The Effect of Ca2+, Lobe-Specificity, and CaMKII on CaM Binding to NaV1.1

Calmodulin (CaM) is well known as an activator of calcium/calmodulin-dependent protein kinase II (CaMKII). Voltage-gated sodium channels (VGSCs) are basic signaling molecules in excitable cells and are crucial molecular targets for nervous system agents. However, the way in which Ca2+/CaM/CaMKII cascade modulates NaV1.1 IQ (isoleucine and glutamine) domain of VGSCs remains obscure. In this study, the binding of CaM, its mutants at calcium binding sites (CaM12, CaM34, and CaM1234), and truncated proteins (N-lobe and C-lobe) to NaV1.1 IQ domain were detected by pull-down assay. Our data showed that the binding of Ca2+/CaM to the NaV1.1 IQ was concentration-dependent. ApoCaM (Ca2+-free form of calmodulin) bound to NaV1.1 IQ domain preferentially more than Ca2+/CaM. Additionally, the C-lobe of CaM was the predominant domain involved in apoCaM binding to NaV1.1 IQ domain. By contrast, the N-lobe of CaM was predominant in the binding of Ca2+/CaM to NaV1.1 IQ domain. Moreover, CaMKII-mediated phosphorylation increased the binding of Ca2+/CaM to NaV1.1 IQ domain due to one or several phosphorylation sites in T1909, S1918, and T1934 of NaV1.1 IQ domain. This study provides novel mechanisms for the modulation of NaV1.1 by the Ca2+/CaM/CaMKII axis. For the first time, we uncover the effect of Ca2+, lobe-specificity and CaMKII on CaM binding to NaV1.1.


Introduction
Voltage-gated sodium channels (VGSCs) are basic signaling molecules in excitable cells and are molecular targets for local anesthetic agents and antiepileptic agents [1,2]. So far, ten isoform members have been identified-Na V 1.1-Na V 1.9 and NaX-forming the VGSCs superfamily [3,4] in which Na V 1.1 Previous research have shown Na V 1.1 IQ domain to bind with CaM [15,29]. Therefore, we first confirmed the binding property of CaM to IQ. As shown in Figure 1, the binding of CaM to IQ was successfully detected, and the molecular weight of GST-IQ (glutathione Sepharose transferaseisoleucine and glutamine) and CaM was 31.98 and 16.7 kDa corresponding to the marker, respectively. As shown in Figure 2B, the binding of CaM to IQ was detected at ≈free, 100 nM, 500 nM, and 2 mM [Ca 2+ ]. The summarized data from the densitometer analyses of replicate gels are shown in Figure 2C and Table 1. The maximal binding estimated B max of CaM to IQ was 2.06 (B max1 + B max2 ), 0.66 (B max1 + B max2 ), 1.08 (B max1 + B max2 ), and 1.38 (B max1 + B max2 ) mol/mol (CaM/IQ) at ≈free, 100 nM, 500 nM, and 2 mM [Ca 2+ ], respectively (n = 4), indicating that apoCaM had the highest affinity with IQ domain. However, the binding of CaM to IQ domain was in a Ca 2+ -dependent manner in the presence of Ca 2+ . The binding affinity estimated as K d value also showed a Ca 2+ -dependent increase in the presence of Ca 2+ (Table 1). Our data showed the binding of Ca 2+ /CaM to IQ was in a concentration-dependent and Ca 2+ -dependent manner, but apoCaM had the highest affinity to Na V 1.1 IQ domain.
