View all hidden authors and organizations

Edited by Richard W. Aldrich of the University of Texas at Austin, approved on June 21, 2021 (reviewed on December 11, 2020)

Cardiac sodium channels (Nav1.5) are essential to produce a regular heartbeat. Therefore, it is not surprising that the Nav1.5 mutation is associated with life-threatening arrhythmias. Interestingly, the activity of Nav1.5 can also be changed by post-translational modifications, such as tyrosine phosphorylation. We combined protein engineering and molecular modeling to reveal that the adverse effects of mutations in patients with long QT3 are only exposed when the proximal tyrosine is phosphorylated. This suggests that there is dynamic crosstalk between gene mutation and adjacent phosphorylation, which may be important in other classes of proteins. In addition, we show that phosphorylation affects the sensitivity of the channel to clinically relevant drugs, and this finding may be important in the development of patient-specific treatment plans.

The voltage-gated sodium channel Nav1.5 initiates cardiac action potentials. Changes in its activation and inactivation characteristics due to mutations can lead to serious and life-threatening arrhythmias. However, despite extensive research work, many functional aspects of this cardiac channel are still poorly understood. For example, Nav1.5 has undergone extensive post-translational modifications in vivo, but the functional significance of these modifications has not been explored to a large extent, especially under pathological conditions. This is because most traditional methods cannot insert metabolically stable post-translational modification mimics, thereby preventing the precise elucidation of the contribution of these modifications to channel function. Here, we overcome this limitation by using Nav1.5 protein semisynthesis in living cells and perform complementary molecular dynamics simulations. We introduce metabolically stable phosphorylation mimics on the background of wild-type (WT) and two pathogenic long QT mutation channels, and decipher the function and pharmacological effects with unique precision. We have elucidated the mechanism by which Y1495 phosphorylation impairs the steady-state inactivation of WT Nav1.5. Surprisingly, we found that although the Q1476R patient mutation itself does not affect inactivation, it enhances the damage of homeostatic inactivation caused by Y1495 phosphorylation by enhancing the debinding of inactivated particles. We also showed that both phosphorylation and patient mutations affect the sensitivity of Nav1.5 to the clinically used antiarrhythmic drugs quinidine and ranolazine, but not flecainide. The data emphasizes that the functional impact of Nav1.5 phosphorylation can be significantly amplified by patient mutations. Therefore, our work may have an impact on the interpretation of the mutant phenotype and the design of future drug treatments.

Cardiac action potentials are caused by the precisely coordinated activities of voltage-gated sodium, potassium, and calcium channels in time. The voltage-gated sodium channel Nav1.5 has reported more than 500 potentially pathologically related point mutations (1) responsible for the initial rapid depolarization observed in cardiac action potentials. Many of these mutations are known or suspected causes of abnormal action potentials and can lead to life-threatening diseases such as Brugada and long QT3 syndrome (BS and LQT3, respectively) (2). The most common non-invasive treatments for such conditions include the use of antiarrhythmic drugs (AAD), such as quinidine, flecainide, or ranolazine (3).

Nav1.5 is a 2,016 amino acid membrane protein with four homologous but different domains (DI-IV) (4). Each domain contains six transmembrane segments (S1 to S6). The first four parts (S1 to S4) form the so-called "voltage sensing domain", where the positively charged residue of S4 is responsible for the voltage sensitivity of the channel. Instead, the last two segments (S5 and S6) and the P ring connecting them form the pore module, which includes the selective filter. After depolarization, the S4 segment will move upward, which will translate into a conformational change that opens the access door. This channel opening is mainly mediated by the S4 helix in DI-DIII, and the slower upward movement of DIV leads to channel inactivation (5⇓ ⇓ –8). The latter is bound by the isoleucine, phenylalanine and methionine (IFM) tripeptide motifs in the DIII-DIV linker to its receptor site adjacent to DIV S6 (9⇓ ⇓ ⇓ – 13) Caused. Subthreshold depolarization causes the channel to experience steady state inactivation (SSI) without first opening. The voltage dependence of SSI is an important determinant of channel availability in the body, and even small changes can cause arrhythmic phenotypes. Similarly, incomplete inactivation can lead to potentially pathogenic continuous currents, called late currents (14).

Interestingly, many known or suspected pathogenic mutations occur in the loop between the cytoplasmic junction and a single transmembrane segment (1, 11, 15⇓ –17). These loops and linkers are not only hot spots for pathogenic mutations, but also contain many post-translational modifications (PTM), such as phosphorylation, methylation, and acetylation (18). Phosphorylation stoichiometry is usually highly variable (19), and previous work has shown that phosphorylation levels in cardiac ion channels (such as Nav1.5 and Kv7.1) can be significantly increased after β-adrenergic stimulation ( 20). Therefore, mutations in PTM sites or dysregulation of PTM levels play an important role in disease states (18, 21). However, despite solid evidence of their existence and relevance in vivo, it is still challenging to directly assess the functional impact of PTM in vitro: traditional mutagenesis can be used to introduce non-modifiable (NM) side chains to prevent modification, but use Naturally occurring PTM mimics PTM side chains are often defective, especially mimic tyrosine phosphorylation (22). Similarly, overexpression of regulatory enzymes (such as kinases or phosphatases) lacks selectivity for target proteins. Recently, we have overcome some of these limitations by using tandem protein trans-splicing (tPTS) to generate semi-synthetic membrane proteins containing stable PTM mimics (23).

