A modular switch for spatial Ca²⁺ selectivity in the calmodulin regulation of CaV channels

Abstract

Ca²⁺/calmodulin-dependent regulation of voltage-gated Ca_V1-2 Ca²⁺ channels exhibits extraordinary modes of spatial Ca²⁺ decoding and channel modulation^1–6, vital for many biological functions^6–9. A single calmodulin (CaM) constitutively associates with the channel carboxy tail^3,10–13, and Ca²⁺ binding to the C- and N-terminal lobes of CaM can each induce distinct channel regulations^2,14. As expected from close channel proximity, the C-lobe responds to the ~100 µM Ca²⁺ pulses driven by the associated channel^15,16, a behavior defined as ‘local Ca²⁺ selectivity.’ Conversely, all prior observations indicate that the N-lobe somehow senses the far weaker signals from distant Ca²⁺ sources^2,3,17,18. This ‘global Ca²⁺ selectivity’ satisfies a general signaling requirement, enabling a resident molecule to remotely sense cellular Ca²⁺ activity, which would otherwise be overshadowed by Ca²⁺ entry through the host channel^5,6. Here, we report that the spatial Ca²⁺ selectivity of N-lobe CaM regulation is not invariably global, but can be switched by a novel Ca²⁺/CaM binding site within the amino terminus of channels (NSCaTE, N-terminal Spatial Ca²⁺ Transforming Element). Native Ca_V2.2 channels lack this element, and display N-lobe regulation with a global selectivity. Upon introducing NSCaTE into these channels, spatial Ca²⁺ selectivity transforms from a global to local profile. Given this effect, we examine Ca_V1.2/Ca_V1.3 channels, which naturally contain NSCaTE, and find that their N-lobe selectivity is indeed local. Disruption of this element produces a global selectivity, confirming the native function of NSCaTE. Thus differences in spatial selectivity between advanced Ca_V1 and Ca_V2 channel isoforms are explained by the presence or absence of NSCaTE. Beyond functional effects, the position of NSCaTE on the channel amino terminus indicates that CaM can bridge the amino and carboxy termini of channels. Finally, the modularity of NSCaTE offers practical means to understand the basis of global Ca²⁺ selectivity¹⁹.

Figs. 1a, b display a prototypic example of global Ca²⁺ regulation induced by Ca²⁺ binding to the N-lobe of CaM, here driving Ca²⁺-dependent inactivation (CDI) of Ca_V2.2 channels³. Upon channel activation by a depolarizing voltage step (Fig. 1a), this CDI manifests as the accelerated decay of Ca²⁺ (middle, red trace) versus Ba²⁺ current (black trace). The baseline decay in Ba²⁺, which binds poorly to CaM², reflects a separate voltage-dependent inactivation process⁷. In population data (right), the fraction of peak Ca²⁺ current remaining after a 300 ms depolarization, r₃₀₀, is a U-shaped function of voltage (red circles), providing a hallmark of CDI²⁰. The corresponding Ba²⁺ relation (black circles) shows a weak monotonic decline, and the difference between Ca²⁺ and Ba²⁺ relations (f₃₀₀) quantifies pure CDI²⁰. In accord with a thus far invariant rule^3,6, N-lobe regulation of Ca²⁺ channels is selective for global Ca²⁺ elevation arising from spatially distant sources. Such global crosstalk is permitted under modest intracellular Ca²⁺ buffering¹⁸ that approximates physiological conditions (Fig. 1a, left, shading with 0.5 mM EGTA). Thus, strong buffering that localizes Ca²⁺ to channel nanodomains^15,16 (Fig. 1b, left, small hemisphere in 10 mM BAPTA) nearly eliminates CDI (Fig. 1b, middle and right).

