β1- and β3- voltage-gated sodium channel subunits modulate cell surface expression and glycosylation of Nav1.7 in HEK293 cells

Voltage-gated sodium channels (Navs) are glycoproteins composed of a pore-forming α-subunit and associated β-subunits that regulate Nav α-subunit plasma membrane density and biophysical properties. Glycosylation of the Nav α-subunit also directly affects Navs gating. β-subunits and glycosylation thus comodulate Nav α-subunit gating. We hypothesized that β-subunits could directly influence α-subunit glycosylation. Whole-cell patch clamp of HEK293 cells revealed that both β1- and β3-subunits coexpression shifted V½ of steady-state activation and inactivation and increased Nav1.7-mediated INa density. Biotinylation of cell surface proteins, combined with the use of deglycosydases, confirmed that Nav1.7 α-subunits exist in multiple glycosylated states. The α-subunit intracellular fraction was found in a core-glycosylated state, migrating at ~250 kDa. At the plasma membrane, in addition to the core-glycosylated form, a fully glycosylated form of Nav1.7 (~280 kDa) was observed. This higher band shifted to an intermediate band (~260 kDa) when β1-subunits were coexpressed, suggesting that the β1-subunit promotes an alternative glycosylated form of Nav1.7. Furthermore, the β1-subunit increased the expression of this alternative glycosylated form and the β3-subunit increased the expression of the core-glycosylated form of Nav1.7. This study describes a novel role for β1- and β3-subunits in the modulation of Nav1.7 α-subunit glycosylation and cell surface expression.


INTRODUCTION
Voltage-gated sodium channels (Na v s) are large glycoprotein complexes responsible for the initial rising phase of the action potential in excitable cells. They are composed of a highly processed α-subunit and are associated to one or more β-subunits (Brackenbury and Isom, 2011). The α-subunit is the poreforming unit of the channel through which the Na + ions pass (Catterall, 2000). Nine genes encoding Na v α-subunits have been found in the human genome. In addition, four genes coding for the different Na v β-subunits have been identified: SCN1B (Isom et al., 1992;Kazen-Gillespie et al., 2000), SCN2B (Isom et al., 1995a), SCN3B (Morgan et al., 2000) and SCN4B (Yu et al., 2003) coding for β1to β4-subunits, respectively. The α-subunit is composed of four homologous domains (Noda et al., 1984). Each of these domains contains six α-helical transmembrane domains (S1-S6). S1-S4 form the voltage-sensing domains and thus regulate α-subunit opening. S5 and S6 form the pore of the channel (Guy and Seetharamulu, 1986;Payandeh et al., 2011). The poreforming α-subunit permits the flow of Na + , but its biophysical properties are modulated by the β-subunits (Isom et al., 1995b), most likely via direct interference with gating . The influence of β-subunits on the biophysical properties of the recorded sodium current (I Na ) vary with cell type, possibly due to different endogenous β-subunit expression and the presence of different partner proteins (Meadows and Isom, 2005). The β-subunits also participate in cell-cell adhesion and cell migration via the interaction with the extracellular matrix and cytoskeletal molecules. They also serve as important signaling molecules (Isom, 2001). Naturally occurring genetic variants in humans and genetically modified animal models have shown that β-subunits are implicated in numerous diseases, i.e., pain, epilepsy, migraines and cardiac arrhythmias (Brackenbury and Isom, 2011). This highlights their importance in the regulation of cellular excitability. The β-subunits are composed of an extracellular immunoglobulin-like domain in the N-terminal region, a single transmembrane segment and an intracellular carboxyterminus tail (Isom et al., 1992). β1and β3-subunits interact with the α-subunit via a non-covalent bond (Hartshorne et al., 1982), while β2and β4-subunits are covalently linked to the αsubunit via disulfide bonds (Hartshorne et al., 1982;Messner and Catterall, 1985;Yu et al., 2003).
Out of the total pool of Na v s, most of the α-subunits are localized intracellularly: in the endoplasmic reticulum (ER) for synthesis, in the Golgi where post-translational modifications occur and in the secretory pathway where they are trafficked to the plasma membrane to exert their main functions (Schmidt et al., 1985;Ritchie et al., 1990;Okuse et al., 2002). In the ER and the Golgi, Na v α-subunits undergo extensive sequential glycosylation (Waechter et al., 1983;Schmidt and Catterall, 1987), a process involving the addition of N-acetylglucosamine capped by sialic acid residues and the sequential addition of oligosaccharide chains. Glycosylation can account for up to 30% of the α-subunit molecular weight (Messner and Catterall, 1985). Protein glycosylation serves various functions such as protein folding, cell signaling, protection from proteases, cell-cell adhesion and regulation. It has also been implicated in development and immunity (Moremen et al., 2012). Glycosylation modifies the gating properties of the Na v α-subunits (Recio-Pinto et al., 1990;Bennett et al., 1997;Zhang et al., 1999;Tyrrell et al., 2001), most likely by interfering with the electric field near the gating sensors (Bennett et al., 1997;Cronin et al., 2005;Ednie and Bennett, 2012).
