Enzyme-inhibitor-like tuning of Ca²⁺ channel connectivity with calmodulin

Abstract

Ca²⁺ channels and calmodulin are two prominent signaling hubs¹ that synergistically impact functions as diverse as cardiac excitability², synaptic plasticity³, and gene transcription⁴. It is thereby fitting that these hubs are in some sense coordinated, as the opening of Ca_V1-2 Ca²⁺ channels are regulated by a single calmodulin (CaM) constitutively complexed with channels⁵. The Ca²⁺-free form of CaM (apoCaM) is already preassociated with the IQ domain on the channel carboxy terminus, and subsequent Ca²⁺ binding to this ‘resident’ CaM drives conformational changes that then trigger regulation of channel opening⁶. Another potential avenue for channel-CaM coordination could arise from the absence of Ca²⁺ regulation in channels lacking a preassociated CaM^6,7. Natural fluctuations in CaM levels might then influence the fraction of regulatable channels, and thereby the overall strength of Ca²⁺ feedback. However, the prevailing view has been that the ultra-strong affinity of channels for apoCaM ensures their saturation with CaM⁸, yielding a significant form of concentration independence between Ca²⁺ channels and CaM. Here, we reveal significant exceptions to this autonomy, by combining electrophysiology to characterize channel regulation, with optical FRET sensor determination of free apoCaM concentration in live cells⁹. This approach translates quantitative CaM biochemistry from the traditional test-tube context, into the realm of functioning holochannels within intact cells. From this perspective, we find that long splice forms of Ca_V1.3 and Ca_V1.4 channels include a distal carboxy tail^10-12 that resembles an enzyme competitive inhibitor, which retunes channel affinity for apoCaM so that natural CaM variations affect the strength of Ca²⁺ feedback modulation. Given the ubiquity of these channels^13,14, the connection between ambient CaM levels and Ca²⁺ entry via channels is broadly significant for Ca²⁺ homeostasis. Strategies like ours promise key advances for the in situ analysis of signaling molecules resistant to in vitro reconstitution, such as Ca²⁺ channels.

Our investigations build on a Ca_V1.4 channel mutation underlying congenital stationary night blindness¹⁵. This mutation yields a premature stop that deletes the distal carboxy tail (DCT) of these retinal Ca²⁺ channels, and produces a surprising emergence of their Ca²⁺ regulation by CaM^11,12 (Ca²⁺-dependent inactivation, CDI). Full-length Ca_V1.4 channels lack CDI^11,12, thereby maintaining Ca²⁺-driven transmitter release at tonically depolarized retinal synapses. Hence, the emergence of CDI likely impairs vision. Mechanistically, the DCT contains an ICDI module that is reported to somehow ‘switch off’ the latent CDI of Ca_V1.4 channels^11,12.

Figure 1 summarizes our initial characterization of ICDI effects. Because Ca_V1.4 channels yield diminutive currents¹⁶, we appended the DCT of the main Ca_V1.4 subunit (α_1F) onto the core of better-expressing Ca_V1.3 channels (main subunit α_1D). This previous approach permits robust investigation of DCT effects^11,12. As baseline, Fig. 1a displays the CDI of core Ca_V1.3 channels (left, α_1DΔ), similar to natural short splice variants⁸. Core channels contain all elements required for CDI^6,17, including the IQ domain for apoCaM preassociation⁶, and EF-hand-like region for CDI transduction¹⁸. Depolarization thereby evoked rapidly decaying Ca²⁺ current (middle, red trace), indicative of strong CDI. Since Ba²⁺ binds CaM poorly¹⁷, the slow Ba²⁺ current decay (black trace) delineates the background inactivation of a distinct voltage-dependent process^6,7 (VDI). Thus, the fraction of peak current remaining after 50-msec depolarization (right, r₅₀) relates intimately to inactivation, with the difference between Ca²⁺ and Ba²⁺ r₅₀ relations indexing CDI (f₅₀). Appending the Ca_V1.4 DCT onto the Ca_V1.3 core (Fig. 1b, left) strikingly reduced CDI (middle and right) versus control (dashes) (Supplemental 1.2). Importantly, the DCT did not altogether abolish CDI as reported before^11,12, but spared a clear residuum. This difference foreshadowed major mechanistic and biological consequences.