Figure 1. Transformation of spatial Ca²⁺ selectivity in Ca_V2.2 channels.

a, Ca_V2.2 CDI, low buffering. Left, cartoon of global Ca²⁺ elevation. Middle, exemplar traces in Ca²⁺ (red) and Ba²⁺ (black). Throughout, current bar references Ca²⁺ trace; Ba²⁺ trace normalized to Ca²⁺ peak. Right, average CDI; f₃₀₀ at 10 mV; data, mean ± SEM throughout; cell number in parentheses. b, Ca_V2.2 CDI, high buffering. Left, cartoon of local Ca²⁺ signaling. c, CDI of cBBBBb. d, Localizing spatial Ca²⁺ transforming element (yellow highlight) within NT_C. Top, exemplar currents for cBBBBb with amino-terminal deletions, (Δ82cBBBBb, removal of first 81 aa of cBBBBb; Δ81bBBBBb, removal of entire amino terminus of Ca_V2.2). Bottom, f₃₀₀ (high buffering) versus residues deleted from cBBBBb; left axis and circles with colors corresponding to exemplars above. K_d,EFF is also plotted on same abscissa; blue right axis and blue squares, with filled symbol corresponding to e. Extreme bottom, localized sequence for both CDI transformation and Ca²⁺/CaM binding within NT_C (yellow highlight).

e, f, FRET for YFP–CaM versus channel amino-terminal segments tagged with CFP. Left, construct schematics. Middle, FRET ratios (FR). Right, exemplar binding curves, with red symbols for Ca²⁺/CaM, and blue symbols for apoCaM (e, arrow shows K_d,EFF).

Previous splice variations, mutations, and chimeras have at best altered the strength of such CDI^2,3,21, making global Ca²⁺ preference appear immutable. Here, however, when the amino terminus of Ca_V1.2 channels (NT_C) was substituted into Ca_V2.2, the resulting ‘cBBBBb’ chimera exhibited strong N-lobe CDI in high buffering (Fig. 1c, Supplementary Information 1.1). Introducing NT_C into Ca_V2.1 produced an analogous result (data not shown), thus generalizing the effect across the Ca_V2 family. This conversion to local Ca²⁺ selectivity is unprecedented, especially given the preponderance of known structural determinants for CDI on the carboxy termini of channels²².

To explore the basis of this effect, we undertook progressive amino-terminal deletions from the NT_C segment within the cBBBBb chimera. Deleting the first 81 residues (Δ82cBBBBb) completely spared CDI in high buffering (Fig. 1d, top, left), where the gray trace shows the full-length NT_C profile. By contrast, removing just one more residue (82W) significantly reduced CDI (Fig. 1d, top, middle), and additional deletion further suppressed CDI (Fig. 1d, top, right). Explicit correlation of CDI strength with NT_C deletion corroborated these trends (Fig. 1d, bottom, left axis, circles), localizing the impact to a short contiguous region (yellow highlight).

What could be the essential mechanistic ingredient of this locus? We reasoned that this segment might orchestrate special molecular interactions with cytoplasmic channel loops or modulatory ligands, and thus probed for such associations using a live-cell FRET two-hybrid assay¹³. Using this approach, with EYFP fused to CaM (EYFP–CaM), and ECFP to various portions of NT_C or the Ca_V2.2 amino terminus (NT_B), we found a unique ability of NT_C to bind Ca²⁺/CaM within the intracellular milieu of HEK293 cells (Fig. 1e). To start, NT_B failed to display FRET interaction with CaM, either in Ca²⁺-free or Ca²⁺-bound states (Fig. 1e, middle column, topmost bars). Specifically, the optical parameter FR was ~1, signifying the absence of FRET¹³. By contrast, NT_C clearly interacted with Ca²⁺/CaM (FR ~2), but not apoCaM (Ca²⁺-free CaM). Trisection of NT_C localized Ca²⁺/CaM binding to the middle third (Fig. 1e, middle column, residues 67–131), and further deletions identified a sharp decrease in FRET upon removal of W82 (Supplementary Information 1.2). Since FR depends on the fractional binding between interacting partners, cell-to-cell variation in expression permitted resolution of binding curves, each specifying a relative dissociation constant K_d,EFF¹³ (Fig. 1e, right column, arrow). Alignments of these NT_C fragments and their K_d,EFF values (Fig. 1d, squares, Supplementary Information 1.2) localized a CaM-binding segment that coincides well with the critical CDI segment (Fig. 1d, yellow highlight). Notably, a W82A point mutation suppressed Ca²⁺/CaM interaction with NT_C (Fig. 1f), fitting with the drop in CDI of the corresponding channel truncation (Fig. 1d, top row, middle). Thus, the transformation of spatial Ca²⁺ selectivity correlates with a novel Ca²⁺/CaM interaction site within NT_C.

