cytoplasmic Ca2+ buffers (Soong et al., 2002; Liang et al., 2003), indicating that the local Ca2+-sensing properties of the C lobe are conserved across the Cav1/Cav2 family. Similarly, in all situations where the N lobe is active (e.g., across all Cav2 family members, where it mediates CDI; Fig. 2) its effects are reduced or eliminated by cytoplasmic Ca2+ chelators (Zühlke et al., 1999; Liang et al., 2003), underscoring the conserved global Ca2+-sensing properties of the N lobe.
What these examples illustrate is that the mechanistic underpinnings of CDF and CDI are complex and channel specific, with seemingly subtle structural differences having dramatic biochemical and biophysical consequences. There is still much to learn from time resolved measurements of channel gating (such as employed in Chaudhuri et al., 2007) and Ca2+-CaM-channel binding (Black et al., 2005) applied to engineered channels carrying unique subsets of the known domains important for CaM's action on Cav channels. Such functional studies will provide the necessary framework for building models that can then be validated by crystallographic snapshots of the interacting pairs.
The first structural pictures of Ca2+-CaM in complex with the Cav1.2 IQ domain have recently been obtained (Fallon et al., 2005; Van Petegem et al., 2005). This work demonstrates that several hydrophobic residues in the IQ domain are involved in the Ca2+-dependent interaction of CaM with Cav1.2. The Ca2+-bound C lobe associates with the IQ domain more tightly than does the N lobe and provides the primary means of anchoring CaM to the channel (Van Petegem et al., 2005). The N lobe, by contrast, appears free to assume different conformations, two of which have been captured thus far. Although, as mentioned above, the N lobe appears to lack functional consequences on wild-type Cav1.2, the Minor lab demonstrated that mutating one aromatic C lobe anchor point (I1624) in the IQ domain of the full-length Cav1.2 channel allowed the normally silent N lobe to evoke CDF (Van Petegem et al., 2005), confirming earlier mutagenesis experiments highlighting the ability of this residue to normally suppress CDF in wild-type Cav1.2 channels (Zühlke et al., 1999).
These structural studies provide a robust predictive framework for evaluating differences in CaM's lobe specificity for Cav1 and Cav2 channels. The dominant interaction demonstrated between the C lobe and the IQ domain is consistent with the dominance of the C lobe in Cav1.2 channel function. Given the predominance of N lobe–mediated CDI in Cav2 channels, however, one wonders whether the N lobe might bind more tightly than does the C lobe in these channels. Consistent with this idea, Van Petegem et al. (2005) note that several of the aromatic anchors for C lobe binding in Cav1.2 are not well conserved in the IQ domains of Cav2 channels. It will be interesting to determine whether mutations in aromatic residues predicted to be necessary for N lobe binding to Cav2.2 and/or Cav2.3 (which normally lack CDF) might breathe new life into the otherwise ineffective CaM C lobe to evoke CDF in these channels. Other key questions remain as well: (a) can mutating aromatic anchoring residues in Cav2.1 further strengthen CDF; (b) does splice variation in Cav1.2, Cav2.2, and Cav2.3 contribute to heterogeneity in these channels' responses to CaM as it does for Cav2.1 and Cav1.3 (Soong et al., 2002; Chaudhuri et al., 2004; Shen et al., 2006); and (c) what are the additional endogenous mechanisms (see, for example, Yang et al. 2006) that naturally regulate CaM-Cav interactions in vivo?
Final Thoughts on the Potential Physiological Impact of Cav Channel Self-Regulation
Calcium's job description as a cellular regulator is formidable: to initiate synaptic vesicle fusion, transcriptional activation, muscle contraction, and membrane permeability changes. Each of these effector responses is optimally activated by a uniquely tailored Ca2+ signal. Furthermore, some types of Ca2+ signals are clearly out of bounds, as an overabundance of Ca2+ is toxic. Not surprisingly, therefore, mechanisms that regulate Ca2+ are assumed to be of great physiological consequence, although definitive tests of this assumption have only begun to emerge.
CDF and CDI offer two opposing, activity-dependent means of altering Ca2+ influx and the many effector responses it triggers. Several aspects of this dual regulatory system are intriguing. CDF is one of nature's rare positive feedback mechanisms. The enhanced opening mechanism proposed by Chaudhuri et al. predicts that both the amplitude and the duration of Ca2+ influx will increase during CDF. This prediction has been confirmed in experiments that directly measure Ca2+ influx during action potential trains. The suggestion is that CDF provides an effective means of bringing about use-dependent enhancement of cellular responses that are mediated by Ca2+, particularly those effector responses (e.g., exocytosis at synapses mediated by Cav2.1 and Cav2.2) that vary as a nonlinear function of Ca2+ influx (Cuttle et al., 1998).
When left unchecked, however, CDF is a potentially dangerous form of regulation, given the toxic nature of sustained elevations in cytoplasmic Ca2+ concentration. Might this explain why CDF appears to be absent from most Cav channel types in vivo (Liang et al., 2003)? Given the sparse expression of CDF amongst the Cav channels, it is also interesting to speculate that the presence of CDF might signal a unique functional role for those rare variants capable of experiencing it. The hazards of runaway CDF may also provide a rationale for CDI as a coexisting countermeasure to terminate Ca2+ influx. Conveniently, CDI develops with some delay relative to CDF, which would allow short-term physiological potentiation of Ca2+ influx, while avoiding long-term pathophysiological consequences.
Unlike CDF, CDI is ubiquitous amongst the Cav1 and Cav2 channels (Liang et al., 2003) and, as such, clearly also has functions outside the context of CDF (Budde et al., 2002). In cardiac myocytes, for example, C lobe–mediated CDI of Cav1.2 channels appears to be an essential regulator of action potential duration (Alseikhan et al., 2002); in nerve terminals, N lobe–mediated CDI appears to underlie use-dependent short-term plasticity of synaptic transmitter release via Cav2.1 and Cav2.2 (Forsythe et al., 1998; Xu and Wu, 2005). Such studies, in combination with the impressive contributions of Chaudhuri et al. in this issue, move us inexorably closer to the era of CaM-Cav channel enlightenment in which we will understand not only how many blades CaM's knife contains but how each is shaped and, thereby, carves out a unique physiological niche.
Abbreviations used in this paper: CaM, calmodulin; CDF, Ca2+-dependent facilitation; CDI, Ca2+-dependent inactivation.