Figure 1. Distal carboxy tail of Ca_V1.4 weakens Ca²⁺ regulation of channels.

a, Core Ca_V1.3 channel contains all known structural elements required for CDI (left schematic of main subunit α_1DΔ, containing NSCaTE¹⁷, EF¹⁸, and IQ⁶) and thereby exhibits robust CDI (right two sub-panels). Average f₅₀ CDI metric at bottom (mean ± sem), with number of cells in parentheses. Throughout, current bars pertain to Ca²⁺ currents, Ba²⁺ currents scaled for kinetic comparison, and tail currents clipped to frame. b, Adding DCT of α_1F (main pore-forming Ca_V1.4 subunit) to core Ca_V1.3 channel weakens CDI. A, B, and ICDI segments of DCT defined in Supplemental 1.1. Dashed curve reproduces baseline from a.

Hints of these consequences came by qualitative consideration of underlying mechanism. As background, we recapitulated coarse structural underpinnings of DCT effects. To confirm DCT collaboration with core channel elements^11,12, we showed the lack of DCT effects on Ca_V2.2 channels (Supplemental 1.3), which presumably lack complementing modules. As well^10,12, only ICDI and A sub-segments (Fig. 1b, left) are required (Supplemental 1.4). Beyond these initial points of clarity, actual DCT mechanisms remain controversial. One group used GST pulldowns of channel peptides to support an allosteric mechanism (Supplemental 1.5), where the ICDI associates with an EF-hand-like module to eliminate CDI transduction¹² (Fig. 1a, left), but leaves apoCaM/channel binding unchanged. By contrast, another group employed channel peptide FRET to advance a competitive mechanism (Supplemental 1.5), where ICDI competes with apoCaM for binding near the channel IQ domain (Fig. 1a, left), thus inhibiting CDI by displacing CaM from channels¹¹.

To aid resolution, we pursued two preliminary approaches. First, live-cell FRET 2-hybrid assays^6,17 tested whether the Ca_V1.4 ICDI (Fig. 2a, CFP fused to ICDI, ECFP–ICDI_F) could bind the presumed apoCaM preassociation module of Ca_V1.3 (YFP fused to PreIQ₃–IQ–A from Fig. 1b, EYFP–PreIQ₃–IQ_D–A_F; Supplemental 2.1). We thus resolved a high-affinity in situ binding curve (Fig. 2a), where FRET strength (FR) is plotted cell-by-cell versus free ECFP–ICDI_F concentration (D_free, free donor). By contrast to prior analyses utilizing single-number FRET indices¹¹, our binding-curve clearly excludes low-affinity interaction, and a similar binding curve held true for partners solely derived from Ca_V1.4 (Supplemental 2.2). We also confirmed avid binding between apoCaM (ECFP–CaM_WT in resting cells) and this EYFP–PreIQ₃–IQ_D–A_F module⁶ (Fig. 2b). More telling, the ICDI (without fluorophore) attenuated the same apoCaM interaction (Fig. 2c, gray zone), suggesting that ICDI and apoCaM could vye for IQ occupancy (Supplemental Fig. 2.2b). In all, these data confirmed the potential for competition, but pertained only to peptides, without guarantee of analogous events within intact channels.

Figure 2. Provisional evidence for competition.

a, FRET, CFP-tagged ICDI of α_1F versus YFP-tagged PreIQ₃-IQ-A from Fig. 1b (Supplemental 2.1). FR ∝ FRET efficiency times fraction of YFP-tagged molecules bound⁶. D_free, relative concentration of unbound CFP-tagged molecules; green bar ~500 nM⁶. b, FRET, apoCaM versus PreIQ₃-IQ-A in a. c, ICDI (without fluorophore) attenuates binding in b. Gray reference curve from b. d, Left, CDI is rescued upon overexpressing CaM with chimera in Fig. 1b. Right, CaM sponge (Ca_V1.2 YFP–PreIQ₃-IQ⁶) eliminates CDI. e, Overexpressing CaM with full-length Ca_V1.4. f₃₀₀, 300-ms version of f₅₀.