This connection was deepened in two ways (Fig. 2). First, the in situ FRET approach was confirmed by in vitro assays, pairing purified Ca²⁺/CaM with a synthetic peptide (SWQAAIDAARQAKLMGSA) spanning the key NT_C binding region. Without peptides, CaM migrated rapidly under non-denaturing PAGE² (Fig. 2a, left gel, lane 1, open triangle). When fully complexed with the IQ domain peptide of Ca_V2.1 (IQ_A), CaM exhibited a known slowing of mobility² (lane 2, filled triangle). Increasing NT_C-peptide produced progressive conversion to the faster mobility species (lanes 3–8), indicating competitive binding to CaM. By contrast, a mutant W82A peptide (NT_{C-peptide-(W82A)}, right gel) was unable to bind CaM. More quantitatively, NT_C-peptide depressed the emission spectrum of dansylated CaM (Fig. 2b, inset), allowing resolution of a 1:1 binding curve with a K_d of 1.2 µM (Fig. 2b, black symbols). NT_{C-peptide-(W82A)} exhibited no such interaction (Fig. 2b, open symbols). These in vitro assays established direct Ca²⁺/CaM binding to the NT_C core region.

Figure 2. Direct CaM binding and mapping of key NSCaTE residues.

a, Competitive gel mobility shift assay, confirming Ca²⁺/CaM interaction with NT_C-peptide (left), not mutant NT_{C-peptide-(W82A)} (right). b, Dansylated Ca²⁺/CaM spectrofluorometry. Binding with NT_C-peptide (filled circles), not with mutant NT_{C-peptide-(W82A)} (open circles), both plotted as mean ± SEM. Inset, NT_C-peptide decreases emission spectrum. c, CDI in high buffering, for Δ78cBBBBb channels with point mutations as labeled. Format as in Fig. 1a; gray trace shows baseline Δ78cBBBBb profile. d, Close correlation of CDI in high buffering with NT_C module affinity for Ca²⁺/CaM₃₄ (Supplemmentary Information 2.1).

Second, further in-depth mapping refined the correlation between spatial Ca²⁺ selectivity and CaM binding to NT_C. Alanine point mutations were introduced into the NT_C domain of cBBBBb channels, targeting key sites inferred from a preliminary report¹⁹. These manipulations reduced CDI (in high buffering) with the rank order: W82A > I86A > R90A (Fig. 2c). Reassuringly, these changes in CDI (f₃₀₀) correlated closely with the relative dissociation constant (K_d,EFF) for CaM interaction with corresponding mutant NT_C peptides, as determined by FRET (Fig. 2d, Supplementary Information 2.1). Given the strong function and affiliation with CaM binding to the core NT_C locus, we named this element NSCaTE (Fig. 2c, top N-terminal Spatial Ca²⁺ Transforming Element).