Because both β-subunits and glycosylation modify the intrinsic biophysical properties of the Na v α-subunit, we hypothesized that β-subunits might directly influence α-subunit glycosylation. This study investigated the effect of the four β-subunits on the Na v 1.7-mediated current when co-expressed in HEK293 cells. Each of the four β-subunits influenced the biophysical properties and kinetics of the Na v 1.7-mediated current to varying degrees, but only the β1and β3-subunits increased Na v 1.7 current density. Cell surface biotinylation and subsequent deglycosylation of the samples revealed the presence of differentially glycosylated forms of Na v 1.7 in the cell; a coreglycosylated and a fully-glycosylated form of Na v 1.7. β1and β3-subunits mediated the differentially glycosylated form of Na v 1.7 and enhanced its expression at the membrane. This suggests that the increase in Na v 1.7 I Na may be explained by a glycosylation-dependent stabilization of Na v 1.7 at the cell membrane. This work reveals a novel mechanism by which Na v β-subunits modulate α-subunit glycosylation and cell surface density.

CELL CULTURE AND TRANSFECTION
Human embryonic kidney (HEK293) cells were cultured in DMEM medium supplemented with 10% FBS, 4 mM Glutamine and 20 μg/ml Gentamicin, at 37 • C in a 5% CO2 incubator (Life Technologies Inc.). For patch clamp experiments, 1 μg of Na v 1.7 cDNA concomitantly with 0.4 μg of a β-subunit and 0.8 μg EBO-pCD-Leu2-CD8 cDNA encoding CD8 antigen as a reporter gene were transfected using the Ca 2+ -phosphate method in a T25 (∼2 × 10 6 cells). For biotinylation and deglycosylation assays, HEK293 cells were transiently co-transfected with 6 μg of Na v 1.7 and 6 μg of each β-subunit or empty vector mixed with 30 μl JetPEI (Polyplus-Transfection) and 250 μl 150 mM NaCl in a P100 dish (∼9 × 10 6 cells, BD Falcon). For a negative control, the cells were transfected with 12 μg of empty vector. The cells were used in patch clamp or biochemical experiments 48 h post transfection.

CELL SURFACE BIOTINYLATION ASSAY
HEK293 cells transiently co-transfected were treated with 0.5 mg/ml EZ-link™ Sulfo-NHS-SS-Biotin (Thermo Scientific) in cold 1X PBS for 15 min at 4 • C. The cells were then washed twice with 200 mM Glycine in cold 1X PBS to inactivate biotin, and twice with cold 1X PBS to remove excess biotin. The cells were then lysed with 1X lysis buffer [50 mM HEPES pH 7.4; 150 mM NaCl; 1.5 mM MgCl 2 ; 1 mM EGTA pH 8; 10% Glycerol; 1% Triton X-100; 1X Complete Protease Inhibitor Cocktail (Roche)] for 1 h at 4 • C. Whole cell lysates were centrifuged at 16,000 g at 4 • C for 15 min. 2 mg of the supernatant was incubated with 50 μl Streptavidin Sepharose High Performance beads (GE Healthcare) for 2 h at 4 • C, and the remaining supernatant was kept as input. The beads were subsequently washed five times with 1X lysis buffer before elution with 50 μl of 2X NuPAGE sample buffer (Invitrogen) and 100 mM DTT at 37 • C for 30 min. These biotinylated fractions were analyzed as Na v 1.7 expression at the cell surface. The input fractions, representing total expression of Na v 1.7, were resuspended with 4X NuPAGE sample buffer plus 100 mM DTT to give a concentration of 1 mg/ml (60 μg/well) and were then incubated at 37 • C for 30 min.

DEGLYCOSYLATION ASSAY
For the total fractions, 60 μg of proteins of whole cell lysates were denatured at 37 • C for 30 min in the presence of 1X Glycoprotein denaturing buffer. The denatured protein lysates were subsequently incubated at 37 • C for 1 h with 1500 units PNGaseF (New England Biolabs) in the presence of 1X NP-40 and 1X G7 buffer to cleave most of the high mannose, hybrid and complex oligosaccharides from N-linked glycoproteins. The reaction was stopped by adding 4X NuPAGE sample buffer plus 100 mM DTT and incubating them at 37 • C for 30 min. For the biotinylated fractions, 35 μl ddH 2 O were added into Streptavidin Sepharose High Performance beads previously incubated with whole cell lysate and denatured at 37 • C for 30 min in the presence of 1X glycoprotein denaturing buffer. The denatured proteins bound to Streptavidin beads were subsequently incubated at 37 • C for 1 h with 2000 units of PNGaseF in the presence of 1X NP-40 and 1X G7 buffer. Following this incubation step, the beads were washed five times with the same lysis buffer used in the biotinylation assay and eluted with 2X NuPAGE sample buffer and 100 mM DTT at 37 • C for 30 min.

