Sodium channel b1 subunits are post-translationally modified by tyrosine phosphorylation, S-palmitoylation, and regulated intramembrane proteolysis

Voltage-gated sodium channel (VGSC)b1 subunits aremultifunctional proteins that modulate the biophysical properties and cell-surface localization of VGSCa subunits and participate in cell–cell and cell–matrix adhesion, all with important implications for intracellular signal transduction, cell migration, and differentiation. Human loss-of-function variants in SCN1B, the gene encoding the VGSC b1 subunits, are linked to severe diseases with high risk for sudden death, including epileptic encephalopathy and cardiac arrhythmia. We showed previously that b1 subunits are post-translationally modified by tyrosine phosphorylation. We also showed that b1 subunits undergo regulated intramembrane proteolysis via the activity ofb-secretase 1 andg-secretase, resulting in the generation of a soluble intracellular domain, b1-ICD, which modulates transcription. Here, we report that b1 subunits are phosphorylated by FYN kinase. Moreover, we show that b1 subunits are S-palmitoylated. Substitution of a single residue in b1, Cys-162, to alanine prevented palmitoylation, reduced the level of b1 polypeptides at the plasma membrane, and reduced the extent of b1-regulated intramembrane proteolysis, suggesting that the plasma membrane is the site of b1 proteolytic processing. Treatment with the clathrin-mediated endocytosis inhibitor, Dyngo-4a, restored the plasma membrane association of b1-p.C162A toWT levels. Despite these observations, palmitoylation-null b1-p. C162A modulated sodium current and sorted to detergent-resistant membrane fractions normally. This is the first demonstration of S-palmitoylation of a VGSC b subunit, establishing precedence for this post-translational modification as a regulatorymechanism in this protein family.

Human variants in VGSC genes are linked to the developmental and epileptic encephalopathies (DEEs) and to cardiac arrhythmia. Loss-of-function variants in SCN1B, encoding b1, result in early infantile developmental and epileptic encephalopathy (EI-DEE) and generalized epilepsy with febrile seizures plus (2,17). Scn1b-null mice model EI-DEE, with severe spontaneous seizures of multiple etiologies, ataxia and sudden death in the third week of life (18). Consistent with loss of b1-mediated cell-cell and cell-matrix adhesion, Scn1b-null mice have neuronal pathfinding and fasciculation defects in the brain (6,8). SCN1B is expressed in the heart in addition to the brain. Scn1b-null mice have prolonged QT and RR intervals. Scn1b-null ventricular cardiomyocytes have increased sodium current (I Na ), altered calcium handling, altered intercalated disk structure, and prolonged action potential duration (13,19,20). SCN1B variants are associated with human cardiac disease, including Brugada syndrome and atrial fibrillation (21)(22)(23)(24)(25). Taken together, these data show that SCN1B is critical for the regulation of excitability in multiple organ systems.
b1 subunits undergo regulated intramembrane proteolysis (RIP) through the sequential activity of b-site amyloid precursor protein (APP) cleaving enzyme-1 (BACE1) and This article contains supporting information. ‡ These authors contributed equally to this work. * For correspondence: Lori L. Isom g-secretase (26). 5 BACE1 cleavage, the rate-limiting step in this process, releases the extracellular b1 Ig domain, which functions as a CAM ligand to stimulate neurite outgrowth (28,29). The remaining membrane-bound C-terminal fragment (b1-CTF) is cleaved by g-secretase in the lumen of the membrane, generating a soluble, intracellular domain, b1-ICD, that translocates to the nucleus to regulate transcription (26,30). Thus, b1 RIP plays important roles in neurite outgrowth, cell migration, cell adhesion, and transcription (31,32). BACE1-and g-secretase-mediated processing of the wellstudied RIP substrate, APP, is regulated by S-palmitoylation, the covalent addition of a 16-carbon fatty acid to cysteine residues via thioester bond formation (33). Palmitoylation targets APP to its proper membrane domains, bringing it in close proximity to proteolytic enzymes for subsequent cleavage (33). Here, we asked whether post-translational modification of b1 subunits by tyrosine phosphorylation or S-palmitoylation could regulate its plasma membrane localization and subsequent RIP. In contrast to b1-ankyrin association, for which b1 tyrosine phosphorylation is critical (10), we found that the tyrosine phosphorylation state of b1 has no effect on its plasma membrane localization, intramembrane cleavage, or ability to modulate I Na . We report for the first time that b1 subunits are S-palmitoylated in mouse brain. Using heterologous cells, we found that substitution of cysteine residue 162 with alanine abolishes b1 palmitoylation, decreases the fraction of b1 in the plasma membrane as assessed by surface biotinylation, and thus reduces the level of b1 that is available for RIP. Treatment of cells with the clathrin-mediated endocytosis inhibitor, Dyngo-4a, restores b1-p.C162A to WT levels at the plasma membrane, suggesting that S-palmitoylation confers plasma membrane stability to b1. Finally, we show that b1-mediated modulation of I Na and b1 sorting to detergent-resistant membrane fractions do not depend on b1 palmitoylation. Taken together, our work suggests that multiple post-translational modification events regulate b1 function. Tyrosine phosphorylation regulates the association of b1 subunits with ankyrin but does not affect their plasma membrane localization. In contrast, S-palmitoylation regulates the cell-surface localization of b1 and consequently its extent of RIP, indicating that b1 cleavage occurs at the plasma membrane. This work provides novel insights into b1 subunit function that may aid in understanding the mechanism of SCN1B-associated pathophysiologies.