While this module impacts chimeric channels, does NSCaTE function within naturally occurring channels? Alignments of various Ca_V1-2 Ca²⁺ channels (Fig. 3a) revealed that NSCaTE is present not only in Ca_V1.2 (α_1C), but also Ca_V1.3 (α_1D). Our attention was foremost drawn to NSCaTE in Ca_V1.3, as these channels exhibit two distinct and robust forms of CDI, each selectively triggered by Ca²⁺ binding to a different lobe of CaM¹⁴. As a preliminary test for the functionality of the Ca_V1.3 NSCaTE, we checked for Ca²⁺/CaM interaction (Fig. 3b). FRET assays showed selective Ca²⁺/CaM interaction with the midsection of the Ca_V1.3 amino terminus (Fig. 3b, NT_D-(35–94)), yielding a K_d,EFF on par with NT_C analogs (Fig. 1e). Consistent with W82A effects in NT_C (Fig. 1f), mutating the analogous tryptophan in NT_D (NT_D-W44A-(35–94)) eliminated Ca²⁺/CaM interaction (Fig. 3b, Supplementary Information 3.1). Substituting NT_D into Ca_V2.2 (yielding the dBBBBb chimera) also produced strong CDI in high buffering (Fig. 3c, Supplementary Information 3.1), confirming the transforming capacity of the Ca_V1.3 NSCaTE.

Figure 3. NSCaTE transforms spatial Ca²⁺ selectivity in native Ca_V1 channels.

a, α₁ alignment, Ca_V1 channels. b, FRET assays. Format as in Fig. 1e. c, CDI of dBBBBb in high buffering. d, Impact of native NSCaTE upon N-lobe CDI of Ca_V1.3. Top, CDI persists in high buffering. Middle, CDI elimination by W44A mutation. Bottom, CDI reappearance in W44A mutant under low buffering. Format as in Fig. 1a, except for 800-msec metrics. e, Native NSCaTE effects in Ca_V1.2. Format as in d, with 300-msec metrics.

Encouraged by these preliminary results, we asked whether the intrinsic NSCaTE imparts local selectivity to the native Ca_V1.3, contrary to the current dogma that the N-lobe invariably senses global Ca²⁺. In Ca_V1.3, the N-lobe form of CDI can be isolated by coexpresing a mutant CaM (CaM₃₄) in which Ca²⁺ binding is restricted to the N-lobe^2,20. Thus far, such CDI has only been studied under modest Ca²⁺ buffering¹⁴. Here, even under high buffering, N-lobe CDI was pronounced (Fig. 3d, top row). More telling, a W44A mutation within Ca_V1.3 nearly eliminated this CDI (middle row), as did deletion of this region (Supplementary Information 3.2). To exclude indiscriminate disruption of CDI, we confirmed that CDI resurfaced under modest buffering permissive of global signaling (Fig. 3d, bottom row, Supplementary Information 3.2). Importantly, CDI was entirely CaM-mediated under reduced buffering¹⁴ (Supplementary Information 3.3); and C-lobe CDI was unaffected by like mutations¹⁹. Hence, the native Ca_V1.3 NSCaTE endows N-lobe CDI with a local selectivity, and reducing Ca²⁺/CaM binding to NSCaTE switches this selectivity towards a global profile.

Returning to native Ca_V1.2, coarse identification of Ca²⁺/CaM binding to the amino terminus of these channels has been reported, but the functional consequences have been unclear²³. In particular, the CaM-mediated CDI of these channels has been attributed only to the C-lobe of CaM^7,20; apparently absent has been an N-lobe form of CDI, the target of NSCATE effects. Here, however, upon co-transfection of Ca_V1.2 with CaM₃₄, to fully isolate a potential N-lobe component, small but unmistakable CDI is seen during prolonged 1-sec depolarization, even in high buffering (Fig. 3e, top row). As well, a W82A mutation abolished this CDI (middle row), but spares it under global Ca²⁺ signaling (bottom row). A more drastic deletion of the native NSCaTE produced an identical effect (Supplementary Information 3.4). Again, the globally signaled CDI remains entirely CaM dependent (Supplementary Information 3.5). Hence, NSCaTE functions similarly in native Ca_V1.2 and Ca_V1.3; the weaker N-lobe CDI of Ca_V1.2 is expected from its lower open probability^19,24. These results generalize the lessons of chimeric channels to the operation of native channels and constitute the first examples where the N-lobe of CaM shows local Ca²⁺ selectivity.

