Subcellular dynamics and functional activity of the cleaved Na+ channel β1 subunit intracellular domain

The voltage-gated Na+ channel β1 subunit, encoded by SCN1B, regulates cell surface expression and gating of α subunits, and participates in cell adhesion. β1 is cleaved by α/β and γ-secretases, releasing an extracellular domain and intracellular domain (ICD) respectively. Abnormal SCN1B expression/function is linked to pathologies including epilepsy, cardiac arrhythmia, and cancer. In this study, we sought to determine the effect of secretase cleavage on β1 function in breast cancer cells. Using a series of GFP-tagged β1 constructs, we show that β1-GFP is mainly retained intracellularly, particularly in the endoplasmic reticulum and endolysosomal pathway, and accumulates in the nucleus. Reduction in endosomal β1-GFP levels occurred following γ-secretase inhibition, implicating endosomes, and/or the preceding plasma membrane, as important sites for secretase processing. Using live-cell imaging, we report β1ICD-GFP accumulation in the nucleus. Furthermore, β1-GFP and β1ICD-GFP both increased Na+ current, whereas β1STOP-GFP, which lacks the ICD, did not, thus highlighting that the β1-ICD was necessary and sufficient to increase Na+ current measured at the plasma membrane. Importantly, although the endogenous Na+ current expressed in MDA-MB-231 cells is TTX-resistant (carried by Nav1.5), the Na+ current increased by β1-GFP or β1ICD-GFP was TTX-sensitive. In addition, β1-GFP increased mRNA levels of the TTX-sensitive α subunits SCN1A/Nav1.1 and SCN9A/Nav1.7. Taken together, this work suggests that the β1-ICD is a critical regulator of α subunit function in cancer cells. Our data further highlight that γ-secretase may play a key role in regulating β1 function in breast cancer.


Plasma membrane expression and activity of β1-GFP
In this study, we used over-expression of β1-GFP in the MDA-MB-231 cell line as a model system in which to study functional consequences of proteolytic processing of β1 by secretase cleavage. MDA-MB-231 cells provide a unique model system to analyse  1-GFP over-expression did not affect the voltage at activation, voltage at half-maximal activation, rate of activation, voltage at half-inactivation, rate of inactivation, time to current peak, or membrane capacitance ( Figure 1D-G and I-M). However, β 1-GFP over-expression caused a hyperpolarisation of the voltage at Na + current peak (P < 0.05; n = 8; t-test; Figure 1H), although the small shift, together with the lack of change in voltage-dependence of activation, suggests that this change is unlikely to be physiologically important. β 1-GFP overexpression also accelerated recovery from inactivation (P < 0.01; n = 8; t-test; Figure 1N, O).
The observations that β 1-GFP over-expression (i) increases Na + current and (ii) promotes transcellular adhesion of MDA-MB-231 cells (39) suggest that it is functionally active at the plasma membrane in this cell line. We therefore examined the subcellular localisation of β 1-GFP, initially focusing on plasma membrane expression. Surprisingly, when live MDA-MB-231-β1-GFP cells were stained with the lipid dye FM4-64, no overlap in fluorescence was detected at the plasma membrane, whereas robust co-localisation was observed within internal vesicles (Figure 2A). In fact, line profile analysis revealed that peak plasma membrane FM4-64 fluorescence and β 1-GFP fluorescence were offset by ~500 nm ( Figure   2B), suggesting that β 1-GFP is not highly expressed at the plasma membrane relative to the cytosol. To ensure that the lack of surface β 1-GFP abundance was not due to FM4-64 quenching GFP fluorescence via FRET, FM4-64 was photobleached and the resulting change in GFP fluorescence monitored. Photobleaching of FM4-64 within internal vesicles caused a modest, but significant, 8.9 % increase in GFP signal (P < 0.05; n = 4; t-test; Figure 2C), suggesting that some FRET did occur between GFP and FM4-64. However, when FM4-64 was photobleached at the plasma membrane, no increase in GFP signal was detected ( Figure 2C), ruling out GFP quenching by FM4-64 as an explanation for the low abundance of β 1-GFP at the cell surface. In summary, although β1-GFP promotes Na + current, most of this protein appears to be retained intracellularly in MDA-MB-231 cells. This observation agrees with a previous study in Madin-Darby canine kidney cells, which showed that β 1 was retained intracellularly, unlike β 2, which was enriched at the plasma membrane (52).