Results
b1 RIP occurs independently of b1 tyrosine phosphorylation b1 tyrosine residue 181, located in the intracellular domain, is important for b1-mediated downstream signaling (10) (Fig.  1A). In previous work, we used phosphorylation-null and phosphomimetic mutant constructs to show that phosphorylation of residue Tyr-181 is a key regulatory mechanism for ankyrin binding (10). Our work in cerebellar granule neurons demonstrated that b1-b1 trans-homophilic adhesion-mediated neurite outgrowth is inhibited by the admin-istration of g-secretase inhibitors and in neurons isolated from fyn-null mice (8,31). Taken together, these data suggested that b1-mediated neurite outgrowth requires association of the b1 intracellular domain with ankyrin via residue Tyr-181, which then triggers b1 RIP. Here, we tested the hypothesis that b1 tyrosine phosphorylation regulates cleavage using a multidisciplinary approach.
We used a cell-free fyn kinase assay (Promega) in which ADP was measured via luciferase activity and positively correlated to kinase activity to determine whether fyn directly phosphorylates a b1 peptide, QENASEYLAITC, at position Tyr-7, which is equivalent to position Tyr-181 in the full-length polypeptide.
Poly-E 4 Y 1 peptide was used as a positive control for fyn kinase activity (Fig. 1B). Inclusion of WT b1 peptide in the assay increased luciferase activity by ;3-fold over the no-substrate control. In contrast, luciferase activity levels in the presence of Y181E b1 peptide (pYb1) were not different from the no-substrate control. These data indicate that fyn kinase can directly phosphorylate b1 at the Tyr-181 position (Fig. 1B).
To understand whether b1 phosphorylation at Tyr-181 affects b1 RIP, we generated phosphorylation-null, b1-p. Y181A-V5-2A-eGFP, and phosphomimetic, b1-p.Y181E-V5-2A-eGFP mutant constructs, based on our previous work (10). Chinese hamster lung (CHL) cell lines stably overexpressing each construct were generated, and plasma membrane localization of each mutant polypeptide was investigated using cellsurface biotinylation assays (29,30). Similar to WT b1, both b1 mutants were detected in the plasma membrane fraction (Fig.  1C). Quantification of these results showed no differences in the plasma membrane association of any of the mutants compared with WT b1 (Fig. 1D). To determine whether b1 phosphorylation at residue Tyr-181 regulates BACE1-and g-secretase-mediated cleavage of b1, each cell line was treated with vehicle (0.1% DMSO) or 1 mM of the g-secretase inhibitor DAPT and analyzed by Western blotting. Treatment with DAPT leads to an accumulation of the intermediary cleavage product, b1-CTF, generated by BACE1 cleavage. 1 We found that levels of b1-CTF were generated similarly to WT in both mutant lines and accumulated similarly to WT following treatment with DAPT, suggesting that neither BACE1 nor g-secretase cleavage of b1 depends on its phosphorylation state (Fig. 1, E and F). b1-CTF fragments generated from DAPT treatment of b1 phosphorylation mutants had variable molecular masses, compared with WT b1-CTF. This effect may be similar to previously observed shifting of the cleavage site resulting from the introduction of mutations into other BACE1 substrates (34) and warrants future investigation. These results also suggest that our previous work, demonstrating that g-secretase inhibitors block b1mediated neurite outgrowth (31), may have implicated g-secretase substrates other than b1. Y181A-V5-2A-eGFP, or b1-p.Y181E-V5-2A-eGFP with hNa v 1.5 increased I Na density compared with eGFP alone, suggesting that the phosphorylation state of residue Tyr-181 does not affect the ability of b1 to modulate I Na density (Fig. 2, A and B). In agreement with the observed increase in peak I Na density, peak conductance was increased by co-expression of WTb1-V5-2A-eGFP, b1-p.Y181A-V5-2A-eGFP, or b1-p.Y181E-V5-2A-eGFP (Fig. 2C). No changes in capacitance were observed (Fig. 2D). We observed no effects of any of the b1 constructs on the voltage dependence of I Na activation or inactivation compared with the eGFP control (Table S1).
b1 is S-palmitoylated in vitro and in vivo APP is a type I transmembrane topology BACE1 substrate that results in the generation of Ab peptides to form protein aggregates that can contribute to Alzheimer's disease pathogenesis (35). BACE1-mediated cleavage of APP is highly dependent on its proper localization in lipid raft microdomains (33). The post-translational lipid modification, S-palmitoylation, is required for proper targeting of APP to lipid rafts (33). In lipid rafts, palmitoylated APP interacts with BACE1 for subsequent cleavage (33). Given its similarity to APP, we asked whether b1 is also S-palmitoylated and, if so, whether S-palmitoylation regulates b1 subcellular localization and RIP. To assess steadystate palmitoylation of b1, we used the acyl resin-assisted capture (RAC) assay, in which free cysteines are first blocked with the alkylating reagent methyl methanethiosulfonate (MMTS). Then thioester bonds between the cysteine residue of the protein and the palmitate are cleaved using the reducing agent hydroxylamine (HA/NH 2 OH) to liberate the previously palmitoylated cysteine residues. The liberated cysteines are selectively captured on activated thiol-Sepharose beads and eluted, allowing for specific immunoblotting of palmitoylated proteins of interest (36). We used endogenous flotillin-1, a known constitutively palmitoylated protein (37), as a positive control for the acyl RAC assay. In cells stably expressing WT b1-V5-2A-eGFP, we observed that b1 is S-palmitoylated, as evidenced by hydroxylamine-dependent binding of b1-V5 to Sepharose beads ( Fig. 3A and Fig. S1).