In all, the traditionally global spatial Ca²⁺ selectivity of channel regulation by the N-lobe^3,6 is not invariant, but can be transformed towards a local selectivity by NSCaTE, a novel Ca²⁺/CaM binding site unrecognized by current motif-detection algorithms²⁵. Four perspectives emerge. First, the presence of NSCaTE in Ca_V1.2 and Ca_V1.3—along with its absence in Ca_V2 channels—underlies the contrasting spatial Ca²⁺ preferences of N-lobe CDI between these two channel clades (Fig. 4a). Previously, N-lobe regulation had been viewed as always having a global Ca²⁺ selectivity^3,6, which is true for Ca_V2 channels (Fig. 4b). We now recognize that the analogous N-lobe CDI of Ca_V1.2/Ca_V1.3 channels¹⁴ exhibits local selectivity, owing to an intrinsic NSCaTE (Fig. 4c). Second, NSCaTE promises vital adjustments of spatial selectivity in diverse biological contexts. NSCaTE would customize Ca_V1.2/1.3 regulation for local Ca²⁺ signaling in cardiac dyads^4,7, and for neuronal Ca²⁺ entry whose precise regulation is essential for long-term synaptic plasticity^26,27. Of further intrigue, a naturally occurring Ca_V1.2 splice variant features a premature stop codon just after the channel amino terminus²¹. The predicted protein product contains NSCaTE, and may exert a dominant negative effect. Although coexpressing NT_C and Ca_V1.3 left CDI unchanged in HEK293 cells (data not shown), additional mechanisms may exist in native tissues to render this splice product active. Third, unlike most functional motifs, which are pervasive from bacteria through advanced mammals²⁵, NSCaTE is only found in a subset of Ca_V1 channels from advanced species (Fig. 4a, red), and in multiple bacterial proteins (Supplementary Information 4). A prokaryotic NSCaTE from Xanthomonas may in fact bind endogenous CaM-like molecules (Supplementary Information 4), hinting that channel and bacterial motifs share a common heritage. Finally, NSCaTE furnishes valuable structural and mechanistic insights. For structure, we deduce that CaM can bridge the carboxy and amino termini of Ca_V1.2/Ca_V1.3 channels (Fig. 4c, right), as follows. This study establishes Ca²⁺-bound N-lobe interaction with NSCaTE on the channel amino terminus. By contrast, prior reports emphasize that regulation by the C-lobe involves its Ca²⁺-bound and Ca²⁺-free interactions with the channel carboxy terminus^12,20,28. Moreover, a single resident CaM orchestrates both C- and N-lobe regulation^10,11, where both coexist in Ca_V1.2/Ca_V1.3 channels^14,20. Finally, channel amino and carboxy termini are in close proximity, and move with channel gating²⁹. Hence, this bridging appears probable, expanding a theme where the lobes of CaM crosslink separate parts of a molecule³⁰. As for mechanism, the modularity of NSCaTE indicates that spatial Ca²⁺ preference is not hopelessly intertwined within holochannel structure. Manipulating this compact element promises valuable insights into the elusive mechanism underlying spatial Ca²⁺ selectivity^6,19.

Figure 4. Functional and structural properties for NSCaTE switching of spatial Ca²⁺ selectivity.

a, Ca_V1-2 dendrogram, based on α₁ subunits (Supplementary Information 4.1). Red, NSCaTE-containing channels. b, Schematic of N-lobe CDI in Ca_V2 channels. Absence of NSCaTE yields global Ca²⁺ selectivity. Black circles on CaM, Ca²⁺ ions. c, N-lobe CDI in Ca_V1 channels that contain NSCaTE (red) exhibit local Ca²⁺ selectivity. CaM bridges amino and carboxy termini of inactivated channels (right).