Subcellular distribution of β1-GFP
The β 1-ICD and β 2-ICD secretase cleavage products localise to the nucleus of heterologous cells and alter gene transcription (46,48). We therefore next investigated whether any β1-GFP signal localised to the nucleus in MDA-MB-231 cells. Prior to anti-GFP antibody labelling, cells were permeabilised with either Triton X-100, which permeabilises all cellular membranes, permitting access to nuclear antigens, or digitonin, which does not permeabilise the nuclear membrane, preventing access to nuclear antigens (53). The inner nuclear membrane protein, lamin B2, was labelled strongly in Triton X-100 permeabilised cells, but not digitonin-permeabilised cells, confirming that digitonin restricted antibody access to nuclear antigens ( Figure 2D). Labelling with the anti-GFP antibody revealed a small but statistically significant 18 % reduction in nuclear:cytoplasmic fluorescence intensity ratio in digitonin-permeabilised cells compared to Triton X-100 permeabilised cells (P < 0.001; n = 26-28; t-test; Figure 2E, F), suggesting that a fraction of β 1-GFP is indeed present in the nucleus. Furthermore, cytoplasmic GFP fluorescence intensity was similar between the two permeabilization conditions (P = 0.86; n = 27; t-test; Figure 2G), whereas nuclear GFP fluorescence intensity was significantly reduced in digitonin-permeabilised cells by 23 % (P <0.05; n = 27; t-test; Figure 2G). Together, these data support the notion that there is a fraction of β 1-GFP signal which localises to the nucleus in MDA-MB-231-β1-GFP cells.
To obtain a better spatiotemporal understanding of γ -secretase-mediated cleavage of β1-GFP, we next followed GFP distribution in live MDA-MB-231-β1-GFP cells using confocal microscopy and fluorescence recovery after photobleaching (FRAP). Regions of interest (ROIs) were photobleached at the leading and trailing edges of control and DAPT-treated MDA-MB-231-β1-GFP cells ( Figure 5D, E). Interestingly, no differences in the proportion of GFP that was freely mobile, or the time taken for half-maximal fluorescence recovery, were detected at the leading or trailing edges (Table 1). This result suggests that DAPT treatment had no effect on spatiotemporal cycling dynamics of β 1-GFP.
Together, these data suggest that β 1ICD-GFP recapitulates the electrophysiological effects of full-length β 1-GFP on Na + current in MDA-MB-231 cells but does not itself promote changes in cellular morphology.

Subcellular distribution of β1ICD-GFP
To determine the extent of β 1-ICD localisation to the nucleus, we imaged β 1ICD-GFP-, β 1-GFP-and GFP-expressing cells by confocal microscopy ( Figure 6F). We used GFPexpressing cells as a control for stochastic movement of small proteins because GFP is known to diffuse throughout the cell, including into the nucleus (63). The nuclear signal for β 1ICD-GFP was higher than for β 1-GFP, consistent with not all full-length β 1-GFP being cleaved at steady state (P < 0.0001; n = 14-17; one-way ANOVA; Figure 6G). In addition, GFP and β 1ICD-GFP had similar nuclear:cytoplasmic signal density ratio (P = 0.98; Figure   6G). These data suggest that β 1ICD-GFP is present within the nucleus. However, it is possible that nuclear localisation of β 1ICD-GFP may be due to stochastic diffusion from the cytoplasm, similar to the case with GFP.
To more accurately evaluate whether β 1ICD-GFP distribution differs from GFP, we compared the mobility of both proteins using FRAP. Initially, a ROI within the cytoplasm of MDA-MB-231-GFP and MDA-MB-231-β1ICD-GFP cells was photobleached and fluorescence recovery measured ( Figure 7A, B). GFP and β 1ICD-GFP displayed similar mobility kinetics within the cytoplasm, with both proteins having a comparable mobile fraction of ~1 (P = 0.07; n = 15; Mann-Whitney U-test; Figure 7C) and time taken for half maximal fluorescence recovery (P = 0.13; n = 15; Welch's t-test; Figure 7D). These data suggest that β 1ICD-GFP and GFP have similar spatiotemporal expression within the cytoplasm of MDA-MB-231 cells.
Next, we compared the nuclear import kinetics of GFP and β 1ICD-GFP. Other cleaved ICDs, such as Notch, are trafficked to the nucleus as part of a heteromeric complex (64). Such a complex is expected to have slower import kinetics than soluble GFP, which can diffuse directly through nuclear pores. We labelled live cells with the nuclear dye, Hoechst 33342, and then photobleached the overlapping nuclear GFP fluorescence and measured fluorescence recovery over time ( Figure 7E, F). Both GFP and β 1ICD-GFP demonstrated a similar mobile fraction of ~ 1 (P = 0.08; n = 10; unpaired t-test; Figure 7G), suggesting negligible immobilised protein is present within the nucleus. However, the time taken for halfmaximal fluorescence recovery was ~2.5 fold greater for β 1ICD-GFP relative to GFP (P < 0.001; n = 10; Mann-Whitney U-test; Figure 7H), implying different nuclear import kinetics for both proteins. According to the experimentally verified cubic relationship between molecular weight and time taken for nuclear import, the five additional kilodaltons of β 1ICD-GFP should increase diffusion time through nuclear pores by ~70 % compared to GFP (65). Therefore, β 1ICD-GFP may be entering the nucleus as part of a larger protein complex.