Accordingly, a second provisional approach specifically targeted the holochannel configuration. Scrutiny of mechanisms (Supplemental 1.5) revealed that manipulating apoCaM levels would affect CDI only in the competitive, but not strict allosteric framework. Indeed, elevating CaM sharply reversed ICDI effects (Fig. 2d, middle), and apoCaM chelation eliminated residual CDI (Fig. 2d, right; Supplemental 2.3). Importantly, augmenting CaM also boosted CDI of full-length Ca_V1.4 channels (Fig. 2e, cf., ref. 19). Overall, both preliminary approaches supported competition, and the residual CDI seen earlier (Fig. 1b) appeared to reflect incomplete competition. Still, these data neither excluded more nuanced allosteric mechanisms²⁰, nor revealed whether biologically relevant CaM fluctuations could modulate CDI.

These limitations might be overcome, if only the free apoCaM concentration could be quantified within the very cells where CDI was measured. If so, one could delineate the holochannel equivalent of classic enzyme inhibition plots²⁰, which rigorously distinguish among mechanisms. Accordingly, we incorporated a recently developed optical FRET-based sensor of apoCaM, BSCaM_IQ. Here, CFP and YFP flank the apoCaM binding site of neuromodulin⁹ (Fig. 3a), such that the overall FRET is determined by free apoCaM concentration. We confirmed the limiting behaviors by coexpressing BSCaM_IQ with excess CaM (Fig. 3b, blue line at FR_min) or CaM-chelating peptides (green line at FR_max; Supplemental 3.1). As expected, FR was nearly independent of isolated CFP fluorescence (S_CFP), an approximate measure of sensor expression in each cell. By contrast, when BSCaM_IQ was expressed alone, FR demonstrated a comforting rise towards FR_max (Fig. 3b, black line), as anticipated for a sensor that itself chelates and decreases free apoCaM. With reassurance of BSCaM_IQ performance in our system, we coexpressed BSCaM_IQ and Ca²⁺ channels, and measured free apoCaM concentration before determining CDI in the same cell (Fig. 3c). If free apoCaM were varied among cells by CaM overexpression or chelation, the resulting CDI versus apoCaM plot would rigorously distinguish among mechanisms. Specifically, using the relation between FRET and free apoCaM concentration in our cells⁹ (Supplemental 3.2), the exact signature of competition²⁰ becomes as shown in Fig. 3d (right, gray curves) and

$$\mathit{CDI}={\mathit{CDI}}_{\max }·\frac{S_{\mathrm{b}}}{S_{\mathrm{b}}·(1-r)+r};r=K_{\mathrm{d}­\mathrm{channel}­\mathrm{apparent}}/K_{\mathrm{d}­\mathrm{sensor}}$$ (1)

where CDI_max is the maximal CDI without ICDI; K_{d-channel-apparent} is the apparent dissociation constant of channels for apoCaM (with competitive inhibitor); K_d-sensor is the dissociation constant of BSCaM_IQ for apoCaM⁹ (2.3 μM); and S_b is the fraction of sensor bound to apoCaM (Supplemental 3.3). As Fig. 3b shows, S_b is directly determined from FR, and ranges from 0 to 1 with increasing apoCaM. If ICDI competition is strong (r > 1), curves will be upwardly concave (Fig. 3d, right, gray curves); if weak (r < 1), curves will be downwardly concave.

Figure 3. Live-cell holochannel biochemistry proves competition.

a, BSCaM_IQ schematic. b, BSCaM_IQ expressed alone (black curve and data), and coexpressed with CaM or CaM sponges (neuromodulin IQ (filled green, Supplemental 3.1) or Ca_V1.2 PreIQ₃-IQ⁶ (open green)). S_CFP, isolated CFP fluorescence⁶. c, Approach to obtain CDI and FRET readouts of apoCaM (FR) in single cells. d, CDI–S_b analysis for α_1DΔ[rat]–(ABI)_F, respectively, for -10 mV steps. Right, family of gray CDI–S_b curves illustrate potential profiles for competitive inhibition, according to Eq. 1. Superimposed red data and fit conform to competitive profile. Left, corresponding exemplar traces, labelled i-iii. e-g, CDI–S_b analysis for α_1DΔ[rat], α_{1D–long[hum]}, and α_{1D–long[rat]}. Format as in d.