We next asked whether b1 palmitoylation occurs in vivo. Using C57Bl/6J adult mouse whole-brain lysates subjected to acyl RAC, we observed b1 palmitoylation, as evidenced by hydroxylamine-dependent binding of endogenous b1 to Sepharose beads (Fig. 3B). These data demonstrate that b1 is S-palmitoylated in vitro, as well as in mouse brain, providing feasibility for investigating the role of S-palmitoylation in b1 localization and proteolytic processing. . b1 phosphorylation at residue Tyr-181 does not affect its RIP. A, schematic of b1 identifying the location of the phosphorylation site, Tyr-181, as well as BACE1 and g-secretase cleavage sites. B, a b1 peptide (QENASEYLAITC) is directly phosphorylated at Tyr-181 (Tyr-7 in the peptide) by fyn kinase in a cell-free assay (n = 3). C, cell surface biotinylation indicates that similar to WT b1-V5, b1-p.Y181A-V5, and b1-p.Y181E-V5 are localized to the plasma membrane (n = 3). D, quantification of C. Plasma membrane fraction was normalized to total protein for that construct (% plasma membrane/total) and normalized again to WT b1-V5 plasma membrane levels. Significance (p , 0.05) was determined using a one-way ANOVA. E, WT b1-V5, b1-p.Y181A-V5, and b1-p.Y181E-V5 are cleaved by BACE1 and g-secretase (n = 4). F, quantification of E. Protein levels were normalized to the loading control and reported as fold change respective to the vehicle-treated group. Significance (p , 0.05) was determined using Student's t test between DMSO-and DAPT-treated constructs. One-way ANOVA was utilized to compare between constructs. IB, immunoblotting; ns, not significant. b1 is S-palmitoylated at cysteine residue 162 To identify the palmitoylated cysteine residue(s) in b1, we first determined the number of palmitoylated sites on b1 using a mass-tag labeling technique, acyl polyethylene glycol (acyl PEG) exchange. In this assay, free cysteine residues are blocked with the alkylating reagent, N-ethylmaleimide. Palmitate groups linked to cysteine residues are subsequently cleaved with hydroxylamine and replaced with a 10-kDa PEG-maleimide group, resulting in a 10-kDa shift in the apparent molecular mass of the polypeptide for each palmitoylated cysteine residue, as detected by Western blotting (38). CHL cells stably expressing WT b1-V5-2A-eGFP were subjected to acyl PEG exchange. We observed a single 10-kDa shift in the apparent molecular mass of b1-V5, which occurred in a hydroxylaminespecific manner, suggesting that b1 is singly palmitoylated ( Fig.  4A and Fig. S2). Based on homology models with the CAM myelin P0, which is S-palmitoylated at cysteine 153, we predicted that b1 would be palmitoylated at the homologous residue, cysteine 162 (39). Using site-directed mutagenesis, we engineered a cDNA construct in which b1 cysteine residue 162 was converted to an alanine and generated a stable b1-p. C162A-V5-2A-eGFP CHL cell line. To test the effects of Figure 2. b1-Mediated modulation of I Na is not dependent on phosphorylation of residue Tyr 181 . HEK-hNa v 1.5 cells were transiently co-transfected with WTb1 (black circles), b1-p.Y181E (blue squares), or b1-p.Y181A (purple triangles). HEKhNa v 1.5 cells transfected with eGFP (light gray circles) were used as negative controls. I Na was recorded in response to a series of voltage steps between 2120 and 130 mV in 5-mV increments, from a holding potential of 2120 mV for 200 ms. A, I Na current-voltage (I-V) relationship. B, peak I Na is significantly increased with co-expression of WTb1, b1-p.Y181E, or b1-p.Y181A over eGFP. C, peak conductance (G max ) is significantly increased with co-expression of WTb1, b1-p.Y181E, or b1-p.Y181A over eGFP. The data were obtained by fitting individual activation or inactivation curves to a Boltzmann equation. G Na was calculated from G Na = I Na /(V 2 V rev ), where I Na is the peak I Na during the test depolarization (V), and V rev is the reversal potential. D, there were no significant differences in capacitance (in pF) measured from each cell. The data were acquired using pClamp 11 (Molecular Devices) software. The data presented in A-C result from at least three separate transfections and are presented as means 6 S.E. **, p , 0.01 versus eGFP; ***, p , 0.001 versus eGFP; ****, p , 0.0001 versus eGFP. Figure 3. b1 is S-palmitoylated in CHL cells and in mouse brain. A, CHL cells stably expressing b1-V5-2A-eGFP were processed for the acyl RAC assay to detect S-palmitoylation. S-Palmitoylation of b1-V5 is detected in CHL cells using an antibody against V5, as shown by the anti-V5 signal in the 1HA lanes, compared with the expected absence of signal in the 2HA lanes. Flotillin-1 is used as a positive control for the acyl RAC assay (n = 3). B, whole mouse brain lysates were subjected to the acyl RAC assay to detect S-palmitoylation. S-Palmitoylation of endogenous b1 is detected in whole mouse brains, using an antibody against the C terminus of b1, as shown by the anti-b1 signal in the 1HA lanes, compared with the expected absence of signal in the 2HA lanes (n = 3). IB, immunoblotting.
the C162A mutation on b1 palmitoylation, we subjected b1-p. C162A-V5-2A-eGFP CHL cell lysates to both acyl PEG exchange and acyl RAC. Using acyl PEG exchange showed a hydroxylamine-dependent PEGylation-induced mass shift in WT b1-V5 but not in b1-p.C162A-V5, suggesting that b1-p. C162A-V5 cannot be palmitoylated ( Fig. 4A and Fig. S2). Fig.  4A demonstrates a faint, yet present "apo" b1-p.C162A-V5 signal in the 2HA lanes, which represents unmodified polypeptide. The control lane is included to show that any PEGylationinduced mass shift observed in the 1HA lanes is hydroxylamine-dependent. In this instance, despite the faint "apo" b1-p. C162A-V5 signal in the 2HA lanes, the b1-p.C162A-V5 signal in the lysate lane is comparable with the WT b1-V5 signal in the lysate lane, suggesting that the lack of mass shift observed in the 1HA lanes for b1-p.C162A-V5 mutant is not due to the lack of starting material but rather due to the loss of the only palmitoylated cysteine residue in b1. We confirmed these results by subjecting b1-p.C162A-V5-2A-eGFP CHL cell lysates to acyl RAC, in which we observed a 92% reduction in the hydroxylamine-dependent signal for b1-p.C162A-V5, compared with WT b1-V5 (Fig. 4, B and C). These results demonstrate that b1 is singly palmitoylated at cysteine 162 and that mutating this site to alanine completely abolishes b1 palmitoylation (Fig. 4D).