Our previous study had examined the effect of I (isoleucine)/E (glutamic acid) mutation on the IQ domain of Ca V 1.2 on the CaM binding to this domain. We had found that the mutation completely abolished CaM binding and confirmed that I1653 in the IQ domain was important for the interaction with CaM [33]. In this study, we mutated I1922 and Q1923 in Na V 1. 1 IQ domain [ 1909TLKRKQEEVSAVIIQRAYRRHLLKRTVK 1936 ] into E ( Figure 2A). As shown in Figure 2D,F, the binding of CaM to EQ (I1922E) and IE (Q1923E) was diminished, confirming that I1922 (and Q1923) were the core amino acids in Na V 1.1 IQ domain for the binding with CaM. The summarized data from the densitometer analyses of replicate gels are shown in Figure 2E,G. with CaM [33]. In this study, we mutated I1922 and Q1923 in NaV1. 1 IQ domain [ 1909TLKRKQEEVSAVIIQRAYRRHLLKRTVK 1936 ] into E ( Figure 2A). As shown in Figure 2D,F, the binding of CaM to EQ (I1922E) and IE (Q1923E) was diminished, confirming that I1922 (and Q1923) were the core amino acids in NaV1.1 IQ domain for the binding with CaM. The summarized data from the densitometer analyses of replicate gels are shown in Figure 2E,G.    with CaM [33]. In this study, we mutated I1922 and Q1923 in NaV1. 1 IQ domain [ 1909TLKRKQEEVSAVIIQRAYRRHLLKRTVK 1936 ] into E ( Figure 2A). As shown in Figure 2D,F, the binding of CaM to EQ (I1922E) and IE (Q1923E) was diminished, confirming that I1922 (and Q1923) were the core amino acids in NaV1.1 IQ domain for the binding with CaM. The summarized data from the densitometer analyses of replicate gels are shown in Figure 2E,G.   To further clarify the regulatory mechanism of CaM on Na V 1.1 channel, we then examined the binding properties of CaM mutants to Na V 1.1 IQ domain. This included CaM 12 and CaM 34 ( Figure 3A) in which Ca 2+ -binding to its N-and C-lobe was eliminated, respectively, and a Ca 2+ -insensitive CaM mutant (CaM 1234 ) ( Figure 3A) [26]. As shown in Figure 3B,C, the binding of CaM 12 to IQ was qualitatively similar to that of the wild-type (wt) CaM. The maximal binding estimated as B max was 1.27, 0.70, 0.82, and 0.91 mol/mol CaM 12 /IQ (n = 4) at ≈free, 100 nM, 500 nM and 2 mM Ca 2+ , respectively, showing an obvious Ca 2+ dependence in the presence of Ca 2+ (Table 1). It is interesting to note that the B max of CaM 12 at ≈free Ca 2+ is~30% was greater than that at 2 mM Ca 2+ , suggesting that like wt CaM, CaM 12 has the highest affinity for IQ in the absence of Ca 2+ . To further clarify the regulatory mechanism of CaM on NaV1.1 channel, we then examined the binding properties of CaM mutants to NaV1.1 IQ domain. This included CaM12 and CaM34 ( Figure  3A) in which Ca 2+ -binding to its N-and C-lobe was eliminated, respectively, and a Ca 2+ -insensitive CaM mutant (CaM1234) ( Figure 3A) [26]. As shown in Figure 3B,C, the binding of CaM12 to IQ was qualitatively similar to that of the wild-type (wt) CaM. The maximal binding estimated as Bmax was 1.27, 0.70, 0.82, and 0.91 mol/mol CaM12/IQ (n = 4) at ≈free, 100 nM, 500 nM and 2 mM Ca 2+ , respectively, showing an obvious Ca 2+ dependence in the presence of Ca 2+ (Table 1). It is interesting to note that the Bmax of CaM12 at ≈free Ca 2+ is ~30% was greater than that at 2 mM Ca 2+ , suggesting that like wt CaM, CaM12 has the highest affinity for IQ in the absence of Ca 2+ .
Next, we examined the binding of CaM34 to IQ domain. As shown in Figure 3D,E, the binding of CaM34 to IQ was also concentration-dependent. The parameters obtained (Table 1) revealed that the maximal binding estimated as Bmax was 0.79, 0.57, 1.33, and 0.79 mol/mol CaM34/ IQ (n = 4) at ≈free, 100 nM, 500 nM, and 2 mM Ca 2+ , respectively. It was noted that this profile of [Ca 2+ ] dependence was different from those of wt CaM and CaM12.  Next, we examined the binding of CaM 34 to IQ domain. As shown in Figure 3D,E, the binding of CaM 34 to IQ was also concentration-dependent. The parameters obtained (Table 1) revealed that the maximal binding estimated as B max was 0.79, 0.57, 1.33, and 0.79 mol/mol CaM 34 / IQ (n = 4) at ≈free, 100 nM, 500 nM, and 2 mM Ca 2+ , respectively. It was noted that this profile of [Ca 2+ ] dependence was different from those of wt CaM and CaM 12 .