Figure 3d also displays the experimental outcome for core Ca_V1.3 channels affixed to the Ca_V1.4 DCT (α_1DΔ–(ABI)_F, from Fig. 1b). In the CDI–S_b plot on the right, each symbol corresponds to a single cell, and together these data fit remarkably well to the competitive scheme (red curve, Eq. 1). On the left, current traces from exemplar cells (i, ii, and iii) explicitly demonstrate the appropriate increase of CDI with growing S_b and apoCaM. Importantly, parallel analysis of core Ca_V1.3 channels revealed far greater apoCaM affinity (Fig. 3e), yielding maximal CDI throughout. Hence, the upward concavity in Fig. 3d is a genuine ICDI effect, not an unanticipated property of the Ca_V1.3 core. Critically, at high apoCaM (S_b ~ 1), CDI of both constructs converged, yielding a hallmark of competition²⁰.

Buoyed by advances for the retinal Ca_V1.4 DCT, we wondered whether CDI–S_b analysis might uncover like DCT mechanisms in other Ca²⁺ channel subtypes, with yet broader distribution and impact. We considered a long splice variant of the human Ca_V1.3 channel¹⁰ (α_{1D–long[hum]}), which contains a DCT homologous to that in Ca_V1.4. This long variant has recently been reported to exhibit decreased CDI¹⁰ compared to a short variant akin to core channels (e.g., Fig. 3e). It thus seemed plausible that a competitive ICDI mechanism could extend to these channels, an important possibility given the wide distribution of Ca_V1.3 channels^13,14, and the predominance of the long variant throughout brain¹⁰. Complicating this view, however, were our prior observations that corresponding long and short variants of rat Ca_V1.3 channels exhibit no difference in CDI⁸. Indeed, all experiments to this point used the rat Ca_V1.3. Accordingly, we undertook CDI–S_b analysis of long Ca_V1.3 variants from both human and rat. The long Ca_V1.3 variant of human (Fig. 3f) clearly adhered to a competitive ICDI mechanism, with maximal CDI equivalent to that of core channels (cf., Fig. 3d). This argues that long forms of Ca_V1.3 and Ca_V1.4 channels do share a common ICDI mechanism. However, the CDI–S_b relation for the long Ca_V1.3 variant of humans differs quantitatively from that with the Ca_V1.4 DCT (cf., Figs. 3f, d), indicating that the strength of ICDI competition is customized by channel isoform. Indeed, the long Ca_V1.3 variant of rat exhibited extreme customization (Fig. 3g). Here, CDI–S_b analysis unmasks competitive inhibition, but the competition is weak enough that CDI remains maximal, except with overt chelation of apoCaM (at S_b ~ 0). The steep saturation of this CDI–S_b relation thus explains prior data that CDI was unaffected by the rat Ca_V1.3 DCT, as no depletion was used⁸. Importantly, the CDI–S_b curve for the long rat variant (Fig. 3g) is distinct from that for the Ca_V1.3 core (Fig. 3e), where CDI stayed maximal throughout. Hence, the rat Ca_V1.3 DCT entails customization, not elimination of the competitive inhibitory mechanism (Supplemental 3.4).

The similarity of DCT elements, particularly of human and rat Ca_V1.3, suggested that minute differences produce extremes of tuning. Indeed, we found that a single valine-to-alanine switch within the ICDI explains the difference (human:rat, ICDI position 47; Supplemental 3.5).

While the CDI–S_b analysis established a competitive inhibitory mechanism at the holochannel level, still critically unresolved was whether the ICDI/IQ peptide interactions studied thus far (Fig. 2c) were relevant to holochannel competition, especially given the multiplicity of CaM sites in channels^6,17,21,22. This is a generic challenge for large signaling complexes. Importantly, enzyme analysis could be extended to resolve even this ambiguity. Fig. 4a summarizes our data for competitive inhibition in holochannels, recasting CDI–S_b data into classic reciprocal-plot form²⁰, where channel CDI (f₅₀) corresponds to enzyme catalytic velocity V, and apoCaM to enzyme substrate S. Visual accord with the textbook signature of competition underscores the insights already provided by in situ holochannel biochemistry. Beyond this, if ICDI/IQ binding actually underlies holochannel competition, enzyme analysis additionally predicts a linear relation between the apparent apoCaM dissociation constant for holochannels (K_{d-channel-apparent}, Eq. 1) and the reciprocal of the peptide dissociation constant (1 / K_{d-FRET-peptide})