Here, we use a combination of tPTS, electrophysiology, and molecular dynamics (MD) simulations to study the functional effects caused by Nav1.5 Y1495 phosphorylation (10 mV right shift in SSI V1/2) (23, 24) Changes in clinically relevant mutations and whether they affect the pharmacological characteristics of AAD in wild-type (WT) and mutation pathways. Our data show that the pathogenic Q1476R mutant (26) has no significant effect on SSI itself compared to the pathogenic ΔK1500 mutant (25). However, when the Q1476R mutation is combined with the phosphorylation of nearby Y1495, this causes a significant right shift of SSI (20 mV) and a significant late current. Our MD simulation attributed these changes in SSI to the unstable binding of IFM-inactivated particles in the 1495 phosphorylated channel, especially when combined with the Q1476R mutation. Finally, we show that the sensitivity of Nav1.5 to both Class I AAD and Ranolazine is altered by phosphorylation and mutations in the tested patients.

Heterologous protein expression does not allow precise control of the PTM range. In the case of Nav1.5 Y1495, although traditional mutagenesis methods can prevent the phosphorylation of tyrosine to phenylalanine mutations, it cannot control the degree of phosphorylation of natural tyrosine. We overcome this obstacle using a recently developed semi-synthetic method that allows the insertion of synthetic peptides carrying single or multiple PTM or PTM mimics into ion channels (23). In this method, the channel is divided into three fragments: the N and C-terminal fragments (NREC and CREC) correspond to the partial channel fragments recombinantly expressed in Xenopus oocytes and the synthetic peptide containing the site of interest ( PSYN), injected into the cytoplasm of the oocyte (Figure 1A and SI appendix, Figure S1). The covalent linkage of the three fragments is mediated by two orthogonal split inteins: CfaDnaE (split intein A) (27) and SspDnaBM86 (split intein B) (28). After assembly, each split intein connects spontaneously and covalently to the connected channel fragments through natural chemistry. This process is also called tPTS (29) (Figure 1A and SI appendix, Figure S1 and S2).

Phosphorylation of Y1495 disrupts the docking of the IFM motif with its receptor site. (A) Schematic diagram of tPTS of Nav1.5 channel used to generate NM (WT NM) or phosphorylation (WT phY) at Y1495. The amino acids (aa) 1 to 101 of CfaDnaE (orange) are merged into the C-terminus of the channel fragment corresponding to Nav1.5 aa 1 to 1471 and used as a heterologous expression of the NREC construct. The PSYN sequence corresponds to Nav1.5 aa 1472 to 1502 and is connected to the C-terminal part of CfaDnaE (aa 102 to 137, orange) at its N-terminal and the N-terminal part of SspDnaBM86 (aa 1 to 11, yellow) at its C-terminal . The corresponding C-terminal part of SspDnaBM86 (aa 12 to 154, yellow) was expressed as a fusion construct at the N-terminal of protein fragment C (Nav1.5 aa 1503 to 2016) to form a CREC construct. (B) SSI (left) and activation (right) curves of WT NM and WT phy constructs, including example traces and chemical structures of aa present in position 1495. Data are shown as mean ± SD; n = 6 to 9. (C) The contact frequency of Y1495 and IFM particle residues with adjacent residues. Phosphorylation reduces the contact between IFM's Y1495 and M1487, and at the same time it causes the IFM particles to increase their contact with the DIII-S6 and DIII-S4-S5 joints. (D) The conformation of the IFM particle at its binding site after 200 ns MD simulation. Phosphorylation of Y1495 moves the loop region of the Dill-DIV linker containing Q1483 residues outward, bringing the IFM particles closer to Dili-S6. The IFM side chain is highlighted in orange. (E) The free energy distribution of IFM debinding, using the distance between the center of mass (COM) of N1659 in the DIV-S5 spiral and F1486 in the IFM particle as the reaction coordinate. WT phY leads to a decrease in the binding energy of IFM binding.

This method allows us to insert PSYN variants containing phenylalanine (which cannot be post-translationally modified, hence the name NM) or phosphonylated tyrosine (30) at position 1495 (phY). The latter is a hydrolytically stable phosphorylation mimic, with almost the same charge and space properties (31). Therefore, whenever we introduce phy, we will mention the effect of phosphorylation throughout the manuscript. In order to improve the splicing efficiency, we introduced the N1472C mutation in the first position of PSYN, which caused the SSI to shift to the right compared to the WT channel (23) (Table 1). This is consistent with the view that the highly similar N1472S mutation was previously reported as a putative long QT-related mutation (32).

Injection of oocytes expressing N and C fragments with NM peptide (WT NM) results in activation and SSI parameters similar to those obtained when all three channel fragments are recombinantly expressed (WTrec) (Table 1 and SI appendix, Figure S3) ). However, when only the N and C fragments were injected, no voltage-dependent current was observed during the similar incubation time, indicating that individual channel fragments could not be assembled non-covalently (23). Consistent with previous work, the phosphonylation of Y1495 (WT phY) caused the SSI curve to shift to the right by approximately 10 mV, while retaining the half-maximum activated WT samples (Figure 1B and Table 1) (23, 33). The time course of inactivation remained unchanged, no later current was observed, and the time course of recovery from inactivation of WT phY was significantly faster than that of WT NM (Table 1 and SI appendix, Figure S4).

In order to clarify the molecular basis of these experimental observations, we turned to the recently determined MD simulation of the Nav1.5 cryo-electron microscope structure, in which the IFM particles are docked at the site between the DIII S4-S5 joint and DIV S6 (11). Due to the SSI return Because of this specific interaction (Figure 1D), we focused on the difference between WT and WT phY in this region. Strikingly, the 200 ns MD simulation showed that phosphorylation caused Y1495 to lose contact with M1487 of IFM particles. In WT phY, there was contact between I1485 of IFM and F1473 and Q1476 (in DIII-S6) (I1485, Figure 1C). On the other hand, the interaction involving F1486 did not differ significantly between the two conditions, except for the formation of contact with A1326 during phosphorylation (F1486, Figure 1C). The phosphate group of WT phY mediates the downward movement of Q1483 and keeps D1484 away from DIV-S6 (Y1495, Figure 1C and D). Therefore, after Y1495 phosphorylation, the phosphate group moves to the binding pocket of the IFM particle (Figure 1D), pushing the IFM particle toward DIII-S6 and contacting Q1476. This causes the N-terminal part of the DIII-DIV connector (Q1483, D1484) to move outward and downward, away from the IFM particle binding pocket (Figure 1D).