Methods

Molecular Biology

The α_1cBBBBb subunit was the published α_1cBBBBb/pcDNA3 construct³¹. PCR was used to truncate the amino terminus of this construct: the forward primer, containing a unique EcoRI site and Kozak start, determined the extent of truncation; the reverse primer annealed downstream of a unique AgeI site. PCR fragments were ligated into α_1cBBBBb/pcDNA3 via EcoRI/AgeI. Mutations to NSCaTE were introduced into the α_1Δ78cBBBBb truncation construct (starts with 78th residue of α_1C), using forward primers as above, but containing desired mutations. For chimeric α_1dBBBBb, PCR introduced a silent and unique BsiWI site at residue 83 of α_1B/pcDNA3, yielding α_1B_BsiWI⁺. The amino terminus of α_1D³² was PCR amplified, then ligated into unique EcoRI and BsiWI sites of α_1B_BsiWI⁺, yielding α_1dBBBBb. The α_1B (Ca_V2.2) used here is a variant of the human α_1B reported previously³³; it is identical but for V354M and R369S variations. This variant backbone was used for all constructs, including α_1dBBBBb and the previously published α_1cBBBBb/pcDNA3³¹. For completeness, we confirmed that transforming NSCaTE effects were also present using the more common α_1B backbone for α_1cBBBBb and α_1dBBBBb. A 141-residue amino-terminal truncation of the Ca_V1.2 α_1C subunit (α_1C[NTΔ]) was generated by PCR amplifying residues 142–611 from α_1C³⁴, followed by ligation into the parental α_1C via unique KpnI and StuI sites. A 44-residue amino-terminal truncation of the Ca_V1.3 α_1D subunit (α_1D[NTΔ]) was similarly constructed by PCR amplifying residues 45–180 from α_1D³², followed by ligation into α_1D via unique NheI and BsiWI sites. The mutant α_1C[W82A] was made by PCR amplifying α_1C residues 1–87, using a reverse primer containing a W82A mutation. This product was ligated into α_1C via unique HindIII and ClaI sites. The mutant α_1D[W44A] subunit was as described¹⁹.

For FRET, all CFP–NT_B/C/D constructs were made by PCR of desired segments of α_1cBBBBb or α_1D, followed by substitution of CaM_WT in a published CFP–CaM_WT/pcDNA₃ clone¹³, using unique Not I and Xba I sites. For CFP fusion to the NSCaTE of the TonB Dependent Receptor (TBDR, amino acids 464–484 of YP_242632.1), annealed synthetic primers encoding this motif (mammalian optimized) were ligated into unique NotI/XbaI sites of CFP–CaM_WT/pcDNA₃. Throughout, all PCR amplified segments were fully verified by sequencing.

Electrophysiology

HEK293 cells were transiently transfected using a calcium phosphate protocol²⁰. Cells were co-transfected with 8 µg of rat brain β_2a³⁵, 8 µg of rat brain α_2δ³⁶, and 1–8 µg of Ca²⁺ channel α₁ subunit. All α_1C and α_1D constructs were co-transfected with 2 µg of SV40 T antigen to enhance expression; 8 µg of cDNA for rat brain CaM₁₂, CaM₃₄, or CaM₁₂₃₄ was added as required²⁰.

Room temperature, whole-cell recordings were performed 1–3 days after transfection, using Axopatch 200A/B amplifiers (Axon Instruments). P/8 leak subtraction was used, with series resistances of 1–2 MΩ after >70% compensation. Currents were filtered at 2 kHz (4-pole Bessel), and sampled at 10 kHz. For high-buffer experiments, internal solutions contained, (in mM): CsMeSO₃, 114; CsCl₂, 5; MgCl₂, 1; MgATP, 4; HEPES (pH 7.4), 10; and BAPTA, 10; at 295 mOsm adjusted with CsMeSO₃. For low-buffer experiments, (in mM): CsMeSO₃, 135; CsCl₂, 5; MgCl₂, 1; MgATP, 4; HEPES (pH 7.4), 10; and EGTA, 0.5. In all experiments, except those with Ca_V1.2/1.3, external solutions contained (in mM): TEA-MeSO₃, 140; HEPES (pH 7.4), 10; and CaCl₂ or BaCl₂, 5; at 300 mOsm, adjusted with TEA-MeSO₃. Due to lower Ca_V1.2 expression, 20 mM CaCl₂ or BaCl₂ was used during high internal buffering, and 10 mM during low buffering. Ca_V1.3 recordings were made in 40 mM CaCl₂ or BaCl₂. A holding potential of −90 mV and 60 s repetition interval were used throughout. Data were analyzed by custom MATLAB software (Mathworks, MA), with average data shown as mean ± SEM.