ELECTROPHYSIOLOGY
Twenty-four hours after transfection, cells were split at low density and whole-cell recordings were performed 48 h after transfection. Anti-CD8 beads (Dynal, Oslo, Norway) were used to identify transfected cells. Whole cell patch-clamp recordings were carriedout using an internal solution containing 60 mM CsCl, 70 mM Cs Aspartate, 11 mM EGTA, 1 mM MgCl 2 , 1 mM CaCl 2 , 10 mM HEPES, and 5 mM Na 2 -ATP, pH 7.2 with CsOH and an external solution containing 130 mM NaCl, 2 mM CaCl 2 , 1.2 mM MgCl 2 , 5 mM CsCl, 10 mM HEPES, 5 mM glucose, pH 7.4 with CsOH. Data were recorded with a VE-2 amplifier (Alembic Instruments, Montreal, Canada) or an Axon amplifier 700A and analyzed using pClamp software (version 8, Molecular Devices), Kaleidagraph (version 4.03) and MatLab. The sampling interval was set to 5 μs (200 kHz) and low-pass filtering to 5.0 kHz. Resistance of the borosilicate pipettes (World Precision Instruments, Sarasota, FL, USA) was 2-6 M . Leakage current was subtracted using the P/4 procedure. I Na densities (pA/pF) were obtained by dividing the peak I Na by the cell capacitance obtained from the pClamp function. Voltage dependence of activation (SSA) curves were determined from I/V curves where the Na + current was evoked from a holding potential of −100 mV to test pulses of 100 ms ranging from −120 to +30 mV in increments of 5 mV. The linear ascending segment of the I/V relationship was used to estimate the reversal potential for each trace. Time constant of inactivation was determined by fitting the current decay with the Levenberg-Marquardt single exponential function. The time constant was plotted against the test voltage, with I = A * exp(−t/τ) + C: where I is the current, A is the percentage of channel inactivation with the time constant τ, t is time and C if the steady-state asymptote. Steady-state inactivation curves (SSI) were measured from a holding potential of −120 mV using 500 ms prepulses to the indicated potentials, followed by a test pulse to 0 mV. To quantify the voltage-dependence of SSA and SSI, data from individual cells were fitted with the Boltzmann relationship, y ( , in which y is the normalized current or conductance, V m is the membrane potential, V ½ is the voltage at which half of the available channels are inactivated and k is the slope factor. Recovery from inactivation curves (RFI or "repriming") were obtained with a standard two-pulse protocol consisting of a depolarizing pulse from a holding potential of −120 to 0 mV for 50 ms to inactivate the channels, followed by a variable duration (from 0.5 to 3000 ms) step back to −120 mV to promote recovery. Channel availability was assessed with the first standard test pulse at 0 mV. The normalized currents of the second pulse at 0 mV were plotted against the recovery interval. We calculated t ½ (ms), the time necessary for half of the channels to recover from the first pulse, by interpolation from a linear relation between the 2 points juxtaposing half recovery (y 1 < 0.5 < y 2 ), using the equation x = [0.5-(y 1 x 2 -y 2 x 1 )/(x 2 -x 1 )] * (x 2 -x 1 )/(y 2 -y 1 ).

QUANTITATIVE REAL-TIME REVERSE TRANSCRIPTION PCR (qRT-PCR)
HEK293 cells transfected with Na v 1.7 (1 μg) alone or with each of β-subunits (0.4 μg) were collected in RNA-later solution (Qiagen, Basel, Switzerland). mRNA was extracted and purified with RNeasy Plus Mini kit (Qiagen) and quantified using RNA 6000 Nano Assay (Agilent Technologies AG, Basel, Switzerland). A total of 600 ng of RNA was reverse transcribed for each sample using Omniscript reverse transcriptase (Qiagen). Na v 1.7 primer's sequence is as follow; 5 -TCTGTCTGAGTGTGTTTGCACTAA-3 and 5 -AAGTCTTCTTCACTCTCTAGGGTATTC-3 . We used GAPDH as reference gene to normalize Na v 1.7 mRNA expression. Gene-specific mRNA analyses were performed using the iQ SYBR-green Supermix (BioRad, Reinach, Switzerland) and the iQ5 real-time PCR detection system (BioRad). Only reactions with appropriate amplification and melting curves determining the amplicon specificity were analyzed. For all conditions tested we used n = 3 samples. All samples were run in triplicate.

STATISTICS
For electrophysiological experiments (current densities and biophysical properties) normality with D'Agostino-Pearson was tested to determine whether a regular One-Way ANOVA and post-hoc Bonferroni tests, or the non-parametric equivalent test (Kruskal-Wallis test and Dunn post-hoc tests), should be performed. For RFI, a Two-Way ANOVA was used to compare Na v 1.7 alone with Na v 1.7 co-transfected with each β-subunit, and the impact of the voltage on this comparison. Biochemical experiments and transcriptional quantification data were analyzed using bilateral Student's t.