$$K_{\mathrm{d}­\mathrm{channel}­\mathrm{apparent}}=K_{\mathrm{d}­\mathrm{channel}}·[\mathrm{ICDI}]·(1/K_{\mathrm{d}­\mathrm{FRET}­\mathrm{peptide}})+K_{\mathrm{d}­\mathrm{channel}}$$ (2)

where K_d-channel is the holochannel dissociation constant for apoCaM without inhibitor, and [ICDI] is the effective concentration of ICDI at the channel preassociation site for apoCaM²⁰ (Supplemental 4.1) Conversely, if peptide interactions are peripheral or the inhibitory mechanism not strictly competitive, this relation will likely fail (Supplemental 4.2). To test this prediction, we noted that CDI–S_b analysis had already determined K_{d-channel-apparent} for channels with three different ICDIs (Figs. 3d, f, g; Fig. 4a). Also, K_{d-FRET-peptide} for the Ca_V1.4 ICDI was measured in Fig. 2a, and the remaining K_{d-FRET-peptide} values are deduced in Fig. 4b. The resulting linear plot (Fig. 4c) argues that ICDI/IQ binding indeed underlies holochannel competition, and apoCaM/channel preassociation involves the IQ with K_d-channel ~10 nM (Fig. 3e, Supplemental 4.1).

Figure 4. Molecular interactions and biology of competitive inhibitory tuning.

a, Reciprocal-plot representation of Fig. 3 (d-g) relations. b, FRET assays characterizing presumed peptide interactions underlying holochannel competition (as in a). c, Linear relation between holochannel and peptide competition parameters. d, Linear format of quantitative tuning relations. Green biological range, 99% boundaries²³. e, Consequences of connectivity in long variant of human Ca_V1.3. Natural variability of native CaM buffers (neuromodulin) affects CDI and Ca²⁺ entry for long variant of human Ca_V1.3. Left, CaM coexpression; middle, endogenous CaM; right, neuromodulin coexpression.

Transforming the reciprocal plots (Fig. 4a) into normal format raises diverse biological implications (Fig. 4d). The dogma has been that Ca²⁺ channels exhibit an ultra-strong apoCaM affinity⁸, ensuring maximal CDI over the green biological range²³, as confirmed for channels lacking ICDI (Fig. 4d, gray curve). Earlier reports that ICDI simply ‘switches off’ CDI^11,12 further promoted this perceived dissociation of CDI and apoCaM fluctuations. By contrast, we show here that the ICDI retunes CDI–[apoCaM] relations, so that natural variations of apoCaM modulate CDI and overall Ca²⁺ entry (Fig. 4d). Such interconnection opens new vistas, given the widespread impact and distribution of Ca_V1.3 and Ca_V1.4 channels^13-15, and the regulation of CaM²³. For example, coexpressing neuromodulin (a bio-molecule that impacts synaptic growth/remodeling/plasticity and buffers apoCaM²⁴) with the long variant of human Ca_V1.3 channels indeed lowers apoCaM sufficiently to eliminate CDI and promote Ca²⁺ entry (Fig. 4e, Supplemental 4.3). This outcome may bear on schizophrenia, where hippocampal neuromodulin is decreased²⁵. Moreover, neurodegenerative diseases are potentially affiliated with Ca²⁺ dysregulation and thereby altered apoCaM²⁶. In Parkinson’s, excess alpha-synuclein is pathogenic²⁷; these molecules bind apoCaM²⁸; and elevated substantia-nigral Ca_V1.3 activity predisposes for disease²⁹. In Alzheimer’s, CaM is depleted²⁶. More broadly, certain heart failure models feature elevated CaM³⁰. In all, exploring the (patho) physiological sequelae of Ca²⁺ channel connectivity with CaM now beckons at the frontier.