b1 S-palmitoylation regulates its plasma membrane localization
We asked whether palmitoylation regulates b1 association with the plasma membrane by comparing b1-p.C162A-V5 to WT b1-V5 in cell surface biotinylation assays. We found the level of b1-p.C162A-V5 polypeptide associated with the plasma membrane to be 77% less than WT (1.00 6 0.1621 for WT versus 0.2271 6 0.0142 for b1-p.C162A), as indicated by the reduced b1-p.C162A-V5 signal in the neutravidin-selected lane (normalized to total protein expression), compared with WT b1-V5 (Fig. 5, A and B, and Fig. S3). HSP90 was used as a negative control, as in previous work, to ensure that no intracellular biotinylation is occurring (40,41). These results suggest that Spalmitoylation promotes plasma membrane association of b1.

S-Palmitoylation regulates b1 endocytosis but not sorting into detergent-resistant membranes
Palmitoylation has been shown to regulate the partitioning of certain proteins to cholesterol-rich lipid raft microdomains (42). We asked whether palmitoylation governed the localization of b1 to lipid rafts, similarly to what has been shown previously for APP (33). b1 is known to localize to detergent-resistant membrane (DRM) fractions of mouse brain and primary neuronal cultures (8,26). To verify the presence of b1 in DRM fractions in CHL cells stably expressing WT b1-V5, we prepared DRMs using density gradient centrifugation and analyzed them by Western blotting using anti-V5 antibody. We found that WT b1-V5 was present in both detergent-insoluble fractions, marked with flotillin-1, and in detergent-soluble fractions, marked with transferrin receptor (TfR), similar to previous results (29) (Fig. 6A). We observed no differences in this distribution for the palmitoylation-null mutant, b1-p.C162A- . b1 is S-palmitoylated at cysteine 162. A, CHL cells stably expressing b1-V5-2A-eGFP or b1-p.C162A-V5-2A-eGFP were processed for the acyl PEG assay to determine the number of palmitoylated cysteines on b1. A single 10-kDa shift in the apparent molecular mass of b1 was observed in a hydroxylamine-specific manner, using an antibody against V5. Compared with WT b1-V5, b1-p.C162A-V5 showed no hydroxylamine-dependent PEGylation-induced mass shift, suggesting loss of the palmitoylation site upon mutation of cysteine to alanine at b1 residue 162 (n = 3). B, CHL cells stably expressing b1-p.C162A-V5-2A-eGFP were subjected to the acyl RAC assay to detect the effect of the mutation of b1 S-palmitoylation. S-Palmitoylation of b1-p.C162A-V5 is not detected in CHL cells using an antibody against V5, as shown by the absence of anti-V5 signal in the 1HA lanes, compared with WT b1-V5 (n = 3). C, quantification of B. The signal from the 1HA lanes was normalized to total protein for that construct and normalized again to WT b1-V5 palmitoylation levels. Significance (p = 0.0003) was determined using a Student's t test. D, cartoon diagram of b1 and its identified palmitoylation site. IB, immunoblotting.
V5, as evidenced by the presence of anti-V5 signal in both flotillin-1-marked DRMs and TfR-marked nonlipid raft domains (Fig. 6A). These data suggest that although palmitoylation of b1 is necessary for its proper association with the plasma mem-brane, it does not regulate the partitioning of b1 into lipid-raft domains.
Because of the observed reduction of b1-p.C162A at the plasma membrane, we compared the extent of WT b1 versus  3-4). B, quantification of A. Quantification represents the amount of b1 at the membrane/total b1 protein expression, normalized to WT plasma membrane levels, to obtain a percent reduction in plasma membrane localization in the mutant relative to WT b1 (n = 3-4 for each construct). Significance was determined using Student's t test. versus b1-p.C162A-V5 accumulation at the cell surface by biotinylation. Anti-HSP90 antibody was used as a negative control for the plasma membrane fraction, and anti-TfR antibody was used as a positive control for endocytosis inhibition with Dyngo-4a. We found that Dyngo-4a administration normalized the level of b1-p.C162A-V5 in the plasma membrane fraction to that of WT b1, implicating clathrin-dependent endocytosis in this process (Fig. 6B). Pulldown experiments are inherently variable. This variability accounts for the apparent presence of higher levels of plasma membrane association of b1-p.C162A in the DMSO-treated control samples compared with WT b1 DMSO-treated control samples. It is important to note that, in this particular instance, pulldown was more efficient in the mutant DMSO-treated samples compared with WT, as evident by the higher TfR signal. This work adds new information to the VGSC field, showing that WT b1 subunits undergo endocytosis via a clathrin-dependent mechanism and suggesting that the palmitoylation may confer plasma membrane stability to b1 polypeptides.
The level of b1-p.C162A RIP is reduced compared with WT We hypothesized that reduction in plasma membrane localization of b1-p.C162A-V5 would reduce its level of RIP. To test this hypothesis, we treated stable b1-V5-2A-eGFP or b1-p. C162A-V5-2A-eGFP CHL cells with vehicle (0.1% DMSO) or 1 mM DAPT and assessed the formation of the ;20-kDa b1 intramembrane CTF by Western blotting analysis. As shown previously, inhibition of g-secretase by DAPT results in b1-CTF accumulation in the presence of normally occurring BACE1 cleavage. 1 If BACE1-mediated b1 cleavage were altered or reduced, DAPT administration would result in reduced levels of b1-CTF accumulation because of a reduction in available substrate for g-secretase-mediated RIP. As expected, DAPT treatment of CHL cells stably expressing WT b1-V5 resulted in b1-CTF accumulation (Fig. 7, A and B). In contrast, using the b1-p.C162A-V5 mutant construct as substrate resulted in an 80% loss in the level of cleavage product compared with WT (Fig. 7, A and B). This result suggests that BACE1 cleaves the small fraction of b1-p.C162A-V5 that is localized to the plasma membrane, generating a reduced level of b1-CTF in response to DAPT treatment, compared with WT b1. These data demonstrate that b1 palmitoylation promotes b1 plasma membrane localization, which allows RIP to occur.