Furthermore, we examined the binding of CaM 1234 to Na V 1.1 IQ domain. As shown in Figure 3F, the Ca 2+ dependence was totally diminished because of the mutation in four Ca 2+ binding sites. As shown in Figure 3G and Table 1, the B max were 1.06, 1.02, 1.06, and 1.03 mol/mol (CaM 1234 /IQ) at ≈free, 100 nM, 500 nM, and 2 mM [Ca 2+ ], respectively. The values of B max and K d were not significantly different among different Ca 2+ concentrations, confirming the Ca 2+ -insensitive nature of CaM 1234 . Our data showed that functional Ca 2+ binding sites of either N-or C-lobe were required for Ca 2+ dependency in CaM binding to Na V 1.1 IQ domain.

Binding of Individual N-Lobe or C-Lobe of CaM to Na V 1.1 IQ Domain
In order to study the effect of specific lobe of CaM on the Na V 1.1 IQ, we first computationally investigated the interactions between N-lobe or C-lobe of CaM and Na V 1.1 IQ domain using Discovery Studio 2017. As shown in Figure 4A,B, the ZDock score and E_RDock for the optimal N-lobe orientation docking into Na V 1.1 IQ domain were 11.28 and −23.31 kcal/mol, respectively, whereas these parameters for the best interaction of C-lobe with IQ domain were 10.66 and −26.21 kcal/mol, respectively. In addition, as shown in Figure 4C, when the N-and C-lobe were docked together into the Na V 1.1 IQ domain, the best ZDock score and lowest E_RDock were 12.78 and −30.72 kcal/mol, respectively. Furthermore, we examined the binding of CaM1234 to NaV1.1 IQ domain. As shown in Figure 3F, the Ca 2+ dependence was totally diminished because of the mutation in four Ca 2+ binding sites. As shown in Figure 3G and Table 1, the Bmax were 1.06, 1.02, 1.06, and 1.03 mol/mol (CaM1234/IQ) at ≈free, 100 nM, 500 nM, and 2 mM [Ca 2+ ], respectively. The values of Bmax and Kd were not significantly different among different Ca 2+ concentrations, confirming the Ca 2+ -insensitive nature of CaM1234. Our data showed that functional Ca 2+ binding sites of either N-or C-lobe were required for Ca 2+ dependency in CaM binding to NaV1.1 IQ domain.

Binding of Individual N-Lobe or C-Lobe of CaM to NaV1.1 IQ Domain
In order to study the effect of specific lobe of CaM on the NaV1.1 IQ, we first computationally investigated the interactions between N-lobe or C-lobe of CaM and NaV1.1 IQ domain using Discovery Studio 2017. As shown in Figure 4A,B, the ZDock score and E_RDock for the optimal Nlobe orientation docking into NaV1.1 IQ domain were 11.28 and −23.31 kcal/mol, respectively, whereas these parameters for the best interaction of C-lobe with IQ domain were 10.66 and −26.21 kcal/mol, respectively. In addition, as shown in Figure 4C, when the N-and C-lobe were docked together into the NaV1.1 IQ domain, the best ZDock score and lowest E_RDock were 12.78 and −30.72 kcal/mol, respectively.   Table 2, the maximal bindings estimated as B max for N-lobe were 0.18, 0.16, 0.31, and 0.17 mol/mol (N-lobe/IQ) at ≈free, 100 nM, 500 nM, and 2 mM [Ca 2+ ], respectively (n = 4). Thus the profile of [Ca 2+ ] dependence of N-lobe was similar to that of CaM 34 .