Using the distance between the center of mass (COM) of N1659 in the DIV-S5 spiral and F1486 in the IFM particle as the reaction coordinate, an umbrella sampling simulation was performed to evaluate the binding free energy of the IFM particle and its docking site (Figure 1E). This specific reaction coordinate was chosen because N1659 forms a hydrogen bond with the carbonyl group of F1486 in the docking state, as suggested by Jiang et al. (11). As expected from the observations of conventional MD simulations, phosphorylation of Y1495 did lead to a decrease in the free energy of binding, indicating that the debinding of IFM particles was alleviated by phosphorylation. This is very consistent with the SSI shift to the right and the accelerated recovery from inactivation observed in the experiment.

In conclusion, the combined experimental and calculated data supports the view that Y1495 phosphorylation disrupts the interaction of IFM motifs with its receptor sites. Specifically, the phosphorylation of Y1495 moves it from DIV-S6 to the DIII-S6 and DIII-S4-S5 joints, thereby shifting the Nav1.5 SSI to the right and accelerating the recovery of inactivation.

The Q1476R mutation in the DIII-DIV junction was previously thought to be the cause of LQT3 in the patient's family (26). Specifically, heterologous expression in mammalian cells showed that compared to WT, the mutant channel showed a 6.5 mV right shift in SSI, along with late currents (26). In view of the physical proximity of Q1476 and Y1495 to IFM inactivated particles and the increased contact frequency of I1485 and Q1476 observed in WT phY, we tried to test whether the phosphorylation of Y1495 in the presence of Q1476R would significantly affect channel function mutations. To this end, we synthesized a PSYN variant containing the Q1476R mutation on the Y1495F (Q1476R NM) or Y1495 phY background (Q1476R phY). The insertion of any peptide variant will generate a strong voltage-gated current within 12 hours after injection (Figure 2A). To our surprise, the introduction of the Q1476R mutation on a non-phosphorylated background did not change the voltage dependence of SSI or induce delayed currents (Figure 2 B and C and Table 1). In contrast, when we introduced the Q1476R mutation in the presence of phosphorylated tyrosine at 1495, we observed a drastic right shift of SSI compared to Q1476R NM (ΔSSI ∼20 mV, Table 1), It clearly exceeds the background we observed on WT (ΔSSI ∼11 mV). We further showed that phosphorylation on the mutant (but not WT) background also significantly slowed the rate of channel inactivation and resulted in a significant delayed current after 400 ms (Figure 2A and C and Table 1).

The Q1476R mutation causes a significant change in SSI. (A) Representative current traces of Q1476R NM (purple) and phY (blue) structures. (B) SSI (left) and activation (right) curves of the specified construct. (C) The late current (top) and deactivation rate (bottom) of the specified structure. Data are shown as mean ± SD in B and C; n = 5 to 10; use unpaired two-tailed Student's t-test to compare data; ns (not significant) P> 0.05, *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001 and ****P <0.0001. Please note that the accurate measurement of late currents in X. laevis oocytes is hindered by slow-occurring, voltage-dependent endogenous currents. Therefore, our later current measurement can only be used as an estimate. (D) The contact frequency of Y1495 and IFM particles with adjacent residues. (E) MD simulates the conformation of the IFM particle at its binding site after 200 ns. Q1476R phY causes the DIII-S6 helix to break, resulting in unstable IFM binding. The IFM side chain is highlighted in orange. (F) The free energy distribution of IFM debinding, using the distance between the COM of N1659 in the DIV-S5 spiral and the F1486 of the IFM particle as the reaction coordinate. Compared with WT phY, Q1476R phY leads to a further decrease in the binding energy of IFM particles, and due to the interruption of DIII-S6 induced by the Q1476R mutation, the equilibrium distance between F1486 and its docking site increases.

The only parameter that is significantly affected by Q1476R alone (Q1476R NM) is the voltage required for half-maximum activation. This is a significant shift to the left compared to WT NM, but it is indistinguishable between Q1476R phY and WT phY (Figure 2B and C and Table 1).

These data indicate that in the presence of pathogenic Q1476R mutations, phosphorylation-induced instability of IFM particles at their receptor sites can be drastically worsened. Our findings further support the idea that compared with WT phY, Q1476R phY has faster recovery from inactivation (Table 1 and SI appendix, Figure S4).

Next, we tried to evaluate whether full-length Nav1.5 containing conventional glutamate as a phosphotyrosine mimic at position 1495 on WT or Q1476R background would produce comparable results (Y1495E and Q1476R/Y1495E, respectively). Although we observed overall similar trends, Y1495E and Q1476R/Y1495E caused even more significant changes in SSI compared to semi-synthetic WT NM, WT phY and Q1476R phY constructs (Table 1 and SI appendix, Figure S3). This emphasizes the advantage of inserting phosphorylation mimics with almost the same charge and steric properties through our semi-synthetic method instead of using conventional mutagenesis.