RESULTS
The functional impact of the co-expression of the four βsubunits on Na v 1.7-mediated I Na was studied by performing whole cell patch-clamp experiments in HEK293 cells. Each βsubunit was independently co-transfected with Na v 1.7 and then compared to Na v 1.7 expressed alone. Figure 1A shows typical traces of Na v 1.7 I Na obtained with a current-voltage protocol. A hastening of the Na v 1.7 current decay kinetics was observed with each of the β-subunits tested ( Figure 1B). The shortening of the Na v 1.7 time constant of current decay was observed for a wide range of voltages and showed voltage-dependency for every β-subunit ( Figure 1B). The shortening was particularly prominent for the β3-subunit. In addition, β1and β3subunits also significantly increased (∼2-fold) Na v 1.7-mediated current density as compared to Na v 1.7 alone or to Na v 1.7 coexpressed with β2or β4-subunits ( Figure 1C and Table 1). We also observed that β2 and β4-subunits did not antagonize β1 and β3-subunits-dependent up-regulation, and that the two latter have additive positive effect on Na v 1.7-mediated current (data not shown).
Whether the I Na density increase mediated by both β1and β3-subunits was also accompanied by alterations of other Na v 1.7 biophysical properties was also assessed. The voltage dependence of macroscopic I Na activation and inactivation (see Materials and Methods) of Na v 1.7 in the absence and presence of each βsubunit was recorded and analyzed. The co-transfection of the β1subunit significantly shifted the V ½ of steady-state inactivation Na v 1.7 + β3 Na v 1.7 + β2 Na v 1.7 + β4 Na Typical whole-cell Na + currents of HEK293 cells transfected with Na v 1.7 alone or Na v 1.7 co-expressed with individual β-subunits elicited with a typical current-voltage protocol. (B) Voltage-dependence of current decay of Na v 1.7 alone compared to Na v 1.7 with each individual β-subunit. Inset: Normalized representative current traces of Na v 1.7 elicited by test pulses at 0 mV. Co-transfection of the β1-(p = 0.011, n = 39), β2-(p < 0.0001, n = 25), β3-(p < 0.0001, n = 16), and β4-subunit (p = 0.006, n = 13) decreased the time constant decay as compared to Na v 1.7 alone (n = 81). Two-Ways ANOVA and Bonferroni post-hoc tests. Data are expressed as mean ± s.e.m. (C) I Na densities from HEK293 cells transfected with Na v 1.7 alone or co-transfected with individual β-subunits. β1-(n = 55) and β3-subunits (n = 34), but not β2-(n = 67) nor β4-subunits (n = 27), increased the Na v 1.7 current densities. p < 0.0001 for β1and β3-subunits with One-Way ANOVA followed by Bonferroni post-hoc tests. Data are expressed as mean ± s.e.m. and were normalized to Na v 1.7 alone for each experiments. Values can be found in Table 1. * p < 0.05, * * p < 0.01 and * * * p < 0.001. toward depolarized potentials by ∼5.8 mV, but had no influence on V ½ of activation (Figure 2A and Table 1). The β3-subunit shifted the V ½ of inactivation toward depolarized potentials by ∼3.5 mV and the V ½ of activation toward hyperpolarized potentials by ∼3.7 mV ( Figure 2C and Table 1). Neither the β2nor β4-subunits affected Na v 1.7 voltage dependence of activation or inactivation (Figures 2B,D and Table 1).
The influence of the β-subunits on recovery from inactivation (RFI) was also tested. Because the RFI relationships could not always be fitted with the exponential functions to the same degree, an interpolation from a linear relation between the 2 points juxtaposing half recovery to obtain the half-time (t ½ ) of RFI was used. Only the β1-subunit significantly hastened t ½ of RFI to 6.19 vs. 7.55 ms for the control (Figures 3A-D and Table 1).
As only minor modifications of the I Na biophysical properties were observed, it is unlikely that the 2-fold increase in the Na v 1.7 current density mediated by β1and β3-subunits is only due to alterations of the single channel properties. Whether the increase of the Na v 1.7 current may have been due to an increase in channel synthesis was investigated. Na v 1.7 mRNA levels remained unchanged with β-subunit co-transfection, as observed by q-RT-PCR (Figure 4), discounting this hypothesis.  to Na v 1.7 co-expressed with individual β-subunits. A minor but significant effect on the V ½ ofactivation wasobserved forβ3-subunit(n = 17, hyperpolarizing shift, p < 0.05) co-expression as compared to Na v 1.7 alone (n = 81). β1-(n = 39), β2-(n = 25)andβ4-subunits(n = 13)didnotmodifytheV ½ ofactivationofNa v 1.7. The effect on the V ½ of inactivation was highly significant for β1-(n = 51, depolarizing shift with p < 0.0001) and β3-subunit (n = 18, depolarizing shift with p < 0.0001) compared to Na v 1.7 alone (n = 92); whereas β2-(n = 43) and β4-subunits (n = 20) had no effect. Individual points are the mean ± s.e.m. of the normalized current at each voltage point. The smooth curves are Boltzmann fits whose equations give both the V ½ of activation and inactivation (midpoints) and their associated slope factors (see Materials and Methods). The V ½ of steady-state activation and inactivation comparing Na v 1.7 alone vs. the co-expression with each subunit were tested by One-Way ANOVA followed by Bonferroni's multiple comparison tests. Values and statistics can be found in Table 1.