We then examined the binding of C-lobe to IQ domain. As shown in Figure 5D, like wt CaM, C-lobe also had the highest affinity with IQ at ≈free [Ca 2+ ]. The maximal binding presented by B max were 0.32, 0.13, 0.13, and 0.15 mol/mol (C-lobe/IQ) at ≈free, 100 nM, 500 nM, and 2 mM [Ca 2+ ], respectively (n = 4) ( Figure 5E and Table 2). Thus, the profile of [Ca 2+ ] dependence of C-lobe was similar to those of wt CaM and CaM 12 . and Table 2, the maximal bindings estimated as Bmax for N-lobe were 0.18, 0.16, 0.31, and 0.17 mol/mol (N-lobe/IQ) at ≈free, 100 nM, 500 nM, and 2 mM [Ca 2+ ], respectively (n = 4). Thus the profile of [Ca 2+ ] dependence of N-lobe was similar to that of CaM34. We then examined the binding of C-lobe to IQ domain. As shown in Figure 5D, like wt CaM, Clobe also had the highest affinity with IQ at ≈free [Ca 2+ ]. The maximal binding presented by Bmax were 0.32, 0.13, 0.13, and 0.15 mol/mol (C-lobe/IQ) at ≈free, 100 nM, 500 nM, and 2 mM [Ca 2+ ], respectively (n = 4) ( Figure 5E and Table 2). Thus, the profile of [Ca 2+ ] dependence of C-lobe was similar to those of wt CaM and CaM12. The parameters (Table 2) revealed that binding of both N-and C-lobe to IQ was also in a Ca 2+dependent manner. Kd of N-lobe was lower than that of C-lobe in the presence of Ca 2+ ([Ca 2+ ] ≥ 100 The parameters (Table 2) revealed that binding of both N-and C-lobe to IQ was also in a Ca 2+ -dependent manner. K d of N-lobe was lower than that of C-lobe in the presence of Ca 2+ ( [Ca 2+ ] ≥ 100 nM), whereas K d of C-lobe was lower than that of N-lobe in the absence of Ca 2+ . Additionally, B max of N-lobe was higher than that of C-lobe in the presence of Ca 2+ , whereas B max of C-lobe was higher than that of N-lobe in the absence of Ca 2+ , indicating that C-lobe was the predominant domain in apoCaM interacting with Na V 1.1 IQ domain, and N-lobe was the predominant domain in Ca 2+ /CaM interacting with Na V 1.1 IQ domain. In the previous study, several CaMKII-mediated phosphorylation sites on Na V 1.1 have been identified [29]. To clarify the modulation of CaMKII on Na V 1.1, we further checked the effect of CaMKII on CaM binding to Na V 1.1 IQ domain. As shown in Figure 6A,B, the binding of apoCaM to IQ barely changed after phosphorylation at ≈free [Ca 2+ ] compared to that in the presence of Ca 2+ , suggesting that phosphorylation of IQ domain by CaMKII had little effect on its binding with apoCaM. By contrast, the binding of Ca 2+ /CaM to IQ was increased in the phosphorylated IQ, indicating that the effect of CaMKII on the CaM binding to Na V 1.1 IQ domain could exert its regulation only in the presence of Ca 2+ (Figure 6C,E,G). The parameters obtained ( Figure 6D,F,H and Table 3) revealed that CaMKII-mediated phosphorylation increased the binding of CaM to IQ, while dephosphorylation by CIP decreased the affinity of CaM with IQ. In addition, K d and B max showed that an increased binding of Ca 2+ /CaM to the channel at higher [Ca 2+ ] was observed compared to that at 100 nM [Ca 2+ ]. nM), whereas Kd of C-lobe was lower than that of N-lobe in the absence of Ca 2+ . Additionally, Bmax of N-lobe was higher than that of C-lobe in the presence of Ca 2+ , whereas Bmax of C-lobe was higher than that of N-lobe in the absence of Ca 2+ , indicating that C-lobe was the predominant domain in apoCaM interacting with NaV1.1 IQ domain, and N-lobe was the predominant domain in Ca 2+ /CaM interacting with NaV1.1 IQ domain.