Methods

Molecular biology

Engineering of the rat Ca_V1.3 long variant (α_1D, AF370009.1) were made as follows. For Fig. 1, a unique XbaI site was introduced by PCR following the IQ domain. The DCT of human α_1F (NP005174) was amplified and cloned non-directionally via the unique XbaI site, yielding the sequence in Supplemental 1.1. For Fig. S3.4, a like process was repeated, except appropriate sections of the DCT of rat Ca_V1.3 long variant (α_1D, AF370009.1) were first PCR amplified with flanking SpeI and XbaI sites (compatible ends), and cloned into the aforementioned unique XbaI site, leaving a unique XbaI site after the inserted section of the rat DCT. Appropriate segments of the ICDI segment of the human Ca_V1.3 long variant (α_1D, NM000718) were then PCR amplified with flanking SpeI and XbaI sites, and cloned into the unique XbaI site, leaving a unique XbaI site after the inserted ICDI segment. For the V41A insertion (Fig. S3.4d), the human ICDI was point mutated via QuikChange^® Mutagenesis (Strategene) prior to PCR amplification and insertion into the channel construct. For the A41V insertion (Fig. S3.4e), the rat ICDI was similar point mutated before cloning into the unique XbaI site of the aforementioned engineered rat Ca_V1.3 long variant. For FRET 2-hybrid constructs, fluorophore-tagged CaM constructs were made as described⁶. Other FRET constructs made by replacing CaM with appropriate PCR amplified segments, via unique NotI and XbaI sites flanking CaM⁶. Details of CaM sponges in Supplemental 3.1. All segments subject to PCR or QuikChange^® were verified in their entirety by sequencing.

Transfection of HEK293 cells

For electrophysiology experiments, HEK293 cells were cultured in 10 cm plates, and channels were transiently transfected by a calcium phosphate protocol^6,8. We applied 8 μg of cDNA encoding the desired channel α₁ subunit, along with 8 μg of rat brain β_2a (M80545) and 8 μg of rat brain α₂δ (NM012919.2) subunits. β_2a minimized voltage inactivation, enhancing resolution of CDI. Additional cDNA was added as required in co-transfections. All of the above cDNA constructs were driven by a cytomegalovirus promoter. To enhance expression, cDNA for simian virus 40 T antigen (1-2 μg) was co-transfected. For fluorescence resonance energy transfer (FRET) 2-hybrid experiments, transfections and experiments were performed as described⁶. Electrophysiology and FRET were done at room temperature 1–2 d after transfection.

Whole-cell recording

Whole-cell recordings were obtained at room temperature using an Axopatch 200A amplifier (Axon Instruments). Electrodes were pulled with borosilicate glass capillaries (World Precision Instruments, MTW 150-F4), resulting in 1–3 MΩ resistances, before series resistance compensation of 80%. The internal solutions contained, (in mM): CsMeSO₃, 135; CsCl₂, 5; MgCl₂, 1; MgATP, 4; HEPES (pH 7.3), 5; and EGTA, 5; at 290 mOsm adjusted with glucose. The bath solution contained (in mM): TEA-MeSO₃, 140; HEPES, 10, pH 7.3; CaCl₂ or BaCl₂, 10; 300 mOsm, adjusted with glucose. These are as reported⁸. To augment currents for the full-length Ca_V1.4 experiments in Fig. 2e, we used 40 mM CaCl₂ or BaCl₂ in the bath solution, while adjusting TEA-MeSO₃ downwards to preserve osmolarity. As well, 5 μM Bay K 8644 was present in the bath throughout to further enhance currents.

FRET optical imaging

FRET 2-hybrid experiments were carried out in HEK293 cells and analyzed as described⁶. During imaging, the bath solution was either a Tyrode’s buffer containing 2 mM Ca²⁺, or the standard electrophysiological recording bath solution described above. Concentration-dependent spurious FRET was subtracted from the raw data prior to binding-curve analysis¹⁷. For simultaneous BSCaM_IQ imaging and patch-clamp recording, three-cube FRET measurements were obtained prior to whole-cell break in, and did not change appreciably thereafter. For ICDI binding curves in Figs. 2a and 4b, unlabelled IQ domain of neuromodulin (sequence in Supplemental 3.1) was coexpressed to reduce interference from endogenous CaM.