Whether the Na v 1.7-mediated I Na upregulation could be due to an increase of Na v 1.7 protein density at the cell membrane was investigated by performing biotinylation of plasma membrane proteins. After lysis, proteins were sampled under reducing conditions known to dissociate the covalently bound β2and β4-subunits from α-subunits (Messner and Catterall, 1985). Co-transfection of β1-, β2-, and β4-subunits significantly decreased Na v 1.7 protein expression in the total cell lysate fraction (input, Figure 5A). The quantification revealed a ∼2-fold decrease for each of these three subunits. Co-transfection of the β3-subunit had no effect on Na v 1.7 expression in the total cell lysate fraction. In the biotinylated membrane fraction two bands at different apparent molecular weights were observed when Na v 1.7 was expressed alone (white and black arrow heads in Figure 5A). These bands correspond to different glycosylated states of Na v 1.7 as demonstrated by using deglycosylating enzymes ( Figure 5B). Endoglycosidase H (EndoH) only cleaves core N-glycans from proteins whereas Peptide-N-Glycosidase F (PNGaseF) does not discriminate between full and core glycosylated proteins. Of the two bands of biotinylated Na v 1.7, only Frontiers in Cellular Neuroscience www.frontiersin.org August 2013 | Volume 7 | Article 137 | 5 Na v 1.7 + β1 Na v 1.7 + β3 Na v 1.7 + β2 Na v 1.7 + β4

FIGURE 3 | Only β1-subunit significantly increases Na v 1.7 recovery from inactivation (RFI). (A-D) RFI from HEK293 cells transfected with
Na v 1.7 alone or co-transfected with individual β-subunits. Smooth curves were fitted incorporating a geometric weight to arrive at the final curve (no equation) and the t ½ was calculated by interpolation on the x-axis from a linear relation between the 2 points juxtaposing half recovery (y 1 < 0.5 < y 2 , see Materials and Methods). Only when co-expressed with β1-subunit (n = 27, p < 0.01) was Na v 1.7 RFI significantly faster as compared to Na v 1.7 alone (n = 72). β2-(n = 34), β3-(n = 16), and β4-subunits (n = 19) did not significantly alter RFI when co-expressed with Na v 1.7. Individual points are the mean ± s.e.m. of the normalized current at each time point. Non-parametric One-Way analysis of variance (Kruskal-Wallis test) with Dunn post-hoc tests to compare each subunit co-expressed with Na v 1.7 vs. Na v 1.7 alone. showing transcriptional levels of Na v 1.7 in control conditions (co-transfected with empty vector) over Na v 1.7 levels when cells were co-transfected with each individual β-subunit. Transcripts were normalized to GAPDH and run in triplicate. Data represent the mean ± s.e.m., n = 3 independent transfections for every condition.
the lower was sensitive to EndoH and was shifted to an apparent lower molecular weight band (compare white arrowhead in the first lane to gray arrowhead in the third lane), indicating that this band corresponds to the core-glycosylated form of the channel (Figure 5B). Because PNGaseF was able to digest both bands, it can be proposed that the higher band corresponds to the fully-glycosylated form of Na v 1.7. The lower band of biotinylated Na v 1.7 migrates at the same apparent molecular weight as the band observed in the input fraction (white arrow heads in Figure 5A) suggesting that most of Na v 1.7 in the intracellular pool is core-glycosylated. This is consistent with the channel being early and rapidly, but only partially, glycosylated after its synthesis. The upper band in the total cell lysate fraction was faint and blurry (Figure 5A), suggesting that the fully-glycosylated channel only represents a small fraction of the total Na v 1.7 cellular pool. It was only by enriching the membrane proteins through the precipitation of the biotinylated membrane fraction (the ratio between the amount of lysate protein loaded and the amount of streptavidin beads needed to precipitate biotinylated proteins was ∼1:30) that the upper band was distinctly observed. Co-expression of the β1-subunit reproducibly shifted the upper band to an intermediate migrating band of lower apparent molecular weight. This suggests that the β1-subunit mediates an alternative glycosylated form of Na v 1.7. When comparing the β1-subunit-modified intermediate band with the upper band of the control condition (Na v 1.7 alone), a significant increase in signal intensity was observed ( Figure 5A, quantification), which is consistent with the increase in the Na v 1.7 current density ( Figure 1C). Co-transfection of the β2-subunit neither modified the glycosylation pattern nor the expression of any of the two bands, consistent with the fact that the current density was not modified. β3-subunit expression also altered the Na v 1.7 band pattern in the biotinylated fractions. The upper band overlapped with the lower band under the migrating conditions used. The β3-subunit significantly increased (∼7-fold) the intensity of the lower band as compared to the lower band of control, consistent with the increase of the Na v 1.7 current density elicited by the β3-subunit ( Figure 1C). Finally, β4-subunit cotransfection led to a small but significant decrease of the lower band.