The Effect of CaMKII on CaM Binding to NaV1.1 IQ Domain
In the previous study, several CaMKII-mediated phosphorylation sites on NaV1.1 have been identified [29]. To clarify the modulation of CaMKII on NaV1.1, we further checked the effect of CaMKII on CaM binding to NaV1.1 IQ domain. As shown in Figure 6A,B, the binding of apoCaM to IQ barely changed after phosphorylation at ≈free [Ca 2+ ] compared to that in the presence of Ca 2+ , suggesting that phosphorylation of IQ domain by CaMKII had little effect on its binding with apoCaM. By contrast, the binding of Ca 2+ /CaM to IQ was increased in the phosphorylated IQ, indicating that the effect of CaMKII on the CaM binding to NaV1.1 IQ domain could exert its regulation only in the presence of Ca 2+ (Figure 6C,E,G). The parameters obtained ( Figure 6D,F,H and Table 3) revealed that CaMKII-mediated phosphorylation increased the binding of CaM to IQ, while dephosphorylation by CIP decreased the affinity of CaM with IQ. In addition, Kd and Bmax showed that an increased binding of Ca 2+ /CaM to the channel at higher [Ca 2+ ] was observed compared to that at 100 nM [Ca 2+ ].   In a previous study, one CaMKII-mediated phosphorylation site S1920 on Na V 1.5 IQ domain had been identified. In addition, CaMKII is a basophilic protein kinase belonging to the Ca 2+ /CaM dependent superfamily of serine/threonine kinases (Herren et al., 2015). Thus, we mutated three potential CaMKII-mediated phosphorylation sites-T1909, S1918, and T1934-into A ( Figure 7A), then Na V 1.1 IQ domain was treated with CaMKII and CIP. As shown in Figure 7B, under free [Ca 2+ ] condition, there was no significant difference in the CaM binding to the IQ domain between the phosphorylated and dephosphorylated peptides as well as 500 nM [Ca 2+ ] condition, indicating that CaMKII facilitated the binding of Ca 2+ /CaM to Na V 1.1 IQ domain due to one or several phosphorylation sites in T1909, S1918, and T1934 of Na V 1.1 IQ domain. In a previous study, one CaMKII-mediated phosphorylation site S1920 on NaV1.5 IQ domain had been identified. In addition, CaMKII is a basophilic protein kinase belonging to the Ca 2+ /CaM dependent superfamily of serine/threonine kinases (Herren et al., 2015). Thus, we mutated three potential CaMKII-mediated phosphorylation sites-T1909, S1918, and T1934-into A ( Figure 7A), then NaV1.1 IQ domain was treated with CaMKII and CIP. As shown in Figure 7B, under free [Ca 2+ ] condition, there was no significant difference in the CaM binding to the IQ domain between the phosphorylated and dephosphorylated peptides as well as 500 nM [Ca 2+ ] condition, indicating that CaMKII facilitated the binding of Ca 2+ /CaM to NaV1.1 IQ domain due to one or several phosphorylation sites in T1909, S1918, and T1934 of NaV1.1 IQ domain.  [Ca 2+ ], a nonphosphorylated state of IQ domain, C-lobe is the predominant domain for apoCaM binding to IQ domain, and the binding of apoCaM to Na V 1.1 IQ is not affected by CaMKII. At high [Ca 2+ ], but a nonphosphorylated state of IQ domain, N-lobe becomes the predominant domain since some of CaM binds with Ca 2+ . Phosphorylation of IQ by CaMKII modulates the binding of Ca 2+ /CaM to the channel. Meanwhile, channel activity is maintained to the basal level. At high [Ca 2+ ], a phosphorylated state of IQ domain, N-lobe is the predominant domain for Ca 2+ /CaM binding to IQ domain. Meanwhile, the effect of CaMKII phosphorylation is further promoted, leading to an increased binding of Ca 2+ /CaM to the channel compared to that at 80-100 nM [Ca 2+ ]. The channel activity would be also increased at high [Ca 2+ ] compared to that at low [Ca 2+ ]. Red and black circles represent Ca 2+ and Na + , respectively. Red oval with a P on it represents activated phosphorylation site.

Expression and Purification of Recombinant GST Fusion Peptides
The above described vectors were transformed into Escherichia coli BL21 (DE3) to express the target peptides as glutathione-S-transferase (GST) fusion proteins. The bacteria were cultured in LB liquid medium at 37 • C overnight until an OD600 of 0.8-1.0. Then, the bacteria were induced with 1 mM isopropyl-1-thio-β-D-galactopyranoside (IPTG) and continued incubating for 4 h at 37 • C before harvesting. The ultrasonic technique was used to harvest GST-fusion peptides. Then, the fusion peptides were purified using Glutathione Sepharose 4B beads (GS-4B, GE Healthcare, New York, NY, USA). The GST regions of CaM and its mutants were cleaved with PreScission Protease (GE Healthcare). GST-IQ, CaM, and its mutants were quantified by Enhanced BCA Protein Assay Kit with BSA as standard with correction factors 1.25 (GST-IQ), 1.69 (CaM and its full-length mutants), 0.82 (N-lobe of CaM), and 0.82 (C-lobe of CaM). ≈ free, 100, 500, and 2 mM). The [Ca 2+ ] was calculated with MaxChelator (http://maxchelator.stanford.edu/index.html). Then, the reaction systems were gently washed twice with the same buffer. Bound CaM (or its mutant) and IQ (or EQ) were resuspended in 5× SDS-PAGE loading buffer and resolved in 15% SDS-PAGE gels. Proteins were stained by Coomassie brilliant blue R (CBB). Protein bands in the SDS-PAGE gels were digitized by the Photoshop software (Adobe, San Jose, CA, USA), and the grey level was quantified by Image J software (NIH, Bethesda, MD, USA) [34][35][36][37]. The optical density values were converted to protein contents using respective correction factors (see below).