b1-mediated modulation of I Na is not affected by palmitoylation
We next asked whether palmitoylation-deficient b1-p.C162A-V5 could modulate I Na . WT b1-V5-2A-eGFP, b1-p.C162A-V5-2A-eGFP, or eGFP were transiently expressed in HEK-hNa v 1.5 cells. We found that WT b1-V5 and b1-p.C162A-V5 increased I Na density to a similar extent, compared with the soluble eGFP control (Fig. 8, A and B). In agreement with the observed increase in peak I Na density, peak conductance was increased by coexpression of WT b1-V5-2A-eGFP or b1-p.C162A-V5-2A-eGFP (Fig. 8C). No changes in capacitance were observed (Fig. 8D). We observed no effect of either b1 construct on the voltage dependence of I Na activation or inactivation (Table S1). These data suggest that the small fraction of b1-p.C162A-V5 that remains properly localized to the plasma membrane (Fig. 5, A and B) is sufficient to modulate I Na .

Discussion
VGSC b1 subunits are multifunctional signaling molecules. In addition to modulating the gating, kinetics, and localization of VGSC a subunits, b1 subunits function in cell-cell and cellmatrix adhesion, cell migration, calcium handling, modulation of potassium currents, neuronal pathfinding, fasciculation, and neurite outgrowth (2). Human SCN1B loss-of-function variants are linked to EI-DEE and cardiac arrhythmia, often resulting in sudden death (2).
It is important to understand how b1 subunits are posttranslationally processed and whether this differential Compared with WT b1-V5, b1-p.C162A-V5 shows little to no cleavage product (n = 3). B, quantification of A. Protein levels were normalized to the loading control and reported as fold change respective to the vehicle-treated group. Significance (p , 0.05) was determined using one-way ANOVA and plotted as the mean 6 S.D. IB, immunoblotting; ns, not significant.
processing affects their functionality. We showed previously that b1 is tyrosine-phosphorylated (10). Here, we extend those findings to show that fyn kinase directly phosphorylates a b1 peptide, supporting our previous, indirect, hypothesis using neurons from fyn-null mice (8). b1 subunits undergo RIP through the activity of BACE1 and g-secretase (26). 1 Initial cleavage of b1 by BACE1 sheds the b1 Ig ectodomain and leaves behind the b1 C-terminal fragment in the membrane, which undergoes subsequent cleavage by g-secretase to generate a soluble intracellular domain, the b1-ICD (26). Recent work by our laboratory demonstrated that the b1-ICD can translocate to the nucleus, where it participates in transcriptional regulation to ultimately modulate sodium, potassium, and calcium currents in mouse ventricular myocytes (44). 1 Overexpression of the b1-ICD resulted in the down-regulation of genes related to proliferation, immune response, and sodium and potassium channels. In contrast, loss of b1-ICD in Scn1bnull mouse cardiac ventricular tissue resulted in the up-regulation of these gene groups, suggesting that the b1-ICD may act as part of a transcriptional repressor complex under normal physiological conditions. Here, we asked whether b1 tyrosine phosphorylation or b1 S-palmitoylation can regulate b1 RIP. Using b1 phosphorylation-null and phosphomimetic mutant constructs, we found that b1 tyrosine phosphorylation at Tyr-181 does not regulate b1 RIP.
Here, we demonstrate that b1 is lipid-modified by S-palmitoylation in the brain, that S-palmitoylation, but not tyrosine phosphorylation, regulates b1 RIP by facilitating b1 localization to the plasma membrane, and that b1 subunits undergo clathrin-mediated endocytosis, at least in the absence of VGSC a subunits. Our results suggest that b1 must be associated with the plasma membrane for RIP to occur and that S-palmitoylation at residue Cys-162 stabilizes b1 plasma membrane association and reduces its level of endocytosis. Given that S-palmitoylation has been shown to contribute to protein stability in other work (34), it is possible that reduction of palmitoylationnull b1-C162A protein expression compared with WT b1 is due to decreased b1 protein stability. It will be important to test this hypothesis in follow-up studies. Residue Cys-162, at which b1 is palmitoylated, is conserved in VGSC b3 subunits and thus may implicate palmitoylation as a similar regulatory mechanism in these proteins. The absence of this conserved residue in VGSC b2 and b4 suggests alternative regulatory mechanisms (Fig. S4). Previous work has shown that palmitoylation of APP promotes its RIP through regulating its subcellular localization (33). Although other RIP substrates, e.g. LRP1 and N-cadherin, have been shown to be palmitoylated, whether palmitoylation also regulates their RIP is not known (30,41).
The biochemical experiments described here were performed in the absence of VGSC a subunits. It will be interesting in future work to consider the effects of a subunit co-expression on b1 subunit post-translational processing. Although a large body of work has shown that b1 subunits function as molecular chaperones for VGSC a subunits to the plasma membrane (46), there is no evidence to support the promotion of b1 subunit cell surface expression by a subunits. We do not know whether the reduction in b1-C162A cell surface expression compared with WT b1 shown here could be due to the absence Figure 8. b1-Mediated modulation of I Na density is not dependent on S-palmitoylation of b1 at residue Cys-162. HEK-hNa v 1.5 cells were transiently co-transfected with eGFP, WTb1, or b1-p.C162A. A, I Na current-voltage (I-V) relationship. B, peak I Na is significantly increased with co-expression of WTb1 or b1-p.C162A over eGFP. C, peak conductance (G max ) is significantly increased with co-expression of WTb1 or b1-p.C162A over eGFP. D, no significant difference in the capacitance (in pF) measured from each cell. The data presented in A-C result from at least three separate transfections and are presented as means 6 S.E. *, p , 0.05 versus eGFP; ***, p , 0.001 versus eGFP; ****, p , 0.0001 versus eGFP. of a co-expressed a subunit. One possible interpretation of our electrophysiological data showing that b1-C162A increases I Na density similar to WT is that the presence of an a subunit changes the behavior of this mutant, resulting in intersubunit, synergistic effects. Biochemical assessment of whether a subunits can promote b1 cell surface expression would be complicated by the noncovalent association of these subunits. Noncovalent a-b1 association precludes separation of the pool of b1 subunits associated with a from those that are not, using standard immunoprecipitation techniques. Furthermore, because VGSC a and b1 subunits are each ankyrin-binding proteins (9,10,49,50), they may associate in a complex, as assessed by co-immunoprecipitation, but not physically interact. Nevertheless, examining the effects of b1 palmitoylation and phosphorylation in the presence of VGSC A subunits, both in heterologous systems and within the channelome complex in brain and heart in vivo, will be interesting future directions of this work.
The effects of b1 co-expression on I Na voltage-dependent properties in heterologous systems are inconsistent throughout the literature (46). Here, we show that co-expression of hNa v 1.5 with b1 or the mutant b1 subunits did not statistically change the voltage dependence of activation or steady-state inactivation compared with a (eGFP) alone. In contrast, we observed increased peak I Na density and peak conductance, in agreement with the well-established role of WT b1 subunits in increasing VGSC function by increasing their plasma membrane expression (46). Interestingly, co-expression of the Tyr-181 or Cys-162 b1 mutants resulted in increased peak I Na density and maximal conductance. Finally, although Na v 1.5 is the major cardiac VGSC, a sizable body of work in recent years has also identified Na v 1.5 in the brain (51)(52)(53), suggesting that our results may be applicable to other VGSCs.
Importantly, Scn1b deletion in mice and heterologous b1 expression in cell culture are not comparable. Scn1b-null mouse cardiac myocytes have increased I Na density because of developmentally regulated increases in the expression of Scn3a and Scn5a mRNA and protein (19,20). As described above, our group has recently shown that b1 is a substrate for RIP by BACE1 and g-secretase in vivo (44). 1 The cleaved C-terminal fragment of b1 can translocate to the nucleus, resulting in reduced expression of a number of genes, including ion channels. We have proposed that the absence of b1 subunits in Scn1b-null animals results in the absence of gene repression and subsequent increased ion channel expression. This situation is very different from acute, heterologous overexpression of VGSC b1 and a subunit cDNAs, in which b1 subunits function to chaperone a subunits to the plasma membrane, as demonstrated here. The required genetic regulatory elements are not present in the cDNA plasmids. Moreover, whereas b1chaperone function is lost in Scn1b-null mouse myocytes, I Na reductions are not observed. It is likely that the presence of a host of other protein components of the channel proteome that retain VGSCs in the cardiac myocyte plasma membrane (54).
In conclusion, S-palmitoylation is a reversible post-translational modification, making it a highly dynamic and tunable process (27,43,45,55). Multiple palmitoyl acyltransferase enzymes, which mediate substrate palmitoylation, as well as protein thioesterases, which depalmitoylate substrates, are implicated in this process. The molecular identities of the enzymes that palmitoylate and depalmitoylate b1 subunits are not known but may be identified in the future to discover novel targets for SCN1B-linked pathophysiology. In addition, we do not yet know whether the level of b1 palmitoylation can be dynamically regulated by extracellular stimuli or by altered excitability, but this information will be important to elucidate because attempts to implicate this post-translational processing in disease mechanisms move forward. It is possible that b1-mediated transcriptional regulation via RIP can be manipulated by altering the level of b1 palmitoylation. Additionally, the effects of SCN1B disease-linked variants on b1 subunit palmitoylation, RIP, and transcriptional regulation should be considered.