In order to rationalize the molecular basis of the effects observed by Q1476R, we performed a 200 ns MD simulation of Nav1.5 using the WT phY and Q1476R phY systems. Strikingly, the Q1476R mutation caused the Dill-S6 helix to break, moving its C-terminus to the IFM binding site. Therefore, compared to the WT phY system, an increase in the contact frequency between G1481/G1482 and the IFM motif was observed in the Q1476R phY system (Figure 2D and E). The Q1476R mutation also caused position 1476 itself to lose contact with I1485, indicating the importance of this contact for stabilizing the IFM motif in the WT phY system. We also observed a decrease in the contact frequency between F1486 of IFM and N1659 of DIV-S6, and the formation of new contacts with L1480 and G1481 (Figure 2D and E). Finally, M1487 reduces the interaction with P1655, Q1483 and 1495 phY, while increasing the contact with G1481 and G1482. Our umbrella sampling results also show that the combination of Q1476R and phY further reduces the free energy of binding between the IFM motif and its binding site, and results in a larger equilibrium distance between F1486 and its docking site, which is in line with the expectations of conventional MD simulations ( Figure 2F).

Overall, we show that in the presence of mutations in Q1476R patients, the effect of Y1495 phosphorylation on SSI is strongly enhanced, while the latter has little effect without Y1495 phY modification. In WT phY, we observed increased contact between IFM and DIII-S6, especially through the contact between I1485 and Q1476. Therefore, mutating Q1476 to R seems to further disrupt the interaction of IFM with its phosphorylated docking site. In short, this means that the pathogenicity of the Q1476R mutation is not caused by the amino acid change itself, but by the function of nearby phosphorylation.

Next, we set out to evaluate whether other patient mutants that are very close to Y1495 will show similar differential regulation through phosphorylation. Specifically, we selected a deletion mutant ΔK1500, which was previously associated with LQT3, BS, and conduction system diseases (25). To this end, we synthesized PSYN variants containing ΔK1500 deletion mutations on Y1495F (ΔK1500 NM) or Y1495 phY background (ΔK1500 phY). For both cases, we observe a strong voltage-gated current (Figure 3A). Interestingly, the midpoint of the voltage dependence of activation and SSI is very similar to the midpoint observed for the corresponding WT construct (compare ΔK1500 NM and WT NM; ΔK1500 phY and WT phY; Figure 3B or C and Table 1). In other words, in the presence of ΔK1500, phosphorylation at position 1495 causes SSI to shift to the right by ~12-mV, which is almost the same as the shift observed on the WT background. However, the ΔK1500 mutation did reduce the slope of the SSI curve and resulted in lower inactivation rates and significant late currents (Figure 3A-C and Table 1), even in the absence of phosphorylation.

The changes in SSI induced by phosphorylation are similar in ΔK1500 and WT. (A) Representative current traces of the ΔK1500 NM (light green) and ΔK1500 phY (dark green) constructs. (B) SSI (left) and activation (right) curves of the specified construct. (C) The late current (top) and deactivation rate (bottom) of the specified structure. Data are shown as mean ± SD in B and C; n = 6 to 10; use unpaired two-tailed Student's t-test to compare data; ns (not significant) P> 0.05, *P ≤ 0.05, ***P ≤ 0.001 and ****P <0.0001. (D) The contact frequency of Y1495 and IFM particles with adjacent residues. (E) The conformation of the IFM particle at its binding site after 200 ns MD simulation. I1485 has increased contact with DIII-S6 residues. The IFM side chain is highlighted in orange. (F) Use the distance between the COM of N1659 in the DIV-S5 spiral and the F1486 of the IFM particle as the free energy distribution of the IFM unbinding of the reaction coordinate. The binding energy of IFM particles in the ΔK1500 phY system is similar to that of WT phY.

Our computational studies further confirmed the observation that phosphorylation affects the inactivation properties of ΔK1500 in a manner similar to WT. We found that the binding free energy of IFM particles and their binding sites is indeed indistinguishable from the binding free energy of WT phY (Figure 3F), although the interaction curve between IFM particles and their binding sites is slightly changed (Figure 3D and E)) .

In conclusion, we show that the ΔK1500 mutation itself slows down the inactivation rate and causes a large amount of late current, which may explain its pathogenicity. However, in contrast to the Q1476R mutation, the functional effects caused by Y1495 phosphorylation are actually the same as those observed in the WT channel.

To test whether the above observations have an effect on the pharmacology of Nav1.5, we next investigated whether the phosphorylation of Y1495 alone or in combination with patient mutations affects the sensitivity of the channel to clinically used AAD. First, we tested Class Ia AAD quinidine (Figure 4A), which showed a clear use-dependent inhibition of Nav1.5. Using a 20 Hz stimulation program to evaluate the inhibitory effect of quinidine, the half-maximal inhibitory concentration (IC50) produced by WT NM was 86 ± 83 μM (Figure 4A and C and Table 2). In contrast, phosphorylation of Y1495 increased the IC50 to 159 ± 47 μM (WT phY; Figure 4A and C and Table 2). The introduction of the Q1476R mutation leads to a further decrease in apparent affinity, but this value is no longer affected by Y1495 phosphorylation (Q1476R NM is 206 ± 196 μM, and Q1476R phY is 195 ± 96 μM; Figure 4A and C and Table 2). Similarly, in the context of the ΔK1500 mutant, quinidine inhibition is actually independent of the phosphorylation of Y1495, with IC50 values ​​of 63 ± 55 μM (ΔK1500 NM) and 43 ± 5 μM (ΔK1500 phY), respectively (Figure 4 A and C And Table 2).

Phosphorylation and disease mutations will affect the pharmacological sensitivity of Nav1.5. (A and B) Concentration response curves of WT, Q1476R and ΔK1500 constructs in the presence of AAD quinidine (A) or flecainide (B) in response to a 20 Hz pulse sequence (see the structure in the left panel). The P50/P1 value is standardized to a range of 0 to 1. (C) The IC50 value obtained from the quinidine data in A. IC50 was only significantly increased by phosphorylation in WT, but not in Q1476R or ΔK1500 constructs. (D) The IC50 value of the flecainide data shown in B. The IC50 value does not change significantly due to phosphorylation in any construct. ns (not significant) P ≥ 0.05 and *P <0.05. Data are shown as mean ± SD; n = 5 to 9. (E) Superposition of flecainide (F) combined with rat Nav1.5 (yellow; PDB code: 6UZ0) and quinidine (Q) and human Nav1.5 (orange; PDB code: 6LQA).