To confirm that the different bands observed when β-subunits are coexpressed, particularly β1 and β3-subunits, represent alternative glycosylated form of Na v 1.7, we again incubated the input and biotinylated fractions with PNGaseF. A small but consistent shift of the Na v 1.7 band into a lower apparent molecular weight band in the total cell lysate fraction was observed (the white arrow heads shifted to the gray arrow heads, Figure 6). Furthermore, when incubating the biotinylated fraction of β-subunit and Na v 1.7 co-expression experiments with PNGaseF, all of the Na v 1.7 bands shifted to a single band (gray arrow heads) of the same molecular weight. This confirms that the β1and β3-subunits modulate differential glycosylation patterns on Na v 1.7 (Figure 6, black and white arrow heads). Noteworthy, when comparing the single band of Na v 1.7 when samples are treated with PNGaseF in the input fraction with the one in the biotinylated fraction, it seems that this band migrates slower in the input as compared to biotinylated fraction when β-subunits are coexpressed (compare bands highlighted by gray arrows for each blots). This may be due to other post-translational modification such as sialylation or palmitoylation of the channel. Further experiments using desialylation or depalmitoylation treatment are needed to confirm this possibility.  FIGURE 5 | β1and β3-subunit mediate differential forms of Na v 1.7 whose expression is increased at the membrane. (A) Representative western blot of a biotinylation assay with total lysate (input, left) and cell surface (biotinylation, right) fractions from HEK293 cells transiently transfected with Na v 1.7 alone, or co-expressed with each individual β-subunit and the associated quantifications. Input: Na v 1.7 is detected in two forms: a fast migrating band (∼250 kDa, that will be referred to as lower band) that consist mostly of the Na v 1.7 immunoreactive signal and a slow migrating band (∼280 kDa, that will be referred to as upper band). β1-(p = 0.006), β2-(p = 0.003), and β4-subunits (p = 0.009) significantly decreased Na v 1.7 expression, whereas the β3-subunit had no effect (p = 0.570). Because the upper band was below the sensitivity threshold, both bands were quantified together. Biotinylation: Na v 1.7 membrane protein is detected in three forms. When expressed alone, one lower band (white triangle, ∼250 kDa) and one upper band (black triangle, ∼280 kDa) were present (for identification of these bands, see Panel B). When the β1-subunit is co-expressed, the upper band was clearly shifted into an intermediate migrating band (∼260 kDa) with increased expression (p = 0.047). β2and β4-subunits revealed the same pattern as when Na v 1.7 was transfected alone and did not change its expression, except for the small decrease of the lower band when the β4-subunit is co-transfected (p = 0.020). The β3-subunit clearly increased Na v 1.7 immunoreactivity of the lower band (p < 0.0001). For input and biotinylation fractions, actin and the α1-subunit of NaK-ATPase were used as biotin leakiness and loading controls, respectively. Data represent mean ± s.e.m, n = 4 independent experiments. Student's unpaired t-test, each condition being compared with Na v 1.7. * p < 0.05, * * p < 0.01 and * * * p < 0.001. (B) Representative western blot and identification of glycosylation state of Na v 1.7 in biotinylated fraction from HEK293 cells transiently transfected with Na v 1.7. EndoH only cleaves the lower band of biotinylated Na v 1.7, demonstrating that this band represents the core-glycosylated form of the channel. The upper band is digested by PNGaseF, demonstrating that it corresponds to fully-glycosylated form of the channel. PNGaseF can also digest the core-glycosylated form of Na v 1.7.

Input biotinylation
Na v 1.7 + β1 Na v 1.7 + β3 Na v 1.7 + β2 Na v 1.7 + β4 Na v 1.7 Input biotinylation PNGaseF non-dig. FIGURE 6 | The different forms of Na v 1.7 observed with β-subunits are due to differential glycosylation patterns. Western blot of a biotinylation assay followed by deglycosylation with total lysate and cell surface fractions from HEK293 cells transiently transfected with Na v 1.7 alone, or co-expressed with each individual β-subunit. Samples were non-treated or treated with Peptide: N-Glycosidase F (PNGaseF) to remove glycosylated residues of the protein. The total lysate Na v 1.7 band (black/white triangle) was slightly shifted to an apparent lower molecular weight (gray triangle) when treated with PNGaseF. In the biotinylation fraction, the pattern of Na v 1.7 glycosylation by the β-subunits was the same as in Figure 4.(black and white triangles). When treated with PNGaseF, all the different bands shifted to the lower band of the same apparent molecular weight, irrespective of the β-subunit co-expressed.