1ICD-GFP expression on tetrodotoxin sensitivity
To further evaluate the involvement of the ICD in regulating Na + current, we over-expressed β 1STOP-GFP, which lacks the ICD (66), in MDA-MB-231 cells ( Figure 8A). β 1STOP-GFP did not significantly increase peak current density compared to GFP (P = 0.89; n = 15; oneway ANOVA; Figure 8B). This result underscores the importance of the  increased the mRNA level of SCN1A (P < 0.001; n = 3; t-test; Figure 8D) and SCN9A (P < 0.05; n = 3; t test; Figure 8I). There was also a small increase in SCN2A and SCN4A expression, although this was not statistically significant (P = 0.052 and P = 0.061, respectively; n = 3 for both; Figure 8E, G). Finally, there was a significant reduction in SCN8A expression (P < 0.001; n = 3; t-test; Figure 8H). However, given that SCN8A has previously been shown to be expressed in a truncated form in MDA-MB-231 cells (50), this reduction is unlikely to be physiologically relevant. We therefore conclude that the elevated TTX-sensitive Na + current present in β1-GFP cells is likely carried by Na v 1.1 and/or Na v 1.7 and that the regulation of these subunits by β1-GFP may, at least in part, be transcriptional.

Discussion
In this study, we show that both β1-GFP and its γ-secretase cleavage product, β1ICD-GFP, are functionally active in MDA-MB-231 breast cancer cells. We show that the majority of kinetics. Furthermore, β1ICD-GFP was necessary and sufficient to increase Na + current measured at the plasma membrane. Finally, both β1-GFP and β1ICD-GFP increased TTX sensitivity of the Na + current. We therefore propose that the proteolytically released β1-ICD is a critical regulator of α subunit function and expression in cancer cells. A strength of our study compared with previous studies (45,46,48), is that, by using GFP-tagged constructs, we were able to visualise β1/β1-ICD dynamics in live cells. However, a caveat with this approach is that we cannot exclude the possibility that the GFP tag may interfere with function of the native protein and its cleavage products. Nonetheless, our findings are generally consistent with these other reports, suggesting that any disruption of β1 function by the GFP tag may be minor. Furthermore, the APP-ICD has been shown to increase Na v 1.6 current when co-expressed in Xenopus oocytes (73). Together, our data support the notion that the β 1-ICD is a functionally active regulator of α subunit expression at the plasma membrane of MDA-MB-231 cells. However, the precise involvement of γ-secretase cleavage on these mechanisms appears complex, given that full-length β1-GFP and β1ICD-GFP both promote Na + current, and that pharmacological inhibition of γ-secretase activity had no effect. Furthermore, the β1mediated targeting of α subunits to the plasma membrane may be cell type-specific and the over-expression systems used may not accurately reflect the stoichiometric balance that occurs between endogenous α and β subunits and/or their subcellular localization. We previously showed that endogenous β1 is present in subcellular compartments in breast cancer cells (74). Further work is required to elucidate the context-dependent localization and trafficking of endogenous β1 subunits in different cell types. have been shown to promote neurite outgrowth (24,25,75). In MDA-MB-231 cells, β1-GFP, but not a mutant lacking the Ig domain, promotes neurite-like process outgrowth (38). In agreement with these observations, we found here that, in contrast to full-length β1-GFP, β1ICD-GFP was not capable of promoting process outgrowth in MDA-MB-231 cells. Thus, regulated proteolysis at the plasma membrane is likely to be a key mechanism by which the CAM function of β1 is modulated to fine-tune neurite outgrowth, neuronal pathfinding, fasciculation and cell migration (15,(19)(20)(21)(22)(23)(24)47). Interestingly, endogenous β1 expression is higher in MCF-7 cells than in MDA-MB-231 cells (39), although γ-secretase activity has been reported in both breast cancer cell lines (76). β1 expression in breast cancer cells has been shown to increase adhesion in vitro and promote neurite-like process outgrowth, tumour growth, and metastasis in vivo (38,39). These effects generally fit with emerging data indicating that expression/activity of VGSCs promotes invasion and metastasis across multiple cancer types where these channels have been shown to be expressed (77). Further work is required to establish whether variation in endogenous β1 expression between different cancer cell types may determine the impact of γ-secretase activity on β1 function.