Experimental procedures
Cell culture CHL cell lines stably expressing b1 or b1 mutants and stable HEK-hNa v 1.5 cells were grown in Dulbecco's modified Eagle's medium with 5% heat-inactivated fetal bovine serum, penicillin/streptomycin, and 600 mg/ml at 37°C, 5% CO 2 . Stable cell lines were generated by transfecting parental CHL cells with 1 mg of cDNA with 5 ml of Lipofectamine 2000. 48 h after transfection, the cells were split into medium containing 600 mg/ml G418 (Gibco). The cells were grown for approximately 1 week or until eGFP-positive colonies were large enough to isolate. Individual colonies were selected and grown until confluent and characterized by Western blotting analysis. Patch-clamp experiments used transient transfection of b1 cDNAs into stable HEK-hNav1.5 cells. 1 mg of cDNA was transfected with 5 ml of Lipofectamine 2000 (Invitrogen). Approximately 12 h posttransfection, the cells were plated for electrophysiological recordings. Patch clamp was completed ;12 h after final plating.

Expression vectors
A synthesis-optimized human WT b1-V5-2A-eGFP cDNA was generated by gBLOCK from Integrated DNA Technologies based on NP_001028.1. The bicistronic cDNA construct included an in-frame b1 C-terminal V5 epitope tag followed by a self-cleaving 2A peptide and eGFP to facilitate immune detection of b1 as well as transfected cells by eGFP. b1-p. C162A-V5-2A-eGFP, b1-p.Y181A-V5-2A-eGFP, and b1-p. Y181E-V5-2A-eGFP were generated by site-directed mutagenesis using the WT b1-V5-2A-eGFP cDNA construct in pENTR-SD/D TOPO as the template. The eGFP alone control was generated by PCR from their respective full-length template cDNAs containing WT b1-V5-2A-eGFP. Using the Gateway cloning system, all constructs were moved from pENTR-SD/D-TOPO to pcDNAdest40 via LR Clonase reaction according to the manufacturers' protocol.
The amino acid numbering scheme for the b1 polypeptides used throughout the paper excludes the N-terminal, 19-amino acid signal peptide, as described in the original report of the b1 cDNA sequence (4).