DISCUSSION
This study demonstrates that Na v β1and β3-subunits modulate the cell surface expression and glycosylation patterns of Na v 1.7 when co-expressed in HEK293 cells. It also confirms that the β-subunits differentially modulate the biophysical properties of Na v 1.7.
The observation that β1and β3-subunits strongly increased Na v 1.7 I Na density contrasts with several recent studies. The study performed by Ho et al. (2012) showed no impact of any of the βsubunits on Na v 1.7 current density in HEK293 cells. The other studies showed no impact using different cell expression systems (Sangameswaran et al., 1997;Morgan et al., 2000;  After synthesis in the rough endoplasmic reticulum (RER), α-subunits are rapidly folded and undergo a first step of glycosylation in the smooth endoplasmic reticulum (SER). It is known that N-acetlyglucosamine and oligosaccharide chains are bound on Asp residues of the protein, a process known as core-glycosylation. Newly synthesized glycoproteins are then translocated into the Golgi network, where they are subject to a second step of more complex glycosylation, involving many different enzymes. Once matured, proteins eventually translocate to the plasma membrane. The present findings suggest that core-glycosylated proteins can also be found anchored at the membrane. By some yet undefined mechanism, the β1-subunit interferes with the second glycosylation step ( * * * full-glycosylation on the scheme) which leads to a modification of the glycosylation pattern of the α-subunits. The β1-subunit enhances the core-glycosylated form of Na v 1.7. This suggests that the β1and β3-subunits already interact with the α-subunits before the step of full-glycosylation of the channel occurring in the Golgi network. By enhancing the differential glycosylation pattern of Na v 1.7, it can be proposed that the β1and β3-subunits promote stabilization of the channel at the plasma membrane. On the contrary, it is likely that the β2and β4-subunits, which have no effect on the glycosylation nor the anchoring, only briefly interact with the α-subunits before translocation of the channel to the plasma membrane. For sake of simplicity, the shown glycosylation patterns are arbitrary. Na v 1.7 is depicted as being "freely" expressed in ER/Golgi and membrane networks for easier interpretation of the scheme. However, Na v 1.7 is embedded in the membranes of the different organelles.
et al. , 2001, 2004). These discrepant observations underline the influence of the cellular background when studying Na v s αsubunit regulation by β-subunits. Even though this study and the one by Ho et al. (2012)  interact with β-subunits. It is also possible that other differences, such as the origin of the HEK293 cells or even passage numbers, may play a significant role in the observed effects. Other studies have shown, however, that β1and β3-subunits can increase the current density of several other Na v isoforms (Nuss et al., 1995;Smith and Goldin, 1998;Fahmi et al., 2001;. A significant shortening of the time constant of current decay was observed when Na v 1.7 was co-expressed with each individual subunit. A rapid rate of inactivation tends to reduce the refractory period of Na v 1.7, meaning that β-subunits can enhance cell excitability via this mechanism. The hastening of RFI by the β1subunit might also reduce the duration of the refractory periods, allowing for faster repetitive firing of neurons. The β1-subunit shifted V ½ of inactivation toward more depolarized potentials, which should increase the number of channels available for opening in response to depolarization at a given voltage near the resting membrane potential (approximately −60 to −70 mV). The β3-subunit similarly influenced this parameter, and also shifted the V ½ of activation toward hyperpolarized potentials, rendering the channel more likely to open at hyperpolarized voltages. These results are consistent with the findings of Ho et al., and account for the shift toward a hyperexcitable state.
The fact that the β1and β3-subunits strongly increased Na v 1.7 current density, but only modestly influenced the biophysical properties, suggests that the single-channel conductance is not altered. In line with the previous, single-channel recordings revealed that β1-subunit did not change the Na v 1.5 open probability despite an important increase in current density (Nuss et al., 1995). As a consequence, it was hypothesized that these two subunits also increase Na v 1.7 channel density at the cell surface. Biotinylation of cell surface proteins was performed and a strong decrease of the Na v 1.7 signal in the input fraction when β1-, β2-, and β4-subunits were co-transfected was observed. This decrease was in contrast with the increase or lack of modification of the Na v 1.7 current density. The reason for this decrease remains to be identified, but one can speculate that the expression of the β-subunits may decrease the ratio of intracellular Na v s to plasma membrane Na v s by hastening forward trafficking and stabilizing the channel at the cell surface. Furthermore, one of the important functions of glycosylation is the proper folding and protection of proteins with respect to proteolysis (Parodi, 2000). The thus far not observed role of β-subunits in altering Na v 1.7 glycosylation might influence the degradation by the proteasome, accounting for the decrease in input. The β-subunits have also been proposed to act as chaperon proteins (Valdivia et al., 2010), further supporting a potential effect on degradation by the proteasome.