GST Pull-Down Assay
The GST-fusion IQ (for control) immobilized to GS-4B beads (40 µL) was phosphorylated in an assay reaction (0.4 µm CaMKIIT286D) in Tris buffer containing 1 mM MgCl 2 and 1 mM Na 2 ATP for 30 min at 30 • C. The reaction was then terminated with the addition of 1× SDS sample buffer and gently washed twice. The dephosphorylation was achieved by adding 5 U/mL calf intestinal alkaline phosphatase (CIP; New England Biolabs, Ipswich, MA, USA) into the reacting mixtures incubating at 37 • C for 30 min. CIP is a nonspecific phosphorylase that commonly exists in calf intestinal mucosa. The reaction was terminated by the addition of same Tris buffer and gently washed twice.

Computational Docking
A homology model of Na V 1.1 IQ was constructed based upon the solved crystal structure of IQ motif of Na V 1.2 (PDB # 2KXW) [15] using the Create Homology Model tool in Discovery Studio 2017 (BIOVIA, Boston, MA, USA). The N-and C-lobe of CaM were derived from the crystal structure of Ca 2+ /CaM-Ca V 1.2 IQ domain complex (PDB # 3DVE) [39]. Docking studies between Na V 1.1 IQ and CaM N-or C-lobe were performed in Discovery Studio using the Dock Proteins protocols. The ZDOCK protocol was used for docking the IQ motif of Na V 1.1 to N-or/and C-lobe of CaM and, subsequently, the RDOCK protocol was applied for further refinement of the 10 best-docked poses. For individual interactions, docking results are displayed as solid ribbons of 1 solution with the lowest RDOCK interaction energy.

Statistical Analysis
Quantified grey level was converted into molar quantities according to the mass of GST-IQ, CaM, N-lobe, and C-lobe of CaM. We found that relative optical densities of the same amount of these proteins on the gel in reference to BSA were 0.80, 0.59, 1.21, and 1.22 respectively, from which the correction factors were determined as 1.25, 1.69, 0.82, and 0.82, respectively. Curve-fitting of the total bound ligand (CaM and its mutants) was performed with the software SigmaPlot 12.0 (version 12, Sigma-Aldrich, Beijing, China). Bound ligand (y) was fitted with the following Hill's equation for the one-site fitting model: y = B max ·x/(K d + x). For the two-sites model, a sum of two Hill's equations was integrated to assume independent binding: y = B max1 ·x/(K d1 + x) + B max2 ·x/(K d2 + x), where B max1 , B max2 , K d1 , and K d2 represents total B max and K d , respectively; x is the concentration of free ligand; K d is the apparent dissociation constant; and B max is the maximum binding for each binding site. We chose either one-site model or two-sites model based on the higher value of R 2 . Hill's coefficient of 1.0 was assumed. Total concentration of ligand was assumed as an approximate of free ligand. The data are presented as mean ± S.E. (n = 4). The SPSS 22.0 software (version 22, Sigma-Aldrich, Beijing, China) was used to evaluate the statistical significance, and p < 0.05 by hypothesis test was considered statistically significant.

Discussion
Our study was aimed at clarifying the molecular mechanism underlying the modulation of Ca 2+ /CaM/CaMKII on Na V 1.1 channel, which is a key issue in understanding the regulatory mechanism of VGSCs. Although the modulation of VGSCs has been a hot spot in ion channel research, the present study examined for the first time the effects of Ca 2+ and CaM on Na V 1.1 channel in a wide range of [Ca 2+ ] using CaM mutants.