Animals
The animals were housed in the Unit for Laboratory Animal Medicine at the University of Michigan. All procedures were performed in accordance with National Institutes of Health guidelines with approval from the University of Michigan Institutional Animal Care and Use Committee.

Measurement of I Na by whole-cell voltage clamp
Voltage-clamp recordings were performed at room temperature in the whole-cell configuration using an Axopatch 700B amplifier and pClamp (versions 11, Axon Instruments, Foster City, CA) with 1.5-2.5 MV patch pipettes. I Na was recorded in the presence of a bath solution containing 120 mM NaCl, 1 mM BaCl 2 , 2 mM MgCl 2 , 0.2 mM CdCl 2 , 1 mM CaCl 2 , 10 mM HEPES, 20 mM TEA-Cl, and 10 mM glucose (pH 7.35 with CsOH; osmolarity was 300-305 mOsm). Fire-polished patch pipettes were filled with an internal solution containing 1 mM NaCl, 150 mM N-methyl-D-glucamine, 10 mM EGTA, 2 mM MgCl 2 , 40 mM HEPES, 25 mM phosphocreatine-tris, 2 mM MgATP, 0.02 mM Na 2 GTP, and 0.1 mM leupectin (pH 7.2 with H 2 SO 4 ). Sodium current was recorded in response to a series of voltage steps between 2100 and 130 mV in 5-mV increments, from a holding potential of 290 mV for 200 ms. A step back to 220 mV for 200 ms was used to determine the voltage dependence of inactivation. Series resistance was compensated 40-65%, and leak subtraction was performed by application of a standard P/4 protocol. Normalized conductance and inactivation curves were generated as described previously (47). Current densities were determined by dividing current amplitude by the cell capacitance (C m ), as determined by application of 110-mV depolarizing test pulses.

Cleavage assays
Stably transfected CHL cells were grown to ;70% confluence in 100-mm tissue culture plates. The cells were treated with vehicle (0.1% DMSO) or 1 mM DAPT (Cayman Chemical), as indicated in the figure legends. After a 24-h treatment, the cells were harvested, and the membranes were prepared. Briefly, the cell pellets were harvested and resuspended in 50 mM Tris, pH 8.0, with Complete protease inhibitors, EDTA-Free (Roche). On ice, the cells were homogenized 10 times with a Dounce homogenizer followed by sonication. To remove nuclei and insoluble material, the lysates were spun at 2,537 3 g for 10 at 4°C. The supernatant was removed and spun at 80,000 3 g for 15 min at 4°C. The supernatant was removed, and the membrane-containing pellets were resuspended in 133 ml of 50 mM Tris, pH 8.0, with Complete protease inhibitors, EDTA-Free (Roche) and sonicated on ice to resuspend the membrane-containing pellets. The samples were heated at 85°C for 10 min and separated using 12% SDS-PAGE gels, and Western blots were performed as described below.

Surface biotinylation assays
Stably transfected cells were grown to 90-100% confluence in 100-mm tissue culture plates. Cell surface proteins were biotinylated using a cell surface protein isolation kit (Pierce) following the manufacturer's instructions and as previously described (47). All samples were heated at 85°C for 10 min and separated on 10% SDS-PAGE gels. Western blots were performed as described above. For endocytosis experiments, prior to cell surface biotinylation, the cells were treated with vehicle (0.1% DMSO) or 1 mM Dyngo-4a for two h in a 37°C incubator with 5% CO 2 . HSP90 was used as a negative control to detect intracellular biotinylation. Any experiments in which intracellular contamination in the plasma membrane fraction was detected (e.g. HSP90 signal in the neutravidin-selected lane) were excluded.

DRM preparations
10-100 mm dishes of CHL cells stably transfected with WT b1-V5-2A-eGFP or p.C162A b1-V5-2A-eGFP were grown to 90-100% confluence. As described previously (48), the cells were washed and resuspended in 2.5 ml of HES buffer (20 mM HEPES, 1 mM EDTA, 250 mM sucrose, pH 7.4) supplemented with 1 mM Na 3 VO 4 and Complete protease inhibitors (Roche). The cells were homogenized by 10 passages through a 22-gauge needle and centrifuged at 245,000 3 g for 90 min at 4°C. The membranes were resuspended with 10 passages through a 22gauge needle in 2.5 ml of MBS buffer (25 mM MES, 15 mM NaCl, pH 6.5) with 1% Triton X-100 and Complete protease inhibitors (Roche) and incubated for 20 min at 4°C. Homogenate was mixed with 2.5 ml of 90% sucrose. 5 ml of 35% sucrose and 2.5 ml of 5% sucrose were overlaid, and the samples were centrifuged at 200,000 3 g in a swinging bucket rotor for 17 h. 1 ml fractions were collected from top to bottom, heated at 85°C for 10 min with sample buffer, and subsequently analyzed by Western blotting.

Western blotting analysis of cell lysates
Cell lysates were prepared either as described above for cleavage assays, surface biotinylation assays, or DRMs, as appropriate. Loading buffer containing SDS, 5 mM b-mercaptoethanol, and 1% DTT was added to samples and heated for 10 min at 85°C. Protein lysates were separated by SDS-PAGE on 10 or 12% Tris-glycine polyacrylamide gels as indicated in the figure legends, transferred to nitrocellulose membrane overnight (16 h, 55 mA, 4°C), and probed with antibodies, as indicated in the figure legends. Incubations with anti-V5 or anti-b1 intra and their respective secondary antibodies were performed using a SnapID with 10-20-min incubations. Anti-a-tubulin, anti-b1 intra anti-TfR, anti-flotillin-1, and anti-HSP90 antibodies were incubated overnight at 4°C. Secondary antibodies were incubated for 1 h at room temperature. Immunoreactive bands were detected using West Femto chemiluminescent substrate (GE Health Sciences) and imaged using an iBrightFL1000 (Invitrogen). Blots for each antibody were individually detected within the linear range using the smart exposure feature included in the software package for the Invitrogen iBright imager. Immunoreactive signals from cleavage assays were quantified using ImageJ and normalized to the level of a-tubulin and subsequently to vehicle-treated samples.