FRET two-hybrid assay

Fluorescence resonance energy transfer (FRET) two-hybrid experiments were carried out in HEK293 cells and analyzed as described¹³. During imaging, the bath solution was a Tyrode’s buffer containing 10 mM Ca²⁺, where 5 µM ionomycin (Sigma-Aldrich, MO) was added for Ca²⁺/CaM experiments. Concentration-dependent spurious FRET was subtracted from the raw data prior to binding-curve analysis³⁷.

Gel mobility shift assay

We used a non-denaturing, 13% acrylamide resolving gel, coupled with a 4% acrylamide stacking gel. 100 µM CaCl₂ was added to the gel to maintain Ca²⁺/CaM binding. Peptide samples were freshly solubilized into a sample buffer: 20 mM Tris-Cl (pH=6.8), 150 mM NaCl, 2% glycerol, 10 mM CaCl_2, 0.001% bromphenol blue dye, and 2 µg purified CaM protein⁷. Binding reactions (×1h) and electrophoresis were performed at ~0–4°C in 25 mM Tris (pH=8.4) and 200 mM glycine, with parameters of 50 V × 20 min, followed by 100 V × 2 h. Gels were stained over heat × 1 h with Commassie R-250 (Bio-Rad Laboratories, CA), and destained overnight²⁰. Peptides were synthesized by the Synthesis and Sequencing Facility at the Johns Hopkins School of Medicine (Baltimore, MD). The wildtype NSCaTE peptide NT_C-peptide was SWQAAIDAARQAKLMGS (rabbit α_1C sequence), and the mutant was SAQAAIDAARQAKLMGS. The IQ_A peptide was KIYAAMMIMEYYRQSKAKKLQ (based on human α_1A subunit²). IQ_A peptide was added to Ca²⁺/CaM in a molar ratio of 8:1 to produce a maximal binding shift. NSCaTE peptide was then added in varying amounts (Fig. 2a).

Dansyl CaM experiments

Purified recombinant CaM protein was dansylated with 0.5 mM dansyl-Cl under alkaline conditions at room temperature (0.3 mM CaCl₂ × 4 hours). The reaction was terminated with 5 mM hydroxylamine, and Tris buffer added before dialysis against 10 mM MOPS (pH 7.2) at 4°C³⁸. Peptides were as above. Spectrofluorometry was conducted at 25°C in 5 mM CaCl₂, 150 mM NaCl, and 50 mM Tris-Cl (pH=7.5). Peptide stocks were freshly prepared and used within 6 h. 100 nM dansyl-CaM was excited at 340 nm, and fluorescence recorded from 400–650 nm (slit size, 3 nm excitation, 15 nm emission; integration time, 0.1 s), using a Fluorolog-3 spectrofluorometer (Horiba Jobin Yvon). No detectable bleaching was observed in controls. Binding curves were determined from the average of five 490-nm measurements (F₄₉₀) at each peptide concentration. Subtraction of background fluorescence utilized a correction factor (m_peptide) derived from measurements of peptide alone (F_{490,bg-corrected} = F₄₉₀ – m_peptide · [peptide]). We calculated F_{490,full-corrected} = F_{490,bg-corrected} · V_tot / V_initial (where V_tot is the final volume, and V_initial was 2.5 mL), so as to correct for dilution by successive peptide additions. Fractional binding (B) was (F_{490,full-corrected,max} - F_{490,full-corrected}) / (F_{490,full-corrected,max} - F_{490,full-corrected,min}); least-square fits were performed using B = [peptide]^k / ([peptide]^k + K_d^k), with resulting k ~1.