Transfection and generation of stably transfected cell lines
MDA-MB-231 cells were transfected using jetPRIME (Polyplus) at a DNA:jetPRIME ratio

Protein extraction and western blot
Protein extraction and western blotting were carried out as described previously, with some Secondary antibodies were HRP-conjugated goat anti-mouse (1:1000, Thermo Scientific) or goat anti-rabbit (1:1000, Thermo Scientific). Chemiluminescence was detected using an iBRIGHT imaging system (Invitrogen) or X-ray film (Fujifilm) following West Dura application (5 min, Thermo Scientific). Densitometry was performed on western blot bands to estimate protein quantity using ImageJ 1.51i (81).

Confocal microscopy
Images were acquired using a Zeiss LSM 880 laser-scanning confocal microscope with Airyscan technology, using a Plan-Apochromat 63x oil immersion objective lens (NA = 1.

Nuclear localisation analysis
The nuclear:cytoplasmic mean GFP fluorescence intensity ratio was calculated from confocal images using the standard ImageJ toolkit. Nuclear fluorescence was calculated by masking the DAPI signal and measuring GFP fluorescence within the mask, both mean fluorescence intensity and total fluorescence. Mean cytoplasmic fluorescence intensity was calculated by subtracting total cellular GFP fluorescence by total nuclear GFP fluorescence and dividing the resulting value by the difference in area of the whole cell and nucleus.

Co-localisation analysis
To quantify the co-localisation between GFP and subcellular markers (Calnexin, GM130, TGN46, EEA1 or LAMP1), Pearson's correlation coefficient was calculated using the "Coloc For FRET, cells were imaged at 2.0 -4.0x zoom factor using a 512 x 512-pixel frame size.
FM4-64 was bleached using 100 iterations of the 561 nm laser (100 % laser power) at the plasma membrane or within internal vesicles. Images were acquired every 0.6 s for 25 s.

FRAP analysis
Analysis for FRAP data was adapted from (84). Images were exported to ImageJ for data acquisition using the FRAP Norm plugin. Three regions were plotted on the image: the photobleached ROI, delineated as the full width at half maximum of the encompassing cytoplasmic GFP fluorescence or the entire photobleached nucleus, for cytoplasmic and nuclear photobleaching experiments, respectively. A control region, placed elsewhere in the cell, was used to calculate the rate of photobleaching at each time point across the time series, relative to the maximal fluorescence intensity at t = 0. Lastly, a background region was placed outside the cell, which was subtracted from the other two regions at each time point. Therefore, at each time point, the ROI could be normalised to the photobleaching rate.
Finally, fluorescence intensity within the ROI was normalised to pre-bleach fluorescence intensity to obtain the final recovery curve. Two parameters were derived from these recovery curves to quantify mobility. The mobile fraction, which defines the proportion of fluorescent protein that is mobile relative to the whole population of fluorescent protein initially in the ROI, was calculated using:

FRET analysis
To analyse FRET, images were exported to ImageJ and analysed using the FRAP Norm plugin. GFP and FM4-64 fluorescence intensities were monitored within the photobleaching ROI for the duration of the time series and normalised against t = 0. FM4-64 signal was monitored to ensure photobleaching occurred. GFP fluorescence intensity before and after photobleaching was then statistically compared.

Whole cell patch clamp recording
Whole cell patch clamp recordings were performed and analysed as described previously (85). Data were collected at a sampling rate of 50 kHz and filtered at 10 kHz. Linear leak currents were removed using P/6 subtraction (86).

RNA extraction and RT-qPCR
Total RNA was extracted from 35 mm dishes of confluent cells using RNeasy Mini kit (Qiagen), according to the manufacturer's instructions. cDNA was generated from 1 µg of RNA using Reverse Transcriptase SuperScript III (RT SS III), random primers (Invitrogen), and dNTPs (Invitrogen). RNA, random primers, and dNTPs were incubated at 65°C for 5 min.

Data availability
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.
AH, CB and WB contributed to the conception and design of the work. ASH, SLH, ALC, LLI, CGB and WJB contributed to acquisition, analysis, and interpretation of data for the work.
ASH, SLH, ALC, LLI, CGB and WJB contributed to drafting the work and revising it critically for important intellectual content. All authors approved the final version of the manuscript.

Conflicts of interest
The authors declare that they have no conflicts of interest.       for t s before second depolarisation to 0 mV. t ranged from 1-500 ms. Data are mean ± SD (unpaired t-test). ns = not significant, * = P < 0.05, ** = P < 0.01.