Acyl RAC
Stably transfected cells were grown to ;90% confluence in 100-mm tissue culture plates. The cells were lysed in buffer containing 100 mM HEPES, 1 mM EDTA, 2.5% SDS, and 2% MMTS (Sigma), adjusted to pH 7.5, sonicated, and left to rotate at 40°C overnight. Acetone precipitation was performed to remove MMTS: 33 volume of cold acetone were used to precipitate the protein for 20 min at 220°C, before spinning down in standard bench-top ultracentrifuge at 5000 3 g for 1 min. The supernatant was discarded, and the pellet was washed three times with 70% acetone, each time discarding the supernatant. Protein pellet was left to dry in air and stored overnight at 220°C. 12 h later, the protein was resuspended in 500 ml of "binding" buffer containing 100 mM HEPES, 1 mM EDTA, and 1% SDS, adjusted to pH 7.5, sonicated, and vigorously shaken for 1 h, before splitting the protein sample into three 1.5-ml tubes, one with 40 ml for "unmanipulated" starting material and two with 220 ml for the palmitoylation assay (one for 1HA and one for 2HA condition). A 1:1 slurry of preactivated thiopropyl-Sepharose beads (GE) was prepared using binding buffer (50-mg beads = 250 ml of binding buffer). 50 ml of the activated bead slurry was added to each 220-ml lysate. 50 ml of freshly prepared 2 M HA (Sigma), adjusted to pH 7.5, were then added to the lysate designated "1HA," whereas 50 ml of 2 M NaCl were added to the sample designated "2HA." Hydroxylamine/bead/lysate mixtures were left to incubate at room temperature for 2.5 h, rotating. To wash out the hydroxylamine and NaCl, the beads were spun at 5000 3 g for 1 min, and the supernatant was removed and discarded. Bead resin was washed five times with 1 ml of binding buffer, each time spinning at 5000 3 g for 1 min and discarding the supernatant to recover the beads. Palmitoylated proteins were eluted using 50 ml of 53 sample buffer supplemented with 100 mM DTT. The samples were heated at 65°C for 10 min, separated on a 10% SDS-PAGE gel, transferred to nitrocellulose, and probed with anti-V5, using anti-flotillin-1 as a positive control.

Acyl PEG exchange
Stably transfected cells were grown to ;90% confluence in 100-mm tissue culture plates. The cells were lysed with blocking buffer consisting of 100 mM HEPES, 150 mM NaCl, 5 mM EDTA, 2.5% SDS, and 200 mM tris(2-carboxyethyl)phosphine (Sigma), adjusted to pH 7.5, sonicated, and rotated at room temperature for 1 h. After 1 h, 12.5 ml of freshly prepared 1 M N-ethylmaleimide (NEM) (dissolved in ethanol) (Sigma) for 25 mM final NEM concentration were added to each lysate, with rotation overnight at room temperature. To scavenge the NEM, 12.5 ml of 2,3-dimethyl 1,3-butadiene (Sigma) were added to each sample and rotated vigorously for 1 h at room temperature. 100 ml of chloroform were added to each sample, vortexed vigorously for 1 min, and centrifuged at maximum speed for 3 min to achieve phase separation. The supernatant on top of the resulting "protein pancake" was split into three 1.5-ml tubes: one containing 40 ml for "unmanipulated" starting material, one containing 100 ml designed for the 1HA condition, and one containing 100 ml designated for the 2HA condition. 20 ml of 2 M freshly prepared HA (Sigma), adjusted to pH 7.5, were added to the lysate designated 1HA, whereas 20 ml of 2 M NaCl were added to the lysate designated 2HA. The HA/lysate mixture was incubated for 2 h at 40°C, rotating. The HA and NaCl were desalted using a pre-equilibrated 40K MWCO Zebaspin desalting column (Thermo Fisher). 10 ml of a freshly prepared 20 mM stock of 5-kDa mPEG-maleimide (Sigma) (dissolved in water) were added to each desalted lysate and incubated for 2 h at 40°C, rotating. 100 ml of 5 3 sample buffer supplemented with 100 mM DTT were added to stop the mPEG-maleimide alkylation reaction and heated for 10 min at 65°C. The samples were separated on a 10% SDS-PAGE gel, blotted to nitrocellulose, and probed with anti-V5.

Fyn kinase assay
Fyn kinase assays were performed according to manufacturer's recommendations (Fyn kinase assay kit, Promega). The reactions were performed in triplicate, and each reaction contained 200 ng of active GST-tagged Fyn kinase, 50 mM Ultrapure ATP, 0.2 mg/ml peptide substrate, 50 mM DTT diluted in a standard kinase reaction buffer (contents of 53 buffer: 40 mM Tris, pH 7.5, 20 mM MgCl 2 , 0.1 mg/ml BSA). b1 peptides corresponded to intracellular domain of b1 surrounding Tyr-181 (amino acids 175-185; b1 peptide, QENASEYLAITC; pYb1 peptide, QENASEpYLAITC). The poly-E 4 Y 1 peptide is a wellcharacterized substrate of fyn kinase. Kinase reactions lacking substrate were used to normalize the kinase activity in substrate-containing reactions. Three independent experiments were performed. Statistical significance was determined with Student's t test.

Statistics
Statistical analysis for cleavage assay experiments were completed with n = 3-4 for each experiment. The data are represented as the means 6 S.E. b1 mutant cleavage experiments were completed as one-way ANOVA with multiple comparisons to WTb1 treated with DAPT. For the fyn kinase assay, three independent experiments were performed. Statistical significance was determined with Student's t test. The data are represented as the means 6 S.E. Electrophysiology experiments had an n of 10-15 cells/condition for each experiment. The voltage dependence of activation and inactivation were compared using nonlinear fit, maximum current was analyzed using one-way ANOVA with multiple comparisons, and current