The analysis of the Na v 1.7 protein at the plasma membrane and the subsequent treatment with the deglycosylating enzyme led to three novel findings: (1) Under normal conditions, Na v 1.7 is present in two glycosylated forms, a core-glycosylated form with a molecular weight of ∼250 kDa and a fully-glycosylated form with a molecular weight of ∼280 kDa; (2) The β1-subunit can further mediate a third and intermediate migrating band, which likely represents an alternative fully-glycosylated form of Na v 1.7; and (3) the β1-subunit increases the membrane expression of this alternative fully-glycosylated form of Na v 1.7; whereas the β3-subunit increases the membrane expression of the core-glycosylated form.
In studies using chimeras between the β1and β3-subunits, it was proposed that different parts of these subunits were involved in the modulation of gating by direct interaction with the αsubunits . The observation in this study that the β1and β3-subunits can mediate the differential glycosylation of Na v 1.7, and despite the well-documented causal link demonstrating that differential glycosylation leads to a modification of Na v s α-subunit gating (Bennett et al., 1997;Zhang et al., 1999;Tyrrell et al., 2001), does not allow us to conclude that the β-subunit-mediated glycosylation of the α-subunit is responsible for the modulation of Na v 1.7 gating. Further studies to identify the potential glycosylation site of Na v 1.7 are necessary to determine whether the effects of the β-subunits on the biophysical properties of Na v 1.7 are also dependent on this mechanism. Nevertheless, the observation that the β1and β3-subunits influence the gating properties of Na v 1.7 and alter its glycosylation pattern, while the β2and β4-subunits do not, supports this hypothesis.
It was recently proposed that only the upper band of biotinylated Na v 1.7 may reflect the functional fully-glycosylated form of the channel, whereas the lower band represents an intermediate and/or immature glycosylated form of the channel which does not participate in Na + conductance (Laedermann et al., 2013). The upregulation of expression of the core-glycosylated form of Na v 1.7 when the β3-subunit is co-transfected, along with the associated 2-fold increase in the Na v 1.7 current density, suggests that there is no such dichotomy and that the link between the glycosylation and functionality of the channel is more complex. It is possible that these distinctly glycosylated forms of Na v 1.7 differentially participate in Na + conductance, but their relative contribution to the overall sodium current has yet to be determined. For instance, quantification of the shifted upper band when the β1-subunit was present revealed an increased signal intensity of ∼30%, which is less important than the 100% increase of I Na measured using the patch clamp approach. This underlines that, in addition to an increased stabilization of the channel, modification of single channel conductance by the intermediate glycosylated form of Na v 1.7 may also partially contribute to the functional 2-fold increase in I Na . This point, in addition to the fact that β-subunits have the ability to shift bands from one glycosylated state to another, demonstrates that quantification of biotinylated proteins needs to be interpreted with caution. It cannot be excluded that due to these shifts, some bands might contaminate the signal of another band.
General kinetic models of biosynthesis (Schmidt and Catterall, 1987) have proposed that the β2-subunit interacts with the αsubunit right before anchoring at the membrane. The present findings are consistent with such a model since both β2and β4subunits did not influence Na v 1.7 glycosylation, and most likely interacted after the Golgi network (Figure 7). The α-β1 complex was previously shown to associate in the ER, enhancing the trafficking to the plasma membrane . Another study demonstrated that β1and β3-subunits increase the efficiency of channel trafficking from the ER to the plasma Frontiers in Cellular Neuroscience www.frontiersin.org August 2013 | Volume 7 | Article 137 | 10 membrane (Fahmi et al., 2001). The β3-subunit has also been shown to mask the ER-retention signal (Zhang et al., 2008). The results of the present work confirm earlier demonstrated interactions between the βand α-subunits, and suggest an additional function of both: β1and β3-subunits interact with the α-subunit in the ER/Golgi, where they regulate the differential glycosylation of Na v α-subunits, which in turn modulates the stabilization of the channel at the cell membrane (Figure 7). The mechanisms underlying their interaction and mediation of α-subunit glycosylation, as well as the identification of other potential isoforms that would be subject to such regulation, remain to be investigated.

PHYSIOLOGICAL RELEVANCE
These results were obtained in cellular expression system. Whether such mechanisms also occur in native cells remain to be investigated. This study used the Na v 1.7 isoform, an important contributor to pain processing (Lampert et al., 2010). The electrophysiological results show that β1and β3-subunits are able to increase Na v 1.7 excitability, as demonstrated by the increase in peak current, kinetics, voltage-availability and repriming rate. Na v 1.7 is expressed in high levels in all types of sensory neurons (Ho and O'leary, 2011). β-subunits are also expressed in sensory neurons, but to a variable extent depending on cell type (Takahashi et al., 2003;Ho et al., 2012). Furthermore, β1and β3-subunit expression has been reported to be increased in pathological pain Coward et al., 2001), where they have been implicated in the generation of hyperexcitability. It remains to be determined if Na v function is mediated by an altered pattern of α-subunit glycosylation, which is an important regulatory process of excitability in dorsal root ganglia neurons (Tyrrell et al., 2001).