Pharmacological sensitivity to clinically relevant AAD

Next, we turn to the inhibition of flecainide, which is a slightly larger and more hydrophobic AAD (Figure 4B). As expected (34), this class Ic drug generally exhibits a higher apparent affinity than quinidine (Table 2). However, in contrast to quinidine, we did not observe phosphorylation-induced changes in WT background apparent affinity, as did the Q1476R and ΔK1500 mutations (Figure 4B and D and Table 2). Finally, we tested ranolazine, another sodium channel inhibitor used clinically. Like quinidine, ranolazine showed a decreased apparent drug affinity for phosphorylation in the WT background. This reduction was also observed in the ΔK1500 mutant, but not in the Q1476R background (SI appendix, Figure S5).

The above shows that the inhibition of WT Nav1.5 by quinidine and ranolazine is sensitive to the phosphorylation of Y1495, while the inhibition of flecainide is not. In addition, our data showed that the Q1476R mutation abolished the phosphorylation-induced apparent affinity reduction of all three tested drugs, while for the ΔK1500 mutation, this only applies to quinidine and flecainide.

In this study, we used a combination of protein semisynthesis and MD simulation to study the effect of phosphorylation on the DIII-DIV linker of Nav1.5 by Y1495. Based on this method, we 1) outline the detailed molecular mechanism of how Y1495 phosphorylation regulates the inactivation of Nav1.5, and 2) prove that when we combine it with patient mutations, the degree of functional impact caused by this modification may be affected. There are big differences, 3) that phosphorylation will affect the pharmacological sensitivity of Nav1.5 to clinically relevant AAD.

Although rapid inactivation (and late current) with individual changes may cause disease (35) (see also Figure 3 AC), it has long been recognized that the pathogenicity of Nav1.5 is usually associated with voltage-dependent changes in SSI. For example, due to the phosphorylation of Y1495, the apparent right shift of SSI has been well confirmed from previous in vitro work (18, 33). However, the exact mechanism of how Y1495 phosphorylation affects SSI (basic process of navigation function) remains a mystery. Here, we show that the significant (~11 mV) shift to the right of SSI voltage sensitivity and the significant acceleration of inactivation recovery due to Y1495 phosphorylation are caused by the instability of the interaction between the IFM particles and their receptor sites This leads to a decrease in the binding energy of IFM particles to their binding sites. Our simulations indicate that phosphorylation causes the IFM particle to shift from its receptor site and move it to DIII-S6 to accommodate the Y1495 phosphate group in the IFM binding pocket. In view of the high sequence conservation of the Dili-DIV linker across mammalian Nav channels (36), we speculate that the mechanism by which Tyr phosphorylation in the Dili-DIV linker affects channel inactivation may remain conserved beyond Nav1.5.

It should be noted that Nav1.5 SSI may involve the side chain in the C-terminal domain (37⇓ ⇓ –40), which may include the interaction between the DIII-DIV linker and the C-terminal of the channel (41). However, our current work cannot directly address the potential role of phosphorylation in this process.

The LQT3 mutant Q1476R has previously been shown to not affect the voltage dependence of activation, but causes SSI to shift to the right by 6.5 mV and a significant late current (26). Obtained in human embryonic kidney (HEK) 293 cells, these findings are consistent with the idea that Nav1.5 LQT syndrome mutants usually result in an gain-of-function phenotype (42). Therefore, we were surprised to find that when Y1495 was completely unphosphorylated (that is, in the NM background), the Q1476R mutation itself did not change the voltage dependence of SSI, change the time course of inactivation, or induce late currents (Figure 2) . In contrast, the introduction of the Q1476R mutation on the phY background can cause a sharp (20 mV) right shift of SSI, a large amount of late current and an accelerated recovery from inactivation. Compared with that observed in HEK293 cells, the more pronounced SSI transition is not unexpected, because in our experiments, 100% of the Q1476R mutant channel was phosphorylated, while the natural phosphorylation level under baseline conditions is usually low ( 19, 20), which will result in a less obvious right shift in SSI.

At the molecular level of our MD simulation, Q1476 seems to stabilize the IFM particles in the WT phY system. However, the Q1476R mutation can disrupt this stability and cause a significant conformational change, resulting in weakened IFM particle binding. The functionally observed SSI shift to the right produces a significant window current of around -40 mV (43). In conclusion, our data strongly suggests that the pathogenicity of the LQT3 mutant Q1476R is not caused by the Gln to Arg exchange itself, but by the combined effects of mutation and phosphorylation.

In order to verify the above findings and determine that the phosphorylation-induced hypershift observed in SSI is not the non-specific effect of the mutation on the Dili-DIV linker, we also studied the LQT3/BS mutant ΔK1500. This mutation has previously been shown to cause a slight right shift of activation voltage-dependent and a small left shift of SSI, slower inactivation and delayed current (25), similar to the deletion mutations observed in classical ΔKPQ, such as (44 ). Our results show that the ΔK1500 mutation does cause a slower inactivation time course and delayed currents in the NM and phY background, although the voltage dependence of activation and inactivation is similar to that observed on the WT background. Most importantly, however, we show that the phosphorylation-induced right shift in SSI is comparable to WT (ΔSSI for ΔK1500 is approximately 12 mV, and WT is approximately 11 mV). This is consistent with our observation that the interaction mode between the IFM motif and its binding site is not further disrupted than WT phY, so it is emphasized that the phosphorylation-induced hypershift in SSI is unique to the Q1476R mutation.