CaM is the most important Ca 2+ binding protein and is involved in the regulation of numerous Ca 2+ -dependent pathways. Its function and structure depend strongly on Ca 2+ concentration [40,41]. In our research, we checked the effect of different Ca 2+ concentrations on the binding of CaM to IQ domain of Na V 1.1. Our data showed that the binding of CaM to Na V 1.1 IQ domain was Ca 2+ -and concentration-dependent. Full-length CaM switches from a simple folding structure at lower [Ca 2+ ] to a rich and complex folding behavior at high [Ca 2+ ] [41]. Accordingly, similar Ca 2+ -dependent conformational changes in CaM between Na V 1.2 and Na V 1.5 have previously been reported [8,40]. Our results showed that the binding of CaM to Na V 1.1 IQ was Ca 2+ -dependent, which reflects the Ca 2+ -dependent conformational change of CaM.
CaM often modulates target molecules only upon conversion to its Ca 2+ -bound form. However, apoCaM binding in itself markedly promotes opening of voltage-gated calcium channels (VGCCs) [9,11,34,41]. VGSCs have also been suggested to adopt a similar modulatory principle [9,16]. Our present data showed that the binding of Ca 2+ /CaM to Na V 1.1 IQ was dramatically decreased compared to that of apoCaM, implying that apoCaM promotes activity of Na V 1.1 channels. On the contrary, studies on Ca V 1.2 indicated that the binding of apoCaM to the channel was significantly smaller compared to that of Ca 2+ /CaM [21,37,42]. This may suggest that CaM binds to different channels in a channel-specific manner, meaning the detailed mechanism of CaM regulation may need to be considered on a channel-specific basis.
It has been reported that regulatory effect of CaM on VGCCs is lobe-specific, and N-and C-lobe of CaM have distinct roles in the regulation of VGCCs [7,40,43]. In addition, a single Ca 2+ /CaM bridges the C-terminal IQ motif of Na V 1.5 to the DIII-IV linker via individual N-and C-lobes, respectively. C-lobe binds to IQ (N-lobe is free) at low [Ca 2+ ], whereas at high [Ca 2+ ], N-lobe binds to IQ (lobe switching) and C-lobe binds to III-IV linker, resulting in depolarization of the inactivation curve [44]. Furthermore, the most prominent Ca 2+ -dependent conformational change is the interaction between the calcified N-lobe of CaM and the Na V IQ domain; the CaM C-lobe remains Ca 2+ -free even in millimolar Ca 2+ , remains bound to the IQ motif, and retains its semi-open conformation. Despite similar Ca 2+ -dependent conformational changes between the Na V 1.2 and Na V 1.5 complexes, the functional effects are isoform-specific, while their mechanistic bases are not clear [40]. However, the lobe specificity of CaM modulation of Na V 1.1 have not been demonstrated. In this study, we applied individual N-and C-lobe of CaM to further explore the role of individual lobes in binding to Na V 1.1 IQ domain. The affinity of IQ for C-lobe binding was higher than that of N-lobe in the absence of Ca 2+ , whereas the affinity of IQ for N-lobe binding was higher than that of C-lobe in the presence of Ca 2+ , indicating distinct responses at different [Ca 2+ ]. Thus, the property difference between the two lobes of CaM might endow the Ca 2+ -dependent and lobe-specific modulation of Na V 1.1 channel. However, we still do not know the reason why CaM binding to the IQ domain is the largest at free Ca 2+ and smaller at 100 nM Ca 2+ . We speculate that C-lobe of CaM might be at least partially occupied with Ca 2+ at 100 nM Ca 2+ . In this scenario, the weaker binding of CaM to the IQ domain would be explained by the fact that the IQ domain has lower affinity for Ca 2+ /C-lobe than for Ca 2+ -free C-lobe. This point may be supported by the CaM 12 and CaM 34 experiments ( Figure 3B-E). In the CaM 12 experiment, the binding property to the IQ at free and 100 nM Ca 2+ was similar to that of wild-type CaM, while this property was less pronounced in the CaM 34 experiment.