It has been previously shown that destruction and inactivation by mutation or other manipulations can weaken the binding of local anesthetics to the Nav channel (45, 46). This is consistent with our finding that the apparent affinity of quinidine for the WT phy background is lower than that of the WT NM background, because the former shows an 11 mV shift to the right in the SSI. However, on channels with Q1476R or ΔK1500 mutations, phosphorylation of Y1495 did not reduce the apparent affinity for quinidine, although phosphorylation caused similar SSI on these channels (ΔK1500) or even compared with WT. Larger (Q1476R) right shift. This suggests that the indirect effects on the apparent affinity of class Ia drugs caused by patient mutations can cover the effects caused by altered inactivation.

Interestingly, we found that the inhibitory effects of quinidine and ranolazine were sensitive to phosphorylation in the context of WT, while the inhibitory effects of flecainide were not. At least for quinidine and flecainide, where structural data is available, this difference can be explained by their different binding positions and postures (11, 13): although flecainide is mainly related to DII and DIII S6 interacts, but quinidine interacts most closely with S6 of DIV (Figure 4E). Because the phosphorylated side chain of Y1495 points to the latter, we speculate that phosphorylation induces unfavorable conformational changes, which are most prominent around the quinidine binding site. In contrast, the more dramatic conformational consequences of the Q1476R mutation appear to be sufficient to eliminate any phosphorylation-induced changes in apparent drug sensitivity.

Although the differences in apparent drug affinity we have observed are relatively small, our data supports the view that both phosphorylation and pathogenic mutations affect the pharmacology of Nav1.5. Although pathogenic mutations have previously been shown to affect navigation pathway pharmacology (26, 34, 47⇓ –49), there is less evidence of phosphorylation-mediated changes (50). In the future, this may affect treatment options for patients with known or suspected pathogenic Nav1.5 mutations, as certain AADs are known to be potentially dangerous in certain patient subgroups (ie flecainide and encainide) The pro-arrhythmic properties of (51⇓ -53).

To date, most confirmed Nav1.5 phosphorylation sites are located within the DI-DII linker. Although multiple pieces of evidence indicate that all intracellular junctions can be phosphorylated (18, 54, 55), studies on the Nav channels of hearts and neurons from natural tissues based on mass spectrometry have not yet found evidence of phosphorylation of the DIII-DIV junction ( 20, 56, 57). This is most likely due to the close association between the DIII-DIV linker and the transmembrane helix (11) and its highly basic sequence content (12/53 amino acids are basic), both of which will reduce the coverage of the MS sequence. In contrast, in vitro studies provide direct evidence of Y1495 phosphorylation by Fyn kinase (33, 58) and the phosphorylation of multiple other Tyr side chains in Nav1.5 (59). In addition, tyrosine kinase inhibitors inhibit sodium current in rabbit cardiomyocytes by shifting SSI to the left (60), which is consistent with our findings and others (23, 33).

The fact that phosphorylation of Y1495 affects the availability of the Nav1.5 channel (Figure 1) and this may be exaggerated by patient mutations (Figure 2) has potential pathophysiological relevance. This is because Tyr kinase has been shown to be more active under pathological conditions, such as ischemia, reperfusion injury or cardiac remodeling (61⇓ –63). The observation that β-adrenergic stimulation leads to increased phosphorylation of cardiac ion channels (including Nav1.5) further emphasizes this aspect (20). Here, we prove that Q1476R will only cause severe changes in channel function when phosphorylated. In short, this may help explain why the clinical phenotype of patients with Nav1.5 mutations changes over time, or why individuals with the same mutations are affected to varying degrees: in the case of Q1476R, the channel may Under basal conditions, but increased phosphorylation levels due to metabolic changes and/or disease states will result in a potentially life-threatening transformation of SSI.

Therefore, our data suggests that for mutations such as Nav1.5 Q1476R, kinase inhibitors may be a better treatment option than traditional AAD (54). In addition, the regulation of Nav1.5 by calmodulin is at least partially mediated through the Dill-DIV linker (64, 65), thus increasing the possibility that the interaction between Y1495 phosphorylation and patient mutations may extend to the Ca2 regulation of Nav1.5 sex. Finally, our work emphasizes that it may be beneficial to tailor future AAD treatment plans based on the patient's specific genetic background (49).

To ensure effective delivery of synthetic peptides, we use microinjection into X. laevis oocytes in the tPTS method. Although this provides a unique "on and off" control of the degree of phosphorylation at the site of interest (0 vs. 100%), we cannot assess that other PTMs may be present in different parts of the protein or if these are in more related mammals The expression system will be different. This is important because the functional impact of the Nav1.5 mutation may be cell type specific (66⇓ –68). In addition, the level of Tyr phosphorylation is usually lower than Ser/Thr (69), so our results may overestimate the extent to which Y1495 phosphorylation affects Nav1.5 function in vivo. Similarly, there is a slight difference in pKa between phosphorylation used in functional experiments and phosphorylation used in computer work (7.5 to 8 vs. 6.5) (31, 70).

As a template for our calculation work, we used a Nav1.5 structure, which was parsed as possibly inactive. This distribution of functions is mainly based on the fact that the IFM particles are docked at the site between the DIII S4-S5 junction and DIV S6. However, we noticed that the structure used to determine the Nav1.5 structure has been carefully designed, resulting in severe gating shifts, prompting us to carefully question this function assignment. Despite these warnings, our calculated data is very consistent with the functional characteristics of phosphorylation and patient mutations, which allows us to confidently propose a mechanistic model of the effects of these modifications.

Our research emphasizes the power of semi-synthetic methods in deciphering complex biophysical interactions. Specifically, data need to be cautious when interpreting the apparent functional effects of potentially pathogenic mutations, because the functional consequences of PTM may exceed those caused by mutations, at least in some cases. This adds another layer of complexity to the investigation and interpretation of mutations in Nav1.5 patients, because PTM levels will change. Given that there are only a large number (> 500) of disease-related mutations in Nav1.5 (1), our findings may be beyond the scope of this study. Our work further stimulated the study of crosstalk between PTM and other protein pathological mutations that has not been explored so far, especially because PTM-mediated effects may not be limited to the spatial proximity of the protein. Finally, our findings provide a starting point for understanding how the Nav1.5 mutation affects the clinically observed patient-specific pharmacological sensitivity.

Gene constructs are generated and used as described in the references. 23; Please refer to the supplementary information for sequence details. PCR was used for standard site-directed mutagenesis and the Q5 site-directed mutagenesis kit (Thermo Fisher Scientific) was used to create deletions. Ambion mMESSAGE mMACHINE T7 Transcription Kit (Thermo Fisher Scientific) was used to linearize and transcribe complementary DNA into complementary RNA (cRNA) for oocyte microinjection.

Unless otherwise stated, all chemicals were purchased from Sigma-Aldrich, Iris Biotech, Rapppolymere, Combi-Blocks, and Chem-Impex. AAD was purchased from Sigma-Aldrich (quinidine sulfate dihydrate, catalog number: Q0875; mexiletine hydrochloride, catalog number: M2727; flecainide acetate, catalog number: F6777; ranolazine dihydrochloride Salt, catalog number R6152) and stored according to product specifications. Dissolve the substance in the ND96 solution (see the two-electrode voltage clamp recording section below for ingredients) and dilute to the specified concentration, and adjust the pH to 7.4. The prepared solution was then stored at room temperature and used for no more than 20 days.

The peptides used for Nav1.5 splicing were synthesized by solid-phase peptide synthesis, as described previously (23). In short, PSYN variants are synthesized by connecting three shorter fragments: the N-terminal intein (IntC-A), the sequence derived from the Nav1.5 ion channel, and the C-terminal intein (IntN-B) , And the Nav1.5 ion channel sequence is the only variable. Regarding the connection strategy adopted, the three fragments were connected in the C to N direction in a “one-pot” manner, using the previously established THz masking group (71). Please note that the WT NM and phY WT peptides also contain the K1479R mutation (23).

Unless otherwise specified, the amino acids used for solid-phase peptide synthesis are Fmoc-Ala-OH, Fmoc-Cys(Trt)-OH, Fmoc-Phe-OH, Fmoc-Gly-OH, Fmoc-Ile-OH, Fmoc-Lys (Boc)-OH, Fmoc-Leu-OH, Fmoc-Pro-OH, Fmoc-His(Trt)-OH, Fmoc-Asn(Trt)-OH, Fmoc-Gln(Trt)-OH, Fmoc-Arg( Pbf )-OH, Fmoc-Ser(tBu)-OH, Fmoc-Thr(tBu)-OH, Fmoc-Tyr(tBu)-OH, Fmoc-Asp(tBu)-OH, Fmoc-Glu(tBu)-OH, Fmoc -Met-OH, Fmoc-Val-OH and Boc-Thz-OH.

Automated peptide synthesis was performed on the Biotage Syro Wave peptide synthesizer using the standard Fmoc/tBu SPPS chemistry previously reported (23). Fmoc deprotection was performed by treatment with piperidine-N,N-dimethylformamide (DMF)-formic acid (25:75:0.95, vol/vol/vol) for 3 12 minutes. Coupling reaction uses Fmoc-Xaa-OH (6.0 is equivalent to resin loading), O-(6-chlorobenzotriazol-1-yl)-N,N,N',N'-tetramethyluronium hexa Fluorophosphate (HCTU, 6.0 equivalents) and N,N-diisopropylethylamine (i-Pr2NEt; 12 equivalents) are coupled for 40 minutes each time. Except i-Pr2NEt which is soluble in N-methyl-2-pyrrolidone, all reagents and amino acids used during SPPS are soluble in N,N-dimethylformamide. The coupling reaction of non-standard Fmoc-protected amino acids is carried out according to the overview of each peptide (SI appendix).

Carry out the thioesterification of the peptide as described in (23, 72), and use 1,1,1,3,3,3-hexafluoro-2-propanol-dichloromethane (20:80, vol/vol) from The protected peptide is cleaved on the resin. Use a mixture of trifluoroacetic acid (TFA)-2,2'-(ethylenedioxy)diethyl mercaptan (DODT)-triisopropylsilane (94:3.3:2.7, vol/vol/v) to carry out the peptide Usually deprotection takes 60 to 90 minutes. After complete deprotection (monitored by matrix-assisted laser desorption/ionization-time of flight), the reaction mixture was concentrated under a stream of nitrogen and the crude peptide was precipitated by adding cold ether. The solid was centrifuged and then washed with cold ether (2x). In the case of partial oxidation of methionine residues, this can be reversed by incubating the peptide in a mixture of DODT (0.2 M) and bromotrimethylsilane (0.1 M) in TFA for 20 minutes (23, 73) . The peptide is then precipitated from ether, and then purified by preparative high performance liquid chromatography.

V-VI stage oocytes from X. laevis oocytes (prepared as previously described in Reference 23) were injected with cRNA and placed in OR-3 (Oocyte Ringer 3) solution at 18°C ​​( 50% Leibovitz medium, 1 mM L-glutamine, 250 mg/L gentamicin, 15 mM HEPES, pH 7.6) for up to 3 days. The lyophilized synthetic peptide was dissolved in Milli-Q H2O to a concentration of 500 to 750 µM, and 9 to 14 nL of the dissolved peptide was injected into the oocytes pre-injected with cRNA using a Nanoliter 2010 micromanipulator (World Precision Instruments) In the cell. Generally, synthetic peptides are injected approximately 24 hours after cRNA injection. Records are made 12 to 20 hours after injection of synthetic peptide.