One of the most intriguing findings of our research is that CaMKII-mediated phosphorylation of Na V 1.1 IQ domain increased binding of CaM to the channel. It has become recognized that both the expression and function of VGSCs is under tight control of protein phosphorylation by protein kinases [45]. CaMKII activated by Ca 2+ /CaM maintains activity of VGCCs [36], CaMKII has emerged as a critical regulator of Na V 1.5, and multiple CaMKII-mediated phosphorylation sites have been identified on Na V 1.5, including S1920 and S1925, which are located on IQ domain and noncanonical CaMKII sites. [29,46]. In addition, CaMKII-enhanced I NaL positively shifts inactivation curve of Na V 1.2 epileptic mutant (Q54) [23,47]. Based on the structural homology of Na V 1.1 and Na V 1.5, we treated Na V 1.1 IQ domain with CaMKII and CIP to examine a possible regulation of Na V 1.1 by CaMKII. Under low [Ca 2+ ] condition, we found there was no significant difference in the CaM binding to the IQ domain between the phosphorylated and dephosphorylated peptides. One possible reason for this may be that the phosphorylation of Na V 1.1 IQ might not affect the conformation of the apoCaM binding region, which interacts mainly with C-lobe of CaM. However, in the presence of Ca 2+ CaMKII-mediated phosphorylation of IQ increased the binding of CaM to IQ domain, while CIP-mediated dephosphorylation of IQ decreased the binding of CaM. Thus, it is possible that Ca 2+ /CaM binding region interacts mainly with N-lobe of CaM, which might be different from the apoCaM binding region. Our data has shown that CaMKII regulates the binding of Ca 2+ /CaM to Na V 1.1 IQ domain due to one or several phosphorylation sites in T1909, S1918, and T1934 of Na V 1.1 IQ domain, indicating that CaMKII-mediated phosphorylation of Na V 1.1 affects CaM binding to Na V 1.1. It is therefore speculated that CaMKII-mediated phosphorylation of Na V 1.1 IQ domain might change the conformation of IQ domain, leading to increased binding of Ca 2+ /CaM to Na V 1.1.
A previous study had shown that CaM overexpression in HEK1.1 cells increases the peak current of Na V 1.1 in a calcium-dependent manner [12]. Our previous study has also demonstrated that neuronal VGSC activity is modulated by CaM in a concentration-dependent manner in normal and low Mg 2+ condition [13]. Thus, we propose the following hypothetical model ( Figure 7C) for the modulation of Ca 2+ /CaM/CaMKII on Na V 1.1 based on our present study and other studies [12,13]: At low [Ca 2+ ]-a nonphosphorylated state of Ca 2+ concentration-C-lobe is the predominant domain for apoCaM binding to IQ domain, and the binding of apoCaM to Na V 1.1 IQ is not affected by CaMKII. At high [Ca 2+ ] but at a nonphosphorylated state of IQ domain, N-lobe becomes the predominant domain since some of the CaM binds with Ca 2+ . Phosphorylation of IQ by CaMKII modulates the binding of Ca 2+ /CaM to the channel. Meanwhile, channel activity is maintained to the basal level at 80-100 nM [Ca 2+ ]. At high [Ca 2+ ]-a phosphorylated state of IQ domain-N-lobe is the predominant domain for Ca 2+ /CaM binding to IQ domain. Meanwhile, the effect of CaMKII-mediated phosphorylation is further promoted, leading to an increased binding of Ca 2+ /CaM to the channel compared to that at 80-100 nM [Ca 2+ ]. The channel activity will also be increased at high [Ca 2+ ] compared to that at 80-100 nM [Ca 2+ ].
In summary, we found that the binding of Ca 2+ /CaM to IQ was Ca 2+ -and concentration-dependent, and apoCaM more preferentially binds to Na V 1.1 IQ domain than Ca 2+ /CaM. In addition, C-lobe of CaM is the predominant domain in apoCaM binding to Na V 1.1 IQ domain, whereas N-lobe of CaM is the predominant domain in Ca 2+ /CaM binding to Na V 1.1 IQ domain. In addition, CaMKII-mediated phosphorylation increases the binding of Ca 2+ /CaM to Na V 1.1 IQ domain due to one or several phosphorylation sites in T1909, S1918, and T1934 of Na V 1.1 IQ domain. Our data provides novel mechanisms for the modulation of Na V 1.1 by the Ca 2+ /CaM/CaMKII axis. For the first time, we uncover the effect of Ca 2+ , lobe-specificity, and CaMKII on CaM binding to Na V 1.1.