Please note that the experimental uncertainty associated with the injection of small amounts of synthetic peptides does not allow us to draw conclusions about expression levels (for example, direct comparison of current sizes between WT and mutant channel variants).

Use the two-electrode voltage clamp of the OC-725C voltage clamp amplifier (Warner Instruments) to record the voltage-dependent current. During the recording process, the oocytes were continuously perfused with ND96 solution (in mM: 96 NaCl, 2 KCl, 1 MgCl2, 1.8 CaCl2/BaCl2, 5 HEPES, pH 7.4). For pharmacological tests, solutions containing different AADs in ND96 were used. Glass microelectrodes with a resistance between 0.1 and 1 MΩ are backfilled with 3 M KCl. To determine the half-maximum activation value, the oocyte is kept at -100 mV, and the sodium current is triggered by a voltage step from -80 mV to 40 mV (in 5 mV increments). The SSI curve is determined by applying a 500 ms pre-pulse from -100 mV to -20 mV (in 5-mV increments), followed by a 25-ms test pulse to -20 mV.

First, the Nav1.5 channel (Protein Database [PDB] ID 6UZ3) (11) is embedded in a homogeneous structure consisting of 400 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC) molecules In the lipid bilayer, the CHARMM-GUI membrane generator (74) is used. Four different systems were prepared: 1) WT Nav1.5, 2) Nav1.5 with Y1495 phosphorylation, 3) Nav1.5 with Y1495 phosphorylation and Q1476R mutation, and 4) with Y1495 phosphorylation And ΔK1500 mutation of Nav1.5. Use CHARMM-GUI Membrane Builder for phosphorylation and mutation. Hydrate the system by adding two layers of ~25 angstroms of water on both sides of the membrane. Finally, the system is ionized with 150 mM NaCl.

The CHARMM36 force field is used to describe the interaction between proteins (75), lipids (76) and ions, and the transferable intermolecular potential 3-point (TIP3P) model is used to describe water particles (77). The system uses steepest descent to minimize 5,000 steps, and uses a constant number of particles, pressure, and temperature (NPT) to balance all four systems for at least 36 ns, during which time the position limit protocol is gradually released according to the default CHARMM-GUI (78). During the equilibration, a time step of 2 fs was used; the pressure was maintained at 1 bar by the Berendsen pressure coupler; the temperature was maintained at 300 K by the Berendsen temperature coupling (79) coupled with protein, membrane and solvent; LINCS Algorithm (80) is used to constrain bonds involving hydrogen atoms. For long-distance interactions, periodic boundary conditions and particle grid Ewald (81) are used. For short-range interactions, a cut-off value of 12 Å was used. Finally, Parrinello-Rahman pressure coupling (82) and Nosé-Hoover temperature coupling (83) were used to perform 200 ns unlimited production simulations for each system. Use GROMACS 2019.3 (84, 85) for simulation.

In order to calculate the free energy distribution of the binding of IFM to its binding site, the distance between the COM of F1486 and the COM of N1659 is used as the reaction coordinate using umbrella sampling. The width of each umbrella sampling window is 1 Å, and the position of the IFM particles is limited by applying a harmonic potential on the reaction coordinate. A short 100 ps long NPT balance was performed using 1 atmosphere pressure maintained by the Berendsen pressure coupler and 300 K controlled by the Berendsen thermostat. A 10 ns production simulation is then performed in each window. The N1659 residue (located at the IFM particle binding site) is constrained throughout the simulation process, and the constraining potential constant is 1,000 kJ ⋅ mol-1 ⋅ nm-2. The weighted histogram analysis method (WHAM) (86) in GROMACS (gmx wham) is used to combine the data from all umbrella sampling windows to calculate the free energy distribution.

When the distance between the Cβ atoms of a residue pair is less than 6.7 Å, it is defined as contact. These are calculated using MD-TASK (87).

Electrophysiology data has been deposited in Zenodo (https://doi.org/10.5281/zenodo.4778147) (88). The molecular modeling and MD simulation files have been deposited in the Open Science Framework (OSF) repository (https://osf.io/hsyx4/) (89).

We thank the Lundbeck Foundation (R139-2012-12390), the Danish Independent Research Fund (7025-00097A and 9039-00335B) (all for SAP), SciLifeLab and the Swedish Research Council (VR 2018-04905 to LD) for funding. The MD simulation was performed on the resources provided by the Swedish National Computing Infrastructure at the PDC High-Performance Computing Center. We thank Professor Christian A. Olsen for his support of peptide chemistry and the members of the SAP laboratory for their helpful comments on the manuscript.

↵1I.G. and HH have made the same contribution to this work.

Author contributions: IG, HH, KK, LD and SAP design research; IG, HH and KC conducted research; IG and KK contributed new reagents/analysis tools; HH, KC, LD and SAP analysis data; IG, HH, LD and SAP wrote this paper.

The author declares no competing interests.

This article is directly contributed by PNAS.

This article contains online support information https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.2025320118/-/DCSupplemental.

This open access article is distributed under Creative Commons Attribution 4.0 (CC BY).

Thank you for your interest in advertising on PNAS.

Note: We only ask you to provide your email address so that the people you recommend the page to know that you want them to see it, and that it is not spam. We do not capture any email addresses.

Feedback privacy/legal

Copyright © 2021 National Academy of Sciences. Online ISSN 1091-6490. PNAS is a partner of CHORUS, COPE, CrossRef, ORCID and Research4Life.